Cable Stayed Bridge

International Association
for Bridge and Structural

Engineering (IABSE)
Proceedings
IABSE Conference
Cable-Stayed Bridges
- Past, Present and Future
The resund Construction Site, October 1998
Malm, Sweden 2-4 June, 1999
Jointly organized by the Danish and Swedish Groups of IABSE
Produced by Congrex Sweden AB
Organising Committee
Hans Ingvarsson
Ole Damgaard Larsen
Ingvar Olofsson
Karl-Otto Sicking
Erik Stoltzner
Henrik Christensen
Chairman, Sweden
Vice Chairman, Denmark
Sweden
Sweden
Vice Secretary, Denmark
Secretary, Sweden
Scientific Committee
Niels J Gimsing
Niels Peter Hj
Joo Almeida-Fernandes
Andrew S Beard
William C Brown
Kent Gylltoft
Manabu Ito
Aarne Jutila
Jrn Lauridsen
Helge Nilsson
Walter Podolny
Gnter Ramberger
Lennart Skogsberg
Man Chung Tang
Ton Vrouwenvelder
Chairman, Denmark
Secretary, Denmark
Portugal
Hong Kong, China
Great Britain
Sweden
Japan
Finland
Denmark
Sweden
USA
Austria
Sweden
USA
The Netherlands
Conference secretariat:
Congrex Sweden AB
Linngatan 89 A
P.O.Box 5619
SE-114 86 Stockholm
Sweden
Phone: +46 8 459 66 00
Fax: +46 8 661 91 25
E-mail:iabse99@congrex.se
Internet:
http://www.congrex.com
International Association for Bridge and Structural Engineering

(IABSE):
IABSE
ETH Hnggerberg
CH-8093 Zrich
Switzerland
Phone: +41 1 633 26 47
Fax: +41 1 633 12 41
E-mail: secretariat@iabse.ethz.ch
Information about IABSE and the Conference is available on internet:
http://www.iabse.ethz.ch
Address by the President of IABSE

The Swedish and Danish Groups have taken the initiative to organise this important conference on CableStayed Bridges - past, present and future.
This is an excellent example of a joint arrangement in keeping with the aim of IABSE to develop and exchange
know-how in order to make civil and structural engineering activities contribute to the development of society.
Several modern construction technologies for both the tunnel and bridge part have been introduced and set the
trend for future major links crossing waterways.
The tunnel was specifically dealt with at the IABSE Colloquium on Tunnel Structures in Stockholm in 1998.
The 8 km bridge with a world-record combined rail and motorway cable-stayed span of 490 m will be almost
completed in 1999, an excellent timing for an international conference on cable-stayed bridges to be held in
Malm in June 1999.
This event will be another high-quality event in the endeavour to assemble the structural engineering profession
globally with the purpose of exchanging know-how and ideas regarding trend-setting structural engineering for
the future.
Klaus H. Ostenfeld
President of IABSE
Welcome Address
The modern cable-stayed bridge has been developed during the second half of the 20th century, and is today the
preferred bridge type for main spans in the range from 200 m to 500 m (and in some cases beyond).
The combined bridge and tunnel project of the resund link for dual mode transport of high speed railway and
motorway is a vital element in the formation of a Northern European financial and commercial centre, the
gateway to Scandinavia and the Scandinavian peninsula. The project is an excellent example where
Scandinavian bridge and tunnel engineering with international contribution is cooperating, resulting in a high
quality modern structural engineering product as a symbol of this new activity for the 21st century.
Most cable-stayed bridges are built to carry roads across rivers and straits, but in a few cases also railways are
crossing over the bridges. Among the cable-stayed bridges carrying both road and railway traffic, the resund
Bridge stands out as the biggest and most heavily loaded bridge of this type. It seems, therefore, to be a good
opportunity to link the completion of this bridge to an international conference covering a wide variety of topics
related to the static and dynamic behaviour of cable-stayed bridges.
For the first time in the history of IABSE two National Groups jointly arrange an international conference. As
chairmen of the Danish and Swedish Groups we cordially invite all engineers interested in cable-stayed bridges
to come to Malm, Sweden in early June 1999.
Niels J. Gimsing
Chairman of the Danish
Group of IABSE
Chairman of the Scientific Committee
Hans Ingvarsson
Chairman of the Swedish
Group of IABSE
Chairman of the Organising Committee
TECHNICAL PROGRAMME
KEYNOTE LECTURES
Wednesday 2 June, 11.20-12.00
Niels J.Gimsing, Denmark, History of Cable-Stayed Bridges
Haifan Xiang, China, Retrospect & Prospect of Cable-Stayed Bridges in China
SESSION 1 - Design and Construction

Wednesday 2 June, 13.30-15.30, 16.00-18.00
1:A Chairman: Aarne Jutila, Finland
Co-chairman: Ingvar Olofsson, Sweden
1:B Chairman: Helge Nilsson, Sweden
Co-chairman: Erik Stoltzner, Denmark
Plenary session
Loizias M.P. Concrete Cable-stayed Bridges in the USA
Chandra, V. & Hsu, R. The Innovative William Natcher Cable-Stayed Bridge
Nagai, M., Xie. X., Yamaguchi, H. & Fujino, Y. Identification of Minimum Width-to-span Ratio of Long-span
Cable Stayed Bridges Based on Lateral Torsional Buckling and Flutter Analyses
Pircher H., Bokan H., Bruer A. Computer Based Optimising of the Tensioning of Cable-Stayed Bridges
Astiz, M.A., Fernndez Troyano, L., Manterola, J. Evolution of Design Trends in Cable-Stayed Bridges
Miyazki M. Aerodynamic and Structural Dynamic Control System of Cable-Stayed Bridges for Wind Induced
Vibration
Hague S.T. Seismic Design for the Cape Girardeau Cable-Stayed Bridge
Reis A.J., Pereira A.P., Sousa D.P. & Pedr J.O. Cable-Stayed Bridges for Urban Spaces
Chen D. A New Method to Assign Initial Cable Forces for Prestressed.Concrete Cable-Stayed Bridges
T. Vejrum & Petersen, A. Bridges with Spatial Cable Systems - Theoretical and Experimental Studies
Christoffersen J., Hauge L., Bjerrum J., Jensen H. E. Design and Construction of a CFRP Cable-Stayed Footbridge
Hansvold C., Faller P., Nilsson H. & Svahn P-O Erection of the Uddevalla Bridge
Brstrup M.W. Cable Stayed GFRP Footbridge across Railway Line
Bergman D.W. Ting Kau Cable Stayed Bridge: Challenges in the Construction Process
Poster Presentations
T. Sugiyama Seismic Response of Partially Earth-anchored Cable-Stayed Bridge
V. Chandra, Ricci A., Menn C. & McCabe R. Charles River Crossing; A Gateway to Boston
Firth I. The Design and Construction of the Lockmeadow Footbridge, Maidstone
Cruz J.S. & Almeida, J. F. A New Model for Cable-Stayed Bridges Control and Adjustment
Auperin M. & Dumoulin, C. Cable Finite Element of High Accuracy
Baumann, K. & Dniker J. Sunniberg Bridge, Klosters, Switzerland
Maeda K., Nakamura H., Konno M., Moroyama Y., Abe M. Structural Countermeasures for Design of a Very
Long-Span Cable-Stayed Bridge under Wind Loads
Larsen S. V. Aerodynamic Performance of Cable-Supported Bridges with Large Span-to-Width Ratios
Sharpe A., Yeoward A.J., & Buckby R. J. Cable Stayed Bridge in Bandung, Indonesia
Fan L.C., Chen D.W., Tham L.G., Au F.T.K. & Lee P.K.K. New Developments of Erection Control for Prestressed
Concrete Cable-Stayed Bridges
Trenkler F., Skrikerud P & Voll D.M. The Lifting, Transport and Placing of the resund Pylon Caissons
Cremer J.M. The Val-Benoit Cable-Stayed Bridge
Han D. & Yan Q. Construction Control Practice for Panyu Cable-Stayed Bridge
Wachalski K., Kaminski J. & Sudak M Some aspects of the design of Martwa Wisla River Bridge in Gdansk
Larose G.L. & Wagner Smitt L. Rain/Wind Induced Vibrations of Parallel Stay Cables of the resund High Bridge
Pulkkinen P. Swietokrzyski Bridge, Warsaw
Sham R. & Monster A. The Design of the Zwolle Cable-Stayed Bridge - Integrating Engineering with Aesthetics
Manabe Y., Hirahara N., Mukasa N. & Yabuno M. Accuracy Control on the Construction of the Tatara Bridge
KEYNOTE LECTURE
Thursday 3 June, 08.30-08.50
Michel Virlogeux, France, Bridges with Multiple Cable-Stayed Spans
SESSION 2 Composite Structures

Thursday 3 June, 08.5009.45
Chairman: William C. Brown, UK
Co-chairman: Henrik Christensen, Sweden
Plenary Session
Svensson, H.S. The Development of Composite Cable-Stayed Bridges
Byers D.D., Hague S.T., McCabe S.L. & Rogowski D.M. Comparison of Slab Participation: Assumed for Design vs.
FEA
Veje E., Mller Nielsen P., Pedersen F. & Fuglsang K. Yamuna Cable Stayed Bridge at Allahabad/Naini, India
de Boer A. & Waarts P.H. Probabilistic FE analysis of a cable stayed composite bridge
Xia G. A & Kindmann R. A Method for the Creep Analysis of Composite Cable-Stayed Bridges
Christensen H., Madsen K. & Petersen C.R. Composite Structures in the resund Bridge
SESSION RESUND
Thursday 3 June, 10.1512.30
Chairman: Niels J. Gimsing, Denmark
Co-chairman: Karl-Otto Sicking, Sweden
Plenary Session
Lundhus P. Build a Link Goals, Principles, Strategies and Results
Falbe-Hansen K. & Larsson . The resund Bridge: Project Development From Competition to Construction
Nissen J. & Rotne G. Getting the Balance Right. The resund Bridge - Design Concept
Gimsing J. The resund Bridge: The Tender Project
Svensson E. From Eurocodes, Special Investigations and Risk Analysis To Design Requirements for the resund
Coast to Coast Structures
Hauge L. & Petersen A. Detailed Design of the Cable Stayed Bridge for the resund Link
Olofsson I. Design Coordination of a Design-build Project
Srensen, L.Th. & Thorsen N.E. The resund Bridge, Erection of the Cable-Stayed Main Span
KEYNOTE LECTURES
Friday 4 June, 08.3009.15
Jrg Schlaich, Germany, Cable-Stayed Bridges with Special Features
Manabu Ito, Japan, Stay Cable Technology Overview
SESSION 3 Cable-Stayed Bridges for Railways

Chairman: Manabu Ito, Japan
Co-chairman: Ole Damgaard-Larsen, Denmark
Plenary Session
Bitsch N. & Hauge L. Design of Girder and Cables for Train Load
Sham R. An Innovative Technique for Fitting Trackwork Alignments Through the Railway Envelope of a Cable-Stayed
Bridge
Gimsing J. & Thomsen A. Comfort Criteria for High Speed Trains on The resund Bridge
Karoumi R. Nonlinear Dynamic Analysis of Cable-Stayed Bridges Excited by Moving Vehicles
Bruno D. Grimaldi A. & Leonardi A. Deformability of Long-Span Cable-Stayed Bridges for Railways
SESSION 4 Stay Cable Technology

Chairman: Manabu Ito, Japan
Co-chairman: Ole Damgaard Larsen, Denmark
Plenary Session
Dumoulin, C. Active Tendon Actuators for Cable-Stayed Bridge
Marchetti M. & Lecinq B. Stay Adjustment: From Design Perspective to On Site Practice
Suzuki Y., Hiyama Y., Kondo T., Kawakami T, Suzuki M., Moriuchi A., Damping Device in Stay Cables of Meiko
Central Bridge
El Kady H.M., Arockiasamy M., Samaan S., Bahie-Eldeen Y., Bakhoum M.M. & El Gammal, M.A. Damping
Characteristics of Carbon Fiber Composite Cables for Application in Cable-Stayed Bridges
Bournand Y. Development of New Stay Cable Dampers
Gonzlez J.L. & Sobrino J. A. Fatigue Reliability Evaluation of Cables in Cable-Stayed Bridges.
Case Study: The Sama de Langreo Bridge
McGuire G.J. PTI Cable Stay Recommendations
Mizoe M., Muroi S., Horii T., Isobe T., Kiyota R. & Imada Y., The Super High Damping Rubber Damper on the
Stay-cables of Meiko East Bridge.
Hemmert-Halswick A. & Sczyslo S. Corrosion Protection of Locked Coil Ropes at Road Bridges
Magonette G., Renda V., Bournand Y., Hansvold C., Jenner, A.G. & Fsterling H. Experimental Analysis of a
Large-Scale Cable-Stayed Mock-up
Stubler J., Domage J.B. & Ladret P. Vibration Control of Stay Cables
Preumont A., Bossens F., Helduser S. Bonnefeld R. & Frsterling H. Active Tendon Control of Cable-Stayed Bridges:
Control Strategy and Actuator Design
Roos, F., Noisternig J.F. CFRP-Tendons -Development and Testing
Bojan J. Bevc L. & Sonda D. Laboratory Tests of the Anchorage Plates for the Cables
Seo-Kyung C. & Seung Wook J. Erection of Composite Deck for Seohae Bridge
SESSION 5 Observation, Maintenance and Repair, followed by Closing

Session
Chairman: Jrn Lauridsen, Denmark
Co-chairman: Hans Ingvarsson, Sweden
Plenary Session
Popa V. & Stanciu M. Bridge Consolidation by Using Cable-Stayed Method
Yamagiwa I, Utsuno H., Endo K. & Sugii K. Application of the Identification of Tension and Flexural Rigidity at
Once to the Bridge Cables
Gentile C. & Martinez F. Dynamic Characteristics of Two Newly Constructed Curved Cable-Stayed Bridges
Suzuki Y., Mizuguchi K., Sakuma S., Maekawa T., Ueda T. & Kobayashi Y. Field Observation on Aerodynamic
Response of Meiko West Bridge
Reinholdt P., Veje E. & Kalvslund J. Rehabilitation of the Luangwa Bridge
Laigaard J. & Pedersen L. Design of Structural Monitoring Systems
Bloomstine M.L. & Stoltzner E. The Faroe Cable-Stayed Bridge -Maintenance Experience with Major Components
Andersen H. & Hommel D.L. & Veje E.M. Emergency Rehabilitation of the Zrate-Brazo Largo Bridges, Argentina
Yamaguchi K. Manabe Y., Sasaki N. & Morishita K. Field Observation and Vibration Test of the Tatara Bridge
Gomez R., Muria-Vila D. Sanchez-Ramirez R. & Escobar J. A. Second Monitoring and Surveillance of the Response
of a Cable-Stayed Bridge
Cunha ., Caetano E., Calada R. & Delgado R. Dynamic Tests on Vasco da Gama Cable-Stayed Bridge
Fuzier J.P., Stubler J. & Grattepanche D. The resund Stay Cables: Design for Fatigue Resistance and Easy
Maintenance
History of cable-stayed bridges

Niels J GIMSING
Professor
BKM, DTU
DK-2800 Lyngby, Denmark
Niels J Gimsing, born 1935, is professor

at the Technical University of Denmark
since 1976. He has at several occasions
acted as specialist consultant during the
design of major bridges.
Introduction
The principle of supporting a bridge deck by inclined tension members leading to towers on either
side of the span has been known for centuries but it did not become an interesting option until the
beginning of the 19th century when wrought iron bars, and later steel wires, with a reliable tensile
strength were developed. A limited number of bridges based on the stayed girder system were built
and more proposed but the system was never generally accepted at that time.
In 1823 the famous French engineer and scientist C.L.
Navier published the results of a study on bridges with
the deck stiffened by wrought iron chains and with a
geometry as shown on the original drawing in Fig.1.
It is interesting to note that Navier considered both a fan
shaped and a harp shaped system in configurations that
today would be denoted multi-cable systems. So the
cable systems were actually up-to-date, but in contrast
to the present practice the backstays were assumed to be
earth anchored, as seen in the lower half of Fig.1.
Naviers final conclusion was that the suspension
system should be used instead of the stayed system [1].
This conclusion was to a large extent based on
observations of stayed bridges that had failed.
Fig.1 Bridge systems investigated by

Navier in the 1820s.
In the early stayed bridges it proved very difficult to

arrive at an even distribution of the load between all
stays. Thus imperfections during fabrication and
erection could easily lead to a structure where some
stays were slack and others overstressed. The stays were
generally attached to the girder and pylon by pinned
connections that did not allow a controlled tensioning.
The problems encountered and the recommendation by Navier resulted in a very limited number of
stayed girder bridges being built up to the 1950s, whereas systems where the suspension system was
combined with the stayed system was used in many major bridges built in the second half of the
19th century.
As an example, Fig.2 shows the Albert Bridge across the Thames in London. In this bridge from
1873 both the parabolic top cable and the stays were made of eye bar chains. The Albert Bridge
still exists so the system has certainly proved its durability.
Fig.2 The Albert Bridge across the Thames in London.

The combination of the suspension and the stayed system was also applied in a number of bridges
built in France in the 1880s, but the most notable bridges of this type were designed by John A.
Roebling and built in the United States among these the longest cable supported bridge of the 19th
century: the Brooklyn Bridge [2].
Introduction of the self anchored cable-stayed bridge system
Around the turn of the century the French engineer A.V. Gisclard developed an earth anchored,
stayed system in which not only the inclined stays but also the tension members at the deck level
were made of cables. In the 1920s the system by Gisclard was developed further by substituting the
horizontal cables by the deck girders and changing the earth anchored system to a self anchored
system with compression rather than tension along the deck for example used in the Lezardrieux
Bridge from 1925. So in reality the system of the modern, self anchored cable-stayed bridges was
developed at that time.
The combined suspension and stayed system used
extensively at the end of the 19th century was
abandoned from the beginning of the 20th century
and substituted by pure suspension systems.
However, in 1938 Dischinger proposed a system
Fig.3 Dischingers proposal for a bridge
in which the central part of the span was carried
between Kln and Mhlheim.
by a suspension system whereas the outer parts
were carried by stays radiating from the pylon top. This system was proposed for a cable supported
bridge with a 750 m main span to be built across the Elbe River in Hamburg.
In connection with the reconstruction of German bridges after the war, the Dischinger system was
proposed at several occasions (Fig.3) but it was never used for actual construction. One of the
reasons is undoubtedly the pronounced discontinuity of the system both with respect to the
structural behavior and to the appearance. The discontinuity reflects Dischingers discontent at the
original Roebling system with its much more continuous configuration achieved by overlapping the
multi-cable stayed system and the suspension system. In the publication of his own system,
Dischinger categorically stated that the stays of Roebling's bridges had proved to be completely
inefficient!
Although never adopted for actual construction, the proposals by Dischinger undoubtedly had a
considerable influence on the subsequent introduction of the pure cable-stayed bridge. Thus, the
Strmsund Bridge, which is generally regarded as the first modern cable-stayed bridge was designed
by Dischinger. The bridge was of the three-span type, a system commonly used for suspension
bridges, and it had a main span of 182.6 m flanked by two side spans of 74.7 m (Fig.4). The stays
were arranged according to the pure fan system with two pairs of stays radiating from each pylon
top. The steel pylons were of the portal type supporting the two vertical cable systems arranged on
either side of the bridge deck. The deck girder contained two plate girders positioned outside the
cable planes to allow an "invisible" anchoring of the stays inside the plate girders.
Fig.4 The Strmsund Bridge.
The start of a new era for cablestayed bridges was to a large

extent due to the improved
technique of structural analysis
allowing calculation of cable
forces throughout the erection
period and thereby assuring the
efficiency of all cables in the final
structure as well as a favorable
distribution of dead load moments
in the deck. Probably, such
calculations were for the first time
made for the erection of the
Strmsund Bridge.
Regarded as a plane system, the Strmsund Bridge is statically indeterminate to the eighth degree,
but by dividing the load into a symmetrical and an antisymmetrical part, the number of redundants
could be reduced to four. This was well within acceptable limits for the numerical work that could
be performed with the slide rule and the mechanical calculators available at the beginning of the
1950s.
The German era
After the Strmsund Bridge the next true cable-stayed bridge to be erected was the Theodor Heuss
Bridge across the Rhine at Dsseldorf - opened to traffic in 1957 (Fig. 5). With a main span of 260
m and side spans of 108 m it was considerably larger than the Strmsund Bridge. Also, the Theodor
Heuss Bridge was more innovative by introducing the harp shaped cable system with parallel stays
and a pylon composed of two free-standing posts fixed to the bridge deck structure. The harp
configuration was chosen primarily for aesthetic reasons giving a more pleasant appearance of the
two cable systems when viewed from a skew angle.
Fig.5 The Theodor Heuss Bridge.

The Theodor Heuss Bridge gave a very clear indication of the cable-stayed bridges' potentials
initiating an impressive development of cable-stayed bridges first in Germany and later throughout
the world in the decades to follow.
The second cable-stayed bridge to be erected in Germany was the Severins Bridge in Kln (Fig.6).
This bridge featured the first application of an A-shaped pylon combined with transversally inclined
cable planes, and it was the first to be constructed as an asymmetrical two span bridge with a single
pylon positioned at only one of the river banks. The cable system of the Severins Bridge was of the
efficient fan shaped type, which is in good harmony with the A-shaped pylon. The cross section of
the deck girder was essentially the same as used in the Theodor Heuss Bridge with two box girders
connected by the orthotropic steel deck. Because of the substantial compression in the girder due to
the one-sided arrangement of the pylon, the application of a steel deck was particularly
advantageous in the Severins Bridge, as axial compression could be distributed over a large crosssectional area. At both ends of the cable-stayed portion, the deck girder was made continuous into
the adjacent box girder spans.
Fig.6 The Severins Bridge in Kln.
Although one of the very first cable-stayed bridges, the Severins Bridge still stands as a most
successful bridge. The design of the pylon with its pronounced dimensions and the way the deck
girder "floats" through the pylon constitute fine solutions to the design problems faced.
The third German cable-stayed bridge, the Norderelbe Bridge at Hamburg, introduced the central
cable plane with pylons and stay cables positioned in the central reserve of the motorway - a system
that in the following years became the preferred system for the majority of cable-stayed bridges to
be constructed in Germany - as well as in several other countries.
In some of its other design features
the Norderelbe Bridge was more
unusual, e.g. with pylons twice as
high as required for structural
reasons and with a cable system
looking as if the main task was to
support the pylon and not the deck
girder (Fig.7).
In the mid 1980s the Norderelbe
Bridge had to go through a major
rehabilitation program and as part
of
this the cable system was
Fig.7 The original Norderelbe Bridge.
modified to a more sensible
configuration. So today the Norderelbe Bridge is less peculiar in its appearance.
After the Norderelbe Bridge came the Leverkusen Bridge (opened in 1964) across the Rhine. This
bridge had the same centrally arranged cable plane, but here the cable system was of the harp
configuration with two sets of stays connected to each pylon. Each stay comprised two individual
cables composed of seven locked-coil strands.
The multi-cable system
In the early cable-stayed bridges built at the end of the 1950s and the beginning of the 1960s each
stay cables was generally composed of several prefabricated strands to achieve the large cross
sections required in these bridges with their limited number of cables. However, the multi-strand
arrangement of the individual stay
gave a number of drawbacks such
as complicated anchorage details
in the girder and difficulties in
replacement of strands. These
drawbacks could be eliminated if
the number of stays was increased
so that each stay cable could be
made of a single strand and this
led to the introduction of the
multi-cable system.
The first two multi-cable bridges
to be built were the Friedrich Ebert
Fig.8 The Rees Bridge.
Bridge and the Rees Bridge both designed by H. Homberg and built across the Rhine. The Friedrich
Ebert Bridge contains a central cable plane with two pylons, each supporting 220 stays with
diameters ranging from 91 to 123 mm, depending on the position of the actual stay. In the Rees
Bridge two cable planes each containing a harp-shaped multi-cable system with 210 stays were
used (Fig.8).
Multi-cable systems lead to a more continuous support of the deck girder, and at the same time the
cable forces to be transmitted at each anchor point are reduced, so that a local strengthening of the
girder at the anchorages can be avoided. During erection advantages are to be found due to the much
shorter deck cantilevers required to reach from one anchor point to the next, and in the final
structure the smaller stay units will ease a replacement. These advantages would subsequently result
in a general acceptance of the multi-cable system in almost all cable-stayed bridges. However, in
that process it should later be realized that the multi-cable system also presented some
disadvantages such as a higher vulnerability to excitations and increased total wind load on the
cable system.
Fig.9 The Knie Bridge in Dsseldorf.
In 1969 a notable cable-stayed

bridge, the Knie Bridge, was
opened to traffic in Dsseldorf
(Fig.9). In this bridge the cable
system was of the harp
configuration with relatively few
parallel stays, but in contrast to
earlier bridges with the harp
system, intermediate supports were
added under every cable anchor
point in the side span. This
increased the efficiency of the harp
system to such an extent that it was
possible to use a very slender deck
girder with an open cross section,
i.e. with insignificant torsional
stiffness.
In the Knie Bridge an asymmetrical layout similar to that of the Severins Bridge was used with the
pylon placed on one of the river banks only. Despite the considerable height of the pylon (114 m) it
was possible to compose it of two free-standing posts without any struts or bracing to stabilize
laterally.
First parallel-wire strands

In 1972 the completion of the Mannheim-Ludwigshafen Bridge across the Rhine marked the first
application of a parallel-wire strand in a major cable-stayed bridge. Each strand (with 295
ungalvanized wires of 7 mm diameter) was anchored by a new type of socket called a HiAm socket
with increased fatigue resistance due to the application of a cold filling material containing epoxy
compound. Furthermore, the Mannheim-Ludwigshafen Bridge introduced an interesting
combination of materials, with the deck girder made entirely of steel in the main span and entirely
of concrete in the side span (Fig.10). This combination was very well justified, as the side span
(through the application of an intermediate pier) had a maximum free span of 65 m, whereas the
main span had a free length of 287
m. Actually, the higher dead load
of the side span proved directly
advantageous as it reduced the
requirement for a vertical
anchoring of the girder.
The combination of concrete
girders with intermediate supports
in the side span and steel girders in
the main span was subsequently
used in several notable cablestayed bridges constructed in the
1980s and 1990s.
The cable-stayed Khlbrand Bridge
in the port of Hamburg exhibits the
first application of the multi-cable
system in a bridge with double cable planes supported by A-shaped pylons (Fig.11). The modified
fan system was one of high efficiency which gave advantages not only in the design of the final
structures but also during erection as no temporary supports or temporary stays were required.
Fig.10 The Mannheim-Ludwigshafen Bridge under
construction.
From the same period is another

remarkable German cable-stayed
bridge: the Dsseldorf-Flehe
Bridge across the Rhine. Despite a
main span length of 367 m it was
chosen to build a two-span cablestayed structure with only one
pylon on one of the river banks.
This necessitated a pylon with a
height of 150 m above ground. In
Fig.11 The Khlbrand Bridge.
contrast to the general German
practice the pylon was made of
concrete, and its lambda () configuration was chosen to give support to the central cable plane with
a harp shaped cable system in the side span and a modified harp in the main span. In appearance the
pylon of the Flehe Bridge is not very harmonic, especially when compared to other, more recent shaped pylons.
For a period of almost twenty years the evolution of cable-stayed bridges was to a very large extent
taking place in Germany but in the following years the activities shifted to other locations on the
globe.
The evolution outside Germany
During the late 1950s and the 1960s a relatively modest number of cable-stayed bridges were built
outside Germany and most of these bridges were based entirely on the German design philosophy.
In the UK the Wye Bridge on the Welsh approach to the Severn Suspension Bridge had been
completed in 1965 and this bridge was quite unique by having only one set of stays leading from the
pylons to the deck. Based on a similar design concept the Erskine Bridge in Scotland (Fig.12)
followed in 1971. Despite its main span of considerable length it also had only one stay leading
from each of the two pylons to the deck girder in the 305 m long main span so the girder had to span
more than 100 m without support from the cable system. Despite this fact, the deck girder was
designed with a depth of only 3.05 m, which is of the same magnitude as found in cable-stayed
bridges with several stays supporting the girder at much smaller intervals. As the stay had to be
made with a very large cross-sectional area it was composed of 24 helical strands each 76 mm in
diameter.
During erection of the
system with only one
permanent stay from each
pylon it was necessary to use
two temporary stays to
reduce the moment in the
deck girder when
cantilevering from the pylon
to the adjacent cable anchor
point in the main span.
In France the completion in
1975 of the Saint Nazaire
Bridge across the Loire River marked a step further for the cable-stayed bridges as it was the first
bridge of this type to span more than 400 m. The pylons consist of an upper A-shaped part of steel
and a lower pier shaft of concrete. The cable system is of the multi-cable fan type with each stay
made of a single locked-coil strand.
Fig.12 The Erskine Bridge.
The first major cable-stayed bridge with an earth anchored cable system, the Indiano Bridge across
the Arno near Firenze (Fig.13), had a 206 m long main span supported by two fans radiating from
the tops of 45 m high pylons leaning slightly backwards. From the pylons, earth anchored back stays
Fig.13 The Indiano Bridge across the Arno at Firenze.

continue to anchor blocks transmitting both the vertical and the horizontal component of the cable
force to the soil.
The special problems related to the construction of cable-stayed bridges with earth anchored cable
systems were overcome in the Arno Bridge by erecting the deck girder on temporary piers before
adding the pylons and the cable system.
Cable-stayed concrete bridges
In the first two decades after the completion of the Strmsund Bridge the evolution of cable-stayed
bridges was to a very large extent dominated by steel bridges with orthotropic decks together with
plate or box girders and cellular pylons.
However, as a remarkable exception from this a cable-stayed bridge of unusual proportions (and
based on a very different design philosophy) had been completed already in 1962: The Maracaibo
Bridge in Venezuela, designed by Riccardo Morandi (Fig.14). Here both the pylons and the deck
girder were made of concrete, thereby introducing a structural material that had not earlier been
used in the main elements of cable supported bridge superstructures. Furthermore, it was the first
multi-span cable-stayed bridge.
Fig.14 The Maracaibo Bridge.

To allow one-way traffic of ships in and out of Lake Maracaibo, it was chosen to build a bridge with
five 235 m long main spans. Each of these spans comprises a double cantilever supported by only
one pair of stays radiating from a triangular pylon structure designed to stabilize the system for
asymmetrical loads. Between the ends of the cantilevers small suspended spans are arranged, so that
the system regarded as a plane system is externally determinate. The application of only one set of
stays necessitated a heavy box girder to span from the pylon to the cable supported point, and during
construction a large truss was required to support the formwork.
The Maracaibo Bridge was later followed by two other major cable-stayed bridges designed by
Morandi, the Polcevara Viaduct in Genova and the Wadi Kuf Bridge in Libya.
However, all of the designs of Morandi were of such a personal style that they did not to any large
extent serve as models for the cable-stayed bridges of concrete to come.
A pioneer among the type of concrete cable-stayed bridge to become more fashionable was the
Donaukanal Bridge in Vienna (Fig.15) with a main span of 119 m. The deck contains a concrete box
girder and the stays are composed of parallel mono strands. The Donaukanal Bridge has a very
pleasing appearance and harmonic proportions, and the construction procedure was quite unique as
the bridge was cast in two halves on either side of the canal and subsequently turned into position
after installation and tensioning of the stay cables.
The application of a multi-cable
system in a cable-stayed concrete
bridge was first seen in the
Brotonne Bridge across the Seine.
Here a central cable plane was
combined with a box-shaped deck
girder, made partially of
prefabricated elements. The stays
were made of parallel seven-wire
strands of a type used for tendons
in post-tensioned concrete.
Corrosion protection was achieved
Fig.15 The Donaukanal Bridge in Vienna.
by inserting the parallel strands in
stainless steel tubes, subsequently filled with cement grout. The anchoring of the seven-wire strands
was initially made by ordinary wedge anchors, but to increase the fatigue strength, especially for
pulsating loads, a supplementary anchoring was established by adding epoxy mortar inside a steel
tube extending from the wedge anchorages.
Another example on the use of the
multi-cable system in a cablestayed concrete bridge can be
found in the Pasco-Kennewick
(Fig.16). Here, the double cable
systems in the fan configuration
assure an efficient support of the
deck both vertically and
torsionally. The stays, each made
of a single parallel-wire strand, are
inside a grouted polyethylene tube
and with HiAm anchors. The deck
girder was erected by the
segmental method using heavy
prefabricated elements having the
full width of the roadway.
Fig.16 The Pasco-Kennewick Bridge.
The twin bridges across the Parana

River in Argentina (Fig.17)
from1978, were in many ways based on the same design philosophy as used for the PascoKennewick design. However, the deck girders of the Parana Bridges were made of steel. They were
the first cable-stayed bridges to transfer heavy railway loading. This gave special design problems
which to a certain extent were accentuated by a one-sided position of the single track subjecting the
two vertical cable systems to traffic loads of different intensity. For this reason it was necessary to
use different dimensions for the stays in the two sides, the heavier cables being required for the
railway side.
After less than 20 years of service one of the stay cables in the Parana Bridges broke without
warning and as a result a major repair work had to be initiated at the end of the 1990s.
The superiority of cable supported
bridges in crossing navigable
waters was clearly demonstrated in
the early 1980s when a new cablestayed Tjrn Bridge was built to
replace the original arch bridge
after it had been hit by a
misnavigated ship. The new bridge
was built with a span of 366 m, 86
m more than the span of the arch
bridge, and this allowed both
pylons to be located on land 25 m
from the coastline.
Fig.17 The Parana Bridge.
The Tjrn Bridge belongs to the

group of cable-stayed bridges with
different structural materials in the
side spans and the main span
(Fig.18). The side spans are
designed as continuous concrete
girders with intermediate column
supports at each cable anchor point
whereas the main span is made as
a steel box with orthotropic steel
deck overhangs.
During the 1980s the activity

within the field of cable-stayed
bridges was considerably reduced
Fig.18 The Tjrn Bridge.
in Europe compared to the
previous decades, and most of the
bridges built did not deviate much in size or design features from those already constructed. There
were, however, a few exceptions from this rule.
In 1984 the completion of the Barrios de Luna Bridge in Spain gave a further indication of the
competitiveness of concrete as structural material not only for the pylons but also in the girder of
cable-stayed bridges (Fig.19). With a main span of 440 m the Barrios de Luna Bridge surpassed the
span of the Saint Nazaire Bridge by a margin of almost 10% and became for a couple of years the
record-holder amongst cable-stayed bridges.
The Far Bridge in Denmark was opened in 1985 and it comprised a 290 m long main span
supported by a central cable plane. The girder had originally been designed by the owner as a
concrete box but an alternative bid based on a steel box proved to be competitive and was chosen
for construction. The concrete pylons form a further development of the diamond-shaped pylons
originally introduced in the Khlbrand Bridge. Thus, in the Far Bridge the lower triangle is
extended all the way down to the water surface (Fig.20) rather than being supported on high pier
shafts. Furthermore, the Far Bridge showed the first application of corrosion protection of the box
girder interior by dehumidification of the air.
Fig.19 The Barrios de Luna Bridge.
Fig.20 The Far Bridge.
Within cable-stayed bridges both the type with a central cable plane above the median reserve and
the type with two cable planes outside the roadway area had been extensively applied in the first
three decades of the modern evolution. To some extent the choice between the two options seemed
to depend on the designer's preference rather than on a rational, unbiased comparison between
advantages and drawbacks.
H. Homberg had clearly preferred the central cable plane concept wherever it was applicable, i.e.
where the road to be carried had a median reserve. It is therefore not surprising that Homberg's
largest cable-stayed bridge, the Rama IX Bridge in Bangkok was designed with a central cable
plane, despite the span of 450 m. The cable system is of the multi-cable, modified fan configuration
and all stays are made of single locked-coil strands, among these the largest diameter locked-coil
strand fabricated so far with a diameter of 174 mm. The deck girder in the main span is a quasi
trapezoidal, five cell box with the full width of the bridge deck (32.5 m) and with a depth of only 4
m.
The American experience
A pioneer among cable-stayed bridges in North America was completed in Montreal already in
1969: the Papineau Bridge (Fig.21) with a main span of 241 m. In several of its design features this
bridge could resemble the Leverkusen Bridge and other German bridges with a central cable plane
and a deep, but relatively narrow, box girder under the wide orthotropic bridge deck. The cable
system was of the fan type with only two sets of stays radiating from each pylon top. Each stay cable
was composed of several helical bridge strands of galvanized wires, and as a novelty each strand
was covered by a hot extruded
polyethylene coating with a
minimum cover of 5 mm - a
protective system that should later
be used extensively.
Fig.21 The Papineau Bridge in Montreal.
Apart from the Papineau Bridge

and a limited number of other
bridges the activity within
construction of cable-stayed
bridges had been very low in North
America during the 1960s and the
early 1970s, but from then on the
situation changed dramatically.
In Florida a ship collision accident had given a clear indication of the inadequacy of the navigation
opening in the 250 m long main span of the Sunshine Skyway. It was, therefore, decided to replace
the existing two parallel bridges by a single bridge having a 360 m long cable-stayed main span.
Two designs were prepared for the bridge, one based on a composite deck and two cable planes
along the edges of the bridge deck, and the other as a pure concrete box and a single central cable
plane. Both designs were put out
for tender and the result showed a
very close race between the two
options.
The final choice was to construct
the concrete bridge according to a
design based on the principles
initially introduced during design
and construction of the Brotonne
Bridge in France. With its main
span of 366 m the Sunshine
Skyway was at its completion in
1986 the longest cable-stayed
bridge in the USA (Fig 22).
The composite girder alternative
for the Sunshine Skyway was
based on a system with two longitudinal plate girders directly under the cable planes and a large
number of transverse girders to give support to the deck slab of reinforced concrete. In its main
features this concept was subsequently applied in another North American bridge, the Alex Fraser
Bridge (Annacis Island Bridge) at Vancouver in Canada. With its main span of 465 m the Alex
Fraser Bridge (Fig.23) became the record-holder among cable-stayed bridges for a period of five
years.
Fig.22 The Sunshine Skyway Bridge.
The potentials of the composite girder concept was clearly demonstrated during the construction of
the Alex Fraser Bridge. Thus, the cantilevering from one cable anchor point to the next was easily
accomplished by the relatively light steel girders, allowing the stay cables to be added before the
heavy concrete deck was erected using precast slabs. At the same time the concrete slab could be
efficiently utilized to transfer the axial compression induced into the girder by the horizontal
components of the stay cable forces.
The advantages of applying composite girders in cablestayed bridges should in the years to follow the
construction of the Alex Fraser Bridge lead to a situation
where this system was gradually being preferred for the
majority of cable-stayed bridges in North America.
In the USA the general trend throughout the 1980s was to
simplify the design of especially the girders in cablestayed bridges. Within concrete bridges a good example
on this trend is the Dames Point Bridge at Jacksonville in
Florida. With a main span of 396 m the bridge surpassed
the Sunshine Skyway as the longest concrete cable-stayed
bridge in North America.
Fig.23 The Alex Fraser Bridge.
The cable system of the Dames Point bridge is a multicable harp system supported by concrete pylons with a
considerable flexural stiffness in the longitudinal
direction. This gave the cable system very good
deformational characteristics so that the girder could be
made with a depth of only 1.5 m corresponding to 1/260
of the main span length.
In principle the structural system of the girder in the Dames Point Bridge corresponds to that of the
Alex Fraser Bridge, i.e. with two longitudinal girders beneath the cable planes and numerous
transverse girders. However, in the Dames Point Bridge the longitudinal girders are made as solid
concrete ribs with a depth of 1.5 m and a width of 2.5 m allowing a most efficient anchoring of the
stay cables.
Seen in comparison with the
Pasco-Kennewick Bridge - the
first major concrete cable-stayed
bridge in North America - the
Dames Point Bridge clearly
shows the simplifications in
girder design.
The Japanese development
In Japan the cable-stayed bridges
were introduced already in the
late 1950s but the first bridges of
Fig.24 The Rokko Bridge in Kobe.
this type were not characterized by special design features so they had little influence on the further
developments. However, in 1977 the Rokko Bridge, the very first double deck cable-stayed bridge,
was completed in Japan (Fig.24). The deck is made as a truss with a depth of approx.8 m to give
ample headroom, daylight, and fresh air on the lower deck. The cable system is of the multi-cable
type with each stay composed of two parallel-wire, mono-strand cables.
In a much larger scale the double deck concept was later used for the twin cable-stayed bridges, the
Hitsuishijima and the Iwagurojima Bridges (Fig.25), that form a part of the Seto Ohashi between
Honshu and Shikoku. Each of the two neighbor bridges has spans of 185 m - 420 m - 185 m. The
traffic is running on a two level truss with a four-lane expressway on the upper deck and a double
Fig.25 The Hitsuishijima and Iwagurojima Bridges of the Seto Ohashi.

track railway (with provisions for a later addition of two more tracks) on the lower deck. The cable
systems are of the modified fan configuration with two vertical cable planes positioned directly
above the deck trusses. Thus, a high efficiency of the cable supporting for both vertical and torsional
loading is achieved.
Fig.26 The Meiko Nishi Bridge in Nagoya.
An elegant cable-stayed bridge was completed in Japan in 1985 across the port of Nagoya, the
Meiko Nishi Bridge (fig.26). Here the roadway is carried by a semi-streamlined box girder
supported by two inclined cable planes radiating from the top of A-shaped pylons. With the chosen
pylon shape and the fan shaped cable systems, the Meiko Nishi Bridge constitutes a fine example of
a highly efficient cable-stayed bridge.
In Tokyo a tricky design problem
was overcome in the late 1980.s by
constructing the worlds first Scurved cable-stayed bridge (the
Katsuhika Harp Bridge)
comprising a central twisted
cable plane and two pylons of
different height (Fig.27).
Fig.27 The S-shaped Katsuhika Harp Bridge in Tokyo.
The double deck configuration was

again applied in the Yokohama
Bay Bridge opened to traffic in
1989. With its main span of 460 m
the bridge was only 5 m shorter
than the Alex Fraser Bridge in
Canada - at that time the recordholder amongst cable-stayed
bridges. The truss of the
Yokohama Bay Bridge has its top
chord made as a 39 m wide and 3
m deep, streamlined box girder,
whereas the bottom chord and the
diagonals are of more conventional
bluff box sections. The total depth
of the truss is 12 m corresponding
to 1/38 of the main span length.
From the point of view of
appearance the Yokohama Bay
Bridge is quite successful as the
truss is well-proportioned and the
pylons have a clear and simple
geometry (Fig.28). Eventually, the
bridge will carry 12 lanes of
vehicular traffic on two decks but
initially only the upper deck has
been opened to traffic.
Fig.28 The Yokohama Bay Bridge
In Japan the parallel-wire strands

have been used extensively and
new types have been developed to
improve the corrosion protection.
Conclusion
With the description of some cable-stayed bridges completed at the end of the 1980s the historical
review shall be concluded, but to show that the evolution of cable-stayed bridges has continued into
the 1990s Table 1 shows the ten longest cable-stayed spans to be found at the turn of the millenium.
It is seen that all of these bridges have been completed during the 1990s.
Longest cable-stayed bridges in the year 2000
No.
Name
Span
Traffic
Country
Year
Tatara Bridge
890 m
Road
Japan
1999
Normandie Bridge
856 m
Road
France
1995
Qingzhou Minjiang Br.
605 m
Road
China
1998
Yangpu Bridge
602 m
Road
China
1993
5
6
Meiko Chuo Bridge

Xupu Bridge
590 m
590 m
Road
Road
Japan
China
1997
1996
Skarnsund Bridge
530 m
Road
Norway
1991
Tsurumi Fairway Bridge
510 m
Road
Japan
1994
9
10
resund Bridge
Iguchi Bridge
490 m
490 m
Road+rail
Road
Denmark/Sweden
Japan
2000
1991
Table 1. The ten longest cable-stayed bridges at the turn of the millennium
It is interesting to note that seven of the ten longest cable-stayed bridges are located in the Far East
(China and Japan), and that the remaining three bridges on the list are from Europe.
In the four and a half decade passed since the Strmsund Bridge was opened the cable-stayed
bridges have developed to become dominating in the span range from 200 m to 500 m. Under
specific conditions the cable-stayed bridges might even be competitive against suspension bridges
up to spans of more than 1000 m. However, it remains to be seen if in the near future the cablestayed bridges will actually pass the present maximum span length of 890 m in the Tatara Bridge.
References
[1]
Troitsky, M.S., Cable-Stayed Bridges, BSP Professional Books, London 1988
[2]
Gimsing, Niels J., Cable Supported Bridges Concept and Design, Wiley,
Chichester 1997
Retrospect and Prospect of Cable-stayed Bridges in China
Haifan XIANG
Professor
Tongji University
Shanghai, China
Haifan Xiang, born 1935,

finished his postgraduate
study in civil engineering at
Tongji University in 1958
Summary
A brief review of cable-stayed bridge construction activities in the last two decades of 20th
century in China is given, and in particular, some large cable-stayed bridges to be probably built
in the first two decades of the 21st century in the country are also introduced. Most of these large
cable-stayed bridges will be built to cross some sea straits in the state highway network along the
pacific coastal line of China. Recent developments of cable-stayed bridges in China including
hybrid system, single pylon system, stay cable system and wind-resistant studies for very longspan cable-stayed bridges are also mentioned.
1.Introduction and Historical Review

The modern cable-stayed bridge was born in 1950s, while Dischinger from Germany designed
the Stroemsund Bridge in Sweden. The construction of modern cable-stayed bridges in China
initiated in 1972, relatively later compared with other developed countries.
The first modern cable-stayed bridge in China was built in 1975, and its technology was
developed through three stages in the past 30 years. In the first period from 1972-1982, some
concrete cable-stayed bridges were built. The completion of Jinan Bridge over Yellow River in
1982 with a main span of 220m may be regarded as a successful conclusion of this learning
period. In the second period from 1982-1990, 19 cable-stayed bridges were constructed in 12
provinces, The main span-length was raised to 260m for Yonghe bridge, and 288m for Dongying
Bridge, which was the only cable-stayed bridge with a steel deck at that time. In the 3rd period in
90s, many cable-stayed bridges with main spans beyond 400m have been built following the
experiences obtained. The successful construction of Nanpu Bridge in Shanghai with a main span
of 423m was a millstone, which encouraged the provincial bridge engineers to design and
construct long-span cable-stayed bridge in their own provinces.
The major cable-stayed bridges in China with span length beyond 400m are listed in Table 1, in
which those bridges under construction or under designing are also listed. 4 single pylon cablestayed bridge with a main span over 200m are also listed in Table 2.
Up to now, more than 100 cable-stayed bridges have been built in China, so China might be the
country of building more cable-stayed bridges than any other countries in the world.
Bridge Name
Location Main Span Year of
Deck Type
Completion
1
Nanpu Bridge
Shanghai
423 m
1991
composite
2
Yangpu Bridge
Shanghai
602 m
1994
composite
3
Yunxian Bridge over Han River
Hubei
414 m
1994
P. C.
nd
4 2 Wuhan Bridge over Yangtse River
Hubei
400 m
1995
P. C.
5 Tongling Bridge over Yangtse River
Anhui
436 m
1995
P. C.
nd
6 2 Chongqing Bridge over Yangtse Chongqing 444 m
1995
P. C.
River
7
Xupu Bridge
Shanghai
590 m
1996
composite
8
Kap Shui Mun Bridge
Hong Kong 430 m
1997
steel
9
Ting Kau Bridge
Hong Kong 475 m
1998
composite
Guangdong 518 m
10
2nd Santou Bay Bridge
u.c(1999)
mixed
11
2nd Nanjing Bridge over Yangtse

Jiangsu
River
rd
3 Wuhan Bridge over Yangtse River
Hubei
Jingsha Bridge over Yangtse River
Hubei
Qingzhoulu Bridge
Fujian
Zhanjiang Bay Bridge
Guangdong
Junshan Bridge over Yangtze River Hubei Prov.
628 m
u.c.(2001)
steel
618 m
500 m
605 m
480 m
u.c.(2001)
u. c.(2002)
u.c
u.c.
mixed
P.C.
composite
P. C.
17 Dafoushi Bridge over Yangtze River Chongqing
460 m
450
u.c
u.c
steel
P. C.
18 Zhenyang Bridge over Yangtze River

Jiangsu
Lingdingyang West Channel
Guangdong
19
400 m
950 m
u.d
u.p
steel
steel
20
1200 m
u.p
steel
12
13
14
15
16
Chongming Bridge over Yangtze

River
Shanghai
Table 1. Major Cable-stayed Bridges in China(L>400m)
Bridge Name
1
2
3
4
Shimen Bridge
Sanxianzhou Bridge
Zhaobaoshan Bridge
Haihe Bridge
Location Main Span

Chongqing
Fujian
Zhejiang
Tianjin
230
238 m
258
310
Year of
Completion
1988
u.c
u.c
u.c
Deck Type
P.C.
P.C.
P.C.
steel
Table 2. Single Pylon Cable-stayed Bridges in China ( L>200m)
2 Long-span Cable-stayed Bridges for 21st Century in China

2.1 Large Crossing Projects on State Highway Along Pacific Coastal Line
To meet the requirements of the rapid development of economy in China, the central government
has planned a new state highway network for 21st century. This new network (Fig.1)will mainly
consist of 5 lines from North China to South China(so called 5 longitudinal lines) and 7 lines
from West China to East China(so called 7 transverse lines) as a skeleton. The 5 lines from the
North to the South are: 1. Tongjiang to Sanya, 2. Beijing to Fuzhou, 3. Beijing to Zhuhai, 4.
Erlianhaote to Hekou and 5. Chongqing to Zhanjiang. The 7 lines from the West to the East are:
1. Suifenghe to Manzhouli, 2. Dandong to Lhasa, 3. Qingdao to Yingchuan, 4. Lianyungan to
Huoerguoshi, 5. Shanghai to Chengdu, 6. Shanghai to Ruili and 7. Hengyan to Kunming. In these
12 lines, 2 longitudinal lines (1 and 3) and 2 transverse lines (4 and 5) as shown in Figure.1 by
thick lines are required to be built in this century, and others have been planned to be completed
in the first two decades of 21st century. Among these lines it is worthy to emphasize the state
highway starting from Tongjiang of Helongjiang Province, Northeast China, and ending at Sanya
of Hainan Province, South China, because this line goes through all big cites along the pacific
coast, the economically developed area of China. On this line there are five large strait crossing
projects, which are more challenging to the engineers.
From north to south, these large projects can be distinguished as Bohai Sea Strait, Yangtse River
Estuary near Shanghai, Hangzhou Bay, Lingdingyang at Pearl River Estuary and finally
Qiongzhou Sea Strait.
Figure 1. State Highway Skeleton Network in China

Yangtse River is the largest and longest river in China, starting from the Qinghai-Xizang Plateau,
and ending into the East Sea at Shanghai. The state highway along the coastal line is planned to
cross over Yangtse river near Shanghai. Because of the high requirement of the navigation at the
river estuary district, a long-span bridge with a main span more than 1200 m and a side span of
500m for the movable channel is needed.
The archipelago of Zhoushan, at the outer fringe of Hangzhou Bay is near Shanghai and very rich
in harbour resources. From Shanghai to Zhoushan of Zhejiang Province, it is needed to build a
highway, and on the way bridges have to be built over Hangzhou Bay and connecting the isles of
the archipelago at Zhoushan. To connect Zhoushan Isles, based on the pre-feasibility study
carried out by the Tongji Bridge Engineering Consultant, several long-span cable-supported
bridges are suggested. Due to the requirement of 300,000t of the navigation ability, one cablestayed bridge with a main span of 900m and a suspension bridge with a main span of 1630m are
proposed.
As the increase of the economy of the Pearl River region and the blossoming trade between the
mainland and Hong Kong, SAR, more border crossings are required. To find the best solution for
the whole route structures, a conceptual design competition has been opened to several bridge
design institutes in China. Up to now, the competition has finished to the first step, the results
show that the bridge over the East Lingding Channel close to Hong Kong might be a hybrid
cable-supported bridge with a main span over 1400 m proposed by the Bridge Design Institute at
Tongji University[Xiang, Chen, 1998]. Figure 2 shows the structural schematic of the bridge
alternative, it is a hybrid system, which is based on the idea of how to increase the torsional
stiffness of the bridge. Another alternative design might be a conventional suspension bridge, but
it seems that it is very difficulty to meet the aerodynamic instability criterion except that a slotted
deck is used, and also the two huge anchorage blocks to be built in the deep water might be very
costly.
Again for the West Lingding Channel, a detail cost analysis between a conventional suspension
bridge and a cable-stayed bridge with a main span of 950 m was carried out. The better
alternative might be the later as shown in Figure 3, which has a mixed deck like the Normandy
Bridge.
Figure 2 Structural schematic of the East Lingding Channel Bridge Proposed by Tongji
University
Figure 3 Structural Schematic of a Cable-stayed Bridge Proposed by Tongji University for West
Lingding Channel
Qiongzhou Sea Strait, located at south of China, might be the most difficult sea strait to be
bridged among all sea straits in China. It separates the mainland and the Hainan Province, an
island which is a special economic zone, with more than 7 million people. This strait, about 20
km wide, with an average water depth of 60 m and maximum water depth of 80-102m, has very
bad natural conditions. The site not only is often hit by typhoons and high tidal current, but also
has possibilities of very strong earthquakes. Up to now there are not any detailed designs for the
strait, what have been done are only some conceptual proposals. Generally speaking, those
proposals include a series of large multi-span suspension bridges with span lengths of 2000m and
3000m, and multi-span cable-stayed bridges with span lengths of 1000m. there are also some
other suggestions combined with bridges and tunnels.
2.2 Major Cable-stayed Bridges over Yangtze River
Since the 1st Wuhan Bridge was finished at the end of 50s, 17 bridge over Yangtze River have
been constructed during the past 4 deades, in which 3 bridges are cable-stayed bridge.
To meet the requirement of the developing of regional economy, the local governments along
the Yangtse River Valley have decided to build more bridges over Yangtse River to form the
local highway network. Several long-span cable-stayed bridge with main spans of more than
400m are now under construction or planning(see Table 1).
2.3 River Crossing Projects for Ring Roads of Big Cities
In China, many provincial capitals or municipalities are located at big riverside, such as Shanghai
at Huangpu River, Hangzhou of Zhejiang Province at Qiangtang River, Nanchang of Jiangxi
Province at Gan River, Changsha of Hunan Province at Xiang River, Nanjing of Jiangsu
Province and Wuhan of Hubei Province as well as Chongqing, a new municipality, at Yangtse
River. All these big cities need multi-ring roads to lighten the increasing pressure of heavy traffic
problems, therefore, some bridges including several cable-stayed bridges have to be built for
connecting both riversides
3 Recent Developments of Cable-stayed Bridges in China

The most important advantage of cable-stayed bridges is its variety of configuration, ease of
construction and competitive cost . So the cable-stayed bridge has been becoming the main type
of long-span bridges within a large range of span-length from 200-800m.
According to the recent level of technology, the cable-stayed bridge with concrete deck has made
a break through 500m. The single-pylon type can be used in the case with a main span from 200300 m, even over 300m. The Haihe Bridge with a main span of 310 m in Tianjing is an example.
It is more appropriate to adopt the composite deck type in the range of main span from 500-700m
in order to decrease the deck weight, the dimension of cable and pylon as well. For the cablestayed bridge with a main span beyond 700m, the steel deck should be considered.
If the side spans are on land or shallows, a mixed deck type with short spans of PC deck in the
side span might be a optimal design by increasing the stiffness and improving the wind-resistant
safety of cantilevering construction.
The A or inverse Y type pylon with two inclined cable plane can make a main contribution in
torsional stiffness of cable-stayed bridges, it provides a good condition in using simple-fabricated
and economic open deck cross section, which can usually meet the requirement of windresistance, instead of closed box deck or separated twin side boxes.
However, for some case of very narrow deck with only 2 lanes. The closed box deck is still
necessary , even together with some additional measures, in order to have enough wind resistant
capacity.
4 Competition and Compromise between Cable-stayed Bridges and

Suspension Bridges
Cable-stayed bridge and suspension bridge are currently the only two available types for very
long-span bridges beyond the span-length of 600m.
Except in the case of having good geological condition, general speaking, the cable-stayed bridge
alternative should be more favorite for span-length less than 700m because of the expensive cost
of anchorage for suspension bridges. The span length between 700 to 1000m is a competitive
zone between these two bridge types. In the recent design competition for west Lingdingyang
Channel, a cable-stayed bridge alternative with a main span of 950m behaves more economic and
stable in strong wind compared with traditional suspension bridge alternatives with a same spanlength. The only disadvantage of very long-span cable-stayed bridge is that, the pylon suffered
large wind loading during the cantilevering erection stage, to which more attention should be
paid. We believe that the penetration of 1000m for the span-length of cable-stayed bridge will
surely be accomplished in the near future.
There were many discussions concerning the limit span of cable-stayed bridge in 70s. Prof.
Leonhardt suggested a cable-stayed bridge alternative with 1800m span-length for the Messina
strait crossing and verified the feasibility for such a extra long-span cable-stayed bridge. Some
researchers reported that, there exists stability problem due to the axial force in the deck, while
the span-length of cable-stayed bridge reaches 1200m and more. In this case the bi-stayed system
could be a solution, which will postpone the limit span of cable-stayed bridge and may also be
regarded as a concession from the fully self-anchored system.
A real compromise between cable-stayed and suspension system is the so-called Dischinger-type
or cable-stayed and suspension hybrid system, in which the main span consists of two cablestayed side sections and a central section suspended on the main cables. The cable-stayed
sections can also be designed as concrete deck, which forms a mixed deck together with central
steel deck section in order to improve the wind-resistant behavior. The main cable may sustain
the deformation of cantilever of cable-stayed sections during erection, which will also reduce the
buffeting response of the pylon under strong wind. Although the fatigue problem in the hangers
near the connection points of two different sections should be solved, the compromise between
two types in their competition zone might be an optimal solution.

Finally, we should say, when good geological condition for anchorage is provided with, the
traditional suspension bridge alternative is still a most natural, reasonable and esthetical type for
very long-span bridges, and 1200m should be a satisfactory limit span for the original selfanchored cable-stayed bridge.
5.Comparison Between Two Stay Cable Systems

The stay cable for cable-stayed bridge was developed continuously in solving its anchor, fatigue,
corrosion protection and wind-induced vibration problems since this bridge type was born in
fifties. Only two cable systems used in China nowadays.
5.1 Parallel Wire Cable System with HDPE Coat
This cable system was initiated in Japan on the basis of electric cable technology. The Shanghai
Municipality decided to establish a new stay cable factory in 1988 for providing the parallel wire
cable system, while Shanghai Nanpu Bridge was under construction. Since then, the loading
capability of cable has been increasing to over 10,000KN with colored skin, and the products was
used in the majority of cable-stayed bridges in China. At the present time, a non-circular, nonsmooth surfaced cable is being developed for solving the rain/wind-induced vibration problem
for the construction of 2nd Nanjing Bridge over Yangtze River with a main span of 628m, which
will be the longest cable-stayed bridge in China.
This type of cable has been adopted in the Tatara Bridge in Japan with a record span of 890m,
and the biggest cable system composed of 421 7 galvanized wires with a loading capacity over
10,000KN was used is Yokohama Bay Bridge in Japan and Third Qiantang River Bridge in
China as well.
5.2 Parallel Strand Bundle Cable System Coated Individually by HDPE
This system was initiated in France with a main advantages in the case of creation for individual
strand and in doing final adjustment by jack with smaller distance of travel. This type of cable
system has been used in many cable-stayed bridges in Europe, especially in Normandy Bridge
with a main span of 856m, and has been spreading in some cable-stayed bridges in China, such
as 2nd Santou Bay Bridge with a main span of 518m. The domestic product is developed by
OVM Corp. in cooperation with Tongji University.
Generally speaking, two types of cable system have their respective advantage. The parallel wire
system is easier and speedy, when the loading capacity is smaller; and the parallel strand system
should be considered for the cable with bigger loading capacity.
6. Wind-resistant Studies for Very Long-span Cable-stayed Bridges

6.1 Nonlinear Theories for Torsional Divergence
Figure 4 shows the structural schematic of the 2nd Shantou Bay Cable-stayed Bridge, locating at
Shantou, Guangdong Province, a strong typhoon-prone area in China, which has a main span of
518 m and a concrete- steel mixed deck. A full aeroelastic wind tunnel model was carried out in
the wind tunnel laboratory at Tongji University. The torsional divergence phenomena was
observed, which happened before fluttering. The critical wind speed of torsional divergence is
120 m/s, which is lower than the flutter wind speed of 134 m/s, and is much lower than that
estimated from the linear estimation[Scanlan], say 209 m/s.

South
North
Figure 4 Structural schematic of the Second Shantou Bay Bridge

It was found that nonlinear effects have to be considered in the analysis, including the nonlinearity of wind loading, which increases not only with the increasing of wind speed , but also
with the deflection of the structure, and the total torsional structural resistance decreases also
with the increasing of wind speed. Figure 5 shows the divergence mode shape of the bridge
obtained from analysis, similar with the divergence mode observed in the wind tunnel. It should
be emphasized that the non-linear torsional divergence should be checked also for cable-stayed
bridges with steel closed box girders.
Figure 5 Divergence mode shape of the Second Shantou Bay Bridge

6.2 Numerical Simulation of Flutter Analysis
We started to try to use CFD technique in bridge aerodynamics in 1997, followed the pioneer
work by Walter and Larsen of Denmark. Figure 6 shows the cross section shape and finite
element grid of 2nd Nanjing Cable-stayed Bridge over Yangtze River, and Figure 7 shows the
comparison of flutter derivatives of the bridge between calculation based on CFD technique and
wind tunnel testing.
Figure 6 Cross section shape and finite element grid of No.2 Nanjing Cable-stayed Bridge over
Yangtze River(Main span 628 m)
0.6
0.0
0.4
-0.1
0.2
-0.2
0.0
0
4
6
Vr=U/fB
10
-0.3
4
6
Vr=U/fB
H*3
-0.2
-0.4
4
6
Vr=U/fB
10
4
6
Vr=U/fB
10
4
6
Vr=U/fB
10
-2
-3
-1
H*2
-1
1.0
0.8
0.6
0.4
0.2
0.0
10
0.0
0
H*1
A *3
0.1
A*2
A*1
0.8
4
6
8 10
Vr=U/fB
calculated
experiment
-2
-3
-4
0
Figure 7 Comparison of flutter derivatives of No.2 Nanjing Bridge over Yangtze River between
calculation and testing
A flutter analysis based on both the flutter derivatives obtained from wind tunnel testing and the
CFD technique was carried out. A satisfied result was obtained. We think that in the near future,
the CFD technique might be widely used in bridge aerodynamics, such as for flutter analysis,
buffeting estimation and wind loading.
6.3 Aerodynamic Selection for Bridge Cross Sections
The design procedure of Qingzhoulu Bridge over Ming River at Fuzhou of Fujian Province is
very interesting. This bridge, now under construction, has a main span of 605 m and a deck of
29m wide, and its wind speed for flutter instability checking criterion is 70 m/s. So, a steel box
girder with a streamlined shape is the first choice. A wind tunnel testing carried out at Tongji
University proved that the critical flutter wind speed is much higher than 70 m/s, but the client is
not satisfied with the design, they think the steel box alternative is too expensive and they want
the bridge has a composite deck with a open cross section as shown in Figure 8. However, a wind
tunnel testing shows that the critical flutter wind speed at +3 degree wind attack angle is only 55
m/s, much lower than 70 m/s.
Different countermeasures are considered to increase the flutter stability, including fairings,
stabilizers, deflectors and skirt plates. Finally, an optimized skirt plate(Figure 8) is found out,
which increases the critical flutter wind speed up to 72 m/s. This example shows that the
aerodynamic measures are efficient to increase the aerodynamic stability.
Figure 8 Cross section of Qingzhoulu Bridge over Ming River
7. Concluding Remarks
z
The construction of modern cable-stayed bridges in China initiated in 1972, relatively later
compared with other developed countries. Up to now, more than 100 cable-stayed bridges
have been built in China. It is still a high tide to build more cable-stayed bridges in the
country, as in this period, a lot of bridges with a main span range of 300-900 m are needed.
So, China will be one of the hot places to build cable-stayed bridges in the next 2 decades.
Based on the experience of cable-stayed bridges in China and other countries, we can expect
that the main span of traditional cable-stayed bridges will go up to 1,200 m, reaching to its
limitation. The hybrid system could be an optimal solution for bridges with a span length
beyond 1,200 m.
Cable-stayed bridges with composite decks and a span length less than 700 m is much
cheaper than those with steel box decks. The aerodynamic stability can be greatly improved
with additional aerodynamic countermeasures.
Also serious wind/rain-induced cable vibration has occurred at several cable-stayed bridges
in China, bridge engineers should pay more attention to this problem, especially for very
long-span cable-stayed bridges.
Reference
[1] Ewert, S., 1997. Neue Grossbrueken in China. Bautechnik 74, Heft 2: 123-127
[2] Lin, Y. P.; Zhang, Z. H. & Ma, B., 1996. Xupu Cable-stayed bridge, Shanghai, SEI, 166-167
[3] Ostenfeld, K. 1992. Aerodynamics of large bridges. SEI, 186-189
[4] Qu, Q. L.; Lin, J. Y., 1996. Chongqing second Yangtse River bridge. SEI, 150-151
[5] Simiu, E. & Scanlan, R. H. 1996. Wind Effects on Structures, Fundamental and Applications
to Design, John Wiley & Sons, Inc. New York
[6] Xiang, H. F. 1991. Cable-stayed bridge in China. Cable-Stayed Bridges, Recent
Developments and their Future, Editor: Ito, M. et al, Elsevier.
[7] Xiang, H. F. 1993. Bridges in China. Tongji University Press, A&U Publication(HK)Ltd.
[8] Xiang, H. F. 1995. Buffeting response analysis and control of long-span bridges. A State of
the Art in Wind Engineering, Ninth International Conference on Wind Engineering, New
Delhi.
[9] Xiang, H. F.; Chen, A. R.& Lin, Z. X. 1998. An introduction to the Chinese wind-resistant
design guideline of highway bridge. Journal of Wind Engineering and Industrial
Aerodynamics 74-76
[10] Xiang, H. F. & Chen, A. R. 1998. 21st century bridges in China, Bridge Aerodynamics,
Proceeding of the international symposium on advances in bridge aerodynamics,
Copenhagen, Denmark, 10-13
Developments in Concrete Cable-Stayed Bridges in the United States

Marcos P. LOIZIAS
Director of BridgeEng
Sverdrup Civil, Inc.
New York, NY, USA
Summary
This paper reviews the developments in cast-in-place and precast concrete cable-stayed
bridges in the United States. It describes characteristic structural details, aerodynamic
considerations and presents the methods of construction of these bridges. Some
significant bridges are discussed in detail along with their cost competitiveness with steel
alternatives.
1.
Introduction
The competitive bidding process between steel and concrete alternatives, previously
required by the Federal Highway Administration for bridges costing over $10 million,
stimulated creative developments in the design and construction of concrete cable-stayed
bridges in the United States.
While many characteristics of the concrete cable-stayed bridges in the United States are
similar to those in other parts of the world, several developments originated and were
further refined in the United States. For cast-in-place construction, they include the use of
flexible girder and open cross-section constructed by a cable-supported formtraveller. For
precast construction, they include the use of delta frames to support separate box girders
and provide for a single plane of suspension.
2.
Cast-in-Place Concrete Cable-Stayed Bridges
The open deck cross-section has dominated cast-in-place concrete construction in the
United States, proven easy to form and structurally efficient. It consists of longitudinal
rectangular edge girders that support transverse floorbeams and a top slab. As the section
is torsionally weak, it requires suspension from two planes of stay-cables. The open
section has a low drag coefficient and is efficient aerodynamically particularly as the
width of the bridge increases and the section approximates a flat plate.
The Dame Point Bridge

The 792.48m long three-span symmetrical Dame Point Bridge in Jacksonville Florida is
the first cast-in-place concrete cable-stayed bridge built in the United States. With a
396.24m long center span and flanking spans of 198.12m each, is still the longest
concrete cable-stayed bridge in the Western Hemisphere. The bridge is a high level
crossing over the St. Johns River and provides for a vertical navigation clearance of
53.34m.
It is the only bridge in the United State where the superstructure is fixed to the towers.
The concrete deck is 32.23m wide and carries six traffic lanes with provisions for a
seventh lane by removing the median barrier. The edge girders are 2.44m wide and vary
in depth from 1.52m to 1.85m at the towers where the compression from the cable thrust
forces becomes the greatest. Post-tensioned transverse floorbeams are spaced at 5.33m on
centers and support the reinforced concrete deck slab that varies in thickness from
229mm to 559mm at the areas of maximum compression.
The stay-cables are spaced at 5.33m along the girder and arranged in a harp configuration.
They are anchored at H-shaped towers with vertical legs extending 91.4m above the deck.
Figure 1. The Dame Point Bridge, Florida

In another unique for the United States application, the stay-cables utilized parallel 32mm
diameter Dywidag thread bars. The grade 1040 thread bars were pressure grouted inside
thick steel pipes that were designed to resist any post-grouting loads on the bridge, i.e.
overlay and live loads, in a composite action between bars and pipe. The steel pipes were
welded on the deck to their required length using full penetration welds and x-rayed to
assure their fatigue capacity under live load. The bars were coupled with mechanical
couplers in 18.28m lengths to their final length and stressed individually in a balanced
procedure to assure the same force in all bars.
The superstructure was built by the balanced cantilever method using a specially designed
light-weight formtraveller that rolled over both edges of previously constructed segments.
The formtraveller was designed to support casting of the entire monolithic pour of edge
girders, floorbeam and deck slab in a typical 32.23m wide by 5.33m long segment.
During casting of a segment, the formtraveller was supported at the front by an extension
of the erected cable and strut and hanger connections to the previously cast segments.
The Dame Point Bridge was initially let in 1979 with the lowest bid for the concrete
cable-stayed bridge at $64.8 million compared to $84.8 million for a steel cable-stayed
alternative. Due to the unfavorable bond interests rate market at the time, the owner of the
bridge had decided to put the project on hold. The bridge was re-let in November1984
during which time the contractors were allowed to bid on the original designs or submit
their own alternate designs in steel or concrete. The lowest bid was for the original
concrete cable-stayed bridge design at $47 million with no bids received for the steel
alternative. The bridge opened to traffic in March 1989.
The Talmadge Memorial Bridge
The Talmadge Memorial Bridge over the Savannah River, Georgia features a three-span
cable-stayed main structure with a 335.28m center span and flanking spans of 143.15m
each. It provides a vertical navigation clearance of 56.4m.
The bridge deck is 24.54m wide and carries four traffic lanes. The edge girders are 1.37m
deep by 1.37m wide. A 282mm thick slab is supported by transverse floorbeams spaced at
8.91m and 8.61m at the center span and flanking spans respectively to match the cable
spacing. The stay-cables are arranged in a semi-fan configuration and anchored at Hshaped towers. The superstructure passes through the towers supported by the stay-cables
and vertical bearings at the roadway strut.
The stay-cables consist of parallel 15mm diameter seven-wire low relaxation prestressing
strands, pressure grouted within a black polyethylene (HDPE) plastic pipe and wrapped in
a light-colored PVF tape. This is the most common type of cable system used in cablestayed bridges in the United States.
The bridge was built by the balanced cantilever method using formtravellers to support
the entire segment consisting of edge girders, floorbeam and deck slab. The movement of
the formtraveller to the casting position required a boggie beam and intermediate cable
adjustments to control the positive bending moment demand in the girders during rolling
of the traveller. At the casting position, the contractor utilized water ballast tanks on the
formtraveller to minimize the need for additional cable adjustments and control negative
moment demand in the girders. The weight of the water was approximately 70% of the
concrete weight and the water was released as the pour of concrete for the segment
progressed. Vertical and diagonal tie-downs were required during construction to stabilize
the deck against wind vibrations.
The bridge was let in June 1987 with the lowest bid for the concrete cable-stayed bridge
at $25.7 million compared to $26.9 million for a steel composite cable-stayed alternative.
The Maysville Bridge
The Maysville Bridge over the Ohio River, Kentucky features a three-span cable-stayed
main structure with a center span of 320.04m and flanking spans of 135.03m each. It
provides for a vertical navigation clearance of 23.5m.
The bridge carries one of the most slender cable-stayed spans in the western hemisphere
with a center span to width ratio of 18. The 17.78m wide deck is designed for four traffic
lanes. The edge girders are 1.37m wide by 1.37m deep. The 254mm thick slab is
supported by transverse floorbeams spaced at 7.69m in the center span and 7.39m in the
flanking spans to match the spacing of stay-cables. A 38mm thick latex concrete overlay
is added over the deck for protection.
The stay-cables are arranged in vertical planes in a semi-fan configuration and anchored
at H-shaped towers that extend 68.6m above the deck. The towers utilize a hollow boxsection above the roadway and solid rectangular section below. The size of the towerhead
at 6.09m long by 2.59m wide was dictated by strength demand as well as the layout of the
stay-cable anchorages to allow stressing and future adjustments of the cables from the
tower.
The wind studies for the bridge included a topographical model of the complex terrain
surrounding the bridge site to assess the directionality of the wind and full aeroelastic and
sectional model tests to confirm the bridges aeroelastic stability against flutter and vortex
excitation.
For the consideration of vortex excitation, peak acceleration limits were set to 0.05g for a
wind velocity up to 13.4m/s and 0.10g for greater velocities, as suggested by the ASCE
Committee on Loads and Forces on Bridges. The corresponding allowable peak
deflection criteria, based on the calculated lowest vertical and torsional frequencies of
0.306 Hz and 0.506 Hz, were respectively 134mm and 0.38 degrees. For the basic
benchmark wind tunnel test conditions, zero degrees angle of wind attack and smooth
flow, vertical vortex-induced oscillations occurred at 9.36 m/s with a peak amplitude of
122mm. The addition of low turbulence was found to significantly decrease peak vertical
excitations. The torsional vortex-induced motions in smooth flow were found well below
the 0.05g criterion. The wind speed for the onset of flutter in smooth flow was in excess
of 53.4m/s compared to the flutter criterion of 37.4m/s based on a 100,000 year return
period 10 minute average wind at deck level.
The bridge was let in December 1996.The concrete cable-stayed bridge bid at $38.3
million was not the low bid. The state went on to construct a steel composite cable-stayed
bridge alternative which was bid at $36.4 million.
The Cape Girardeau Bridge
The Cape Girardeau Bridge over the Mississippi River, Missouri features a three-span
cable-stayed main structure with a center span of 350.52m and side spans of 142.65m
each. It provides for a vertical navigation clearance of 18.9m.
The bridge site is closed to the New Madrid region which has been the most seismically
active region in the central and eastern North America. The bridge is designed for a
ground acceleration coefficient of 0.36g with a 90% probability of not being exceeded in
250 years which essentially corresponds to a repeat of the largest earthquake that shook
New Madrid in 1811 at a surface wave magnitude (Ms) of 8.5.
The 28.6m wide deck carries six traffic lanes. The edge girders are 1.52m wide by 1.52m
deep. The 254mm thick slab is supported by transverse floorbeams spaced at 7.62m in the
center span and 7.24m in the flanking spans to match the spacing of stay-cables. A 76mm
thick silica fume overlay is added over the deck for protection.
at the H-shaped towers that extend 75m above the deck.
The superstructure passes through the towers supported by the stay-cables and vertical
bearings at the roadway strut. Longitudinal fixity at the two towers is accommodated
through shock transmission devices (STU) capable of transmitting a longitudinal force of
15,000 KN at each tower leg. The STU devices provide for translation fixity for all fast
moving (transient) loads and could accommodate a total movement of 178mm under the
combined effects of temperature fall and creep and shrinkage of concrete. Tie-downs at
the anchor piers provide free longitudinal translation.
The geo-seismic studies for the bridge included evaluations of near-fault rupture
potential, soil liquefaction, site response and spatial incoherence of ground motion. The
seismic analysis was based on a time history response analysis and considered multiple
support excitations and substantial forces (22,000 KN) from sloughing of liquified soil. In
addition to the longitudinal and transverse horizontal motions, the design included a
vertical motion equivalent to 70% of the longitudinal motion. To preserve the integrity of
the bridge without significant damage, the substructure and connections were designed
for the elastic earthquake loads.
From the wind tunnel tests, the wind speed for the onset of flutter in smooth flow was
found in excess of 62.4m/s compared to the flutter criterion of 41m/s based on a 100,000
year return period 10 minute average wind at deck level. At zero degrees angle of wind
attack and smooth flow, vertical vortex-induced oscillations occurred at 18.3 m/s with a
peak amplitude of 91mm, but were completely suppressed with the addition of low
turbulence.
The bridge was let in June 1996.The concrete cable-stayed bridge bid at $51.5 million
was not the low bid. The state went on to construct a steel composite cable-stayed bridge
alternative which was bid at $50.85 million.
The Owensboro Bridge
The Owensboro Bridge over the Ohio River, Kentucky features a three-span cable-stayed
main structure with a center span of 365.76m and flanking spans of 147.83m each. It
provides for a vertical navigation clearance of 17.49m.
The 23.06m wide deck carries for four traffic lanes. The edge girders are 1.37m wide by
1.45m deep. The 279mm thick slab is supported by transverse floorbeams spaced at
8.08m in the center span and 7.85m in the flanking spans to match the spacing of staycables. A 38mm thick LMC overlay is added over the deck for protection.
at the H-shaped towers that extend 75.36m above the deck. The superstructure passes
through the towers supported by the stay-cables only. Longitudinal fixity at the towers is
provided through shock transmission devices.
The bridge was let in July1997.The concrete cable-stayed bridge bid at $58.98 million
was not the low bid. The state went on to construct a steel composite cable-stayed bridge
alternative which was bid at $55.45 million.
The Chelyan Bridge
The Chelyan Bridge over the Kanawha River, West Virginia features a three-span cablestayed main structure with a center span of 181.05m and flanking spans of 75.4m each. It
provides for a vertical navigation clearance of 22.98m above the normal pool elevation.
The 22.68m wide deck carries four traffic lanes and a 1.52m wide sidewalk. The edge
girders are 1.37m wide by 1.52m deep. The 279mm thick slab is supported by transverse
floorbeams spaced at 7.69m in the center span and 7.39m in the flanking spans to match
the spacing of stay-cables. A 32mm thick LMC overlay is added over the deck for
protection.
The stay-cables are arranged in a semi-fan configuration and anchored at H-shaped towers
with vertical legs extending 42.98m above the deck. The tower legs are stiffened at deck
level with a stiff strut. There is no strut above the roadway, providing thus a clean and
simplistic appearance to the structure. The superstructure passes through the towers
supported by the stay-cables and vertical bearings at the roadway strut.
From the wind tunnel tests, the wind speed for the onset of flutter in smooth flow was
found in excess of 45.9 m/s compared to the flutter criterion of 41m/s based on a 100,000
year return period 10 minute average wind at deck level. At zero degrees angle of wind
attack and smooth flow, vertical vortex-induced oscillations occurred at 20.9 m/s with a
peak amplitude of 152mm, but were reduced to negligible levels with the addition of low
turbulence.
The bridge was let in May 1995.The concrete cable-stayed bridge bid at $26.89 million
was not the low bid. The state went on to construct a steel truss alternative which was bid
at $25.9 million.
Other Cast-in-Place Concrete Projects
Another cast-in-place concrete cable-stayed bridge built in the United States is the
Cochrane Bridge over the Mobile River in Alabama. It features a 238m long center span
and flanking spans of 110m each. The deck section consists of twin trapezoidal box
girders connected with transverse diaphragms. The bridge was opened to traffic in August
1991 at a construction cost of $70 million.
The Sidney Lanier Bridge over the Brunswick River in Georgia is currently under
construction. It provides for a three-span symmetrical cable-stayed main span structure
with a center span of 381m long and flanking spans 190.5m each. The bridge was let in
October 1996 with the lowest bid for the concrete alternate at $65.5 million compared to
$70.1 million for a steel composite cable-stayed alternate.
3.
Precast Segmental Concrete Cable-Stayed
The most distinctive cross-section for precast construction is the torsionally stiff box
girder supported by a single plane of stay-cables along the centerline of the bridge.
Variations in the cross-section include a single cell box and twin boxes connected with
delta frames for wider bridges. The box girder has large lateral strength and is fairly
efficient aerodynamically with low drag coefficient.
The Sunshine Skyway Bridge
The Sunshine Skyway Bridge in Tampa Florida is the largest precast concrete cablestayed bridge in the United States. It features a three-span cable-stayed main span
structure with a center span of 365.76m and flanking spans of 164.59m each. It is a high
level crossing providing for a vertical navigation clearance of 53.35m.
The bridge is conceptually very similar to the Brotonne Bridge in France with the
exception that the entire segment is precast in a monolithic unit. A typical segment is
29.9m wide by 3.66m long. It consists of a single cell box with inclined webs and internal
struts provided to transfer the stay-cable forces from the anchorage area at the top slab to
the bottom of the girder and over the full depth of the section. The depth of the box girder
is 4.26m.
The bridge is supported by a single plane of cables along the centerline of the bridge. The
cables are arranged in a fan configuration and are continuous through the pylon with the
angle of stays with respect to the deck varying from 22 to 47 degrees. The cables are
anchored at the box girder at 7.31m centers. The stay-cables consist of 60 to 80 15mm
diameter seven-wire low relaxation prestressing strands pressure grouted in steel pipes.
The cables were overstressed before grouting and then released after the grout had set in
order to introduce permanent compression in the grout. Damping devices are provided at
deck level to control cable vibrations.
The towers consist of a single shaft pylon above the deck and twin elliptical box pier
shafts below. The pylon, the box girder superstructure and the twin pier shafts are all
rigidly connected together. Longitudinal movement is accommodated by the flexibility of
the twin pier shafts.
Figure 2. The Sunshine Skyway Bridge, Florida

The bridge was constructed by the balanced cantilever method. The precast segments
were lifted to position from barges by a pair of winches attached to the end of the
cantilevers. The segments were basically prestressed in the precasting yard in the
transverse and vertical direction, while a limited amount of longitudinal prestress in the
top slab permitted the cantilever construction before placing the permanent cables.
External longitudinal tendons, laid above the bottom slab and draped inside the box
section to be anchored at the stay-cable anchor block at the top slab, were prestressed to
resist the combined effects of live load bending moments, temperature gradients and
creep and shrinkage of concrete.
The bridge was let in 1982 with the lowest bid for the concrete cable-stayed bridge at
$106.6 million compared to $109.3 million for a steel composite cable-stayed alternative.
The bid price included the cost for the high level approach spans. The bridge was opened
to traffic in April 1987.
The Neches River Bridge
The Neches River Bridge in Texas features a five-span continuous cable-stayed bridge
with a center span of 195.07m, side spans of 85.34m and end spans of 42.67m. It is a high
level bridge with a vertical navigation clearance of 43.6m.
The 17.1m wide deck consists of a precast segmental trapezoidal box girder with inclined
outside webs connecting directly to the edges of the top slab without cantilevers. Internal
struts provided intermediate support to the top slab. The 2.44m deep box is supported by
two vertical planes of stay-cables anchored directly at the connection of webs and the top
slab. The cables are arranged in harp configuration and are continuous over saddles at the
pylons. The typical segment length is 3.05m, with cable spacing along the girder at 6.1m.
The piers and pylons are made of precast segments assembled by vertical post-tensioning.
The side spans, considered as a natural continuation of the approach spans were
constructed up to the pylons by the span-by-span method of construction with temporary
piers and assembly trusses. The center span was then constructed from the pylon in onedirectional cantilever. The center span segments were raised from barges by winches
attached to the cantilever.
The construction cost for the project was $22.8 million.
The Chesapeake and Delaware Bridge
The 1,417m long Chesapeake and Delaware Bridge in Delaware features a 502.92m long
precast concrete cable-stayed bridge with a center span of 228.6m. It is a high level bridge
with a vertical navigation span of 42.1m.
The 38.4m wide superstructure features twin parallel trapezoidal box girder segments.
Each box girder is 17.92m wide and 3.65m deep. One of the most significant features of
the bridge is that the box girders are independent of each other in the approach spans
while in the cable-stayed main span they are tied together with precast delta frames that
transfer the loads from the girders to a single plane of stay-cables. The stay-cables
arranged in harp configuration, are continuous over a curved saddle in the pylon and
anchored to the precast delta frames, allowing stressing of the cables at the deck level.
The stay-cables carried up to 85 15mm diameter seven-wire strands pressure grouted in
steel pipes. The typical segment length is 3.05m, with the cable spacing along the girder
at 6.1m.
The transition piers are made of 3.05m long precast segments assembled by vertical posttensioning. The pylon is cast-in-place in 3.05m increments to match the precast posttensioned piers. The pylons extend 56.2m above the deck.
The construction of the 228.6m long center span began from each pylon in a onedirectional cantilever method of erection. With the entire erection done from above, box
girder segments were transported with segment haulers over the previously completed
portion of the bridge and set in-place simultaneously at the tip of the cantilever using
crawler cranes. The cable-stayed side spans leading to the pylon were constructed as a
continuation of the approach spans, built in typical span lengths of 45.7m by the span-byspan method of erection using an overhead gantry.
The bridge was let in December 1991 with the lowest bid for the concrete cable-stayed
bridge at $57.9 million compared to $64.7 million for a steel composite cable-stayed
alternative. The bid price included the cost for the high level approach spans. The bridge
was opened to traffic in 1995.
The James River Bridge
The James River Bridge in Virginia is very similar to the C & D Canal Bridge, featuring
twin parallel trapezoidal box girder segments connected with delta frames and supported
by a single plane of stay-cables along the centerline of the bridge.
The 466.3m long cable-stayed bridge has a center span of 192.0m and provides for a
vertical navigation clearance of 44.2m above the James River. The superstructure is
38.4m wide. Each box girder is 17.92m wide by 3.65m deep and carries three traffic
lanes. The cables are spaced at 6.1m along the deck and are anchored every other segment
at the delta frames. The stay-cables consist of parallel seven-wire prestressing strands
pressure grouted in PE pipes. The cast-in-place pylons extend 40.8m above the deck.
The bridge was opened to traffic in July 1990 at a construction cost of $34 million.
The Cooper River Bridge
The 5,029m long Cooper River Bridge in South Carolina also features twin parallel
trapezoidal box girder segments connected with delta frames and supported by a single
plane of stay-cables along the centerline of the bridge.
The 518.16m long cable-stayed bridge has a center span of 243.84mm and provides for a
vertical navigation clearance of 47.24m above the Cooper River. The superstructure is
33.03m wide. Each box girder is 14.35m wide by 3.05m deep and carries three traffic
lanes. The cast-in-place pylons extend 50.4m above the deck.
The bridge was let in November 1986.The concrete cable-stayed bridge bid at $106.61
million was not the low bid. The state went on to construct a steel truss alternative which
was bid at $89.42 million. The bid price included the cost of the high level approach
spans.
Other Precast Projects
Two other precast concrete cable-stayed bridges were built in the United States. They are
the Pasco-Kennewick Bridge in Washington with a 229m long center span and
triangular-shaped box girder superstructure supported by two planes of stay-cables, and
the East Huntington Bridge in West Virginia with a center span of 274.4m and featuring
an open cross-section with transverse steel floorbeams.
Several other precast segmental concrete cable-stayed bridges were carried through final
design but not built. The projects included the 6th Street Bridge in West Virginia bid at
$28.6 million compared to $24.5 million for a steel truss alternative, the Roosevelt Lake
Bridge in Arizona where a steel arch at $18.7 million came over $5 million lower than the
cable-stayed bridge alternative, and the Baytown Bridge in Texas where a steel
composite cable-stayed bridge was built at a construction cost of $91.25 million while no
bids were received for the concrete alternative.
4.
Conclusion
Concrete cable-stayed bridges have been proven very competitive with steel alternatives
for spans in the range of 180m to 400m. With dramatic increases in concrete strengths
and quality, good aerodynamic performance, superior durability, ease of inspection and
maintenance and contractors familiarity with this type of construction, concrete cablestayed bridges are expected to be competitive in upcoming bridge projects that require
main navigation spans in the range of 500m.
The Innovative William Natcher Cable-Stayed Bridge
Vijay CHANDRA
P.E,. Sr Vice President
Parsons Brinckerhoff
New York, NY, USA
Received an M.S. degree
(Advanced Structures)
from the University of
London and a B.E.
degree from the
University of
Mysore, India
Ruchu HSU
P.E., Supv. Structural
Engineer
Parsons Brinckerhoff
New York, NY, USA
Received an M.S. degree
(Civil Eng.) from
Polytechnic University
of New York and a
B.S. degree from
National Cheng-Kung
University, Taiwan
Introduction
For the Kentucky Transportation Cabinet (KTC), Parsons Brinckerhoff (PB) designed a steel
state-of-the-art cable-stayed bridge (See Figure 1) that successfully competed against a concrete
alternate cable-stayed bridge. The bridge spans the Ohio River connecting Owensboro,
Kentucky and Rockport, Indiana. The contract was bid in September 1997 and construction
began in November of the same year.
There were three bidders for the project. Two of the lowest bids were for the steel alternate. The
selected bid price was $55.45 million, excluding the main tower foundations, which were bid at
$14.55 million and built earlier due to funding issues.
Figure 1. Elevation View of the Natcher Bridge

Being one of the longest cable-stayed spans over the US inland waterway system, we had to
carefully evaluate the design to develop an efficient, safe and durable bridge. Also, one of our
original objectives was to make the bridge user-friendly for the contractor, inspectors and
maintenance personnel. To accomplish this, we went beyond normal US stay cable technology.
Some of the bridges noteworthy features include:

Simple and flexible details of the girder-to-stay cable anchoring system
Efficient prefabricated composite steel stay cable anchoring system in the towers
Simple and efficient superstructure fixity connection at the Kentucky side tower
Continuity of the superstructure at the anchor piers with the approach girders
Flexible stay cable specification to accommodate all present anchoring systems
Bridge Configuration
Total Width of Structure

Number of Traffic Lanes
Superstructure Approaches
A. Land:
Prestressed Concrete Beam
Spans
Depth
Spacing
B. Water:
Steel I-Beam Spans
Depth
Spacing
Superstructure Cable-Stayed
Steel Edge I-Girder with
Floor Beams
Spans
Girder Depth
Floor Beam Spacing
Deck Slab Thickness
Latex Modified Concrete
Overlay Thickness
Substructure
Foundations
Kentucky
Approach
21.28 m
4
Main Span
23.68 m
4
Indiana
Approach
21.28 m
4
Four equal spans

of 33.5 m
1.68 m
Six at 3.11 m
Five equal spans of

41.8 m
1.83 m
Seven at 2.67 m
84-152-84 m
3.7 m
3.74 m
84 m
3.7 m
3.74 m
203 mm
None
152-366-152 m
1.52 m
4.57 m
229 mm
38 mm
203 mm
None
Pile bents over

land
Hammerhead
piers at anchor
piers
Pile bents over land
Hammerhead
piers
in water
Diamond
shaped piers at
tower piers
Hammerhead piers
in water
Drilled shafts
1.22 to 2.44 m
diameter up to
40 m deep
Drilled shafts
1.83 m
diameter
up to 25 m
deep
Drilled shafts
1.22 to 1.83 m diameter
up to
34 m deep
Aesthetics
Function was the overriding factor in defining various bridge elements. For instance, the shape
of the tower piers was based on economy, functionality, constructibility, inspectability,
maintainability, torsional deck stability, cable connectivity, etc. However, aesthetics also played
a role once the shape was selected. Our in-house architects sculptured the exposed face at the
top by giving it definition in the form of reliefs and striations. Our experience shows that
architects play a useful role in shaping a bridge, as long as form is not put ahead of function.
Design Criteria
The following design criteria were used on the project:
American Association of State Highway and Transportation Officials (AASHTO) HS-25
truck load
Barge impact forces over the length of the total bridge, specifically:
Land approach: single 180-ton empty jumbo barge moving with the current at 100-yearflood condition
Water approach and main crossing: 15 (3 x 5) fully loaded jumbo barges with tow
moving under power with the normal current or single 1,700-ton jumbo barge moving
with the current at 100-year-flood condition (Maximum force = 13,300 to 17,800 kN)
Seismic Forceper AASHTO Seismic Performance Category B
Restricted channel closing to barge traffic
100-year return wind speed = 130 kph
Scour plus barge impact = 50% scour depth for normal flow and 100% scour depth for 100year-flood flow
Gravity, thermal and transient loads
Stay cable replacement with only the two far lanes fully live loaded
Accidental loss of a cable with all lanes fully loaded
Hydraulic Analysis
When obstructions are placed in a river or floodplain, the flow characteristics of the river are
affected. Flow is of particular concern to bridge designers because it affects scour, which is the
scooping out of soil around bridge foundations. The faster the current, the worse scour can be.
Therefore, care must be taken to avoid creating fast-flow conditions when choosing the span
arrangement for a bridge. Hydraulic analysis determines the flow velocities associated with
various span arrangements in order to determine the optimum span placement. The flow
velocities are then used to calculate the severity of anticipated scour and the depth needed for the
bridge foundations.
We performed a state-of-the-art hydraulic analysis and scour evaluation of the 6.4-kilometerwide floodplain of the Ohio River in the area of the Natcher Bridge. This evaluation included
the main river crossing and the relief structures in the overbank areas on the Kentucky floodplain
side.
We performed a 2-D finite element hydraulic analysis of 89 square kilometers of the Kentucky
floodplain in the vicinity of the crossing, using the Finite Element Surface Water Modeling
System (FESWMS). The computer model was validated and calibrated using known flood
information for the Ohio River in the vicinity of the bridge. The program provided flow and
discharge vectors at every nodal point. Although more than 14 alternatives, including 8 to 10
options each, were studied, only two adequately met the specified criteria:
Flow rate not to exceed 1.22 meters/second immediately downstream of the bridge
Backwater elevation not to exceed 76 millimeters/second
After a detailed comparative evaluation was performed, examining cost, hydraulics, operations
and maintenance, KTC selected the alternative with five flood relief structures in the overbank
area on the Kentucky side. Utilizing information provided by the FESWMS model and the
Federal Highway Manual on scour (HEC18), scour at each main bridge pier and at the flood
relief bridges were computed for both contraction as well as general scour. The information
provided by the velocity vectors and the scour analysis also assisted in determining ship impact
forces.
Tower Piers
The importance of tower piers and their shape cannot be
overemphasized because they are the most dominant element of
a cable-stayed bridge. After a careful evaluation of tower
shapes during preliminary design, PB selected a diamondshaped (also known as an A-shaped) tower (See Figure 2). The
shape of the tower improved the torsional stiffness of the
superstructure and ensured the stability of the bridge during
construction and to seismic and wind loads when completed.
The Natcher Bridge will have two identical concrete A-shaped
towers supporting three superstructure spans through stay
cables. The towers will be 100.6 meters high, rising 79.3
meters above the bridge deck. Each tower has two inclined
legs above the deck and a trapezoid-shaped head to contain all
the cable anchors. Below the deck, the two tower legs bend
inward and are held in position with a tie strut.
Figure 2. Tower Piers
The superstructure passes through each tower between the legs

and sits on top of the tie strut with bearings under each edge
girder. Bearings at the Kentucky tower are fixed, while those
at the Indiana tower are expansion type. At the base of the
tower head, a 1.5 x 1.5 meter opening is provided for easy
equipment access from the roadway level. Meanwhile,
lightening arresters and air beacons are housed at the top of the
tower.
The tower legs are 4.88 x 2.44 meters above the roadway and 4.88 x 3.66 meters below the
tension strut, which is prestressed to a jacking force of 70,300 kN to resist dead load plus live
load tension. Polyethylene sheathing is used for prestressing tendons in the struts.
Superstructure Continuity
The superstructure consists of steel edge girders with steel floor beams (See Figure 3). The edge
girder web is inclined at 8 for the inclined stay cables anchored at the towers. The floor beams
are spaced at 4.57 meters. To control torsion in the floor beams during precast slab erection, a
central beam is added.
The deck slab consists of precast concrete units with cast-in-place (CIP) infills. The slab is
connected to the edge girders and floor beams by shear studs. A 38-millimeter overlay of latex
modified concrete will be placed on the deck slab.
A unique feature of the Natcher Bridge involves the transition from the cable-stayed back spans
to the adjacent approach spans at the anchor piers (See Figure 4). In cable-stayed bridges,
challenging design issues arise in the vicinity of anchor piers where a transition occurs from the
cable-supported structure to a
conventionally supported structure.
Therefore, many important elements are
located in this area, including:
Counterweights and tie-downs to
secure the anchor cables
Windlocks to prevent relative
transverse movement between the
deck and anchor pier
Bearings for both the cable-stayed
spans and the approach spans
Figure 3. Superstructure Detail
Figure 4. Transition at Anchor Pier
We rigidly connected the approach girders

to the cable-stayed edge girders and floor
beams because they were designed to be
continuous over the anchor piers. There is
no relative rotation between the two
adjacent spans. The bearings at the anchor
piers allow the structure to slide
longitudinally over the anchor pier.
In the transition area, six longitudinal
approach girders are framed into the two
cable-stayed edge girders through a series
of three floor beams. Moments are
transferred by upward and downward
forces on these floor beams. The approach
girders, edge girders and floor beams are
all 3.7 meters deep at this location, beyond
which the two cable-stayed edge girders
gradually decrease to a typical depth of
1.52 meters.
The beauty of the connection to the approach girders is that the approach dead load reaction
resists part of the uplift. The remainder of the uplift is resisted by concrete counterweights. The
counterweights are integrated into the superstructure and placed so that their center of gravity
coincides with the centerline of bearing at the anchor pier. The counterweights are provided at
each anchor pier to completely balance the uplift, even for the worst loading case with a full live
load exclusively on the main span. This system offers several advantages at the anchor piers,
including:
Absence of deck joints
Approach spans assist in resisting uplift
Live load stress range reduced in anchor cables
Deck slope continuity maintained
AASHTO also requires provisions to resist an additional uplift force equal to the maximum live
load upward reaction. This was addressed by adding a simple cable tie-down device, which is
installed slack to allow for thermal movement of the deck.
Longitudinal Fixity
The longitudinal forces at the Kentucky tower are resisted by two brackets that drop down from
the edge girder and hug a heavily reinforced concrete pedestal. Steel reinforced elastomeric
bearing pads that rest between the steel brackets and the pedestal transfer the longitudinal force.
The pad has a Teflon surface that bears against the bracket so that rotation of the girder is not
hampered.
Windlocks at Anchor Piers

One windlock is provided at each anchor pier, located at the centerline of the bridge, under the
floor beam. As the name implies, windlocks provide transverse restraint against wind; however,
they also restrain transverse movement caused by impacts and seismic events. The controlling
load case was barge impact, and the windlocks were designed to transfer a 4,400-kN impact
force to the superstructure at ultimate capacity. We also anticipated that some elements of the
windlock might be damaged in the event of a full barge impact. Therefore, these parts were
designed to be easily replaced. Service load design was used for normal wind load.
The wind locks were designed to accommodate necessary longitudinal and vertical movement, as
well as free rotation of the superstructure about the anchor pier. Each is comprised of two
assemblies:
A lower assembly, attached to the pier cap with anchor bolts
An upper assembly, bolted to the underside of the floor beam bottom flange
Stay Cables and their Connection

Early in the design process, we specified greased and sheathed strands (or flo-fil epoxy-coated
strands) with grout in a PE pipe. Since coextruded PE pipe had recently become available in the
US, both black PE pipe with Tedlar tape wrapping and coextruded pipe with a white exterior
were specified (subsequently, the contractor opted to use the coextruded pipe). Also, while
spiral beads on the outside of the pipe were accepted as a means to reduce rain/wind vibration,
cross ties were provided to reduce the effects of galloping. During design, total flexibility was
provided for the stay cable anchorage, wedge, wedge socket and Hi-Am types. In addition, both
individual and multi-strand jacking were included in the specifications to provide flexibility.
Cable-to-Girder Connection (Non-Stressing End)
PB designed a simple bolted splice
connection between the connection plate
and girder web (See Figure 5). This
completely eliminates torsion of the
girder, allowing the connection to be
located between floor beams. To provide
for shear flow in the edge girder where
the top flange has been slotted, we
designed an angle to be bolted to the
connection plate and to the top flange,
along the slot. The connection plate is a
flat steel plate that passes through a slot
in the top flange of the edge girder as an
extension of the edge girder web. A
bolted splice connection is used to avoid
Figure 5. Cable-to-Girder Connection
stress concentration and reduce the risk
of cracking in the weldment. The other end of the plate is cut into a tuning fork shape with two
prongs, between which a thick-walled pipe is welded. The cable is installed by inserting its
anchor head into the steel pipe. A ring nut or a shim plate then supports the anchor head to bear
against the end of the pipe.
Two additional plates welded to the connection plate and to the pipe stiffen the pipe against
squashing and reduce the required thickness of the connection plate. These plates give the cross
section of the connection a cruciform shape. They are tapered and stop above the top of the castin-place portion of the concrete deck, which is poured around the connection plate. The
cruciform shape is an open section that allows easy access for inspection. Below the edge girder
top flange, the splice connection can also be easily inspected.
The cable anchorage is located above the deck. Construction workers, inspectors and
maintenance personnel can access the cable anchors directly without using expensive special
equipment. The cables are also protected against accidents and vandalism because the steel
pipes enclose the cable to a distance of 2.4 meters above the deck.
Once again, maximum flexibility is provided with this cable anchorage system. Depending on
the contractors preference, the anchorage can be shipped with the steel edge girder floor beam
assembly, shipped and erected separately, or attached to the cable prior to installation.
Cable-to-Tower Connection (Stressing End)
For the Natcher Bridge, PB designed steel frames attached to the interior walls of the cable
anchorage chamber by shear studs. These frames carry the horizontal component of the cable
force, and transfer the vertical component to the concrete (See Figure 6). They can also transfer
unbalanced cable forces during cable replacement or loss.
Each tower head contains 12 steel
frames, each supporting two side span
cables and two main span cables. A
frame consists of two built-up
channels with flanges inclined to
match the slope of the inner tower
walls. A cap plate with a steel pipe is
welded to each end of the channels,
and the inclined flange and cap plates
are attached to the tower walls with
shear studs. The cable bears against
inclined support plates that are
sandwiched between the channel
flanges. Maximum flexibility was
provided in the fabrication, assembly,
Figure 6. Cable-to-Tower Connection
transporting and erecting of these
steel frames. They can be preassembled in the shop to whatever height the fabricator and
contractor are comfortable with or they can be assembled in the field.
The closed chamber at the top of the tower piers provide a protected environment for the cable
anchorage and is a convenient location for cable stressing operations. Platforms located at each
cable level will facilitate the inspection and maintenance of the frames and anchorages.
Wind Tunnel Testing

To ensure aerodynamic stability, a wind study was performed for the bridge that included wind
data collection and analysis and wind tunnel tests. Rowen Williams Davies and Irwin, Inc.
(RWDI) was retained to perform the wind study.
Wind tunnel tests were performed on both a sectional model and a full aeroelastic model. The
aeroelastic model was used for testing the bridge during four construction stages, as well as when
completed. The various criteria used for testing the Natcher Bridge encompassed:
Design wind speed of 132 kph to compute static wind loads
Design flutter speed of 154 kph to study the dynamic wind effect
Structural peak acceleration limits of 5% of gravity up to 48 kph and 10% of gravity above
48 kph for studying vortex excitation
All models were tested for smooth and turbulent air flows. The vortex excitation was within the
criteria and occurred around 72 kph. Flutter speed was around 192 kph, well above the 154 kph
predicted every 100,000 years at the site.
Though baffle plates in the mid-third of the main span were found to be necessary during the
sectional model tests, they were later removed during aeroelastic model studies due to the 3-D
action of the stay cables.
Construction Sequence
Designing for anticipated construction stages is as important as designing the structure itself,
particularly on complex structures such as cable-stayed bridges. Without careful planning, many
delays can ensue during construction, leading to increased costs. To limit such impacts, we
closely evaluated construction conditions.
Because access to the Ohio Rivers shipping channel must be maintained during construction,
the bridges cable-stayed spans were designed to be constructed using the balanced cantilever
segmental method (ensuring that shipping lanes will remain free of temporary construction
supports).
The construction sequence for the balanced cantilever method involves the following steps:
1. The tower is constructed.
2. A centrally positioned superstructure segment is erected at the tower and supported by
temporary bracing in the area of the towers.
3. The remaining superstructure segments are erected sequentially on alternate sides of the
tower until connections are made at the approach pier and, finally, at mid span.
Additionally, each superstructure segment is erected in the stages listed below:
1. A steel frame is lifted into position by
a barge-mounted crane and field
spliced to the edge girder of the
previously installed segment. It is
allowed to cantilever freely.
2. Two stay cables connecting the steel
frame to the tower are installed and
tensioned to an initial length.
3. The six precast deck panels are
installed and cast-in-place concrete
closure strips are placed between them
and along the edge girders of the
previously installed segment.
4. The two cables are tensioned to their
final length, and the procedure is
repeated on the opposite side of the
tower.
Two-level tensioning of the cables is
planned to control both erection and
locked-in stresses in the deck and edge
girders. Also, to reduce bending moments
in the tower caused by wind loads and
Figure 7. Typical Construction Stage
construction sequence imbalances, a
temporary tie-down is connected to the superstructure of the back span (See Figure 7). In
addition, weights are strategically positioned and repositioned on the deck to control stresses in
the superstructure.
Construction Status
Presently, Traylor Brothers, the low bid contractor, has completed the Kentucky side approach
pier foundations, some of the approach piers and the inclined legs of the two main tower piers
(See Figure 8). The tension strut of the Indiana tower pier is under construction, while the main
tower pier foundations were completed under a previous contract. The overbank approach
embankment and flood relief structures are also complete.
Conclusion
The design of the William Natcher Bridge
was a challenge in which PB faced tough
competition from the concrete alternative.
Because early completion of the main
tower foundations under a separate
contract eliminated one of the major
advantages of a steel cable-stayed bridge,
namely a less expensive foundation due to
lighter loads, we needed to be bold and
innovative in other areas without
sacrificing safety, strength and durability.
We also wanted the bridge to be easy to
inspect and maintain.
Hands-on inspection and future
maintenance are critical for bridge
Figure 8. Kentucky Approach Pier Foundations
longevity. When taken together with the
fact that the cable anchors of cable-stayed
and Inclined Legs of Two Main Tower Piers
bridges are traditionally very difficult to
access, PB made the accessibility and maintainability of these areas two of the most important
factors to be addressed during design. We developed innovative ideas for cable-to-tower and
cable-to-deck connections as well as for the elimination of uplift and deck joints. Our design
allows inspectors and maintenance crews to perform hands-on work by walking directly to the
cable anchors without using any special equipment. These innovations will result in the first
user-friendly cable-stayed bridge in the United States.
Acknowledgements
The authors express our sincere thanks to the Kentucky Transportation Cabinet, Mr. Steve
Goodpastor and Mr. Henry Phillips, for allowing us the opportunity to design the bridge and for
their cooperation and assistance during the design phase. We also appreciate the assistance of
Mr. Jim Lyle of KTC during construction and Mr. Charles Raymer for his assistance during the
hydraulic study. In addition, we wish to express our sincere appreciation to the American
Institute of Steel Construction; the National Steel Bridge Alliance; the contractors, fabricators
and erectors; Dr. Schlaich and Partners; and many in-house experts for their contributions and
important suggestions during the design phase.
Identification of minimum width-to-span ratio of long-span cable-stayed

bridges based on lateral torsional buckling and flutter analyses
Masatsugu NAGAI
Xu XIE
Professor
Nagaoka University of Technology
Nagaoka, Japan
Kaihatsu Consultant Co., Ltd.

Tokyo, Japan
Hiroki YAMAGUCHI
Yozo FUJINO
Professor
Saitama University
Urawa, Japan
Professor
University of Tokyo
Tokyo, Japan
Summary
This paper describes static and dynamic instability analyses of long-span cable-stayed bridges
such as finite displacement analysis under displacement-dependent wind load and flutter analysis
based on modal coordinate. Using a 1400-meter cable-stayed bridge model, in which four cross
sections of the girder having different widths with a fixed depth of 3.5 meters are selected, static
and dynamic instability analyses are carried out. Instability behaviors of them are made clear
and, finally, the design material for identifying a minimum width-to-span ratio of the girder is
presented, which ensures safety against the instabilities.
1. Introduction
In the design of long-span cable-stayed bridges, ensuring safety against static and dynamic
instabilities under wind load is an important issue, because the shape and dimension of the girder
are controlled mainly by above instabilities. However, static and dynamic instability phenomena
of long-span cable-stayed bridges have not been made clear so far. In this paper, using a 1400meter cable-stayed bridge model, static and dynamic instability analyses such as a nonlinear
static analysis under displacement-dependent wind load and a flutter analysis based on multimode coordinate are carried out. Four types of the cross section of the box girder having
different widths of 25,28,32 and 35 meters with a fixed depth of 3.5 meters are chosen. A spanto-width ratio is from 56 to 40, and a span-to-depth ratio is 400. It is recommended, for ensuring
safety against out-of-plane instability under wind load, that the span-to-width ratio should be less
than 401). However, in this study, the larger values are employed. If the value of 40 is inevitable
for ensuring safety against instability under wind load, the width of the girder become large
regardless of the number of traffic lanes. This means that long-span cable-stayed bridges become
less competitive compared with other alternatives such as suspension bridges. Hence, an
identification of the minimum width of the girder becomes an important issue. The employed
box girders are preliminary designed, in which the yield point of steel is only selected to be an
instability criterion. By carrying out above instability analyses, the critical wind velocities of
lateral torsional buckling and flutter are investigated. Finally, the design material for obtaining
minimum cross-sectional shape and dimension of the girder is presented.
2. Analytical Procedure
Static nonlinear analysis under displacement-dependent wind load and flutter analysis based on
modal coordinate had been explained in our previous papers2)3)4). Hence, we explain the
procedure briefly.
2.1 Finite Displacement Analysis Under Displacement-dependent Wind Load
When the girder is subjected to the wind load, it displaces in the horizontal direction and starts
rotating. Due to the rotation of the girder, three components of aerodynamic forces such as the
drag force, lift force and aerodynamic moment vary, because they are dependent on the angle of
wind attack. This displacement-dependent characteristics of wind loads are taken into account in
the analyses. Furthermore, the wind load acting on the cable and the change of the tension in
cables are also considered.
Fig.1 shows aerodynamic coefficients experimentally measured for a cable-stayed bridge5). In
the figure, CD,CL and CM are aerodynamic coefficients with respect to the drag force, lift force
and aerodynamic moment, respectively, and is the angle of wind attack. In this analysis, these
coefficients are used for the calculations.
CD ,CL ,CM
5
4
3
2
1
0
-1
-2
-3
CD
CL
CM
-15
-10
-5
angle of attack (degree)
10
15
Fig.1: Aerodynamic coefficients

2.2 Flutter Analysis Based on Modal Coordinate
A fundamental equation of flutter analysis is derived based on modal coordinate. The unsteady
drag force of the girder is derived based on quasi-steady theory, and the unsteady lift force and
aerodynamic moment are derived based on flat plate theory. The unsteady drag and lift forces of
the cables are derived based on quasi-steady theory.
The fundamental equation of flutter analysis is given by
q
q
[ ]T [(M BC FR ) iFI ][ ] + [ ]T [K BC ][ ] = {0}
qC
q C
(1)
Where, {q} is the generalized displacement, which corresponds to the global vibration, {qC } is
the generalized displacement of the cables, which corresponds to the cable local vibration,
[M BC ] and [K BC ] are mass and stiffness matrices, respectively, and [FR ] and [FI ] are matrices
consisting of real and imaginary parts of the unsteady aerodynamic forces, respectively, [ ] is
the modal matrix which is obtained by carrying out eigenvalue analysis of the whole structure, in
which cable local vibration is neglected.

Assuming the reduced frequency, complex eigenvalue analysis is carried out, then we obtain
complex eigenvalue of = R i I . When the sign of the damping ( = R
changes from plus to minus, flutter occurs.
R + I )
3. Bridge Model
Fig.2 (a) shows a side-view of a cable-stayed bridge model. Center and side spans are 1400 and
680 meters, respectively. In the side span, three intermediate piers are installed at a distance of
100 meters in order to increase in-plane flexural rigidity. Fig.2 (d) is a front view of the tower,
and the tower height from the deck level is one fifth of the center span length. Fig.2 (b) shows
the cross-sectional shape of the girder. A depth of the girder is 3.5 meters. A span-to-depth ratio
of the model is 400, which is larger than that used in the design of conventional steel cablestayed bridges. The selected widths of the girder are 25, 28, 32 and 35 meters, respectively, and
the span-to-width ratios of them are from 56 to 40.
These four models are preliminary designed. The cross-sectional properties of the girder are
determined by applying the following design conditions.
1) The thickness of 12mm is used for both deck plate and lower flange. Taking into account of
longitudinal ribs, which are expected to bear the axial force, their effective thickness is assumed
to be 20mm. The thickness of the web plate is 15mm.
2) The dead load per unit length ( WD ) is calculated by Eq. (2).
(2)
WD = 1.4 AS S + 70 (KN / m )
Where, AS is the cross-sectional area of the girder, a coefficient of 1.4 is to take account for the
load from diaphragms, cross beams and so on, S is the weight density of steel and a constant
values of 70 KN / m is the superimposed dead load such as the pavement, handrail, curb,
attachment and so on.
3) The live load per unit length is 38 KN / m .
4) The design wind velocities of the girder and cables are assumed to be 60 and 70m/s,
respectively. Those of them at the stage of erection are 70% of above values.
5) The drag coefficients ( C D ) of the girder, cable and tower are assumed to be 0.8, 0.7 and 1.4,
respectively.
Dimension of the girder is determined by using the following criteria:
( 1 = 1.7)
D + L < y 1 19.6 ( MPa)
D + w < y 2 19.6 ( MPa)
( 2 = 1.15)
!
"
Where, D , L and W are stresses from dead, live and wind loads, respectively, y (=451Mpa)
is the yield point of the employed steel and is the factor of safety. A value of 19.6MPa is the
margin, because the bending moment under dead load, shear lag effect, shear stress and so on are
not taken into account at this preliminary design stage.
To satisfy Eq.(4), the thickness of the plate is increased as shown in Fig.2(c). This is in order to
increase out-of-plane flexural rigidity of the girder efficiently. In the bridge axis direction, the
section of Xu as shown in Fig.2 (a) is reinforced. Table 1 shows the cross-sectional properties of
the girder preliminary designed. In the table, the figures in the parenthesis are ones for the
reinforced girder.
(m)
10
C
L
132
3@100
=300 680
Xu
380 20
Xu
20
20
660
20
1400
148
(a) Side-view
Bu
x
2%
7
40
Hw
46
(b) Cross section

0.02
5
Tup
0.015
12
0.04
B
30
10
0.008
(d) Tower
(c) Increase of plate thickness for region 'Xu'
Gider
Fig.2: Bridge model

Bu
(m)
Hw
(m)
25
3.5
28
3.5
32
3.5
35
3.5
tower (per one column)
A
(m2)
1.314
(2.243)
1.433
(2.147)
1.642
(2.070)
1.761
(2.046)
1.760
Ix
(m4)
2.560
(4.050)
2.849
(3.985)
3.269
(3.943)
3.939
(4.432)
30.67
Iy
(m4)
75.65
(177.3)
102.6
(204.6)
151.5
(234.6)
193.2
(261.1)
40.32
J*
Iw
W
Xu
(m4)
(m6)
(KN/m)
(m)
5.767 90.431 210.3
200
(9.395) (314.4) (281.8)
6.542
131.3
223.1
180
(9.365) (358.8) (278.1)
7.583
212.7
245.6
120
(9.561) (383.2) (278.6)
8.330
282.0
258.4
80
(9.739) (421.5) (280.4)
39.27
189.6
* longitudial ribs are neglected
Table.1: Cross-sectional properties of the girder and tower
Cross-sectional area of the cable is designed under the condition that a live load-to-dead load
ratio is 0.25 and the allowable stress is 588MPa. The allowable stress assumed is around 7%
lower than that of 628MPa, which has been used in Japan.
In-plane load-carrying capacity of one of the models, which has a width of 25 meters, is
examined by employing 3D elasto-plastic finite displacement analysis6)7). Under uniformly
distributed load applied throughout the bridge length, the load parameter of 2.9 is obtained,
which is a ratio of the applied load to dead load. Fig.3 shows interaction curves between axial
force and bending moment for different three points, A, B, C in the girder shown in Fig.2 (a). In
the figure, N and M are the produced axial force and bending moment, respectively. Ny and MP
are the yield axial force and full plastic bending moment. The stress produced in the girder at the
tower point, C, reaches yield point, followed the redistribution of the stress resultants and,
finally, the bridge collapses when the axial force at points A and B reaches the yield axial force
(N/Ny = 0.98). Fig.4 shows an incremental displacement at ultimate state. It is seen, at points A
and B, that the vertical displacement increases rapidly. The obtained value of 2.9 is considerably
large. Hence, it is concluded that the safety of the employed models against in-plane instability is
ensured. From this result, it is also interesting to know that the ultimate strength of the girder will
be controlled by local buckling instability of stiffened plates and that the larger span-to-depth
ratio exceeding 400 is expected to be used.
N/Ny
N/Ny
1.0
1.0
1.0
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.0
0.0
N/Ny
N 0.85M
+
= 1.0
Ny
Mp
0.2
0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
1.0
0.0
0.2
0.4
0.6
M/MP
M/MP
(a) at Pt.A
0.8
(b) at Pt.B
0.8
1.0
M/MP
(c) at Pt.C
Fig.3: Interaction curves
Fig.4: Incremental displacement at ultimate state
4. Results and Discussions

4.1 Lateral Torsional Buckling Instability Under Wind Load
Fig.5 shows lateral displacements, vertical (upward) displacements and rotational angles at the
middle of the center span of the bridge. Fig.6 shows those at the tip of the cantilevered girder of
the cable-stayed bridge under construction. In case of the completed bridge in Fig.5, at the wind
velocity of around 60m/s, nonlinear behavior of the vertical displacement and rotational angle
becomes prominent and, in the range of the wind velocity from 75 to 80m/s, they diverge. This is
lateral torsional buckling. In case of the bridge under construction, since the system is flexible,
the larger lateral displacement is obtained. At the wind velocity of around 50m/s, nonlinear
behavior of the vertical displacement becomes prominent and, in the range of the wind velocity
from 65 to 70m/s, the bridge becomes unstable. In all models, the critical wind velocities
calculated are high enough compared with the design wind velocity.
Figs.7 and 8 show the tensions in cables. In case of the completed bridges, at the wind velocity
from 75 to 80m/s, they start decreasing rapidly. In case of the bridge under construction, in the
range of the wind velocity from 65 to 70m/s, they also start decreasing. These rapid decreases of
the tensions in the cables are due to rapid increase of the upward displacement of the girder.
30
25
20
15
10
5
0
(m)
10
8
6
4
2
0
-2
Bu=25m
Bu=28m
Bu=32m
Bu=35m
(m)
10
Bu=25m
Bu=28m
Bu=32m
Bu=35m
(degree)
Bu=25m
Bu=28m
Bu=32m
Bu=35m
8
6
4
2
0
0 10 20 30 40 50 60 70 80 90
0 10 20 30 40 50 60 70 80 90
0 10 20 30 40 50 60 70 80 90
(a) lateral displacement
(b) vertical displacement
(c) rotational angle
wind velocity (m/s)
wind velocity (m/s)
wind velocity (m/s)
Fig.5: Displacement at the middle of center span (after completation)

60
50
40
30
20
10
0
(m)
30
25
20
15
10
5
0
Bu=25m
Bu=28m
Bu=32m
Bu=35m
(m)
14
12
10
8
6
4
2
0
Bu=25m
Bu=28m
Bu=32m
Bu=35m
0 10 20 30 40 50 60 70 80 90
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)
wind velocity (m/s)
(a) lateral displacement
(degree)
Bu=25m
Bu=28m
Bu=32m
Bu=35m
0 10 20 30 40 50 60 70 80 90
(b) vertical displacement
wind velocity (m/s)
(c) rotational angle
Fig.6: Displacements at the tip of the cantilevered girder (under construction)

12
10
8
6
4
2
0
(MN)
Bu=25m
Bu=28m
Bu=32m
Bu=35m
12
10
8
6
4
2
0
(MN)
Bu=25m
Bu=28m
Bu=32m
Bu=35m
0 10 20 30 40 50 60 70 80 90
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)
wind velocity (m/s)
(a) windward
(b) leeward
Fig.7: Tensions in uppermost cable in the center span (after completation)

(MN)
10
10
4
2
0
Bu=25m
Bu=28m
Bu=32m
Bu=35m
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)
(a) windward
4
2
(MN)
Bu=25m
Bu=28m
Bu=32m
Bu=35m
0
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)
(b) leeward
Fig.8: Tensions in uppermost cable in the center span (under construction)
Fig.9 shows the maximum normal stress in the girder at the tower point. In case of the completed
bridge, when the wind velocity of all models reaches around 70m/s, the normal stress exceeds
yield point. Since the plastic region spreads in a limited range of the girder, it will not lead to
immediate collapse of the bridge. However, strictly speaking, elasto-plastic finite displacement
analysis is recommended for identifying exact buckling instability. In case of the bridge under
construction, the maximum normal stress exceeds yield point when the wind velocity is around
60m/s. In this case, elasto-plastic finite displacement analysis is also recommended.
700
600
(MPa)
700
600
500
right hand side of eq.(4)
500
(MPa)
right hand side of eq.(4)
400
400
300
300
Bu=25m
Bu=28m
Bu=32m
Bu=35m
200
100
0
Bu=25m
Bu=28m
Bu=32m
Bu=35m
200
100
0
0 10 20 30 40 50 60 70 80 90
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)
wind velocity (m/s)
(a) completed bridges
(b) bridges under construction
Fig.9: Maximum normal stress in the girder at the tower point

4.2 Effect of Aerodynamic Coefficients on Instability Behavior
In the analysis carried out in 4.1, the aerodynamic forces presented in the paper5) are used.
However, it is natural to consider that the aerodynamic coefficients vary depending on the cross
sectional shape and it is interesting to identify how the behavior changes depending on the
aerodynamic coefficients. Unfortunately, since the aerodynamic coefficients for the employed
cross section are not obtained, we try to carry out instability analysis using differently assumed
values. It is seen, from Fig.5, when the rotational angle exceeds 3 degrees, that nonlinear
behavior of it becomes prominent. Hence, when the angle of attack is larger than 4 degrees, the
values of CD and CL are multiplied by 1.5 and between 3 and 4 degrees, linear interpolation is
used. Fig.10 shows thus assumed aerodynamic coefficients and solid notations are new ones.
Since the coefficient of aerodynamic moment (CM) is small, it is kept to be the same.
CD,CL ,CM
5
4
3
2
1
0
-1
-2
-3
CD
CL
1.5CD
1.5CM
CM
-15
-10
-5
angle of attack (degree)
Fig.10: Assumed aerodynamic coefficients (solid circles)
10
15
Figs.11 (a) and (b) are upward deflections and rotational angles at the middle of the center span
of the completed bridge and Fig.12 is upward defections at the tip of the cantilevered girder. In
case of the bridge after completion in Fig.11, the deflections diverge in the range of the wind
velocity from 65 to 70m/s. This wind velocity is around 10m/s lower than those obtained in 4.1.
In case of the bridge under construction, the deflections diverges in the range of the wind
velocity from 55 to 60m/s. This is also around 10m/s lower than those obtained in 4.1. From
these results, though the obtained wind velocity is still higher than the design wind velocity, it is
known that instability behavior strongly depend on the aerodynamic coefficients.
10
8
6
4
2
0
-2
(m)
10
Bu=25m
Bu=28m
Bu=32m
Bu=35m
8
6
Bu=25m
Bu=28m
Bu=32m
Bu=35m
4
2
0
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)

(a) vertical displacement
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)

(b) rotational angle
Fig.11: Displacement at the middle of the center span (after completion)

30
25
20
15
10
5
0
(m)
Bu=25m
Bu=28m
Bu=32m
Bu=35m
0 10 20 30 40 50 60 70 80 90
wind velocity (m/s)
Fig.12: Vertical displacements at the tip of the cantilevered girder (under construction)
4.3 Flutter Onset Wind Velocity
Table 2 shows the flutter onset wind velocity of the completed bridge and the bridge under
construction, respectively. In the table, the figures of 30 and 20 are the number of modes
employed which are needed to obtain the converged wind velocities. The figures in the
parenthesis are the results when the effect of the cable local vibration is taken into account. From
the results, it is found that flutter onset wind velocity is higher than the critical wind velocity
under static wind load. It is also found that the effect of the cable local vibration is prominent.
Finally, it is concluded that the dimension of the girder is controlled by instability under static
wind load.
Model
Bu=25m
Bu=28m
Bu=32m
Bu=35m
Completed
30-mode
Selberg
120
130
(141)
120
130
125
132
127
133
Under construction
20-mode
Selberg
99
40
(151)
109
95
105
98
107
100
Table.2: Flutter onset wind velocity
5. Concluding Remarks
The followings are main results obtained from this study.
1) In all models, the flutter onset wind velocity, even in case of the bridges under construction,
exceeds around 100 m/s and is higher than the critical wind velocity of lateral torsional buckling
under static wind load. Hence, the static instability under displacement-dependent wind load
controls the dimension of long-span cable-stayed girders.
2) On condition that the bridge is designed based on the procedure explained in Sec.3, the girder
with a width-to-span ratio of around 1/55 can be used. However, for saving the steel weight, if
we employ the steel with higher strength, the ratio of 1/55 has to be reconsidered. Because it
results in the reduction of flexural and torsional rigidities of the girder. Hence, in such case,
instability analyses have to be employed.
3) Aerodynamic coefficients affect critical wind velocity. Hence, for attaining higher critical
wind velocity, controlling the aerodynamic characteristics of the girder becomes an important
issue.
References
[1].
[2].
[3].
[4].
[5].
[6].
[7].
F.Leonhardt and W.Zellner : Past, present and future of cable-stayed bridges, CABLESTAYED BRIDGES (ed. by M.Ito et al.), Elsevier, pp.1-33, 1991
X.Xie, H.Yamaguchi and M.Ito : Static behaviors of long-span cable-stayed bridges, Proc.
of JSCE, No.537/I - 35, pp.205-215, 1996 (in Japanese)
X.Xie, H.Yamaguchi and M.Nagai : Static behaviors of self-anchored and partially earthanchored long-span cable-stayed bridges, Int. Jour. of Structural Engineering and
Mechanics, Vol.5, No.6, pp.767-774, 1997
M.Nagai, X.Xie, H.Yamaguchi and Y.Fujino : Static and dynamic instability analyses of
1400-meter long-span cable-stayed bridges, IABSE Symposium Kobe 1998, IABSE
Reports, Vol.79, Kobe, Japan, pp.281-286, 1998
V.Boonyapinyo, H.Yamada and T.Miyata : Nonlinear buckling insatiability analysis of
long-span cable-stayed bridges under displacement-dependent wind load, Jour. of
Structural Engineering, JSCE, Vol.39A, pp.923-936, 1993
X.Xie, M.Nagai and H.Yamaguchi : Ultimate strength analysis and behavior of long span
cable-stayed bridges, Jour. of JSCE, No.598/I - 44, pp.171-181, 1998 (in Japanese)
M.Nagai, X.Xie, H.Yamaguchi, K.Nogami and Y.Niida : Elasto-plastic behavior and
strength of 1400-meter long-span cable-stayed bridges, Proc. of Nordic Steel Construction
Conf. 98, Vol.2, Bergen, Norway, pp.573-578, 1998
Computer Based Optimising of the Tensioning of

Arne BRUER
Civil Engineer
Teknisk Data AS
Oslo, Norway
Heinz PIRCHER
Senior Partner
TDV GesmbH
Graz, Austria
Heinz BOKAN
Partner
TDV GesmbH
Graz, Austria
Summary
A numerical approach to reduce the calculation effort when attempting to minimise the number
of stressing operations during the erection of cable-stayed bridges is shown. The proposed
method is illustrated with sample calculations from a small example and from the Uddevalla
bridge which is currently under construction.
The Problem
The solution for the optimum tensioning strategy for long span cable-stayed bridges can be an
extremely tedious and time-consuming process for the following reasons:
Practical reasons
Tensioning one cable affects the forces in all the other cables
Cables can not, in reality, withstand compressive forces but stressing an adjacent cable may
apparently cause this condition.
Stressing of the stay cables is an expensive procedure due to the difficulty in the cable
stressing procedure.
Analytical reasons
A minimal cable tensioning strategy whilst saving a considerable amount of time and money
during the construction phase greatly complicates the analytical phase of the design process.
Definition of the tensioning strategy is interrelated with the chosen erection method and the
simulation of the erection procedure using the structural model can be very complicated.
The deck girder and pylon system must behave reasonably during all phases of construction.
i.e. the deflections should be neither excessive nor incompatible with type of construction.
Creep and Shrinkage (applicable where some or all the bridge elements are concrete or
partially concrete members) greatly complicates the analytical process.
Uplift conditions could exist at the temporary supports which further complicates the
analysis. Whilst special Non-tension members could be used, this greatly increases the
degree of indeterminacy and hence the speed of analysis.
All of the above practical and analytical reasons obviate the need for a consistent, standard, non
trial and error type approach to the solution for these complex structures. It is possible to use a
unit load system of analysis tied to the bridge construction method, relate this to an estimate of
the max/min final live load moment envelope and through this, where possible, minimise the
cable stressing operations.
The proposed method will always achieve a solution, which must then be checked for structural
consistency by the user. The results can be structurally unacceptable as the solution is directly
achieved from a set of simultaneous equations. The structural unacceptability may arise from
such things as compression in the cables or unacceptably high stresses etc.
Structurally acceptable results clearly demonstrate that the parameters chosen to define the
structure and its construction are correct and also define the required tensioning and
construction strategy
Structurally unacceptable results will point the way for modification of the parameters to be
used in the next analysis. (The modification would typically be to the Ideal Bending
Moment Diagram Refer below)
Choosing the System and manipulating the Moment Diagram.

The basic bridging system must be chosen and optimised before the stressing strategy design can
be found. The system is chosen through a series of considerations such as required bridge
functionality, availability and cost of materials, Clients requirements etc. The bridging system is
considered, from this analytical viewpoint, as basic given information.
Integral with the bridging system choice is the concept that almost any moment diagram can be
achieved in the deck and in the pylon by adjusting the following Degrees of Freedom:
The tensioning forces in the stay cables and their stressing procedure
The support movements (translation longitudinal and vertical)
Prefabrication shape of the deck girder and the pylon
The erection procedure of the deck and the pylon
Finding the Ideal Moment Diagram for dead load.

Once the basic information has been defined in principle, the effects of the traffic / pedestrian
loads and any additional loading balustrade / guard-railing / surfacing / etc. on this
fundamentally defined structure can be estimated.
The load capacities of the deck sections and the pylon sections can then be compared with the
live plus additional dead load envelopes and the ideal dead load force diagrams may then be
defined. The sign of the ideal dead load force diagrams may well be of opposite sign to the
diagrams resulting from normal load directions. This demonstrates the distinct advantage of
cable supported structures where the initial dead load moment diagram can be easily manipulated
to suit the design needs.
Establishing the Unit Force Equations Principles

General
When the ideal dead load force diagrams have been defined then the system of unit forces can
be mathematically equated to these ideal dead load force diagrams.
The process for defining the tensioning sequence and amount, the deck and pylon construction
sequence as well as any required deck/pylon prefabrication effects then begins.
Process Principle:
The unit loading system is first defined for the final stage structure in order to establish
reasonable member sizes. This process usually involves certain re-definitions of the member
sizes and program re-runs to prove structural integrity.
Once reasonable values have been achieved the unit force method can be extended to the
construction stage analysis.
Each construction stage can be checked and proven for design compliance.
Degrees of Freedom
The most commonly selected unit forces or Degrees of Freedom (N.B these DOFs are not the
same as the structural system DOFs) in the structural system include:
A unit shortening of the cable (causing an axial cable tension) or a unit tensioning causing
an axial cable shortening.
A unit translation of a rigid support (transverse or longitudinal movement at a pier or
abutment support). A longitudinal force applied at the end of the deck changes the
moments by changing the cable forces which act on the deck.
Setting up the Unit Force Equations
Define the unit loading cases and the ideal moment diagram. The same number of unit
loading cases must be defined as the number of Fixed Moment points chosen on the
structural model to represent the ideal moment diagram (or vice-versa!).
Principles in Example below:
The ideal dead load bending moment diagram is defined for the deck girder by bending
moments at 9 points along the girder (positions A, B, C ...... I).
Nine unit loading cases are selected for setting up the simultaneous equations.
The 8 unknown required stay cable forces chosen in this case to be 1000 kN.
One unit translation at the end support chosen in this case to be 50 cm settlement.
The solution to the equations (the unknowns) will be the factor by which the unit loads should be
factored to achieve the ideal dead load bending moment diagram.
Note: There is no fixed prescription for the selection of the unit loading cases. The designer is
free to choose whatever he wishes. This flexibility is demonstrated in the example by the nonselection of the two stay cables adjacent to the pylon (positions E and F).
The 1000 kN unit cable force was selected to be of the same order of magnitude as the final
cable force because of 2nd Order considerations. (Refer below for further description of 2nd Order
effects.)
The ideal moment diagram for dead load is given below and is very different from the dead
load bending moment diagram (MP) which would result if the loading was applied to the
structure with un-stressed cables.
N.B. Care must be taken in the selection of sensible and unrelated ideal moments as if one is
related to another (i.e. dependent on it) then a singularity in the equations will result and
there will be no solution. Provided there is no singularity, a solution will be reached.
Figure 1.
The following system of linear equations is set up:
MA = MP + MT1=1 . X1. + MT 2 =1 . X 2+ ........ MT 8=1 . X 8+ MT J . X 9
.
.
.
.
MI = MP + MT1=1 . X1. + MT 2 =1 . X 2+ ........ MT 8=1 . X 8+ MT J . X 9
MA....... MI
MP
MT1=1... MT 8 =1
MJ
Final stage moment at the current position (including tensioning + jacking).

Permanent load moment at the current position (without tensioning or jacking).
Bending Moment due to each unit tensioning at the current position.
Bending Moment due to unit jacking of the end support at the current position.
The X 1 ...... X 9 factors set up in the unit loading cases are the unknowns in the set of linear
equations which are found by the solution of these equations.
Note that the system of equations is not symmetrical and the diagonal coefficient may be zero.
This is to be considered when solving the equations. This basic solution defines the cable forces
and the jacking force for the final stage and, at the moment, does not include the effects of:
the sequence of construction stages, the creep, 2nd Order Theory or the non linearity of the cables
due to the sagging effects.
The basic principles must therefore be extended to accommodate these effects.
Construction Stage analysis.
A similar system of unit loading cases can be defined for the construction stage analysis. The
unit loading cases are, in this case, applied to the different structural systems which exist at the
individual construction stages. The sketches below, which show a few of the construction stages,
demonstrate the principle of the method of analysis. The loading cases for each construction
stage are combined to form the set of simultaneous equations which must be solved to find the
required multiplication factors for the unit loadings.
Figure 2
Creep a rational linear approach

There is a general belief that creep design is a non-linear problem and therefore it is often
approached in an empirical manner using some rules of thumb or past experience to assess
its affects. This approach is particularly prevalent where the structural concrete is subjected to
the many and varying loads which occur during the multiple construction stages of large cablestayed bridges.
Taking account of creep effects using the CEB-FIP model code is even more complex than was
the case using more traditional methods. Inspite of the above statements it can be shown through
a series of mathematical equations that the effects of creep can be treated in a linear manner.
The derivation for the effects of creep is founded on the known fact:
{e} . = {c}
(Elastic Strain * Creep Factor = Creep Strain)
Decomposing the structure down to element level, the above equation is applied to each
individual element by applying the generalised displacement method rules for calculating initial
strain type loads:
Define {e} over the element.
Define {c} = {e} . over the element.
The member end displacements {c} are found by weighted integration of the strain vector over
the element length in the usual way.
The member end forces are calculated and the system of equations are assembled and solved for
nodal displacement {} in the usual way.
({} - {c}) . [k] = {FI} gives the internal forces due to creep.
The system of analysis is completely linear up to this point.
In the specific case of creep, cognisance must be taken of the age differences in the concrete as
well as the various ages of different parts of the structure at the time of each increment of load
application (e is no longer constant but varies with time). A finite difference approach in time is
applied here and using a linear variation over a time interval, we can say:
{t} = {0} . t 1 t +{1} . t t 0
t
t
This equation can then be put into the basic displacement equation: at time t0 , {e} is known
and then by solving the equations for {} at time t = t1 a recursive formula can be derived which
results in a linear relationship at time t1 such that the equation including the effects of creep is
the same as the original equation with the exception that the modulus E is replaced by
E
A detailed description of the whole procedure is given in Ref. 2 and the
1+*0.5
theoretical background for the finite difference approach to solve initial value
problems is described in Ref. 3.
The essence of the above statement is, that all the creep influences on the final distribution of
internal forces and displacements are related in a linear manner to the elastic strain which itself
initially caused the creep.
The principles of linear superposition may, in consequence, be applied and the total creep
occurring during a single time step may be decomposed into single contributions:
Considering one of the prescribed ideal moment positions:
M creep = M p + M c t =1 . X 1 + M c t =2 . X 2 + M c t =3 . X 3 .............etc.
M creep therefore consists of one part which is related to the permanent load and the other parts
are related to the unit loads described above which are linearly coupled to the same unknown
factors X 1 ...... X 9.
As before the basic concept can now be applied; the effects of creep for permanent loads and for
unit loads are decomposed into separate contributions from each time interval and then summed.
The system of equations for defining X 1 ...... X 9. therefore remains linear. The only
approximation made is the assumption that the behaviour within any single time step is linear
which is consistent with the usual application of the finite differences in time approach.
Second Order Theory and cable non-linearity (due to cable sag)

Since the element stiffness depends on the axial force (in the case of 2nd Order Theory as well as
for cable sag), the basic displacement method equations become non-linear. The equations
defining the solutions for X 1 ...... X 9, which were proved to be linear for the creep case above,
also become non-linear. An iterative approach must therefore be applied:
The Simple Approach
Estimate X 1 ...... X 9, and use this estimate of the unknowns to find the variable stiffness
which on substitution into the equations hopefully gives a solution which is close to the final
behaviour.
Correct the estimate of X 1 ...... X 9 and calculate again.
A Better Approach
Use the tangent stiffness for calculating the influence of the application of a small increment to
each unit loading case. The equations can then be transformed to define the iterative correction
for X 1 ...... X 9 and a procedure such as the Newton Raphson method can be set up.
The tangent matrix for the 2nd Order Theory or even large deflections (with respect to suspension
bridges) can be similar to that usually applied in the Large Deflection Theory. E.g the
corrective term N/L is added into the appropriate position in the element stiffness matrix. The
cable sagging effects can be accommodated by deriving dS/d x from the well known Peterson
Formulae (Ref. 4). Where S means the Cable force and x is the cable extension. Convergency
is accelerated and guaranteed, when using the tangent matrix with the Newton Raphson approach
as long as a real solution exists.
The Results from the sample analysis.

This particular example was chosen not only to demonstrate the principles of analysis but also to
demonstrate the effects of 2nd Order Theory and of creep on the structure.
The results from a few selected points have been chosen for demonstrating these principles:
Final Stage cable forces (kN) resulting from the different analyses
Cables
Posn E
Posn F
Table 1.
1st order theory

& creep
1073.9
1000.8
2nd order theory

no creep
1079.8
1003.5
2nd order theory

& creep
718.6
663.9
Initial Stage cable forces (kN) resulting from the different analyses
Cables
Posn B
Posn I
Table 2
1st Order Theory

& creep
1775.52
1788.02
2nd Order Theory

no creep
1484.44
1468.22
2nd Order Theory

& creep
1833.65
1840.82
Pylon Moments (kNm) resulting from the different stage analyses (design system)
Construction
stage
1
2
3
4
5
Final
Table 3
1st Order Theory

& creep
-500.00
-471.0
-466.2
-463.1
-2151.6
-2019.0
2nd Order Theory

no creep
-41
-32
-33
1493
-617
-617
2nd Order Theory

& creep
-503.0
-472.0
-464.0
-538.0
-2413.0
-1977.0
Pylon Moments (kNm) resulting from the different stage analyses (1st system)
Construction
stage
1
2
3
4
5
Final
Table 4
1st Order Theory

& creep
0
0
0
10492.4
10651.0
244.1
2nd Order Theory

no creep
-
2nd Order Theory

& creep
0
0
0
15657.0
15567.9
1126.0
Minimum Deck Girder Moment Envelope (kNm) (design system)

Construction
stage
1
2
3
4
5
Table 5
1st Order Theory

& creep
-4338
-4129
-4194
-3951
-3792
2nd Order Theory

no creep
-4851
-4877
-4835
-5018
-5089
2nd Order Theory

& creep
-4347
-4223
-4228
-3982
-3819
Maximum Deck Girder Moment Envelope (kNm) (design system)

Construction
stage
1
2
3
4
5
Table 6
1st Order Theory

& creep
1876
3232
2906
3675
3906
2nd Order Theory

no creep
1238
2420
2756
2860
3146
2nd Order Theory

& creep
1935
3401
3098
3775
4018
These above results highlight:

The importance of accurate creep action assessment and shows that creep effects are critical
to the structural integrity and must be accurately calculated and can not simply be assessed
from some arbitrary rules. It can be seen that the creep in the deck affects the cable forces
which in turn affect the deck and pylon moments significantly. The pylon moments are
modified to such a degree that they are even reversed in construction stage 4.
The importance of consistent construction stage checks as the moments in the pylon, whilst
being quite acceptable in the final stage are excessive in the construction stage under the 1st
system of analysis.
The significant changes to the pylon moments caused by 2nd Order effects. (The pylon is
highly compressed and therefore sensitive to additional moments from the deflected shape).
The easy parameter design check:
Whilst a solution to the 1st system of analysis was found, the pylon failure in construction
stage 4 & 5 was easily identified.
Inspection of the 1st system showed that the translational fixity at the pylon was the cause
of the excessive moments. Removal of this fixity proved to be an adequate modification
to the design system.
Construction Stage Analysis forwards or backwards?

The traditional method of carrying out the construction stage analysis is to start at the Final
stage structure and gradually reduce the structure (going backwards) stage by stage until the
first construction stage is reached.
It is argued that this method is the most likely to achieve the fastest result as it starts from a
structurally correct solution - the final stage which may possibly have been defined using the
unit load method described above. Whilst this argument does have much merit, it falls down
when problems are subsequently found at a particular construction stage. The check then reduces
to a trial and error method. Using the proposed unit load method, the forwards or backwards
solution are equally possible and equally simple as even the creep principles described above can
be adapted for the backwards solution by solving the equations for 0 instead 1.
The Uddevalla Cable-Stayed Bridge

A bridge which was designed using the principles described above is the Uddevalla CableStayed Bridge.
This cable-stayed bridge is the central part of a continuous 1712m crossing over the Sunningesund waterway between Uddevallamotet north and Uddevallamotet south in Sweden.
The approach viaducts, comprising twin steel box girders with a concrete slab, are rigidly
connected to the main bridge on either side and provide overall longitudinal structural stability.
The cable-stayed bridge portion comprises a 414 metre main span, symmetrical side spans of 179
metres and two 85 metre high (above the deck girder) diamond shaped concrete pylons which
anchor the fan shaped stay cable arrangement. The stay cables, which support the bridge deck on
either side, are anchored at 13.32 metre centres in the longitudinal direction.
The bridge deck structure carries 6 lanes of traffic and comprises a composite, open steel grid
structure with a 240 mm thick concrete top slab which spans longitudinally over the diaphragms.
The deck edge beams (I type beams) also have a thin walled shell structure connected to the side
which in addition to acting as a wind spoiler provides some torsional stiffness to the edge beams.
More comprehensive descriptions and details of the Uddevalla Cable-Stayed Bridge can be
found in Ref. 4.
Given below is a summary of the principles used in the analysis of the Uddevalla Cable-Stayed
Bridge using the unit load method:
The Degrees of Freedom (or unknowns) chosen for the unit load analysis were:
All the stay cables a unit tensioning
Translation at one cable-stayed bridge pylon support ( X and Y directions)
The Uddevalla Cable-stayed Bridge construction requires a 3 stage stressing procedure:
Stage I stressing provides support for the new steel portion of the deck during assembly. The
cables are initially stressed to provide support and to counteract excessive deflection before
making the welded connection to the existing deck.
Stage II stressing provides support for the whole structural self weight comprising the steel plus
pre-cast concrete top slab elements. The procedure is simple as the stressing jacks from the
previous stressing operation are still connected. The first Unit load analysis to find the cable
forces is carried out at this stage.
Stage III stressing is required for counteracting the superimposed dead load and creep effects on
the pylon deflection. The procedure is required because the stringent minimal pylon moment
criteria precludes a sufficient pylon pre-camber. The second Unit load analysis to find the retensioning cable forces is carried out at this stage.
The Ideal Moment diagram chosen for the initial dead load is shown below together with a
general bridge arrangement. Note the unusual shape in this Ideal Moment diagram in the deck
girder was dictated by a strict limitation prescribed for the pylon moments which takes
cognisance of the very severe environmental conditions for reinforced concrete
weathering/corrosion. In order to comply with this stipulation, the Ideal Moment diagram had
to include a minimal moment condition in the pylon.
Figure 3
Ideal Moment Diagram
Figure 4
The diagrams below show the time dependency of a few characteristic results from the analysis.
The time axis is not to scale but shows a sequence of the different actions. Stages 1-16 are the
cable tensioning and deck cantilevering stages. The deck construction is complete at the end of
stage 16, the additional dead load is applied in stage 17 and stage 18 is for creep and shrinkage
up to time infinity.
Cable force variation in side span cables 5, 6, 7 and 8 (numbering from the pylon)
Figure 5
Main span moment variation at cable 2 north and cable 2 south (numbering from the pylon)
Figure 6
Moment variation in pylon at the top of the footing
Figure 7
References:
[1] C. Hansvold, Sunnungesund Cable-Stayed Bridge, IABSE Symposium Kobe 1998
[2] Heinz Pircher, Finite Differences to simulate creep and shrinkage in pre-stressed concrete
and composite structures, Proceedings of the Int. Conf. on computation Modelling of
Concrete Structures, edited by H. Mang, N. Bicanic, R. De Borst, Pineridge Press 1994
[3] O. C. Zienkiewicz, R.L. Taylor, The finite Element Method, Fourth Edition Volume 2
[4] C. Peterson, Abgespannte Maste und Schornsteine Statik und Dynamik, Berlin: Ernst +
Sohn 1970
Evolution of design trends in cable-stayed bridges
Miguel A. ASTIZ
L. Fernndez TROYANO
Javier MANTEROLA
Prof. Dr. Civil Engineer
Dr. Civil Engineer
Prof. Dr. Civil Engineer
Carlos Fernndez Casado S.L. Carlos Fernndez Casado S.L. Carlos Fernndez Casado S.L.
Madrid, Spain
Madrid, Spain
Madrid, Spain
Miguel A. Astiz, born in 1950,
received his civil engineering
degree from the Polytecnical
University of Madrid in 1973.
Leonardo Fernndez Troyano, born

in 1938, received his civil
engineering degree from the
Polytechnical University of Madrid
in 1963.
Javier Manterola, born in 1936,

degree from the Polytechnical
University of Madrid in 1961.
Summary
This paper presents an overall perspective of how cable-stayed bridge design has evolved
during the last twenty five years. This perspective is supported mainly on the authors
personal experience as well as on well known cases. Aesthetics, technogical advances both in
structural analysis and materials, reliability improvements and increased public interest are
considered as the factors which have developped an impressive activity in this field. At this
point specific codes seem to be necessary to maintain the level of quality which has been
reached until now.
1.
Introduction
Athough 40 years old, modern cable-stayed bridges have undergone a radical evolution along
the last 25 years. Technological advances applied to analysis, materials fabrication and
construction have driven the designers to face very different problems along these years. In
fact, the design of a cable-stayed bridge is a rather standard activity nowadays while it could
be considered exceptional a few decades ago.
What are the reasons for such a change? Practice and experience are good reasons. Knowledge
on cable-stayed bridges has progressed steadily; conferences and symposia on cable-stayed
bridges are being organized and they get a great success. As a consequence of all this
universal engineering effort, the world record for the span length has increased by a factor of
about three in thirty years and, what is more important, the cable-stayed bridge has become a
very common alternative to be considered when designing a new bridge.
Our personal experience reflects this general trend. We entered into the cable-stayed bridge
field in 1974 with the Glorias Catalanas footbridge in Barcelona and we had the opportunity
to design, three bridges and nine years later, the Barrios de Luna bridge, a world record span
length at the time. After this very rapid growth, the cable stayed bridge has become a common
alternative in many of our projects.
We will present in this paper a general perspective of how aesthetics, general structural
design, codes, structural analysis, cable technology and construction methods may influence
the design of a cable-stayed bridge and how the influence of these topics has changed along
the last decades. The question to be finally asked refers to the relative importance of
engineering and architectural concepts in the design of cable-stayed bridges.
2.
Bridge design
For a bridge designer, the bridge has to be seen mainly as a means for fulfilling a number of
functional and safety requirements. Then the static aspects have to be present during the
creative phase of conceptual design. This may not always be true and examples can be found
where statics was not even considered during the conceptual design; the resulting bridge may
badly engineered but it may also be a great architectural achievement.
What is the reason for choosing a cable-stayed alternative? In some cases this is the obvious
solution: there are no alternatives in a certain range of span lengths: roughly from 300 to 900
meters. But most of the cable-stayed bridges which are being built have a span length which is
lower than 300 meters. At this stage the cable-stayed bridge has to be compared to other
classical alternatives such as the girder or the arch bridge.
Apart from economical considerations which are supposed to be the decisive arguments when
we decide on the best alternative to solve an engineering problem, the cable-stayed bridge
adds a bonus which is very difficult to evaluate: cable-staying opens a very broad field for
crativity in the design of the towers and the cable system and, to a lesser extent, in the design
of the deck.
From a purely static point of view, the cable system is responsible for one of the most
important changes which may be appreciated in the evolution of cable-stayed bridges: the
multiple stay concept. At the beginning, the cable-stayed bridge was understood as a
continuous girder which was supported at a limited number of supports by the cables; many
years ago, engineers found out that it was much more effective to design many stays, located
at small distances; as a consequence, the deck behaves like a beam on an elastic foundation
and bending moments are reduced and better controlled during construction.
An important side effect of this evolution is the increased aesthetical importance of the cable
system. The fact that there are many cables allow the designer to shape new plastic effects
while preserving the original static role of the cables. A very illustrative example can be found
in the Ebro and the Lrez bridges [1] which were designed at a 20 year interval with the same
concept for the pylon and for the static cable system but with different cable distributions. The
Ebro bridge has two planar backstay systems while in the Lrez bridge the backstay system
shape hyperbolic paraboloid surfaces. There are some reasons to explain the differences
between these two cable systems: the Ebro bridge is a motorway bridge with important
constraints with respect to the horizontal alignment of the carriageway while the Lrez bridge
is located in an urban environment and there is a roundabout just behind the pylon where
larger counterweights may be fitted. All these functional requirements still leave some room
for to achieve an important visual impact in the Lrez bridge(fig. 1).
Fig.1. The Lrez bridge, Pontevedra, Spain (1996)

A single tower seems to be a good departure point to produce new staying systems. In the
Malecn footbridge [2] we combined a complex deck structural work (torsion combined with
an arch effect in the horizontal plane) to hang it from one edge. In this way the stays define a
surface which may be observed by the pedestrians who are crossing the bridge. The backstays
oppose a radically different concept (fig. 2). Some of these ideas were also present in the
Glorias Catalanas footbridge. This is obviously not the most economical alternative to cross
the river but it shows how it is possible to depart from classical solutions for a small budget
increase (at least in footbridges and for small span lengths).
Fig.2. The Malecn bridge, Murcia, Spain (1996)

The tower is the other important aesthetical feature of a cable stayed bridge. It complements
the cable system and both are mutually dependent. As a matter of fact it is very difficult to
consider both elements as separate entities either from the structural or from the aesthetical
point of view. As pylons in cable stayed bridges are tall and many tension forces are applied
on them at different points, the structural problem is by no means simple but the possibilities
for asthetical expressivity are still high.
The structural and the functional problems have made engineers to design different alternative
arrangements: H-shaped towers for bridges with two planes of cables (fig. 3a), A (fig. 3b),
inverted Y and diamond shaped towers both for single and double plane of cables and vertical
pylons for any kind of bridges (fig. 3c). From a purely structural point of view, the inverted Y
shape seems to be the most effective since it has been used in most of the major cable-stayed
bridges recently built.
Fig. 3.
Tower design: a) Barrios de Luna bridge (1983); b) Sama de Langreo bridge

(1986); c) Papaloapan bridge (1995); d) Bocairente.
As a matter of fact, the single vertical or slightly inclined pylon is a cable-stayed mast and this
is one of the most stable structures which can be designed; the tent is a representative
example. Spatial staying allows a perfect absorption of all the horizontal forces and it allows a
large variety of cable arrangements (fig. 1). Only one aspect has to be kept in mind: the
horizontal components of cable forces have to be balanced either at deck level or below to
avoid expensive foundations. Nevertheless this is not a very restrictive constraint since there
are many ways to comply with it.
As the spans become longer, it is not possible to think in terms of truly spatial cable
arrangements; it is still possible to maintain a certain degree of three-dimensionality by
anchoring the cables on both deck edges and on the vertical pole of an inverted Y pylon as it is
being done on the longest cable-stayed bridges.
But at this point we may also discover the beauty of the single pole in such bridges like the
Brotonne bridge. The pylon is usually subjected to moderate transverse forces and it can be
designed as a free standing tower in the transverse direction. This idea has brought some great
german bridges such as the well known Bonn Nord and Oberkassel. This solution may be
extended to a double plane of cables alternative as in the Nordbrcke, in the Kniebrcke or in
the more recent Queen Elizabeth II and Oresund bridges. Our Papaloapan bridge (fig. 3c, 4) is
also an example of a cable-stayed bridge with unconnected vertical pylons; it is important to
notice that this alternative was shown to be valid even in the presence of very significative
wind and seismic forces.
Fig. 4. Papaloapan bridge, Mexico, 1995

An aspect which we consider very important in the design of the towers is the way to fit cable
anchorages. If we want to enhance the design of the tower and its aesthetical relation with the
cables, an effort should be produced to hide the anchorages as much as possible and to avoid
any perturbation in the aesthetics of the tower. Many good ideas have been proposed along the
history of cable-stayed bridges; a good example is the internal gallery of the Faro and
Normandie bridges among many others.
In some of our bridges we have hidden the cable anchorages in vertical slots (fig. 5a, 5b). In
another case we designed a steel saddle which was embedded in the tower concrete (fig. 5b) to
serve as cable anchorage both for front and back stays. Another alternative for medium span
bridges consists in making the cable continuous through the pylon and establishing the
conection by means of an external tube (fig. 5c). In large bridges this is a minor problem since
the dimensions of anchorages become negligible as compared to the tower dimensions.
Nevertheless the structural and geometrical problems which arise when both cable fans cross
each other inside the pylon still exist and the previously mentioned alternatives are good
solutions to solve them.
From a purely aesthetic point of view, the design of the deck is not as important as the design
of the tower or the cable system. This is mainly an engineering problem. In this sense we only
consider today two deck types: the closed box and the slab, either alone or on top of
longitudinal edge girders, with or without transverse beams. The closed box is a somewhat
universal solution since it is used both for edge stayed and for center stayed decks; although
the shape depends on the staying system. It is probably a mandatory solution for long span
bridges where torsional deck stiffness is necessary (Normandie, Tatara, Skarnsundet). For
smaller spans, the slab is a very attractive solution since it allows a higher slenderness, as in
the Evripos bridge (1/477), a simpler construction (either cast-in-place or prefabricated) and a
good wind performance. The slab on edge girders combines the bending stiffness of the closed
box and the possibility to concentrate longitudinal compression stresses in the line of action of
deck anchorages. Although many of our first designs were closed boxes as in the Barrios de
Luna bridge, most of our recent projects are slabs (fig. 6); the closed box is still necessary in
those cases where the staying system requires a torsional contribution from the deck; these
cross sections show a general trend towards higher slenderness (the Ozama bridge deck is
heavier because it carries highway and railway traffic).
Fig. 5. Cable and tower interaction: a) Cross section of the Lrez bridge tower; b) Steel
saddle of the Papaloapan bridge; c) Cable connection to the tower in the Ozama
bridge.
Fig. 6. Deck cross section of a) Barrios de Luna bridge, b) Sama de Langreo bridge, c)
Papaloapan bridge, d) Ozama bridge, e) Bocairente bridge
Prefabricated decks can find very interesting applications in cable-stayed bridges.
Prefabrication has already been used in many bridges but at a small scale. We begin to think
in terms of large deck units which can be operated and assembled for short spans and for a
reduced number of cable stays. Our Bocairente bridge (fig. 6e, 7) is an example of a radical
use of prefabrication since the deck, the pylons and the struts are prefabricated. This also an
example of how cable system and pylon design are related to define a unique concept both in
terms of structural design and formal expression.
Another important development which can be observed in the last decades is the growing
applications of composite structures. The composite deck is specially well suited to the cablestayed bridge since it combines a good capacity for axial compression forces, lightness and
many possibilities for prefabrication and for quick erection. Annacis, Houghly and Ting-Kau
are good examples of such application. Another iteresting and completely different type of
composite construction for the deck is the one which was used in the Normandie bridge where
concrete was used in the side spans and in part of the main span. Applications of composite
construction to pylons are also increasing mainly in relation with the interface between cables
anchorages and tower concrete.
Another possiblity which has been proposed as a way to limit the compression forces to be
transmitted by the deck consists in designing the cable system in such a way that the deck
would be tensioned in the middle of the span. This is a very attractive possibility for long
spans but the erection process would be more difficult and cable forces would increase.
Fig. 7. Bocairente bridge, Spain, 1999.
The span length distribution is the origin of most differences between cable-stayed bridges.
The ration of side span length to main span length has important consequences in terms of
structural behaviour and aesthetics. Shorter side spans are usually necessary to increase global
stiffness and for long spans; they also give to the bridge a more powerful character. Longer
side spans create somewhat more equilibrated schemes; stiffness may be increased by means
of intermediate supports and deck ballasting in the side spans. Span distribution is generally
determined by topographical constraints and it gives to the bridge most of its identity.In this
respect, the single pylon symmetric bridge is a possibility which has also been considered with
very positive results (Isre, Alzette among others).
3.
Codes
There are no specific codes for cable-stayed bridges. We may qualify this situation as not
surprising since it also happens with other bridge types. Twenty years ago the ASCE
Recommendations [3] were a useful (although too simple) reference since the cable-stayed
technology was relatively new for most engineers.
The first topic which began to be worrying for engineers and administrations was long term
cable behaviour with respect to fatigue and corrosion. The PTI Recommendations [4] were an
effort to give an answer to this concern. These recommendations reflected the state of the art
cable technology which was reached in the late seventies and they still are a good reference,
specially with respect to fatigue problems [5].
The cable-stayed bridge field is still open to innovation and we find new ideas almost in any
new bridge which is presented to the engineering community. It is very difficult to write
standards for such structures and these standards would probably constrain new developments.
Bridge engineers and administrations tend to consider the cable-stayed bridge as any other
type of bridge; then loads and material specific design rules are the same as, for instance, in a
girder bridge. This decision is reasonable since the present methods of analysis allow a very
precise knowledge of stresses in any element of the bridge. In level by designing the concrete
elements in agreement with, say EC2, and by limiting the cable stress to 45% of tensile
strength, which is an old but generally admitted rule? What is the safety level against windinduced vibrations? Many questions are still unanswered. our usual mathematical models,
cables are modelled by means of truss elements and deck and pylons are modelled by means
of beam elements but all of them are treated in a similar way. The cable-stayed bridge is a
complex structure and our models give us information on stresses in any part of it.
Methods of structural analysis have also changed very significantly in the last two decades.
The finite element method has become available for every engineer and not only for linear
analysis but also for geometrical and material non-linear analysis. Partial factor methods based
on reliability studies have become widespread and we now may use codes which are based on
this approximation: CEB-FIP Model Code, Eurocodes, AASHTO LRFD Specifications.
But for the cable-stayed bridge several questions arise. Are all these standards really
applicable for a highly redundant structure? Do we get the same safety
A good approximation to the safety problems may come from the use of the possiblities of
structural analysis. Ultimate states can be modelled today and it has been done in real projects
[6,7]. This is a rigourous method to define the safety level of any structure although it usually
gives less conservative results that the application of standard design rules, specially for steel
and composite structures. This approximation is being used with any kind of bridge to solve
local and global problems. In the case of the specific problems of cable-stayed bridges, global
ultimate state can be investigated in this way. As an example of such applications, the
transverse stability of single pole pylons and decks may be studied on the basis of a fully nonlinear model (geometrical and material) with a higher accuracy and reliability than by trying to
define the bifurcation point through a linear model based on the geometric stiffness matrix
[8].
Although the cable-stayed bridge is highly redundant, non-linear analyses as well as scale
model testing [9] show that with the present design methods the cables would yield first in the
ultimate state. The slender deck would reach its ultimate state soon afterwards. Nevertheless
all these analyses show that the global safety factor is greater than 2, which is coherent with
the design rules which are being used for the cables. This figure may seem too high but it
somehow takes into account the uncertainty about the long term behaviour of the cables.
Another point of raising concern is the possibility of a local buckling problem in the deck
which could trigger a global instability of the whole bridge (obviously in steel or composite
decks). Such possiblity should be considered by a modification of the corresponding partial
factors.
The cable-stayed bridge has also some specific properties which should be considered in the
codes. One of them is the dead load control. The construction method which is mostly used
includes a very precise control of cable stresses and dead loads; a small error in the deck dead
load will be detected immediately through the cantilever vertical displacements and the top of
pylon horizontal displacements. In such circumstances, the partial factor to be applied to dead
load should be close to unity, as it has been done for the Oresund project.
In relation also with the balanced cantilever method of construction, some standards consider
an unbalanced live load to be applied during construction; sometimes this unbalanced load is a
small fraction, say 5% or even 2%, of total dead load. This unbalanced load may be caused by
temporary construction loads but this possibility can be avoided through a tight control on site
and any unbalance would be detected through vertical displacements of the deck. Anyway
some stiffness has to be provided against this effect to take into account wind forces; for such
effect, a buffeting analysis or a reduced vertical load could solve the problem. The
AASHTO/ASBI Specifications for Segmental Bridges are a good reference although not
specifically written for cable-stayed bridges [10]. Reliability analyses are necessary at this
point to define well founded values for these unbalanced loads.
Cable-stayed bridge design and analysis is quite well known by now and, in spite of this fact,
it is still an open field with almost no standards. It is necessary to build an international effort
to reflect this accumulated knowledge in a code which might serve as a worldwide reference.
4.
Cable technology
The reliability increase of the cables is one of the factors which may be responsible for the
present day development of cable-stayed bridges. All the cable fabricators proposed in the
seventies anchoring systems to avoid transmitting live loads through jaws. They obtained a
good success by filling the anchorage with some propietary compound which allows direct
transmission of variable loads to the structure of the anchorage.
The possible corrosion problems were dealed with two alternative solutions: either by using
locked-coil galvanized cables or by protecting the cables with a polyethilene duct filled with
mortar. The second solution has been giving good results as we had the opportunity to test the
cables of the Las Glorias footbridge when it was dismantled 20 years after construction to
move it to another location [11]. We are also presently involved in the process of changing
one locked-coil cable of the Ebro bridge to analyze it and to define its present condition.
Today other alternatives exist: stainless steel tubes, individually coated strands, two and three
level protection systems. Individual protection of the strands has a major advantage since it
avoids local damage when handling the cables and increases fatigue resistance of steel. We
can say today that long term behaviour of the cables is no more a source of concern provided
we use the right alternatives. All the problems suffered by the cables have not only brought an
important technological lap from the fabricators; today bridge designers have to leave open
the possibility of changing the cables. When long term behaviour of the cables will be known
more precisely, changing the cables will be either avoided or considered as a normal
maintenance operation to be done at very long time intervals.
A problem which is still not completely solved comes from cable vibrations. The use of
dampers (Brotonne, Sunshine Skyway) is opposed to the design of a net of transverse wires
(Normandie) although the design rules for such transverse system are still not clear; cable
fabricators are beginning to develop new connection systems which make these complex cable
systems easier to build and more reliable. In any case none of these is probably the optimum
solution; research on aerodynamic devices (protuberances) seems more promising since it tries
to fight the cause of the problem and not its effects.
5.
New applications
As in any active field, new applications arise every day. Many of them are architectural.
Others are new engineering solutions. Both of them are interesting. As engineers, we should
understand that the public interest on cable-stayed bridges is partly due to the input of
architects. In any case we will focus here on the engineering solutions.
The cable-staying concept may be undestood as a kind of external prestressing. We have
recently built at Osormort (Spain) a continuous concrete girder bridge with eleven, 40m long,
spans; each span is stayed by creating a sort of vertical pylon in the center of the span,
underneath the deck (fig. 8). The stays are anchored at the deck, close to the piers. This
solution allows a very slender deck design (1/25) and it is very effective to support the dead
load, which is an important part of total applied load. As cable stress is almost constant,
standard prestressing steel cables are used instead of stay cables.
A new field for application of cable-stayed bridges may be perceived from this example.
Extradosed prestressing generates a kind of bridges which are formally similar to cable-stayed
bridges but with some important structural differences. The most important comes from the
fact that cables are not very effective in supporting live loads and, consequently, stress
variations are small; this fact allows higher working stress levels [13]. This type of application
seems to be very promising for medium range continuous girders (120 to 150 m span) where
the cable system may be understood mainly as a way to create a variable depth cross-section;
these cables would only help in supporting the dead load but they would allow a significant
reduction in the girder depth and weight. The Bocairente bridge which was shown before (fig.
7) may be an example of such application.
There is still an open field for innovation between the girder bridges and the cable-stayed
bridges. We are presently building a bridge with a 180m long main span(fig. 9) where the
towers are lower than usual (h/l=1/6.6). This is not an extradosed bridge; the reason for such a
design is mainly a search for formal compatibility with an older suspension bridge which is
located nearby.
Fig. 8. Osormort bridge, Spain, 1995
Fig. 9. Ozama bridge, Dominican Republique, 1999

Among some of the projects we are working on, we find interesting a proposal for a multispan cable-stayed curved bridge. Transverse cable forces are balanced by means of a stay
connecting the pylon top to the outside edge of the deck. But, to enhance the curved character
of the bridge, we have leaned the tower outwards (fig. 10). This tower inclination slightly
helps in balancing cable forces but we find that it offers a new perspective of the cable system
and its relation with traffic on the deck.
Finally every designer has many proposals which were formulated at some time and were not
accepted or did not win the design competition. It happened because they were not costeffective or because they were judged too risky or, simply, because they were not the best.
Nevertheless many of them are interesting and their only defect is not being born. We could
find real treasures of imagination in many of these unknown projects and they will probably
arise as new cable-stayed bridges will be built.
Fig. 10. Proposal for a curved multi-span cable-stayed bridge.
6.
Conclusions
Cable-stayed bridge design has already arrived to what we could call a classic period. General
design parameters and techniques are well established. There is still room for innovation and
this fact is very challenging for engineers. This type of bridges has attracted general public
interest and as new bridge design and construction is partly becoming a social or political
event, we may be entering into a baroque period, specially for the short span range.
Contributions from architects may seem disturbing for many engineers but they should be
looked with interest since they bring a new insight into a problem which we, as engineers,
tend to see mainly as a statics and optimization problem. Very often the client, a public
administration (and, by this way, the society) is not asking for the optimum solution in
engineering terms, but just for good design. We should try to fulfill these demands. At this
point it is very important to define some rules for design to keep cable-stayed bridges as safe
as they have been till now.
References
[1]. Fernndez Troyano L, Manterola J. & M.A.Astiz, ", The Inclined Towers of the Ebro
and Lrez Bridges, Structural Engineering International, 4/98, 258-260, 1998
[2]. Fernndez Troyano L. & Manterola J., Spatial Cable-Stayed Bridges, Spatial
Structures:Heritage, Present and Future, de. G.C.Giuliani, SGE, Milan, pp. 1019-1026
(1995)
[3] ASCE, Tentative Recommendations for Cable-Stayed Bridge Structures, Proc. ASCE,
Journal of the Structural Division, 103, ST5, 929-959, 1977
[4] PTI, Recommendations for Stay Cable Design, Testing and Installation, Post-Tensioning
Institute, 1990
[5] Elices M, Llorca J. & Astiz M.A., Fatigue of steels for concrete reinforcement and
cables, in Handbook of Fatigue Crack Propagation in Metallic Structures, ed. A.
Carpinteri, Elsevier, 1994.
[6] Kovacs I., Svensson H.S. & Jordet E., "Analytical Aerodynamic Investigation of CableStayed Helgeland Bridge", Journal of Structural Engineering ASCE, 118, 147-168, 1992.
[7] Biwer R., Crmer J.M., Hubert F. & de Ville de Goyet V., Cable-Stayed Bridge upon
Alzette, . ", Proc. Conf. Cable-Stayed and Suspension Bridges, Deauville, 413-420, 1994
[8] Ren W.X., "Ultimate Behavior of Long-Span Cable-Stayed Bridges", Journal of Bridge
Engineering, ASCE, 4,No.1, 30-37, 1999
[9] Walter R., Houriet B., Isler W. & Moa P., "Cable-Stayed Bridges", Thomas Telford, 1988
[10] AASHTO/ASBI, "Guide Specifications for Design and Canstruction of Segmental
Bridges, American Segmental Bridge Institute, 1998
[11] Fernndez Troyano L, Manterola J. & Astiz M.A., "Footbridge of the Glorias Catalanas,
Barcelona", FIP Notes, 1996/2, 13-14, 1996
[12] Matsumoto M, Hikami Y. & Kitazawa M., "Cable Vibration and its
Aerodynamic/Mechanical Control", Proc. Conf. Cable-Stayed and Suspension Bridges,
Deauville, 439-452, 1994
[13] Ogawa A., Matsuda T. & Kasuga A., The Tsukuhara Extradosed Bridge near Kobe,
Structural Engineering International, 3/98, 172-173, 1998
Aerodynamic and Structural Dynamic Control System

of Cable-stayed Bridge for Wind Induced Vibration
Masao MIYAZAKI
Gen. Mgr, Steel Struct.Group
Sumitomo Heavy Industries Co., Ltd.
Japan
Masao Miyazaki, born 1948 received

his Dr. Eng. from the University of
Tokyo in 1991.
Since 1976 he has been acting as a
researcher in the aerodynamic design
field of bridges and steel structures.
He is head of Bridge Eng. Division of.
Member of JSCE, JSSC and JSWE.
Summary
This paper reports on the wind endurance and vibration control measures for the tower and
cables of a 325 meter-long continuous three-span steel cable-stayed bridge with a central span of
175 meters. Deflectors were installed on the tower where horizontal members were omitted due
to design requirements, and aerodynamic vibration control measures by the use of U-stripes were
taken for the cables that were likely to be subject to rain vibration. It was so arranged that both
measures should be maintenance-free.
1. Introduction
This is a continuous three-span steel cable-stayed bridge with a central span of 175 m and a side
span of 75 m. Since the location is a scenic spot in Seto National Park, attentive consideration
was given from the viewpoint of landscape. And finally a cable-stayed bridge based on the image
of a bow was adopted. The towers and side spans were constructed in large blocks by the use of a
floating crane while the center span was installed by the cantilever method.
Figure 1 General Drawing of the Bridge
2. Vibration Control Measures for the Tower

The stability of the tower against winds both during construction and after completion was
examined through wind tunnel testing. It is often considered that the tower is relatively free of
problems in terms of wind endurance after completion due to this bridge size. However, it has
only recently been learned that intentional omission of horizontal members in this type of bridge
will considerably reduce wind endurance.
Figure 2 Shapes of Tower and Deflector

Based on the results (Figure 3) of the wind tunnel testing, it was predicted that the tower of the
completed bridge would be subject to galloping due to wind in the direction of the bridge axis,
i.e., at right angles to the tower, when the wind velocity exceeded about 18 m/s. This is a socalled destructive vibration, which diverges quickly. When structural damping was increased, the
responses moved into vortex-induced vibration in the low wind speed range and galloping in the
higher one. When the structural damping increased to 4% in logarithmic terms, both types of
vibration were almost completely suppressed at wind velocities below the design wind speed. As
vibration control measures against galloping, changes in the shape of the tower or in the crosssectional shape of tower poles or installation of dampers are proposed. It was decided to install
deflectors at four corners of the tower out of consideration for landscape design, influence on the
substructure, and ease of future maintenance. The shape of the tower legs and optimum intervals
between them were determined by means of wind tunnel testing. In the experiments, the weights
of the girders and cables were taken into account as factors affecting vibration, in addition to the
weight of the main tower. The results (Figure 4) of the experiments showed that deflectors
installed at a distance of approximately 40% the tower height from the top will ensure the
necessary wind endurance for the completed bridge.
Figure 3 Vibration of Completed Bridge (without vibration control measures)
Figure 4 Vibration of Completed Bridge (with vibration control measures)

The cantilever method was employed for the construction of the main girder in the center span of
this bridge. When this construction method is used, the typhoon season is generally avoided if
wind endurance comes into question. It is unavoidable, however, that the tower should stand in
isolation for a certain period of time. In this state, it is expected that galloping, i.e., in-plane
vibration of the tower at right angles to the bridge axis, will be generated under the influence of
the wind in the direction of the bridge axis. This situation is avoided in advance by the
installation of deflectors, which are usually installed on completed bridges. On the other hand, it
was predicted that the wind at right angles to the bridge axis would generate vortex-induced
vibration due to bending motion in the direction of the bridge axis at around the wind velocity of
10.5 m/sec to 12.6 m/sec, creating stresses in excess of the resisting moment at the foot of the
tower. However, this vibration is suppressed when two lower side cables are installed, increasing
the mass that works against the aerodynamic force to damp the vibration associated with
aerodynamic instability.
Therefore, two small passive tuned mass dampers (TMD) were installed at each tower top to
control the vibration while the tower stands in isolation during construction. In designing the
TMD, the allowable amplitude of the tower, defined in terms of acceleration tolerance,
was assumed to be 50 gal during work, and 300 gal in a non-working condition
according to the standards adopted by Honshu-Shikoku Bridge Authority. The variables
such as damping forces necessary for the design of the dampers were determined by wind tunnel
testing. The damping effect was verified through experiments during construction.
Figure 5 Results of Wind Tunnel Test (Sc-A Diagram)
Figure 6 TMD for Tower during Construction
Figure 7 Deflector for the Tower of the Bridge completed
3. Vibration Control Measures for Cables

3.1 Aerodynamic Vibration of Cables
It is well known that vibration of cables on cable-stayed bridges poses problems. Vibrations
generated on cables are classified as follows according to the mechanisms of their
generation.(1)Vortex-induced vibration, (2)Wake galloping, (3)Rain vibration
Vortex-induced vibration, which is induced by trailing vortices of cables, and wake galloping,
which is generated when downstream cables are placed in the wake flow of upstream cables
arranged in parallel, have been known for a relatively long time. Of these, wake galloping was
studied only in relation to power transmission lines where distances between cables were as large
as 10 to 20 times the cable diameters [2] [3]. The wake galloping that occurs on cable-stayed
bridges is characterized by relatively small cable-to-cable distances of less than 6 times the cable
diameter. On the other hand, rain vibration, which is associated with aerodynamic instability, was
first recognized during the construction of major cable-stayed bridges in Japan such as
Meikonishi Bridge and Iwaguro-Jima Bridge and later observed on cables of many cable-stayed
bridges [4] [5]. Such aerodynamic vibrations are generally caused by strong winds that involve
rain. Although vibration of a large amplitude was also observed on Higashi-Kobe bridge in high
winds without rain [6], it is distinguished from rain vibration.
Regarding the vibration of cable-stayed bridges associated with aerodynamic instability, it is
possible to devise control measures by clarifying the mechanism of their generation. Thus,
vortex-induced vibrations may basically be controlled by suppressing the generation of trailing
vortices or by disturbing the simultaneity of flow separation. However, the mechanisms of wake
galloping and rain vibration (including similar vibration generated by winds without rain) still
remain to be clarified.
3.2 Rain Vibration
Rain vibration is a type of aerodynamically induced vibration, whose occurrence has recently
been confirmed on cable-stayed bridge cables coated with polyethylene tubes. Typically, it is
generated in high winds with rain. Its characteristics appear to differ from those of vortexinduced vibration and wake galloping. The actual bridge where rain vibration was first observed
in Japan was Meikonishi Bridge, a 758 m long cable-stayed bridge with a central span of 405 m
[7]. The cables on this bridge weighed 5.1 kg/m after grouting and their logarithmic decrement
of vibration was approximately 1%. The cables were 125 to 165 mm in diameter with typical
cable diameter being 140 mm.
Formerly, it was thought that several conditions must be satisfied for vibration to occur and that
rain vibration would occur only under extremely limited circumstances.
The conditions of vibration occurrence that were known then include:
(1) Vibration is generated on cables that have a downhill grade in the wind direction, its
amplitude reaching a maximum when the wind blows at an angle of 45 degrees to the
surfaces of the cables.
(2) For occurrence of vibration, it is indispensable that rivulets (small stream) of rain should be
formed flowing down the top and bottom surfaces of the cables.
(3) The vibration is an in-plane vibration (oscillations in the inner direction of the cables) in a
relatively lower mode and has a larger amplitude than vortex-induced vibration.
(4) The rivulets vibrate together with the cables.
(5) The formation of rivulets is affected greatly by the conditions of cable surfaces, wind
velocity, wind direction, and rainfall.
Since the formation of rivulets, especially on the top surfaces of cables, was indispensable for
generation of vibration, it was thought that vibration similar to galloping was generated due to
asymmetry of the burble point and relative motion of the elevation angle when the flow separates
from the rivulets on the top and bottom surfaces.
Later, along with the progress of studies on the characteristics of rain vibration and on the
mechanism of vibration generation, vibration control measures changed form structural dynamic
measures (which suppress vibration forcefully by dampers and the like) to aerodynamic measures
(which try to eliminate the source of vibration). The first such measure was the parallel
protrusion method employed for Higashi-Kobe Bridge. However, it was replaced by hydraulic
dampers and viscous shear dampers because of its high cost of fabrication and difficulty of
construction. Rubber with high damping capacity came into use to avoid installation of awkwardlooking dampers on cable-stayed bridges, which attach importance to landscape. Even in this
case, the fact that physical properties of rubber depend on temperature presented a problem.
However, this problem has largely been solved.
3.3 Vibration Control Measures for Cables the Bridge
Aerodynamic vibration control measures were used for the top-layer cables, which are most
liable to rain vibration because of their length and diameter.
Regarding other cables, rubber dampers with high damping capacity were installed at the tips of
anchor tubes of the cables in the middle three layers where there remains possibility of rain
vibration. In a vibration experiment, = 0.02 to 0.03 was achieved by the rubber with
high damping capacity.
Figure 8 Aerodynamic Vibration Control Measure for Cables

One of the distinguishing features of rain vibration is the formation of rivulets on the top and
bottom surfaces of cables. The existence of rivulets appear to be closely related to the generation
of vibration and their effects can be examined through reproduction of rain vibration in a wind
tunnel by the use of the equipment shown in Figure 9. Figure 10 shows the results of an
experiment and formation of rivulets when water was applied to the top and bottom surfaces of
cables. The cables used were 3 m long and 150 mm in diameter. To make it easier to check the
effect of Scruton numbers, cables lighter than actual ones were used. It can be seen from the
Figure 10 that rain vibration started to occur at wind velocity of around 9 m/s. The vibration
increased its amplitude sharply with increase in the wind velocity. However, observation was
made only up to a wind velocity of 20 m/s. And no properties were observed from that point on.
This resulted in a serious lack of information, which fact was learned later. For clarification of
phenomena encountered for the first time, sufficient consideration is necessary. As shown in the
figure, rivulets form only on the bottom surfaces of cables in low winds and they form both
surfaces of cables only when the wind velocity is sufficiently high. This indicates that rivulets
form in such a way as to counterbalance the wind pressure.
Figure 9 Cable Model Installed in the Wind Tunnel
Figure 10 Relationship between Wind Speed and Formation of Rivulets

While one might be tempted to conclude that formation of rivulets on top and bottom surfaces of
cables is the necessary and sufficient condition for the generation of rain vibration, actually the
phenomena of aerodynamic instability are often very complicated. So is rain vibration. An
example of aerodynamic force generated is shown in Figure 11. Under the circumstances where
rain vibration is generated, the aerodynamic force is 2% at the most in terms of the logarithmic
decrement although it has a small amplitude. This is important when considering the vibration
control measures for cables as described later.
Figure 11 Aerodynamic Force in Rain

Vibration
Figure 12 Drag Coefficient and Reynolds Number
3.4 Mechanism of Vibration Damping [8] [9]

In this method, the polyethylene tubes that coat cables are provided with V-groove stripes (Ustripes). The damping effect of U-stripes on the vibration associated with aerodynamic instability
is understood as follows.
(1) Damping of vortex-induced vibration: U-stripes act as surface roughness to raise an apparent
Reynolds number up to the supercritical Reynolds-number region. As a result, the Karman
vortices that existed in the wake flow area in the subcritical Reynolds-number region
disappear, and thus generation of vortex-induced vibration is restrained.
(2) Damping of rain vibration: The effect of surface roughness raises an apparent Reynolds
number up to the supercritical Reynolds-number region, shifting the burble point further
downwind than the subcritical Reynolds-number region. Consequently, pressure
distribution on the surface changes due to reattachment, preventing the formation of
rivulets. Apart from this, the U-stripe grooves forcefully guide the water flowing down the
cable surfaces, preventing rivulets from forming in particular places.
4. Conclusion
This paper reported on the stability of the cable-stayed bridge against winds. The vibration
control measures designed for the bridge based on the results of wind tunnel testing are
summarized as follows.
(1) For the tower, deflectors were installed to protect the completed bridge against galloping and
tuned mass dampers (TMD) were used to restrain vortex-induced vibration during construction.
(2) For the cables, U-stripes were provided in the surfaces of the polyethylene tubes on the toplayer cables as countermeasures against rain vibration while rubber with high damping capacity
was installed on the middle-layer cables.

No vibration was observed on the bridge under construction or the completed bridge.
References
[1] "Design, fabrication, and construction of Yuge Ohashi Bridge." Sumitomo Heavy Industries
Technical Report, 1995. 11
[2] Simpson, A "Stability of subconductors of smooth circular cross section." Proc. The
Institution of Electrical Engineers, Vol. 117, pp. 741 - 750, 1970
[3] Simpson, A. and T. V. Lawson "Oscillations of twin power transmission lines." Proc. Wing
Effects on Buildings and Structures, Vol. 2, 1968
[4] National Land Development Technology Center "Research report on wind endurance of
cable-stayed bridges." 1989. 2
[5] Civil Engineering Research Center "Reports on vibration control measures for long-span
cable-stayed bridges." 1993. 3
[6] Matumoto, M., Y. Hikami and M. Kitazawa "Cable vibration and its aerodynamic/mechanical
control." in Cable-stayed and suspension bridge (Deauville), 1994. 10
[7] Higami, S. "Rain vibration of cable-stayed bridge cables." Journal of Wind Engineering , No.
27, 1986
[8] Miyazaki, M. "A study on distribution of wind pressure acting on bridge structure and wind
endurance." dissertation at Tokyo University, 1990. 12
[9] Miyazaki, M. "Aerodynamic control method for vibration of bridge cables." 1st Inter. Con.
Struct. Control, 1995.7
Seismic Design for The Cape Girardeau Cable-Stayed Bridge

Steven T. HAGUE
Prof. Eng.
HNTB Corporation
Kansas City, MO, USA
Steve Hague was born in 1959,

earned his Bachelor and Masters
degrees at Texas A&M University
and is a licensed Professional Engineer
Summary
In 1927, the Missouri Highway Department, now the Missouri Department of Transportation,
constructed a 1450-meter crossing of the Mississippi River near Cape Girardeau, Missouri.
Now this two lane bridge is scheduled for replacement with a new four lane cable-stayed
structure.
The proposed structure has an overall length of 1206 meters, and was designed in both
concrete and steel alternatives for competitive bidding purposes. The main span unit is
comprised of a three-span, 636-meter cable-stayed unit with a 350-meter navigation span.
The approaches are of typical steel plate girder construction.
Figure 1
This new bridge is located within the New Madrid Seismic Zone, the location of three of the
largest seismic events to occur within the interior of a tectonic plate and the site of the most
violent series of earthquakes ever recorded, and is a candidate to experience a significant
earthquake within its design life. Although not actually recorded, studies of the available data
indicate that the events of the winter of 1811-1812 had surface wave magnitudes (Ms) of
about 8.6, 8.4, and 8.7 and it is suggested that the recurrence interval of magnitude 8
earthquakes in this region is approximately 550 to 1200 years.
In addition to the probability of a significant earthquake, the geology of the site may be
characterized as having deep, liquifiable soils which are subject to frequent flooding and the
potential for extensive scour. These site conditions, combined with the significance of the
design earthquake event, generated some unique design challenges.
The design issues presented will demonstrate the methodology used to consider the
significance of the design earthquake, the site specific ground motion, and the effect of
liquefaction and lateral spreading forces on this structure.
Introduction
Although often overshadowed by the seismicity of the American west coast, the New Madrid,
Missouri region is a very real and significant seismic threat to the midwestern region of the
United States. The general public, even if aware that earthquakes often occur in the central
United States, does not readily admit the potential destruction that would follow a major
event. Fortunately, state departments of transportation, the U.S. Federal Highway
Administration (FHWA), and many local building code officials are acutely aware of the risk
and the damage - in terms of loss of life as well as economic losses - which would follow
even a moderate event in the New Madrid region.
New structures are now designed in accordance with current seismic guidelines developed
from observation of structural behavior during earthquakes and millions of dollars worth of
research. However, both the research funding and empirical evaluations focus on
conventional structures. In this paper, we will look at a bridge structure which is somewhat
outside the norm for earthquake design and the methods used to ensure that the structure is as
capable of resisting seismic loading as our current state of knowledge will permit while
maintaining the reliability that the travelling public has come to expect from its infrastructure.
The Project
The relocation of Missouri Route 74 - Illinois Route 146 will cross the river approximately
200 meters downstream of the existing bridge at an angle of approximately 15 degrees to the
direction of flow. The proposed structure has an overall length of 1206 meters and is
comprised of a three-span, 636-meter steel/concrete composite cable-stayed unit and 570
meters of conventional steel plate girder approach structure. The main span unit will be a 4lane, symmetrical cable-stayed unit supported by two planes of cables 28 meters apart. The
cables are attached to the steel edge girders at a uniform spacing of approximately 10 700
millimeters. The Illinois approach structure has 11, 52-meter steel plate girder spans
supported on concrete piers and founded on deep, large diameter drilled shafts.
The city of Cape Girardeau, Missouri is located within the New Madrid Seismic Zone, the
location of the most violent series of seismic events ever recorded and is a candidate to
experience a significant earthquake within the not so distant future. Studies of the available
data indicate that the three most significant events of the winter of 1811-1812 had surface
wave magnitudes (Ms) of about 8.6, 8.4, and 8.7. It has been suggested that the recurrence
interval of magnitude 8 quakes in this region is between 550 and 1200 years.
P ro je c t L o c a tio n
Figure 2
In addition to the probability of a significant earthquake, the geology of the site may be
characterized as having deep, liquifiable soils which are subject to frequent flooding and the
potential for extensive scour. These site conditions, combined with the significance of the
design earthquake event, generated some unique design challenges for both the structural and
geotechnical engineers.
The Site
The Mississippi River is one of the world's great rivers. Flowing with commerce; it provides
a major transportation corridor for inland barge traffic and generally contributes to the
economy of the entire midwest. The Mississippi River flows for some 3600 kilometers,
draining approximately 40 percent of the continental United States. At St. Louis, Missouri,
the Mississippi is joined by the Missouri River, the longest river in the U.S., which adds
approximately 1800 m3/s to the average discharge of the river. Near Cape Girardeau,
Missouri, the Mississippi River drains more than 1 850 000 square kilometers spread over
twelve states and three Canadian provinces. At the site, the channel is 600 meters in width
with a 1100-meter wide floodplain. The floodplain is bounded by high bluffs on the west and
controlled by a levee on the east.
The river is also a significant route for inland shipping with some 75 million tonnes of cargo
shipped through the region annually.[1] This cargo is transported in barge tows up to 370
meters in length; typically comprised of up to a dozen barges, tied three across, and powered
by a single tug. Therefore, the navigation requirements for this location are critical as
demonstrated by the U.S. Coast Guard requirement for a channel width of 250 meters normal
to the flow of the river. By establishing the navigation span at 350 meters it was possible to
achieve considerable savings in the foundations by not having to construct a major foundation
in the deepest section of the Mississippi River channel. In addition, a privately owned drydock facility is located immediately adjacent to the project right-of-way, downstream of the
bridge and an added benefit of the longer span was to continue to allow access to the drydock.
Scour
At Cape Girardeau, the 5-year flood discharge is over 17 000 m3/s at a mean velocity of about
2 m/s. However, at times the discharge may be as much as 32 300 m3/s with an average
channel velocity approaching 3 m/s. Several hydraulic models were developed using the U.S.
Army Corps of Engineers HEC-2 computer program. The models utilized the velocity
distribution and normal bridge routines to generate the water surface elevations, flow depth,
and stream velocity. The FHWA publication "Evaluating Scour at Bridges," HEC-18, was
used to predict scour for both the 100-year and 500-year flood frequencies.
The scour analysis indicated that the total scour depth, defined for this project as the sum of
the effects of long-term scour (aggradation/degradation) and local scour, may be as much as
15 meters near the main channel and up to 8 meters near the Illinois levee. Since the new
bridge will span from the east levee, near the existing bridge abutment, to well beyond the
west bank, it was determined that the contribution of contraction scour to the total scour value
would be negligible.
Site Geology
Geologically, the project is located on the eastern edge of the Ozark uplift and the
southwestern boundary of the Illinois basin. The bedrock formations at the site are mostly
limestone, with minor amounts of shale, upon which the new bridge is to be founded. The
limestone is overlain by a granular, liquifiable material to a depth of approximately 25 to 30
meters. Although the area is heavily faulted, the faults are considered to be inactive.
The bridge is located within approximately 80 kilometers of New Madrid where there is a
significant probability for a devastating earthquake within the next few years. During the
series of events of 1811 and 1812, there were more than 200 moderate to large earthquakes
and some 2000 total events with well documented evidence of liquefaction and having effects
being felt as far away as Washington, D.C. [2]
Design Criteria
The Bill Emerson Memorial Bridge represents a significant investment of public funds, and
as such, required that the Missouri Department of Transportation (MoDOT) develop a design
criteria to protect the travelling public, both roadway and navigation traffic, and their
investment against those external events which could reasonably be expected to occur.
Additionally, the criteria recognizes the importance of the structure to the economic wellbeing of the region as well as the difficulty of certain types of post-seismic repair. Therefore
the general criteria for the bridge were established as follows:
provide for six lanes of AASHTO HS20-44 (modified) live load
provide minimum navigation clearances of 250 meters normal to the flow of the
river and 18 meters above the 2 percent flowline
protect against barge impact based upon a 365-meter tow, travelling at 5 meters
per second at the 2 percent flowline elevation
design for the 100-year scour condition, with only one-half of the anticipated scour
during earthquake design
design for earthquake forces in accordance with the "Geotechnical Seismic
Evaluation" report
provide for operation of the structure following the design event
resist seismic forces in the tower piers within the elastic range
Design Earthquake
The New Madrid region has been the most seismically active region of central and eastern
North America. The events of the winter of 1811-1812 are well documented and have been
the subject of a significant volume of research over the years. Nuttli's study [3] of the damage
and felt effects of these events indicates that the surface wave magnitudes were on the order
of 8.5. Other studies have reached similar conclusions regarding the magnitude of that series
of events.
Between 1813 and 1990 over 23 earthquakes having magnitudes of 4.5 or greater were
documented in the New Madrid area. [4] By using a map of acceleration contours having a
90 percent probability of not being exceeded in 250 years, it can be shown that the peak rock
acceleration at the site is approximately 0.36g. Based on input from the project design team,
MoDOT selected this as the design event and, considering that Ms 8 or larger events are
anticipated every 550 to 1200 years, [5] the design earthquake is essentially a repeat of the
1811 and 1812 events.
Woodward-Clyde Consultants, the project seismic subconsultant, then developed response
spectra for the site based on published data for the central and eastern U.S. with the peak rock
acceleration of 0.36g. Exploratory borings were made and shear wave and compression wave
velocity tests conducted. The results of these investigations were used to develop three
separate spectrum compatible site specific acceleration time histories for the seismic analysis
of the bridge. These time histories were derived from the 1985 Michoacn, Mexico (Mexico
City) earthquake and the Val Pariso and Pichulema records of the 1985 earthquake in Chile.
These records were selected for their epicentral distances and the magnitude of the recorded
event; however, these events do not necessarily represent a large earthquake on a continental
intraplate source.
The time history files provided were in the form of accelerations, given as a fraction of
gravity, over a period of about sixty seconds. These time histories were established for two
orthogonal directions with consideration given to the directional uncertainty of the design
event. These files also included the effect of spatial incoherency and the phased effect of the
ground motion due to the piers being located over such a long expanse. Although the vertical
component of the design earthquake was not directly considered it was included in the model
by applying a percentage of one of the horizontal accelerations in a vertical direction
simultaneously with the separate horizontal components.
In addition to the rock accelerations, Woodward-Clyde generated surface spectra which
included the soil amplification effects of the soils along the Illinois bank of the Mississippi
River.
Liquefaction and Lateral Spreading

As previously noted, the Illinois side of the site consists of some 30 meters of alluvium;
primarily loose to medium-dense sands. Both the comprehensive geotechnical investigation
and the investigation conducted to evaluate specific geological conditions related to the
seismic evaluation for the bridge revealed Standard Penetration Test (SPT) blow counts as
low as 4 with only thin seams of material having blow counts above 16 in the upper 25 to 30
meters of alluvium. With these poor soil conditions and the high level of shaking which is
expected to occur during the design earthquake, widespread liquefaction is anticipated to a
depth of up to 25 meters below grade.
In addition to the liquefaction, lateral spreading is also anticipated. The gently sloping banks,
especially between the main channel and the levee on the Illinois shore, could flow as much
as 3 meters toward the channel while in a liquefied state. Clearly this will produce large
horizontal forces on the bridge foundations at a time when there is little lateral support.
Preliminary Design
As with all projects, the design process for a bridge of this magnitude is an iterative one,
requiring multiple revisions and redesign of many components along the way. Because of the
location of the bridge, and the potential for a significant earthquake, the design team
attempted to minimize backtracking by working with the seismic subconsultant at the earliest
stages. After the development of the basic structural concept, input from the seismic
engineers was necessary to confirm the preliminary design and to prepare the models for final
design analyses.
It was noted that liquefaction presents little problem for the cable-stayed unit since the three
supporting piers are founded on huge footings keyed into rock; however, the approach spans
are considerably different. As noted earlier, these foundations are located in an area with very
deep, highly liquefiable soils. When combined, the liquefaction and the depth of anticipated
scour eliminated spread footing type foundations from consideration. After extensive studies
of various soil improvement techniques, it was determined that any soil improvement would
be ineffective due to repeated degradation and aggradation of the channel. Thus, the early
input from the geotechnical engineers permitted the elimination of both spread footings and
driven steel piles as viable foundation alternatives and led to the selection of large diameter
drilled shafts socketed into rock.
Design Methods
It was obvious at the earliest stages of design that the governing load case would be a
combination which included earthquake forces. Other combinations, those including scour
and barge impact, were also considered significant but not viewed as potentially governing
the design of the bridge. Due to the large number of seismic related load combinations, those
combinations with and without scour and those with and without liquefaction, it was
determined that the design would be for one event only, with a final force check with the two
secondary events.
The computer program used for the analysis of the structure, T187, was developed by HNTB
specifically for the design and analysis of segmental and cable-stayed bridges. Within its
dynamics module, the program performs a linear time history analysis based upon support
accelerations. Using the Wilson-theta method, the program computes and stores velocities
and displacements for each degree of freedom at each time step, thereby allowing the user to
stop and restart the dynamic analysis and to modify the structure at any predetermined point
within a dynamics run. The program also allows the user to accelerate each supported degree
of freedom with a different transient load and to begin the acceleration at different times.
Based upon user input of estimated damping percentages, the program calculates the
appropriate Rayleigh damping coefficients, assuming damping to be proportional to mass and
stiffness, and applies these coefficients to the mass and stiffness matrices during the run. The
program will then compute displacements, forces and reactions for each time step and provide
the user with his choice of maximum or minimum values for a given list of members. Since
the joint displacements are saved by time step, the user may elect to open the dynamic
displacement file at a later date in order to compute additional results.
For design, it was determined that this bridge is an essential structure, thereby requiring that
the bridge remain serviceable following a moderate earthquake, and sustain only minor
damage as a result of the maximum credible event. By minor damage, we intend that the
structure would remain operational although expansion joints, bearings and other easily
repaired components could sustain some damage. And since there is very little data regarding
the confinement of large, hollow concrete sections, or the performance of such sections
beyond the elastic range, the tower piers were designed to remain elastic thorughout the
design earthquake. Additionally, the approach span piers are sufficiently large that they
remain elastic under all load conditions.
Cable-Stayed Spans
The initial steps in the analysis were to confirm that the acceleration time history files
provided by Woodward-Clyde Consultants were being read correctly by the analysis program.
This included a preliminary run which computed maximum relative joint displacements
between the accelerated supported joints and generated a plot of the relative displacements
throughout the event. These were compared to, and corresponded well with, the 10
centimeter maximum relative displacement and the continuous record of relative
displacements predicted by the seismic subconsultant. Computed absolute displacements at
the supports were also compared to the predicted values, and again the values correlated well.
These investigations provided the confidence that the 10 000 points in each of the
acceleration records provided were being correctly read by the analysis program.
Initial earthquake design runs for the cable-stayed unit indicated that without any longitudinal
restraint at the tower piers, the design preference, the bridge would experience movements up
to 1200 millimeters in each direction at the ends of the unit. Further study indicated other
undesirable effects with full fixity and full longitudinal restraint at the tower piers. These
conditions caused live load rotations and temperature rise and fall to place higher, often
undesirable, demands upon other bridge elements. The erection analyses concluded that
construction of the bridge with full fixity would generate forces much higher than those
observed with no restraint.
However, the fixity studies also revealed that there were some advantages to fixity as well.
The wind induced motion of the bridge could be reduced while the flutter velocity threshold
increased and longitudinal displacements under the various live load combinations could be
minimized. Reduced movements would then require smaller expansion joint devices and
relieve the required movement capacity of the side span tie down devices. These studies led
to the conclusion that the cable-stayed spans should be restrained longitudinally, either with
some type of bearing or key. Development of preliminary alternative details for this type of
restraint indicated that the most effective solution would be one which allowed limited
translation, that caused by slowly applied loads such as a uniform temperature change, and
free rotation under all loading conditions.
Several types of seismic isolation and damping systems were considered; however, it was
determined that force transfer would provide the most efficient solution. Both isolation and
additional damping were studied to evaluate the potential impact on the design and cost of the
structure. The overall stiffness contribution of the stay cables and the length of the main span
reduced the effectiveness of both alternative solutions.
It was determined that force transfer was the most practical solution, and for this bridge it is
achieved through the use of an earthquake shock transfer device. This device, comprised of a
cylinder filled with silicon putty and a piston, is capable of transferring forces in both tension
and compression. Therefore, the "double action" of the unit simplified the design of the
connections to the structure and permitted transfer of earthquake forces at an elevation much
lower than the stay cable connections.
The tower piers, which support the bulk of the load, are supported by a spread footing and a
dredged caisson on rock at Piers 2 and 3, respectively. Other foundation types were studied,
however, it was determined that another foundation type was not justifiable given the
magnitude of the earthquake forces and the depth of water and alluvial soils.
Approach Spans
Liquefaction presents little problem for the cable-stayed unit since the supporting piers are
founded on huge footings keyed into rock; however, the approach spans are considerably
different. As noted earlier, these foundations are located in an area with very deep, highly
liquefiable soils. When combined, the depth of liquefaction and anticipated scour eliminated
spread footing type foundations from consideration. Even soil improvement techniques were
considered ineffective due to degradation and aggredation of the channel. After studying
several possible alternative solutions including pile and drilled shaft foundations and various
soil improvement methods, it was determined that piers supported on drilled shafts to rock
would be the most economical solution.
These shafts, which are drilled through water-bearing sand to a depth of up to 30 meters will
require casing for installation. Therefore, the decision was made to require that the casing for
the shafts be left in place and used as additional confinement steel during extreme condition
seismic events. These conditions are generally after loss of lateral support of the shafts due to
either scour or liquefaction. The analysis of the structure for the extreme event condition was
conducted within the T187 "dynamics" routine. Since the solution method is based upon the
values of displacement, velocity and acceleration of the joints at the previous time step, it was
possible to simulate liquefaction and loss of lateral support in the foundation by varying the
foundation stiffness during the run.
In this way, we were able to analyze the structure with various foundation support conditions,
fixity at the piers, and multiple restrainer combinations. Ultimately it was determined that the
drilled shaft option, placed in permanent casing was the best overall foundation for the site.
We found that no one foundation support condition governed all aspects of the design.
Primarily the half-scour and liquefied states governed the design of the piers and foundations
while the forces in the structure with full support controlled the design of the bearings and
superstructure connections.
Conclusions
Although the New Madrid fault zone lies within approximately 80 kilometers of the bridge
site, the design seismic event is for a magnitude 8.5 earthquake, and the bridge must
withstand the seismic forces within the elastic range, the design of the Cape Girardeau
replacement bridge was a success. Each point of the design criteria was met without creating
unnecessarily complex details and without significant added cost to the structure. The
electronic transfer of files related to the design earthquake and the cooperation of all parties,
including MoDOT, HNTB Corporation, and both the wind and seismic consultants all helped
to identify and solve problems before they became too difficult or costly to handle. In this,
the seismic analysis, wind studies and final design and detailing of the bridge remained on
schedule and within the client's budget.
Acknowledgments
The author extends his thanks to the Missouri Department of Transportation for permission to
share their experiences and knowledge of the seismic evaluation and design of the bridge and
Woodward-Clyde Consultants for much of the data, testing and graphics used in the
compilation of this paper and its presentation.
[1].
[2].
[3].
[4].
[5].
Based on U.S. Army Corps of Engineers data for locks at River Mile 274, at St. Louis,
Missouri in 1989 and 1990.
Nuttli, Otto W., The Effects of Earthquakes in the Central United States, 2nd ed., May
1990, Center for Earthquake Studies, Southeast Missouri State University
Nuttli, Otto W., The Mississippi Valley Earthquakes of 1811 and 1812: Intensities,
Ground Motion and Magnitudes, Bulletin of the Seismological Society of America,
Vol. 63, No.1, 1973
Nuttli, 1990
Johnson, A.C. & Nava, S.J., Recurrence Rates and Probability Estimates for the New
Madrid Seismic Zone, Journal of Geophysical Research, Vol. 90, No. B8, July 10,
1985
Cable-Stayed Bridges for Urban Spaces
Antnio REIS
Armando PEREIRA
Techn.Dir
GRID-Cons. Eng.
Lisbon-Portugal
Civil Engineer
GRID-Consulting Engineers
Lisbon-Portugal
Jos PEDRO
Daniel SOUSA
Civil Engineer
Lisbon-Portugal
Civil Engineer
Lisbon-Portugal
Summary
Design concepts and case studies for cable stayed bridges, in which aesthetics and environmental
conditions required particular consideration related to its integration in urban spaces are reported.
The solutions adopted for decks, towers and cable-stayed arrangements are compared. Structural,
aerodynamic behaviours and execution methods are discussed for concrete cable stayed bridges.
1.
Introduction
The design of a cable-stayed bridge in an urban space requires particular consideration of aesthetics
and environmental integration. Complex geometrical constraints, due to in-plan curved alignments,
traffic maintenance requirements during erection stages and aspects related to the structural
behaviour and the competitiveness of cable-stayed solutions, are a challenge for designers. Very
often, classical cable-stayed solutions namely, symmetrical solutions with two pylons, are not the
most efficient ones to overcome difficult urban site conditions. Besides, for short to medium spans,
Owners and Designers tend to avoid cable-stayed solutions for urban sites due to economical and
environmental integration reasons so, beam and slab type of bridges are usually selected for urban
viaducts. However, as referred by Christian Menn [1], the general public was never captivated by
modern bridge construction. Beam bridges were largely perceived as boring. These aspects shall
be taken into consideration by civil engineers at the conceptual design stage. The architecture of
bridge design is a civil engineering problem.
2.
Aesthetics and Environmental Demands
In the last few years, a tendency to adopt cable stayed bridges for spans ranging from 150 to 500
meters has been observed. For urban spaces, cable stayed bridges have been adopted most
frequently over rivers and not very often in urban viaducts where long spans are unnecessary.
However, even for small to medium spans, some advantages of cable stayed solutions should be
considered:
the transparency of the solution, reducing the number of piers.
the slenderness of the deck.
the reduction of traffic disturbances during execution.
Open spaces with very few constraints, are ideal locations for cable-stayed solutions with a 3 span
scheme and two masts. However these conditions are infrequent in places where urban bridges and
viaducts have to be built. The lack of symmetry at some of these locations, namely due to existing
constructions, the skew alignment of the upper and lower roadways, the existing of the interchanges
or roundabouts, may require asymmetric solutions for reducing the visual impact. Besides, urban
bridges should not spoil the sites but to improve the quality of the environment for the benefit of the
citizens.
The classical rule-form follows function, should not be adopted as a limiting principle at the
conceptual design stage, but as a guideline. Bridges are very often landmarks for the cities;
innovation and creative ideas are necessary [2].
3.
Stay Cables: Arrangement and Visual Impact
A single plan of stays-axial type cable arrangement is likely to be the best choice for integration
into an urban space where multiple visual obstructions exist. Besides, avoiding the visual impact of
the crossing of stays for skew views of the bridge, is a well-known advantage. However, to adopt an
axial staying scheme an increased torsional stiffness of the deck is required for transverse load
distribution and aerodynamic stability. All the bridges herein discussed were designed according to
this concept. In Fig. 1, a cable stayed solution for a 92m main span urban viaduct was shown to be
the best solution for environmental integration.
Figure 1 -A symmetrical solution for a cable-stayed viaduct for a highway in the city of Funchal,
Madeira Island.
Figure 2 Symmetrical stay-cable arrangement for the Viaduct of Fig.1.

The stays (31 to 43 strands - 15mm), with an axial arrangement, are 5 m distance apart only, at
the deck level (Fig.2). This bridge is currently under construction in the city of Funchal. However,
in-plan curvature of the deck, makes the adoption of axial cable-stayed solutions more difficult. For
the case shown in Fig. 3 and 4, The St Tirso Bridge, in the North of Portugal, a 3D arrangement of
stays was adopted to solve a 61m end span. A set of backstays, in a single plan, was introduced for
longitudinal stability of the pylon. Concentrating the anchorage of the stays, at approximately 3
points, creates order and a 3D image of the bridge consistent with the helicoidal access ramp. The
architectural integration of the bridge structure with the railway station (Fig.3), was achieved. Six
(2x3) cable stays were adopted at the end span, 37 strands (15mm) each. The 3 backstays have 70
strands each. All the strands were stressed at the top of the mast. The anchorages at the deck level
(Fig.5) are located in such a way to reduce permanent bending actions in the pylon. The resultant
transverse and longitudinal horizontal forces induced by the stays at the top of the mast, were
minimised by adjusting the cable forces at the end of the construction. This bridge was opened to
traffic in the summer of 1998.
Figure 3 - The St Tirso Bridge over the river Ave - a 3D stay cables arrangement.
Figure 4 - Longitudinal section of the St Tirso Bridge.

Another design case of an urban viaduct,
where a cable-stayed solution was selected in a
design competion, is shown in Fig. 6 and 7. A
main span of 120m allows the crossing of a
highway and several railway lines. For this
particular case, we designed a completely
asymmetric solution. It was important to
understand the urban planning issues, since the
viaduct will be built along an avenue with a
green park.
Figure 5 - Anchorages at the deck level of the St Tirso Bridge.

A classical symmetric solution, for the stay-cable arrangement, with a single mast, considered as
well as an alternative solution at the preliminary design, would never attain the aesthetical
objectives of the urban planning.
The Owner decided to select the asymmetric solution, even taking into consideration its cost being
about 15% higher than the symmetric solution. The staying scheme, consists in a single layer of 37
to 43 strands of 15mm
diameter, spaced at 8,0m
along the main span; the
back stays (2x3) arranged in
two planes, have 67 strands
of 15mm each. These
backstays are anchored at the
deck level by prestressed
concrete crossbeams and
have a 3D arrangement of the
anchorage at the top of the
tower.
Figure 6 -An asymmetric stay-cable solution for the Praa das
Flores Viaduct in Oporto.
Figure 7 Praa das Flores Viaduct: longitudinal section with the arrangement of the cablestays.
4.
Decks: Aesthetics and Structural Performance
For cable-stayed bridges in urban spaces, if an axial arrangement of stays is preferred, a sufficient
torsional stiffness for the deck for aerodynamic stability is required. Besides, transverse cross
section deformations under asymmetric live loading shall be taken into consideration if an axial
stayed scheme is adopted. Single cell box girders with prestressed steel diagonals, as we adopted
several years ago in the Socorridos Bridge [3], are usually the simplest solution. This type of
solution was adopted for the Praa das Flores Viaduct (Fig.8).
Figure 8 - The Praa das Flores Viaduct: A trapezoidal box girder deck.
For viaducts located in urban areas, if a very slender deck is required, the standard trapezoidal
single cell box girder may be replaced by a triangular type box girder (Fig.9) as we adopted for the
Viaduct in Funchal. The main advantage is aesthetical, because keeping the same depth (2,0 to
2,20m minimum for cast in situ prestressed concrete decks) the triangular superstructure is
apparently much more slender. However, the torsional stiffness is reduced for the triangular shape
and so the torsional vibration frequency. In Table 1, a comparison is made for the 3 types of decks
adopted for the bridges previously referred to, in what concerns the main parameters controlling the
aerodynamic stability. For short spans, say up to 70 meters, a slender concrete voided slab is a very
simple and feasible solution. However, the dead weight may be 30 to 40% more than a box girder
resulting in additional cost for the stays.
Figure 9 - Cross-section for the Viaduct in Funchal - A triangular shape box girder deck.
For the St Tirso Bridge, a voided slab was adopted for all the spans including the end span (61m)
that is elastically supported at the mid span section by two planes of 3 stay cables (Fig.5 and 10).
The natural torsional frequency (Table 1) is higher for the trapezoidal box girder deck due to the
increased J compared to the other sections.
Figure 10 - Cross-section for the St Tirso Bridge - A voided slab cable stayed deck.
Besides, the torsional frequency of the St Tirso Bridge is significantly affected by the transverse
flexibility of the pylon. With a rigid pylon, fb increases from 1.29Hz to 1.98Hz. This type of
solution - an axial cable-stayed slab is feasible. Introducing an axial central rib as adopted by
Cremer et al [4] at the Kortrijk Bridge may increase the torsional stiffness.
frequencies.
DL(kN/m)
DL/m2
J (m4)
fb(Hz)
ft (Hz)
Pr.Flores Viaduct
18.0
253.0
14.1
13.61
0.84
2.66
Viaduct in Funchal
21.5
326.5
15.2
4.80
0.80
1.58
St Tirso Bridge
13.0
245.0
18.8
4.20
0.76
1.29
b (m)
Table 1 - Deal Load (DL), torsional stiffness factor (J) and natural bending (fb) and torsional (ft)
The structural behaviour under asymmetric loading and the aerodynamic stability of the triangular
box girder was studied. However, this deck presents good relationships span/width =4.3 and
torsional / flexural vibration frequencies ft /fb = 2 for aerodynamic stability.
This was confirmed [5] at the wind tunnel where a sectional model (1.45m long; 1/50 scale) was
tested for aerodynamic stability and to determine the aerodynamic coefficients. The wind angles of
attach were varied between - 5 and 6 and the wind speeds correspond in the prototype to 133km/h
to 266km/h. The drag coefficient varies between 1.1 and 1.3, being the maximum observed for
232km/h. No aerodynamic instabilities were detected or any kind of vortex shedding; the analysis
of drag and lift coefficients shows a stable behaviour also with respect to gallop instabilities.
5.
Towers : Aesthetics and Functional Requirements
The geometry of the towers shall result from aesthetics, structural and functional requirements.
Functional requirements shall be here understood as including constructional demands, namely
clearances for anchorages at the top of the pylons, horizontal and vertical clearances at the deck
level and safety with regard to structural instability. The height of the pylon, herein referred as the
part of the tower above the deck, is related to the acceptable minimum angle of the cables with the
deck. For classic cable stayed solutions, this angle may be of the order of 20, resulting in a pylon
height of about 20% of the span for solutions with two towers (Fig. 2) and 35 to 40% of the span for
single pylon solutions. If the stay cables do not extend throughout the span (Fig. 6 and 7), what may
be convenient for aesthetics in urban viaducts, the height of the pylon may be reduced.
Figure 11 - Anchorages at the top of

the pylon of the Viaduct in
Funchal.
Figure 12 - Anchorages at the top of the

pylon of the St Tirso Bridge.
Figure 13 - The tower of the Praa das Flores Viaduct.
The anchorages at the top of the pylon and the horizontal clearances at the deck level may require a
variable width for the pylon. That may be required to cross the cables if a 3 plane arrangement is
adopted as in Fig. 12 and 13 or to have enough space at the inside of an anchorage steel tower head
for stressing operations as in Fig. 11. For these cases the width of the pylon increases toward the
top. A different approach, due to functional requirements, was adopted for the Praa das Flores
Viaduct. We decided to insert a central walkway. Besides, an innovative shape for the tower (Fig.
13) was adopted as a result of aesthetical and structural requirements - the tower should be an
emblematic element for the urban park; the tower should have a sufficient longitudinal bending
stiffness to reduce the forces at the backstays, and the bending moments in the vertical shaft of the
pylon. The balance of horizontal components at the top of the mast for the permanent actions was
achieved by adjusting the stay forces. In Fig. 14 one compares the bending actions under live load
in the vertical shaft of the tower with and without the inclined leg. To accommodate the walkway a
door was open in the tower. A detailed finite element analysis of this part of the tower was done
to find out the flow of internal forces and the local bending actions. The tower shaping results from
two triangles connected by their bases, resulting in an appearance consistent with the triangular
arrangement of the stay cables. The door has a triangular (gothic) appearance as well.
H[m]
-2066
- 10000
H[m]
70
70
60
60
50
50
40
40
30
30
20
20
- 5000
5000
10000
- 10000
- 5000
6402
0
10
10
5000
10000
Figure 14 - Comparision of the bending actions in the vertical shaft of the tower of
Fig 13, with and without the inclined leg.
6.
Construction: Site Constraints and Execution Methods
The construction of bridges located in urban spaces is very much influenced by site constraints,
namely disturbance in existing traffic conditions, environmental noise impact and safety during
erection stages. For the bridge cases previously reported, two different methods were adopted - span
by span, cast in situ deck on a formwork supported from the ground for the St Tirso Bridge and
balanced cantilever, with cast in-situ segments, for the other two viaducts. To stress the stay cables
of the St Tirso Bridge, it was necessary to control the interaction of the deck with the formwork in
order to control the decompression limit state at the end span. The maximum allowable vertical
deflection at the anchorage section was evaluated as being 30mm. After reaching this value, the
formwork girders were lowered at the supports and a 2nd stressing operation was carried out. The
process was continued until the specified stresses in the stay cables have been reached.
For the viaduct in Funchal, the 5,0m segments are casted in two stages - 1st the lower flange and
webs and 2nd the top flange, due to the shape of the cross section. Each stay is stressed only after
moving the equipment to the next segment.
The construction scheme for the viaduct in Oporto requires the casting of the 8,0m segment in two
stages - 4.0+4.0m, stressing the adjacent cable stay before casting the 2nd stage.
7.
Conclusions and Final Remarks
Aesthetics, environmental integration, structural performance and execution methods were

discussed for the design of cable-stayed bridges of short to medium spans located in urban areas.
Three case studies of recent bridge designs were reported. The advantages of using axial cablestayed solutions and asymmetric configurations were shown; these concepts have been adopted by
the authors in another bridge currently being designed (Fig. 15). The deck, 30m width, is a
composite box girder deck where the webs are replaced by a 3D steel truss. The lower flage has a
pedestrian function, since the bridge will be located between two green parks. The solution select in
a design competition is expected to be a landmark for the city.
Figure 15 - A model for the Europa Bridge in Coimbra, Portugal.
8.
References
[1]
C. Menn - Functional Shaping of Piers and Pylons. Structural Engineering International,

Vol.8, n 4, 1998.
A.J.Reis - Designing Post-tensioned Concrete Bridges for Innovation, in Post-Tensional
Concrete Structures, FIP Symposium, Vol. 1, pag. 963, 1996.
A.J.Reis, A.Pereira - Socorridos Bridge: A Cable-Panel Stayed Concept. International
Conference A.I.P.C. - F.I.P. Deauville 1994, Proceedings - Vol. 1, pag. 343/350.
J.M.Cremer, V.Ville de Goyet, A.Lothaire, L.Ney, V.Radu - Some Innovative CableStayed Bridges. International Conference A.I.P.C.-F.I.P. Deauville 1994, Proceedings - Vol.
1, pag. 235/260.
F.Branco, P.Mendes, J.Ferreira Ensaios em Tnel de Vento do Tabuleiro do Viaduto
sobre o Caminho do Comboio. Relatrio IC-IST, EP n 10/97.
[2]
[3]
[4]
[5]
A new method to assign initial cable forces

for prestressed concrete cable-stayed bridges
Dr. Dewei Chen, born
1956,
received his degrees from
Tongji University of
Shanghai.. He is now an
Associate Professor of
the department.
Dewei CHEN
Associate Professor
Tongji University
Shanghai, China
Summary
The determination of initial cable forces in a prestressed concrete cable-stayed bridge for a
given vertical profile of deck under its dead load is an important but difficult task that affects
the overall design of the bridge. A new method utilizing the idea of force equilibrium is
presented in this paper for their determination. The method can easily account for the effect of
prestressing and the additional bending moments due to the vertical profile of the bridge deck.
It is much more rational and simpler than the traditional zero displacement method, and it is
able to achieve bending moments in the bridge deck approaching those in a continuous beam
over rigid simple supports.
Introduction
The cable-stayed bridge is a modern form of bridge which is both economical and aesthetic.
It has been extensively employed in the construction of long-span bridges in the past few
decades. However this kind of structures are highly statically indeterminate, and therefore
many schemes of initial cable forces are possible. In the particular case of prestressed
concrete cable-stayed bridges, it is especially important to choose an appropriate scheme of
initial cable forces while the bridge is under dead load only. Owing to shrinkage and creep,
the deflections will change with the passage of time and the internal forces may also
redistribute. Should an inappropriate scheme of initial cable forces be chosen, an
unfavourable pattern of internal forces may be locked in subsequently, for which there may be
no simple solution.
Theoretically it is possible to search for a stable scheme of initial cable forces under which
there is the minimum redistribution of internal forces and time-dependent displacements.
However it is usually very difficult in view of the many factors affecting the subsequent timedependent deformations. For example, many cable-stayed bridges are constructed using cast
insitu segmental cantilever construction, which gives rise to complex effects of shrinkage and
creep because of the different ages of concrete. The presence of longitudinal prestressing also
complicates the problem further. Inevitably some simplifying assumptions have to be made.
Review of Existing Methods

The scheme of initial cable forces giving rise to bending moments in the bridge deck
approaching those of an equivalent continuous beam with all the supports from cables and
towers considered as rigid simple supports is generally acknowledged to be both rational and
practical, as the long term behaviour of the bridge is reasonably stable. The problem hinges
upon how to achieve this scheme of initial cable forces. There are two main categories of
methods in achieving an appropriate scheme of initial cable forces in prestressed concrete
cable-stayed bridges [1-6], namely the optimization method [2-5] and the zero
displacement method [6].
In the optimization method [2-5], the initial cable forces are chosen based on the optimization
of certain objective functions which may either be related to structural efficiency or economy.
In this method, the total strain energy is often one of the objective functions to be minimized.
It is necessary to impose the constraints for optimization very carefully, or else the resulting
schemes may sometimes become impractical.
On the other hand, the traditional zero displacement method [6] is more straight forward in
theory, and it enables the designer to fine-tune the initial cable forces as well as the structural
configuration. If a straight and horizontal bridge deck is supported on a number of stay
cables, the horizontal components of the cable forces have little effect on the bending
moments of the deck, and hence the bending moments are primarily governed by the vertical
components of the cable forces and the dead load. In the zero displacement method, an
appropriate scheme of initial cable forces is obtained by making the deflections at the cable
anchorages vanish. When the deck gradient is negligible, the resulting bending moments in
the deck are essentially those of an equivalent continuous beam with all supports from cables
and towers considered as rigid simple supports. However, when the vertical profile of the
bridge deck is significant by reason of traffic requirements or otherwise, the basis of this
method is itself questionable. As the horizontal components of the cable forces will induce
additional bending moments in the deck, the resulting bending moments are likely to cause
substantial redistribution in the long run. In this case, what really matter are the bending
moments because they will affect the long term behaviour of the bridge. Whether the
corresponding displacements are zero or not is immaterial, as they can be adequately
controlled by appropriate precamber or preset of the deck during construction.
In this paper, a new method utilizing the idea of force equilibrium is presented for the
determination of a stable scheme of initial cable forces. The method can easily account for
the effect of prestressing and the vertical profile of the bridge deck, and therefore it is much
more rational as well as simpler than the traditional zero displacement method. Two
numerical examples using real cases of prestressed concrete cable-stayed bridges are
presented to demonstrate the versatility of the proposed method.
The Force Equilibrium Method

In the force equilibrium method, the cable-stayed bridge is modelled as a planar structure.
The method works on an evolving substructure eventually comprising the bridge deck and
towers, and searches for a set of cable forces which will give rise to desirable bending
moments at selected locations of the substructure. As the method works only on the
equilibrium of forces rather than deformation, there is no need to deal with non-linearity
caused by cable sag and other effects. The method is therefore computationally efficient.
First of all, certain sections of the bridge deck and tower are chosen as control sections where
the bending moments are adjusted by varying the cable forces. Consider a typical single
tower cable-stayed bridge, as shown in Figure 1, in which the connection between the bridge
Figure 1. A typical single tower cable-stayed bridge.
Figure 2. Stage 1 model for cable-stayed bridge shown in Fig. 1.

deck and tower is monolithic. To established the target bending moments, only the bridge
deck is considered. All supports from the cables and tower are replaced by rigid simple
supports, as shown in Figure 2. This is regarded as the Stage 1 model for the sake of
subsequent discussions. The prestressing to be applied during construction is also taken into
account. The bending moments caused by dead load in the bridge deck under such modified
support conditions are then taken to be the target bending moments. It is noted that the
prestressing to be introduced after the completion of the bridge deck is not taken into account
here. These target bending moments are adopted because the effects of creep and shrinkage
of concrete tend to change the bending moments towards these target values in the long term
anyway [1]. If the initial bending moments in the towers can be controlled at the same time,
the scheme of initial cable forces is reasonably stable. It is further assumed here that factors
such as the differences in age among deck segments are insignificant in the long term and
therefore they are neglected.
Fig. 3 shows the same bridge as in Fig. 1, except that all cables are taken away and replaced
Figure 3. Model for stage 2 and stage 3 for cable-stayed bridge.

by the internal forces. This simplified model applies to both Stages 2 and 3. The only
difference between these two stages lies in the degree of sophistication. The cable forces are
taken as independent variables for adjustment of bending moments at the control sections.
Normally the bending moment at each deck section where a cable is anchored is treated as a
control parameter. It should be pointed out that wherever a model consists of a back-stay
anchored at the deck above an end pier, where the deck carries no bending moment, the
corresponding cable force can be treated as an additional variable to improve the structural
efficiency further. For example, the bending moment at the deck-tower junction or that at the
tower base may be taken to be an additional control parameter as they are critical sections
affecting the long term behaviour. The target bending moments at the deck sections are those
obtained from the Stage 1 model whereas the target bending moment at the chosen tower
section is normally set as zero.
The above arguments can also be extended to other configurations of cable-stayed bridges. In
a symmetric single tower cable-stayed bridge without back-stays anchored above end piers,
the bending moments in the tower should normally be zero under dead load, and therefore
there is no need to treat any of these as a control parameter. In a two-tower cable-stayed
bridge of symmetric arrangement, it is only necessary to consider one half of the bridge with
appropriate boundary conditions at the middle section to account for symmetry, and the above
reasoning can similarly be applied.
The main purpose for setting up the Stage 2 model is to evaluate the approximate influence
coefficients, which are the bending moments at the control sections caused by a unit load in a
certain cable. In order not to introduce the non-linearity of cable stiffnesses, some
simplifying assumptions are made. The self-weight of each cable is neglected, and hence the
forces at the ends are roughly equal. The bending moments in the deck are primarily
determined by the cable forces acting on the deck, and to a lesser degree by the cable forces
acting on the tower. Therefore the cable forces acting on the tower are neglected in the
calculation of bending moments in the deck. Similarly in the calculation of bending moment
at the control section at the tower, only the cable forces acting on the tower are taken into
account. The errors introduced by these simplifying assumptions will be almost eliminated
by iterations in the next stage.
Considering the equilibrium of the Stage 2 model, the following equation can be written
{ M } = [m] {T} + { M }
0
(1)
where { M 0 } is an N-dimensional vector containing the target bending moments M i0 derived

from the Stage 1 model, [m] is an NN matrix containing approximate influence coefficients
mij for the Stage 2 model in which mij is the bending moment at the ith control section caused
by a unit force in the jth cable, {T } is an N-dimensional vector containing the cable forces
Tj , { M d } is an N-dimensional vector containing the bending moments M id caused by only
dead load and prestress in the Stage 2 model, N is the number of cables considered in the
model, i is the subscript corresponding to the ith control section and j is the subscript
corresponding to the jth cable.
If { M 0 } contains the bending moments of the equivalent continuous beam on rigid simple
supports as obtained from the Stage 1 model, and the control sections are well chosen so that
the matrix [m] is non-singular, an initial estimate of the cable forces {T 0 } can be calculated
from the Stage 2 model as
(2)
{T 0 } = [m]1 { M 0 } { M d }
However the cable forces {T 0 } obtained above are only rough estimates as the Stage 2
model does not take into account the interaction among tower, cables and deck. It is therefore
necessary to build the Stage 3 model.
In the Stage 3 model, the interaction among tower, cables and deck is taken into account by
iterations. The cable forces at the deck anchorages are taken as independent variables in the
optimization process, and the self-weight of each cable can also be introduced. Using the
initial estimate of the cable forces {T 0 } , as well as the bending moments { M d } caused by
dead load and prestress in the Stage 2 model, the updated deck bending moments { M 1} can
be calculated from the Stage 3 model. Such bending moments are normally different from the
target bending moments
{M } ,
0
and hence it is necessary to introduce some adjustments
{T } of the cable forces given by

1
{T } = [m] ( { M } { M })
1
Using the updated cable forces
(3)
{T } given by
1
{T } = {T } + {T }
1
the updated deck bending moments
(4)
{M }
2
can then be calculated again from the Stage 3
model. Notice that the approximate influence matrix [m] for the Stage 2 model has been
used in the Stage 3 model, and hence further iterations are necessary. Further adjustments
{T } may be obtained by
2
{T } = [m] ( { M } { M })
1
resulting in more accurate cable forces
(5)
{T } given by
2
{T } = {T } + {T } + {T }
2
(6)
This process can be repeated until the updated deck bending moments
{M } .
0
{M }
n
converge to
This is summarized in the flow chart shown in Figure 4.
Numerical Examples
Two numerical examples are presented to demonstrate that the present method is both
rational and reliable. Both are taken from existing prestressed concrete cable-stayed bridges
in China, but some minor simplifying modifications are made.
Example 1. A single tower prestressed concrete cable-stayed bridge with harp
arrangement
The first example is a single tower prestressed concrete cable-stayed bridge, situated in
Ningbo City, China, with spans of 90m and 105m. The moment of inertia (ID), the cross
sectional area (AD) and the Youngs modulus (ED) of the deck are 4.706m4, 12.145m2 and
3.5107kN/m2 respectively. The stay cables are of the harp arrangement. Three types of stay
cables are used, and their respective cross sectional areas (AS) are 0.013m2, 0.0166m2 and
0.0208m2. The Youngs modulus of the stay cables (ES) is 2.1108kN/m2. The tower is
stepped with the biggest section below the bridge deck and the smallest section over the
length where the cables are anchored. The moments of inertia (IT) of the tower are 11.212m4,
19.939m4 and 79.688m4. The corresponding cross sectional areas (AT) are 14.46m2, 19.0m2
and 45m2 respectively, while the Youngs modulus (ET) is 3107kN/m2. The information on
the prestressing is omitted for brevity.
Three different vertical profiles of the bridge deck have been considered. The bridge deck is
straight and horizontal in Case 1. In Cases 2 and 3, both the vertical profiles consist of a
symmetric parabolic summit curve of 180m horizontal length and a straight tangent of 15m.
The highest point is precisely at the tower location. The gradients of the straight tangents for
Cases 2 and 3 are 3% and 9% respectively. Figure 5 shows an elevation of the bridge for
Case 3.
Figure 5. A single tower P.C. cable-stayed bridge with harp arrangement.

The present method was applied to optimize the bending moments in the bridge deck for the
three cases, and the tolerance value used to terminate iterations was 5kNm. The results for
Case 3 are shown graphically in Figures 6-8. Notice that the deck bending moments after
optimization agree well with the target values obtained from an equivalent beam on rigid
simple supports, except at the tower section which was not chosen as a control section. The
abrupt jumps in bending moment are caused by prestressing. The bending moment at the
tower base is also very close to zero after optimization.
The three cases were also analyzed by the zero displacement method using 0.001m as the
Figure 6. Internal forces in bridge deck of example 1 (Case 3).

a) Target bending moment in deck in kNm; (b) Bending moment in kNm;
(c) Shear force in kN; (d) Axial force in kN.
Figure 7. Cable forces of Example 1 (Case 3) in kN.
Figure 8. Internal forces in tower of Example 1 (Case 3).

a) Bending moment in kNm; (b) Shear force in kN; (c) Axial force in kN.
tolerance value to terminate iterations. The initial cable forces for the three cases obtained by
the present method are tabulated in Table 1 and compared to those obtained by the zero
displacement method. It is observed that when the bridge deck has no slope, i.e. Case 1,
results from the above two methods are effectively the same. There are, however, marked
differences in the other two cases, especially in the cables close to the tower.
Case 1
Case 2
Case 3
Cable
Present
Zero Disp.
Present
Zero Disp.
Present
Zero Disp.
No.
Method
Method
Method
Method
Method
Method
14045
14045
14483
14398
15458
15351
3931
3931
3969
3975
4044
4065
8832
8832
8919
8915
9130
9116
6757
6757
6798
6800
6836
6841
7242
7242
7200
7200
7126
7125
6900
6900
6825
6818
6637
6630
7727
7727
7561
7590
7283
7311
6676
6676
6605
6497
6224
6117
6707
6707
6016
6416
5428
5826
10
2889
2889
4251
2754
3971
2484
11
14618
14618
11288
14323
11143
13758
12
14619
14619
11298
14339
11118
13769
13
2887
2888
4245
2747
3961
2474
14
6706
6706
6013
6414
5422
5820
15
6680
6680
6607
6499
6222
6115
16
7714
7714
7552
7582
7270
7299
17
6948
6948
6867
6860
6678
6671
18
7064
7064
7024
7025
6931
6934
19
7304
7304
7340
7339
7427
7421
20
6998
6998
7027
7032
7028
7044
21
10768
10768
10922
10912
11560
11524
22
8924
8924
9306
9311
9908
9924
Table 1
Initial cable forces for Example 1 (kN)
Example 2. A single tower prestressed concrete cable-stayed bridge with semi-fan

arrangement shown in Fig. 9
The second example is also a single tower prestressed concrete cable-stayed bridge, situated
in Jilin Province, China, with spans of 95m and 132m. The moment of inertia (ID), the cross
sectional area (AD) and the Youngs modulus (ED) of the deck are 5.1m4, 10.579m2 and
3.5107kN/m2 respectively. The stay cables are of the semi-fan arrangement. Three types of
stay cables are used, and their respective cross sectional areas (AS) are 0.020m2, 0.019m2 and
0.013m2. The Youngs modulus of the stay cables (ES) is 2.1108kN/m2. The tower is
stepped in a manner similar to Example 1, and the moments of inertia (IT) are 17.92m4,
24.01m4 and 47.73m4. The corresponding cross sectional areas (AT) are 17.92m2, 13.44m2
and 37.20m2 respectively, while the Youngs modulus (ET) is 3107kN/m2. The effects of
prestressing is not considered in this example for simplicity. The vertical profile consists of a
symmetric parabolic summit curve of 190m horizontal length and a straight tangent of 37m.
The highest point is again precisely at the tower location. The gradient of the straight tangent
is 6%. The results obtained using the present method are shown in Figure 6, indicating very
good agreement between the deck bending moments and the target values, shown in Figure 912.
Figure 9. A single tower P.C. cable-stayed bridge with semi-fan arrangement (Example 2).
Figure 10. Internal forces in bridge deck of Example 2.

a) Target bending moment in kNm; (b) Bending moment in kNm;
(c) Shear force in kN; (d) Axial force in kN..
Figure 11. Cable forces of Example 2 in kN.
Figure 12. Internal forces in tower of Example 2. (a) Bending moment in kN; (b) Shear force in kN;
(c) Axial force in kN.
Conclusions
A new method utilizing the idea of force equilibrium is presented for the determination of an
optimum scheme of initial cable forces in a prestressed concrete cable-stayed bridge for a
given vertical profile of deck under its dead load as well as prestress. In the proposed
method, the stiffnesses of the cables do not enter into the calculations, and it therefore
obviates the need for introducing non-linearity into the algorithm. The bending moments,
rather than the displacements, of the deck are taken as parameters to be controlled. The
additional bending moments caused by the vertical profile of the deck can also be taken into
account. Two real prestressed concrete cable-stayed bridges have been investigated using the
proposed method, which demonstrate that it is both rational and practical.
It is also observed that, as far as the initial bending moments of the tower are concerned, the
harp arrangement is less favourable than the fan or semi-fan arrangement, as the cables are
anchored over a larger length in the former case. The proposed method is also a handy tool
for optimizing the bending moments in the tower.
Acknowledgements
The financial support of the block grant from the Scale B, The Scientific Committee of P.R.
of China is acknowledged.
References
[1].
[2].
[3].
[4].
[5].
[6].
Analysis of Secondary Stresses in Prestressed Concrete Cable-stayed Bridges due to

Creep (in Chinese), Shanghai Institute of Design and Research in Municipal
Engineering, P. R. China, 1983.
Furukawa, K., Sugimoto, H., Egusa, T., Inoue, K. and Yamada, Y., Studies on
optimization of cable prestressing for cable-stayed bridges. Proceedings of
International Conference on Cable-stayed Bridges, Bangkok, 1987, 723-734.
Lu Q. and Xu Y.G., Optimum tensioning of cable-stays (in Chinese). Chinese Journal
of Highway and Transport, P. R. China, 1990, 3(1), 38-48.
Simoes, L.M.C. and Negrao, J.H.O., Optimization of cable-stayed bridges with boxgirder decks. Proceedings of the 1997 5th International Conference on Computer
Aided Optimum Design of Structures, Rome, Italy, 1997, 21-32.
Negrao, J.H.O. and Simoes, L.M.C., Optimization of cable-stayed bridges with threedimensional modelling. Computers and Structures, 1997, 64 (1-4), 741-758.
Wang, P.H., Tseng, T.C. and Yang, C.G., Initial shape of cable-stayed bridges.
Computers and Structures, 1993, 46(6), 1095-1106.
Begin
Input properties of cable-stayed bridge (Figure 1)
Set up the Stage 1 model (Figure 2)
Calculate the target bending moments {M 0}
from the Stage 1 model (Figure 2)
Calculate the approximate influence matrix
[m] from the Stage 2 model
Calculate the bending moments {M d} caused by dead load
and prestress from the Stage 2 model
Calculate an initial estimate of the cable forces {T 0} from
{T 0} = [m]-1 ({M 0}-{M d})
Calculate the updated bending moments
{M } caused by {M d} and {T 0} from the Stage 3 model
1
Set i = 1
Calculate the adjustment {T i} to the cable forces from
{T i} = [m]-1 ({M i}-{M 0})
Calculate the updated cable forces {T i} from
{T i} = {T 0} + { T j}
Set i = i + 1
Calculate the updated bending moments {M i} caused by
{M d} and {T i-1} from the Stage 3 model
No
Check convergence
||{M i}-{M 0}|| < ?
Yes
End
Figure 4. Flow chart describing the present method.

10
Bridges with Spatial Cable Systems - Theoretical and Experimental Studies

Tina VEJRUM
Dr. Eng.
COWI AS
Lyngby, Denmark
Tina Vejrum, born
1968 obtained her
degree in 1993,
Ph.D. in 1997.
Joined COWI in
1996. Project
Engineer, Major
Bridges,
Aerodynamics
Anton PETERSEN
Civil Eng.
COWI AS
Lyngby, Denmark
Anton Petersen, born
1950 obtained his
degree in 1974 and
joined COWI in
1975.
Chief Engineer,
Technical Manager,
Bridges.
Summary
In cable-stayed bridges with small width-to-span ratios the girder becomes inefficient in
transferring lateral loads in bending. Furthermore, the critical load for lateral buckling decreases.
A solution could be to apply a so-called spatial cable system that provides both vertical and
lateral support for the girder.
1.
Introduction
The present trend within design of cable supported bridges moves towards decreasing width-tospan ratios, see Figure 1.1. This lateral slenderness is either the result of a very long span with a
standard deck width or it may be due to an extremely narrow girder used in connection with a
moderate span. The first situation becomes relevant because an increased span does not
necessarily call for a wider deck since the deck width is more a question of requirements for road
and rail traffic. The latter situation could occur in areas with small traffic intensity and thus
limited demands concerning deck width. Taking into account the developments in construction
techniques and cost, it may become feasible from an economical point of view to build these
bridges despite the low traffic volume.
In an earth-anchored system, the lateral wind load is transferred partly by the girder in transverse
bending and partly by the cable system due to the deflection of the cable planes. In a traditional
self-anchored system with vertical cable planes, the wind load has to be transferred entirely by
the girder in transverse bending as there will be no pendulum effect. Furthermore, in a selfanchored cable system the girder is subjected to a considerable compressive normal force
induced by vertical loads.
As the girder becomes more narrow the transfer of lateral loads in bending looses in efficiency.
Adding to the lateral load on the girder itself is also approximately half of the wind load on the
cable system, as each stay cable will transfer half of its wind load to the pylon and half to the
girder. As a consequence, the static and dynamic behaviour of the structure both during
construction and in the completed stage might turn out to be unacceptable, Gimsing (1997) and
Gimsing (1994).
Width-to-Span Ratio
0.25
Completed Bridge
Buchenauer
Erection stage
0.2
0.15
Nord-Elbe
Friedrich Ebert
Theodor Heuss
Severin
0.1
Stromsund
0.05
0
1955
Erskine
Duisburg-Neuenkamp
Maracaibo
Severin (erec.)
1960
1965
Pasco-Kennewick
Sunshine Skyway
2nd Severn
Faro
Rama IX
Luling
Baytown
Annacis
Brotonne
Oresund
Rande
Barrios de Luna
Kohlbrand
Tampico
Ikuchi
Wadi Kuf
Tjorn
Quincy
Karnali River
St. Nazaire
Knie (erec.)
Helgelands
Tatara
Normandie
Skarnsundet
Karnali River (erec.)
Rees Bridge
Knie
1970
1975
1980
1985
1990
1995
2000
Year of Completion
Figure 1.1 Development in lateral slenderness for cable-stayed bridges. Adapted from Larsen
(1997).
A possible solution to problems associated with lateral wind load on cable supported bridges
with small width-to-span ratios is to apply a so-called spatial cable system that provides both
vertical and lateral support for the girder.
Figure 1.2 Architectural model of a cable-stayed bridge

having a spatial cable system.
A full three-dimensional
support of the girder will
require at least three mutually
inclined cable planes forming a
spatial network of cables.
However, to achieve symmetry
four cable planes will generally
be preferable, see Figure 1.2.
Until present pseudo-spatial cable systems have been applied for pedestrian and pipeline
suspension bridges some of these spanning more than 300 m. This paper presents the results of
studies on a prototype cable-stayed bridge with a spatial cable system having an 800 m main span
and a girder width of 8 m. This gives a width-to-span ratio of 1:100 which is close to a factor of
2.5 compared to the width-to-span ratios found in cable-stayed bridges built until present, see
Figure 1.1. Side spans of the prototype bridge are 250 m long.
The investigations are divided into three parts: Analytical analyses and related parametric
studies, FE-calculations and finally a model test. The research was carried out at the Technical
University of Denmark as part of a Ph.D. project, Vejrum (1997).
2.
Analytical Investigations and Parametric Studies
In order to determine the range of inclination of cable planes

that is realistic to consider for spatial cable systems, analytical
analyses and parametric studies on the prototype bridge are
carried out. These show that optimum height of a pylon
supporting a spatial cable system does not differ from what is
found for a pylon supporting a traditional cable system with
vertical cable planes. Thus for the prototype bridge a pylon
height above bridge deck of 120 m is chosen, see Figure 2.1.
120 m
30 m
An evaluation of material cost and of deflections due to wind

load indicates that the pylon width, b, should range from half
the pylon height to the full pylon height. This results in a
lateral inclination of cable planes between 1:4 and 1:2, see
Figure 2.2.
Figure 2.1 Basic geometry of

spatial cable system.
The total cost includes contributions from stay and anchor cables and from pylons. The cost of
the girder is assumed to be independent of the cable system geometry and is not included in the
parametric study on material cost.
Deflections are due to elongation of stay cables and to rotation of the pylon caused by elongation
of anchor cables. Both contributions and the total deflections are shown in Figure 2.2.
2,0
1,6
1,4
Stay and anchor cables
Pylon
Lateral deflections at midspan [m]
1,8
Normalized cost
10
Total cost
1,2
1,0
0,8
0,6
0,4
0,2
Total deflections
Stay cables
Anchor cables
7
6
5
4
3
2
1
Price ratio: r
0,0
0
40
80
120
160
Pylon width, b [m]
200
240
40
80
120
160
200
Pylon width, b [m]
Figure 2.2 Parametric study on prototype bridge. Pylon height fixed to 120 m.
Left: Total cost of structure. Price ratio: Unit price of cable steel / unit price of structural steel.
Right: Lateral deflections due to wind load. Design wind speed at girder level: u = 45 m/s.
240
3.
FE-analyses
General presentation of spatial cable systems

3.1
Four different layouts of the spatial cable system are studied and compared by means of
numerical analyses, see Figure 3.1. Focus is on the behaviour for wind load, in particular with
respect to deflections. The dead load of the girder including railings, surfacing and bridge
equipment is 42.7 kN/m. All cables are designed to have a dead load stress of 450 MPa.
a)
b)
c)
d)
Figure 3.1 Four different layouts of the spatial cable system.

The basic static behaviour for wind load on the four layouts is explained in Figure 3.2. System a)
is considered a "fully spatial (three-dimensional) cable system". We define the term "fully
spatial" as cable systems where lateral loads are transferred by the same cables as are used for
carrying vertical loads. This way, the dead load of the girder and the cables themselves is used to
prestress the cables and thus make transfer of lateral forces possible by increasing or decreasing
the prestress. With system a) the lateral force can be transferred by the cable system without any
aid from the girder.
In system b) the girder needs to have a certain torsional stiffness, because the two cables per section
of cable anchorage do not intersect at the centre of gravity of the girder and thus torsion arises.
Consequently the girder has to transfer torsion from one section of cable anchorage to the next,
where the torsional moment is reversed.
With cable system c) the transfer of lateral forces by the cable system results in local vertical
bending of the girder. This is not expected to cause difficulties, since the girder will have a certain
vertical bending stiffness to allow transfer of vertical loads from the loading point to the sections of
cable anchorage.
Finally, in cable system d) the girder needs to have a considerable torsional stiffness, if lateral
forces are to be transferred by the cable system. In this cable system the torsional moment is not
reversed at the next section of cable anchorage as it was the case for cable system b), so with cable
system d) the transfer of lateral forces by the cable system gives rise to global torsional moments to
be transferred by the girder.
Figure 3.2 Basic static behaviour of four layouts of spatial cable system subjected to lateral load,
U.
As described above, cable system a) is the only system of the four that can be considered as truly
fully spatial. However, it has some disadvantages compared to the other three. Most importantly,
the number of cables is twice the number required for the other systems. This leads to a higher
wind load on the cable system. Furthermore, the process of adjusting cables when mounted is
more time-consuming and complicated when the number of cables is doubled.
Cable system a) and b) have crossing cables which might lead to a relatively complicated
structural detail. Furthermore, the requirements for clearance have to be met. Advantages and
disadvantages of the four spatial cable systems are listed in Table 3.1.
Cable system
Advantages
Disadvantages
a)
Fully spatial (lateral forces can be

transferred without any aid from the
girder)
Larger wind load on the cable system

Adjusting of cables more complicated
Crossing cables
b)
Only half the number of cables

compared to a)
Only partially spatial (local torsion)

Crossing cables
c)

compared to a)
No crossing cables
Only partially spatial (local bending)

Unsymmetrical with respect to bridge
axis
d)

compared to a)
No crossing cables
Only pseudo-spatial (global torsion)
Table 3.1 Comparison of advantages and disadvantages related to the four spatial cable systems.
Deflections due to wind load

3.2
The following structural features of the girder are used in the study:
A = 0.35 m2
Ilat = 2.6 m4
Ivert = 0.31 m4
Itor = 0.92 m4
Girder deflections due to wind load are shown in Figure 3.3. Deflections of cable system b) and
c) are identical. The larger deflections of system a) are related to the higher wind load on the
cable system because the number of cables is the double. Wind on the cable system causes 66%
of the midspan deflection in the case of system a). Wind on the girder accounts for 31% while
3% of the deflection is due to wind on the pylon. The relatively large deflections of the pseudospatial cable system d) are due to global torsion that has to be transferred by the girder. A
parametric study reveals that torsional stiffness of the girder has to be increased by a factor of 10
to reduce deflections to the same level as found for the other three spatial systems.
Studies on the aerodynamic behaviour of bridges with spatial cable systems are presented in
Larsen (1999) and Larsen (1997).
Lateral deflection of girder [m]
Cable system a)
Cable system c)
Cable system b)
Cable system d)
14
12
10
8
6
4
2
0
-700
-600
Abutment
-500
-400
-300
Pylon
-200
-100
100
200
Distance from midspan [m]
300
400
500
Pylon
600
700
Abutment
Figure 3.3 Lateral deflections for wind load on structure. Comparison of four layouts of the
spatial cable system. Note: The girder has no lateral support at the pylons.
Buckling stability
3.3
In Table 3.2 the buckling stability of the girder is compared for the four layouts. The FEcalculation is an eigenvalue buckling analysis where compression is induced by a uniformly
distributed vertical load. The critical loads are normalized by the lowest critical load for vertical
buckling which is identical for all four layouts due to the design criterion for the cable cross
sections.
Cable system
Lateral asymmetric
Lateral symmetric
Vertical symmetric
a)
0.91
0.93
1.0
b)
0.93
0.96
1.0
c)
0.93
0.97
1.0
d)
0.81
0.76
1.0
Table 3.2 Critical loads according to an eigenvalue buckling analysis. Loads are normalized
with respect to the lowest critical load for vertical buckling (a symmetric mode).
The analyses show that the modelled bridge type having an extremely narrow girder supported by
a spatial cable system is not likely to exhibit any stability problems in its completed stage, since
the critical loads for buckling instability equal approximately 12 times the characteristic traffic
load. However, as a distinctive feature related to a bridge having a narrow girder, the critical
loads for lateral and vertical buckling are practically identical. This is in contrast to existing
cable-stayed bridges where the critical load for lateral buckling is significantly higher than for
vertical buckling.
4.
Model Test
Introduction
4.1
A comparative experimental
study on both lateral and
vertical girder instability
phenomena is carried out on a
model of the bridge in the
erection stage. The geometrical
length scale is 1:80. The
parameter to be varied is the
lateral inclination of cable
planes, see Figure 4.1. Seven
tests with different geometries
of the cable system were carried
out.
Figure 4.1 Test setup used in comparative experimental study
The aim of studying both lateral
on bridges with plane and spatial cable systems. The model
and vertical instability with girder is 5 m long.
basically - the same model
influenced fundamentally on the design of the model itself as well as on the test setup and
procedures. The experimental study focuses on spatial cable systems but with two geometries of
plane cable systems as reference tests (laterally free or restrained). Thus in order to facilitate the
change from plane to spatial cable systems the pseudo-spatial system d) was chosen.
Design of the model
4.2
Only vertical load was used to induce compression in the girder. To ensure the possibility of both
vertical and lateral instability phenomena to develop, no restraint - lateral nor vertical - arising
from loading arrangement or measuring equipment are allowed. Load was applied in the form of
plummets placed on scales at each cable set.
In the prototype bridge, the distance between the cable sets is 20 m at the girder. In the physical
model in scale 1:80 this would cause some difficulties regarding spacing and joints.
Consequently, the number of cable sets in the model is reduced to 5.
The expected critical loads for the prototype bridge exceed the yield strength of the stay cables by
a factor of two to three. In order to be able to study instability phenomena experimentally the
load carrying capacity needed to be increased without changing the axial stiffness of the cables. It
was decided to use solid rods and model the axial stiffness of the cables with tension springs.
The needed load carrying capacity was out of range for standard spiral springs available. Instead,
tension springs were constructed from a pile of steel discs working in compression. The discs are
placed inside a steel cylinder and are activated by pulling an axle through their central holes, see
Figure 4.2. Cable forces and in situ stiffness of the cylindrical springs are measured by means of
transducers placed in series with the springs.
Figure 4.2 Principle of cylindrical tension springs.

The aim of the model test requires a quite accurate 3D-measuring method for the girder position.
Furthermore, no lateral nor vertical restraints arising from the measuring equipment are allowed.
These requirements make surveying methods a logical choice.
The girder is modelled by a solid rectangular aluminium profile. Dimensions are chosen in
accordance with the ratio between critical load for lateral and vertical buckling of the prototype
bridge girder. Sectional forces in the model girder are determined from strain gage
measurements.
Test program and results
4.3
At first two reference tests on a plane cable system were carried out in order to provide the range
of the critical loads. Pure lateral instability was observed for the girder without any lateral
support, whereas pure vertical instability was obtained by restraining the girder laterally with
steel wires fixed to the columns of the test setup in order to prevent instability in this direction.
Results for all geometries of the spatial cable system will lie within this range of critical loads.
Then the pylon width or lateral inclination of cable planes was gradually increased with the
intention of increasing the critical load for lateral instability which would finally exceed the
critical load for vertical instability, see Figure 4.3 left.
We believe this is the first time an experimental parametric study on the buckling stability of a
narrow cable supported girder has been carried out. In Figure 4.3 expected and actual test results
are shown.
Critical load
Critical load per scale

300
290
Vertical instability
II
280
Lateral instability
(expected results)
VI
270
260
Lateral instability
III
Plane cable
systems
VII
IV
240
Spatial cable systems
250
230
I
220
210
Laterally free plane system

Laterally restrained plane system
Pylon width, b
200
0
10
20
30
40
50
60
70
80
90
100
Pylon width, b [cm]
Figure 4.3 Left: Critical load and type of instability as function of pylon width (i.e. inclination of
cable planes as the pylon height is kept constant).
Right: Test results. Laterally free plane system, Test I. Laterally restrained plane system, Test II
and VI. Spatial systems, Test III, IV, V and VII.
Test I is the laterally free plane system, while Test II and VI are the laterally restrained plane
system. The model girder had to be changed after Test V and therefore one of the reference tests
was repeated with the new girder. The result of Test III is in very good agreement with the
critical load predicted by FE-calculations. However, from this point no increase in critical load
was observed when cable planes were further inclined.
Measurements of in situ stiffness of the cylindrical springs led to a possible explanation. During
calibration of the springs it was observed that the response was not as stable as intended.
Through the test series the relative stiffness of the two springs forming a set increased with the
stiffer springs located in the same side of the girder in four out of five cable sets. This difference
in axial stiffness induced lateral forces on the girder making lateral instability more critical than
vertical instability. Furthermore, the lateral forces arising from a certain difference in axial
stiffness of the cable elements increase when cable planes are further inclined. This probably
explains why an increase in critical load was not observed for the more spatial systems, Test IV,
V and VII. FE-calculations confirm the destabilising effect due to difference in axial stiffness.
Based on the test results and FE-calculations it seems that the spatial cable system can provide
the necessary elastic support for a girder with a small width-to-span ratio to prevent lateral
buckling of being more critical than vertical buckling when lateral inclination of cable planes is
around 1:4. Thus the requisite inclination of cable planes to reduce the lateral deflections to an
acceptable level also stabilises the narrow girder with respect to lateral buckling that would
otherwise have a lower critical load than for vertical buckling.
5.
Conclusions
Based on the investigations carried out in the present work it is concluded, that arranging a
spatial cable system is a promising way of solving problems related to applying a girder with a
small width-to-span ratio.
6.
Acknowledgement
The Ph.D.-project was carried out with financial support from The Danish Research Councils
(STVF). The advisors Professor Niels J. Gimsing and Associate Professor, Ph.D. Henrik Stang
are gratefully acknowledged for their supervision.
7.
References
[1].
Gimsing, N.J. (1997): "Cable Supported Bridges - Concept and Design", 2nd ed., John
Wiley & Sons Ltd., Chichester, England.
[2].
Gimsing, N.J. (1994): "Suspended Bridges with Very Long Spans", International
Conference on Cable-Stayed and Suspension Bridges, Deauville, France, Proceedings vol.
1, pp. 489-504.
[3].
Larsen, S.V. (1999): "Aerodynamic Performance of Cable-Supported Bridges with Small

Width-to-Span Ratios", Proceedings of the IABSE Conference on "Cable Stayed Bridges,
past, present, future", Malm, Sweden.
[4].
Larsen, S.V. (1997): "Long and Narrow Cable Supported Bridges Subjected to Wind
Load", Ph.D. thesis, Danish Maritime Institute and Department of Structural Engineering,
Technical University of Denmark.
[5].
Vejrum, T. (1997): "Bridges with Spatial Cable Systems. Theoretical and experimental
studies with special emphasis on lateral buckling stability of the girder", Ph.D. thesis,
Series R, No. 19, Department of Structural Engineering and Materials, Technical
University of Denmark.
Design and Construction of a CFRP Cable Stayed Footbridge
Fig. 1
Jens CHRISTOFFERSEN
Civil Engineer, Ph.D., HD
COWI
Lyngby, Denmark
Lars HAUGE
Civil Engineer
COWI
Lyngby, Denmark
Henrik ELGAARD JENSEN

Civil Engineer, Ph.D
COWI
Lyngby, Denmark
John BJERRUM
Civil Engineer
Danish Road Directorate
Copenhagen, Denmark
Artist's impression of Herning footbridge,

courtesy Mller & Grnborg Architects & Planners
Summary
The paper describes a research and development project carried out in order to gain practical
knowledge of the use of non-corrosive reinforcement. As a key element of the project, a
footbridge with total length of 80 metres has been constructed
1.
Introduction
The first bridge in Denmark, and one of the first in the world to be built with extensive use of
Carbon Fibre Reinforced Polymer (CFRP) materials is in the final stage of construction in the
Danish town of Herning. The cable-stayed bridge has one central pylon, dual cable planes and a
total length of 80 m. The bridge will facilitate pedestrians and emergency vehicles crossing a
railway switchyard.
The bridge will be the longest so far constructed by the exclusive use of CFRP stay cables. The
bridge deck is post-tensioned with 6 CFRP tendons, and a 40 meter section of the bridge deck
reinforced with CFRP bars and stirrups. The opposite 40 meter section will be reinforced with a
combination of conventional steel and stainless steel reinforcement.
The bridge is a key element of a R&D project, initiated by the Danish Road Directorate in 1997.
A consulting team, headed by COWI, has been awarded a contract for investigating the possible
use of FRP materials in bridges, and subsequently to design a footbridge with extensive use of
FRP materials. The construction contract for the trial bridge was awarded to Skanska A/S. The
aim of the R&D project is to evaluate the use of non-corrosive materials in bridge construction.
The Danish Road Directorate has an intensive interest in non-corrosive materials, being
responsible for the operation and maintenance of the Danish main road network which includes
more than two thousand bridges. The heavy use of de-icing salts in the winter periods, combined
with frequent freeze-thaw cycles, have rendered traditional reinforced concrete bridges prone to
damage, initiated by reinforcement corrosion.
The most frequent damage to reinforced concrete bridges in Denmark can be linked to corrosion
in bridges parts that are exposed to the chlorides in the de-icing salts. Edge beams and lower
sections of bridge columns are especially prone to damage. The deck reinforcement corrodes
only severely infrequent, due to strict use of bituminous water proofing membranes on the bridge
decks. The water proofing membranes are usually replaced at 25-30 years intervals, and the
associated operations constitute a considerable part of the total maintenance costs and cause
severe traffic restrictions. The substitution of steel reinforcement with advanced composite
materials or high grade stainless steel reinforcement may render the use of bituminous
membranes superfluous and possibly result in lower maintenance costs and less traffic
disturbance during maintenance operations. Whether the total net present value of bridge
construction and life time operation cost will be reduced by use of non-corrosive materials is
influenced by the cost of the alternative materials as well as the increase/decrease in construction
costs due to different aspects within the construction process.
Furthermore, the objective of the R&D project is to obtain experience in using composite posttensioning and stay cables. During the last 30 years a vast number of post-tensioned bridges as
well as cable supported bridges have been damaged due to tendon or cable corrosion world wide.
The possibility to use non corrosive cable materials might be feasible despite the high initial
costs of CFRP materials due to the reduced risk of very expensive cable or tendon replacements
during the service life of bridges.
2.
Bridge Articulation
The bridge primarily consists of a deck, supported by 16 cable stays anchored to a central pylon
as shown in the figure below. The walkway is 3.5 metres wide and the total width of the deck
approximately 5 metres. The dual cable planes support the bridge deck at 9 meter intervals. Due
to aesthetic reasons, the bridge is designed without the traditional back span piers. The
asymmetric live load is balanced by anchoring the outer stay cables to the foundation structure
below the abutment and thereby fixing the top of the pylon. The connection between the CFRP
stay cables and the foundation is achieved by using stainless steel bar tendons, suited for the
exposed position below the bridge deck.
Fig. 2
Elevation and Section of Herning Footbridge. Required clearance for rail tracks shown
below the deck.
3.
Initial R&D project
After being awarded the contract, COWI carried out a thorough survey of literature on FRP
materials and the use of the materials in civil engineering, [97.1]. While carrying out this survey
the team at COWI had very good access to the latest information through the associated team of
experts, acknowledged below. The investigation focused on Glass Fibre Reinforced Polymer
(GFRP), Aramide Fibre Reinforced Polymer (AFRP) and Carbon Fibre Reinforced Polymer
(CFRP). Each of the mentioned FRP materials can be produced by several types of resins of
which epoxy, vinyl-ester and poly-ester are the most common. Epoxy based resin was chosen as
the most promising matrix material due to its very good mechanical and chemical resistance
properties.
Tensile Strength
Young's Modulus
Density
CFRP
1700-3000 MPa
140-300 GPA
1600 kg/m3
AFRP
1200-2100 MPa
50-120 GPA
1300 kg/m3
GFRP
1500 MPa
50 GPA
2400 kg/m3
Table 1. Typical properties of the most common FRP materials

The possible fibre materials were evaluated on a number of factors of which the most important
are shown in the table below.
GFRP
AFRP
CFRP
Environmental
resistance
Tensile strength
++
Fatigue strength
++
Young's modulus
++
Creep/relaxation
++
Stress fatigue
++
Density
++
++
Material price
++
Table 2. Comparison of FRP properties: - not good, 0 neutral, + good, ++ very good.
Considering the material parameters, CFRP was chosen to be the FRP material with the highest
potential for future use in bridges, although the material price of CFRP is astounding at present.
The primary reason for rejecting GFRP was doubt regarding the long term environmental
resistance when cast in concrete, and its proneness to low level stress fatigue. AFRP was
determined to have a low Young's modulus that might may be a limiting factor in bridge design.
It was also noted that most manufacturers are focusing on CFRP and GFRP.
4.
Design of the Footbridge
Prior to the design, a project specific design basis was compiled, as the Danish codes of practice
do not cover structures of advanced composite materials. A design level of reliability
corresponding to "High Safety Class" in the Danish codes was chosen, and the corresponding
partial safety factors determined by probabilistic analyses. Material parameters were obtained
from potential manufacturers.
4.1
Stay Cables
Special attention was devoted to the ultimate limit state, governed by cable stay failure, by
anchorage slipping or by damage due to vandalism.
Anchoring of CFRP cables to facilitate the full use of the extremely high tensile and fatigue
strength is one of the most challenging problems within CFRP cable development. Several
manufacturers have solved the problems caused by the low transverse and inter laminar shear
strength of CFRP with unidirectional fibres. High gradients of shear at the anchors have
notoriously lead to reduction in the static tensile strength as well as in the fatigue strength, but
various manufacturers have solved the problem by developing special anchors, typically of the
socket type, Erki & Rizkalla [1993].
The possibility of damage to the cables by fire/heat or by mechanical damage due to impact or
cutting by hand held tools was identified as a serious risk early in the project. To reduce the risk
of stay failure due to vandalism, the stay cables are protected as shown in the figure below. A
HDPE sleeve is extruded on the free stay cable between the anchors. A 2 mm stainless steel tube
in two parts is clamped onto the HDPE sheath. Finally, the lower part of the stay cables is
protected by a 5mm stainless steel tube that serves the dual purpose of protecting the cables and
reducing the stay movement in the anchor zone. The protective measures are not eliminating the
possibility of damage, but are intended to discourage vandals before major damage occurs.
Regardless of the protective measures indicated above, the bridge has been designed to prevent
collapse in case of abrupt failure of one stay cable or static failure of two adjacent cables.
Fig 3.
CFRP cable stay with protective enclosures. The load cell will be used to monitor stay
forces.
4.2
Bridge Girder
The deck is basically a slab, supported at both edge beams by the stay cables at nine metres
intervals. The weight of the deck is reduced by using seven cut-out ducts. The ducts are
discontinued between the stay anchors to form internal cross beams in the deck at these points.
The design of the CFRP reinforced part of the bridge girder has been considerably influenced by
the intrinsic linear elastic nature of the CFRP reinforcement which would result in a bridge deck
with a brittle flexural failure mode, if a usual reinforced concrete design was carried out. Prior to
the design two philosophies were evaluated. The first was to cope with the brittle failure mode by
raising the total level of safety of the deck. This does not eliminate the possibility of brittle
failure, but reduces the probability of undesired events. The second evaluated method, and
eventually adopted, was originally proposed by Mitsuyoshi et. al. [1993]. It aims at asserting
ductile flexural failure, despite the linear CFRP reinforcement. By confining the compression
zone in the concrete with stirrups, the ultimate strain of the concrete can be raised substantially.
Combining this effect with the amount of tensile reinforcement that ensures an over-reinforced
failure, a ductile failure governed by concrete crushing can be developed, as shown in the figure
below of the original Japanese tests.
Fig 4. Test results of unconfined

(No. 6) and over-reinforced and
confined beams( No. 3 & 5),
Mitsuyoshi et. al. 1993
Fig. 5
7-wire CFRP reinforcement as used in
Herning Footbridge, courtesy Tokyo
Rope Mfg. Co.
The confined and over-reinforced cross sections were adopted in the longitudinal direction as
well as in the internal cross beams between the stay cables. This design is in accordance with the
traditional Danish design philosophy, based on ductile failure modes. However, in comparison
with the design method described first, the increase in cost is considerable due to the elevated
CFRP quantities.
The CFRP reinforcement, longitudinal bars and stirrups, are all 7 wire strands of varying
dimensions, supplied by Tokyo Rope Mfg. Co. The stirrups are shaped at the factory prior to
setting of the matrix resin and cannot be modified on site. In the future, the use of thermo-plastic
resins may result in CFRP reinforcement products that can be modified on site; a most desirable
possibility.
Fig 6.
Typical cross section of the bridge deck showing CFRP reinforcement of type CFCC
from Tokyo Rope Mfg. Co.
While designing the bridge deck using CFRP stirrups, focus was on the reduced strength of the
stirrups due to transverse action at the corners. Test results found in the literature suggested a
strength reduction of up to 60% of the axial capacity of the stirrups due to the transverse action.
To get specific knowledge of the strength reduction of the used stirrup type, an experimental
program was carried out at the Technical University of Denmark. The results indicated that a
reduction factor of 0.4 would be sufficiently conservative, and this factor was applied in the
design in excess to the partial safety factor on the tensile strength of the reinforcement. Further
tests at the Technical University of Denmark were carried out in order to verify the bond and
anchor capacity of the CFRP reinforcement. The 7-wire strands were found to bond excellently
to the concrete due to mechanical interlock.
Fig 7. Deviator for unbonded CFRP tendon in internal cross beam.

The bridge deck is post tensioned by six 7-wire tendons. The tendons are placed un-bonded in
the cut-out ducts with saddle points at the internal cross beams between the stay anchor points.
The post tensioning reduced the amount of plain reinforcement in the longitudinal direction that
was necessary to ensure over-reinforced cross sections. The use of longitudinal un-bonded posttensioning facilitates the structural monitoring as the individual tendons can be replaced in the
future and the original tendons inspected. Use of un-bonded tendons has the added benefit in
comparison with traditional bonded tendons of smeared tendon strain in the ultimate limit state.
Thereby the risk of brittle tendon failure at the point of maximum flexure of the deck prior to
concrete compression failure is reduced.
4.3
Pylon
The pylon is constructed of weathering steel that fits in well at the switchyard area. Due to the
limited free space between the rail tracks, the pylon had to be very slender. This was
accomplished by anchoring the top to the foundations below the abutments by connecting the
outer stay cables with "back stays" below the bridge deck. The pylon is an airtight structure,
assembled in a shop and transported to the bridge site and erected in one piece.
5.
Construction
The bridge is being constructed by Skanska A/S as main contractor. Strict restrictions have been
imposed on the contractor with respect to operations on the rail areas. With the exception of short
shutdown periods, while the contractor performs specific operations, service of all tracks is
maintained during the construction period.
The abutments and pylon foundation were built using conventional reinforced concrete. All
foundations are piled due to soft top layers of ground.
The contractor opted to build the 40 ton steel pylon in a shop and transport it to the bridge site at
night in one piece after securing the necessary permits from relevant authorities. The pylon was
erected by a mobile crane, levelled and bolted to the foundation.
All CFRP materials were delivered by ship from Japan. Stay cables, post tensioning tendons and
longitudinal CFRP bars were shipped on coils, whereas the stirrups were shaped by the
manufacturer and shipped duly secured in wooden boxes. After arrival the CFRP products were
stored in a protective environment to prevent damage from mechanical impact that could occur if
stored as conventional steel reinforcement on the construction site.
The contractor placed the CFRP reinforcement and tendons in the form work and secured the
reinforcement by using plastic strips instead of the usual binding twine. A curious, but not
trifling measure, was to secure the very low density CFRP reinforcement against uplift while
casting the concrete. The operations regarding reinforcement placement and concrete casting and
vibration had been adjusted after a trial casting of a 2 meter section of the bridge deck. The
CFRP reinforcement was protected from damage by encasing the vibrators in approximately 1015 mm of synthetic material. The contractor's staff was prevented from stepping directly on the
CFRP reinforcement prior to setting of the concrete.
The post-tensioning tendons are tensioned by using a hydraulic jack. The elongation of the 80
meter CFRP tendons during tensioning is considerable due to the relatively low modulus of
elasticity. Consequently an adoption bar between the socket type anchor and the jack was used.
In the future, development of permanent wedge type anchors for CFRP tendons would be an
improvement both with respect to the jacking operations and by being more flexible than the
fixed length tendons with socket anchors fitted by the manufacturer. At present the socket type
anchor has superior fatigue strength and was consequently chosen.
The cable stays are placed and tensioned stage wise. The tensioning was carried out based on
geometrical requirements to ensure symmetric loading of the pylon and optimal longitudinal
profile of the deck under dead load.
6.
Structural Monitoring
A program for structural monitoring of the bridge in the coming years has been set up. It has the
dual purpose of ensuring detection, if a severe structural deficiency should occur in the structure,
and to facilitate information about the function and durability of the CFRP components.
The first purpose will be achieved by regular measurements of the stay cable forces, tensile stress
in chosen CFRP reinforcement bars, and precision measurement of the bridge's geometry.
Information about the performance of the CFRP components will be gathered by the following
proposed initiatives:
Replacement of an original stay cable at 5 year intervals followed by close inspection of the
cable and the anchors.
Replacement of an original post tensioning tendon at 5 year intervals followed by close
inspection of the cable and the anchors.
Pull-out tests on protruding CFRP bars to determine long term bond strength.
Measurements on built-in corrosion cells to determine the time at which conventional steel
reinforcement would corrode.
Periodic general inspections of the bridge.
7.
Scope for Future use of Non-Corrosive Reinforcement in Bridges
During the last decade much attention has been devoted to transferring FRP technology from the
aero-space and defence industries to the construction industry. The focus has primarily been on
material characteristics and mechanical behaviour of single structural members. Initially, the
extreme strength to density and stiffness to density ratios were identified and theoretic efforts
initiated to determine the envelope of possibilities by using advanced composites, Meier [1987],
Meier [1992]. Secondly, the very good durability of certain FRP products was identified and use
as substitutes for steel in exposed structural parts proposed. Finally, the use of composites for
post-strengthening of concrete structures was proposed, and experimental programs carried out to
determine feasible operation and design methods.
So forth the most successful area of FRP use in the construction industry has been post
strengthening of concrete structures by bonded laminates for flexural strengthening of beams or
by improving the ductility and capacity of concrete columns by column winding methods, Sieble
[1995]. In these areas, FRP solutions have proven to be able to compete with traditional solutions
within technical and economic terms. The successful FRP use for post strengthening structures
can be tributed to devoted research, large scale trial projects and development of commercially
available products with associated design and operation guidelines. Finally, the post
strengthening methods, bonded laminates as well as column winding, take advantage of
situations at which alternatives are costly. Consequently, the high material cost does not prohibit
the use.
A similar path may be feasible for other FRP applications. The present Danish trial project is an
opportunity to gain information on all aspects of FRP-use in bridges from design to construction.
Based on knowledge from this and other trial projects, advanced composites could find their
place in bridge engineering. The interesting challenge at present is to develop a FRP application
that not only substitutes steel but takes advantage of the composites' intrinsic parameters in
structural solutions that conventionally are either expensive or technically deficient.
8.
Acknowledgements
The financial support from the Danish Agency for Development of Trade and Industry
(Erhvervsfremme Styrelsen) is kindly acknowledged. Furthermore the valuable technical and
scientific support from professor Atsuhiko Machida (University of Saitama), professor Urs Meier
(EMPA), professor Frieder Seible (UCSA), professor Henrik Stang (Technical University of
Denmark), Ris National Laboratories and Fiberline Composites A/S are acknowledged.
9.
References
[1].
[97.1] 'FRP materials for cable stays, prestressed- and plain reinforcement in
Danish Road Directorate, 1997.
bridges',
[2].
Meier [1987] 'Proposal for a Carbon Fibre reinforced Composite Bridge Across the strait
of Gibraltar at the Narrowest Site', Proceedings Institute of Mechanical Engineers, Vol. 01
No. B2, 1987.
[3].
Meier [1992] 'Carbon Fiber-Reinforced Polymers: Modern Materials in Bridge

Engineering', Structural Engineering International, No. 1, 1992.
[4].
Erki et. al.[1993] 'Anchorages For FRP', Concrete International, June 1993.
[5].
Seible et. al.[1995]'Earthquake Retrofit of Bridge Columns with Continuous Carbon Fiber
Jackets - Vol. 2, Design Guidelines', Advanced Composites Technology Transfer
Consortium, Report No. ACTT 95/08, UCSD, 1995.
Erection of the Uddevalla Bridge

Petter FALLER
Production Mgr Steel Constr.
Alfr. Andersen A/S,
Oslo, Norway
Carl HANSVOLD
Eng. Mgr
Johs Holt A/S
Oslo, Norway
Helge NILSSON
Design co-ordination Mgr
Skanska Teknik AB
Gteborg, Sweden
Per-Ola SVAHN
Design Mgr
Skanska Civil Engineering AB
Gteborg, Sweden
Summary
The Uddevalla bridge, located on the Swedish west coast close to the town Uddevalla, is part of
the motorway E6 between Oslo and Gothenburg. Construction of the bridge started in mid 1997
and scheduled completion is summer 2000. This paper gives a description of the bridge and the
erection methods. Emphasis has been put on erection of the bridge superstructure.
Introduction
The E6 highway between Oslo in Norway and Malm in the southern part of Sweden has during
the last ten years been continuously upgraded to a motorway standard. The highway acts as an
important link for the communication from Scandinavia to Europe.
One critical stretch is the by-pass of Uddevalla, which is located in the end of a deep fjord. The
existing road passes east and through the central part of Uddevalla. After extensive
investigations during the last thirty years it was decided that the new road will pass west of
Uddevalla.
The new route is approximately 9 km and will save 12,8 km in total length for the road E6.
The route passes through very sensitive locations. The southern side of the fjord has both
archaeological and geological values that are classified as of national interests. At the northern
side of the fjord there is a location of established dwelling houses, which partly interfere with the
suggested route.
The Swedish Road Administration performed a conceptual design and the tender work started in
May 1996. The tender was an alternative design proposed by the Swedish contractor
SKANSKA. The design-build contract was signed in January 1997. The detailed design of the
bridge is carried out by SKANSKA Teknik AB, a subsidiary of SKANSKA AB, in co-operation
with the Norwegian consulting firm Johs. Holt A.S.
2
2.1
Description of the Bridge

Overall configuration
The bridge is a high-level bridge of total length 1712 m carrying 4 traffic lanes, figure 1.
The central cable-stayed section provides a navigation clearance 190 m wide and 52 m high over
the Sunninge Sund. The approach bridges to either side of the cable-stayed section at the centre,
have a total length of 506 m at the south side and 434 m at the north side. The spans increase
from 50 m at the north abutment to typical 88 m towards the centre. The cross-section of the
superstructure is constructed using two separate steel box girders with concrete deck cast in situ.
The central cable-stayed bridge is made up of a 414 m main span and two 179 m spans either
side. The cables are arranged in slightly inclined cable planes nearly parallel to the tower legs.
They are anchored at equidistant intervals along the bridge deck of 13,32 m except for the outer 3
back stays which are concentrated at the anchor piers N5 and S5.
Figure 1. Elevation of the bridge

The entire structure is continuous with expansion joints only at the abutments. Continuity
between the approach bridges and the cable-stayed bridge is provided by a heavy concrete
transition structure. The six piers at axes N5-N7 and S5-S7 are hinged to the bridge
superstructure and contributes to the stability of the bridge in the longitudinal direction. All
other piers are equipped with sliding bearings.
2.2
Bridge superstructure
The bridge cross-section in the stayed spans, figure 2, is a composite structure of an open steel
grid and prefabricated concrete slab elements. The wind nose, connected to the outer
longitudinal I-beam, is a load bearing thin walled shell structure. The stay cables are directly
connected to the web of the 1,7 m high longitudinal girders. The slab elements are spanning
longitudinally and are connected by loop reinforcement in the cast in place joints. The thickness
of the concrete elements is 240 mm and longitudinal reinforcement of 20 mm bars at spacing 160
mm is generally provided to achieve satisfactory strength and limitation of crack widths to
0,20 mm.
Figure 2. Bridge cross-section.

2.3
Towers
The towers are made of concrete grade K55 according to

the Swedish Standard BBK 94. They are diamond shaped
and rise to elevation 140, figure 3. At the tower top the
stay cables are anchored inside steel boxes fixed to the
concrete by shear studs. The tie-beam between the tower
legs is fully post-tensioned for the outward thrust from the
tower legs.
2.4
Stay cables
The stay cables consists of 22 to 77 strands (15,7 mm),

individually galvanised, waxed and sheathed. The bundle
of strands is covered by an external HDPE pipe.
The void between the strands and the external pipe is
maintained empty. The stay cables and anchorage system
is delivered and installed by VSL.
Cable vibrations have been a problem on several cablestayed bridges. To our knowledge, no well documented
theory exists for evaluating the risk for such vibrations.
Thus the following precautions have been taken:
Figure 3. Tower S1
The external HDPE pipes have ribs to prevent so called rain-wind vibrations .
The cables will be equipped with a friction damper attached to the lower cable anchorage.
The cable system shall allow for future installation of transverse stiffening ropes if deemed
necessary.
Overall erection scheme of the bridge
The erection of the bridge comprises the following main operations:

1. Foundation work and construction of abutments, piers and towers using climbing formwork.
2. Launching of the approach bridges and casting of the concrete slab.

3. Construction of the transition structure at pier N5 and S5.
4. Erection of the cable-stayed bride. Connection to the transition structure.
5. Connection of the approach bridges to the transition structure. The entire bridge is now
continious.
6. Completion works such as installation of concrete barriers and pavement of asphalt layer.
The situation at the bridge site in autumn 1998 is shown in figure 4.
Figure 4 The bridge site in autumn 1998

The actual design of the main bridge superstructure needs several of different professions. Steel,
concrete and cables need skilled workers and normally different contractors have to co-operate
closely during the erection. Furthermore close co-operation with the designer is necessary during
all phases of erection. For the Uddevalla bridge the erection is organised as follows:
The main contractor SKANSKA is responsible for the overall co-ordination of the erection.
They deliver the concrete panels for the bridge deck and perform the casting of the joints on site.
The steel contractor Alfred Andersen delivers the steel grid. On site they are responsible for the
lifting works with the derrick cranes and welding of the steel sections together. The cable
contractor JV Internordisk Spnnarmering/VSL delivers anchors, staypipes and strands. On site
they are responsible for the erection of the cables.
Erection of the approach bridges
The superstructure of the two approach bridges comprises a steel box girder with a composite
deck of concrete. The steel box girders are fabricated in 16 21,5 m long pieces with weights up
to 82 tons. They are welded together on the site behind respectively abutment and launched to
the final position. The deck is casted in-situ with a movable formwork.
The spans of the approach bridge vary from 50 m to the most typical 88 m span.The bridge is
performed in a horizontal radius of 1750 m, which have a significant influence on the launching
work. The steel box girder is equipped with a launching nose, shaped with a slope versus the
steel girder to take care of the deflection of the cantilever during the launching. The length of the
nose is determined by the strength of the steel box. The actual length of the nose is 30,5 m,
which gives a maximum deformation of the nose tip of approximately 2,1 m during the
launching.
The launching is performed on temporary sliding bearings, which are placed on the permanent
neotopf bearings. The sliding bearing has a length of 1,2 m and consists of a steel plate with
sliding plates of PE-material on top. The painted steel girder slides direct on these plates. A
critical point for the steel girder is the risk for buckling in the web due to the patch load from the
bearing. The web has been temporarily reinforced near the lower flange over a length of
approximately 24 m, where the maximum reaction forces will appear. When the steel girder is
launched to its final position the temporary sliding bearings are removed.
The deck is casted in-situ with a movable formwork in steps of 22 m. The height between the
ground and the bridge deck together with the need of an efficient production lead to a design
with a continuos casting from respectively abutment. To avoid high tensile stresses in the
concrete above the piers, where the girders are continuos, the steel girder was pre-loaded with a
weight of 300 tons. The load was placed to give a negative bending moment in the steel girder
during casting and hardening of the deck above these piers. This method has been performed
with a good result.
Launching of steel box girder is shown in figure 5.
Figure 5 Launching of steel box girder
Erection of the cable-stayed spans.
The superstructure of the cable supported spans is constructed by the balanced cantilever method.
Two starter segments on either side of the tower are first constructed on fixed scaffolding. After
tensioning of the cables the scaffolding is removed and the lifting derricks assembled. The
standard erection cycle, consisting of 5 main stages, then follows:
A. A steel grid of length typical 13,32 m, width 26,06 m and weight about 70 tonnes consists of
two longitudinal edge-beams, three cross-beams and two cable anchorages. It is lifted by the
derrick and temporarily fixed to the previous steel section. After control of local geometry in
elevation and plan welding of the main beams can begin.
The steel grid is a rather flexible structure, and auxiliary bracings are used to ensure correct
geometry during lifting and installation. The bracings also provide the necessary lateral
support of the slender cross-beams when loaded with the concrete elements.
B. The cables are installed and stressed to a first stressing stage.
C. The derrick lifts the concrete elements in place. Geometry and cable forces are checked, and
possible adjustments performed.
D. The joints between the concrete elements and the edge beams are cast.
E. When the concrete has reached a compressive strength of 25 MPa the cables are stressed to
their final length. The derrick is then moved in position for lifting of the next segment.
The cables are stressed at the tower head in two steps, at stage B and E, using monostrand jacks.
In addition, a final tuning of the back-stays will be performed in parallel with completion works,
e.g. installation of concrete barriers and pavement of asphalt layer.
At stage C the rather flexible steel grid is loaded by the dead load of the concrete elements and
the cable forces applied at stage B. The cable forces have to be carefully determined such that no
harmful deflections or stresses are built into the deck section when composite action is
established at stage D. Stage C is thus an important control stage as regards geometry and cable
forces.
The first stressing operation will be based on measurement of force. For the second stressing and
other cable adjustments, both cable forces and elongations (jack stroke) will be specified with
the latter as the prime control parameter.
For the stages described above local geometry and cable forces will be surveyed. For every third
segment a complete survey of the entire cantilever will be carried out.
A detailed analysis of all structural systems during erection and for the completed bridge has
been performed. The analysis takes into account the various loads at each erection stage, the
cable forces and the effects of temperature, creep and shrinkage. Procedures for survey
containing all necessary data for construction as well as data for easy corrections have been
established. Exchange of data between the bridge site and the design office in Oslo will be based
on electronic mail transmission.
The bridge deck is temporarily fixed to the tower during free cantilevering. Extensive analyses
of wind buffeting show that additional supports of the cantilevers are necessary. Temporary
struts to the ground are installed at a distance of 40,5 m to either side of the tower in order to
reduce the wind-induced bending of the tower in the vertical plane. At cantilever length
124,86 m these struts are replaced by a new support at distance 107,1 m from the tower on the
landside. This support takes vertical and horizontal forces thus reducing both vertical and
transverse oscillations of the cantilever. It also allows for vertical jacking of the bridge deck in
order to ease the connection of the bridge to the transition structure at pier N5 (S5).
Stiff steel supports were selected instead of more simple tie-down solutions. One reason for this
is the rather strict project specifications allowing only limited tensile stresses but no cracking of
the towers during the entire construction period.
After connection of the main bridge to the transition structure at pier N5 (S5) erection of the
remaining segments in the main span follows. Before closing at midspan the cantilevers are
jacked approx. 75 mm in outwards in order to compensate for creep and shrinkage effects.
Concluding remarks
The construction work is now well under way. The situation on site in beginning of February is
that launching of the steel box girders of the approaches is nearly completed, and about 50% of
the concrete slab has been cast. The north tower N1 was completed in Dec. 1998 while the south
tower S1 will be completed in March 1999. Erection of the main span is in progress. Scheduled
opening of the bridge for traffic is May 2000.
Acknowledgements
Owner:
Contractor:
Steel contractor:
Stay cables:
Structural design:
Swedish Road and Bridge Administration, Region West.

Skanska Civil Engineering AB, Bridge Department.
Alfred Andersen mek. verksted & stberi A/S.
Internordisk Spnnarmering AB / VSL International Ltd.
Skanska Teknik AB in co-operation with Johs. Holt A.S.
Cable-Stayed GFRP Footbridge across Railway Line

Mikael W BRAESTRUP
Dr. Eng.
RAMBLL
Copenhagen,.Denmark
Mikael W Braestrup, born 1945, is a

Senior Consultant at the RAMBLL
Bridge Department. Current
assignments include the 16 km
resund Link between Denmark and
Sweden, in particular the design basis,
involving application of the Eurocode
system.
Introduction
The use of fibre reinforced plastic (FRP) in bridge building is fairly novel. The first European
example is the 113 m long Aberfeldy Footbridge, a cable-stayed bridge spanning 63 m over a
stream on a Scottish golf course, installed in the early nineties.
In June 1997 a new FRP bridge for pedestrians and cyclists was opened at Strandhuse near the
Danish town Kolding. As the first advanced composite bridge in Scandinavia it has the further
distinction of being the first FRP bridge crossing a busy railway trunkline. The cable-stayed
bridge is constructed entirely of glass fibre reinforced plastic (GFRP), and it is the result of a
collaboration between a local producer of pultruded GFRP profiles, a major consulting
engineering company, and a public owner who was willing to consider an innovative
replacement of a footbridge removed due to the increased clearance profile resulting from
electrification of the railway. The erection of the bridge was carried out in just 18 hours. The
total capital costs are 5 -10 % higher than alternative designs in steel or concrete, but this is
offset by the resistance of the GFRP material to water, frost, and de-icing salts, implying that
cosmetic maintenance only is envisaged for the next 50 years. The bridge, known as the
Fiberline Bridge after the producer, is shown in Fig 1.
Description
The width of the bridge is 3.2 m, and the length is 40.3 m, with spans of 27 m and 13 m. The
stays are 100 x 100 mm2 GFRP cables. The 1.5 m deep girder and the 18.5 m high,
asymmetrically placed pylon are constructed from standard GFRP profiles bolted together.
The 12 different profiles used are shown in Fig 2, and Fig 3 shows the bridge seen from
below. The only steel components are the bolts and the inserts in the concrete foundations.
The total weight is 12.5 t, less than half of a corresponding steel structure.
Fig. 1. Fiberline Bridge at Kolding, Denmark
Fig 2. Standard Profiles
Fig 3. Underside of bridge
Design
The bridge is designed for a live load of 5
kN/m2 plus a 50 kN moving point load,
representing a snowplough or the occasional
ambulance. In accordance with Danish code
tradition limit state design based upon the
partial coefficient method is used. Thus the
load factors are 1.0 on dead loads and 1.3 on
live loads. The material partial coefficients on
strength and elastic modulus are 1.8 and 1.9,
respectively. The two pairs of stays on either
side of the pylon minimize the deflections
Fig 4. Design Manual
which would otherwise result from the low
stiffness of the GFRP material.
A general Design Manual, see Fig 4, for structures in pultruded FRP profiles has been
developed, including shaping and connection, and resistance to chemical attack and fire
exposure.
Manufacture
Fig 5. Pultrusion of GFRP Profile
Fig 6. Connection by Bolting
The GRFP stays, as well as the profiles used for bridge girder and pylon, are produced by
pultrusion. Pultrusion is a continuous process whereby the fibreglass reinforcement is pulled
trough a permanent form, into which the polyester resin matrix is injected, see Fig 5. The
matrix is cured by means of catalyst addition and heating. The profiles are cut to length and
shaped by ordinary hand tools. The profiles may be coloured as desired, but an aesthetically
more pleasing finish is obtained by surface coating, and light grey and blue tones were chosen
for the bridge. Assembly was carried out by stainless steel bolts, see Fig 6. Bolting is clearly
not the most suitable joining method for the highly anisotropic pultruded profiles, and glued
connections are under development. The strength, stiffness and bolt pull-out strength of the
profiles are subject to regular testing. Fig 7 illustrates the testing of a beam for flexural
properties.
Fig 7.
Test Rig for Flexural Testing
Installation
Due to the continuous operation of the railway, construction activities on site were only
allowed during 8 hours of the night between Saturday and Sunday. The bridge was
manufactured in three pieces, comprising the pylon and the two girder spans. Fig 8 shows a
bridge section being loaded onto a trailer for transport to the bridge site. The erection took
place during three night sessions, see Fig 9.
Fig 8. Bridge Section
Fig 9. Erection of Bridge
Cost
Fiberline has compared the actual costs of the bridge with corresponding designs in steel and
concrete, see Table 1 below. The conversion to Euro is done at the rate of DKK 7.50 to one
Euro. For the FRP option the development costs are included under design.
Costs in 1000 Euro

Design
Foundations
Materials
Shaping and Joining
Erection
Surface Treatment
Miscellaneous
Total
FRP
55
55
105
55
25
10
25
330
Steel
25
70
15
80
55
25
35
305
Concrete
20
80
80
80
15
35
310
Table 1. Cost Comparison

Monitoring
Fig 10. Mounting of Strain Gauges
Fig 11. Monitoring from Control Station
The fact that the bridge crosses a busy railway line causes sharp focus on safety and
reliability, and a system is established to monitor structural stresses and deformations due to
changing loads. Key bridge components, including the stays, are fitted with strain gauges, see
Fig 10, wired to a permanent control station, shown in Fig 11. Universities are invited to use
the bridge as a test site. To monitor temperatures, wind speeds and directions a weather
station is installed at the top of the pylon, see Fig 12.
Fig 12. Weather Station on Top of Pylon

Conclusion
Although data on long-term performance is still outstanding, the Kolding experience indicates
that FRP is a viable material for minor bridges with a premium on swift erection and minimal
maintenance. To make full use of the potential of this novel structural material it is pertinent
to develop technology for gluing of profiles, as well as new architectural forms.
Ting Kau Cable Stayed Bridge: Challenges in the Construction Process

Don BERGMAN
Buckland & Taylor Ltd.
Hong Kong
Introduction
The Ting Kau Cable-Stayed Bridge in Hong Kong provides a vital new link between the Western
New Territories and the expressway linking Kowloon with the new Chek Lap Kok Airport on
Lantau Island. The US$225 million design construct contract for the bridge was awarded by the
Hong Kong Highway Department in August 1994 with completion scheduled for July 1997. The
bridge opened to traffic in May 1998.
Fig. 1: General Arrangement

The bridge is a light and innovative structure spanning the 900 m wide Rambler Channel. At
1177 m in length, the bridge is one of the longest cable stayed structures in the world and is one
of only a few multi-span cable-stayed bridges in existence (Fig. 1).
The design has several unique features which are responsive to the difficult site and schedule
constraints for the project. These design features however created specific challenges for the
construction team:
Monolithic steel tower heads which house the upper cable stressing anchors permit shop
fabrication of these complex elements under controlled conditions. (require 190 tonne heavy
lift into place at the top of the 200 m towers).
Stabilizing stays with deck level struts provide the necessary transverse stiffness and strength
for mono-leg towers which could be easily slip formed. (require heavy lift of 50 tonne struts
and load transfer to stabilizing stays)
The mono-leg tower places a minimal footprint in the Rambler Channel where foundation
and ship impact criteria are onerous. (requires temporary stabilization of slender mono-leg
tower during construction)
Twin decks with intermittent transverse cross connecting members and four stay planes
(Fig. 2) minimize the weight of the transverse deck members and provide excellent
aerodynamic performance in the severe wind climate. (geometry and detail difficulty with
four girders, simultaneous installation of four stays)
Longitudinal stabilizing cables (LSCs) diagonally connecting the top of the main tower to
the deck at the side towers, provide longitudinal stiffness and capacity to the balanced central
main span cantilever. (the LSCs are the longest cable stays erected to date, the Main
cantilever is longest balance cantilever erected to date)
Deck grids could be shop prefabricated and preassembled under controlled conditions to deal
with the complex 3D deck and cable anchorage geometry which includes a superelevated,
curved, variable width deck at the Ting Kau side span (requires quality prefabrication and
preassembly)
Fig. 2: Deck Arrangement

The ability to prefabricate and preassemble had the greatest potential to impact construction.
Ideally it would permit fast efficient erection with relatively unskilled crews. In order to achieve
this however it was necessary to obtain high quality steel fabrication from the start. The
components had to be fabricated and preassembled in the correct sequence, with assurance that
quality welding and geometry control was being achieved.
Steel Fabrication
Success relied on the choice of a steel fabricator with the necessary qualifications and facilities.
That choice in this case was based primarily on economics. The fabricator had relatively little
relevant bridge experience and the outcome of the project was largely determined by this
fundamental decision.
The first and most critical elements to be fabricated were steel tower heads which are large
complex pieces with many heavy restrained welds requiring carefully planned weld procedures
and sequences (Fig. 3). As fabrication progressed testing revealed weld cracking. A time
consuming program of NDT, repair and reinforcement substantially delayed the delivery of the
tower heads. The slip formed Ting Kau tower, which was completed first, waited six months to
receive the first tower head.
Under the original schedule, erection of the tower heads for the three cantilevers was to be
staggered. This spread the resource demands at the three cantilevers and permitted the re-use of
crews and equipment. To recover the lost delivery time for the tower heads, it became necessary
to reduce the scheduled stagger making additional heavy lift equipment necessary.
Contractual payment clauses for fabrication encouraged early completion of the simplest, highest
tonnage deck components, the cross girders, even though most were not immediately required for
pre-assembly and erection. Delay in the preassembly of the first deck grids resulted. These grids
2
were subsequently shipped with many of the secondary fixtures having not been preassembled.
The result was costly and cumbersome assembly of these elements in place.
The fabricators organization made use of multiple levels of subcontractors. Control of the work
was difficult. The lower level subcontractors tended to feel no direct responsibility to the main
sub contractor if more than one level separated them.
Pre-assembly of the steel deck grid presented unique
challenges. Due to variations in deck width, cross girder
spacing, cross girder length, and deck slab thickness, the
camber of the cross girders varied significantly. It was
necessary to preassemble the deck unstressed, i.e. with no
deck concrete in place. If one imagines a complete deck
with the concrete slab in place, with correct as-built girder
and cross girder geometry, and then imagines removing all
of the concrete deck slab to unstress it, the result would be
the theoretically correct unstressed geometry. The
difficulty is that the cross girders which have different
cambers want to deflect and end rotate by differing amounts.
The result was an unstressed deck grid where the girders
are trying to restrain the differing cross girder rotations. The
unstressed grid would therefore have built-in moments in
the cross girders and built-in torques in the main girders.
This unstressed geometry is in fact stressed and cannot be
achieved in preassembly without unreasonable effort.
Ultimately a truly unstressed preassembly geometry was
Fig. 3: Tower head
derived. The resulting preassembly geometry was a
compromise which attempted to minimize the warping of the main girders, the amount of error in
final cross girder camber, and the error in target girder profile.
Welding quality control was thrown into question late in the fabrication after defects were found
in previously accepted welds. Many of the critical welds such as those for the connection of the
cable anchorages to the main girders were found to be defective and a large and costly program
of weld repairs was required some of it on steel which was already erected.
Tower Head Erection

The first steel erected was the tower heads which provide the stressing anchorage at the top end
of the stay cables. The steel tower heads are boxes up to 31 m in height, 4 m in width and 1.5 m
in depth. After lifting and positioning, the tower heads are concreted in and stressed to the tower
top. The tower heads weighed 150 tonnes each for the Ting Kau and Tsing Yi towers and 190
tonnes for the Main central tower.
The tower heads were lifted into place using a pair of 200 tonne heavy lifting strand jacks on
beams mounted on the tower top (Fig. 4). The beams carried the jacks on a sliding cross frame at
each end. The cross frames allowed the tower head to be slid into position for concreting and
stressing onto the tower top.
The tower heads were transported to the site by water, lifted

onto self propelled heavy load platform trailers which were
used to position and hold the unit while rotating to the vertical
(Fig. 3).
Fig. 4: Tower head heavy lift
Lifts began only after confirmation of a three day site specific

wind forecast for winds less than 14 m/s. A secondary cable
wind restraint system was used to stabilize the tower head
during the lift. Prior to lifting, tower head weights and centers
of gravity were confirmed. The as-built tower head weights
tended to be greater than the design weights in part as a result
of the repairs and reinforcement noted above. The lifting
frames were designed for the theoretical tower head weight and
erected by the time the as-built weights were known. As a
result it was necessary to re-design the wind restraint system for
the heaviest lift at the Main tower to minimize any additional
vertical wind loads applied to the lifting frame.
Prior to lifting the Main tower head, inspection revealed

cracking in several of the welds for the lifting beams. A detailed program of NDT and weld
repairs was implemented prior to lifting.
Delays in the delivery of the tower heads allowed a significant layer of corrosion to build up on
the lift strands. While there was no damage or significant section loss, the corrosive layer had to
be power washed off the strands prior to lifting so that the anchor wedges would not fill and slip
during the repeated gripping and regripping required during the lift.
Starter Panel Erection

Following preassembly, the starter panel grids were
transported to the site and positioned at the tower base.
Using the tower top heavy lifting gear again the grids
were rotated to vertical, lifted to final deck level,
lowered into deck level hinges on falsework, and rotated
to horizontal (Fig. 5). The grids were lifted with the
transverse bearings attached to the inside of the interior
girders. To orient the interior girders for splicing the
cross connecting girders and grouting the bearings to the
tower, the outside girders were temporarily supported
below final elevation.
Precast panels were then installed, the starter panel stays
installed and the heavy lift strands released. Large
vertical downwards displacements of the starter panel
Fig. 5: Starter panel erection
were expected when the lifting equipment was erected
on the starter panel and the adjacent deck grids were
lifted. The starter panel stays were therefore temporarily installed short of final length to lift the
starter panel and provide vertical clearance between the panel underside and the tower transition.
Care was taken in positioning and squaring the starter panel grids to match the preassembly
geometry prior to erecting the precast panels. Errors in the starter panel plan geometry would be
passed on to all subsequently erected panels either side.
Tower Strut Erection

Once the starter panels were erected heavy lifting beams and strand jacks were mounted on the
starter panel to lift the horizontal struts for the transverse stabilizing cables (Fig. 6). A pair of
precast panels were left out of the starter panel deck in order to pass jacking strands through the
deck to the strut. As with the tower heads, the jacks were positioned on sliding cross beams
allowing the strut to be correctly positioned for grouting and stressing onto the tower leg.
The struts were released from the heavy lift gear
as soon as a minimum number of strands for the
transverse stabilizing stays were installed. This
permitted removal of the heavy lift gear and start
of erection of the derrick supports (sleds) and the
derricks themselves.
A sequence of strand installation for the transverse
stabilizing stays was chosen which minimized the
movement of the tip of the strut and permitted
installation directly to force. Specific final forces
are required in the stabilizing stays such that under
typhoon, the leeward stays will just come slack. It
was necessary to calculate installation forces
which after all elastic, creep, and shrinkage
shortening of the tower had taken place, would
result in the specified final force.
Fig. 6: Strut heavy lift
During construction the transverse stabilizing stays initially exhibited significant rain/wind
vibrations. The vibrations were temporarily restrained using nylon ropes connected some
distance up each stay. A system of permanent dampers was subsequently designed, fabricated,
and installed just after the opening of the bridge. It is interesting to note that these were the only
stays on the bridge which did not have HDPE sheaths with raised spirals to inhibit rain wind
vibrations.
Derrick Cranes
Deck mounted stiff leg derrick cranes were chosen for lifting steel grids and deck panels.
Derricks do not rely on counterweights for stability but rather tie down to and use the deck
structure itself to resolve the lifting reactions. The arrangement is therefore light and minimizes
the bending effects imparted to the deck by the construction loads. Derricks are typically
mounted on a sled structure which distributes the reactions to the deck framework and also
facilitates movement of the derrick after completion of each new deck grid (Fig. 7).
The twin deck arrangement of the bridge made it necessary to use a pair of derricks at each
cantilever tip. Twelve stiffleg derricks were therefore required for the three towers. Erection of
the sleds and derricks, load testing, and commissioning of the derricks was directly on the critical
path for the project. This process took 7 weeks for the first pair of derrick cranes. The last pair
was completed in 4 weeks. Because the process of erecting and commissioning the derricks was
so time consuming, the decision was taken to eliminate the last pair intended for the Ting Yi land
span. Instead a crawler crane with 66 m of main boom and 54 m of luffing jib was positioned on
the approach road in the Tsing Yi land span and used to erect the land span superstructure. This
decision reduced the time to erect the Tsing Yi cantilever by approximately 4 weeks.
After each segment was completed the derricks were
released from the deck, lowered onto Hillman rollers
positioned over each cross girder, and pulled ahead to the
next position using hydraulic tirfors. Release, lowering and
preparation to roll would take approximately 4 to 6 hours.
Rolling, and tiedown for the next segment would take
approximately 8 to 10 hours. The required strength of the
newly cast infill joints was 20 MPa for rolling and 30 MPa
for lifting.
In total, erection of each starter panel, including the derrick
cranes which were needed to begin the typical deck erection
sequence, took 18 weeks for Ting Kau, 19 weeks for Main
and 11 weeks for Tsing Yi which used only two derricks.
Note that in order to make room for the second set of
derricks at Main and Ting Kau, a second deck grid needed to
be erected and the first derricks rolled onto it. Complete
erection of the three cantilevers including the starter panel
durations above, took 37 weeks for Ting Kau (22 grids), 34
Fig. 7: Derricks at starter panel
weeks for Main (39 grids) and 25 weeks for Tsing Yi (24
grids). Approximately one half the time to erect each
cantilever was therefore consumed by the start up process. This is an aspect of this type of
project which does not always receive due attention, but where full optimization can yield
substantial schedule benefits.
Extensive efforts were put into insuring safe operation of the derricks. These included load
testing, operator training, instrumentation for load and reach limits, detailed checks and signoffs
for derrick translation, tiedown, and operation, and certification by the Hong Kong authorities.
Despite these efforts, a serious accident occurred during erection of the last sea side grid for the
Tsing Yi cantilever. A failure occurred in the derrick sled which resulted in the grid being
dropped to the barge below. The grid was destroyed and the barge severely damaged.
Fortunately no workers were injured. The failure was traced to a missing weld detail, which was
hidden inside a box beam anchoring the rear derrick tiedown. The accident occurred on 3
January 1998. A replacement grid was fabricated in Spain, flown piece small to Hong Kong, and
erected by 27 February 1998. The last closure between the Main and Tsing Yi cantilevers was
delayed slightly as a result, but by carrying out some of the finishing work such as placement of
the waterproof membrane and asphalt surfacing prior to closure, no time was lost on the overall
schedule.
Typical Deck Erection Cycle

The critical path for erection of the superstructure ultimately passed through the Main cantilever,
which is the longest of the three. The challenge was therefore to minimize the erection cycle
times there. Because full welded grids were erected on both ends of the cantilever and lifting off
the water at both ends made delivery relatively simple, it was possible to consistently produce
four day erection cycles (Fig. 8). This was one of the great successes for the project. Operations
were carried out over a twenty four hour period where necessary.
Fig. 8: Typical four day deck cycle
Cycle times for the piece

small grids in the Ting Kau
and Tsing Yi land spans
were typically longer
ranging between 5 and 8
days. The longest cycle
times were required for the
Ting Kau land span where
material delivery to the
derricks was complicated
by the terrain, the wide
deck and the large
members.
Stay Cable/Precast Panel Installation

After each steel grid was erected and bolted to the previous grid, the next step was to install stays
and precast panels. One precast panel and a stay installation platform was first installed over
each of the four girders. This provided the work area for the stay crew. The platforms carried a
cutting table, a deviator wheel for guiding the strands into the stay sheath and a winch for
returning the pulling cable back to the deck after each strand.
The HDPE sheath containing the master strand was first erected and the master strand tensioned
to force. The installed length was then checked against the reference length. All master strands
were carefully measured and marked when cut by the supplier. If the length confirmed the
tension the remaining strands were pulled into place and stressed to the master strand tension by
matching the monostrand jack pressure that of a load cell remaining on the master strand.
The stay installation procedure for a composite deck must insure that the correct non-composite
girder bending moments and geometry are locked in when the joints are cast. The procedure
which requires simultaneous installation of precast deck panels and cable strands must also
insure that no overstresses occur in the girders at any intermediate stage. If too may panels are
installed without strands to carry the load, the girder will be overstressed in hogging. If too many
strands are installed without panels to load the strands, the strand force will drop to a level where
proper anchorage cannot be assured. For Ting Kau the lower bound strand force was set at 20
kN by the cable supplier. Computer modeling the sequence of strand/panel installation for each
of the grids showed that an upper bound strand force could be established which would insure no
overstress. The upper and lower bound strand forces were then used to control strand/panel
7
erection. When the strand force approached the upper bound force, panel installation would stop
while strand installation continued until the strand force dropped. When the strand force began
approaching the lower bound force, strand installation would stop while panel installation
continued and the strand force increased. This method of control proved to be simple and
effective. The result was a stay installed to the correct force, insuring correct non-composite
bending moments in the girder.
After installation of the stays and panels, and casting of the infill joints, the stay was then
stressed to the final calculated length, not force. Final stressing to length has proved to be the
most effective method of achieving correct final geometry.
Demands placed on the cable installation crews were severe, particularly at the Main cantlever
which operated on a four day cycle and required some of the largest stays. In December of 1997,
approximately 1200 tonnes of cable stays (more than many complete bridges) were erected in a
single month.
Precast Panel Support

The precast panel support detail makes use of shims welded to the cross girder top flange over
the web. Four stiffened rebar loops in the corner of each panel rest on these shims to support the
precast at the correct level. This detail relies on close tolerances in as-built cross girder camber
to achieve acceptable deck geometry. The actual as-built camber of the cross girders was in
general less than theoretical and in some instances even negative. This resulted in high shim
stacks, stud extensions and extra rebar in the haunches created over the cross girders. Formwork
was now also required between the soffit of the precast and the top flange of the girder rather
than the simple sealing strips originally intended.
Temporary Wind Restraints

The stiffness and stability of a typical cable stayed bridge in its complete form is derived from
two elements - a heavy main span - typically longer than the side span - combined with anchor
stays connecting the tower top to a fixed anchor point in the side span. Neither of these two
elements are present during erection by balanced cantilever. Stability during erection is therefore
Fig. 9: Temporary wind restraints

8
often more critical than for the completed bridge. This was particularly important for Ting Kau
where the main cantilever is the longest balanced cantilever ever erected and the wind climate is
so severe.
Stability under typhoon was required for wind speeds up to 95 m/s. This could only be insured
through the use of a system of temporary cable restraints. Arrangements were proposed and
wind tunnel tested at the design stage. The final system (Fig. 9) has several significant aspects:
Prior to installation of the cable restraints, the deck was stabilized by fixed longitudinal
connections of the interior girders to the tower.
Once the cable restraints were installed, the longitudinal connections at the tower were
disconnected so as not to attract a disproportionate share of the restraining forces.
In order for the Main cantilever to progress beyond 170 m, the permanent longitudinal
stabilizing cables (LSCs) had to be installed to restrain the rocking response of the cantilever
(longitudinal tower bending). This in turn required that to resist the horizontal anchor forces
generated by the LSCs, the TY deck be temporarily connected longitudinally to the abutment
and the TK cantilever have its full cable restraint system in place. Note that the LSCs are
self anchoring with the deck in the final condition.
In order for the Main cantilever to progress beyond 170 m, the Main deck also had to be
connected to the Ting Kau deck by means of cross connecting cable ties to restrain the
weathervaining response of the cantilever (plan deck rotation). The Ting Kau deck also had
to be connected laterally to the anchor pier in order that the Ting Kau cantilever top could
provide restraint.
The typhoon season in

Hong Kong is between
June and November.
Peak gust wind speeds
by month as measured at
Waglan Island together
with the erected length
of the Main cantilever
are shown (Fig. 10).
The critical 170 m
length was not to be
reached until the end of
November. The peak
wind speeds for
December through
Fig. 10: Monthly wind speed versus erected Main cantilever length
March drop to as little as
one half of the peak typhoon wind speed. The temporary wind restraint system, which was
designed for full typhoon could therefore be reduced as follows:
The cross connecting cables were eliminated. This was a substantial saving. These cables
would have been installed slack at all times, to be tensioned only in a wind event. The cables
would have interfered with deck erection and would have required cumbersome moveable
temporary anchorages.
The LSCs needed only to be started at a cantilever length of 200 m and complete by a
cantilever length of 260 m. This meant that erection of the main cantilever need not stop
while the LSCs were being installed.
The temporary cable restraints were installed to force. Master strand tensions were calculated for
the installation stage and remaining strands were balanced to the master strand force. With the
exception of concrete placement, deck erection was permitted to proceed simultaneously with
temporary cable installation once the master strand was in place.
Closures
The Ting Kau and Tsing Yi side spans are permanently tied down with link plates and rocker
bearings connecting the girders to the TK anchor pier and TY abutment. The rocker bearings
were first installed and stressed to the anchor pier/abutment, the anchor grids were erected on
adjustable falsework over the bearings and the rocker bearings field welded to the girders.
Sidespan closure was made by jacking the anchor grid towards the cantilevered deck and
splicing. The Main/TY closure was made by releasing the post-tensioning bars tying the TY
deck to the abutment. The Main/TK closure was made by releasing the TK longitudinal restraint
cables in the side span.
Geometry and Stay Forces

The bridge geometry was checked closely for each stage of construction, as follows:
After erecting a new grid the plan alignment was checked. Adjustments were made before
the precast panels were erected and the alignment locked in. Alignment was controlled by
adjustment of temporary cross bracing installed in each grid during preassembly. Alignment
adjustments of up to 20 mm over a grid were common. Demand on the cross braces was the
greatest in the TK land span where the wide curved deck reached 60m in width and 6% in
crossfall. Coordinates of the tower anchorage and the deck anchorage for the new stay were
checked for use where the master stand force did not agree with the reference length.
After installation of stays and panels, the deck and tower geometry, and final stay installation
force were confirmed against theoretical before placement of infill concrete was permitted.
After placement of the infill concrete the global deck and tower geometry were confirmed
against theoretical.
Detailed global surveys of the deck profile were done before placement of the asphalt wear
surface. The asphalt was placed in two lifts a nominal 55 mm base course and a constant 30 mm
friction course. Using the deck survey, base course thicknesses were determined along the deck
to produce the best ride possible while staying within the +/- 15 mm thickness tolerance.
After the final deck closures a global survey of the bridge was carried out and stay forces
checked by monostrand liftoff. A set of final stay adjustments were calculated to tune the deck
geometry and stay forces. After closure 145 of the 428 deck stays were adjusted. Of these stays
129 were shortened by monostrand jacking and 16 were lengthened using a full size annular jack.
Stay adjustments took approximately two weeks to complete and were done independently of the
10
other finishing work on the bridge. Final adjusted deck geometry was typically within 50 mm of
theoretical.
Conclusions
The complete success of an innovative design such as the Ting Kau Cable stayed bridge must
come from careful consideration of both the final product and the construction process. In this
case the design innovations produced an efficient final design and also presented opportunities
and challenges to the construction team to optimize the construction process. While many of the
challenges were met, opportunities to fully realize the potential construction benefits of the
design were in some cases missed.
11
Seismic response of partially earth-anchored cable-stayed bridge

Toshiyuki SUGIYAMA
Associate Professor
Yamanashi University
Takeda, Kofu JAPAN
Toshiyuki Sugiyama, born

1954 received his Dr. Eng
degree from Univ. Tokyo
1984
Summary
The dynamic characteristics of partially earth-anchored cable-stayed bridge subjected to strong
earthquake motion are discussed based on the results of time-history response analysis. The
bridge type is three spans continuous girder type with multiple cables and its main span length is
1000 meters. Finite Element Method is applied to the dynamic analysis. Hyogo-ken Nanbu
Earthquake (Kobe Earthquake) record including both horizontal and vertical components is
adopted as input earthquake motion. Only the direction of motion parallel to the bridge axis is
considered. Soil-structure interaction and phase-lag of input earthquake motion that arises among
two piers and two anchored points of side-span cables are neglected here.
It has been revealed that the maximum vertical displacement at the center of main span of
partially earth-anchored cable-stayed bridge is enough smaller than the deformation limit
although the maximum deformation of partially earth-anchored cable-stayed type is larger than
that of self-anchored one. The results also show that the stress resultants of partially earthanchored cable-stayed bridge are considerably smaller than those of self-anchored one. These
results indicate that no problem may occur from seismic viewpoint in case of the application of
partially earth-anchored cable-stayed bridge to long span bridge with main span length of about
1000 meters. And it is also cleared that the consideration of only horizontal earthquake motion is
sufficient in case of the execution of dynamic response analysis of cable-stayed bridge with
1000m mainspan length.
1. Introduction
When cable-stayed bridge is applied to long span bridge with about 1000m center span length,
large axial force that acts on the main girder probably becomes a serious problem as it may cause
the buckling of main girder. To reduce this axial force, partially earth-anchored cable-stayed
bridge has been proposed by Gimsing[1]. And a few studies have been carried out to discuss the
static characteristics of this type of bridge by Kaneko et al. [2]. However, the seismic
characteristics of partially earth-anchored cable-stayed bridges have not been estimated in detail.
The purpose of this study is to discuss the dynamic response of partially earth-anchored cablestayed bridge subjected to strong earthquake motion based on the results of time-history response
analysis. Partially earth-anchored cable-stayed bridge with 1000-meter main span length is
selected as the subject of study and its seismic characteristics are compared with those of selfanchored one. The bridge type is three spans continuous girder type with multiple cables. Finite
Element Method and Newmark's method are applied to the dynamic analysis. Hyogo-ken
Nanbu Earthquake (Kobe Earthquake) record including both horizontal and vertical components
is adopted as input earthquake motion.
2. Analytical model
Fig. 1 shows the partially earth-anchored cable-stayed bridge with main span length of 1000m.
The upper six cables in each side span are anchored to the earth. Self-anchored cable-stayed
bridge with 1000m main span length is illustrated in Fig. 2. The cross sectional area and second
moment of area of each structural element are shown in Table 1. These two types of cable-stayed
bridges are selected as the subject of investigation here and their dimension and cross sectional
properties are the same as the bridges taken into account in reference [2]. Especially, partially
earth-anchored type as shown in Fig. 1 was one of the bridges that were
Fig. 1 Partially Earth-anchored Cable-stayed Bridge
Fig. 2 Self-anchored Cable-stayed Bridge
Main grinder
Cable
upper part
Pylon
lower part
Cross Sectional Area

0.8626 ~ 0.9996 m2
0.0097 ~ 0.0141 m2
1.1760 m2
Second Moment of Area

1.1166 ~ 1.3226 m4
1.4700 m2
12.411 m4
9.9810 m4
Table 1 Cross Sectional Properties of Cable-stayed Bridge

proposed for Tatara Bridge planning by Dr. Kaneko's bridge design group, although this type was
not adopted in practice.
The following two supporting conditions are considered in this study;
1) one end of main girder is fixed by hinge bearing and other points are supported by movable
hinge bearings;
Fig. 3 Hyogo-ken Nanbu Earthquake Record

2) one pylon and main girder are rigidly connected.
As any significant difference between self-anchored type and partially earth-anchored one has
not been recognized regarding to the dynamic response in the transverse direction of bridge axis
[2], only the direction of motion parallel to the bridge axis is considered here.
Hyogo-ken Nanbu earthquake record is adopted as input earthquake motion. Both horizontal and
vertical earthquake motions are taken into account. Figs. 3(a) and (b) show the horizontal and
vertical component records of Hyogo-ken Nanbu earthquake, respectively. Fig. 4(a) and (b) are
the power spectrum density diagrams corresponding to Fig. 3(a) and (b), respectively.
Soil-structure interaction is neglected. And phase-lag of input earthquake motion that arises
among two piers and two anchored points of side-span cables is also neglected here.
In the analysis based on Finite Element Method, main girder and pylon are assumed to be one
beam and one column whose cross sectional properties are shown in Table 1.
Fig. 4 Power Spectra of Hyogo-ken Nanbu Earthquake Records

Each cable is assumed to be a member that resists only tensile force. The length of each finite
element is 10[m] according to the result of investigation regarding to the convergence of lower
order eigenvalues. The initial values of stress resultants of main girder, pylons and cables are
obtained from the static analysis in which only dead load acts on each type of cable-stayed
bridge.
In order to estimate the effect of vertical earthquake motion on the dynamic response, the
following two cases are taken into account in time history analysis; that is, a) dynamic response
under only horizontal earthquake motion (designated CASE-1 hereafter) and b) dynamic
response under both horizontal and vertical earthquake motions (designated CASE-2).
3. Results of dynamic response analysis

3.1 Vertical Displacement of Main Girder
Fig. 5 shows the maximum vertical displacement of main girder along bridge axis under Hyogoken Nanbu earthquake motion. In this figure, left-hand part shows the vertical displacement in
case of supporting condition 1), and the right-hand part presents the one in case of supporting
condition 2). And in Fig. 5, each line corresponds to the following cases;
solid line : partially earth-anchored type in CASE-2
broken line : partially earth-anchored type in CASE-1
dotted line : self-anchored type in CASE-2
one dot chain line : self-anchored type in CASE-1
(These four kinds of lines in Figs. 6 9 correspond to the cases described above, too.)
From Fig. 5, it can be recognized that maximum vertical displacement at the center of centerspan in case of supporting condition 2) is smaller than those in case of supporting condition 1).
This fact means that the supporting condition 2) is more preferable than supporting condition 1)
from the point of view of smaller deformation. Accordingly, only the results in case of
supporting condition 2) are discussed hereafter.
From Fig. 5, the following facts are also found;
1) the maximum vertical displacement of partially earth-anchored type is about twice as large as
that of self-anchored one in case that only the horizontal earthquake motion is input;
2) in case that both horizontal and vertical motions are considered, any significant difference
between self-anchored type and partially earth-anchored one is not recognized.
3) the maximum displacement of main girder of partially earth-anchored type in CASE- 2 is
approximately the same as that in CASE-1, although the former is a bit larger than the latter;
4) the maximum vertical displacement at the center of center-span of partially earth- anchored
type is enough smaller than the deformation limit although the value of 3[m] for deformation
limit in Japanese specification is not expressed in Fig. 5.
Fig.5 Maximum Vertical Displacement of Main Girder

3.2 Horizontal Displacement of Pylon
Fig. 6 presents the maximum horizontal displacement of pylon along its height. From this, it can
be recognized that the horizontal displacement of pylon of partially earth-anchored type becomes
larger than that of self-anchored one. However, it is also known that the maximum displacement
at the position where the lowest cable of main span is anchored to the pylon takes approximately
the same value.
Fig. 6 Horizontal Displacement of Pylon

3.3 Axial Force of Main Girder
The maximum values of axial force that acts on the main girder are shown in Fig. 7. From this, it
can be recognized that the maximum axial force in case of partially earth-anchored type is at
most about 50% of the one in case of self-anchored type. It is also found that any significant
difference between CASE-1 and CASE-2 is not recognized. This means that the consideration of
only horizontal earthquake motion is enough when the dynamic response analysis of cablestayed bridge with 1000m main span length is executed.
Fig. 7 Axial Force of Main Girder

3.4 Bending Moment of Main Girder
Fig. 8 illustrates the maximum values of bending moment of main girder. Regarding to the
bending moment, the values in CASE-2 are considerably smaller than those in CASE-1. Its
reason may be as follows; in case that the vertical and horizontal earthquake components are
input simultaneously, the dynamic behavior of main girder becomes complex and the effects of
both horizontal and vertical motions are offset.
Fig. 8 Bending Moment of Main Girder

3.5 Stress Resultants of Pylon
Fig. 9 shows the axial force, shear force and bending moment which act on the pylon. Except the
shear force, the tendency is the same as the bending moment of main girder. In case of shear
force, the differences among four cases are very small.
Fig. 9 Stress Resultants of Pylon
4. Concluding remarks
The dynamic response of partially earth-anchored cable-stayed bridge subjected to strong
earthquake motion is investigated based on the results of time-history response analysis. It has
been revealed that the maximum vertical displacement at the center of center span of partially
earth-anchored type is enough smaller than the deformation limit although the maximum
deformation of partially earth-anchored cable-stayed type is larger than that of self anchored one.
The results also show that the stress resultants, i.e., axial force of main girder and axial force and
bending moment of pylon of partially earth-anchored cable-stayed bridge are considerably
smaller than those of self-anchored one. These results indicate that no problem may occur from
earthquake resistant viewpoint in case of the application of partially earth-anchored cable-stayed
bridge to long span bridge with center span length of about 1000 meters. And it is also cleared
that the consideration of only horizontal earthquake motion is sufficient in case of the execution
of dynamic response analysis of cable-stayed bridge with 1000m main span length.
References
[1] Gimsing, N.J.: Cable Supported Bridge -Concept and Design-, John Wiley & Sons, 1983.
[2] Kaneko, S., Nakayama, T., Mukoyama, T., Iwaki, T. and Takekawa, S.: Applicability of
Partially Earth-anchored Cable-stayed Bridge to Long-span Bridge, Journal of Construction
Management and Engineering, No. 510/VI-26, pp.113-124, March 1995 (in Japanese).
Charles River Crossing:A Gateway To Boston

Vijay CHANDRA
Chief Bridge Engineer
Bechtel/Parsons Brinckerhoff
Boston, MA, USA
Anthony RICCI
Manager of Structures
Massachusetts Highway Department
Boston, MA, USA
Christian MENN
Consultant to the Central Artery/Tunnel Project
Zurich, Switzerland
Raymond McCABE
Senior Vice President
HNTB, Inc
Fairfield, NJ, USA
1.
Introduction
Bostons Central Artery/Tunnel (CA/T) project involves many innovative underground

structures. However, the projects keystone is the Charles River Crossing, featuring a hybrid
cable-stayed bridge that will be the first of its type in the US. The river crossing will provide
virtually the only access to Boston from the north, straddling the Charles River in an historic area
where Paul Revere began his famous ride and the Battle of Bunker Hill took place. Accordingly,
there is great interest in ensuring that the bridge features a distinctive design while also providing
a dramatic gateway to the citys downtown.
The immense CA/T project consists of many kilometers of tunnels, six major interchanges, a
long-span parallel crossing of the Charles River connecting Storrow Drive to the north and the
cable-stayed bridge. Major project funding is being provided by the Federal Highway
Administration (FHWA). Parsons Brinckerhoff Inc., in joint venture with Bechtel (B/PB), is
serving as project management consultant for the Massachusetts Turnpike Authority (MTA), the
owner.
B/PB prepared the bridge type studies, performed preliminary designs of selected structure types,
managed and oversaw preparation of the final design by HNTB, Inc./Figg Engineers, and is
managing construction. MTA is providing oversight of the projects design and construction
phases.
2.
Design Constraints
Major physical constraints for the bridge in the Charles River crossing area include:
Orange Line subway ventilation building adjacent to the south main pier
Orange Line tunnel alignment traversing the bridge alignment at an angle, requiring the north
main pier foundation to straddle it
A 0.92-meter-diameter waterline located below the south main pier
Steep 5% grade entering a tunnel at the south end and tying into a three-level interchange at
the north end
Potential interference with a CA/T highway tunnel exit ramp to be constructed at the south
end of bridge
Existing double-deck bridge and its ramps, which must remain in service during construction
of the new bridge
3.
Existing Charles River lock and dam system abutting the bridge on the east side
Bridge Type Selection
The Charles River Crossing must meet the objectives of numerous state and federal regulatory
agencies, including the FHWA. The bridge design must present sound engineering solutions to
numerous site constraints while also meeting community expectations that the structure create a
distinctive signature on Bostons skyline. To fulfill these goals, the project team conducted a
bridge type study and assembled a multidisciplinary team of experts in structural engineering,
highway design and engineering, urban planning, construction, environmental engineering,
architecture, and cost and scheduling control.
Initially, the project team identified 16 bridge types, each with a main span measuring 227
meters. The sixteen designs included three arch bridges, four truss bridges, a segmental box
girder bridge, a fin-back concrete bridge, a suspension bridge and six cable-stayed bridges. The
engineering team then analyzed each option for a wide variety of impacts, including alignment,
design, environmental, urban design, constructibility, ease of inspection, maintenance and cost,
before narrowing the field to seven bridge types for further study. They included a two-hinged
arch, a simple span truss and five cable-stayed bridges. The cable-stayed bridges, both
symmetrical and asymmetrical designs, varied between 10 and 12 traffic lanes and featured
single- and double-tower configurations. It was also decided that the cable-stayed bridges could
be built using steel, concrete or a hybrid of both, and that tower pier shape was flexible. An
evaluation matrix was prepared based on priority factors as well as a quality rating for the
impacts previously mentioned.
From the field of seven candidate
bridges, two cable-stayed designs
emerged that best met the
demanding site conditions: an
asymmetrical 10-lane bridge with
twin, inverted Y-shaped towers
and two cantilevered lanes, and a
symmetrical 10-lane bridge with
a single A-shaped tower. From
these two finalists, the
Massachusetts Highway
Department selected the twin
tower asymmetrical design, citing
its smaller inverted Y towers as a
Figure 1. The First Hybrid Cable-Stayed Bridge in the US
benefit in reducing the visual
scale of the bridge to be more appropriate with its surroundings (See Figure 1). The initial
concept for this asymmetrical bridge was proposed by the renowned Swiss engineer, Dr.
Christian Menn, who was retained as a consultant during preliminary design to help ensure that
the bridge met community expectations.
Preliminary designs were then prepared for steel, concrete and hybrid alternatives for the
asymmetrical cable-stayed structure. Although each alternative posed unique challenges, a
committee composed of international experts convinced the project team that only the hybrid
design should be continued due to its relatively short back spans compared with the main span.
The back spans will feature heavy cast-in-place post-tensioned concrete construction to
counterbalance the light main span constructed of steel floor beams and edge girders with a
precast concrete composite deck. By using this configuration, the bridge will become the first
hybrid cable-stayed structure built in the US.
4.
Site-Specific Seismic Study
The Charles River Crossing will be a lifeline to Boston. Like all major interstate highway
structures, it is considered important for seismic considerations. This means that the bridge
should be serviceable after the design earthquake, sustaining only minor damage. Therefore, due
to the structures critical role and its unusual design features, a site-specific seismic study was
undertaken.
Evaluating seismic sources in the New England area was based on the latest developments in
seismic source zone characterization and attenuation relationships for the eastern US. Both 2%
and 5% damping for the 500-year operating design earthquake and 2,000-year maximum design
earthquake return periods were developed for the Charles River Crossing.
5.
Aesthetics
The shape of the tower piers and the cable arrangement (main span cables splayed out from the
tower with back span cables centered in the median) evolved from technical requirements.
For example, the proximity of the Orange Line vent building to the south tower and the need for
the foundation of the north tower to straddle the subway alignment below, necessitated truncating
the columns below the roadway. The tower shape was itself dictated by the torsional rigidity
required in the transverse direction due to the cantilevered ramp on the east side. In addition, the
cable arrangement resulted from the proximity of Leverett Circle ramps (approximately 0.61
meters) on the south side, which must remain in service while the new structure is built.
Splaying the cables in the back span was further impacted by existing Leverett Circle ramp
supports.
Technical issues aside, aesthetics played an important role in shaping the structure, making it a
true gateway to Boston. This concern even extended beneath the structure, where openings were
provided in the wide main span deck to prevent a permanent shadow from being cast on the
water.
6.
Bridge Configuration
The Charles River cable-stayed bridge features a 56.4-meter-wide, five-span hybrid

superstructure with a main span of 227 meters; two south back spans of 34.2 and 39.6 meters;
and two north back spans of 51.8 and 76.2 meters (See Figure 2). The tower piers are an inverted
Y shape (See Figure 3) and the back spans consist of multi-cell concrete box girders, 3 meters
deep and 38.4 meters wide. Main structural elements include a 3-meter-wide central spline beam
and four secondary webs with internal diaphragms spaced at 4.6-meter intervals (See Figure 4).
The spline beam, in turn, is supported by a single plane of cables spaced at 4.6 meters.
Figure 2. Bridge Elevation
Figure 3. Tower Elevation
The main span consists of precast concrete

deck panels acting compositely with
longitudinal steel box edge girders and
transverse steel floor beams (spaced at 6.1
meters on center) by means of cast-in-place
closure strips (See Figure 5). The box edge
girders are supported by cables anchored on
the outside at 6.1-meter intervals. On the
main span side, the two-lane ramp (SA-CN)
is carried on floor beam extensions
cantilevered to one side of the main line
deck. Lightweight precast concrete deck
panels are used for the ramp to minimize
eccentric dead loads. On the back spans,
the ramp is a single-cell concrete box
girder, independent from the cable-stayed
structure, with roadway joints at the tower
interfaces. Open-grid fiberglass closure
panels partially cover the underside of the
main span superstructure to create a more
aesthetic underbelly.
The many design challenges of the Charles River cable-stayed bridge are discussed in the
following pages.
Figure 4. Concrete Back Span Typical Section
Figure 5. Main Span Typical Section

6.1
Foundation and Drilled Shaft Design
The tower foundations consist of footings on 2.44-meter drilled shafts designed to carry a
working load of 2,270 tonnes each. At the north tower, the footing and the supporting drilled
shafts straddle the Orange Line tunnel. Measures were taken to ensure that the tunnel is not
adversely impacted by foundation construction activities. The drilled shafts closest to the Orange
Line tunnel are placed outside a 1.52-meter buffer zone and the project team investigated the
effect of lateral forces transmitted from the drilled shafts through the surrounding soil to the
Orange Line tunnel. As a result, the two shafts closest to the tunnel are to be installed within
2.74-meter-diameter isolation casings. Meanwhile, the south tower foundation is very close to
the existing subway ventilation building on the west side, while also bridging a 0.9-meterdiameter waterline. As a result, the three drilled shafts closest to the ventilation building will
also be isolated within a 2.74-meter-diameter isolation casing.
6.2
Strut at Tower Piers
The change in direction of the tower legs at the deck level produces large tension forces in the
tower strut, which also serves as the transition from the main span composite steel superstructure
to the post-tensioned concrete box girder back spans. These forces, which are carried in the
spline beam of the back spans, need to be transferred to the edge box girders of the main span or
vice versa. The edge girders, in turn, are then attached to the main-span cables. As a result,
special attention was focused on strut design, while shear lag effects and tension stresses in the
concrete were also carefully evaluated.
In addition, the imbalance of bending, shear and axial load forces in the main span and back
spans under different loading combinations produce torsion, bi-axial bending and bi-axial shear
stresses in the tower strut. Due to the critical structural nature of the strut, which will be posttensioned in stages to a total initial jacking force of 25,000 tonnes, limiting principal tensile
stresses to pre-determined values under the working loads was an important design
consideration.
6.3
Concrete Back Spans
Two major challenges in the

design and detailing of the
south back span were due to
the physical overlap of the
plan area of the proposed
bridge, a tunnel ramp at the
south interface, and the
presence of numerous
Figure 6. Spline Beam
existing, temporary and
future ramps under the north back span. At the south interface, the design solution features an
early termination of the main line bridge deck (by approximately 15.24 meters), while extending
the central spline beam the full length as a cantilever to receive and anchor the first three cables
(See Figure 6). Heavyweight ballast concrete with a density of 4,000 kg/m3 will be placed in the
box girder cells within the last three floor beam bays to counteract the local reduction in
superstructure weight due to the early termination of the roadway. The cantilever extension of
the spline beam is housed in a vault built into the slab on grade roadway of the adjoining project
contract. To reduce the impact of north back span construction to ramp traffic by limiting ramp
closures and detours, final design and detailing of the north back span was conducted based on
the incremental launching construction method. However, after a detailed evaluation, the
contractor proposed and the project team accepted a value engineering proposal to cast-in-place
the north back span using falsework.
6.4
Steel Edge Girders and Floor Beams
The edge girders are asymmetrical steel box sections with an inclined bottom flange and an
inclined fascia web. Typical edge girder field sections are 18.3 meters long, supported by three
stay cables. To achieve a full moment connection between the tower and the edge girders, a base
plate connection with 35-millimeter-diameter, 1,034 Mpa post-tensioning bars is used. Floor
beams, spaced at 6.1 meters longitudinally, will span 42.7 meters between box edge girders and
cantilever approximately 13.7 meters to carry two lanes of traffic outside the east plane of the
stay cables. Additionally, an edge beam provided at the fascia of the cantilevered section will
distribute truck loads to multiple floor beams.
6.5
Stay Cables
Project design documents required stay cables strands to be either greased and sheathed or
completely filled with epoxy. Also, flexibility was provided concerning the type of anchorage,
wedge or socket or wedge/socket to be utilized. After the project was bid, Kiewit/Atkinson, the
successful bidder, proposed using ungrouted stay cables employing the Freyssinet Iso-tension
stressing method. Implementation of this proposal means that this will be the first US project to
use ungrouted stays and the Iso-tension method of stressing. The contractor also successfully
proposed using coextruded PE pipe with a spiral bead to reduce stay cable vibration.
The CA/T project has opened many doors and advanced the state-of-the-art of stay cable
technology. Presently, two of the three stay cable fatigue/tension tests have been performed
successfully with the last one underway.
6.6
Cable Anchorage at Towers
The vertical leg at the

tower top varies from 3.2
meters square to
approximately 4.9 meters
square at its base. Because
of the limited room to
anchor cables, a
prefabricated steel anchor
box will be built into the
tower, acting compositely
with the concrete section
by means of shear
connectors. The cables
Figure 7. Tower Head Cable Anchorage
will be anchored by
bearings at the inner end of
structural tube sections built into an anchorage girder (See Figure 7). This detailing offers the
following advantages:
Reduced torsional moment due to closer transverse spacing of the cables
Improved geometry control of the cable anchorages
Elimination of complicated forming of the inside walls
Elimination of post-tensioning in the tower cross section
However, torsion in the tower leg due to the cantilevered ramp on one side posed a challenge.
The east side cables of the main span carry a 30% greater load than the west side cables. This is
overcome by using a lightweight concrete deck slab in the cantilever and offsetting the geometric
centerline of the back span and main span cables by 76 millimeters.
To avoid external cable anchorages and related maintenance issues in the inclined legs of the
tower, a non-uniform cable spacing scheme is implemented. This entails gradually increasing
vertical spacing from the standard 1.68 meters for the uppermost ten cable pairs up to 2.9 meters
for the lowermost cable pair. Additional minor geometric adjustments were also implemented.
6.7
Girder-to-Cable Anchorages
The cable anchorages on the main span
box edge girders are mounted on the
outside and are detailed as a pipe
assembly bolted to the side of the
girder (See Figure 8). The cables are
then passed through the anchor pipe,
with the cable anchor bearing against
the lower end of the pipe (which also
forms a part of the structural system
resisting local forces due to the cable
anchorage). The pipe will be
connected to the base plate with a
single web plate.
This detail was selected due to its

visual appeal over typical box-type
cable anchorages; fabrication and erection considerations; and for allowing easy access for
inspecting all critical welds and bolts.
Figure 8. Cable-to-Edge Girder Anchorage
7.
Aerodynamic Evaluation
Wind tunnel tests of both the sectional and aeroelastic models were performed for the final
structure as well as for intermediate construction stages. Vortex excitation occurred at about 128
kph, within criteria, while flutter speed was measured at 715 kph, well above the requirement of
210 kph. Smoke flow visualization tests also indicated that wind flows were not significantly
altered by changes to the deck section, such as deck openings and porous closure panels on the
underside.
The assessment of the potential for cable vibration, considering the use of coextruded PE pipe
with a spiral bead to reduce rain/wind vibration and ungrouted stay cables, resulted in cross-tie
requirements to offset galloping. After a study of the Freyssinet viscoelastic dampers was
conducted by RWDI and Construction Technology Laboratories, it was concluded that providing
dampers at all lower anchorages, coupled with some cross-tie arrangements, would best meet
project needs.
8.
Erection Scheme
The cast-in-place back spans will be constructed on falsework concurrent with tower
construction. The tension strut at the tower piers will be post-tensioned in stages. Afterwards,
the superstructure of the main span will be erected in a cantilever fashion. As sections of the
main span are erected, stay cables will be installed and tensioned.
The superstructure of the main span will be erected in stages. Initially, the edge box girders will
be cantilevered out, followed by precast slab placement and, finally, cast-in-place closure strip
construction. After the superstructure is erected in the main span, barriers and latex-modified
concrete overlay will be constructed.
9.
Construction
Construction of the Charles River cable-stayed bridge was started in September 1997 by
Kiewit/Atkinson, a joint venture. The low bid price for the bridge was $86.7 million. The
drilled shaft foundations for the main piers, back span piers, and walls are nearly complete. The
foundation slab of the south tower is complete and work has started on the north tower. The
south tower lower inclined legs have also been constructed and work on the tension strut is
underway. Bridge completion is scheduled for late 2001.
10.
Conclusion
Boston, in the forefront of the American Revolution over two centuries ago, is now in the
forefront of another revolutionin the field of cable-stayed bridge technology. A highly
complicated structure, unique in the world, has been successfully designed and is under
construction. New technologies and innovations have become hallmarks of the Charles River
Crossing.
The Design and Construction of Lockmeadow Footbridge, Maidstone
Ian P.T. FIRTH

Partner,
Flint & Neill Partnership
London,
England.
Ian Firth graduated from the

University of Bristol in 1979 and
obtained a Masters degree in
Structural Steel Design at Imperial
College in 1982. He has been
responsible for many bridge
projects with Flint & Neill
Partnership, including the Poole
Harbour Bridge in England. He is
also responsible for the design of
two other footbridges soon to be
constructed in Maidstone.
Summary
This competition winning design for a cable stayed aluminium footbridge uses aluminium
extrusions in an innovative and effective way to reduce both initial construction costs and future
maintenance requirements.
Introduction
The bridge crosses a bend of the River Medway in the centre of Maidstone in Kent, adjacent to
the Grade 1 listed Archbishops Palace at a location where historical and archeaological issues
predominate. The site has been described as one of the most sensitive sites for a bridge
anywhere in England, and after careful analysis the governing visual design criteria of
slenderness and lightness evolved. (Figure 1) A mediaeval lock used to occupy the river at this
point (hence the name Lockmeadow) and more recently a pedestrian ferry boat service
operated here, and both of these also had an influence on the design.
Figure 1: Architects model image of the finished bridge with the Archbishops Palace behind
The design was developed in conjunction with Chris Wilkinson Architects and was the winning
entry in an invited design competition held by Maidstone Borough Council in early 1997.
Inevitably, any design competition scheme attracts a lot of critical interest from among the
architectural and engineering professions, as well as from a wider audience, and the visual
quality as well as the elegance of the engineering solution have received and will continue to
receive much scrutiny.
The contract was let by Maidstone Borough Council to Christiani and Nielsen in July 1998 for a
construction cost of 630,000. Their principal sub-contractors are D&B Darke for the fabrication
and deck erection, and Nedal Aluminium BV for the aluminium supply.
The bridge provides a pedestrian crossing of the river from the town centre to a new leisure
development under construction on the west bank adjacent to the Lockmeadow market site.
The Conceptual Design
The brief called for a navigation clearance of 4m without intermediate supports in the river, and
disabled access requirements dictated a maximum gradient for the approach ramps of 1:20. Due
to local constraints at the east bank, where the abutment is set among trees behind an ancient
stone river wall, it was however agreed that short lengths of steeper ramp could be tolerated with
suitable landings in between.
The land behind the river wall on the east bank is a Scheduled Ancient Monument with very
difficult access conditions, and the stability of the old stone retaining wall could not be easily
determined. The bridge thus needed a very light touch on this bank, permitting the use of a
small abutment with a minimum of excavation and vertical mini piles in sleeves to prevent any
lateral loads on the wall.
The river is known to flood regularly, and the west bank is part of a flood plain which was to be
kept as clear of obstruction as possible. It thus became necessary for the bridge to span not only
the river itself, which at this point is about 40m wide, but also part of the flood plain as well.
Access to the west bank was not such a problem, and permitted the construction on the river
bank of a substantial pier to compensate for the need for a light touch on the other side. Thus the
solution of a twin span cable stayed bridge with one span over the river and the other over the
western flood plain began to emerge, with the masts partly hidden in the trees. This also suited
the visual and historical references which demanded a strong focus for the bridge on the west
bank of the river. This west bank pier is shaped like the bow of a boat to face upstream and act
as a cutwater under flood conditions, and on its back it carries the access stairs to the bridge
from the towpath. (Figure 2)
A cable stayed solution enabled the deck depth to be kept to a minimum, owing to the
intermediate stay supports, and this not only suited the desired visual lightness and transparency,
but also reduced the lengths of the approach ramps. The supporting mast on the west bank was
split into two, one each side of the cutwater stairs, and these were inclined forwards and
outwards to provide the most effective positions for suspending the deck.
The budget set by the Client at the competition stage was extremely tight, and this was partly
responsible for the decision to adopt a deck system involving the assembly of simple repeatable
components with no added finishes. A lightweight solution was also necessary to minimise
foundation costs. The final bridge solution adopts a unique and very shallow extruded
aluminium deck system as a result of these factors.
Figure 2: Plan and elevation

Crossing a footbridge such as this is not just a means of getting from one side to the other but
also an experience in itself. Indeed part of the interest in a footbridge is the texture, colour and
feel of the handrail and ballustrade, as well as the visual appearance of the bridge. In this case,
the ballustrade also needed to be relatively transparent and permeable to the passage of the flood
water. The chosen solution is a stainless steel handrail with black carbon fibre posts and
stainless steel wedge wire infill panels. These seemingly expensive but attractive materials
turned out to be cost effective partly because of their low maintenance costs and partly because
their specialist manufacturers were keen to supply components for this high profile project.
The bridge has a slight plan curvature which was added at a late stage to tie in with the landscape
scheme, to add extra interest, and also to direct the eye away from the rather ugly adjacent leisure
building on the west side.
The Aluminium Deck
The extruded aluminium sections are placed side by side longitudinally and stressed together
transversely with stainless steel bars. The parapet posts are fixed via couplings to the ends of the
bars and simultaneously retain the outer edge extrusions. The top flange of the extrusion is
ribbed and cross-cut transversely to provide a slip resistant surface, and there is no need for any
secondary structure or for any added finishes. This is one of the beauties of the system;- the
aluminium extrusion provides the primary structure, the secondary structure and the finish all in
one section, and this provides an elegant and economic solution.
This system leads to the formation of a cellular deck cross-section. Early discussions with
manufacturers showed that closed section extrusions would be at least twice as expensive as
open sections, so a series of back-to-back open channel sections was adopted. We did not wish
the soffit to have the same ribbed surface as the top, so a doubly symmetric shape which would
have enabled the use of tongue and groove type interlocking sections was not possible. Instead
we provided small continuous grooves in the section and introduced shear keys to improve
transverse rigidity. It was also necessary to introduce inclined webs and a central blocking cell
so that when assembled they together formed a pattern of X-bracing. (Figure 3)
Figure 3: Typical Section

The maximum cross-section size of extrusion which can be formed is limited by the size of the
die through which the aluminium is pushed. In the UK there are no large diameter presses
operating, but in other parts of Europe there are several. The limiting size is defined by the
circumscribing circle, and it became clear from early discussions with manufacturers that an
economic solution would need to fit within a 350-400mm circle. Thus the final section is
300mm deep and 105mm wide.
The maximum length of extrusion is governed by the amount of aluminium in the cross-section.
An ingot of aluminium pushed through a die will produce a longer length if there is less material
in the section. There is also a certain amount of wastage at each end of the section which is cut
off after extrusion. In our case, the maximum finished length of extrusion is about 7m, and most
finished lengths are approximately 6.4m. The section weighs 15kg/m, so each typical length
weighs about 96kg which can easily be handled by two men - another beauty of the system which
considerably facilitates deck assembly.
The aluminium alloy selected is 6082 T6 to BS 1474 which gives good material strength
properties, although with a slightly lower quality of finish than some of the other architectural
grades. The outer edge sections are anodised to give a good uniform finish, but it was decided to
leave the remainder as mill finished to reduce costs. The 6082 material contains copper in the
alloy, and tends to take on a slightly greenish colour after anodising.
The transverse pre-stressing bars are 24mm diameter S316 martensitic stainless steel with a yield
strength of 800 N/mm2. The guidance given in the design standard BS8118 regarding isolation
of dissimilar metals was followed. This recommends that the stainless steel and aluminium need
only be separated when fully immersed in water or where high levels of atmospheric pollution
exist. In view of the fact that the bridge would occasionally be immersed in flood water at its
lower end, it was decided to place the bars within sleeves and apply a layer of bituminous
material under the contact surface at the ends of the bars as an extra safeguard.
Friction tests on samples of the extrusions showed a lower than expected coefficient of friction of
only about 0.2. This led to the introduction of additional shear keys to assist in transferring
transverse and longitudinal shears between the extrusions at points of high shear such as at stay
anchor positions and at bearings.
Figure 4: Typical Modal Response Analysis Plot

Our analysis of the dynamic behaviour and time history response calculations indicated that the
design just failed to meet the criteria contained in the Highways Agency standard BD49/93 for
excitation by pedestrians. This analysis integrated the response caused by loads travelling over
the bridge in phase with the natural frequencies of the bridge to determine the peak accelerations.
(Figure 4) Several options were considered for alleviating this issue, but we found that the
simplest was to add mass over the midspan section which improved the response so that peak
accelerations were brought back within the specified limits. This extra mass is added in the form
of wrapped steel reinforcing bars inserted inside the voids in the deck section over the central
17m at midspan. The calculated fundamental frequency of the finished bridge is 0.98 Hz.
The Cable Stay System
The cable stays are 45mm locked coil ropes with cast steel sockets all supplied by Bridon
International. Length adjustments are made using externally threaded cylindrical sockets at the
deck anchors which are fabricated steel sections fixed to the outside of the deck.
The back stays which hold the masts in position are in pairs for added redundancy in the case of
damage or wilful vandalism and also so as to facilitate future replacement. These are adjusted by
threaded bars at their anchorage beside the cutwater stairs. The top horizontal stays between the
mast heads are also paired to permit future replacement one at a time, and the bridge is designed
for accidental or planned removal of any one stay.
The Steel Masts
Continuing the transparency theme, the 15m long steel masts were designed as skeletal members
using 70mm diameter solid legs held apart by special cam-shaped spacers. The masts are cigar
shaped for efficiency as a pin ended strut, and this further enhances the appearance of
slenderness. The spacers are double 10mm plates in pairs with 6mm webs between to provide
sufficient vierendeel stiffness to reduce the buckling length of the individual legs to 1055mm. A
solid section at the base discourages vandals from attempting to climb the masts, and a light
fitting is contained within it to illuminate the mast head through holes in the cam-shaped spacers.
The masts are pinned at their base with a simple spherical bearing surface machined onto a
125mm diameter pin which bears on a slab cast into ears on the side of the concrete cutwater,
and the stays are attached to flat plates welded to a solid section at the masthead. Each mast
weighs about 3.5 tonnes, and was delivered to site in one piece.
Aluminium Fabrication
As already mentioned, the aluminium extrusions were machined on their top surface to introduce
added roughness to achieve a non-slip surface. The contract required careful control of the
finished cross cut pattern, particularly in respect of the regularity of spacing and depth of the
cuts, and in the event this presented some difficulties. The equipment available to the specialist
sub-contractor and the tolerances in the extrusions meant that they had to be machined
individually rather than in pairs as originally planned. However, after the initial difficulties, the
necessary quality was achieved with acceptably good results.
A trial assembly of a section of the deck was specified under the contract and was carried out at
the fabrication works. This enabled the development of a suitable jig for the assembly of the
aluminium sections, and allowed the fit up of the aluminium extrusions and the associated
processes to be checked. (Figure 5)
Particular care had to be taken in the handling of the aluminium sections to avoid damage.
Scratches and dents could occur relatively easily with careless handling against unprotected steel
rollers or equipment during drilling or machining, and this was unacceptable because of the
desire to achieve a high quality of finish throughout. It was therefore necessary to ensure that
extra handling protection was added to prevent such damage, and that all operatives were
properly informed about the importance of this factor.
Figure 5: Deck trial assembly
Construction
It was originally envisaged that the west bank section would be assembled in situ on temporary
trestle supports, and that the river section would be assembled in one 30m long piece on the river
bank and then floated out and lifted into place in one operation. However this method required
the use of barges which would cause a temporary restriction of the river flow, and this was not
permitted by the Environment Agency because of the relatively high risk of flooding. It was
therefore decided to erect the bridge by incremental launching from the west bank instead since
this only required the installation of a single temporary steel bent at mid-river causing negligible
blockage to the flow.
The aluminium sections were assembled by hand in a special jig mounted on the west abutment
where the working environment and processes could be carefully controlled. The assembled
deck was then pushed out along a temporary trestle support and the next section assembled
behind it. Thus the deck was in effect doubly extruded because this launching process on site
reflected the extrusion process in the aluminium manufacture.
A king post arrangement was used to enable the deck section to span the distance from the
cutwater to the mid-river support and on to the east abutment.
Once the deck was in place, the masts were erected and the stays attached to the deck in a single
operation. The deck was then lifted off the temporary supports by stressing the stays, adjusting
their lengths to achieve the desired profile.
The original design required the aluminium extrusions to be pre-cambered vertically so as to
achieve a straight deck in the finished bridge. However, this turned out to be too costly and the
pre-cambering had to be omitted. The result of this was that the deck profile could only be
adjusted to be what it would naturally adopt between the support positions. The maximum
calculated dead load deflection of 43mm over the 16m between the stay positions was considered
to be acceptable, although it would have been better if this could have been avoided by precambering.
One other consequence of this change was that the bearing at the cutwater had to be fitted prior
to lifting the deck by stressing the stays. Originally it had been intended that it should be left out
until after stay adjustment so as to achieve higher dead load tensions in the steeper stays. In the
event this was not possible, and the stay dead load tensions ended up lower than intended. To
compensate for this, some further additional mass was added inside the deck at the side span
inner stay anchor position in the same manner as at midspan.
Conclusions
As with all innovations, there have been several lessons learnt which would lead to potential
improvements next time. Nevertheless, this unique design has worked well and has proved the
efficiency and effectiveness of the original concept of assembling the bridge deck from a series
of identical aluminium extrusions. The expected economies have been achieved mainly due to
the small number of different components, the avoidance of any added finishes, and the ease of
assembly and erection.
The project proves the effectiveness of aluminium as a structural material for bridges, and the
logic of using large section extrusions as the principal structural component.
All parties have sought to achieve the desired high quality throughout, and the result should be a
bridge which is faithful to the designers intentions and in keeping with the Clients expectations.
A New Model For Cable-Stayed Bridges Control and Adjustment

Joo Srgio CRUZ
Assistant Professor
Civil Eng. Depart./IST
Lisbon, PORTUGAL
Joo F. ALMEIDA
Associate Professor
Civil Eng. Depart./IST
Lisbon, PORTUGAL
Joo Srgio Cruz, born 1961

Received his civil engineering
degree in 1985, the MSC in
1989 and the Ph.D. in 1998
Partner of JSJ-Consult
Joo F. Almeida, born 1957

Received his civil engineering
degree in 1981, the MSC in
1985 and the Ph.D. in 1990
Partner of JSJ-Consult
Summary
The present work concerns the analysis and control of cable-stayed bridges during construction. The
most currently used construction procedures and adjustment criteria are briefly summarised,
illustrating how they can induce important geometrical and stress variations that cannot be
neglected.
A model for the non-linear incremental analysis during construction is presented. The model is
three-dimensional and takes into account all relevant time-dependent and geometrical non-linear
effects.
Based on the model, new techniques for the adjustment of cable-stayed bridges are formulated.
Those techniques simulate the construction sequence, allowing the direct definition, in every phase,
of the segments geometric position and the calculation of the forces that should be applied in each
cable, in order to achieve an appropriate internal forces distribution and the required longitudinal
profile.
The procedure is generalised to include the correction of geometrical and cables tensions deviations
occurred during or after construction.
A practical application concerning a case study of a composite cable-stayed bridge recently built is
presented. The results are compared with values obtained from the construction site, showing the
adequacy of the proposed models.
1. Introduction
Cable-stayed bridges can be built using different construction techniques, which are chosen
according to local conditions and bridge characteristics. Therefore, the relevance of the geometric
control study increases with the complexity of the erection procedure.
Medium span cable-stayed bridges crossing easy obstacles are generally built with simple
construction methods. After the towers completion, the common procedure lies on the construction
of the whole deck on temporary supports or on scaffoldings, that can be adjusted in order to achieve
the correct position for the later cable tensioning. Thereafter, the mounting cable forces are precisely
evaluated to balance the vertical deck reactions on the temporary supports, leading to the pretended
geometry and stresses distribution [2,3].
Otherwise, cable-stayed bridges over any kind of obstacle can be built by the cantilever method,
especially those with long spans. Then, the erection procedure produces deflection and stress
histories, which must be carefully evaluated. In this situation, the geometric control becomes an
important aspect of the bridge analysis, in order to obtain the correct geometry and stresses
distribution [2,3].
Cable-stayed bridges, enable the design of flexible and slender decks and towers, which allow
important adjustments to correct construction differences. Nevertheless, this slenderness, associated
to the variability of the material properties, thermal effects, actions and uncontrollable mistakes
produced during construction, can lead to sensible deviations relative to the theorectical geometrical
profile and stresses distribution.
This fact, on itself, implies a detailed geometrical control study, concerning all the erection
sequence and relevant actions and effects, in order to support decisions during construction to
minimize and correct deviations. Regarding the thermal response behavior of the cables, the study
must include a sensibility analysis to thermal effects in terms of structure stresses and
displacements.
2. Control and adjustment criteria

The geometric control study includes the simulation analysis of the structure erection sequence,
allowing the knowledge of the initial mounting segment position of the deck and towers,
simultaneously with the initial tensioning cable forces. Among others, this study has the main
purpose of evaluating the initial conditions conducting to a pretended final geometric configuration
and stress distribution for a determined temporal horizon t.
It is currently accepted that cable-stayed bridges adjustment criteria lies on the cable force
distribution, for a predefined geometry under dead loads, which minimizes bending moments in
towers and deck, corresponding to a certain time instant t.
This adjustment criteria, established for the final geometry, leads to permanent loads equilibrium
mainly based on axial compression forces in towers and deck, in association with tensions in the
stay cables. Small, span independent, permanent bending moments in the deck, between consecutive
stay-cables, appear at the end. Another direct consequence of this design condition is the
minimizing of the self-induced bending deflections due to creep and non-linear geometric behavior
of the compressed members.
Fig.2.1 illustrate the static system and the variables used for the establishing of final equilibrium
conditions.
Figure 2.1 Cable-stayed bridge static system
Equations (2.1) and (2.2) express the equilibrium of a central span deck n segment.
in
in + 1
(2.1)
sin c,n + cos c,n in
G + Nn + 1 sin
Tc,n =
c,n
cot gc , n + 1
G + Nn + 1 sin c , n 1 +
cot
g
c
,
n
(2.2)
Nn =
cot gc , n
sin c , n 1 +
cot gc , n
The recursive nature of these expressions needs a previous estimate of the axial force in the central
segment at time t, in order to evaluate all the forces Tc,n and Nc,n, from mid span to the towers. As a
remark, these forces applied to the deck only produce bending moments due to the loads in the span
(Hc,n) as indicated in Fig 2.2.
Figure 2.2 Bending moments diagram layout for dead loads in cable-stayed bridge
After the evaluation of the central span stay-cable forces Tc,n with equation (2.1), detail B of the Fig
2.1 indicates the towers central and lateral stay-cables connections, from which equilibrium
equation (2.3) is derived, providing a null bending moment in the towers above the deck.
Tl , n = Tc , n
cos c , n
cos l , n
(2.3)
Finally, the lateral span equilibrium is established with a symmetrical stay-cable arrangement, and
intermediate supports, or even with the help of counterbalance loads distribution.
3. Geometrical and stay-cables forces adjustment model

Traditionally, a backwards analysis would be applied [2,3], consisting in a logic numerical
simulation of the bridge dismounting, from the final state, associated with the design criteria at time
t, through all the intermediate phases in the opposite sequence as it was built.
Such analysis evaluates all stresses increments, computing the stay-cable tensions at the installation
together with the associated position of the structure.
Nevertheless, the backwards analysis is inadequate for the correct evaluation of time-dependent
effects in concrete towers and deck. The nonlinear geometrical effects, sometimes important during
construction, become incorrect by the misevaluation of non linear material behavior [2].
The above aspects lead to a successive series of analysis, regarding dismounting and mounting, with
correction of the initial conditions, which is an heavy process, even using automatic calculation.
This paper presents an alternative adjustment technique based in a convergent iterative process,
which lies on a structural analysis simulation model, regarding all the relevant issues, such as
[1]:
The structure tridimensional nonlinear geometrical behavior is modeled through the
establishment of the equilibrium and compatibility equations on the structure deformed shape.
The geometrical nonlinear cable effect is evaluated through the tangent Ernst and secant
moduli.
Structural system evolution and modification along the time construction is considered.
Loads and action variations are taken into account.
Concrete time-effects like creep, shrinkage, prestress losses due to steel relaxation, are
evaluated with a time incremental analysis. Concrete creep is modeled with an association of n
reological Kelvin models and one Hooke model.
Bearings are modeled with geometrical and physical nonlinear behavior. The effect of the top
plates relative slip displacement with friction is considered.
The prestress cables inside concrete can be considered with or without relative slip friction
inside the gains.
The iterative process starts with a first bridge mounting simulation analysis, based on initial
conditions for the towers, deck geometry and stay-cable forces equal to those predefined at time t.
Obviously this first analysis results on a final geometrical profile and cable-stay forces distribution,
not coincident with the adjustment design criteria. The differences between the obtained solution
and the pretended one, are used to correct the initial conditions of the previous iteration. Hence,
there is a new data block ready to simulate again the mounting construction. This method converges
to the right solution accordingly with the specified tolerances.
Figure 3.1 Indicative scheme with the geometric control variables

The currently used variables in the adjustment procedure are indicated in Fig.3.1, Robj - Tobj, Rti,
Rnfi, Tti and Tn0i. They are, respectively, the final pretended positioning vector and final stay-cable
forces, the positioning final vector obtained at iteration i, the positioning vector after mounting the n
stay-cable at iteration i, the final stay-cable force at the iteration i and mounting stay-cable force n at
the iteration i.
The principal static and cinematic differences concerning the towers displacements, deck geometry
and stay-cable forces are grouped in R i and T i vectors, indicated in (3.1) and (3.2) equations.
Ri = Robj Rti
(3.1)
Ti = Tobj Tti
(3.2)
It is as well possible to define other static vectors of variables differences, like element action
effects Xi and stresses i, indicated in expressions (3.3) and (3.4).
Xi = Xobj Xti
(3.3)
i = obj ti
(3.4)
Based on the fact that the reciprocal influence of the initial segments positioning coordinates,
grouped in R0I, can be neglected, the correction in these initial conditions R0i+1 can be done by
simple adding the differences Ri to the previous position vector R0i. Then, for the next iteration,
the initial geometric conditions vector is obtained by the expression (3.5).
R 0 i + 1 = R 0 i + Ri
(3.5)
This procedure provides the adjustment of current frame bridges, erected by the cantilever method.
However, in cable-stayed bridges the stays mounting forces do not remain constant, resulting on
final forces quite different from the initial ones. Even in this situation, this convergence process can
be enforced in order to obtain an almost correct geometric profile but with wrong stresses.
Therefore, concerning cable-stayed bridges it is necessary to correct the stay-cable vector mounting
forces T0i, in order to achieve a new initial vector T0i+1, leading to the right final stresses distribution
simultaneously with the correct geometrical profile.
Similarly to the correction process of the geometrical initial conditions, the stay-cable convergence
can be simply obtained by adding the force differences Ti to the initial mounting forces vector T0i,
producing a new initial mounting forces vector T0i+1, for the next construction simulation. An
important improvement in the convergence efficiency of this process is made by the introduction of
a well known influence matricial operator Ci differentiating the stay-cable mounting forces
reciprocal influence [1,2].
Nevertheless, in the present technique, the influence matrix concept operator Ci (3.6), has a tangent
matrix significance, including all the information concerning the erection sequence, nonlinear
physical and geometrical effects and specially the influence of the initial conditions of iteration i.
Each column of Ci represents the influence of a stay-cable installation force on the values of the
adjustment control variables [4].
The selection of the bridge n stay-cable forces determines the number of matrix columns. The
number of lines can be selected, corresponding to n control variables, like n-2 stay-cable anchor
deck vertical coordinates and 2 longitudinal towers top coordinates or towers base bending
moments. By this way, the obtained subgroup of Ci operator is a square nonsingular matrix Cni.
Tf
T 10
Xf
Ci = T 10
f
T 10
Rf
T 10
Tf
T 20
Xf
T 20
f
T 20
Rf
T 20
Tf
(n lines )
Tn0
Xf
. .
( m lines )
Tn0
f
. .
( k lines )
Tn0
Rf
. .
( l lines )
Tn0
. .
(3.6)
When a stay-cable is installed, for an iteration i, the vector variables Tf, Xf, f, and Rf, indicated in
expressions (3.7) to (3.10), are, respectively, the actual n stay-cable forces, the linear frame element
m internal forces, the k stresses and the actual structural position l coordinates.
Tf 1
Tf 2

Tf = (3.7)

Tfn
Xf 1
Xf 2
Xf = (3.8)

Xfm
f 1
f 2

f = (3.9)

fk
Rf 1
Rf 2

Rf = (3.10)

Rfp
Once the square matrix operator Cni is assembled, it is simple to conceive an evaluation method for
the installating stay-cable forces correction vector Ti, accounting for the reciprocal force influence.
The differences in the selected adjustment control variables Ri Ti, Xi and i, are grouped in the
Gi vector, allowing the establishment of the linear equation system given by expression (3.11),
Cni Ti + 1 = Gi
(3.11)
providing a solution vector Ti+1 that is added to the previous initial stay-cable installing forces
vector T0i, as indicated in (3.12), in order to improve the convergence for the next iteration.
T 0 i + 1 = T 0 i + Ti + 1
(3.12)
The improved convergence is obtained measuring the different influence that each stay-cable
mounting force has on the other adjustment control variables. Another important aspect, is the
possibility of mixing static and cinematic control variables.
The above procedure can be enlarged to situations where cable forces are installed in multiple steps.
The adjustment process can be applied by choosing the last intervention in the stay-cables to
calibrate the initial force, in order to obtain the final design conditions.
Finally, after the evaluation of the geometrical segments position and mounting or retensioning staycable forces, grouped in the R0 e T0 vectors, it is possible to initiate the construction.
All the above adjustment control procedure lies over the assumption of an uniform thermal
distribution, a load planning and a rigorous mounting sequence. However, uncertainties during
construction are inevitable, having consequences in terms of geometry and stresses deviations.
It is common knowledge that cable-stayed bridges are very sensitive to thermal actions, especially
due to the cables low thermal inertia. Even the stay-cable exterior protection color influences the
thermal behavior. Due to this fact, the mounting stay-cables and retensioning operations must be
executed with low thermal gradients, usually in the morning before the sunrise.
After the bridge structure completion, all the detectable deviations should be checked in order to
permit a final adjustment. That purpose can be performed based on a pre-established stay-cable
intervention sequence, on the actual geometric profile and stay-cable stresses distribution, using the
presented adjustment technique. In this case, it is necessary to redefine the objective adjustment
static and cinematic conditions and simulate the correction sequence iterative process, as explained
before.
4. Geometrical and cable-stay adjustment of pereira-dosquebradas bridge

The present example refers to the adjustment and geometric control study made for the construction
of the new cable-stayed Pereira-Dosquebradas bridge in Colombia.
The four lane bridge has a linear plan alignment and a general 1.5% linear profile with small
concordances adjustments. The structure is a composite concrete-steel deck, lateral stayed by 36
stay pairs on two diamond shaped towers. The symmetrical span arrangement has two small
approach 30.05m spans, two 83.25m lateral spans and a 210.9m central span, in a total of 437.5m
between expansion joints. The towers are reinforced and prestressed concrete box girders. The deck
is made of a steel structure grid and a reinforced concrete slab, connected by full shear connections.
The 0.25m thickness pavement concrete slab was built in a first 0.1m thickness precast slab and a
second 0.15m thickness cover concrete cast in situ.
Figure 4.1 General geometry of the Pereira-Dosquebradas bridge

4.1 Deck erection sequence
The deck segment erecting cycle is presented in Fig.4.2. Due to the small weight of the steel
structure and equipment, along with the need for reducing the stay-cables interventions, it was
necessary to use temporary extraweights during the front steel segment mounting. In that way, it was
possible the installation of half of the stay-cable strands with a convenient stress level. In fact, it is
important to have a minimum number of strands, with a minimum stress value (0.20fpuk), in all staycables to provide a confortable stiffness during construction, which limits, among other factors, the
mounting deck displacements. This fact is strongly associated with the control of the front staycables unload, when a new pair of stays is installed.
The mounting sequence of the bridge deck was chosen to achieve, as soon as possible, the
maximum dead load, minimizing the stay-cables unload during erection. This objective was reached
by concreting the nearest front slab cast in situ layer before the stay-cables retensioning and
mounting phases. Fig.4.2 illustrates the deck segments erection sequence:
1 Derricks positioning on segment i.
2 Elevation and assembling of the steel segment i, together with an 150KN extraweight.
3 Pre-cast slabs positioning on the segment i-2.
4 Concrete cast in situ over the segment i-2 and extraweighs moving to the next segment.
5 Prestress application on segments i-2 and i-3.
6 Retensioning the segment i-2 stay-cables.
7 Installing the segment i stay-cables.
Figure 4.2 Deck cycle segment construction scheme

4.2 Final design bridge adjustment
The achievement of the correct geometric profile and stresses distribution is based upon the criteria
presented in chapter two, consisting in an almost zero bending moment in the towers above the deck
and a rigorous dead load balance of the stayed central span, at a 4000 days time.
Since stay-cable forces are installed in two steps, the last one was chosen, according to the proposed
adjustment technique presented in the third chapter, to define the Gni matricial operator. Therefore,
the n stay-cable forces are also the adjustment control variables. The convergence tolerances on the
geometrical coordinates and stay-cable forces are, respectively, 10 mm and 0.5%.
Fig.4.3 illustrates the comparison between theorectical and measured vertical deck displacements
after the T6L and T8C stay-cables mounting at tower 10. As a final remark, the maximum vertical
displacement due to the T9 stay-cable last retensioning phase, was about 2 meters. This order of
displacement magnitude value, clearly shows the importance of geometric nonlinear effects during
construction.
Figure 4.3 Theorectical and measured vertical deck displacements

after the T6L and T8C stay-cables installation
Figure 4.4 - Longitudinal stresses in the deck steel beams and concrete slab
Figure 4.5 - Towers longitudinal bending moments M3 and longitudinal displacements
Figures 4.4, 4.5 and 4.6 show the convergence process in 8 iterations, in terms of, respectively,
longitudinal stresses on the deck steel beams and concrete slab, towers longitudinal bending
moments M3 and longitudinal displacements, and stay-cable final forces at 4000 days.
A effective convergence is obvious on the towers longitudinal deflections and bending moments. In
each iteration the final geometric profile is always ensured, but only the last one complies with the
aimed stay-cable stresses distribution.
A final stay-cable adjustment was done after the central deck connection, in order to minimize the
overall deviation on the checked control variables.
Figure 4.6 - Stay-cable final forces after 4000 days
5. Acknowledgment
The authors acknowledge Mr. Armando Rito for his valuable suggestions and contributions to this
work. To Mr. Jim Curto the acknowledgment for his contribution on the Pereira-Dosquebradas
example presentation.
6. References
[1] - Cruz, Joo S. N. D. Construction Control of Cable-stayed Bridges (in portuguese) - Ph.D
thesis, Instituto Superior Tcnico, Lisbon-Portugal, August 1997
[2] - Ito, M.; Fujino, Y.; Miyata, T.; Narita, N. Cable-stayed bridges, Recent developments and
their future Seminar, Yokohama, Japan, December, 1991.
[3] - Virlogeux, Michel - Erection of cable-stayed bridges, the control of the desired geometry International Conference A.I.P.C. - F.I.P. - Cable-stayed and suspension bridges, Deauville, October
12-15, 1994
[4] - Fujisawa, Nobumitsu e Nakamura, Nobuhide - Computer system for cable adjustment of cablestayed bridges during erection - International Conference A.I.P.C. - F.I.P. - Cable-stayed and
suspension bridges, Deauville, October 12-15, 1994
Cable Finite Element of High Accuracy

Michel AUPERIN
Special Studies Managing Director
BOUYGUES Travaux Publics
St Quentin en Y., France
Claude DUMOULIN
Civil Engineer
BOUYGUES Travaux Publics
Summary
This paper describes a Stay-Cable Finite Element of High Accuracy with only two nodes based
on the equilibrium of a string subjected to loads distributed linearly along the cable and taking
into account the sag effect. For all static structural analyses, it is convenient to use only one
single finite element, whatever its length is. Regarding dynamic analysis, the inertia forces
contribute to the stiffness matrix. Assuming that these forces vary linearly between two nodes is
sufficient for computing the stiffness. Usually, an important reduction of the node numbers of a
finite element model modifies largely the dynamic behaviour. This paper outlines that an
appropriate choice of the mass matrix leads to an excellent accuracy regarding dynamic analysis,
even with a very small number of nodes. Concerning a stay-cable, a quite exact value of the N
first modes of each type (which means a total of 3N modes) could be obtained with only 2N finite
elements of high accuracy.
1- Introduction
During the eighties, the authors have developed a stay-cable finite element based on the wellknown formula governing the static equilibrium of a string subjected to a uniformly distributed
load. This element is defined by two end nodes, whatever its length is. The suppression of all the
intermediate nodes is worthwhile to reduce the computation time because the main characteristic
of these nodes is their very low stiffness in a plane normal to the cable. The static structural
analyses of Pont de Normandie have been carried out using this element for all the construction
and service loading cases.
Regarding dynamic analysis, this element loss a large part of its advantages, because the
distributed loads include the inertia forces, varying along the cable. It is necessary to model a real
stay-cable by a succession of a large number of cable finite elements based on a uniform applied
load. A new stay-cable finite element was therefore developed taking into account the following
objectives:
Regarding static analysis, a real stay-cable must be modelled by only one element if the
applied loads are linearly distributed.
The dynamic behaviour of a real stay-cable must be simulated with a minimum number of
intermediate nodes.
Regarding static analysis, the objective has been reached with no difficulty. Regarding dynamic
analysis, if a real stay-cable is modelled by a succession of 2N new cable elements, the frequency
value of each of the N first longitudinal modes and transverse modes (in-plane and out-of-plane)
is known with a relative error lower than 0.25 %.
2- Stay-Cable Finite Element of High Accuracy

2.1- Static Equilibrium
Neglecting the torsion and bending stiffness, the cable is assumed to behave like a taut string. Let
us call chord the line joining the two anchorages of the stay-cable and L the distance between
them. x is the relative abscissa along the chord, varying from 0 at the first anchorage to 1 at the
second one. Let us assume that:
The cable is subjected along its length to a distributed load whose each component varies
proportionally to x.
The Hookes law is applied to the stay-cable constitutive material.
Let us call X1k, X2k and Xck (k varying from 1 to 3) the co-ordinates of the two anchorages and of
any ordinary point of the cable expressed in an orthonormal co-ordinate system having one axis
parallel to the chord. Let us write the following relationship between these co-ordinates:
X ck = (1 x ) X 1k + x X 2 k + x (1 x ) Fk [x ; X 1 , X 2 , EA , S 0 , Q ]
(1)
Fk is a x polynomial whose coefficients are depending on the parameters:
X1 : co-ordinates of the first anchorage.
X2 : co-ordinates of the second anchorage.
EA : product of the material Youngs modulus E by the cable cross-section area A.
S0 : cable length obtained when the cable tension is equal to zero.
Q : set of all the cable linearly applied loads.
The distributed load parallel to the chord may be expressed as:
p(x ) = (1 x ) p1 + x p 2
(2)
Let us define a force H by writing the projection Tx along the chord of the tension force T as:
p + p2
p p2
Tx (x ) = H + 1
L (1 2 x ) + 1
L 1 6x + 6x 2
(3)
4
12
Let us then write that the nodal forces, tangent to the cable, equilibrate the tension forces at the
anchorages. Having decided to express the functions Fk as polynomials of degree 2 of teh vriable
x, three conditions concerning the moments have to be defined: the moments are equal to zero at
the anchorages and at the cable mid-span. The polynomial coefficients are then solutions of
linear equations whose parameters are the distributed loads and the force H.
To compute the stay-cable length S, let us assume that the square of the cable slope related to the
chord is small versus 1 and let us expand the S formula until the order 4 according to the
polynomial coefficients. Let us then write the Hookes law, i.e. the average tension Tm along the
cable is proportional to the cable length variation:
S S0
Tm = EA
(4)
S0
Let us define the expressions s and e as:
(5)
S = sL
S 0 = (1 + e ) L
(6)
and a function h as:
( p1 + p 2 ) + ( p1 p 2 )
s 1
+
s
s
and are expressions a little bit complex according to the polynomial coefficients (and
consequently to H) which may be neglected in most of the practical cases. The governing
equilibrium may be expressed as:
h = EA e
This allows computing of:
e and consequently the length S0 if the tension is known at one point of the cable (cable
tensioning problem).
H and consequently the tension and the deformed shape of the cable if e is known.
h = H EA
(7)
(8)
2.2- Cable Stiffness

Let us define the parameter r by the following formula:
dh
dh
dL
1
=

=
(9)
r
d H c
d L c
d H c
The subscript c denotes that the derivatives are evaluated assuming that the loads are constant
according to intensity and direction. This coefficient r plays an essential role in the stay-cable
analysis. It is always positive and its value is lower or equal to 1. If the stay-cable is only
subjected to its dead load, r may be expressed according to the well-known Irvines parameter 2
as:
12
r =
(10)
12 + 2
It could be easily proved that, if L is the extension of the chord, the total loads along the cable
changing neither in intensity nor in direction, the projection Tx of the tension force along the
chord truncated to the first order might be expressed as:
EA
Tx = r
L
(11)
S0
The extension stiffness of a beam where A is the cross-section area, S0 the length and made of an
elastic material whose Youngs modulus is E, may be expressed as EA/S0. The expression (11)
exhibits that the cable extension stiffness is r times those of the beam.
Regarding a stay-cable whose slope versus the horizontal axis is defined by the angle and
whose applied load is only its own dead load, the tangent stiffness matrix of the cable may be
expressed in the vertical plane as:
A A
K =
(12)
A A
where A is a symmetric matrix (2,2) whose coefficients aik are given by:
EA
H
H
a11 = r
a12 = a 21 = (1 r ) tg
a 22 =
(13)
S0
S0
L
In the previous expression of A, the X-axis is assumed parallel to the chord.
2.3- Mass Matrix
There is a complete analogy between the equation of motion of a bar subjected to dynamic axial
deformations and a taut string subjected to dynamic transverse deformations. The mass matrix M
of a straight bar element with two nodes, of length l and mass m per unit of length, may be
expressed as:
a
0.5 a
M = ml
(14)
0.5 a
a
where a is a coefficient whose value is often taken equal to 1/2 or 1/3. Let us consider a bar of
length L, fixed at each end, and let us divide it into N elements of the same length l. Computing
the frequencies of this finite element model (N two nodes bar elements) subjected to axial
deformations under free-vibration conditions, the following results are exhibited:
To obtain accuracy about 0.25 % on the frequency of the first mode, the bar must be divided
into 13 elements.
The error values are the same, except the sign, when the two well-known values of a are
used.
When N is large versus the mode order k, the convergence towards the exact solution is of order
N-2. It is of order N-4 if the value 5/12 is used as a value, i.e. the mean value of the two well
known values. In fact, the objective is a little bit different: what is the most efficient value for the
parameter a in order to obtain the first k natural frequencies with the minimum elements number
N and for a given accuracy? A detailed analysis shows that the best a value is 0.40733. Then the
relative error related to the computed natural frequency is more low than the order of the mode
small and is not larger than 0,25 % concerning the mode whose order is equal to the integer part
of N/2 (for instance, mode 2 for 4 or 5 elements). A similar analysis, of course more
sophisticated, has been developed for straight beams, exhibiting also interesting results.
Let us come back to the stay-cable problem. Considering a stay-cable only subjected to its own
dead load, it can be demonstrated that the first in-plane natural frequency f1 of the cable with
fixed anchorages may be expressed as:
1
H
f1 =
(15)
2L m r
The difference between the result obtained by this formula and that given by the more
sophisticated expression proposed by Irvine is insignificant, even when the static sag of the staycable is large.
Finally, putting together all these results and assuming that the co-ordinates system is
orthonormal with an X-axis parallel to the chord, the mass matrix of the cable finite element may
be expressed as:
M1 M 2
M = m S0
(16)
M 2 M1
where M1 and M2 are two diagonal matrixes (3,3):
1
(1 , 1 , 1) Diag (M 1 )
(17)
2
The ry and rz coefficients depend on r and on the curvatures in the planes (XY) and (XZ). If the
stay-cable is not loaded in the plane (XY) for instance, ry is equal to 1 et rz is equal to r.
Diag (M 1 ) = 0.40733 (1 , ry , rz )
Diag (M 2 ) =
Concerning dynamic analyses, the cable element stiffness matrix is that obtained from the static
study and the mass matrix is that given by (16) and (17). Modelling a stay-cable with only two
elements, the frequencies of the first in-plane mode, of the first out-of-plane mode and of the first
axial mode are obtained with a relative error of 0,25 %. The result obtained for the axial
vibrations of bars could be generalised. To have a relative error lower than 0,25 % on the
frequencies of the first k in-plane modes, of the first k out-of-plane modes and of the first k axial
modes, only 2k cable elements of high accuracy are required.
3- Example
Consider a stay-cable with the following characteristics:
Chord length at rest
L = 210.50 m
Slope angle versus horizontal
= 22.3
Stiffness
EA = 2018.25 MN
Tension at upper anchorage
T2 = 5.200 MN
Cable mass (not tensioned)
m = 95.7 kg/m
(total masse: 20094 kg)
The e parameter of the stay-cable has a value of 0.25 %. The cable static sag related to the chord
is equal to 0.930 m. These static results are obtained whatever the number of cable elements are
used (even with one element). The r coefficient, related to the complete stay-cable, is equal to
0.96. The frequency computed for the first in-plane eigenmode depends on the number of
elements:
2 elements: 0.5620 Hz
The transient analysis of the stay-cable vibrations induced by the lower anchorage moving
vertically at the frequency computed for the first in-plane mode and with 2-cm amplitude, has
been carried out. The duration of the analysis was 900 seconds. The initial state was at rest i.e.
the static equilibrium. The damping viscous ratio was very low ( = 0.1 %). The structural
analysis software package used was Pont/ARC, which is a Bouygues own developed code. The
time integration method is derived from the Bulirsch-Stoer method, which is an extension of the
well-known Runge-Kutta method. The stiffness and mass matrixes have been computed at each
time step. Figure 1 details the results.
One notices that the maximum amplitude of the displacement and the tension decreases when the
number of elements increases, stabilising rather quickly: there are only small differences between
the responses obtained with four or six elements. The difference is more important concerning
the responses obtained with two elements. Nevertheless, a small increase of the excitation
frequency (0.001 Hz in our example) modifies largely the amplitude of the response. It is a wellknown phenomenon due to the high non-linearity behaviour of the dynamic stay-cable motion.
The amplitudes of the displacements and the tension variations obtained with two elements are
close enough to those obtained with four elements with a frequency a little bit increased.
Mid-Span Displacements (unit meters)

Figure 1
Tension (unit MN)
Stay-cable response to a vertical excitation with 2-cm amplitude of the lower anchorage. Viscous
damping ratio: = 0.1 %. Left column: mid-span displacements, normal to the chord. Right
column: tension at the upper anchorage.
(a) and (b)
two finite elements model. Excitation frequency: 0.562 Hz
(c) and (d)
four finite elements model. Excitation frequency: 0.564 Hz
(e) and (f)
four finite elements model. Excitation frequency: 0.565 Hz
(g) and (h)
six finite elements model. Excitation frequency: 0.5637 Hz
4- Conclusion
The Cable Finite Element of High Accuracy developed in this paper meets to the prescribed
objectives. Regarding static analysis, as far as the loads applied on the cable could be assumed to
be linearly distributed, one single element is sufficient to model the stay-cable, whatever its
length is. Regarding dynamic analysis, the elements number depends on the goal:
If only an estimation of the displacement or tension amplitudes is wanted, only 2N elements
are necessary for results till modes of order N.
If a good accuracy of the transient response is expected, then the number of elements has to
be increased, for instance by doubling the previous value.
5- References
[1]
Irvine, H. Max, 1981, Cable Structures, M.I.T. Press.
[2]
Stoer, J., and Bulirsch, R. 1980, Introduction to Numerical Analysis, New York: SpringerVerlag.
Sunniberg Bridge, Klosters, Switzerland

Karl BAUMANN
dipl. Ing. ETH
Bnziger+Kppel
+Brndli+Partner
Chur, Switzerland
Karl Baumann was born
in 1960. He received his
Civil Engineering degree
in 1984 from the ETH
Zrich. 1991 he joined
BKB as Project Manager
for bridge projects.
Jrg DNIKER
Bauing. HTL
Stahlton AG
Zurich, Switzerland
Jrg Dniker, born 1946,
received his BS in Civil
Engineering in 1969.
After two years of
experience in design of
steel structures, he
worked in Australia for a
Consultant. Since 1974
his activities concentrate
on special post-tensioned
Structures and QualityManagement
Bridge Concept
The Sunniberg Bridge forms part of the main road connection between Landquart and Davos. It
crosses the Prttigau valley close to Klosters in a sweeping curve, high above the valley floor.
The Bridge is the most visually impressive structure of the Klosters by-pass project.
Fig. 1
View of the finished bridge
In view of the prominent location of the bridge and the importance of the surrounding, still
largely unspoiled, alpine landscape, the aesthetic quality of the design was of particular
importance. When viewed from the ascending approach road, the bridge appears shorter than it
actually is. Span lengths of significantly greater than 100m were therefore desirable. An elegant,
modern and original structure was conceived, which convinces the beholder that, in addition to
1
its strengths in farming and tourism, this mountainous region can also contribute to technical and
cultural development. The convincing aesthetic appearance justifies the additional costs of about
15%, compared to the least expensive alternative.
Given the topography of the site and the chosen route, the desired transparency and longer spans
could only be satisfactorily achieved by a cable-stayed bridge. Multi-span cable-stayed bridges
present problems with regard to statical behaviour and appearance. The statical problem is that
the load-carrying system at each pylon must be stabilised under localised loading. Visually, the
ratio of the tall piers to the normally equally tall pylons can be disharmonious and rather
unconvincing.
The curved form in plan allows for a monolithic structure, fixed at both abutments, without
expansion joints. It is thus possible to fix the pierheads at deck level in both the longitudinal and
transverse directions. Changes in temperature lead to horizontal deflections perpendicular to the
bridge axis, without causing appreciable secondary stresses in the deck cross-section. The piers,
however, are subjected to horizontal displacement at the pierhead, and must consequently be
designed as slender frame constructions. Viewed from the side, the piers become narrower from
top to bottom, corresponding to the force flow in the piers under unbalanced traffic loading.
At the upper extremity, the widening pier section is consistently continued into the pylon. This
results in a broad, stiff pylon diaphragm, which facilitates a very gentle stay cable gradient.
As a result of the fixation of the pierheads by the deck, and the stiff arrangement of the pylons,
the deformations under traffic loading over a single span can be kept within acceptable limits,
even with very gently inclined stay cables.
The combination of pylon, piers, deck and cables results in a balanced structure with impressive
architectural qualities.
Fig.2
Pylon model
2
Description of the Bridge

General
The Sunniberg Bridge crosses the valley at a height of about 60m over the Landquart river, from
the Bel plateau to the slope of the Gotschna, close to the Drosbach stream. The bridge is curved
in plan, with a radius of 503m, measured to its axis. The deck has a longitudinal inclination of
3.2%, and a transverse inclination of 7%.
The span lengths are 59m, 128m, 140m, 134m and 65m, resulting in a total length of 526m.
Fig. 3
Longitudinal section with construction programme
This arrangement of spans is ideally suited to the bridge location. The locations of the Bel and
Drostobel abutments were chosen to be as close as reasonably possible to the tops of the valley
slopes. The spans are generous and contribute to the transparent appearance of the structure.
The usable width of the Sunniberg Bridge is 9.0m. The deck is bounded on either side by a 1m
high New Jersey barrier, capped with a steel tube railing.
Bridge Deck
The bridge deck is formed by a 12.375m wide plate, simply reinforced in the transverse direction,
between two edge-beams. The thickness of the plate varies in the transverse direction between
0.40m at the centreline and 0.32m at the edge beam. Outside the New Jersey parapet, at the edges
of the deck plate, the thickness varies between 0.4m and 0.5m. For statical reasons, the thickness
of the bridge plate is increased close to the pylons. The plate is 0.55m thick at the centreline over
the initial 13m length either side of each pylon, before the first stay cable anchors. During the
first two balanced-cantilever construction stages, the thickness of the plate at the centreline
reduces linearly from 0.55m to 0.40m.
3
Fig. 4
Bridge cross-section
The stressing niche zones, together with the continuous edge beams, form two massive edge
reinforcements, each about 1.90m wide. The stressing niches for the stay cables are located
outside the edge beams, and are in section slightly asymmetrically positioned.
In the mid-span areas, the missing axial force is compensated by longitudinal post-tensioning:
2x3 pcs. 1'900kN cables, pulled in and stressed following the closure of the joint between the
two cantilevers.
Stay Cables
The stay cables are arranged in a harp configuration, with a 6m spacing between the deck
anchorages. In view of the radius of the bridge deck, this configuration is required to ensure that
the planes of stay cables on either side of the bridge deck give the impression of continuous and
reassuring "walls". The average inclination of the cables is 1:5, with variations arising from the
longitudinal inclination of the bridge deck.
The dominant feature of the system is the stiff pylon. Changes in cable forces on one side of the
pylon therefore have practically no effect on the forces in the cables on the other side, but are
rather compensated by bending of the pylon. The commonly encountered, and for stay cable
systems often crucial, backstay cables can be totally omitted.
The maximum working load of the cables was set at a conservative level of 50% of the ultimate
capacity. The maximum stay cable forces at the Sunniberg Bridge is 5'000 kN.
Pylon and Transverse Beam
The pylon rises about 15m above the deck, in the form of two diaphragms outside the deck plate.
These two diaphragms are inclined outwards at an inclination of 8:1. The inclined arrangement is
prescribed on the one hand by the geometry of the stay cables for the curved structure, and on the
other hand by the overall aesthetic appearance of the pier and pylon system.
Fig.5
Pier and pylon
With a width of 5.95m - 8.00m in the longitudinal direction, the pylon diaphragm resists the
bending forces resulting from non-symmetrical traffic loading. Its thickness of 1.75m allows it to
accommodate the bending moments caused by the transverse component of the stay cable forces.
The fixed upper anchorages of the stay cables are situated in the central part of the pylon
diaphragm. This arrangement leaves the outer parts of the pylon available, unreduced, for the
resistance of bending forces. The anchorages are accommodated in two back-to-back steel
anchorage boxes which are connected together by steel plates (S355J2G3), each 500x30mm.
Fig. 6
Upper cable anchorages
The very massive transverse beam, with a height of about 3.1m and a width of about 2m,
transforms the large transverse bending moments of the pylon diaphragms into unequal axial
loading on the legs of the pier. The inner leg supports about 60% of the total axial load.
Piers
In the longitudinal direction, the lines of the edges of the piers are parabola-shaped. Below the
deck level, the piers become narrower with decreasing elevation, changing to a slight widening
above the foundations, in the case of the tall P2 and P3 piers. The variation in width, from 3.30m
to 5.95m, is achieved by varying the flange width and the width of the recessed central part of the
pier leg. The edge details and the transverse width of the pylon leg (1.60m) remain constant.
In the transverse direction, the transition from the vertical pier legs above the foundations (total
width 8.80m) to the 8:1 inclined pylon occurs over a vertical distance of 36m between the lower
transverse connector and the pylon transverse beam (total width 13.42m at the base of the pylon).
Abutments
The abutments are connected monolithically to the bridge deck, and are the anchor locations for
the horizontal stabilisation of the bridge system. The size and shape of the abutments have been
largely determined by the tensile anchoring forces required at each abutment (at design level:
17'700 kN for the Bel abutment, 14'600 kN for the Drostobel abutment).
The abutments consist in principle of earth-filled containers, each with floor, sidewalls and
central support wall.
Foundations
The 3 piers P2, P3 and P4 are each founded on 6 bored piles (dia. 1.50m) between 14m and 16m
long. The massive pilecap (11.6m x 7.2m x 3.0m) is offset in plan 0.75m towards the inner side
of the curve of the bridge, since the inner leg of the pier supports a considerably larger part of the
total vertical load.
Pier P2 is situated close to the bank of the river Landquart, and the piers P3 and P4 are situated
close to the Drosbach stream. The foundations have been designed to withstand even very severe
flooding, with several meters of ground loss.
6
Pier P1, closest to the Bel abutment, is situated on a flat area of ground, adjacent to the steep
slope into the Prttigau valley. The foundation is constructed in the form of two pits, about 17m
deep. The upper 4m of each pit has a diameter of 4.75m, to suit the dimensions of the foot of the
pier. Otherwise the diameter of the pits is 3.5m, due to construction considerations.
Finishing Works
The bridge deck has been sealed with polymer-bitumen mats and has been paved with 3 layers of
asphalt, with a total thickness of 16cm. The edge detail at the New Jersey barrier construction has
been sealed with a flexible joint strip. The edge of the deck plate over the stressing niches has
been treated with a 4mm thick coat of 2-component epoxy-polyurethane varnish, sprinkled with
quartz sand.
The drainage of the bridge is achieved by means of drainage gullies spaced at about 30m,
connected to a HDPE longitudinal drainage pipe. The drainage pipes and the protection pipes for
various services are fixed to the underside of the bridge, between the edge beams.
For the purpose of maintenance, a special movable platform, supported at the edges of the deck
plate, is foreseen.
Design of the Bridge

Gently inclined stay cables
Due to the gentle gradient of the stay cables, the vertical deformation of the deck under traffic
loading on a single span is a significant design criterion. A maximum allowed vertical
deformation of 1/400 the span length was agreed with the client. The design load on a single span
is composed of a uniformly distributed load of 2 kN/m2, combined with a point load of 360 kN
(including an impact factor of 1.2). The vertical deflection is calculated for the system without
cracks, with a Modulus of Elasticity of the concrete (Eco) of 35'000 N/mm2, ignoring the
concrete parapets.
Fig. 7
Deformations
In the longest span of 140m, the maximum deflection is 225mm, corresponding to 1/600 the span
length. The deflection results 40% from the deformation of the pylon and pier and 60% from the
elastic deformation of the stay cables. The two neighbouring spans exhibit upward vertical
displacements of max. 60mm in this case. It was not considered reasonable to increase the steel
section of the cables, just to further restrict the deformations.
Is is clearly apparent that the chosen system, with a cable gradient of 1:5, has little in reserve
with regard to allowable deformation, and that the fixed connection of the deck to the abutments,
with the consequent fixation of the pierheads, was a necessary measure.
Curvature
The deflected forces resulting from the curvature of the bridge lead to large transverse bending
moments in the lower region of the pylon diaphragm. At the foot of the pylon on the outer side of
the curve, the bending moment at design level reaches ca. 50 MNm (including secondary effects).
This bending moment is taken up by vertical pre-stressing cables in the pylon diaphragm and
transformed by the massive transverse beam, pre-stressed by 6 cables of capacity 2'350 kN, into
unequal vertical loads on the two pier legs.
Fig.8
Bending moments Myd in pylon and transverse beam
Connection of the deck to the abutments

Due to the curvature in plan, with a radius of 503m, the bridge deck can be connected
monolithically to the abutments, without causing appreciable secondary stresses in the deck
cross-section. The missing axial force in the stage between the last stay and the abutment is
provided by 2 x 3 x 1'900 kN cables.
Fig. 9 Secondary forces

The foregoing diagrams plot the secondary axial force, the concrete stresses at mid-arch, and the
horizontal displacement of the bridge deck, as a function of arch length, for a restrained arch. The
cross-sectional dimensions of the Sunniberg Bridge and a temperature variation of 10 are
assumed. For the system without piers portrayed in the diagrams, the secondary axial force
almost disappears, for an arch length of 530m. The horizontal displacement at the crown of the
arch is ca. 90mm.
In the actual system constructed, with the change in radius being restrained by four piers, the
secondary axial force is 2'200 kN, while the horizontal displacement is 50mm.
Construction of the Bridge

Construction schedule
The client scheduled the construction of the bridge for the period between June 1996 and August
1998, with the deck insulation and paving works being performed in the autumn of 1998, so that
the bridge would be ready for hand-over to the contractors performing the construction of the
adjacent Gotschna Tunnel in November 1998. Material excavated from the tunnel must be
brought over the completed bridge.
The time schedule demanded a well planned erection concept from the contractor (see Fig. 3).
Working with two major groups, one in charge of the pier and pylon construction and the other in
charge of the cantilever construction, a good solution was found, which enabled the works to be
performed in a repetitive weekly cycle - an important factor in cantilever construction.
The main items of equipment used during the construction were:
- two pairs of form travellers,
- two large tower cranes, situated close to the pylons, 78 m high and with reaches of 60 m and
70 m,
- a smaller tower crane,
- a concrete batching plant.
Thu.
Concreting of edge beams and deck plate, with simultaneous reduction of ballast.
Fri.
Removal of frontal formwork, positioning of the support rail of the travelling formwork for the next
stage, mounting of stressing jacks at the stay cable anchorages, and initial stressing.
Mon.
Stressing of stay cables in 4-6 steps, with simultaneous increase in ballast (up to concreting stage 5),
lowering of travelling formwork onto the support rail, and moving forward of travelling formwork;
positioning of formwork and cable trumpets at one cantilever arm.
Tue.
Positioning of formwork and cable trumpets at the second cantilever arm, placing of reinforcement for
the first cantilever arm, preparation of stay cables for installation.
Wed.
Placing of reinforcement for the second cantilever arm, installation of stay cables for next stage.
Uncoiling of stay cables for the next stage.
Fig. 10 Weekly cantilever cycle (The weekly cycle for stay cable installation is shown in Fig. 13)
An unconventional form traveller concept was proposed and subsequently implemented by the
contractor. The form traveller extended over two 6m stages, with the leading edge beams and the
trailing deck slab being poured in each concreting cycle. This procedure has the following
advantages:
- the form traveller is better balanced
- the weight of the form traveller (40 tons) can be introduced into the cantilever near the stay
cables of the previous stage
- installation and stressing of the stay cables can take place at the leading edge beams, before
moving the traveller forward into the next stage.
Fig. 11 Travelling Formwork

Stay cable fabrication
For the stay cables of the Sunniberg Bridge, a parallel wire system with DINA anchorages was
chosen. The special features of this system are:
- cables are factory-fabricated, and ready for installation upon delivery to site.
- cables and anchorages exhibit high fatigue resistance.
- the cable has a high, constant modulus of elasticity, corresponding to the modulus of elasticity
of the constituent wires.
10
A total of 148 stay cables, each constituted of between 125 and 160 wires, each with diameter
7mm, were fabricated. Cable lengths vary between 11.4m and 67.8m, and the guaranteed
breaking loads of the cables vary between 7'695 kN and 9'850 kN.
Cold drawn, galvanised wires with a nominal strength of 1'600 N/mm2 and a minimum zinc
coating of 280 g/mm2 were cut to a predetermined length. This predetermined length is
calculated taking the distance between the anchor plates on the completed bridge structure, as
well as the corresponding cable forces, into consideration. The individual wires were bundled
together before being pulled through a bath of molten corrosion inhibiting compound, directly
into the HDPE stay tube.
The corrosion inhibiting compound is a microcrystalline wax with a dropping point of approx.
80C. The same material was also injected into the HDPE tube, following the fitting of the DINA
anchorages to the ends of the cable. This injection of corrosion inhibiting compound fills the
spaces between the wire bundle and the duct.
Fig 12 DINA anchorage

The wires are anchored in the DINA anchorages by means of button-heads. A special epoxy
compound prevents the ingress of oxygen into the anchorage zone, eliminates fretting between
wires and the steel anchorage body, and facilitates a smooth introduction of cable forces into the
anchorage. The DINA anchorages are designed to withstand fatigue stresses of up to 250 N/mm2
over 2 million load cycles.
Coiled on reels, the stay cables were ready for transport to site and installation. The fabrication of
the cables in a permanent plant guarantees a high and controllable standard of quality.
Cable installation
Uncoiling of the 4 cables per stage was performed by means of a reel stand and a winch on the
bridge deck. Following the permanent welding of the telescopic joints in the HDPE stay pipe at
either cable end, the upper end of the cable was placed on a deviation saddle, which was then
lifted to the anchorage location at the pylon. By screwing the lock nut onto the threaded DINA
anchor, the upper end of the cable was secured in its final position (see Fig. 6). The lower portion
of the cable was then lifted by crane onto a tubular steel scaffolding, equipped with rollers,
positioned over the starter reinforcement of the New Jersey barrier, with the lower anchorage
being pushed into the mouth of the steel trumpet. On the day following the placing of concrete,
the lower anchorages were pulled through the trumpets and temporarily secured.
11
Fig. 13 Stay Cable Erection Stages

Cable stressing
Three days after concreting, the concrete strength required to allow the tensioning of the cables
was reached. Using four 5000 kN hydraulic jacks, the cables were stressed in several stages,
determined by work which had to be performed on the form traveller, and by the placing of
ballast. These operations had to follow a carefully planned procedure. After each step, various
measurements, such as cable forces, deformations of the cantilever, pier and pylon, and cracking
of the concrete, were made and compared to the expected values. The regulation of cable forces
was very efficient, as increases and reductions of cable forces were easily and accurately
possible, by adjusting the position of the lock nut on the continuously threaded anchorage (see
Fig. 10).
Stability during construction
The 70m long cantilever on the 60m high piers had to be stabilised against dynamic wind loads.
This was achieved by securing the deck at the quarter-span location to the foundation plate, itself
anchored with soil anchors, by means of crossed pre-stressing cables, each made up of four 0.6"
strands. These stabilisation cables had to be de-tensioned each time before taking deformation
measurements, as they significantly influenced the free movement of the whole structure.
12
Costs
The final total cost, including taxes, of the Sunniberg Bridge was SFr. 20 million, corresponding
to a unit price of SFr. 3'075/m2
Split up of the Total Costs
Site set-up
Special foundations
Ground works
Abutments
Piers, pylons incl. foundations
Bridge deck, incl. travelling formwork
Stay cables
Insulation and paving
Railings
8.3%
3.0%
1.5%
1.4%
20.6%
33.5%
23.1%
7.4%
1.2%
Credits
Outline Design
Detailed Design
Site Supervision
Main Contractor
Stay Cables, Post-Tensioning and Ground Anchors
Tiefbauamt Graubunden
Consultant: A. Deplazes, dipl. Arch. ETH
Concept: Prof Dr. C. Menn, Chur
Bnziger + Kppel + Brndli + Partner, Chur
Wst + Trb + Partner, Schaffhausen
Arge Sunniberg (Vetsch, Klosters; Preiswerk + Cie. AG
Brckenbau, Siebnen)
Stahlton AG, Zurich
References
[1]
Schweizer Ingenieur und Architekt SI+A, Nr 19 and 44/1998
13
Structural Countermeasures for Design

of a Very Long-Span Cable-Stayed Bridge under Wind Loads
Ken-ichi MAEDA
Professor Dr.Eng
Tokyo Metropolitan Univ.
Hachiouji, Japan
Hitoshi NAKAMURA
Research Associate M. Eng
Hachiouji, Japan
Yu MOROYAMA
Graduate Student
Hachiouji, Japan
Makoto ABE
Research Engineer
Chodai Co. Ltd.
Japan
Makoto KONNO
Research Engineer M. Eng.
Nippon Kokan K.K.
Japan
Summary
The purpose of this study is to present adequate structural countermeasures for reduction of the
stress resultant under design wind loads, which becomes dominant in the static design due to the
decreased width-to-span ratio, and for improvement of the static and dynamic aerodynamic
stability in the wind-resistant design. In this paper, by using the example of a trial-design bridge
with a center span of 1,500 m which is considered the critical span length, the authors clarified
the usefulness of the proposed countermeasures from the viewpoint of cost efficiency and windresistant stability, and confirmed the realizability of very long-span cable-stayed bridges in the
near future.
1. Introduction
The Tatara Bridge being constructed in Japan with a center span of 890m, which is the longest
span length of cable-stayed bridges in the world, is scheduled for completion in the spring of
1999. The development of cable-stayed bridges has been rapid, and the class of bridges with a
center span of 1,000m is already planned for construction, according to the latest information
[1]. The critical span length for cable-stayed bridges is reported to be about 1,500m, mainly
because the in-plane buckling stability of main girders is degraded with increasing compressive
axial-forces under dead and live loads for the static design [2].
However, with decreasing width-to-span ratios of main girders due to the increased span length,
it is predicted that the stress resultant under design wind loads becomes dominant in the static
design and influences the cost effectiveness [3]. In addition, it is also predicted that ensuring
safety against lateral-torsional buckling instability and coupled flutter under strong winds will
become very important. Moreover, to satisfy these requirements, bad influences of the
flexibility of stay cables as determined by their own weights and caused by wind actions cannot
be neglected [4].
The aim of this study is to present adequate structural countermeasures for reduction of the
stress resultant under design wind loads in the static design, and for improvement of the static
and dynamic aerodynamic stability in the wind-resistant design. Also, the authors intend to
clarify the effectiveness of these structural countermeasures, and to confirm the realizability of
very long-span cable-stayed bridges with center spans of 1,000m to 1,500m in the near future.
For this purpose, the authors first developed a basic design model based on the trial design of a
cable-stayed bridge with A-type towers and a center span of 1,500m, which is considered the
critical span length for cable-stayed bridges. Then it was confirmed that the stress resultant due
to design wind loads became dominant in a wide region of main girders near each tower.
Next, to flexibly change the out-of-plane support conditions between main girders and towers,
which are considered to greatly affect the static and dynamic deformation characteristics under
wind loads, the authors produced an alternative design model with three-dimensional (3-D) Atype towers. Cable-stayed bridges with 3-D A-type towers are attracting attention due to their
increased in-plane rigidity.
The structural countermeasures proposed and investigated by using the above-mentioned basic
and alternative design models in this paper are as follows:
(1) Application of elastic out-of-plane supports between main girders and 3-D A-type towers,
positioned away from the central line of each tower.
(2) Application of new auxiliary cable systems for controlling the flexibility of stay cables
caused by wind actions in the out-of-plane direction.
The analytical modeling was performed so that not only the design conditions peculiar to cablestayed bridges, such as cable prestresses, are satisfied, but also the flexibility of stay cables in all
directions is taken into consideration. Then, with respect to the effectiveness of the first
countermeasure to reduce the stress resultant under design wind loads, the authors compared the
values of out-of-plane bending moments and out-of-plane lateral displacements of main girders,
and examined the optimal position of the elastic supports and the optimal spring constant.
With regard to the effectiveness of the second countermeasure in improving the static
aerodynamic stability, the authors roughly investigated the optimal number of steps of stay
cables for which new auxiliary cable systems are installed, and then compared the values of
lateral-torsional displacements and critical wind velocities against buckling instability, by
applying static aerodynamic forces as a function of the angle of attack. Furthermore, with
respect to the effectiveness of the second countermeasures in improving the dynamic
aerodynamic stability, the authors carried out multi-mode flutter analyses by applying unsteady
aerodynamic forces, and investigated the natural vibration characteristics and the coupled flutter
behaviors.
2. Basic Design Model and an Alternative Model

To verify the necessity and effects of the structural countermeasures proposed, the authors first
performed the trial design of a cable-stayed bridge with A-type towers and a center span of
1,500m, and developed a basic design model, in accordance with the specifications for longspan bridges in Japan [5]. In this process, the in-plane buckling stability of main girders against
compressive axial-forces due to dead and live loads for the static design was checked according
to the specifications for highway bridges in Japan [6].
Fig.2.1 shows a general diagram of the basic design model, and Table 2.1 lists major crosssectional properties. Fig.2.2 shows a sectional view of a box-shaped main girder and its
components (steel type; yielding stress, tensile strength, thickness of web, upper and lower
flange plates, according to JIS). In this figure, the hatched region is that in which stresses due to
out-of-plane bending moments under design wind loads become dominant in the static design of
main girders.
In addition, to flexibly change the out-of-plane support conditions between main girders and
towers, the authors also produced an alternative design model which has three-dimensional (3D) A-type towers. Because the out-of-plane support conditions of main girders are considered
to greatly affect the static and dynamic deformation characteristics under wind loads. Figs.2.3
(a) and (b) show the shapes of towers in the basic and alternative design models, respectively; in
the alternative design model, the cross-sectional properties of the region where each tower is
three-dimensionally separated into two pillars were set as half those of each tower in the basic
design model.
300
C
L
40
Girder
Tower
90
20@21=420
(1)
1.43
6.1
113.7
12.2
(2)
1.82
7.8
144.7
16.7
(3)
2.10
9.1
166.7
19.2
(4)
2.39
10.3
188.7
23.0
Lower
1.79
29.2
45.4
40.4
9.6
9.6
26.5
Member*
12.8
80
20@35=700
550
20
20@35=700
80
20@21=420
90
550
1500
Cable
A(m ) Iin(m ) Iout(m ) J(m )
Upper 1.54
1st 0.015 36th 0.030
*(1) - (4) : section No. (see Fig.2.2)
Fig.2.1 General diagram of the basic design model
Table2.1 Major properties
C
L
23.0
9.0
1.5
9.0
tf
tw
4.1
21.8
SM490(3200,5000)
tl
4.1
30.0
(Unit : m)
Grade of material
y*(kgf/cm2)
SM570
SM490
SM570
SM490
4600
3200
4600
3200
tu*(kgf/cm2)
Section No.
5800
5000
5800
5000
(1) (2)
(3)
tf 12
14
14
Plate thickness
19
19
tw 16
(mm)
16
22
tl 10
* y: yeileding stress, tu: tensile strength
(4)
(3)
(2)
(1)
17
14
14
12
22
19
19
16
26
22
16
10
: region where stresses due to wind loads are dominant.
Fig.2.2 Sectional view of main girder and its components
300
300
40
40
(a) Basic model (b) Alternative model
(a) Basic model
(b) Alternative model
Fig.2.4 Skeleton diagrams of both analytical models
Fig.2.3 Shapes of towers

in both design models
Next, two analytical models for the basic and alternative models were constructed by modelling
box-shaped main girders as fish-bone structures. For these analytical models, to carry out
analyses taking into consideration the flexibility of stay cables as determined by their own
weights and caused by wind actions, all stay cables were modelled as linked structures by
equally dividing each cable into 8 axial-force members. Then the initial conditions of both
analytical models were determined so that the design conditions peculiar to cable-stayed
bridges, such as cable prestresses, are satisfied [7]. Figs.2.4 (a) and (b) show skeleton diagrams
of both analytical models under the equilibrium conditions of completion in which dead loads
and cable prestresses are applied.
3. Reduction of Stress Resultant in Static Design

3.1
Countermeasure for Stress-Resultant Reduction
Based on the results of the trial design for producing the basic design model described in the
previous chapter, it was confirmed that the magnitude of the stress resultant under design wind
loads greatly influences the cost efficiency in the static design, with decreasing width-to-span
ratios of main girders due to the increased span length. Furthermore, it was evident that this is
due to the installation of wind-shoes in towers, which fix and support main girders in the out-ofplane direction.
Therefore, as a structural countermeasure for reducing out-of-plane bending moments of main
girders under design wind loads and consequently increasing the cost efficiency, the authors
devised a method for elastically supporting main girders in the out-of-plane direction away from
the central line of each tower. Namely, in this proposed method, main girders are elastically
supported at optimal positions in 3-D A-type towers in the alternative design model as shown in
Fig.3.1; hence, tower-links, which support main girders in the vertical direction, should be
displaceable both in the bridge-axis direction and the out-of-plane direction perpendicular to it.
Fig.3.2 shows an example of the structure of tower-links in which displacements in the two
directions are possible.
3.2
Analytical Results and Optimum Condition
To verify the effects of applying elastic out-of-plane supports between main girders and 3-D Atype towers, finite displacement analyses, based on the approximate updated Lagrangian method
(AULD method) [8][9], were carried out by using the basic and alternative design models.
Design wind loads were applied to main girders, stay cables and towers in the out-of-plane
direction, according to the new specifications for long-span bridges including the aforesaid
Tatara Bridge in Japan [10]. Actual values of the design wind loads per unit length with respect
to the member-axis are listed in Table 3.1.
First, by using the basic design model in which main girders were elastically supported by a
provisional support positioned away from the central line of each tower, the authors examined
the optimal position of elastic supports and estimated the overall effects of the countermeasure.
In the examination, positions of the provisional elastic support were horizontally moved along
the bridge-axis, and a spring constant of 2,000tonf/m was set temporarily.
As part of the analytical results, Fig.3.3 shows the relationship between the amount of
eccentricity at each elastic support position from the central line of each tower and the following
factors: out-of-plane bending moments at the elastic support position and out-of-plane lateral
displacements at the midpoint of main girders. In the figure, the analytical result for the case
when main girders were fixed and supported by wind-shoes at the center of towers is also
described for reference.
The results in Fig.3.3 indicate that, as a result of applying this method, the absolute value of the
minimum out-of-plane bending moment and the maximum out-of-plane lateral displacement are
significantly decreased, and that the countermeasure is most effective when the position of
elastic supports is far from the central line of each tower and close to the center of bridge.
Though the analytical results are not shown, the authors compared the effects of elastic supports
at multiple positions near each tower; however, the countermeasure was most effective when
there was only one support near each tower.
Member
Wind load
Girder
1.208 (tonf/m)
Tower
12.094 (tonf/m/tower)
Cable 0.0516 - 0.0816 (tonf/m/cable)
Tower link
Table 3.1 Design wind loads

Tower side
an
e sp
Sid
Elastic supports
s
ter
Cen
pan
Girder side
Fig.3.1 Concept of elastically supported girder Fig.3.2 An example of the structure of tower-links
Out-of-plane lateral displacement

at midpoint of center span
(by wind-shoe)
-150000
15
Central line of tower
Side span
Center span
10
5
Out-of-plane bending moment at elastic suppot position

Out-of-plane lateral displacement at midpoint of center span
-50000
0
-80
-60
-40
-20
20
40
60
Amount of eccentricity from central line of tower (m)
Fig.3.3 Examination into the optimal position
-400000
Out-of-plane bending moment at elastic support position
30
With tower-link
Without tower-link
Out-of-plane bending moment (tonf . m)
20
-350000
-250000
25
Out-of-plane lateral displacement (m)
Out-of-plane bending moment at fixed support position

(by wind-shoe)
Out-of-plane bending moment (tonf. m)
-450000
-300000

With tower-link
at elastic support position
20
Without tower-link
-200000
10
-100000

at midpoint of center span
With tower-link
Without tower-link
0
1.0E+02
80
0
1.0E+03
1.0E+04
1.0E+05
Spring constant (tonf/m)
1.0E+06
Fig.3.4 Examination into the optimal spring constant
Next, on the basis of these results, by using the alternative model in which main girders were
elastically supported by a one-sided pillar of each 3-D A-type tower close to the center of
bridge, the authors examined the optimal spring constant of elastic supports; namely, the amount
of eccentricity at each elastic support position from the central line of each tower is 60m. Then,
by using the alternative model in which tower-links were temporarily eliminated and main
girders were supported directly by a cross beam of each tower in the vertical direction, the
authors compared the effects of elastic supports to examine the influence of out-of-plane lateral
displacements of tower-links.
As part of the analytical results, Fig.3.4 shows the relationship between the spring constant for
elastic supports and the following factors: out-of-plane bending moments, out-of-plane lateral
displacements at the elastic support position and out-of-plane lateral displacements at the
midpoint of main girders for the cases with and without tower-links. In the figure, the analytical
result for the case when main girders were fixed and supported by wind-shoes at the center of
towers in the basic design model is also described for reference.
The results in Fig.3.4 indicate that the absolute value of the minimum out-of-plane bending
moment decreases due to the decreased spring constant; however, when the spring constant
decreases to below 1,000tonf/m, the out-of-plane displacement at the elastic support position
increases and the influence of recovery forces due to the inclination of tower-links becomes
apparent. Accordingly, to avoid excessive charges to tower links, a spring constant of
1,000tonf/m is considered to be optimal in this numerical experiment.
3.3
Effect of Structural countermeasure on Economy
Figs.3.5 and 3.6 show the distributions of out-of-plane bending moments and out-of-plane
lateral displacements of main girders due to the design wind loads, respectively, for both the
analytical models in which fixed supports of wind-shoes or elastic supports with the spring
constant of 1,000tonf/m were applied. Both kinds of supports were positioned at the center of
Alternative model
by windshoe
by elastic support
0
150000
300000
Elastic support position
-150000
Bridge-axis
10
Elastic support position
-300000
20
by windshoe
by elastic support
Basic model
Out-of-plane bending moment (tonf . m)
-450000
Bridge-axis
Basic model
0
-5
by windshoe
by elastic support
Alternative model
by windshoe
by elastic support
Fig.3.5 Out-of-plane bending moments of main girders Fig.3.6 Out-of-plane lateral displacement of main girders
each tower for the basic design model, and on the one-sided pillar of each tower close to the
center of bridge for the alternative design model.
The results in Fig.3.5 indicate that some effects from the method in which wind-shoes are only
replaced by elastic supports, or from the method in which wind-shoes are only moved to a
position away from the central line of each tower, can be expected, and that the absolute value
of the minimum out-of-plane bending moment can be significantly reduced by combining these
methods. In Fig.3.6, we can clearly see that excessive out-of-plane lateral displacements are not
generated around the elastic supports, and that the maximum out-of-plane lateral displacement at
the midpoint is the largest in the basic design model in which main girders are supported
elastically.
Based on these results, the structural countermeasure of elastically supporting main girders in
the out-of-plane direction at an optimal position away from the central line of each tower is very
effective for reducing the out-of-plane bending moments in the region of main girders, where
stresses due to the design wind loads become dominant. Therefore, since the additional
construction costs due to adopting 3-D A-type towers are not very high, the countermeasure is
very useful for increasing the cost efficiency. Moreover, the 3-D A-type towers are expected to
increase the in-plane rigidity of cable-stayed bridges and decrease their construction cost.
4. Improvement in Wind-Resistant Design

4.1
Countermeasure against Aerodynamic Instability
As a structural countermeasure against unstable static and dynamic phenomena in the windresistant design, the authors devised a method for controlling the flexibility of stay cables caused
by wind actions in the out-of-plane direction. This proposed method aims at improving the
wind-resistant stability by installing an auxiliary cable system in which two stay cables in each
step are woven together in a twilled-weave manner, as shown in Figs.4.1 (a) and (b) for the
basic and alternative design models, respectively. These auxiliary cables, named lacing cables,
are installed to avoid the loss of torsional resistance when two stay cables approach each other
due to the application of aerodynamic forces.
The cross-sectional area of all lacing cables was set at 0.001 m2 in both the analytical models.
Because it was previously confirmed that, by installing the lacing cables, the deformation
characteristics under live loads and wind loads only negligibly change in the static design,
except that torsional deformations under eccentric live loads are slightly decreased. Moreover,
in this case, since the diameter is sufficiently small, the landscape wouldnt be greatly denuded
of its picturesqueness, and the drivers view wouldnt be seriously troubled due to the
installation of lacing cables. In the modeling of lacing cables, dead loads were neglected, and
their initial stresses when aerodynamic forces are not applied were also neglected; however, it
was assumed that they can resist not only the tension, but also the compression.
CD,CL
5.0
CM
0.25
4.0
0.20
3.0
0.15
2.0
0.10
1.0
0.05
0.0
0.00
-1.0
-0.05
-2.0
-0.10
CD
CL
CM
-3.0
-4.0
-5.0
-15
(a) Basic model with lacing cables (b) Alternative model with lacing cables
Fig.4.1 Image views of both models with lacing cables
-10
-5
10
Angle of attack (deg.)
-0.15
-0.20
-0.25
15
Fig.4.2 Static aerodynamic coefficients
4.2
Lateral-Torsional Buckling Stability
4.2.1 Design Model Used and Static Aerodynamic Forces

To investigate the effects of controlling the out-of-plane flexibility of stay cables on the lateraltorsional buckling stability, the basic design model was used, in which the lacing cables were
installed in steps from the top to the 16th stay cable. Prior to detailed investigations, the authors
had developed some analytical models with various numbers of steps installed, and examined
the difference in the effects of lacing-cables depending on the number of steps; as a result, it was
found that 16 steps are the most effective.
By expressing the drag force, lift force and pitching moment coefficients as CD, CL and CM
dependent on the angle of attack , respectively, three components of the static aerodynamic
forces per unit length of main girders can be obtained with respect to the wind-axis as follows:
D( ) = 1 2 U Z2 An C D ( )
L( ) = 1 2 U Z2 BC L ( )
M ( ) = 1 2 U Z2 B 2 C M ( )
(1)
Z 7
U Z = U 10
10
= 0 + x
(2)
(3)
where D, L and M are the drag force, the lift force and the pitching moment dependent on ,
respectively; is the air density (=0.000125tonf.sec2/m4); An and B are the vertical projected
area and the width, respectively. Then the mean wind velocity UZ at the mean height Z
corresponds to the design wind velocity, which can be given by eq. (2) expressing U10 as the
basic wind velocity at the height of 10 m.
In this analysis, the above-mentioned static aerodynamic forces were applied to the main
girders, in which An, B and Z were set to 5m2, 30m and 40m, respectively, and CD, CL and CM
obtained from a wind-tunnel test were employed as shown in Fig.4.2. Also, the drag forces
independent of were applied to stay cables and towers, in which the drag force coefficients
were set to 0.70 and 1.20, respectively. Since the relative evaluation of the values of lateraltorsional displacements and critical wind velocities against buckling instability was the first
objective, the authors set the wind-axis in the horizontal direction and the angle of incidence 0
to be 0 degrees. Therefore, the angle of attack given by eq. (3) was identical to the torsional
displacement x of main girders.
4.2.2 Analytical Results and Effect of Structural Countermeasure
In the lateral-torsional buckling analyses, the set values of the basic wind velocity U10 were
increased gradually; the numerical calculations were performed until the angle of attack ,
namely the torsional displacement x of main girders, could not converge [11].
6
Torsinal displacement (deg.)
With lacing cables
(Straight cable model)
99m/s
80m/s
Without lacing cables
86m/s
3
2
1
0
50
60
70
80
90 U10(m/s)100
Fig.4.3 Torsional displacements

at the midpoint of man girders
As part of the analytical results, Figs.4.3 and 4.4 show the relationships of torsional
displacements and out-of-plane lateral displacements at the midpoint of main girders to the basic
60
86m/s 99m/s
50
40
80m/s
30
20

With lacing cables
10
0
50
60
70
80
90
100
U10(m/s)
Fig.4.4 Out-of-plane lateral displacements

at the midpoint of man girders
Torsional displacement (deg.)
wind velocity U10, respectively. Fig.4.5 shows

the distribution of torsional displacements of
main girders when the basic wind velocity U10
is 80m/s. In these figures, the analytical result
for the special analytical model, named the
straight cable model in which each stay cable
was modelled as one straight axial-force
member with the purpose of neglecting the
effects of flexibility, are also described for
reference.

With lacing cables
2
Bridge-axis
0
The results in Figs.4.3 and 4.4 indicate that, as
a result of installing lacing cables, the out-of- Fig.4.5 Torsional displacements of man girder
plane lateral displacements of main girders
increase a little; however, the torsional displacements are fairy
reduced due to the control of flexibility of stay cables.
Accompanying this phenomenon, in these figures, the critical wind
velocity against buckling instability increases from 80m/s to 86m/s. In
Fig.4.5, we can clearly see that, at the same wind velocity, the
torsional displacements of main girders are greatly reduced by
installing lacing cables, and their maximum value is even lower than
that of the straight cable model in which the flexibility of stay cables
was neglected.
The reason for the difference in torsional displacements of main

girders with and without lacing cables can be explained using Fig.4.6,
which is a projection figure of the lateral-torsional deformations on
the vertical plane at the center of bridge when the basic wind velocity
U10 is 80m/s (the same as that in Fig.4.5). Based on the results in
Fig.4.6, when lacing cables are not installed, the torsional resistance is
With lacing cables
lost because two stay cables approach each other in each step, as
predicted; in addition, main girders are pulled in the direction of the
Fig.4.6 Lateral-torsional
cable-axis.
Deformations
Accordingly, the structural countermeasure of controlling the flexibility of stay cables in the outof-plane direction is very useful for improving the lateral-torsional buckling stability. The
critical wind velocity obtained in this numerical experiment is sufficiently high, even when
lacing cables are not installed. However, in cases where the lateral-torsional rigidity becomes
weaker due to smaller width-to-span ratios or shallower depths of main girders, critical wind
velocities might be lower than the allowable limit [12]. In such cases, the proposed
countermeasure is extremely effective without greatly degrading the cost efficiency because of
slight design changes. Furthermore, this method can be applied as a temporary countermeasure
during construction and as a supplementary measure after completion.
4.3
Coupled Flutter Behavior
4.3.1 Design Models Used and Unsteady Aerodynamic Forces

To investigate the effects of controlling the out-of-plane flexibility of stay cables on the coupled
flutter behavior, not only the basic design model, but also the alternative design model was used,
in which lacing cables are installed in steps from the top to the 16th stay cable. The position of
elastic out-of-plane supports and the spring constant in the alternative design model were the
same as those evaluated in the previous chapter.
As the dynamic aerodynamic forces in this analysis, the lift force and the pitching moment,
which are the unsteady aerodynamic forces based on the plate-wing theory, were applied to main
girders. Also, the drag force based on the semisteady theory was applied to main girders. For
the structural damping, a logarithmic decrement ST was set at 0.02.
4.3.2 Analytical Results and Effects of Structural Countermeasures

In the analysis, the authors first obtained natural frequencies and natural vibration modes up to
50 dimensions by means of natural vibration analyses; then, by using these results, multi-mode
flutter analyses were carried out on the basis of a modal analysis technique [13].
As part of the analytical results, Fig.4.7 shows the relationship between the wind velocity U at
the mean height of main girders and the aerodynamic damping (logarithmic decrement) of
flutter mode branches. Table 4.1 lists the values of critical wind velocities against coupled
flutter evaluated from the U- curves. In this table, the values of natural frequencies of the 1-st
symmetric and antisymmetric vertical deflection and torsion modes are also listed for reference.
The results in Fig.4.7 and Table 4.1 indicate that, due to the control of out-of-plane flexibility of
stay cables as a result of installing lacing cables, the flutter critical wind velocity of the basic
design model increases by about 23% from 140m/s to 172m/s, and that of the alternative design
model with 3-D A-type towers and elastic supports increases by about 16% from 137m/s to
159m/s. In Table 4.1, we can clearly see that, due to this control, frequencies of the 1-st
symmetric and antisymmetric torsion modes increase beyond the values, which were expected
from the decrement in static torsional deformations under eccentric live loads when the crosssectional area of lacing cables was evaluated.
Logarithmic decrement
Therefore, the structural countermeasure of

0.4
Basic model without lacing cables
controlling the out-of-plane flexibility of stay
Basic model with lacing cables
Alternative model without lacing cables
cables is extremely effective for improving the
Alternative model with lacing cables
dynamic wind-resistant stability, and can obtain
a significantly high flutter critical wind
0.2
velocity, regardless of the application of elastic
out-of-plane supports between main girders and
3-D A-type towers. The critical wind velocity
obtained in this numerical experiment is fairly
0
high
even
without
any
structural
0
100
countermeasures; however, it is evident that the
Wind velocity (m/s)
two proposed structural countermeasures can -0.1
indeed increase the safety level against Fig.4.7 U- curve
dynamic instability due to wind actions.
1-st symm. deflection mode

Freq. 1-st antisymm. deflection mode
(Hz) 1-st symm. torsion mode
1-st antisymm. torsion mode
Flutter critical wind velocity (m/s)
200
Analytical model
With lacing cables
Basic model Alternative model Basic model Alternative model
0.1040
0.1309
0.1034
0.1321
0.1102
0.1511
0.1113
0.1522
0.4886
0.4015
0.6321
0.5122
0.8800
0.6058
0.9223
0.6702
139.9
136.8
172.4
159.0
Table 4.1 Natural vibration frequencies and critical wind velocities against flutter
5. Conclusions
By using a trial-design bridge with a center span of 1,500m, in this study, the authors
investigated the effects of the proposed structural countermeasures on static and dynamic
actions due to wind loads, and discussed the cost effectiveness and the wind-resistant stability of
very long-span cable-stayed bridges with small width-to-span ratios of main girders. The
authors reached the following conclusions based on the analytical results obtained.
(1) In the static design of cable-stayed bridges, due to the decreased width-to-span ratios of main
girders accompanying the increase in their span length, the stress resultant under design wind
loads becomes dominant in a wide region of main girders near each tower, and greatly
influences the cost effectiveness. At this time, the proposed structural countermeasure of
applying elastic out-of-plane supports between main girders and 3-D A-type towers is very
useful for reducing out-of-plane bending moments of main girders generated by design wind
loads, and therefore for increasing the cost efficiency.
(2) The countermeasure of applying elastic out-of-plane supports between main girders and 3-D
A-type towers increases its effectiveness, if the supports are positioned close to the center of
bridge and as far as possible from the central line of each tower. In addition, an optimal
spring constant can be evaluated according to not only the degree of stress-resultant
reduction, but also the allowable amount of out-of-plane lateral displacement at the elastic
support position of main girders under design wind loads.
(3) The structural countermeasure of controlling the flexibility of stay cables in the out-of-plane
direction is very useful for improving the lateral-torsional buckling stability. For cases in
which excessive lateral-torsional displacements occur or critical wind velocities become
lower than the allowable limit due to smaller width-to-span ratios of main girders, the
proposed countermeasure is extremely effective without greatly degrading the cost efficiency
because of slight design changes. Furthermore, this method can be applied as a temporary
countermeasure during construction and as a supplementary measure after completion.
(4) The critical wind velocities against coupled flutter in cable-stayed bridges are extremely
high; they rarely decrease to below the allowable limit, even when their center span length
approach the critical value. However, adopting the structural countermeasure of controlling
the out-of-plane flexibility of stay cables leads to a higher flutter critical wind velocity,
regardless of the application of elastic out-of-plane supports between main girders and 3-D
A-type towers; thus it ensures the safety against dynamic instability under strong winds.
In this study, a 3-D A-type tower was used, the one-sided pillar of which supports main girders
elastically; however, other structures are also possible, in which ordinary plane A-type towers
are used and main girders are supported elastically at the edge of extended concrete piers. In
addition, as a method for controlling the out-of-plane flexibility of stay cables, an auxiliary
cable system was used, in which two stay cables in each step are coupled by lacing cables;
however, other methods are also possible. Furthermore, to realize an era of very long-span
cable-stayed bridges with 1,000m- 1,500m center spans at an earlier stage, it will be necessary
to device more appropriate structural systems and to investigate the efficiency of their
construction and maintenance.
References
[1] J.M.Muller Very Long Span Bridges - Concepts, Materials and Methods, Proc. of IABSE Symposium on
Long-Span and High-Rise Structures, Kobe, 1998.
[2] M.Nagai Possibility and Limitations of Long-Span Cable-Stayed Bridges Based on Static and Dynamic
Instability analyses, Proc. of International Seminar on Long Span Bridge Aerodynamics Perspective '98, Kobe,
1998.
[3] K.Nomura, S.Nakazaki, N.Narita, K.Maeda and H.Nakamura Structural Characteristics and Economy of
Cable-Supported Bridges with Long-Span, Journal of Structural Engineering, Vol.41A, 1995. (in Japanese)
[4] H.Nakamura, K.Maeda, M.Konno, M.Hayashi and N.Narita Lateral-Torsional Buckling Stability of a Long
Span Cable-Stayed Bridge with Flexible Cables under Wind Action, Journal of Constructional Steel, Vo1.5,
1997. (in Japanese)
[5] Honshu-Shikoku Bridge Authority Specifications on Superstructure Design, 1989, and Specifications on
Wind-Resistant Design, 1976. (in Japanese)
[6] Japan Road Association Specifications for Highway Bridges, Part 2 Steel Bridges (English Edition), 1987.
[7] K.Maeda, M.Hayashi, H.Setouchi, H.Nakamura and N.Narita Three-Dimensional Finite Displacement
Analysis of Long-Span Cable-Stayed Bridges, Journal of Structural Engineering, Vol.41A, 1995. (in Japanese)
[8] Y.Maeda and M.Hayashi Finite Displacement Analysis of Space Framed Structures, Journal of Structural
Mechanics and Earthquake Engineering, No.253, 1976. (in Japanese)
[9] P.Jetteur, S.Cescotto and V.Degoyet Improved Nonlinear Finite Elements for Oriented Bodies Using an
Extension of Marguerre's Theory, Computers & Structures, Vo1.17, No.1, 1983.
[10] Honshu-Shikoku Bridge Authority Specifications on Wind-Resistant Design of the Onomichi-Imabari Route,
1994. (in Japanese)
[11] V.Boonyapinyo, H.Yamada and T.Miyata Nonlinear Buckling Instability Analysis of Long-Span Cable-Stayed
Bridges under Displacement-Dependent Wind Load, Journal of Structural Engineering, Vol.39A, 1993.
[12] M.Nagai, Xu Xie, H.Yamaguchi and Y.Fujino Static and Dynamic Instability Analyses of 1400-meter LongSpan Cable-Stayed Bridges, Proc. of IABSE Symposium on Long-Span and High-Rise Structures, Kobe, 1998.
[13] M.Iwamoto Prediction of Aerodynamic Behavior of Cable Supported Bridges, Ph.D.thesis, University of
Tokyo, 1995. (in Japanese)
Aerodynamic Performance of Cable-Supported Bridges with Large

Span-to-Width Ratios
Sren V. LARSEN.
M.Sc., PhD.
Danish Maritime Institute, Lyngby,
Denmark
Sren V. Larsen, born 1966,

obtained his degree in 1991,
and his Ph.D. in 1997.
He joined DMI in 1995.
Project Manager, Hydro- and
Aerodynamics
Summary
This paper presents some results of an experimental study of three different types of cable
supported bridges. A traditional suspension bridge, a traditional cable-stayed bridge and a new
concept of a cable-stayed bridge with four inclined cable planes have been studied with the
emphasis on the aerodynamic behaviour. All bridges have an identical stiffening girder, which is
extremely slender compared with the main span (span-to-width ratio of 100). The paper focuses on
the results of full bridge aeroelastic model tests performed at a scale of 1:100.
1. Introduction
After the collapse in 1940 of the first Tacoma Narrows Bridge, it became evident for the modern
bridge designer that the dynamic effects of the wind are of immense importance. Bridges with
tendencies of significant wind-induced movements were modified in the following years. Such
modifications included the Golden Gate and the Bronx-Whitestone bridges.
In recent years, the spans of erected and/or proposed bridges have become longer and longer.
Especially for cable-stayed bridges, an increase of the span over the last 40 years is truly
remarkable, from 183 m in 1955 (Strmsund) to a record-holding span of 890 m expected in 1999
(Tatara, Japan). For suspension bridges the increase in span has been more moderate in this period
of time from around 1300 m (Golden Gate and Mackinac) to 1624 m (Storeblt, Denmark) and
1991 m (Akashi Kaikyo, Japan) in 1998. Though, studies have been made for spans of 3.3 km
(Messina) and even up to 5 km (Gibraltar).
A measure of the slenderness in the lateral direction may be defined : the span-to-width ratio is the
ratio between the length of the main span and the width of the bridge girder (or deck); this ratio
accounts for the lateral slenderness of a bridge system. The bending moment of inertia of the deck
is possibly a better parameter to compare with the span length. However, such detailed data as
lateral bending stiffness is not readily available for many bridges. Figure 1.1 illustrates the
development in the span-to-width ratio for suspension bridges.
100
90
80
Tacoma 1
span-to-width ratio l/B
70
60
Mackinac
50
40
30
20
Brooklyn
Humber
Hoga Kusten Akashi Kaikyo
Askoey
Storebaelt
Golden GateChesapeakeTacoma Tagus

2
VerrazanoSevern
Bosporos 1
Fyksesund
Minami-Bisan
Francisco- Tancarville
George WashingtonSanOakland
Kita-Bisan
Detroit River
Emmerich
BronxBosporos 2
Forth
Whitestone
Williamsburg
Manhatten
Lillebaelt
10
0
1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020
year of completion (or expected year of completion)
Figure 1.1 Development in lateral slenderness (span-to-width ratios) for suspension bridges.
As seen from Figure 1.1, the suspension bridges were becoming increasingly slender up to the
Tacoma Narrows failure, after which bridges were built with more moderate lateral slenderness.
However, the lateral slenderness was not decreased to the same extent as the vertical slenderness
was. The first Tacoma Narrows Bridge had a span-to-width ratio of 72. Only 17 years later the
Mackinac Bridge was completed having a span-to-width ratio of 56. The lateral slenderness of the
Mackinac Bridge is similar to that of newly completed long span bridges: the Storeblt and the
Akashi Kaikyo bridges
For many years, cable-stayed bridges have been somewhat "short and wide" compared to
suspension bridges. Cable-stayed bridges were often seen having span-to-width ratios five times
less than that of the most slender suspension bridges. The development of the span-to-width ratio
for cable-stayed bridges is illustrated in Figure 1.2. For cable-stayed bridges the most critical stage
with respect to lateral slenderness is during erection assuming that free-cantilevering method is
employed. During erection the free cantilever is prone to lateral deflections. For mono-tower cablestayed bridges during erection, the maximum length of the free cantilever is equal to the length of
the main span, whereas for cable-stayed bridges with two towers, the free cantilever is only a half of
the main span. Therefore, in Figure 1.2 mono-tower cable-stayed bridges are shown with both a
"span-to-width ratio" and a "double span-to-width ratio", where the latter exemplifies the erection
stage.
60
Karnali River (erec.)
50
span-to-width ratio l/b
Tatara
Skarnsundet
Normandie
40
Helgelands
Knie (erec.)
St. Nazaire
30
20
10
0
1950
Karnali River
Tjorn
Wadi Kuf
Quincy Ikuchi
Tampico
Kohlbrand
Barrios de LunaBaytown Oresund
Brotonne Rande
Annacis
LulingFaro
Maracaibo ReesKnie
Rama IX
2nd Severn
Bridge
Stromsund
PascoDuisburg- KennewickDames Sunshine Skyway
Severin
Neuenkamp
Point
Erskine
Theodor Heuss
Friedrich
Ebert
Nord-Elbe
Buchenauer
Severin (erec.)
1960
1970
1980
1990
year of completion (or expected year of completion)
2000
2010
Figure 1.2 Development in lateral slenderness (span-to-width ratios) for cable-stayed bridges.
2. Scope and Extent of Study

When the span-to-width ratio becomes larger and the bridge consequently becomes more slender in
the lateral direction, the demand for additional lateral support becomes more pronounced.
As described in [1], three mutually inclined cable planes are required to have a three-dimensional
support of the girder, i.e., support against vertical, lateral and torsional loads. This has so far only
been used in pipeline bridges and minor pedestrian bridges. Such cable systems consisting of
mutually inclined cable planes are referred to as spatial cable-stayed bridges or spatial systems.
Three cable planes are, as mentioned, sufficient to establish a spatial system, but based on
considerations of symmetry, four cable planes seem reasonable. A spatial cable-stayed system can
be obtained with various stay cable configurations and the pylons of a spatial system can be
designed in many ways, see [3] and [4] for extensive discussions.
The objective of the present study is to investigate bridges with extremely narrow girders with
respect to the aerodynamic behaviour. An important part of the study is the examination of a new
improved cable-stay system. For comparison, two traditional cable supporting systems are included
in the study. The primary study of the three bridges is performed in an extensive three-dimensional
aeroelastic wind-tunnel study with full bridge models at a relatively large scale.
The project is concerned with bridges having a span-to-width ratio of 100. This is very close to a
factor of 2 to what has been seen in suspension bridges, and 2.5 to the extreme in cable-stayed
bridges. The study comprises three cable supported bridges with an assumed full-scale dimensions
of 800 m main span and 250 m side spans. The three bridges concerned are:
Suspension bridge
Traditional modified fan ("plane") cable-stayed bridge
Cable-stayed system with four inclined cable planes ("spatial")
Traditional cable supported bridges, i.e., suspension bridges and cable-stayed bridges with one or
two cable planes, can give only vertical or vertical and torsional support to the bridge deck.
Additionally, as a secondary effect, these systems can also give some restraint against lateral loads
if the cable system is earth-anchored. So far for major cable supported bridges, only suspension
bridges are built as earth-anchored systems.
The suspension bridge in the present study is an earth-anchored system, and the two cable-stayed
systems are self-anchored. The two cable-stayed bridges will be referred to as plane and spatial
systems, respectively. The three bridges are illustrated in Figure 2.1.
Figure 2.1
Illustrations of the three cable-supported bridges studied. Upper: suspension

bridge, middle: plane (or traditional) cable-stayed bridge, lower: spatial cablestayed bridge.
For the prototype bridge girder a simplified deck was assumed with a constant cross-section
equivalent to a 2N8 m steel box girder, with 16 mm side and bottom flanges and a 20 mm upper
flange. The chosen cross-section was intended to represent an idealized deck, hence mass and
stiffness contributions from diaphragms, etc., was assumed to be included in the flanges. The bridge
deck is studied in two cases: as a rectangular section (bluff) and as a streamlined section, see Figure
2.2. The streamlined section represents a section being aerodynamically preferable to the bluff, but
not necessarily an optimized cross-section. The purpose of including two cross-sections in the study
was to examine two deck types with significantly different aerodynamic characteristics
Figure 2.2
Illustrations of the two bridge deck configurations. Upper: bluff box girder, lower:
streamlined box girder.
3. Section Model Tests

Initially, the two cross sections were studied in an extensive section model test programme. The
load coefficients were measured for 11 angles of attack of the incoming flow: 10 in steps of 2. In
Table 3.1, the results for 0 angle of attack are listed.
Section
flow
CD
CL
CM
dCL/d
dCM/d
Table 3.1
Streamlined girder
smooth
turbulent
0.20
0.20
-0.28
-0.30
-0.054
-0.078
7.59 rad-1
7.40 rad-1
2.83 rad-1
2.71 rad-1
Bluff girder
smooth
turbulent
0.36
0.36
-0.25
-0.26
-0.010
~0
3.10 rad-1
5.45 rad-1
-1.01 rad-1
-0.45 rad-1
Static coefficients and slopes, streamlined bluff section. CD = Fdrag/( 1U2Bl), CL =

Flift/( 2U2Bl) and CM = M/( 3U2B2l), where 4 is the air density, U is wind
speed, B is a characteristic width of the model (8 m full scale used for both cross
sections) and l is the length of the section model.
4. Full Bridge Models

The three aeroelastic full bridge models were produced at a geometric scale of 1:100. The two
cable-stayed models could be modified into modelling a critical phase of the erection stage. The
models were designed using Froude scaling.
The models were tested in two flow conditions, namely smooth and turbulent flow. Turbulence
intensity in the turbulent flow was about 13% for the u-component at deck level.
The bridge deck was produced with an aluminium spine covered with deck segments of a light
foam material. Furthermore, a standard model railing was mounted onto the deck segments. The
aluminium spine simulates the structural properties of the deck and the deck segments and the
railings simulate the cross-sectional geometry. The deck segments were attached to the spine with
small aluminium crossbeams; every second crossbeam was attached to the hangers or stay cables.
The bridge deck could be modified from bluff to streamlined cross-section. The sketches in Figure
4.1 show the cross-sections of the bridge deck.
Figure 4.1
Cross-section of bridge deck in full bridge model tests. The upper sketch shows the
bluff section and the lower the streamlined section.
Figures 4.2 through 4.4 shows photographs of the full bridge models installed in DMIs very wide
boundary layer wind-tunnel.
Figure 4.2 Suspension Bridge Model
Figure 4.5 Spatial Cable-Stayed Bridge Model

Figure 4.4 Plane Cable-Stayed Bridge Model
5. Results of Full Bridge Models

It was possible only in two cases to reach the stability limit of the individual systems with the
various configurations in turbulent flow. Instability developed for the suspension bridge and the
plane system, where the wind speed was increased until the response was found unacceptable. The
instability occurred only with the bluff cross-section of the girder. For the spatial system with bluff
cross-section, the maximum wind speed of the wind tunnel is believed to be close to the stability
limit.
Table 5.1
Bridge
Configuration
Stability limit
Suspension bridge
Plane cable-stayed bridge
bluff girder
bluff girder
42 m/s
58 m/s
Determined stability limits (turbulent flow) in full-scale.
In the case of the suspension bridge with bluff cross-section, the instability started to develop at
approximately 42 m/s in turbulent flow. The instability occurring was in torsion.. Tests performed
in smooth flow showed that the stability limits were not different from those measured in turbulent
flow. For the plane system mounted with the bluff deck, the instability started to develop around 58
m/s in turbulent flow. Similar to the suspension bridge, the instability has a torsional nature.
All systems, where the streamlined section was employed, did not exhibit instability within the
range of wind speeds of the tests. Also the spatial system with bluff section was found stable,
however, as mentioned, it appeared close to instability. In the case of the erection stage of the plane
system, the tests were stopped when the lateral deflections were found unacceptable, therefore
aerodynamic instabilities at high wind speeds could not be studied in these cases.
The buffeting response was measured in the turbulent flow condition for all bridges in all
configurations. These measurements give information about the quasi-static mean and dynamic
response in the simulated flow. Figures 5.1 through 5.8 show the buffeting response measured at
RMS vertical displacement (m)
vertical buffeting response - streamlined cross section

0.80
suspension
plane
0.70
spatial
0.60
0.50
0.40
0.30
0.20
0.10
0
10
20
30
40
50
60
mean wind speed at deck level (m/s)
70
RMS lateral displacement (m)
Figure 5. 1
lateral buffeting response - streamlined cross section
0.50
suspension
0.45
plane
0.40
spatial
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0
10
20
30
40
50
60
70
Figure 5. 2
0.40
0.30
0.20
0.10
10
Figure 5. 3
2.00
1.50
1.00
0.50
0.00
0
10
20
30
40
50
60
70
mean lateral buffeting response - streamlined cross section

suspension
12.0
plane
spatial
10.0
8.0
6.0
4.0
2.0
0.0
RMS lateral displacement (m)
RMS vertical displacement (m)
0.50
2.50
10
20
30
40
50
60
70
Figure 5. 5
vertical buffeting response - bluff cross section

0.80
suspension
plane
0.70
spatial
0.60
0.00
torsional buffeting response - streamlined cross section

4.00
suspension
3.50
plane
spatial
3.00
Figure 5. 4
mean lateral displacement (m)
0.00
RMS torsional displacement (degrees)
the centre of the span in the completed bridges. However, it should be noticed that the bridge deck
with the bluff section was around 10% lighter than with the streamlined section. The mass moment
of inertia of the two cross-sections varied more than 30%.
20
30
40
50
60
70
lateral buffeting response - bluff cross section

0.50
suspension
0.45
plane
0.40 spatial
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0
10
20
30
40
50
60
70
Figure 5. 6
mean lateral buffeting response - bluff cross section

mean lateral displacement (m)
RMS torsional displacement (degrees)
torsional buffeting response - bluff cross section

4.00
suspension
3.50
plane
3.00 spatial
2.50
2.00
1.50
1.00
0.50
0.00
0
10
20
30
40
50
60
Figure 5. 7
70
12.0suspension
plane
10.0 spatial
8.0
6.0
4.0
2.0
0.0
0
10
20
30
40
50
60
Figure 5. 8
The two cable-stayed systems with streamlined deck exhibit very similar behaviour in dynamic
vertical response, whereas the suspension bridge with streamlined deck exhibits a larger response;
at 40 m/s the response is approximately 40% larger than that for the cable-stayed systems. The
dynamic lateral response of the suspension bridge and the plane system with streamlined deck are
both larger and more scattered than that of the spatial system. It is also found that the three systems
with streamlined deck all exhibit relatively small dynamic responses in torsion.
6. Conclusions
Laterally slender bridges seem to be an area which until now has not had very much attention.
Therefore, it was found valuable to carry out a comprehensive study of an important topic related to
such bridge structures: the aerodynamic characteristics of cable supported bridges with narrow
bridge decks.
The study revealed that the proposed spatial cable supported system has a distinct advantage
compared to the traditional cable systems in terms of its aerodynamic resistance. This was
confirmed in the following aspects. Firstly, a spatial system exhibits a smaller dynamic response
than a suspension bridge. Secondly, the lateral deflection, which can be critical for the design of
extremely slender bridges, is reduced by approximately a factor of 2. Thirdly, during the critical
erection stage, the spatial system is stable at least up to the design wind speed of the completed
bridge, even with an extremely slender girder, which will cause large problems during erection for a
traditional cable-stayed system. Finally, for the spatial system, the design of the girder cross-section
becomes less important because even with a bluff cross-section, the system is still aerodynamically
stable, whereas the suspension bridge had only an acceptable stability limit with the streamlined
deck. Therefore, the spatial system can be built with a less material-consuming girder than the
suspension bridge. Assuming similar flange thicknesses as for the bluff section, a streamlined
section will have a cross-sectional area about 30% higher than that of the bluff section.
Consequently, the spatial system offers the possibility of saving a significant amount of steel
material since a girder with simpler geometry is acceptable.
Generally, it may be concluded that a spatial system in terms of aerodynamics may be very
advantageous, especially with a slender girder which can cause problems for traditional cable
supported systems. It was also found that the observed torsional instabilities were not coupled with
70
the lateral degree of freedom of the deck despite the fact that the bridges were very flexible in this
direction.
7. Acknowledgements
This study has been supported by the Research Academy, Aarhus and the Danish Maritime
Institute. Furthermore, the COWIfoundation has contributed with financial support of the model
construction and installation The work has been performed under the supervision of Professor Niels
J. Gimsing, Dr. Techn. Clas Dyrbye of Technical University of Denmark and Professor, Dr.
Hiroshi Tanaka of University of Ottawa. All are acknowledged for their support. The Danish
Maritime Institute and A/S Storeblt are also recognised for making the Very Wide Wind Tunnel
available for the tests.
8. References
[1]
[2]
[3]
[4]
[5]
[6]
Gimsing, N.J. "Cable Supported Bridges with Spatial Cable Systems" - Bulletin of the
International Association for Shell and Spatial Structures, Vol. 33 n.1., 1992.
Gimsing, N.J.: "Suspended Bridges with Very Long Spans", International Conference on
Cable-Stayed and Suspension Bridges, Deauville, France, Proceedings vol. 1, pp. 489-504,
1994.
Vejrum, T. & Pedersen, A.: "Bridge with Spatial Cable Systems - Theoretical and
Experimental Studies", Proceedings of the IABSE Conference on "Cable Stayed Bridges,
past, present, future, Malm, 1999.
Vejrum, T. : "Bridges with Spatial Cable Systems. Theoretical and experimental studies
with special emphasis on lateral buckling stability of the girder", Ph.D. thesis, Series R, No.
19, Department of Structural Engineering and Materials, Technical University of
Denmark, 1997.
Larsen, S.V. & Gimsing, N.J.: "Static and Dynamic Behaviour of Cable Supported bridges
With Small Span-to-width Ratios", International Conference on Cable-Stayed and
suspension Bridges, Deauville, France, Proceedings vol. 1, pp. 569-576, 1994.
Larsen, S.V.: "Long and Narrow Cable Supported Bridges Subjected to Wind Load", Ph.D.
thesis, Danish Maritime Institute and Department of Structural Engineering, Technical
University of Denmark, 1997.
Cable-stayed Bridge in Bandung, Indonesia

Alan SHARPE
Halcrow Group Ltd.
Senior Bridge Engineer.
London, UK
Andrew J YEOWARD
Halcrow Group Ltd.
Engineering Manager,
Bridges Department.
London, UK
Roger J BUCKBY
Halcrow Group Ltd. Director
Bridge Engineering
London, UK
Summary
The construction of a 2.1km long elevated road to carry traffic along the Pasteur Cikapayang
Surapati corridor in Bandung has been authorised by the Government of the Republic of Indonesia.
The focal point of the scheme is a 400m long cable-stayed bridge crossing over the Cikapundung
valley.
The paper describes the general arrangement of the 106m mainspan cable-stayed bridge and key
features of the design and construction methods adopted for the bridge
Introduction
Bandung is a rapidly expanding city of 3 million inhabitants, located in the West Java province of
Indonesia. Studies carried out by the Directorate General of Highways (Bina Marga), as part of the
governments development plan, indicated that priority should be given to providing a new east-west
elevated arterial road in the northern part of Bandung. Bina Marga appointed Sir William Halcrow
and Partners in association with INCO of Kuwait, INDEC and Associates of Indonesia and LAPIITB to plan and develop a detailed design for the project, and to supervise its construction. Funding
for the design and construction of the scheme is being provided under a loan from the Kuwait
Government.
Project Description
The project consists of:
A western approach viaduct of 1.35km in length, starting at

Junjunan Road in the west, closely following the alignment of
the existing Pasteur Road to the Cikapundung valley. Figure 1
illustrates the precast glued segmental structural form of the
approach viaduct.
A 400metre long elevated bridge across the Cikapundung valley

which includes a cable-stayed bridge with a length of
161metres.
An eastern approach viaduct of 800m following the existing

Cikapayang Road.
The alignment of the scheme is shown in Figure 2.
Figure 1. Western of Approach Viaduct (Pasteur Road).

The approach viaducts which carry two lanes of traffic in each
direction and have typical spans of 44.5 metres between pier
supports, are designed to be erected by the balanced cantilever
method.
Figure 2. Scheme Layout
The cable-stayed bridge over the valley carries three lanes of traffic in each direction and has a
single tower located in the highway median, supporting the main span via a single plane, semiharp
arrangement of stay cables. The tower is anchored to the backspan pier by an arrangement of parallel
stay cables. The superstructure of the bridge is designed to be constructed using precast, prestressed
concrete glued segments, with the main span being erected by cantilevering from the pylon.
Intermediate on / off ramps are provided at various locations to allow local traffic to access the
elevated road. On either side of the valley, a pair of these ramps is provided with the ramp slip roads
forming the third lane in each direction over the cable-stayed bridge.
Cable-stayed Bridge
Span arrangement
The span arrangement for the cable-stayed bridge was finalised after careful consideration of a
number of options. The asymmetrical cable configuration with a single vertical pylon was
considered to provide an aesthetically pleasing structure across the valley, with continuity of
structural form with the approach viaducts. The resulting asymmetric structure has a main span of
106 m and a back span of 55m. Figure 3 shows computer-generated impression of the bridge.
Further details are given in Figure 4.
Figure 3. Cable-stayed bridge.
Figure 4. Elevation and plan of cable-stayed bridge.
Pylon
The height of the pylon was restricted by
the close proximity of Bandung city
airport. This results in the longest cables
on the main span being at an unusually
low angle of approximately 20 degrees to
the horizontal.
The pylon shaft rises 53m above the valley
floor. Its height above the deck is 37m,
which is 0.35 of the main span length. The
anchorage region of the pylon shaft has
been designed such that the cables pass
through the shaft and are anchored on the
opposite face. This has the advantage of
putting the shaft laterally into compression
in the anchorage zone avoids the splitting
forces frequently associated with this area
in of cable-stayed bridge pylons. Figure 5
shows a cross section through the deck at
the pylon location and details the cable
anchorages at the pylon top.
Figure 5. Pylon Details

The pylon shaft is connected monolithically to the superstructure to provide torsional stiffness to the
bridge multi cell box girder deck. Below deck level, the pylon remains solid in section. Grade 50/20
concrete is specified for the pylon over the full length of the shaft.
Superstructure
The superstructure is a three cell precast segmental concrete box girder with an external shape
similar to that of the viaducts (Figure 6). Each longitudinal segment is constructed in two sections
with an in-situ central stitch enabling the same external formwork moulds as the viaduct to be used.
The adoption of an in-situ stitch also facilitates easy construction of the tapered deck sections at each
end of the cable-stayed bridge and allows incorporation of the cable anchorages in-situ. The box
girder span longitudinally between the pylon diaphragm and the adjacent piers and are partially
supported by the cables to enable a shallow depth deck to be maintained. The box is stressed by
internal bonded prestress within the top flange.
Figure 6. Precast segment cross sections

The segments are all 2.95 metres in length to match that those on the approach viaducts, each having
a 500mm deep transverse rib positioned at the centre to support the cantilever slab. The rib also
carries the concealed drainage pipe from the gully pots to the longitudinal drainage system, which is
situated within the box section. The box has an average web thickness of 450mm, top slab of
250mm and bottom slab of 250mm. The bottom slab is increased to 600mm at pier locations to
reduce the compressive stresses due to hogging over the piers.
The inner webs are increased to 1250mm as the box approaches the pylon as 90% of the shear was
attracted to the pylon through these webs, the remainder being carried via the deck diaphragm.
At each pier location special diaphragms segments were required to transmit the forces from the
webs into the bearing plates. In each of these diaphragms access holes are provided to allow
inspection throughout the box structure.
Diaphragms are also incorporated at each cable anchorage to ensure that the support provided by the
cables is transmitted into all four webs. A three dimensional analysis was carried out which
confirmed that slightly more bending was attracted to the inner webs. The prestress was distributed
accordingly. Grade of 50 concrete was required for the main span deck segments to accommodate
the high compressive stresses during construction.
The profile of the cable-stayed bridge segments was carefully developed to match that of the
viaducts. The long cantilever and curved soffit was finally adopted after consideration of many
alterations to produce a section that would reduce the overall visual impact of a very wide
superstructure. This was particularly important given the proximity of the existing residential road
corridor which the viaduct follows.
Stay cables and anchorage positions
The main span cables consist of 9 cables along a central single plane spaced at intervals of
8.85metres to coincide with every third deck segment. The back span cables consist of 10 cables in
pairs on two central planes and are tied directly back into the backspan pier diaphragm. Cables are
made up of galvanized 7-wire strands protected by HDPE sheathing filled with wax after cable
tensioning. The cable sizes are of two types, 73T15 in the backspan and 85T15 in the mainspan.
During detailed design, consideration was given to providing intermediate anchorages in the
backspan but this was found to introduce undesirable hogging moments in that span and increased
the main span sagging moments.
The design brief specified that the structure should be designed for the effects of any cable being
removed with normal loading under the serviceability and ultimate limit states.
The cable anchorages are constructed within the central in-situ stitch. These are tied into the precast
units by a steel frame which takes the cable forces directly by tension into the concrete deck. The
insitu stitch along the centre of the deck allows the incorporation of these steel anchorage frames
when the stitch is made thus negating the need for precast segments with pre-positioned anchorages.
Cable-stayed bridge Construction methods

Pylon and Piers
The pylon and piers are designed to be supported on bored pile foundations with cast-in-situ
climbing formwork for the shafts. At the pylon head, formwork tubes will be cast in under carefully
controlled geometrical conditions to ensure accurate alignment of the cables.
Deck
As there are no intermediate stays in the back span and with the varying width it was not appropriate
to specify construction by the cantilever method. The backspan was therefore designed to be
constructed on falsework supported from the ground and post tensioned with the falsework removed
during cantilevering of the mainspan. The main span is built by erecting the two precast segments
followed by casting of the in-situ stitch.
Temporary prestress is required to hold the boxes in place and permanent prestress is applied only
after the in-situ stitch has reached its target strength. This ensures that the in-situ portion is also
effective for the prestress design.
At the pylon connection to the deck a cast in-situ length of deck will be constructed to tie into the
pylon reinforcement and to create a working platform from which the cantilevered main span boxes
will be erected. The length of the cast in-situ deck was chosen at 9.2m projecting approximately
500mm beyond the pylon shaft.
As the main span cantilever increases
temporary props are specified to
support the proportion of the dead
load that is later taken by the
longitudinal beam action. These props
will be removed as subsequent
segments are placed relieve the load.
The sequence of erection is shown in
Figure 7.
Cantilever construction continues
until it reaches the in-situ stitch joint
is located 13.75 metres from the first
east pier. The final distribution of
loading in the cables and the box
girder is achieved by jacking the main
span onto the protruding cantilever
from the first east pier. This lifts the
deck from the remaining support and
relieves the cable forces, thereby
introducing the desired longitudinal
moment distribution into the box
girder. The jacking forces relieve the
cable tensions and provide the
additional capacity in the cables for
superimposed and live load effects.
Figure 7. Cable-stayed bridge construction

Deck Prestress
During construction of the main span, the deck moments vary from predominant hogging during
cantilevering to a permanent sagging at the deck completion. The short-term hogging moments are
reduced by introducing temporary props during erection. There is also a requirement for temporary
prestress in both the bottom and top flanges to maintain the allowable stress range in the concrete
box during construction.
The permanent prestress comprises 7-wire strand tendons with tendon sizes of 22T15 and 19T15
chosen to match those in the viaduct.
Where possible prestress tendons and anchorages are contained within the top and bottom slab
haunches to minimise the web thickness and weight of the segments.
At the pylon, where the prestress is at its greatest, some of the tendons are located within the webs.
For cable installation and cable replacement the design specifies that stressing takes place from deck
level. Sufficient space has been provided for a full size jack to be attached to the anchorage.
Project Implementation Programme

The detailed design of the elevated viaducts and cable-stayed bridge, including all tender documents
was completed in Bandung by Halcrow Indec - Inco in the period September 1996 August 1997.
Construction is scheduled to start early in 1999, with a 24-month contract period.
New Developments of Erection Control for Prestressed Concrete CableStayed Bridges

L.C. FAN
Tongji University
Shanghai, China
D.W. CHEN
L.G. THAM
Tongji University The University of Hong Kong
Shanghai, China
Hong Kong, China
F.T.K. AU
P.K.K. LEE
The University of Hong Kong The University of Hong Kong
Hong Kong, China
Hong Kong, China
Summary
In the construction of prestressed concrete cable-stayed bridges, the simultaneous control of
geometry and internal forces is one of the most important issues to address. This paper describes
the recent developments in construction control of prestressed concrete cable-stayed bridges in
China. An adaptive control system has been developed utilizing modern engineering cybernetics
theory. The system makes use of the structural parameters identified for the continuous
adjustments of the mobile carriages for insitu cantilever construction of the bridge deck. The
system has been successfully applied to the construction of several long span prestressed
concrete cable-stayed bridges.
1. Introduction
With the rapid development in the construction techniques for prestressed concrete cable-stayed
bridges in the past two decades, China has become one of the countries having the largest stock
of such bridges. From the experience gained so far, the simultaneous control of geometry and
internal forces during the cantilever construction of prestressed concrete cable-stayed bridges has
been recognized as one of the key problems to address. The importance of construction control
has been amply demonstrated by a few mishaps during the construction of a few of such bridges
in recent years.
Very rigorous numerical simulation of the insitu cantilever construction is usually carried out to
estimate the amount of preset in the fixing of the mobile carriage. However in spite of such
efforts, it is almost impossible to eliminate the discrepancies between the theoretical predictions
and the actual structural responses. Should such discrepancies be not corrected in a timely
manner and thus be allowed to accumulate, the geometry and internal forces of the bridge may be
out of control.
The discrepancies between the theoretical predictions and the actual structural responses can be
attributed to the following factors:
1. The assumed structural parameters for the design may be different from the actual values
achieved on site. Such parameters include, but are not limited to, moduli of elasticity of
concrete, steel reinforcement and prestressing tendons, dead loads, construction live loads,
1
shrinkage and creep of concrete, moments of inertia of the segments as well as the
temperature distribution in the bridge deck and tower.
2. The values for the control of geometry and internal forces during cantilever erection are
usually given only at certain important milestones such as the beginning of each segment
construction cycle. However in reality, the structural responses of the bridge are changing
continuously according to the variations of applied loading and environmental conditions.
3. Errors may also be introduced by simplifying assumptions made for the structural model and
the method of structural analysis.
It is believed that to achieve the designed geometry and internal forces of the bridge within
reasonable tolerance, certain parameters need to be continuously monitored in order to determine
the appropriate preset in the fixing of the mobile carriage. Factors such as the amount of
prestress, alteration of the structural configuration, creep, shrinkage, etc. should all be taken into
account.
The construction control of cable-stayed bridges has long come to the attention of bridge
engineers [1-7]. Three main categories of practical control methods have evolved from the work
in the past two decades. They include the Kalman Filter method, relaxation of geometric
tolerance and the cybernetics approach.
The Kalman Filter method attempts to achieve the intended deck geometry by continuously
adjusting the cable tensions. In other words, it tries to achieve the design deck geometry at the
expense of the cable tensions. However, apart from significantly increasing the workload on site,
this method may also cause adverse distribution of tensile forces among the cables.
The second method involves the relaxation of geometric tolerance. The bridge is so designed
that ample tolerance is allowed in the levels of the bridge deck and the cable tensions. The
profile of the final running surface is then made good by a certain regulating course.
Construction control therefore becomes less onerous. This approach may work well for cablestayed bridges with steel or composite bridge decks. However this strategy is not suitable for
prestressed concrete as the stress limitations impose a lot of restrictions on the allowable
tolerance. Likewise the cables are normally prefabricated to the required lengths and any
significant deviations from the design values will cause much inconvenience in construction.
The third method works on an adaptive control system utilizing modern engineering cybernetics
theory [8]. In essence, the profile of the bridge deck and/or the cable tensions are continuously
monitored during the construction stage in order to identify the major design parameters and to
predict the discrepancies between the design and the completed structure. Corrective actions are
then implemented in order to minimize such discrepancies in respect of both the levels of the
bridge deck and the cable tensions.
Therefore an adaptive control system is likely to be more suitable for prestressed concrete cablestayed bridges in which a lot of restrictions are imposed by the stress limitations. In the adaptive
control system, the identification of the structural parameters is a crucial component. Tomaka
and Kamei [2] proposed a least square method of structural system identification whereby the
cable tensions could be adjusted. In this method, the structural model together with the
discrepancies has to be assumed before hand. Entire magnitudes of the structural parameters are
used and it gives rise to certain difficulties in actual practice.
This paper describes an adaptive control system, which utilizes the technique of transfer matrices
[9, 10] commonly used in structural analysis. Instead of the entire magnitudes of structural
parameters, observed increments in various steps of each construction cycle are used to identify
the parameters. The method requires less field measurements and hence it is practical but
simpler and effective. A package for the adaptive control system has been developed with Visual
C++ for use under Windows 95 or Windows NT [11]. The package has also been tested and
verified in the construction of a few prestressed concrete cable-stayed bridges.
2. An Adaptive Control System for the Construction of Prestressed Concrete

Apart from ensuring the safety during the cantilever construction of a prestressed concrete cablestayed bridge, the construction control system also strives to control the final geometry of the
bridge and the cable tensions to be within acceptable tolerances of the design values. Figure 1
shows the flow chart for such a typical system. It consists of numerical simulations of the
construction process, the building up and subsequent updating of a database of expected
characteristics, field measurements of levels, stresses and cable tensions at selected locations,
identification of structural parameters, real-time corrections and adjustments of specifications on
site as well as the management, output and storage of data.
The success of the system hinges on recent improvements in system identification and the realtime adjustments of the mobile carriages for insitu concreting of bridge segments. Trappl [8]
defined cybernetics as the "the science, craft, and art of communication, computation, and
control in the machine, the living being, and the organization". In the present system, modern
engineering cybernetics theory is used to resolve the problem of inconsistencies between the
predicted and measured values of the control variables, and to implement real-time adjustments
of the mobile carriages in order to bring the control variables to within acceptable tolerances of
the design.
2.1
Correction of Soffit Level of the Mobile Carriage
The surveying and subsequent adjustment of the soffit level of the mobile carriage is the most
crucial step in the cantilever construction of bridges. The soffit level requires continuous
monitoring round the clock so that corrections or adjustments can be implemented. Such
corrections should take into account changes in dead and live loads acting on the cantilever,
temperature distribution within the cross section, the actual stiffness of each mobile carriage as
well as the effects of construction inaccuracies of the previous segments.
Output
Measurements on Site
Bearing capacities
Moduli of elasticity
Stresses
Temperatures
Levels of bridge deck
Deflections of tower tops
Cable tensions
Database of
Expected
Characteristics
Comprehensive
Analysis
Error
Analysis
Adjustment of
Expected
Characteristics
Adjustment of
System Parameters
Prediction
Real Time
Adjustment on Site
Construction
Activities on Site
Figure 1. Flow chart for the self-correcting control system for the construction of
prestressed concrete cable-stayed bridges.
Completed
segment
Design deck
profile
Segment to be
cast
Node (i-2)
Node (i-1)
i-2
Constructed
deck profile
Node i
i-1
i-2
i-1
li-1
li
Segment (i-1)
Segment i
Figure 2. Real time adjustment of deck

Figure 2 shows the relationship between the design deck profile and the constructed deck profile
for the completed segment (i-1) and the segment i yet to be cast. The upper solid line denotes the
design deck profile. The dash line denotes the expected deck profile taking into account the
effects of changes in temporary loading, temperature difference and gradient, as well as the
deformation of the mobile carriage. The lower solid line stands for the constructed deck profile
with certain construction error. The total amount of preset i at the tip of the mobile carriage
for segment i is given by
4
i = i + g +
in which
i
(1)
is the adjustment to account for the effects of changes in temporary loading, and
temperature difference and gradient;
is the required preset of the mobile carriage to account for its deformation; and
is the adjustment to account for the construction inaccuracies of the previous
completed segments.
Assuming that there is no error due to construction inaccuracies, the adjustment i to account for
the effects of changes in temporary loading, and temperature difference and gradient can be
obtained from the geometric relationship of structural deformation [6] as
i = i-1 + (i-1 - i-2) li / li-1
(2)
where
i-1 and i-2

li and li-1
are the adjustments for segments (i-1) and (i-2) respectively to account for
the effects of changes in temporary loading, and temperature difference
and gradient; and
are the lengths of segments (i-1) and (i-2) respectively.
In actual case, there are discrepancies i-1 and i-2 due to construction inaccuracies for segments
(i-1) and (i-2) respectively. The geometric relationship of structural deformation shown in
equation (2) can therefore be amended as
i + = (i-1 + i-1) + [ (i-1 + i-1) - ( i-2 + i-2) ] li / li-1
(3)
When the length of segment is constant, equation (3) can be simplified as
i + = (2i-1 - i-2) + (2i-1 - i-2)
(4)
In case the mobile carriage is in the form of long suspended falsework [12], the required preset of
the mobile carriage is further given by
g = a - b
(5)
in which a is downward preset to account for the additional inelastic deformation due to the
tensioning of stay cables for the mobile carriage and other sources, and b is upward preset to
account for the additional inelastic deformation due to the insitu concreting of the next deck
segment and other sources. The equation for real time adjustment can therefore be obtained by
substituting equations (4) and (5) into equation (1), i.e.
i = (2i-1 - i-2) + (a - b) + (2i-1 - i-2)
(6)
There are other factors that may affect the correct preset of soffit level of the mobile carriage.
They include, for example, the tension and inclination of stay cables for the mobile carriage, and
these can be estimated from the database built up from site measurements and the learning
process in implementation. It can be observed from equation (6) that the discrepancy i-1 at
segment (i-1) will manifest itself at segment i as double of i-1. The importance of timely
correction to prevent further accumulation and propagation of errors is therefore amply

demonstrated.
2.2
System Identification using Transfer Matrices
An important feature of the adaptive control system as outlined in Figure 1 is the system
identification method. This is the essential rationale behind the control system for simultaneous
control of both the geometry of and internal forces in the cable-stayed prestressed concrete
bridge. Through the site measurements carried out at various stages of each construction cycle
and subsequent computational checks, important structural parameters such as the creep
coefficients, segment weights, flexural rigidities of the towers and decks, etc. can be estimated
and refined. During the cantilever construction of a bridge, each cantilever essentially comprises
a chain of segments, and such a configuration favours the use of transfer matrices [9,10] in its
analysis and monitoring.
v
Mi(li)
Mi(0)
Ni(0)
Node i
Ni(li)
Node i+1
li
Qi(0)
Qi(li)
Figure 3. Nomenclature of displacements and end forces acting on segment

Figure 3 shows the nomenclature of displacements and end forces acting on an existing segment i
of the bridge. During a certain step of the construction cycle of the segment at the cantilever tip,
the major load increment is not applied on segment i. The state vector Si+1 at node (i+1) is
related to the state vector Si at node i through the transfer matrix Gi by
Si+1 = Gi Si
(7)
or
u i ( l i ) 1
v (l ) 0
i i
i ( li ) 0
=
(
)
N
l
i
i
0
Qi (li ) 0

M i ( li ) 0
0
1
0
0
0
0
0 li
li
1
0
0
0
/ Ei Ai
0
0
1
0
0
0
l / 6 Ei I i
l / 2 Ei I i
0
1
3
i
2
i
li
0
ui ( 0)
l / 2 E i I i v i ( 0)
li / Ei I i i (0)
0
N i (0)
Qi (0)
0
1
M i ( 0)
2
i
(8)
where
Ei, Ii and Ai
ui(0) and ui(li)
vi(0) and vi(li)
i(0) and i(li)
are the modulus of elasticity, second moment of area and cross

sectional area, respectively, of segment i;
are the axial displacements at the ends of segment i;
are the vertical displacements at the ends of segment i;
are the rotations at the ends of segment i;
6
Ni(0) and Ni(li)

Qi(0) and Qi(li)
Mi(0) and Mi(li)
are the axial forces at the ends of segment i;

are the shear forces at the ends of segment i; and
are the bending moments at the ends of segment i.
In particular, the vertical displacement and rotation at node (i+1) are given respectively by
vi (li ) = vi (0) + lii (0) + li3Qi (0) / 6 Ei I i + li2 M i (0) / 2 Ei I i
2
i
i (li ) = i (0) + l Qi (0) / 2 Ei I i + li M i (0) / Ei I i
(9)
(10)
At each step of a construction cycle, site measurements are carried out and these can be
attributed to certain increment or decrement of loading, such as tensioning of stay cables. On
rearrangement, equations (9) and (10) give
Ei I i =
li3Qi (0) + 3li2 M i (0)

6 (vi (li ) vi (0) lii (0) )
(11)
v (l ) vi (0) lii (0)
i (li ) = i (0) + 3 (li2Qi (0) + 2li M i (0) ) i 3 i

2
li Qi (0) + 3li M i (0)
(12)
The loads acting on the cantilever during construction as well as their subsequent variations are
closely monitored. For example, the force applied on tensioning each cable is accurately
measured. The subsequent variation of cable force is detected by regular monitoring of ambient
vibrations. The actual amount of concrete and other construction materials that goes into the
bridge deck is also closely monitored and refined. The bending moments Mi and the shear forces
Qi can be estimated from information on such loading. The vertical deflections vi can be
regularly monitored from precision surveys. The rotation at each node i can be calculated from
equation (12) based on the rotation at the previous node. For symmetrical balance cantilever
construction, the rotation i(0) at the tower-deck junction can be taken to be zero, and hence the
rotations at other nodes can be calculated according to equation (12). Equation (11) can also be
utilized to estimate the flexural rigidity EiIi. Iterations are carried out until the discrepancies
between the predicted values and measured values are eliminated. In most cases, the tensile
forces in a few cables are particularly sensitive to concrete creep. Therefore, through continuous
monitoring of these cable forces, it is possible to obtain reliable estimates of the creep coefficient
[5].
3. System Control Software with Graphical User Interface

Advancements in modern computer technology have enabled the implementation of the adaptive
construction control of prestressed concrete cable-stayed bridges using a personal computer, in
spite of the complexity of the problem. A computer package to realize the strategy outlined in
Figure 1 has been developed [11]. The enormous amount of computation for the simulation of
the construction process is carried out mainly to confirm the correctness of the design erection
sequence and to ensure the safety of the structure during erection. The package was developed
using OpenGL for the application programming interface in order to make it user-friendly. The
reliability of the package has been tested in the construction of around 10 cable-stayed bridges.
4. Case Study
The use of the above system in the construction of a real bridge is described below. It is a
prestressed concrete cable-stayed bridge that spans over a wide river. The bridge is an
asymmetric single-tower cable-stayed bridge with two vertical planes of stay-cables. The main
crossing consists of 5 spans, namely 74.5m, 258m, 102m, 83m and 49.5m. The 29.5m wide
bridge deck is "floating", comprising two spine box beams of 2.5m depth transversely connected
by cross girders at 4m spacing. The 148.4m tall tower is of "H" shape with inclined legs. There
are altogether 102 stay-cables arranged in a modified fan pattern. The typical spacing between
cable anchorages at deck level is 8m, as shown in Figure 4.
Figure 4. Elevation and section of a prestressed concrete cable-stayed bridge over a wide river.
During the construction of segment 11 of the bridge, the levels of segments 9-11 were obtained
from precision survey and they are shown in Table 1.
Tip of segment no.
Design level (m)
Measured level (m)
Discrepancy (m)
9
37.990
37.903
-0.026
10
37.926
37.816
-0.015
11
37.872
37.784
Table 1. Levels during construction of segment 11

From measurements obtained in previous construction cycles, the required presets a and b of
the mobile carriage were estimated to be 30mm and 70mm respectively. Substituting these
values into equation (6),
= (2i-1 - i-2) + (a - b) + (2i-1 - i-2)

= [2(37.926-37.816) - (37.990-37.903)] + (0.03-0.07) + (-20.015+0.026)
= 0.089m
The corrected soffit level of the mobile carriage should then be (37.872m-0.089m) or 37.783m.
The set-up soffit level was 37.784m, implying a discrepancy of 1mm only.
An electronic balance on site closely monitored the actual weight of concrete that went into each
segment. The average weight of concrete used in each segment was 3390kN. However from the
8
numerical simulation of the erection process taking into account various site measurements, the
adjusted weight of concrete in each segment was only 3351kN. It indicated around 1.2% of loss
of concrete mainly inside the conveying pipes for concreting.
The discrepancies of deck profile when construction reached segment 20, i.e. when the cantilever
was 168m long, are shown in Figure 5. The discrepancies of deck profile were within 40mm and
the cable tensions were all within 5% of the design values.
0.05
'#
19
'#
17
'#
15
'#
13
'#
11
9'
#
7'
#
5'
#
3'
#
1'
0#
2#
4#
6#
8#
#
10
#
12
#
14
#
16
#
18
20
-0.05
Figure 5. Discrepancies of deck profile in metre when the cantilever was 168m long.
5. Conclusions
In the construction of prestressed concrete cable-stayed bridges, simultaneous control of the
geometry and internal forces is very important. An adaptive control system has been developed
utilizing modern engineering cybernetics theory. The system makes use of the structural
parameters identified for the continuous adjustments of the mobile carriages for insitu cantilever
construction of the bridge deck. The system was developed utilizing extensive graphical
capabilities to make it user-friendlier. The system has also been successfully applied to the
construction of several long span prestressed concrete cable-stayed bridges.
6. References
[1]. F. Leonhardt, Die Spannbeton-Schrag-Kabel Bruken uber den Columbia River Zwischen
Pasco und Kennawich in Etaat Washington, U.S.A., Beton und Stahllelonban, Heft, 1-4,
1980.
[2]. H. Tomaka and M. Kamei, Cable Tension Adjustment by Structural System Identification,
Proceedings, International Conference on Cable-Stayed Bridges, IABSE, Bangkok,
Thailand, Nov., 1987.
[3]. Y.P. Lin, The Application of Kalman Filter in the Construction of Cable-Stayed Bridges (in
Chinese), Tumu Gongcheng Xuebao (Journal of Civil Engineering), China, Vol. 3, 1983.
[4]. D.W. Chen, X.G. Zheng and H.F. Xiang, Construction Control of Concrete Cable-Stayed
Bridges (in Chinese), Tumu Gongcheng Xuebao (Journal of Civil Engineering), China, Vol.
26, No. 1, 1993.
[5]. D.W. Chen and H.F. Xiang, Application of Construction Control in the Cable-Stayed
Bridge over Yong River (in Chinese), Proceedings of the 1992 National Conference on
Bridge Structures, Tongji University Press, 1992.
[6]. D.W. Chen, L.C. Fan and H.F. Xiang, The Construction Control of the Single-Tower
Cable-Stayed Bridge at Sanshui of Guangdong (in Chinese), Journal of Tongji University,
Vol. 25, No. 1, 1997.
[7]. D.W. Chen, Z.Y. Han, D.J. Huang, Y.L. Qian and L.C. Zhang, The Construction Control of
Zhaoboshan Bridge at Ningbo (in Chinese), Annual Conference of Municipal Engineering,
Chinese Society of Civil Engineering, Tianjin University Press, 1998.
[8]. R. Trappl, Cybernetics Theory and Applications, Hemisphere Publishing Corporation,
1983.
[9]. R.K. Livesley, Matrix Methods of Structural Analysis, Pergamon Press, 1975.
[10]. S.W. Cai, Matrix Method in Structural Mechanics (in Chinese), China Communication
Publisher, 1975.
[11]. D.W. Chen, X.T. Wang, W.Y. Chen and L.C. Fan, A Software with Graphical User
Interface for Construction Control of Cable-Stayed Bridges (in Chinese), Journal of Tongji
University, Vol. 26, No. 5, 1998.
[12]. D.W. Chen, D.J. Huang and H.F. Xiang, A New Method for Erecting Cantilevers with
Long Suspended Falsework in P.C. Cable-stayed Bridges (in Chinese), Tumu Gongcheng
Xuebao (Journal of Civil Engineering), China, Vol. 29, No. 6, 1996.
10
The Lifting, Transport and Placing of the resund Pylon Caissons

Ferdinand TRENKLER
Civil Engineer
VSL (Switzerland) Ltd.
Lyssach, Switzerland
Ferdi Trenkler, born 1941, received
his civil engineering degree in 1966
from the Swiss Federal Institute of
Technology. Since 1977, he has
worked with VSL (Switzerland) Ltd.
Presently he is Chief Engineer with
Heavy Lifting, one of VSL's worldwide activities.
Dr. Petter SKRIKERUD

Civil Engineer
Structural Engineering AS
Lysaker, Norway
Petter Skrikerud, born 1947,
received his civil engineering degree
in 1970, and the Ph.D. degree in
1982, from the Swiss Federal
Institute of Technology. He is
presently senior partner of Structural
Engineering AS, and is mainly
engaged with projects for the
offshore oil industry.
Capt. Dan M. VOLL

Master Mariner
Neptun AS
Stavanger, Norway
Dan Magne Voll, born 1952, seaman
experience since 1968, received his
Master Mariner degree in 1974. His
experience encompasses master of
various marine operations with
command of several vessels and
equipment. He is presently tow
master / salvage master and is
involved in offshore oil industry
projects.
Summary
The two 204 metre tall pylons of the central, cable stayed bridge of the new resund crossing
between Denmark and Sweden rest on concrete caisson foundations. The caissons have plan
dimensions of 35 by 37 metres, are 22.5 (21) metres high and have a dry weight of
approximately 20'000 tonnes.
The two caissons were transported one by one from the prefabrication site - the Kockums dry
dock in Malm - to their permanent offshore location in resund. This operation was
accomplished by means of a purpose-built catamaran, suitably equipped with hydraulic
lifting/lowering equipment, carrying the partly submerged caissons with high-strength tendons.
The present paper describes the caisson LTP operations (Lifting, Transport and Placing) which
were successfully carried out in April, 1997.
Introduction
1.1
The Fixed Link across resund
The 16 km long coast to coast link between Denmark and Sweden consists of two railway tracks
and a four-lane motorway, and comprises the following principal elements (Figure 1); a short
peninsula at the Danish coast, an immersed tunnel, an artificial island, and a 7.8 km long bridge.
The tunnel starts near Copenhagen airport and the bridge ends just south of Malm. The resund
Link is scheduled to be opened for public traffic in the year 2000.
The bridge part of the resund Link consists of two approach bridges leading to a central, 1'092
m long, cable stayed "High Bridge" spanning the new, dredged ship lane Flinterenden. The two
204 metre high pylons on either side of the 490 m long central bridge span rest on two
prefabricated concrete caissons.
Figure 1 The resund Link

1.2
The Caissons
Although the technique of placing prefabricated concrete structures onto seabed foundations is
not new, the resund caissons represented a new scale in terms of size, weight and installation
accuracy.
Each caisson consists of a foundation base, measuring 35 m x 37 m in plan, extending into two
base sections for the pylon towers (Figure 2). The total height is 22.5 m for the west caisson and
21 m for the east caisson. After placing of the caissons into pre-dredged trenches (approx. 10 m
deep) in the limestone seabed, the two tower sections extended 4.0 m above sea water level, thus
allowing concrete casting work to continue.
The dry weight of each caisson was about 20'000 tonnes, including temporary fixtures (working
platform, crane pedestal, etc.). During transportation to the installation site the caissons were
partly submerged (draught about 6.4 metres), and the net design lifting force was specified as
12'200 tonnes.
The caisson design included horizontal post-tensioning in both directions as well as vertical posttensioning of the outer walls. The anchors of these vertical post-tensioning cables were utilised
as couplers to the tendons of the lifting/ lowering equipment.
Figure 2 Schematic View of Pylon Caisson
1.3
Organisation and Scope of Work
The overall responsibility for the caisson LTP operations (Lifting, Transport and Placing) was
with the Main Contractor for the resund Bridge, Sundlink Contractors, which is a consortium
of Skanska AB (Sweden), Hochtief AG (Germany), Monberg & Thorsen A/S (Denmark) and
Hjgaard & Schulz a/s (Denmark).
Sundlink Contractors engaged Neptun Heavy Lift AS (Norway) for carrying out the marine
works, and VSL (Switzerland) Ltd. for the lifting and lowering tasks of the caisson LTP
operations. Neptun subcontracted the engineering and design to Structural Engineering AS
(Norway), and engaged Gdansk Shiprepair Yard (GSY) Remontowa (Poland) for the necessary
conversion of the marine equipment.
Due to the substantial weight of the caissons, the possibility of using a conventional crane vessel
for their installation was found not feasible. Furthermore, the caissons could hardly be made to
float freely, and if so at a draught, which would be larger that the available water depth in
resund. Hence, an alternative solution had to be established in order to bring the caissons from
the prefabrication site in the dry dock of the Kockums yard in Malm to the installation site in
resund, and place them safely onto the pre-installed foundation pads in the excavated trench
within the given tolerance of 75 mm.
The solution arrived at was to convert (lengthen) and strengthen two existing Neptun heavy lift
barges, Goliat 18 and Goliat 19, both having the overall dimensions 80 m length, 17 m breadth
and 15.8 m height after conversion. The two barges were then joined by means of two large
space truss structures fore and aft, forming a catamaran (Figure 3).
Figure 3 Finite Element Model of Catamaran
Catamaran
2.1
Design Resume
The catamaran was designed to carry the 12'200 t net weight of the partly submersed caisson at a
common draught of 6.4 m. The response of the catamaran was determined in three different
environmental and operational conditions; fully loaded and with the caisson seafastened in
transit to the installation site, empty in transit back from the installation site, and for the caisson
hanging freely in the tendons during lowering (at several intermediate draughts) with the
catamaran moored over the installation trench.
The first two analyses, with the catamaran and the caisson modelled as one common body, were
used to design the catamaran itself, to check the dynamic load in the lifting tendons, and to
design the horizontal seafastening. Thereby the hydrodynamic pressures and accelerations for
both the maximum transit and the holding conditions were transferred to the finite element
model of Figure 3 for determination of stresses and deflections.
For the third hydrodynamic analysis, lowering of the caisson onto the foundation pads, the two
structures, catamaran and caisson, were modelled as two separate bodies interconnected by the
lifting tendons. The analysis served to verify that both the relative and absolute movements of
the caisson were within the tolerances for all stages of the lowering operation. In particular, the
analysis served to determine the maximum tolerable sea state at touch-down, for an installation
within the given tolerances.
As a result of the design calculations, both barges had to be strengthened. The main areas of
strengthening were; the side shell under the load carrying hydraulic jacks and towards the
connecting trusses, the deck, and the local area of connection between truss and barge.
The two connecting space truss structures were after the LTP operations removed from the
barges and used as temporary support for the main bridge girder.
2.2
Hydraulic Lifting / Lowering System
In order to lift the caisson up from the bottom of the flooded dry dock, and to lower it into the
cavity at the installation site, each barge of the catamaran was equipped with 20 VSL hydraulic
strand lifting/lowering jacks, located on cantilevering supports at the inner side shell (Figure 4).
Each hydraulic strand jack had a nominal lifting capacity of 330 t (total for 40 jacks 13'200 t, or
108% of the design load), whereas the nominal breaking load of an individual tendon, made up
of 31 high-strength strands, was 930 tonnes (total 37'200 t, or 305% of the design load).
All 40 strand jacks were hydraulically coupled to a 3-point, static determinate system during
lifting and lowering (20 on one barge and two times 10 on the other). The system was remotely
controlled from a single control centre, which also included electronic level and pressure control.
Figure 4 The 20 Strand Jack Assemblies on Goliat 19

During sea transportation, the strands of the tendons were mechanically locked, thus forming a
structural entity with the catamaran in the vertical direction. To control the relative motion
between the catamaran and the caisson in the two horizontal directions, surge and sway stoppers
(seafastening) were installed on the inner side shells of the barges (Figure 5).
The connection between the tendons and the caisson was solved with a specially designed screw
on/off type coupler. The anchor heads of the vertical post-tensioning cables in the caisson walls
served as mating pieces for the couplers. Whereas the coupling onto the anchor heads was
effectuated in the dry dock by closing the gate and lowering the water level (Figure 6), easy and
quick uncoupling by divers in 12 m water depth was a main requirement for the design of the
couplers.
Figure 5 Surge and Sway Stoppers

2.3
Figure 6 Installing Tendon Couplers
Conversion and Outfitting
The strengthening and conversion of the two barges Goliat 18 and Goliat 19, the fabrication of
the two space truss structures, the assembly of the catamaran, and the steel outfitting took place
at the conversion yard GSY Remontowa in Gdansk, Poland, in the period mid September, 1996
to mid February, 1997. The final outfitting of the catamaran was carried out in Malm during
March, 1997.
The work at GSY Remontowa consisted of;
lengthening of Goliat 18 by a centre section of 26.6 m to the same length as Goliat 19 (80 m)
fabrication of the two space trusses (overall size: 70 x 10 x 16 m, and weighing about 375 t
each)
fabrication and installation of cantilevering supports for the strand jacks (40 off)
fabrication and installation of surge and sway stoppers (for horizontal seafastening)
supply and installation of bottom valves, and amendment of ballast tank compartment, to
enable fast ballasting of the barges during caisson lowering
assembly of the catamaran
steel outfitting (various gangways, ladders, platforms, etc.) of the catamaran
For the installation of the supports for the strand jacks, the as-built locations of the posttensioning (coupling) anchors on the caissons (average of the two) were taken into account.
Thanks to close co-operation between all involved parties, the entire conversion and outfitting
work at GSY Remontowa, from ordering of the first steel until delivery of the catamaran, was
completed according to schedule in just over five months.
Lifting, Transport and Placing (LTP) Operations
3.1
Lifting of Caissons in Kockums Dry Dock in Malm
After the final outfitting of the catamaran in Malm harbour had been completed, and the two
caissons were ready for the LTP operations, i.e. complete with work platforms and crane
pedestals for the continuation of the construction of the pylons, the Kockums dry dock was filled
with water (Figures 7 and 8).
Figure 7 Ingress of Water
Figure 8 Water Filled Dry Dock
The buoyancy reduced the net weight of the caissons to about 9'000 tonnes. After opening of the
dock gate the catamaran was manoeuvred into the dock by means of tugs, guides/fenders and
winches. In order to avoid clash between the leading space truss of the catamaran and the
protruding reinforcing bars of the pylon bases, the catamaran had to be trimmed to about 6
degrees by ballasting the end ballast tanks.
When the catamaran was in position over the caisson, it was brought to even keel by
deballasting, the dock gate was closed, and the water level was lowered by about 2 metres to
allow coupling of the lifting tendons to the post-tensioning anchor heads of the caisson (Figure
6). Then water was again filled into the dock, and the caisson was lifted by means of the strand
jacks to a common draught with the catamaran (6.4 m). The net (lifting) weight thereby
increased to about 11'800 tonnes (west caisson) and 11'500 tonnes (east caisson).
For the first (west) caisson, a test lift was conducted to verify the strength of all lifting and
structural elements. This task was effectuated by increasing the lifting height of the caisson
(reducing draught and thus buoyancy) by 1 metre, thereby fully utilising the total lifting capacity
of the strand jacks, thus achieving a lifting load of about 110% of the actual.
After shimming the outer caisson wall against the surge and sway stoppers (Figure 5) to achieve
a horizontal seafastening, and fixing the lifting tendons at their current position (mechanical
locking of the individual strands), the dock gate was opened and the loaded catamaran was
manoeuvred out of the dock by means of winches and tugs, Figure 9. Well outside the dock, but
still within the Malm harbour, the five tugs of the towing fleet were connected to the catamaran,
ready to start the tow to the installation site.
The operational environmental criteria for leaving Malm harbour were specified as a 72 hours
weather window with forecast Beaufort 5 or less and favourable conditions for the following 2
days. In addition, the 12 km long tow to the installation site in resund via the very busy ship
lane Flinterenden, had to be co-ordinated with other ship traffic, because the 70 m wide tow
utilised the entire ship lane.
Aft View
Figure 9 West Pylon Caisson Leaving the Dry Dock
3.2
Arial View
Tow-out, Transportation and Mooring
The towing fleet for the transport of the loaded catamaran consisted of 5 tugs; one lead tug of
about 60 tonnes bollard pull, and four assisting tractor tugs hooked up to the catamaran of about
35 tonnes each (Figure 10). The towing fleet was dimensioned to safely manoeuvre the
catamaran in the strong currents of the resund (design value 1m/s, holding condition 1.75 m/s).
Figure 10 Towing Configuration
The towing route had to be carefully selected due to the large draught (6.4 m) and the generally
shallow waters of resund. As primary navigation and positioning aid, satellite navigation (GPS
system) was used.
Prior to starting the LTP operations, the tugs and vessels of the Neptun fleet had installed and
tested the mooring spread for the catamaran for the west pylon location. The pre-laid mooring
spread consisted of 8 legs, each of which made up of a 12 tonnes Stevprice anchor and 76 mm
chain (length between 150 and 200 m), which were picked up via surface buoys and coupled to
the catamaran mooring wires. In east-west direction, the latter were tensioned by four mooring
winches on board the catamaran itself. In north-south direction, the mooring legs were connected
to two of the tractor tugs on one side, and to a common anchor vessel connected to the towing
bridle on the other side. The north-south mooring legs were tensioned by means of the winches
on board the tugs and the anchor vessel, respectively. Figure 11 gives an overview of this
mooring configuration.
Figure 11 Mooring Configuration

During the lifting of the east pylon in the dry dock, the mooring spread was re-located to the
installation site of the east pylon and tested.
3.3
Placing on Pre-Installed Foundation Pads
The operational weather window for releasing the seafastening and lowering the caisson onto the
pre-installed foundation pads in the cavity was specified as 20 hours of Beaufort 3 or less.
During lowering of the caisson, the catamaran was concurrently filled with ballast water in order
to avoid unnecessary lowering distance. The ballasting was effectuated by opening the bottom
valves of the barges and redistribution of the ballast water to other ballast tanks as required using
portable pumps. Thus the catamaran was kept on constant draught 6.4 m throughout the lowering
operation. At the end of the 12 m lowering distance the net weight of the caisson was reduced to
about 3'000 tonnes.
Just before touch-down, the position of the caisson was checked and any corrections made by
means of the mooring lines, and, if necessary, with four tugger winches with wires attached to
the caisson. The achieved installation accuracy was well within the given tolerance of 75 mm:
42 mm for the west and 24 mm for the east caisson.
Immediately following touch-down, the tendons were paid out by several full strokes of the
strand jacks and the barges were flooded by opening the bottom valves, in order to quickly
slacken the lifting tendons and thus stabilising the caisson.
The couplers between the lifting tendons and the post-tensioning anchor heads were released by
divers. Due to the strong currents experienced in resund, the divers had to tie themselves to the
tendons to be able to release the coupler. When a coupler had been released the lifting tendon
was retracted to surface and subsequently prepared for the next lift.
Conclusions
The described LTP operations for the two pylon caisson of the High Bridge of the resund Link
took place in April 1997. Weather conditions, as usual at this time of the year, were sometimes
quite rough. Picking the right weather window became one of the most difficult tasks of the
operation.
The fact that the whole operation was a real success is the result of a great teamwork of
specialists from not less than 7 nations.
The Val-Benoit Cable-Stayed Bridge

Jean-Marie CREMER
Civil Engineer
Engineering Office Greisch, Lige,
Belgium
Jean-Marie CREMER, born 1945, received his Civil Engineers diploma at

the University of Lige, 1968. He presently carries on the function of managing director at the ENGINEERING
OFFICE GREISCH. He performs the
direction of the bridges department.
Professor at the bridges chair of the
University of Lige.
1. Situation
The city of Lige in Belgium is located at the intersection of 7 highways leading to important European
cities.
A connecting highway goes round Lige on the south-west, along a very important railway axis. Along
this 4,1 km long connection, there are among others a double tunnel under the hill adjoining the city, a
bridge to cross the river and a tunnel under an important railway junction.
Figure 1.
2. Choice Of The Type Of Bridge
Figure 2.
The site constraints and the transportation requirements over and underneath the bridge are numerous
and contradictory. Squeezed between two tunnels, the bridge crosses the river and roads on both
banks. The grade profile on the bridge does not allow slopes over 6,5 % to enter or exit the tunnels.
Under the bridge, the road clearance, mainly on the left bank, and the navigation clearance allow a
maximal thickness of 2,50 m for the deck. The river, over 140 m wide, has to be navigated in the future
by pushed convoys of 9.000 tons. There is a sharp curve in the river at that place. Just below the new
bridge, two older ones, with two piles each in the river, are already considered as security obstacles for
the navigation. For the new bridge, piles in the river had obviously to be avoided, if it was possible.
Figure 3.
Various types of bridges have been studied, a classical one with a variable height girder and a support in
the river, a arch-tied bridge, a cable-stayed bridge with a pylon on each bank and a cable-stayed bridge
with a single pylon.
In addition to the sharp curvature of the river, the site is characterised by a sharp housing dissimetry. On
the right bank, the outer side of the curve, the quite flat ground is mainly occupied by industrial
constructions, dominated by the important railway junction. On the left bank, the inner side of the curve,
the urbanistic texture is mainly made of habitations, squeezed between the river and the hill.
The numerous technical requirements and the site constraints have led to the choice of a cable-stayed
bridge with a single pylon on the right bank.
Its particularity of highway bridge in urban site has also led to the need to combine simplicity,
impeccable appearance and high quality finishing materials.
Figure 4
The esthetical quality of the bridge is related to the following particularities :
- The necessary readability of the structure has led us to choose a single cable plane, located on the
axis of the bridge.
- The search for thin structural elements bordered by more monumental abutments is satisfied by the
stays and the very thin circular pylon, which slenderness is accentuated by the truncated cone shape.
On the contrary, the choice of a single cable plane thwarts this search for slenderness for the deck.
Indeed, the height of the deck box-girder, required by the torsional stiffness, does not match this
objective.
However, the design of the cross-section allows to limit its visual impact. The deck slab is located
around mid-height of the box-girder and is supported by very sloping lower steel tubes which improve the lower part and accentuate the impression of slenderness, as the upper part is hidden by the
lateral security barriers.
- The balancing span replaced by a balancing abutment which, on the one hand, points out the
entrance portal of the tunnel and, on the other hand, acts as an acoustical protection for the near by
habitations. The visible concrete facings are concreted on site in plank-structured formworks; the
lower face of the deck is particularly being taken care of as it is very closely perceived by the
walkers.
Figure 5.
- The faces of the abutments are realised with country natural stone as the roof of the balancing
abutment is entirely covered with vegetation.
- The pylon, shaped as a slightly truncated cone, is covered with glazed glass, fixed with bolts.
- The sheathes of the stays as well as the external sleeve of the steel tubes supporting the deck slab are
made of stainless steel.
- A specific lightning device points out the bridge and respects the habitation neighbourhood.
- Absorbent coverings are widely used to limit at best the acoustical nuisances.
- Landscaped arrangements such as pedestrian and cycle tracks, rest zones, street furniture and
vegetation complete the urbanisation of the site, strongly perturbed by the monumental bridge.
- An esplanade, widely open on the river, clears the base of the pylon.
Figure 6.
3. Technical description
Figure 7.
The main span of the bridge over the river is 162 m long, in continuity with a short 31.5 m long span
above the left embankment and a very short 12 m long span between the pylon and the balancing abutment. This abutment is 122 m long and is the first part of a tunnel that goes on under the railway junction.
The whole structure bears through bored piles on a gritty shistous foundation rock of good quality,
however crossed by thin coal veins. The left bank abutment is founded on 1,28 m vertical and inclined
piles. The other bearings, that means the pile P1, the pylon and the balancing abutment, are founded on
1,50 m vertical piles.
Figure 8.
The prestressed concrete deck, 25,90 m wide, has a rather uncommon cross-section, with a small central box-girder, 3,75 m high and 4,50 m wide, and two cantilever slabs, located 1,25 m beneath the
upper level of the box-girder.
The prolongation of the box-girder above the cantilever slabs is necessary to obtain a torsion rigidity
sufficient to resist to the transversal loads of balances. The cantilever slabs are supported every 3 m by
steel tubes, they are also transversally prestressed by 4T15 cables every 50 cm.
On most of its length, the deck is longitudinally centrally prestressed. The end span which is not cablestayed and the zone around the bearing on left bank have to be fitted with 16 complementary undulated
cables 19T15. The whole prestressing is located inside the concrete and is injected with a cement
grouting.
Figure 9.
The pylon, located on the axis of the bridge, has a total

height of 82 m. It is rigidly restrained to the deck. It is covered with glazed glass. Its double foundation sole allows a
slight longitudinal translation due to the effects of shrinkage,
creep and temperature variation in the part of bridge that
separates the pylon from the fix point around the middle of
the balancing abutment.
The deck is supported, above the river, by 22 stays, balanced on the pylon by 22 other stays which are anchored in
the balancing abutment. The 44 stays are made of sheathed
greased galvanised strands in an external 2 mm thick stainless steel sleeve. The sections of the cables vary from
81 cm (54T15) to 137 cm (91T15). The external diameters of the sleeves are 219 and 254 mm.
At the lower part of the stays, a 5 m long double sheath,
filled with wax, insures an excellent dumping towards parasitic vibrations.
Figure 10.
4. Execution
The deck is positioned by incremental launching

and here are the main stages :
- foundation of the pylon and temporary
supports for the launching,
- construction of the deck on the right bank
and incremental launching in 18 stages,
- construction of the pylon and the balancing
abutment,
- fitting of the stays and adjustment of the
geometry,
- removal of the temporary supports and
fitting of the bridge equipment's,
Figure 11.
a) Foundations
The design of the pylon foundation and the construction
method have mainly been dictated by the level of the
Meuse river (60,0), influencing directly the level of the
water-table in the gravel's behind the embankment walls.
The difficulty to realise a water-tight cofferdam and the
need of the foundation slab to be concreted out of the
water have dictated a lower level of the foundation slab
above level 55,5.
Since the foundation rock is crossed by a 50 cm thick
coal vein of weak mechanical characteristics, the use of
1,50 m bored piles has allowed, on the one hand, to
bridge locally the coal vein and, on the other hand, to
build the foundation slab out of the water.
Figure 12.
b) Deck
The deck is made of 18 pieces, out of which 16 ones are 12 m long, and is built on a fix site located on
the right bank and positioned by incremental launching 12 m at a time.
The site is divided in five 12 m long working zones :
- zone 0, assembly of the reinforcement steel for the floor of the box-girder,
- zone 1, formwork fitting and concreting of the central box-girder and tightening of the launching
longitudinal prestressing,
- zone 2, free zone,
- zone 3, formwork fitting and concreting of the cantilever slabs and tightening of the longitudinal
prestressing of the slabs,
- zone 4, fitting of the lower steel tubes and tightening of the transversal prestressing of the slabs.
To allow the launching, three temporary supports are placed in the bed of the river, one temporary support on the left bank, two on the right bank, plus the launching abutment, where are placed the jacks,
able to produce the 600 tons horizontal effort, necessary for the launching. The front end of the boxgirder is prolonged by a 32 m long steel nose.
The succession of the operations is such than in a cycle of one week, a 12 m long piece is realised and
launched, as the tightening of the longitudinal prestressing, the removal of the formworks and the
launching are realised each Monday morning.
c) Pylon
The pylon, culminating 70 m above the deck, is built once the deck
has reached its definitive position.
The pylon has a circular cross-section, shaped as a slightly truncated
cone. It is concreted with climbing formworks, about 4 m at a time. A
38 m high steel structure is incorporated in the concrete, in order to
facilitate the fitting of the stays.
The glazed glass facing is planned for the end of the works, once most
of the contractions in the pylon have occurred.
Figure 13.
d) Balancing abutment
Figure 14.
The balancing abutment is a classical reinforced concrete structure, founded on 1,50 m bored piles.
Started at the beginning of the works, the construction goes on during the erection of the pylon.
e) Stays
As soon as the construction of the pylon and the balancing abutment is finished, the stays are placed,
starting with the shortest ones and symmetrically with regard to the pylon.
The first adjustment, as the stays are being placed, allows to make up for 70 % of the lengthening under
permanent loads. The definitive adjustment is relaxed as all the stays are placed, the temporary supports
are removed and all the equipment's are fitted.
5. Studies
The design of this cable-stayed bridge is unconventional and the implantation site has some particular
characteristics. It has required important specific
studies.
In addition to the difficulties related to the particular
shape of the deck and to the tridimensional behaviour of the balancing abutment, some uncommon
problems have been met.
Figure 15.
A railway bridge is located 80 m down the construction site. The important wind turbulences, provoked
by the very angular shapes of this bridge, risked to have a negative effect on the aerodynamical behaviour of the new bridge. An analysis of this behaviour, in situ measures and wind-tunnel tests have shown
that the turbulences created by the railway bridge risked not to cause tangible disagreements for the
cable-stayed bridge.
A 50 cm thick coal vein goes askew in the foundation rock under the pylon. A very detailed simulation
of the foundation soil behaviour has been realised in collaboration with the University of Lige. This
study has allowed to precise the effect of the coal vein and to determine the settlements (millimetres) to
be considered in the calculation of the superstructure.
Construction Control Practice for Panyu Cable-stayed Bridge

Dajian HAN
Prof. Dr.
Soth China Univ.
Technology
Guangzhou, China
Quansheng YAN
Assoc. Prof. Dr.
South Cina Univ.
Technology
Guangzhou, China
D.J.Han, born 1940,

received her B.S. 1963
Peking Univ. China M.S.
1982, Ph. D. 1984 Purdue
Univ. USA
Q.S.Yan, born 1968,

received his B.S. 1985,
M.S. 1988, Ph. D 1994
Changsha Railway Univ
China
Summary
This paper presents the methods used to simulate the erection process, to monitor and adjust
the geometry and cable tensions of Panyu cable-stayed bridge. The bridge with 380 meters
long of the main span is built using the balanced cantilever construction method. Bridge
segments are cast in-situ with traveler form. Due to the flexibility of deck, cable-supported
carriages are used. It brings more difficulty into the construction control. Thus it is important
to carry out careful and detailed simulation of the construction processes. It shows that
detailed calculation and continuous monitoring of the erection process can lead to high
precision and give good knowledge of the static behavior of the actual bridge. Such a method
makes it possible to achieve a high level of accuracy for construction of PC cable-stayed
bridges.
1.
Introduction
The general view of Panyu cable-stayed bridge is shown in Fig.1. The cable-stayed portion of
the bridge has a total length of 702.0 meters. The main span between towers is 380.0 meters
and the side anchor spans are 161.0 meters. The side anchor spans consist of two spans of
90.0 and 71.0 meters long. An auxiliary pier is in between. The deck is composed with two
edge girders and a deck plate. The edge girders are laterally stiffened by a T-shape PC girder
with 6 meter spacing. The edge girder is a solid section whose height is 2.2 meter and width
varies from 2.6 meter at intersection of girder and pylon to 2.0 meter at middle span. The deck
plate is 28cm thick. The width of the deck is 37.70 meter out to out with 8 traffic lanes.
Spatial 264 stay cables are arranged in a semi-fan configuration. The pylon shape like a
diamond with an extension mast. All the cable stays are anchored in the mast part of the
pylon. The stay cables are attached to the edge girders at 6.0 meter spacing. At the side anchor
span an auxiliary pier is arranged to increase the stiffness of the bridge. Two anchorage
segments of deck at both ends of the bridge are set up to balance the lifting forces from
anchorage cables.
For a bridge with multiple cable systems, the girder with lower height becomes more flexible
and make it more difficult to control the geometry and stay tensions properly. And due to
inevitable errors between design values and actual ones at the construction stage, it is
necessary to carry out a detailed simulation analysis and a continuous monitoring throughout
the erection process [1][2]. Thus the real state of the structure can be obtained in order to
determine the most suitable adjustment of cable tensions and of the elevation of the given
segment.
In this paper the detailed simulation analysis of the erection process, the methods used in the
control and adjust the deck profile and the stay of Panyu cable-stayed bridge are discussed.
Photo 1. The general view of PanYu cable-stayed bridge after completion
2.
Erection Process and Simulation Analysis
2.1 Erection process

The deck of the bridge is erected by the balanced cantilever method utilizing cable-stayed
form carriages. Due to its large width of 37.7 meters, each segment is nearly 4200.0 kN
weight. But the thickness of deck plate is only 28.0 cm. It is necessary to provide more rigid
form carriages in order to reduce local deformations. And the weight of a form carriage is
about 2500.0 kN. The construction process is briefly described as follows:
. To build the towers
. To cast in-place the first segment on timbering support
. To erect the No. 1 cable and stress to its final length
. To hoist the traveling carriages and locate their positions.
. To erect the girder segments one by one on the two sides of the pylons
. To connect the cantilever ends of the side span with the anchorage parts at two ends.
. To continue to erect the remaining girder segments in center span
. To connect the cantilever ends of the center span
. To remove traveling carriages and temporary supports
. To connect the girder with the auxiliary piers.
. To cast pavement and set up fence etc.
A typical erection stage of one segment is described as follows:
. To move the traveling carriage forward and set up the form at proper elevations.
. To erect the stay cables, connect them with the traveler and partially stress them.
. To place reinforcement, post-tensioning bars and couple the stressed bars with
those of the previously completed segment.
. To cast in-situ the concrete
. To stress the stay cables and adjust the girder segments to proper elevations
. To cure the concrete of segment and stress the longitudinal and lateral bars and strands.
. To loosen the connection between the stay cables and the traveling carriages
. To stress the cable stays to their final length
The above erection step are repeated until the bridge is closed at middle span.
2.2 Detailed Construction Simulation in Site
Design final state of cable-stayed bridge under permanent loads
In the Panyu cable-stayed bridge, the final situation is defined as such a state that the
geometry of deck and pylons is a prescribed profile by designer and there is no or a little
bending forces in the pylons and the deck under permanent loads (including self-weight of
structure, pavement, fence etc.). Such a situation will reduce the second-order effects as well
as the time-dependent effects, such as creep and shrinkage of concrete etc.[1]
From the final bridge situation, the erection situations can be evaluated by a precise structural
analysis.
Software for simulation analysis
The erection analysis can be performed by using a conventional step-by-step method, such as
the forward assemblage analysis according to the construction process or inversely, the
backward dismantling analysis[3]. But for the cable-stayed bridges erected utilized cablesupported form carriages, the installation, remove and movement of carriages must be
carefully considered. A software which implement for the monitoring and adjustment of
construction procedure of Panyu bridge is specifically developed. In this software the
carriages are simulated as a part of the constructed structure and the behavior of carriage can
modeled efficiently and automatically.
Primary factors on construction such as the construction loads (weight of equipment and
traveling carriages, temporary loads etc.), effects of concrete creep and shrinkage, are
considered in detail. The software also implements the methods of forward assemblage and
the backward dismantling analysis.
Simulation of Erection Scheme for Panyu Cable-Stayed Bridge
The construction procedure described in Section 2.1 has been simulated stage by stage. Since
the creep and shrinkage of concrete occurs and the second part of the dead weight is loaded on
the bridge girder after completion of the structure, a downward displacement is induced.
Therefore, as the erection is just finished the elevation of the girder profile is set higher than
that of the design profile and the pylons is leaning toward the side spans. As for Panyu bridge,
the maximum value which is set higher than the designed profile in the middle of the bridge is
about 35.4cm, while the displacement of pylon top leaning to anchorage span is about 7.8 cm.
Through the simulation of construction process, a theoretical reference for each construction
stage is established. They include the information about elevations for laying form and for
cable tensions etc.
To give a precise modeling of all site operations, the volumes of concrete actually
implemented, and the temporary loads on the deck etc. need to be available. Then theoretical
calculations are performed at site for every stage. Creep effects are also taken into account.
And when there a need for slight modifications of the actual construction, the simulation of
actual construction process also provides the information on the adjustment of cable tensions
and laying forms.
3.
Monitoring and Adjustment
There are four sets of instrument installed on Panyu bridge to measure parameters as listed
below:
- elevations of the deck and displacement of the pylons.
- tensions of the cables
- stresses of the concerned section in the deck and pylons
- temperatures and gradients in the deck, pylons and cables.
3.1 Elevations of the deck and displacements of the pylons
Usually the elevations are measured at the previous three segments to monitor the
configuration of deck during the concrete casting. After the longitudinal strands and bars are
stressed and the stays are stressed to their final length in each segment, the elevations of deck
are measured at five previous segments of the front deck ends. And the displacements of
pylon ends are also measured. For every 5 segments have been completed, the geometry of
the completed structure are monitored. In order to eliminate the thermal effects as much as
possible, the measurement have been done at 6:30-7:00 a.m. before the sun rises.
Using the careful measurement can guarantee the geometry of bridge under the control of
engineers in site.
3.2 Cable Tensions
At Panyu bridge, the semi parallel wire cables (SPWC) with Hi-am anchorage are used. The
cable tensions are measured by frequency method. First the lowest 10 natural frequencies of
the cable are measured, then using the calibrating coefficient to evaluate the cable tension.
Each cable is calibrated after it is just installed and partially stressed.
Measurement sensors fixed on the five pairs of previous stays near the carriage enable to
establish the actual stay tension precisely correspond to those predicted by the design model.
Before and after the closure of side spans and main span, tensions of all the installed cables
are measured.
3.3 Temperature
There are six thermocouples in different parts of segments, eight thermocouples distributed
over the pylons, two thermocouples inserted in the cables. The average temperature
measurements are used in monitoring and the calculation of adjustment.
3.4 Stresses in pylons and deck
In several concerned sections of the deck and pylons, strain gauges are embedded into and the
strains of the structure are measured during the whole construction period. All strain
measurements, including the collection, storage and processing of measured data, are finished
by PC computers automatically. Generally the measurement was done once every 1 hours.
Thus the stress of the structure can be monitored.
3.5 Supplementary measurement of structural parameters
In an actual cable-stayed bridge, the discrepancies of parameter values between designed and
actual values such as the modulus of elasticity of concrete, the mass density of concrete, the
weight of girder segments may give rise to disagreements between the actual structural
response and the theoretical prediction.
At Panyu bridge, when the traveler carriage is lifted upon the deck, the reactions of each form
carriage are measured. Thus the weight of each carriage is obtained. At each stage the mass
density of concrete and the elasticity modulus of concrete are tested in laboratory in situ. And
the quantity of concrete and steel bar used in each segment are recorded and measured. All
the measured parameters are used in the calculation of construction and adjustment. Thus the
actual weight of deck are quite precise. The uncertainties of loads are diminished as much as
possible.
4.
Results of the Panyu cable-stayed bridge
4.1 Excellent agreement of the theoretical prediction and actual measurement

With the software described in Section 2 and the structural parameters evaluated by
measurement, detailed analyses are carried out during the construction processes. For every
segment, many plans to tension the stays which link with the form carriages temporarily are
calculated, and every concerned factors, such as the stresses in the deck and form carriage,
the cable tensions and the displacements of deck and pylons etc., are obtained. Thus the
engineers in site select an optimum scheme.
Table 1 shows the results from the theoretical prediction and measurement of actual structure
when the 14th deck segment is competed. The result of comparison shows good agreement. In
the practice, the agreement for every segment is very well. These provide with a solid
foundation to ensure the safety and smoothness of construction.
Cable
Cables Tension(kN)
Elevation of deck(m)
Number
Tm
Tp
Tm-Tp
Hm
Hp
Hm-Hp
S10
2865.0
2793.0
72.0
34.703
34.703
0.000
S11
2980.0
2948.0
32.0
34.518
34.519
0.001
S12
3241.0
3230.0
11.0
34.344
34.349
0.005
S13
3489.0
3503.0
-14.0
34.187
34.186
-0.001
S14
3681.0
3657.0
24.0
33.983
33.994
0.011
Cable
Cables Tension(kN)
Elevation of deck(m)
Number
Tm
Tp
Tm-Tp
Hm
Hp
Hm-Hp
M10
2634.0
2742.0
-98.0
37.408
37.407
-0.001
M11
2788.0
2862.0
-74.0
37.491
37.487
-0.004
M12
3072.0
3116.0
-44.0
37.552
37.551
-0.001
M13
3368.0
3308.0
60.0
37.617
37.623
0.006
M14
3506.0
3532.0
-26.0
37.693
37.702
0.009
In Table 1, T represents cable tension, H represents elevation.
sub-index m represents measured value, p does theoretical predicted ones.
Table 1.Comparison of the measured and predicted values (at 14th segment)
4.2 Final geometry within 4 cm of theoretical figures

After ten-month long erection of deck segments, the two ends of the cantilever parts meet at
the middle within only 3.9cm deviation in elevation. The errors in the axis of the deck are
limited to 1.5cm. The two ends of cantilever before the closure of side spans is well agreed
with the design requirement within error less than 2.0 cm.
The spatial position of a deck segment is given by reference to the previous segment; namely
by the relative geometry, there are no great errors in the successive segments. At every
segment the errors of laying form are limited at the prescribed construction tolerances. The
final profile of the deck is smooth in elevation. At each stage the position of the deck is in
good agreement with the design requirement (within 3.0 cm variation).
4.3 Cable tensions under control (<7%)
The problem of cable tension is more than just the accuracy of the readings of a pressure
sensor of the jack. When a cable is stressed to its final length, the tension of the cable is
obtained from the reading of the jacks which are calibrated every two months and from the
tension measurement system by frequency method. These results of tensions are corrected
each other and an exact tension value of a cable can be obtained.
As construction proceeded, the geometry of deck is monitored and the forecast final geometry
is compared with the final designed geometry. In every phase, the standard deviation of the
differences in each segment is less than 3.0 cm.
The convergence of the results from geometry monitoring and cable tension measurement
substantiates the conclusion that tension is correct to within better than 7%.
5.
Conclusion
During the erection of Panyu cable-stayed bridge, detailed simulation and continuous
monitoring of construction process has been carried out. These made it possible to achieve
high precision of construction. From the results of construction control practice of Panyu
cable-stayed bridge, some conclusion can be induced as follows:
1. In cable-stayed bridges, and particularly in PC cable-stayed bridge with relatively flexible
deck, the construction of the concrete cantilevers is complicated due to the use of the cablestayed form carriages, continuous geometrical monitoring is absolutely necessary in order to
obtain acceptable geometry and tension conditions for the structure.
2. This type of continuous monitoring enables the engineers to treat any errors that may arise
during the construction process and can make more suitable decision for the adjustment in
site.
3. Since there are many deviation in parameters, such as the mass density, elasticity modulus
of concrete etc., must be measured in the erection of each segment of deck. And the practice
of construction control for Panyu cable-stayed bridge shows that it is very important and
necessary to carry out a detailed simulation of construction process.
4. The good adjustment of the Panyu cable-stayed bridge is made possible by the co-operation
of all those involved. Such continuous monitoring and detailed simulation of erection process
make it possible to reach a high level of accuracy of construction in PC cable-stayed bridges.
References
[1] Virlogeux,M. Erection of cable-stayed bridges: the control of the desired geometry. Proc.
of the Seminar on cable-stayed and suspension bridges, Oct. 1994.
[2] Tang,M.C. The 40-Year Evolution of Cable-Stayed Bridges, in 1994 International
Symposium on Cable-Stayed Bridges, Lin Yuanpei et. al.(Editors) pp30-11,Shanghai.
[3] Walther,R., Houriet,B., Lsler,W., and Moia,P., Cable-stayed Bridges, Thomas Telford,
London, 1988.
Elevation
Plane
Fig.1 General View of Panyu Cable-Stayed
Some Aspects of The Design of Martwa Wisla River Bridge in Gdansk
Krzysztof WACHALSKI
Chief Engineer, Bridges
BPBK S.A.
Gdansk, Poland
Jacek KAMINSKI
Civil Engineer
BPBK S.A.
Gdansk, Poland
Marek SUDAK
Civil Engineer
BPBK S.A.
Gdansk, Poland
Computer visualisation
Summary
The intention of paper is to impart same of specific problems as regards designing cable-stayed
bridge in Gdansk, and also especially with seeking own computational methods and construction
details adequate to the Polish reality. We would like to pay special attention to the method of
concrete pylon computation and computational model of foundations as well as constructional
solutions of such details as support of vertical load variable sign (pressure and anchor) with
considerable horizontal displacement.
1.
Introduction
Polish economic changes of last years have caused the necessity of development of, among other
things, seaports. As far, goods transport out of The Seaport in Gdansk to the country interior has
been executed by train. In the current circumstances, a reconstruction of the port is connected
with the necessity of the road network reconstruction. In 1994 a project was started the aim of
which was efficiency increase of the communication system in the area of The Seaport in
Gdansk. The first element of the reconstruction is Crossing over the Martwa Wisla River.
The crossing of about 1 km length is to connect the port areas with the national road Warszawa
Gdansk. Assumptions concerning the new route localisation are based on the its connection
with highway A1 (European highway) to be executed in the future - the connection shall
perfectly pass by the city. Additionally, the designed crossing shall constitute an element of the
north-western ring of Gdansk. As it has been presented, the investment, necessary for the port,
appears profitable for the whole agglomeration of Gdansk.
The basic element of the designed crossing is a bridge over the Martwa Wisla River. Various
variants of the river crossing were considered, such as a typical bridge with supports in the river
current, drawbridge, cable - stayed bridge and tunnel.
During evaluation of the above - mentioned conceptions, a cable-stayed bridge was chosen.
Advantages of the conception were, among others, significant technical attractiveness,
a challenge for Polish engineers (it is one of the first bridges of that type to be built in Poland),
and finally economic justification, as the most important factor. Cable-stayed bridge enabled,
due to its small constructional height, shortening of the access roads to the bridge and
elimination of inclusion lanes within the bridge, which shall diminish its width. This is the
reason why the total cost of the whole investment with a cable - stayed bridge is not bigger than,
and in some circumstances it may appear smaller in relation to, a bridge of typical construction.
A very important factor as regards the advantages of a cable-stayed bridge is its foundation.
Because of significantly high costs of supports in the river current and very inconvenient
geologic conditions, it was advisable to use a construction of as few supports in the river current
as possible. Apart from the factor, whose importance should be evaluated and classified as equal
to that of economy (if not as more important), there are also aesthetic advantages. In the
localisation, in the area of entry to the city, a cable-stayed bridge, with its architecture, fits in
with the city panorama very well, constituting its characteristic element. The executed study of
architecture and landscape confirmed the conclusions.
Finally, General Directorate of Public Roads, as the representative of Polish Government
(Client), decided in favour of construction of a cable-stayed bridge over the Martwa Wisla River
in Gdansk. The project is financed by Polish Government and supported by financial credit from
World Bank. The technical and bidding documentation has been prepared in Municipal Design
Office (BPBK S.A.) in Gdansk. The beginning of the bridge construction is planned to be in the
middle of 1999 and its completion and putting the whole crossing into service at the end of 2001.
2.
General characteristics of bridge
The designed bridge is one-pylon construction of the main span length equal 230 m. Total length
of the bridge equals about 380 m. In cross-section, an composite bridge deck consists of two
girders and a reinforced concrete bridge deck slab of 23 cm thickness. The height of bridge deck
construction equals 2,39 m. Each steel girder consists of two welded plate girders. Crosswise, the
bridge deck is braced by steel cross-beam spaced every 4,0 and 4,33 m. Total width of the bridge
deck equals 20,31 m. The width includes two carriageways, two traffic lanes each, (2 x 7,0 m),
a reserve lane (median strip) and service footways, 0,75m each.
Cable stays system has been designed as semi-harp pattern, of dense type with two-sided
outside stays (two-planed). As stays, cables of parallel 7-wire strands, 15,5-mm diameter each
strand, are used. In the longer span, 15 stays have been used, spaced every 12 m. In the back
span 8 stays have been used, spaced every 13 m. Stays passive anchorage is placed in the pylon
and active anchorage in the bridge deck. The anchorage in the bridge deck has been designed in
girders axis (plane), which prevents from its torsion.
Bridge supports are founded on reinforced concrete drilling piles of 1500 and 1600 diameters.
The piles lengths equal from 25 to 30 m. The bridge main support - pylon, has been designed
with its total height of about 100 m and inside service. The pylon is A-shaped. In cross-section,
the pylon is a reinforced concrete box of 3,60 x 5,0 m dimensions.
In the areas of stays anchorage in the pylon, additional strengthening has been applied by means
of steel frames, which are to transfer great tensile forces.
The pylon has been put on a concrete block - footing of variable thickness from 3,5 to 7,0 m,
founded on 59 piles of 1600 mm diameter and 30 m length.
Bridge abutments have been designed as box abutments. The abutment on the side of the back
span has been fixed to the bridge deck and constitutes a fixed bridge support. The abutment on
the side of the main span is loaded with variable vertical reaction (lift up and pressure) and
considerable deck horizontal displacements. Additionally, the back span has been stabilised with
3 anchoring supports, spaced every 26 m. The bridge assembly as regards the ground part has
been designed as assembly on scaffolding and as regards the water part as using the cantilever
erection method. Assembly segments to be subsequently fixed using the cantilever method shall
be transported by waterway on pontoons. They consist of full composite section (steel girders
and reinforced concrete deck slab). The assembly segments length equals 12 m, and total weight
of a single segment equals about 200 T.
Characteristic indexes of the bridge construction are :
ratio of the bridge construction height and the main span length 1:96
ratio of the pylon height and the bridge stays length 0,46
ratio of the bridge deck width and the main span length 1:11,3
SCAFFOLDING ER ECTION
Figure no. 1 Side view
CANTILEVER ERECTION
SCAFFOLDING ER ECTION
Figure no. 2 Cross section of deck
A-A
B-B
Figure no. 3 View of pylon
3.
Computational model of pylon foundation
The bridge foundation geotechical parameters are not advantageous. In the vicinity, there are
poor grounds depositing to the depth of about 20 m below the area level. They are constituted by
interbedded aggregated mud and fine sands of loose consolidation. Only below the sands are
there consolidated medium and coarse sands. Because of the possibility of the base displacement
under the bridge main support, the pylon foundation flexibility was taken into consideration in
the computational analysis.
The bridge computational model includes substitute resultant elastic reactions, calculated for
each direction in space (3D) co-ordinate system and imposed at the place of the pylon rest on the
foundation block.
Calculation of the substitute elastic reactions-springs was executed according to the following
algorithm:
A single pile was modelled in an elastic space defined by elastic reactions. The reactions are
equivalents of ground layers of thickness above 0,25 m in the pile upper part to layers of 1 m
thickness at a pile base. The reactions corresponded to the vertical component - friction on
a side surface, and to two horizontal components - lateral passive earth pressure. Elastic
reactions also occurred as support in a pile base.
Loads of unit forces and moments on a pile head in all directions (two horizontal loads, one
vertical, three rotations) and the displacements obtained were the basis of rigidity
determination :
1
Ki =
where :
i degree of freedom (X, Y, Z, RX, RY, RZ)

i - displacements from a unit load
In such a way, substitute rigidities of each pile for all the other degrees of freedom (3D) in elastic
geometric space were obtained.
A model (3D) of the pylon base was created. It consisted of a concrete block supported by
elastic supports that corresponded to each foundation pile.
Unit loads were executed at the place where legs are rested on the base and the
corresponding displacements, according to the principle described above, were the basis of
calculation of substitute rigidities of the whole support in all directions (3D).
Figure no. 4 Model of pile and footing
The obtained substitute (resultant) rigidities of foundation enabled active consideration of

support displacement. The basis of estimation of ground elastic reactions-springs parameters
(as regards the pile model) were formulas specified by Polish Standards and Polish publications
concerning geotechnics.
The elastic reactions envelope obtained out of a static and dynamic analysis was used to load
the described models of a base and a pile and to calculate inner forces in them.
4.
Computation of stresses in pylon
The state of stresses in the pylon reinforced concrete construction, constituted by two-way
eccentric compression, conditions of Polish Standards, which require execution of strength
calculations according to the Linear Stresses Method, as well as the fact that pylon construction
is of extreme importance for the whole bridge resulted in preparation of special software for
computation of stresses in concrete and reinforcement steel for optionally shaped sections.
A principle was assumed that section modelling should consist in division into minute concrete
elements and particular reinforcement bars. For such an assumption, the problem was described
theoretically, an algorithm was made, and finally, a computer program was prepared. Popular
software for PCs, such as EXCEL 7,0 for WINDOWS95, with elements of VISUAL BASIC,
was applied here. Preparation of the algorithm was executed in co-operation with
Prof. K. Wysiatycki from Technical University of Gdansk. In the following lines, we present
theoretical bases and exemplary computational results.
Figure no. 5 Theoretical model
We have 3 equations of equilibrium :
w
x
y
ij
wij
= F
(1)
ij
w ij = F x o
(2)
ij
w ij = F y o
(3)
yr
Zij
yj
c
xi
xr
Zij = a xi + b yi + c
xr
y
b = 2r
2
r
r
2
2
2
r = xr + yr
The plane
(4)
a =
Z = ax + by + c
(5)
a=tg b=tg
Conditions for points (xi , yj) :
if
if
Zij > 0
Zij 0
then
then
Zij : = Zij
Zij : = 0
(6)
(7)
We assume comparative stress p and
wij = x x zij p (8)
Then we find equations solution (1) (2) (3) by use (4) and (8). Searching solution (1) (2) (3)
needs check below conditions :
if wij > F
if wij xi> F xr
if wij yj> F yr
then
then
then
reduce c parameter
reduce xr parameter
reduce yr parameter
The iteration method of change xr and yr parameters helps us to find solution automatically.
Figure no. 6 Stresses of lower part of pylon
Figure no. 7 Stresses of upper pylon part
5.
Analysis of horizontal force transfer from stay onto bridge deck
The analysis of horizontal forces distribution from stays, at the place of their anchors in the
bridge deck, was carried out by means of programme Robot V6, applying Finite Element
Method. Shell model was applied for the elements such as upper flanges, bottom flanges and
webs of the main girders and cross-beams. Plate model was applied for the reinforced concrete
slab. Connectors of the slab integration to upper flanges were assumed as strips, modelled also as
a shell. The conclusions drawn from the stress results analysis were used for dimensioning of
bridge deck elements. The most significant fact is that normal stresses, in the slab, in the nearest
area of a stay anchor, turned out to be about twice as big as uniformly imposed stresses. Such
assumptions were confirmed as regards angle of force distribution in the bridge deck slab, which
approximately equals 45o. Additionally, introduction of an axial force into the slab takes place
(by means of connectors) mainly in the nearest area, which totally takes up about 6 m (in front
and behind the point of application of the stay force).
Figure no. 8 MES model of deck
7.
Detail of bearing of variable reaction and great horizontal displacements
In order to transfer horizontal forces (reactions) of variable sign from the superstructure onto the
abutment on the main span side, a special bearing was constructed. The bearing consists of two
standard pot bearings and prestressing bars. Additionally, the bearing must ensure the possibility
of the bridge deck horizontal displacement, whose shift absolute value equals 450 mm. Because
of additional vertical tensioning of the pack of two pot bearings, the bearing is free from
knocking effect during a change of reaction sign. Optionally, a variant with a single pot bearing
and prestressing bars was considered. However, load capacity of such a bearing would be almost
twice as high, due to additional tensioning of the bearing against the whole tensing force.
Because of the above-mentioned great horizontal shift on support and relatively short
prestressing bars possible, the solution was rejected.
Figure no. 9 Abutment anchor bearing
8.
Conclusions
In this article, we wanted to popularise the project of cable-stayed bridge, the building of which
shall be soon started. The discussed issues constitute a small part of the solved designing
problems. These are, most probably, standard issues solved in the case of constructions of that
type. However, we aimed at explanation of the methods used in Polish reality. Among interesting
issues, there are :
1. Pylon foundation estimation method influence on static operation of the bridge superstructure
Support model with about 25000 freedom nodes, substituted by resultant rigidities during
superstructure analysis, enabled optimum (time and software) determination of inner forces
and stresses in the construction.
2. Computation of stresses in pylon concrete and reinforcement steel. A special computer
programme, based on popular and available personal computer (PC) software, was prepared.
3. Computation of stays anchorage details, by means of MES, enabled determination of precise
distribution of forces near anchorage.
9.
References
[1].
Detailed technical project of cable-stayed bridge over Martwa Wisla River, prepared by
BPBK S.A. Gdansk 1998.
[2].
Bridge over Martwa Wisla River in Gdansk Scientific Conference Gdansk-Jurata,

September 3-5, 1997.
[3].
Bridge over Martwa Wisla River in Gdansk Magazine Inzynieria i Budownictwo,

6/1998.
Rain/Wind Induced Vibrations of Parallel Stay Cables

Guy L. LAROSE
Technical Manager
Danish Maritime Institute
Lyngby, Denmark
Guy L. Larose, born 1961,
graduated in mech. eng. from
Laval Univ., Qubec, MESc
from Univ. of Western Ontario
and PhD from Tech. Univ. of
Denmark. He joined DMI in
1992 where he is a wind
engineering specialist.
Leif WAGNER SMITT

Chief Naval Architect
Danish Maritime Institute
Lyngby, Denmark
Leif Wagner Smitt, born 1939,
graduated in naval architect.,
MSc, from the Tech. Univ. of
Denmark. He joined DMI in
1964 where he is a renowned
specialist of physical modelling
of fluid-structure interactions.
Summary
The main findings of a series of wind-tunnel experiments focusing on the rain/wind induced
vibration phenomenon of stay cables of a large cable-stayed bridge are presented in this paper. In
particular, the phenomenon is studied for stay cables in tandem arrangement, one cable on top of
the other, where the effectiveness of an aerodynamic means to mitigate the excitation has been
verified.
1.
Introduction
Cable vibrations associated with a combination of wind and rain have been reported on several
occasions since the middle of the 1980s [1,2,3]. The lightly damped stay cables of large cablestayed bridge are particularly susceptible to these oscillations, especially when the steel strands
forming the stays are covered by a smooth synthetic tube, often polyethylene high-density
(PEHD), and when the cables form an angle of 20 to 30 with the horizontal plane. Remedial
measures have been devised, experimented and implemented on many structures, most of them
with great success and relatively low cost [2,4]. The excitation is, in general, weak and needs the
simultaneous combination of several parameters to occur. Some prefer to increase the modal
damping by a factor of five or so to damp out the vibrations [e.g. Erasmus Bridge [6]), others
prefer to mitigate the excitation at its source by aerodynamic means (e.g. several Japanese
bridges [4]) and others select a combination of both approaches (e.g. Pont de Normandie [7]).
The conditions at which these oscillations can occur for a cable-stayed bridge are well
determined and were pointed out as early as 1984 by the late Hikami [1]: smooth stay cable,
cable inclination of +20 to +45, wind yaw angle 20 to 40 directed towards the decreasing
cable height, cable natural frequencies between 0.4 and 3 Hz, mean wind speed between 8 to 15
m/s, light rain or mist, and high wettability of the cable surface. The vibrations are thought to be
caused mainly by the formation of water rivulets on the upper and lower surfaces of the cable.
The rivulets change the aerodynamic shape of the cable to an unstable shape, inducing acrosswind vibrations (lift) due to variations of the mean pressure distribution on the cable. The
problem is also amplified by the fact that the rivulet changes position as a function of wind speed
and with the cable motion. Also, the presence of secondary axial flows has also been observed
for inclined cables [5]. These axial flows can also create mean pressure distribution changes that
can induce vibrations for inclined cables (30 to 45 or so), even without rain.
A final element that adds a bit more
complexity to an already complex problem
is that, for the majority of the stay cables of
cable-stayed bridges, the transition between
sub-critical to critical Reynolds numbers
occurs in the 8-15 m/s range of wind speed,
depending on the cable surface roughness
and turbulence intensity. This transition
corresponds to an abrupt reduction of drag
force (50% or so) and a reduction of the
strength of the shedding of vortices in the
wake. The latter is equivalent to a reduction
of the loading mechanism associated with
fluctuating pressure. This reduction leaves
room for weaker excitation to dominate the
loading such as the excitation encountered
during the rain/wind-induced vibration
process. The presence of the water rivulet
or a slight change of surface roughness can
trigger this transition from sub-critical to
critical Reynolds number.
The resund High Bridge is a cable-stayed
bridge with a main span of 490 m flanked by
Figure 1: View of the cable system of the
side spans of 301 m each. The two-level
resund High Bridge during construction.
truss-girder bridge deck will carry vehicles
and train traffic and will be supported by two vertical cable planes anchored to two 204 m high
H-shaped pylons. The cable system has a harp configuration, each cable forming an angle of 30
with the bridge deck. The bridge has 40 stays per cable plane, each stay being composed of two
parallel cables placed one on top of the other with a 670 mm centre-to-centre spacing. The steel
strands of the stay cables are covered with a polyethylene high-density (PEHD) tube, 250 mm in
diameter. Fundamental natural frequencies of the stay cables will range from 0.5 Hz to 2.5 Hz.
Figure 1 shows a view of the cable system arrangement for the partially constructed bridge.
Figure 2: The 2.1 mm thick double helical fillet

fitted to the PEHD tubes of the resund Bridge
The combination of cable angle, polyethylene

surface, low natural frequencies and high
probability of occurrence of light rain with
moderate winds at the bridge site set the stage
for possible rain/wind-induced vibrations of
the stay cables. Based on experience, it was
decided at an early stage in the detailed design
of the superstructure to fit the PEHD tube with
an aerodynamic countermeasure to prevent
rain/wind-induced vibrations, namely a double
helical fillet, 2.1 mm high (see Figure 2),
similar to the fillet used for the stay cables of the Pont de Normandie in France, with the same
pitch but the fillet is 0.7 mm higher. To verify the effectiveness of the proposed countermeasure
for 250 mm diameter cables (the Pont de Normandie cables have a diameter of 160 mm) in a
tandem configuration, a series of wind-tunnel tests was initiated by Sundlink Contractors and
carried out by the Danish Maritime Institute (DMI).
2.
Experimental Procedures
2.1
Dynamic Test Rig
A 6 m long section model of a stay

was built at a geometric scale of
1:1 and was mounted in a
purposely designed test rig fitted
with suspension springs. The rig
was designed such that only one of
the cables of a pair could oscillate
while the other, when present, was
kept fixed and only acted as a
dummy to simulate adequately the
surrounding flow field. All windtunnel tests were carried out in
the Velux Wind Tunnel in
stbirk, Denmark, which has a 4
m x 4 m open jet cross-section, a
Figure 3: Test rig in A/S Velux Wind Tunnel
30 m/s maximum wind speed and
a rain facility. A view of the test rig in the wind tunnel is shown in Figure 3 and a sketch of the
rig general arrangement and sign convention is given in Figure 4.
2.2
Scope of Wind Tunnel Tests
The parameters investigated during the

rain/wind vibration tests were:
the influence of the tandem cable

configuration on the vibrations;
the influence of wind incidence, 40 in
the horizontal plane;
the influence of wind speed and rain
intensity; and,
the influence of structural damping and
frequency of oscillations.
Initially, the test programme focused on the

reproduction of rain/wind-induced vibrations
observed elsewhere for an isolated smooth
PEHD tube forming an angle of 30 with the
horizontal plane in yawed winds and light
Figure 4: Sketch of test rig and sign convention
rain. This was followed by a series of tests aimed at defining a systematic test procedure including
surface treatment of the PEHD tube. The test procedure was applied for a series of exploratory tests
where the worst case conditions were sought for the cable fitted with the helical fillet. Finally, tests
aimed at comparing the level of aerodynamic damping between a dry and a wet cable with rivulet
were performed for various cable configurations and various levels of structural damping.
2.3
Cable Surface Treatment
Past experience with the simulation of rain/wind-induced vibrations in wind tunnels had shown
that the wettability of the cable specimen was an important test parameter. In nature, a
combination of dust accumulation, saline deposit, acid rain or the like, and sunlight increases the
wettability of the PEHD tube surface with time. It was also observed that the more wettable the
cables are, the larger the possibility of formation of a water rivulet (upper and / or lower) on the
cable and therefore possible rain/wind-induced vibrations. Based on this it can be conclude that
black PEHD tube are more susceptible to rain/wind excitation than white tubes since they have a
larger proportion of carbon, an element that happens to increase the wettability.
Since the PEHD tube supplied for the experiments were brand new, a wetting agent had to be
applied to the surface to simulate the prototype conditions. For the Pont de Normandie
experiments, soot from an oil furnace mixed with water proved to be a very effective wetting
agent. For the present investigations a different approach was used. Firstly, the PEHD tubes
were lightly sanded with a fine grade sandpaper to simulate natural erosion and dust particles.
Subsequently, the cable surface was treated with a thin coat of polyvinyl alcohol simulating a rise
in surface energy of the cable
equivalent to oxidation. This surface
treatment increased the apparent
wetting characteristics of the PEHD
tube, helping the formation of the
upper rivulet.
Figure 5 compares the flow patterns
on a smooth untreated PEHD tube to
the flow patterns on a treated tube.
The polyvinyl alcohol used in this
study was Gosenol KP-06 from
Nippon Gosei applied with a brush
from a 5% solution in ethanol. The
polyvinyl alcohol coating was applied
systematically after about two hours of
testing.
2.4
Figure 5: View of lower rivulet on a new PEHD

tube (top) compared to flow pattern for a PEHD
tube after treatment with polyvinyl alcohol (bottom)
Scaling Parameters
The aerodynamic phenomena studied here is due to a combination of rain running down on the
PEHD tube covering the stay cables and wind with low turbulence intensity in the 8 to 15 m/s
range. The difficulty in scaling rain droplets and the complexity of the phenomenon, which is
surely dependent on Reynolds number, suggests that the only advisable scale for geometrical
scaling is 1:1. The same remark prevailed for the choice of the velocity scaling, the water
patterns on the cable being a function of the volume of rain droplets being carried by the wind.
A velocity scale of 1:1 was thus selected, implying a frequency scale of 1:1 also.
An advantage of the 1:1 geometrical scale is that the PEHD tubes fabricated for the prototype
structure could directly be used for the model cable. However, to ensure an adequate aspect ratio
of the model (length of the model divided by its diameter) a cable model of at least 6 m (6 / 0.25
= 24) should be used. This forced the wind-tunnel tests to be conducted in a very large wind
tunnel, namely the 4 m x 4 m Open Jet Wind Tunnel of the window manufacturer Velux A/S in
stbirk, Denmark.
The mass per unit length of the prototype cables is on average 80 kg/m. For a 6 m long model
cable this would mean a mass of 480 kg. For practical reasons, the mass scaling could not be
kept at 1:1. The model cable, including the springs and the diverse fixing components had a total
mass of 84 kg, resulting in a mass scaling of 1:5.7.
It can be assumed that the mass damping parameter:
4 m
,
D2
(1)
where =air density, D = cable diameter, m = cable mass per unit length and = cable damping
as a fraction of critical, governs the dynamic modelling of wind-induced vibrations of cables.
Based on this parameter, the lower mass of the cable model can thus be compensated for by a
higher modal structural damping in comparison with the expected structural damping of the
prototype cable. This links directly the mass scaling, 1:5.7, to the damping scaling, 5.7:1.
However, to take into account unavoidable three-dimensional effects due to the flow passing by
the extremities of the finite length model cable, the damping scaling was corrected. Since, the 3D effects are likely to render things better in the wind-tunnel than for the prototype cables, a
damping scaling of 3.8:1 was adopted. This means that a model structural damping of 0.6% of
critical would be equivalent for the prototype to a structural damping of approximately 0.16% of
critical.
The main scaling parameters were thus as follows: geometry, velocity, frequency and density,
1:1; mass, 1:5.7 and damping, 3.8:1.
2.5
Modelling of the Wind and Rain
The 4 m x 4 m open-jet of the Velux

Wind Tunnel provided a relatively
smooth air flow with a mean
turbulence intensity of 1% measured
0.2 m downstream of the inlet. Zones
of higher turbulence intensity
(approximately 2%) are found at the
edges of the inlet while the lowest
turbulence zones are near the centre of
the cross-section. The turbulence
intensity increased further downstream
but was believed to be in the lower
range of turbulence intensity expected
for prototype cables. The room
Figure 6: View of medium intensity rain and rain rig

in the Velux Wind Tunnel.
around the jet is 7.5 m wide by 7.5 m high. These flow features made the Velux Wind Tunnel
well suited for rain/wind induced vibration experiments. The wind tunnel is also equipped with a
rain generation facility, where demineralised water is supplied under pressure to an array of
adjustable nozzles to provide a wide variety of rain intensity and droplet sizes.
For the present experiments, a rain rig was purposely built by DMI to provide a medium to light
rain in a plane following the cable inclination of 30 and parallel to the cable. At first, the rain
rig was composed of two tubes with 10 nozzles
each. This proved to provide rain with too high
intensity (see Figure 6), even though the water
supply pressure was reduced to a minimum
(flow rate of 3.5 litre/min). After several
iterations, the optimum rain rig was composed
of only one tube with 8 spray nozzles (Figure
7).
3.
Main Findings
3.1
Smooth Cable
Figure 7: View of the light rain rig in the

wind tunnel.
Rain/wind-induced vibrations of a smooth

PEHD tube, 250 mm in diameter were observed for an angle of wind incidence of +30 and wind
speeds varying between 9 and 12 m/s. The vibrations developed rapidly, within a few cycles, up to
250 mm, after the formation a small coherent rivulet on the upper and lower surfaces of the tube.
There was no apparent along-wind motion of the rivulets. The large vibrations started after an
equivalent medium intensity rainfall had wet the cable and was stopped for approximately one
minute. The structural damping of the cable was very low, 0.025% of critical.
Large rain/wind vibrations (up to 250 mm) were also observed for smooth cables in a tandem
configuration (670 mm cable spacing). The damping level was adjusted so that the amount of
energy dissipated per cycle for the experiments was equivalent to the prototype conditions,
assuming a prototype structural damping of 0.16% of critical. Subsequently, it was observed that
an increase of structural damping up to an equivalent prototype damping of 0.58% of critical was
not sufficient to damp out completely the rain/wind-induced vibrations for smooth cables in
tandem configuration as seen in Figure 8.
100
50
-50
-100
50
100
150
200
250
Time (sec)
Figure 8: Time histories of lower cable displacements due to rain/wind excitation for smooth
cables in tandem configuration and structural damping of 0.58% of critical.
3.2
Cable With Helical Fillet
The tests conducted with the PEHD tube with a double helical fillet, 2.1 mm thick, showed a strong
reduction of the rain/wind-induced vibrations observed with the smooth tube. The helical fillet
disrupted the formation of a coherent upper rivulet, therefore mitigating the excitation at its source.
These observations are in accordance with the results of the wind tunnel investigations made for the
stay cables of the Pont de Normandie. The helical fillet proved to be effective in reducing the large
rain/wind induced oscillations even for the cases where the structural damping of the model was as
low as 0.025% of critical.
Exploratory tests indicated that the helical fillet was as efficient for a wind angle of +30 as for its
mirror image configuration at -30. The tests indicated also that the process was only slightly
affected by the frequency of oscillations of the cable, the helical fillet being as effective at 1.2 Hz as
at 0.66 Hz.
For some conditions, the rain/wind-induced excitation persisted even with the cable fitted with the
helical fillets. The amplitudes of vibrations were limited, however, when compared to the results of
the smooth PEHD tube tests. The worst cases observed were: 1) +30, 12 m/s; and 2) +20, 11 m/s,
(see Figure 9). In both cases, the cable was in its tandem configuration and oscillated at a frequency
of 0.66 Hz. It is believed that the lower rivulet, which remains coherent even with the helical fillet
in place, causes these oscillations.
50
-50 11 m/s, dry tube

0
20
40
60
80
100
120
140
160
180
60
80
100
Time (sec)
120
140
160
180
50
-50 11 m/s, tube wet, slowly drying

0
20
40
Figure 9: Comparisons between time histories of vertical displacement of a dry cable versus a wet
cable at 11 m/s for a wind yawed angle of +20, in tandem configuration, with helical fillet.
It was observed that the cable immersed in a steady-state rain field of light to medium intensity had
a tendency to lift. This lift was combined with an apparent damping of the buffeting vibrations
when the intensity of the rain reached medium level. This can be observed in Figure 10 where a
mean cable lift of 20 mm was recorded at a 12 m/s wind speed.
40
20
0
-20
-40
-60
12 m/s, no rain
0
20
40
60
80
100
120
140
160
180
60
80
100
Time (sec)
120
140
160
180
40
20
0
-20
-40
-60
12 m/s, with medium rain

0
20
40
Figure 10: Effect of rain on vertical displacement of the cable configuration for the resund High
Bridge, 12 m/s and wind angle of +30, in tandem configuration, with helical fillet.
Figure 11 compares decay traces at 12 m/s wind speed for a dry cable to a slowly drying cable
where a distinct lower rivulet is present. For these tests, the cables were excited by hand and
released. It clearly indicates that the total damping is reduced when the cable has been exposed to
rain, the decay traces being shorter for a dry cable.
100
12 m/s, dry tube, hand excitations
50
0
-50
-100
120
100
140
160
180
200
220
240
260
280
300
200
220
Time (sec)
240
260
280
300
12 m/s, slowly drying tube, hand excitations
50
0
-50
-100
120
140
160
180
Figure 11: Comparisons between decay traces of a dry cable versus a slowly drying cable with
lower rivulet and helical fillet, 12 m/s and wind angle of +30.
3.3
Effect of Added Structural Damping on the Rain/Wind-Induced Vibrations
A damping device composed of a plunger

oscillating in a silicone oil bath was built and
fitted to the rain/wind cable vibration rig (see
Figure 12). An extremity of the plunger was
fixed directly to the moving cable while the other
extremity was fitted with an array of 2 to 6 pins;
the number of pins being varied to change the
level of damping. A damping device was fitted
at each end of the cable.
Figure 12: Silicon oil damper
In addition to the inherent damping level of the
rig without damper (0.49% of critical), three
other levels of damping were achieved, namely 1.31%, 1.62% and 1.94%. The damping referred
to here and later on in this text is the mean damping value of the model cable for oscillations
between 60 mm and 20 mm in amplitude. While the damping was found to vary only slightly
with amplitude for the tests in still air without rain, the tests with wind and rain showed an
important variation of total damping with amplitude.
The effect of increasing the damping from 0.49% to 1.31% of critical in model scale is depicted
in Figure 13 for the cable with helical fillet, at 12 m/s wind speed, with rain. In general, the
amplitudes of vibrations were slightly less than halved by this 2.6 increase of damping. For all
the tests with increased structural damping, no self-induced vibrations due to the rain/wind
combination were observed. Based on mass damping similitude, 1.31% of critical in model scale
is equivalent to 0.33% of critical in full-scale.
40
20
0
-20
40
no damping added, 0.49% of crit. in still air, model

50
60
70
80
90
100
110
120
130
140
90
Time (sec)
100
110
120
130
140
40
20
0
-20
40
with damper, 1.31% of crit. in still air, model

50
60
70
80
Figure 13: Comparisons between time histories of vertical displacements of cables in tandem for
two levels of structural damping, with rain and 12 m/s wind speed.
To study the influence of the structural
damping on the rain/wind-induced
vibration phenomenon, aerodynamic
damping tests were carried out. The tests
were performed systematically for four
levels of structural damping for the cable
with helical fillet, in tandem
arrangement, with a +30 azimuth angle,
with and without rain.
1.6
1.2
0.8
0.4
0
-0.4
Total damp. at 12 m/s w/o rain - struct. damping in still air

Total damp. at 12 m/s w/ rain - total damp. at 12 m/s w/o rain
-0.8
The results are summarised in Figure 14.

-1.2
At 12 m/s without rain, an important
-1.6
contribution to the total damping was
0
0.5
1
1.5
2
Structural damping (% of critical)
attributed to the aerodynamic damping.
The decay tests with 12 m/s wind speed
Figure 14: Variations of aerodynamic damping for
and rain showed a drop of total damping
dry and slowly drying cables for four levels of
generally 2 or 3 minutes after the rain had structural damping.
been stopped. This drop in total damping
typically corresponded to the level of aerodynamic damping at 12 m/s for a dry cable.
From this observation, it can be concluded that the presence of the lower rivulet transforms the
aerodynamic shape of the cable so that a lift force (upward) is created. This lift force varies as a
function of the apparent angle of attack of the wind when the cable is oscillating vertically. This
variation of lift appears to annihilate the drag-induced lift that is the source of the aerodynamic
damping.
4.
Conclusions
Rain/ windinduced vibrations of a stay cable were reproduced in the laboratory for a single
cable and for cables in tandem configuration. The vibrations were mitigated either by using an
aerodynamic appendage to the cable, namely a 2.1 mm double helical fillet, or by increasing the
level of modal damping. No marked differences were observed between the single and tandem
cable configuration with regards to rain/wind vibrations. If anything, the propensity of
developing axial flow might be larger for the tandem configuration, increasing the possibility of
vibrations.
Acknowledgements
Sundlink Contractors, the constructors of the resund Bridges, commissioned this wind-tunnel
study. Their support and permission to write this paper is greatly acknowledged.
References
[1] Hikami Y. & N. Shiraishi, Rain-wind vibrations of cables in cable-stayed bridges, J. of
Wind Engineering. and Industrial Aerodynamics, 29 (1988), 409-419.
[2] Langsoe H. E. & O.D. Larsen, Generating mechanisms for cable stay oscillations at the
Faroe Bridges, in Proc. of Intll Conference on Cable-stayed Bridges, Bangkok, Nov.
1987, pp. 1023-1033.
[3] Yoshimura T., T. Tanaka, N. Sasaki, S. Nakatani and S. Higa, Rain-wind induced vibration
of the cables of the Aratsu Bridge, in Proc. of 10th Japanese National Conf. On Wind
Engineering, Tokyo 1988, pp 127-132.
[4] Matsumoto M., K. Yokoyama, T. Miyata, Y. Fujino and H. Yamaguchi, Wind-induced
vibrations of cable-stayed bridges in Japan, in Proc. of Canada-Japan Workshop on
Bridge Aerodynamics, Ottawa, Sept. 1989, pp. 101-110.
[5] Matsumoto M, C.W. Knisely, N. Shiraishi, M. Kitazawa and T. Saito, Inclined cable
aerodynamics, in Proc. of 1989 Structures Congress, San Francisco, May, pp. 81-90.
[6] Geurts C.P.W., A.C.W.M. Vrouwenvelder, P.C. van Staalduinen & J.H. Reusink,
Numerical modelling of rain-wind induced vibrations: Erasmus Bridge in Rotterdam,
Structural Engineering International, Vol. 8, No. 2, May 1998.
[7] Flammand O., Rain-wind induced vibration of cables, J. of Wind Engineering. and
Industrial Aerodynamics, 57 (1995), 353-362.
Swietokrzyski Bridge, Warsaw

Pekka PULKKINEN
M.Sc. Civil Eng.
MESTRA Engineering Ltd
Helsinki, Finland
Pekka Pulkkinen, born 1955,

received his Civil Eng.
degree at the University of
Oulu 1980.
Summary
A bridge competition was organised 1997 to find technically the most innovative and progressive
bridge solution over the river Wisla in Warsaw. A single pylon cable-stayed bridge proposed by
Finnish and Polish designers was the winner of the competition. Many modern technical solutions
were proposed in this bridge. The cable anchorage structures as well as the cross section of the
superstructure have been designed in most economic and effective way. The aesthetics of the bridge
was investigated very thoroughly.
1.
Introduction
The bridge is located in the heart of the city of Warsaw. The new bridge will be built just beside the
existing bridge named Syreny bridge. The old bridge will be demolished after the new bridge is
opened. The bridge will give a new outlook for the city and river banks. The building of the new
bridge is a part of the bigger building project to improve the traffic conditions in Warsaw.
The bridge is a cable stayed bridge of composite construction. The cable spans are 180 and 140
metres in length. The total length of the bridge is 448 metres. The total effective width of the bridge
deck is 29.8 metres, consisting of four traffic lanes and bicycle and pedestrian lanes on both sides
of the bridge deck.
The bridge will be constructed in extremely short time, during 1998-2000.
Fig 1. Elevation
2. Aesthetics of the bridge

During the competition phase many bridge types were studied. In order to fulfill all technical and
aesthetic requirements set in competition documents a cable stayed bridge with single pylon was
chosen as a final bridge type.
Raising of the street level at bridge site was limited because the new bridge is a part of existing
street connection. On the other hand the size of the navigation channel was determined to be
relatively large. These reasons caused that the bridge should have quite a long main span and a
slender superstructure.
When considering the type of the pylon A-shape seemed to be superior. It gave good possibilities to
design the shape of the pylon to be a land mark of the river crossing. At the top of the pylon there is
a black cover plate which illustrates a key of piano. This reminds people of the remarkable musical
background of Warsaw. The colour of the pylon will be light or natural white.
The cable anchorages on the deck level are placed between the traffic and bicycle lanes. When
approaching the bridge the passengers have a possibility to see the cables and the pylon fluently.
Close to the pylon legs there are extensions of the deck designed to be places for rest and viewing
the river banks. The rest stops are equipped with benches and they are covered by shelters made of
steel.
Fig 2. The pylon
3.
Superstructure
The superstructure of the main bridge is a composite steel - concrete girder. In cable spans the cross
section has two main longitudinal steel beams at the distance of 19.0 metres. The deck slab is a
reinforced cast in situ concrete slab. The distance between cross beams is 5.0 metres. The cross
beams have cantilevers outside the main girders.
The aerodynamic behaviour of the bridge was analysed in the conceptual design phase. The ratio
between height and width of the cross section is only 0.08. In order to improve the ratio of natural
frequencies of bending and torsion the weight of the cantilever slab and the height of the edge beam
were reduced.
In order to get more stiffness to the cross section outside the cable-stayed part two additional beams
have been placed to the cross section. Four beams are also used at the negative moment area at the
ballast abutment. The total amount of structural steel in cross section is only 180 kg/ m2, which is a
relatively small amount.
The superstructure is fixed to the ballast abutment. The uplift force at the abutment is balanced by a
foundation slab and earth filling. There is only one expansion joint in the main bridge.
The steel superstructure will be installed by launching. Launching will be carried out by using
temporary supports at the main spans. The launching is proposed to be executed from both sides of
the pylon. The concrete deck slab will be cast in 20 metres long sections.
Fig 3. Cross section of cable spans

The cable forces are anchored directly to the webs of the main steel girders. The anchorage
structure is simple and consists of stiffened steel web and guide pipe. The locations of cable
anchorages dont affect the spacing of cross beams. The stressing of cables will be carried out at the
pylon top, therefore the space needed for cable anchors is minimised on the deck level.
Forces due to eccentricities of guide pipes are eliminated by using short external centring pipes,
which are installed after the stressing of cables.
Fig 4. Cable anchorage to the main girders
4.
Pylon and cables
The A-pylon is a 87.5 metres high concrete tower. The cross section of tower legs is hollow with a
hole of 1.25 metres for maintenance purposes. In order to get smaller inclination in legs they are
forced to penetrate the deck slab at the pylon. The bearings of superstructure have been placed on
external cantilevers.
The cable anchorages at the pylon top will be fixed to the concrete structure, there will be no steel
boxes inside the tower. The cables are anchored to the concrete tower by penetrating cable guide
pipes through the tower. In the crossing area the cables from Warsaw city side are placed to the
inner side of the pylon.
The pylons will be constructed by using climbing formwork. Each casting section will be about 4.2
metres in height.
The stay cables consist of high quality parallel wires, which are protected against corrosion with
hot-dip galvanising, grease and HDPE pipes. The colour of cables is white or light colour. The
cable installation and stressing work will be made after the casting of the deck slab. The cables will
be stressed at the pylon top.
Fig 5. The principle of cable anchorage at the pylon
5.
Foundations
The ground conditions of the bridge site are interesting. At first there are fills on both sides of the
river. The filling material is mainly coarse; sand, gravel, concrete etc. Under the fills and the river
base there is a sedimentary layer of sand. The material varies from fine sand to gravelly sand. The
layer thickness changes a lot. The base layer consists of hard plastic cohesive material down to 100
metres depth. The soil is mainly clay and sandy clay.
All supports are founded on cast in situ bored piles. The piles act partly as end-bearing and partly
as cohesion piles. The diameter of piles is 1.5 metres, except at ballast abutment where the diameter
is 1.2 metres. Raked piles are used for ballast forces, collision loadings and for launching forces
during construction. Vertical piles are used only at pylon and abutment S1.
Because of the soil conditions, a lot of attention has been given to the settlements of the pile
foundations. FEM-analyses have been made to determine total settlements and deformations during
the construction period. Full scale test loading of piles will be implemented during the piling works.
Fig 6. Swietokrzyski Bridge
The Design of the Zwolle Cable-stayed Bridge

- Integrating Engineering with Aesthetics
Robin SHAM
Technical Director
Maunsell Ltd
Beckenham, UK
Robin Sham, born 1954,
received his BSc in 1978
(Birmingham)
and PhD (Structural
Engineering)
in 1989 (Imperial
College). He is
Technical Director
responsible for Bridges &
Special Structures and was
Maunsells Project
Engineer for Zwolle
Bridge.
Arie MONSTER
Grontmij Consul.Eng.
De Bilt, The Netherlands
Arie Monster, born 1946,
BSc in Civil Engineering.
He works for the Structures
for Roads & Waterways
Department and is an
experienced Project
Manager for all types of
bridges, viaducts and
tunnels. He was Project
Manager for the Zwolle
Bridge
Introduction
Between Stadshagen and Zwolle in the
Netherlands, a landmark bridge now graces the
environs of Zwarte Water and forms a subject for
study on bridge aesthetics. As construction
progressed the striking profile of the Zwolle Bridge
emerged from Zwarte Water and captivated the
admiration of the local residents and visitors alike.
At dawn and in the golden sunset, the scene was
one of the most spectacular of all bridge sites.
Fig 1 Zwolle Bridge, The Netherlands

The Zwolle Bridge is an asymmetrical cable-stayed bridge with a single main span of some 56m
and a continuous east approach span of 25m. The project also consists of a west approach viaduct
and a bascule bridge. The steel bascule span closes the 18m gap between the cable-stayed main
bridge and the west approach viaduct. The superstructure of the cable-stayed bridge consists of
twin longitudinal spine beams 1000mm deep, with a concrete slab varying in depth from 250mm
to 330mm, and cross girders at typically 4375mm centres. Longitudinal bending, shear and axial
compression are primarily resisted by the twin spine beams and top flange. Transverse actions
between the cable planes are resisted by the stiffening cross girders. The superstructure is
monolithic with the bascule chamber, which forms the substructure to the pylon, and is
continuous over the intermediate pier in the east approach span.
The unique elegance of the Zwolle Bridge has been instrumental in the marketing of the
Stadshagen development area. It has received extensive coverage in the regional and national
press which has increased the profile of the development area. Local residents have shown great
interest and pride in their new structure, especially as it is the first asymmetrical cable-stay
structure in the whole of the Netherlands.
Design
Consideration of bridge architecture dictates that the pylons are
located as close to the Stadshagen side of Zwarte Water as
possible, to compliment two apartment development areas.
Clarity, light, space and water are the key elements considered in
the architectural design of the structure. The shape of the pylons
is architecturally unique and brilliant. The flow of forces are
well communicated by the shape and form. The rationale for the
overall aesthetics is motivated by the desire to maximise the
intensity of light that can be cast on the pylons - leaf shafts which
represent masts are therefore adopted to maximise the surface
area while providing a sufficient horizontal cross section to
enable column action. Careful consideration has been given to
the projection of light on the structure, including shadow during
Fig 2 - Effect of Light &
daylight and illumination at night. Each leaf shaft is 17 metres
wide, tapering to a knife blade-like point at the top. A number of Shadow
options for the curved profile would be feasible - circular, polygonal and parabolic. A parabolic
profile was adopted by virtue of the smooth flow which it introduces to the system.
To improve the stability of the asymmetrical cable-stayed
superstructure, the weight of the main span is counterbalanced by
inclining the pylons backwards. This combines structural efficiency
and aesthetics. The self weight of the pylons are not sufficient to
balance the overturning moment from the deck. To compliment this
and thereby to control the tension in the near face, each pylon shaft
is prestressed vertically from the base of the Bascule chamber to a
height of 22.5m above deck level. The cellar was designed such that
the cables could be drawn up the pylons from within the Bascule
chamber. Prior to cable installation the prestress in the pylons are
inefficiently positioned. The prestress adds to the tension on the
front face.
The centre of action of the forces on the bascule chamber lies
outside and behind the footprint of the bascule chamber. The leaf
Fig 3 - Stillness versus
shafts would normally have been continued down to pile cap level
Motion
had it not been for the need for a bascule chamber for
accommodating the electrical and mechanical installations for the adjacent bascule bridge. The
pylon shafts are cast monolithically with the bascule chamber which thereby reduces their
effective buckling length. The need for a bascule chamber is therefore exploited in the design to
optimise functional requirements and structural efficiency.
If the pylons lean forwards, they will still be aesthetically interesting although structurally
inefficient in terms of the counterbalance which they provide for the weight of the main span. In
the conceptual development of Zwolle Bridge, it is logical to consider rotating the pylons
backwards, away from the navigation span. As the pylons become vertical, the centroid of the
pylon stems shift longitudinally outside the main span. Further rotation backwards instigates a
counterweight action whereby the self-weight of the pylons assists in stabilising the
superstructure system. A ground investigation showed that the site consists primarily of wellcompacted sands. However, layers of soft silt are found in the river channel. During the design,
the possibility of adopting a spread footing for the foundation of the bascule chamber was
examined. It was judged and then proven by analysis that the behaviour of the cable-stayed
bridge would be sensitive to stiffness of the foundation. A piled foundation was therefore
chosen. The piles are in tension during construction and this condition determines their length.
In order to achieve a clean profile, the twin leaf shafts are designed to
remain stable without the need for cross bracing. The pylon shafts are
at a maximum height of 43.35m above deck level and 50m above
mean water level. The rear face of each pylon shaft is notched to
accommodate the bascule steel span in its opened position, thus
completing a smooth pylon outline whenever the bascule span is
raised. The notches add further character to the profile and provide
visual relief where it is warranted.
The cross section of the superstructure,
as well as the highway layout over the
deck, are asymmetrical. The design
allows for future widening of the deck
Fig 4 - Rear Face of
on one side to accommodate a revised
Pylon Shaft Notched to
highway cross section. (Should this be
Fit Bascule Span
required with further development of
the Stadshagen area). The lateral position of the pylon shafts has
been configured to permit deck widening. The final concept is
one in which the shafts are inboard of the edges of the running
deck. If the shafts were located outboard of the deck, the need
for additional clearance to allow for deck widening would have
resulted in an increased horizontal separation of the shafts. This
would risk the introduction of horizontal bracing for stability
against buckling and thereby ruin the clean aesthetics of the
pylon design.
Fig 5 - Tour de Force
There are five cables to each pylon shaft and each cable is
threaded obliquely through, in deviator pipes, and anchored at
recesses in the rear face to permit access to cable anchors for
inspection, maintenance and potential re-stressing. At the lower
end access would be more difficult as the anchorages will be
over Zwarte Water. Here the cables are anchored on the
underside of the deck slab, outside the longitudinal spine beam
girders, at every second cross girder. The in situ concrete
anchorages are functional as well as being a deliberate
architectural statement. They are integrated with the grillage
beam system, thus accommodating the geometric variation and
injecting the cable loads directly into the main beams.
Construction
Construction of the 17m-long by 15m wide bascule chamber
commenced through pouring a 1m-thick concrete slab under
water.
Fig 6 - Variation of
Geometry with Viewpoint
Fig 7 - Cable Anchorage - A Deliberate Statement Fig 8 - The Bascule Chamber under Construction
A further 2.3m was added to form the reinforced concrete base slab. The bascule camber is some
10.5m high, with side walls rising to form the monolithic base of the pylon shafts.
Pylon construction commenced in September 1997 and was completed by the end of the year.
The pylon shafts were constructed in 4.5m lifts at an average rate of one pour per pylon per
week. Once construction had reached 22.5m above deck level each shaft was prestressed. Pylon
construction was completed after grouting of the prestressing tendons.
The lift size was influenced by the architectural imprint
which appears on the pylon leaves. Aesthetics
considerations also dictated the layout of the shuttering
boards - to ensure that the pylon faces are embossed with a
particular quilt like pattern. The pylon was constructed
using climbing falsework which moved up the pylons as
the construction progressed.
All the cable stays
appear to be the same
size despite the actual
number of strands they
contain. The visually
smooth appearance of
the cables is enhanced
by providing protective
casings over cable
correction details, which
eliminate sharp changes Fig 9 - Construction of Pylon Shafts
in profile.
Fig 10 - Embossed Pylon Face
The construction of the superstructure was investigated and

carefully controlled by an erection sequence analysis. Over
half of the 56m cable-stayed span, to a point just beyond the
third cable, was constructed on falsework. After three cables
were fixed from each pylon shaft to the edge of the deck, the
falsework was then removed.
Zwolle bridge was officially opened on 19 September 1998 and it has added to the worlds
repertoire of landmark structures.
Fig 11 Superstructure under Construction
Fig 12 - Maximising the Intensity of Light
Acknowledgement
Client:
Consultant:
City of Zwolle, the Netherlands

Grontmij Traffic & Infrastructure, The
Netherlands
Cable-Stayed Bridge Sub-Consultant:
Maunsell Limited, United Kingdom
Architect:
Maarten Struijs, Gemeentewerken
Rotterdam, The Netherlands
Fig 13 - Combining
Functionality with Aesthetics
Accuracy Control On the Construction of Tatara Bridge

Yasuhito MANABE
Mukaishima Office, HonshuShikoku Bridge Authority,
Hiroshima, Japan
Nobuyuki HIRAHARA
Tokyo Office, HonshuShikoku Bridge Authority
Tokyo, Japan
Nobuo MUKASA
Mitsubishi Heavy Ind. Co.Ltd
Hiroshima, Japan
Masashi YABUNO
Ishikawajima-Harima Heavy Ind.
Hiroshima, Japan
Summary
Tatara Bridge is a worlds longest steel-concrete hybrid cable stayed bridge, constructed by
Honshu-Shikoku Bridge Authority on the Onomichi- Imabari root. The center span is 890m
long, and a part of side span is concrete . Deck girder section is structured by 3-chambers.
This bridge is much flexible because not only by its length but also the low girder- depth (the
girder-depth / span-length ratio is about 1/300). This flexibility cause a large deformation while
the erection of girder. The largest vertical deformation at the edge of girder caused by the loading
of deck-block while erection was more 2 m.
Such a flexibility of structure, it was difficult to complete the bridge accurately with controlling
cable tension as used on general sized cable stayed bridges. Because the geometric error would
be large with only controlling the tensions.
So, for the accurately erection of Tatara Bridge, we gave account on the geometrical controlling
included length of each member. By those controlling, Tatara Bridge was completed accurately.
Fig-1 General view of the Tatara Bridge
Fig-2 The procedure of erection of superstructure
1. Outline of erection of superstructure

Fig-2 shows the procedure of erection of superstructure.
At first, the lower part of tower and inclined support were erected by floating crane(F.C.), and
then large block of deck around tower was erected. Upper part of tower was erected block by
block.
The side span was consisted by steel and PC deck. Immediately after the completion of the
tower, steel deck between P2 to 2P was erected by F.C. by large block, because the length of
steel section was shorter than 3P side.
For the erection of side span of 3P side, the span was long that it was impossible erected by a
large block. So, at the first stage of the erection near the tower, balancing erection method was
applied. It was the method erecting deck by each short block alternatively adjusting the balance
between center and side span.
When the distance from the edge of erected deck to the PC deck became 100 m then large steel
deck block was erected by F.C. The deck of center span was erected by traveler crane in short
block. The deck was jointed by welding in upper flange, bolted by high tensile bolt in web and
lower flange.
2. Basic philosophy for accuracy control

2.1 Structural characteristics of Tatara Bridge
For the controlling of accuracy of this bridge, the structural characteristics should be cleared. So,
the calculation considering some error factor was carried out.
a) Tolerance of member length
The error of deck length effect for the geometrical accuracy. If the deck was 2mm longer in 10m
length block, the deck level came to be 320mm higher at the center of span. The effect for the
cable tension is few, so negligible.
The error of stay cable length also much effective for the geometrical accuracy of the deck. The
cable tension is insensible, so the controlling of cable tension is not applicable for the accuracy
controlling.
b) Tolerance of straightness
The deck structure shows high convergency. While if the each deck block have same tolerance
of angle, geometrical tolerance of whole of deck is negligible it expected making large curvature.
Local tolerance of angle remain after completion and it is hard to be collect. But the effect also
locally and negligible in whole structure.
c) Tolerance of Dead weight
If considering the 5% error in dead weight, vertical 170mm deformation will be caused at the
center of middle span. Then the tension change about 18 ton.
The dead weight of PC deck is large but they were supported themselves, so the effect for the
whole structure was few and negligible.
d) Tolerance of cable tension
The effect of dead weight of deck for the cable tension is few. If the dead weight changed 5%,
the tension changed only 2 - 3% . So, the tolerance of cable tension caused by the dead weight is
negligible for the geometry of the deck. But if the cable tension is controlled satisfying the design
value without considering the error of dead weight, large deformation of deck would be caused
(in order of meter). Because the tension not correspond to the dead weight play as the prestressing.
So the tension controlling was not suitable by the effect to geometrical accuracy.
e) Tolerance of deck closing work
The structure is so flexible that the force pulling the deck for closing is small and easy to do.
But the tolerance caused by the closing work remain locally around the closing joint.
f) Tolerance of stiffness
Calculation considering the 5% tolerance of bending and elongation stiffness for deck and tower
was done. The result shows that the effects for sectional force and deformation were few and
negligible. Calculation considering the 1% tolerance for the elongation stiffness for stay cable
also done, and the effect was negligible.
g) Geometrical tolerance of PC deck
The PC deck of side span have high bending stiffness because of its short span. So the
controlling of geometry is difficult by its rigidity.
The characteristics of structure got by calculations are summarized as follows;
the effects of tolerances of dead weight and stiffness are few for the geometrical accuracy.
The change of stress also in the range of margin considered at design.
the tolerance of member length effect for the geometrical accuracy, but few for the tension of
stay cable and sectional forces of deck and tower.
a large deformation would be caused by the controlling the cable tension, adjusting for the
design value without the consideration of tolerance of dead weight . So such a usual and
traditional method was not suitable for this bridge.
controlling of the globally geometric of structure effect few for the sectional forces (the
change of them are in the range of margin considered at design) .
2.2 Basic philosophy for accuracy control
By the result of the study for the characteristics of structure, we decided the basic philosophy
for accuracy control of erection of deck.
a) at site, geometric of deck are mainly controlled, but cable stresses are not. The tension of stay
cables are observed for the references only but not for the controlling.
b) the controlling of length of member are emphasized.
c) controlling of member length is based on the fabrication data observed in workshops because
of the accuracy of data itself. Also the observed value at site considered supplementary. Off
cause, the data of PC deck was given only in site.
d) the measuring and controlling were done when a same level of stays of center and side span
were erected.
e) if the controlling by spacer plate at anchor of stay cable was required, it was planed to be done
only a stay cable at edge block. (actually, only measurement work done in night, but such a
installing work never done because the controlling at daytime satisfied the requirement.)
f) geometrical data were measured while night time, under constant temperature state.
g) controlling of length of stay cable by spacer plate were done for collecting the geometry of
deck, because the structural characteristics and for the performance of the road.
h) controlling of geometry of PC deck is impossible. So the setting direction of jointing girder
was decided to be not effective for whole structure. Also, the coordinate of stay anchor in PC
deck was measured before erection of stay, and the error in them are collected by spacer plate.
i) effect of deformation by creep is negligible, so it never considered in controlling in each step.
The distinction of geometrical accuracy controlling of Tatara Bridge is given by the controlling
of length of the member fabricated in workshops. The measurement in site is only a checking of
erected geometry. By such a controlling, the works in site for accuracy controlling would be
reduced.
pre-measuring data in workshops
the height of tower (accumulated),steel deck length (accumulated),
PC deck length (measured in site),length of stay cables (the length of gage wire)
definition of thickness of spacer plate
erection of cable
measurement for controlling

(at each erection step)
lean of tower
camber of deck
tension of stay cable
temperature of bridge
controlling by
spacer plate
measurement for tolerance factor

axial force of inclined support
straightness of deck
the length of deck
shrinkage by site welding
weight of deck block (in workshops)
weight of stay cable
calculation of tolerance
collection for temperature
NO
judgment
YES
next erection step
Fig-3 Flow chart of accuracy control
3. Target of accuracy control in erection

Target allowable range of accuracy in erection is;
sectional forces: in a range of margin equal 5% of design sectional forces
geometry of deck: allowable range decided for the completed state are applied for erection
-110
+110
+110
-110
[unit:mm]
+235
+55
+80
-55
1A P1 P2
2P
50m 50m 170m
-235
-80
3P
890m
1480m
270m
P3 4P
50m
Fig- 4 Target value of accuracy control
4. Measuring system
Whole construction work of superstructure was divided into two joint venture, 2P and 3P side.
Each of them had own measuring system, but the logic of calculation and the basic method for
accuracy controlling were same.
Table-1 shows the characteristics of measuring system of both.
lean of tower
2P JV
optical range finder
(measure 2 point automatically)
camber of deck
3P JV
vertical collimator
(collimate directory with CCD
camera)
water gage
(pressure sensor system
:measuring the difference of
pressure)
water gage
(magnetic strain system
:measuring the water level by
magnetized float )
tension of stay
accelerometer
cable
(calculate the tension by natural frequency)
temperature
thermocouple
(automatically )
calculation
non-linear calculation
(systematized by EWS)
others
automatically measuring for all data
automatically except for the lean of
tower
Table-1 Comparison of characteristics of measuring system
5. Actual result of accuracy control

5.1 Workshops
In workshops, the length of each member was measured and controlled with cumulated
tolerance.
The cumulated tolerance were under 10 mm in total length of tower and deck, it means the each
parts were fabricated with few tolerances.
6
The tower blocks jointing edge were faced by facing (cutting) machine. The deck blocks were
welded under temperature control, the difference of temperature of upper and bottom flange is
less than 2 degrees. The accuracy of data itself got at workshops were high, so the reliability of
the accumulated data also high.
The measuring of length of stay cable were impossible because of its length, the longest one is
about 460m. So they were controlled by the length of gage wire.
The actual measuring of length of stay cables were impossible. But, as mentioned below, it was
estimated that the major uncertainty factor was originated by this tolerance of stay length because
the tower and deck were measured accurately.
5.2 Result of measurement for accuracy control in erection
In balancing erection of 3P side, the edge of deck of center span deformed downward, side span
upward and the tower top leftward (2P side). They means the total structure leaning leftward. It
was caused by the weight of traveler crane on the edge of deck of center span. The state in
balancing erection was easily deform by small load, so the estimation of factor cause the
tolerance
Fig-5 The geometrical tolerance and the cable tension tolerance (when the center span was
closed)
7
was difficult. Considered such a situations, the controlling of length of stay cable were never
done in balancing erection.
In cantilever erection state, after the side span was jointed with PC girder, edge of deck of
center span deformed upward, side span downward and the tower top lean outer side of bridge,
both 2P and 3P. Such a tendency was shown clearly after the erection of stay cable anchored in
PC deck. If they were left without any control, the deck level at the center estimated was over
allowable range, 235 mm when the bridge completed.
So the length of stay cable were controlled by the erection of stays anchored in PC deck,
installing the spacer plate. The thickness of spacer plate were calculated based on the tolerance of
deck level.
The tolerance of deck level at the edge of deck of center span was 150mm upward than design.
By the result of measuring of them after closing of span, the decreased of tolerance was about
100 - 150 mm locally around the center.
Fig-5 shows the geometrical tolerance when the deck was closed
6. Analysis of tolerance factors

6.1 Outline of tolerances
As mentioned above, in cantilever erection state, the structure lean outer side of bridge, both 2P
and 3P. Such a tendency was shown clearly with the progress of erection.
The deck near the center of main span deformed upward just before the closing. And it was
decreased 100 - 150 mm by closing locally around the center.
The factor of tolerance after the closing of deck, was needed to be considerate special local
condition. So the analysis of the tolerance factor was done individually, before closing and after
closing.
6.2 The analysis of tolerance factor at cantilever erection
In cantilever erection state, edge of deck of center span deformed upward, both 2P and 3P. This
tolerance grew with the progress of erection. This tendency shown clearly after the erection of
stay cable anchored in PC deck. So the analysis of tolerance factor was done at the erection step12(2P) and step-14(3P). The number 12 and 14 means that the stay cable number counted from
the lower, so the step-12 means the erection of 12th stay cable.
By the analysis, the half of tolerance was caused by these 4 factors; dead weight of steel deck,
temperature of bridge, weight of erection facility and creep of PC deck. The factor cause another
halfs was assumed as uncertain factor. Uncertain factor was conversed for the thickness of
spacer collecting the length of stay cable. After those steps, the thickness of spacer (it means the
effect of uncertain factor) was controlled.
The tolerance caused by the factor already had been cleared were left without any controlling.
By the calculation considering those effects, the deck would be deformed upward about 150mm
at just before the closing. Actually this tolerance was measured for such situation.
Fig-6 Chart of analysis of tolerance factor

6.3 The analysis of tolerance factor after the closing
The tolerance of deck level at the edge of erection of center span was 150mm upward than
design. This value equal to the analysis of tolerance factor mentioned above.
But, by the measurement after closing of span, the tolerance was decreased about 100 - 150 mm
locally around the center. The tolerance factor of such behaviors was analyzed.
The closing work was done in summer, the hottest season. For the welding work, the root-gap
was set as 3mm, but it became narrow by the elongation of deck caused by the high temperature.
The decks touched each other and welding work came to be difficult. So the gap was recreated by
gas cutting for easy welding.
It was assumed this gas cutting done in site cause the error of angle of welding face, and it cause
the deformation to downward at center block. So calculation considering such deformation angle
was done.
By the result of calculation, the angle of joint welding face changed 6mm, then the deck of
center would be deformed about 100mm locally. This result explains the actual situation. This 6
mm is appropriate consider the original 3 mm + gas cutting 3 mm.
Another calculations, considering the error of length of center block, or another factor, never
explain actual situation. So the factor cause this local tolerance is the change of face angle of
welding.
Conclusions and acknowledges

Tatara Bridge is the worlds longest cable stayed bridge not only the 890m center span, but also
the length of cantilever erection. The structure is much flexible and this flexibility cause the large
deformation. The accuracy control was difficult by this flexibility.
From the phase of fabrication in workshops, the length of every member were measured and the
tolerance was controlled severely. Also in the phase of erection at site, worked for the reproduce
of the accuracy of fabrication in workshops. By those endeavors, Tatara Bridge was completed
with high accuracy. Everyone relate for this work pride this result.
This result and experiences, construct with the controlling the member length geometrical, will
help the construction of long span cable stayed bridge in future.
We thank everyone who related this work for the completion with high accuracy, and no
accident.
References
[1].
Yukikazu Yanaka, Tsutomu Takazawa and Nobuyuki Hirahara ; Erection of the Tatara
Bridgs Superstructure, Proceedings of the IABSE Symposium Kobe ,1998
10
Bridges with Multiple Cable-Stayed Spans

Michel VIRLOGEUX
Consulting Engineer and Designer
President of fib
Bonnelles, France
Michel Virlogeux, born 1946,

worked as civil servant in Tunisia
(1970-1974) and then in France at
the SETRA. Head of the large
bridge division (1980-1994), he
designed many bridges among
which the Normandie Bridge and
the R Island Bridge. Now
Consulting engineer, he worked
as consultant for the Portuguese
Administration for the Vasco de
Gama bridge.
Summary
This paper is devoted to a very important development of cable-stayed bridges, bridges with
multiple cable-stayed spans. Beginning with historical reference to pioneer bridges by Ricardo
Morandi, it evokes the very few bridges built with several cable-stayed spans and the projects
which were proposed without success. It ends with the presentation of recent and important
projects which evidence the possibilities of this new concept for wide river and sea crossings.
1.
Historical background
As everyone knows, the first attempts to erect cable-stayed bridges in the beginning of the 19th
century were unsuccessful with the collapse of the Tweed and Saale bridges; engineers ignored
at the time the real flow of forces and did not seriously consider wind effects even with a
simplified and purely static approach. The famous French scientist Navier "demonstrated" that
cable-stayed bridges were unsafe and that suspension bridges were to be preferred. This stopped
the development during a very long time, and cable-stays were only used in some suspension
bridges close to the pylons to stiffen the system; the best example is the Brooklyn Bridge, but
many others could be cited. In France, at the turn of the century, Gisclard increased the role of
cable-stays in his personal composite design associating suspension and cable-staying. The first
very pure cable-stayed bridge has been built in Spain by Eduardo Torroja in 1925, in concrete, a
cable just replacing a pier which could not be installed due to the site.
But the real and scientific development of cable-stayed bridges came with the ideas and the
papers by Franz Dischinger in the late thirties and beginning forties. Surprisingly, the first
application was in France by Albert Caquot, in 1952 and in reinforced concrete, for the bridge
over the Donzre Canal, some years before the well-known Stromsund bridge in Sweden.
Everyone knows the fantastic development of cable-stayed bridges which followed, in Germany
in a first step and in the whole world later.
But in the same time as the concept of modern cable-stayed bridges was being developed, with "flexible"
pylons and a continuous deck - and later with multiple cable-stays following Helmut Homberg, total
suspension initiated by Fritz Leonhardt and very flexible decks developed by Ren Walther and Jorg
Schlaich - Ricardo Morandi developed his own concept in a different direction, with extremely rigid
pylons (inverted V shape longitudinally, with an additional V to support the deck), rigidly connected to a
deck section cantilevering on both sides, and with simply supported spans to close bays between the
different cantilevers tied to their pylons. The first application of this concept was the Maracaibo Bridge,
designed by Morandi and completed in 1962, with six pylons and five main cable-stayed spans 235 metres
long (figure 1). The same principle was used by Morandi, but with two pylons only, for the Wadi Kuf
Bridge in Libya (main span 282 metres in 1971), and for the Polcevera Creek viaduct near Genoa in Italy,
with three pylons and two main cable-stayed spans (208 metres in 1964). It has been reproduced only
once by another designer for the Chaco Corrientes Bridge in Argentina with two pylons (245 metres in
1973).
Figure 1 - Structural concept of the Maracaibo Bridge

2.
The specific problem of multiple cable-stayed spans
As we shall see, the concept of Morandi's bridges is perfectly adapted to the specific problem of
bridges with multiple cable-stayed spans. Though evident, these problems must be evoked.
2.1 In a classical cable-stayed bridge with three spans, loading the main span produces its
downwards deflection and due to the tension variation in cable-stays the pylons bend towards the loaded
span; the cable-stays which suspend side-spans receive a tension variation to balance horizontal forces in
the pylons but due to the limited rigidity of the deck it deflects upwards and tension variations are
concentrated in the backstays, anchored at the abutment and which have anchorages fixed vertically due
Figure 2 - Structural behaviour of a classical three-span cable-stayed bridge
to their position (figure 2). This unequal distribution of tension variations in the cable-stays which
suspend the side-spans produces important bending forces in the pylons: the backstays, anchored on top,
balance alone the tension variations in the cable-stays suspending the main span. This is why it is
necessary to concentrate cable anchorages when the cable-stayed bridge has this classical configuration:
the reduction of the distances between anchorages in the pylons reduces the bending moments.
When a side-span is loaded, it deflects downwards, and due to the tension variation in the corresponding
cable-stays, the adjacent pylon deflects towards the loaded span; thus the tension decreases in the
backstays and in the same time the main span deflects upwards.
The backstays have a very specific role to stabilize the pylons, and receive the largest tension variations in
the bridge.
We have the same situation with cable-stayed bridges continuously extended by access spans on both
sides, where the group of cable-stays anchored close to the first pier on each side acts as backstays; and
with cable-stayed bridges having two spans, one shorter with the backstays anchored at the corresponding
abutment.
As demonstrated in many occasions, for example for the Seyssel Bridge (Travaux, October, 1988),
2.2
the design of classical cable-stayed bridges can be improved by the installation of intermediate supports in
the side-spans: when the main span is loaded, all cable-stays anchored in the side-spans act as backstays
since they are tied to an almost fixed deck at their lower anchorage; the deflection of pylons towards the
loaded span is thus reduced and the efficiency of cable-staying increased; even the downwards deflections
of the main span is reduced by the greater rigidity of the system. And when side-spans are loaded, the
load directly passes in the intermediate supports and there is practically no effect in pylons and main span.
The same of course applies to a cable-stayed bridge with two spans, when intermediate piers are installed
in the shorter one.
If we consider now a bridge with multiple cable-stayed spans, the situation is very different.
2.3
When one span is loaded, it deflects downwards; the corresponding cable-stays receive an increased
tension; the adjacent pylons deflect towards the loaded span and the adjacent spans upwards without any
other restraint than their own rigidity. There is no more any backstaying effect (figure 3). Deflections can
be only limited by the rigidity of pylons or deck, or of pylons and deck.
Figure 3 Structural behaviour of a bridge with multiple cable-stayed spans
The situation appears even more critical with the effects obtained when loading one of the adjacent spans:
the span which was deflecting downwards with the load now deflects upwards with large bending
moments, and the adjacent pylons deflect in the opposite directions.
The design must produce the necessary rigidity. One of the possible solutions consists in rigidly
2.4
connecting the pier below the pylon, the deck and the pylon, so that the pier rigidity takes part in the
restriction of deflections. But this immediately produces a new problem: the structural system must adapt
to the length variations in the deck due to the elastic shortening produced by prestressing forces installed
after span closure, to temperature variations and also to concrete creep and shrinkage.
3.
The Morandi's concept
Clearly the concept developed by Ricardo Morandi perfectly answers the different questions: the
3.1
pylons are extremely rigid and can directly balance the effects of live loads on either sides; and with the
simply supported spans between the cantilevers supported by the pylons, length variations can freely
develop. The single drawback of this solution is its high cost and weight; this is why - though it must be
considered as the pioneer one for multiple cable-stayed spans - it has not been reproduced.
3.2 But it inspired many designs, though none of them received an application.
- We can mention the project proposed in 1967 by Ulrich Finsterwalder for the Great Belt Bridge, with
solid and rigid pylons suspending a series of spans 350 metres long, with a very flexible deck and
expansion joints at mid-span in each bay ([5] p. 38; [7] pp. 17-20-33; [9] p. 142).
- And also the project for a bridge across the river Ganges in India, designed by Fritz Leonhardt with
ten pylons and nine central spans 159 metres long. Pylons were also solid and rigid, allowing for a
limitation in the number of expansion joints, only in some spans ([4] pp. 28-42-45; [7] pp. 33-34).
More recently, the Grands Travaux de Marseille - GTM- developed two important projects which had no
more success than the previous ones for the Great Belt and the River Ganges.
- The first one is part of the GTM Channel Project prepared in the early eighties, which comprised a
bridge on each side of the Channel to give access to a central immersed tunnel. Each of these bridges was
made of a series of complete cable-stayed cantilevers - composite deck rigidly connected to the concrete
pylon, and cable-stays -, totally prefabricated and installed with the help of heavy floating cranes on the
corresponding piers; the cantilevers were joined by drop-in spans to constitute a series of typical spans
520 metres long. This project was in the same time a prefiguration of the Rion-Antirion conceptual
design, and one of the first attempts to develop heavy prefabrication techniques which received large
applications later with the Storebelt Western Bridge, the Second Severn Crossing and the Confederation
Bridge in Canada.
- The second one, very much inspired from the Channel project, has been jointly proposed by GTM
and Campenon-Bernard for the R Island Bridge in 1986. The same concept was used of complete cablestayed cantilevers - totally in prestressed concrete this time -, prefabricated and installed on the piers. But
with spans limited to 140 metres, the cantilevers were only joined by expansion joints at mid-span in each
bay. The cross-section of the deck, proposed by Jean Muller and inspired from a previous idea by Pierre
Xercavins, consisted in a flat slab stiffened by multiple floor-beams and with side walks at a lower level
to produce the desired rigidity; we developed and applied this concept for the Burgundy Bridge at Chalonsur-Sane (Travaux, October, 1991 and July-August, 1992).
- GTM came back to these principles and very close to Morandi's ideas with the conceptual design of
the Rion- Antirion bridge, developed in the late eighties by Jean-Paul Teyssandier, Franois Lemprire
and Yves Maury's team. The bridge is made of four cable-stayed cantilevers, each resting on a large
foundation caisson which constitutes a pier in the same time, and of simple spans between the cantilevers.
Each cantilever consists in a four-legged pylon, rigidly connected to the composite deck, and of two
cantilever arms, 255 metres long from the pier axis. Each central span, 560 metres long, is made of two
cantilever arms - coming from the two adjacent pylons - and of a simple span 50 metres long. Each sidespan is made of one cable-stayed cantilever hanging from the corresponding pylon, also 255 metres long,
Figure 4 The peliminary design of the Rion-Antirion Bridge
and of a single span 50 metres long to join the cantilever with the end support (figure 4). The sole
difference with Morandi's design is that the cantilever - with its pylon - is not rigidly connected to the pier
below; to reduce seismic forces, the cantilever is installed on sliding bearings with a system of large
dampers in both directions - longitudinal and transverse - to limit seismic movements. We shall see later
how this initial concept has been amended and improved for a much better design.
4.
Typical solutions
4.1 Some other solutions than the Morandi's concept can be considered, though many of them are not
extremely elegant (figure 5).
4.2 One consists in introducing an intermediate support at mid-span in every second span. Of course
this is not always possible, and this is certainly the weakest way to introduce the necessary rigidity.
Fortunately, nobody dared doing it.
4.3 The second solution is inspired from suspension bridges, which are even less adapted to the concept
of multiple spans than cable-stayed bridges. To prevent pylons from bending towards the loaded spans,
they are connected from head to head by horizontal cables, headcables ("cbles de tte" in French).
Several French bridges built in the first half of the century have several suspended spans with such headcables ; we can cite the bridges at Chteauneuf-sur-Loire, Langeais
The same could be done for cable-stayed bridges, though this solution is probably not so efficient as for
suspension bridges since the structural rigidity of cable-stays is greater ; the additional effects of headcables may be more limited. In addition this does not look so elegant, with the introduction of a new line
in the structure, reducing the architectural simplicity. A unique project referred to this technique, the
winning design of the Poole Harbour competition, but construction is not yet decided.
4.4 A third solution consists in introducing, in addition to the classical cable-stays distributed to carry
the deck loads, diagonal cable-stays which are only installed to stiffen the pylons; a typical diagonal cable
is anchored on top of a pylon and at the deck level in one of the two adjacent pylons. Once again, this
solution introduces a new line in the structure, reducing the architectural simplicity. It has been adopted
by Jorg Schlaich for the design of the Ting Kau Bridge in Hong Kong. Since it has only three pylons and
two central cable-stayed spans, only the central pylon had to be stabilized by diagonal cables of this type.
The composite deck is supported on the piers by classical bearings so that length variations can easily
develop.
Almost the same idea consists in installing cable-stays from each pylon beyond the mid-span section of
the two adjacent spans; the central part of each bay is thus suspended from both adjacent pylons. But this
can be efficient only if the deck has a rather large rigidity.
4.5 Fritz Leonhardt proposed a last solution many years ago ; it consists in amending the distribution of
spans with a longer and a shorter one for each group of two. With a ratio of about 0.90 to 1.10 or 0.85 to
1.15, the shorter span stiffens the longer one. But this system also has serious drawbacks ; the differences
in span lengths and in the distribution of stays, is not so elegant, and in addition the distribution of
permanent loads is not well balanced, calling for serious amendments.
The Macau Bridge - designed by Jos Luis Cancio Martins with two central cable-stayed spans 112
metres long - can be related to this concept ; with two pylons and a very short span between the two main
spans, it works like two successive and independent cable-stayed bridges and cannot be considered as a
real reference for bridges with multiple cable-stayed spans. It has been completed in 1994 ([11] p. 52;
[12]).
Figure 5 A series of more or less acceptable solutions for multiple cable-stayed spans
5.
Distribution of rigidities
5.1 Finally, the best solution appears to be the research of an adapted distribution of rigidities between
deck, piers and pylons to resist bending forces and limit deflections (figure 6). From one extreme to the
other, several solutions can be considered :
- on the one hand, we can have a very rigid deck and flexible pylons on condition that spans are not too
long. Then the deck can be simply supported on the piers with pylons rigidly connected to the deck for
simplicity.
- Rigidity can be distributed between piers, deck and pylons with a careful attention to the deck length
variations.
- And on the other hand, we can have a very flexible deck on condition to have rigid pylons, either by
their shape (an inverted V, longitudinally) or their dimensions. Of course, bending moments must pass
from pylons to piers, either through two lines of bearings to adapt to length variations or with a rigid
connection between pylon and pier on condition that the design of piers adapts to the deck length
variations.
Figure 6 Distribution of rigidity between piers, deck and pylons

5.2 As already mentioned, length variations are produced in the deck by the elastic shortening induced
by prestressing forces installed in the structure after the span closure, by temperature variations and by
concrete shrinkage and creep. The design must be such that they can develop almost freely. Three
solutions can be proposed (figure 7).
- The first one consists in installing between piers and deck special bearings, sliding except on one, two or
three central piers - the number depending on the piers flexibility - where fixed bearings can be
introduced. There may be only one line of bearings on each pier if the deck is extremely rigid as already
shown (paragraph 5. 1) but there must be two lines of bearings to take advantage of the piers rigidity.
This solution is not so simple due to two different problems: if the bridge is very long, the displacements
produced by length variations produce load eccentricity (the deck and pylons move on the piers and
receive excentered reactions) ; and due to the heavy loads on the supports, friction on sliding bearings can
produce important bending forces in high piers.
- The second solution is more efficient and more elegant. It consists in producing a rigid connection
between the deck - which may be rather flexible - and piers made of two flexible parallel shafts. Such
piers are extremely rigid as regards rotations, but rather flexible as regards length variations in the deck.
This concept of twin flexible shafts was developed by Jacques Mathivat in the early sixties.
Figure 7 Solutions to allow for length variations with rigid piers

- The last solution consists in introducing an expansion joint in some - few - spans. But to avoid an
increase in vertical deflections, the continuity of bending moments can be restored by introducing a steel
continuity beam in the deck (figure 8) ; as done for example by Jean Muller for the Rogerville viaduct, a
rather classical prestressed concrete box-girder bridge. But this is possible only with a box-girder deck of
rather large dimensions, just to leave the necessary place.
Figure 8 A continuity beam to transfer bending moments through a joint

5.3 Though some of these ideas already appeared in one or two early projects, such as twin flexible
shafts, no real application was made of this global concept.
5.4 We can only mention that after the first competition for the Storebelt crossing another project was
proposed in Denmark for the Samso Belt, in 1972, this time with a continuous deck ; the project had four
spans 264 - 624 - 624 and 264 metres long ([5] pp. 313-314). The lateral pylons were stabilized by
backstays but the central one had to receive a very large rigidity. Of course - and as for almost all the other
bridges which will be evoked in this paragraph - the situation is extremely favourable with only four
spans.
5.5 Since this time some medium-span cable-stayed bridges have been built, almost unnoticed, with
several cable-stayed spans.
The first one is the Kwang Fu bridge in Taiwan, designed by T.Y Lin and completed in 1978 ([11] p. 10).
It has three pylons and two central cable-stayed spans, 134 metres long ; pylons have classical shapes and
a limited rigidity ; the effects of traffic loads are balanced by bending forces in pylons and deck - which
has a rather large flexural rigidity as compared to the span - and also by the side-spans with cables acting
as backstays due to the high deck rigidity. These backstays control the deflection in the lateral pylons,
only the central pylon being really flexible. Such a design has been reproduced in Spain for the Colindres
Bridge completed in 1993 with three pylons again and two central spans 125 metres long ([11] p. 49). But
the most important application has been for the construction of the Mezcala Bridge in Mexico, still with
three pylons and two main spans 312 metres long, completed in 1993 ([11] p. 44). Due to some specific
site conditions controlling the distribution of spans, the central pylon is taller than the lateral ones, as in
the Ting Kau Bridge.
We must insist on the favourable situation of these bridges with only three pylons and two central cablestayed spans. This is only an intermediate step between classical cable-stayed bridges with two pylons and
a central span and the real multispan cable-stayed bridges. The single application of really multiple cablestayed spans is the Arena viaduct in Spain, designed by Juan Jos Arenas and completed in 1993, with six
pylons and five central spans, 105 metres long ([11] p. 48). But the reduced span length limits the rigidity
problems in this bridge and prevents learning much from its design: the classical rigidity of deck and
pylons is perfectly adapted to the forces in such spans.
We must add that in all these bridges - Kwang Fu, Colindres, Mezcala and Arena - the deck is supported
on the piers with classical bearings to adapt to length variations.
6.
Geneva and Millau
6.1 Two very large projects developed in the nineties produced a gigantic step forward, for the Millau
viaduct over the River Tarn valley, and to cross the Lake of Geneva. We developed the concept for the
Millau viaduct in 1990-1991 but the design remained preliminary until 1993, due to the many obstacles
met by the project. Jean- Franois Klein and Pierre Moia took inspiration from it to design a bridge across
the Lake of Geneva in a project competition which they won ; they developed in 1993-1994 an excellent
project with a completely detailed design. Being in the jury of this competition, we have taken inspiration
from their project for the later development of the design of the Millau viaduct so that these projects
helped each other as it happens frequently.
Figure 9 The bridge designed to cross the Lake of Geneva
6.2 The Pont de la Rade in Geneva has four pylons and three central spans 350 metres long. It has a
slightly curved alignment for the bridge elegance (R = 900 metres). The deck is extremely wide, 33.46
metres. Its design is specially elegant, balancing rigidity between a relatively slender deck (an elegant
streamlined box-girder, 3.50 metres deep) and rather rigid piers and pylons (figure 9). Length variations,
produced by temperature, shrinkage and creep are permitted by the relatively limited distance between the
central point and the extreme pylon but also by soil conditions. Unfortunately a general votation is
necessary in Switzerland to build very large structures and the Geneva population voted against the
project for financial reasons.
6.3 The Millau viaduct is even more ambitious; almost 2.5 kilometres long, it comprises seven pylons
and six central spans 342 metres long with two piers about 240 metres tall. The development of the
project has been extremely complex, with an initial design by the SETRA and two design competitions, a
rather informal one in 1993 and a more formal one in 1995-1996. Five teams of engineers and architects
were constituted for this second competition, from the result of the first one and each in charge of
developing a different type of solution. The cable-stayed solution with multiple spans, developed from our
conceptual design by SOGELERG - Europe Etudes Gecti - SERF and the British architect Sir Norman
Foster, was selected in July, 1996 and we developed the project with this team between the end of 1996
and September, 1998.
Figure 10 The Millau viaduct (prestressed concrete solution)
Two alternatives are proposed, the deck being either in prestressed concrete or in steel with almost the
same design adapted to the specific conditions of multiple cable-stayed spans and to the extreme wind
forces due to the high position of the bridge in the valley. The rigidity is distributed between the deck,
piers and pylons. The deck is a trapezoidal box-girder with a rather narrow bottom flange so that it is
almost triangular; it is about 4.50 metres deep. The pylons, 90 metres tall, have the shape of inverted V
for a very high rigidity. The design of piers is more complex since the taller ones have to resist important
forces due to wind and to second order effects ; and the extreme ones - about 90 metres high - must adapt
to very important length variations due to the bridge size (about 0.80 metres). As soon as in 1992-1993,
with Emmanuel Bouchon we decided to have these extreme piers made of two parallel, flexible shafts
with a unique line of fixed bearings on top of each to increase their flexibility.
The architect later preferred to have the same design for all the piers; this led to the final design of solid
piers which divide into twin shafts in the upper part, 90 metres high (figure 10).
This very elegant bridge will be built - if decisions taken are applied - in the years to come with a
concession.
Figure 11 - The four pylons of the Rion-Antirion Bridge
7.
Total suspension
7.1 A last idea must be evoked to complete this overview : the total suspension concept. It has been
initiated with the Pasco Kennewick bridge and soon after for the Alex Frazer Bridge in Canada. It must
be clear that it adapts very well to the concept of multiple cable-stayed spans since it allows for free
length variations without any interference with the rigidity of piers and pylons.
This concept has been proposed by Bouygues and Pierre Richard for the R Island Bridge in 1986. They
proposed a cable-stayed bridge with a continuous deck, almost 2800 metres long and with a series of
central spans 210 metres long. Unfortunately, just after the successful construction of the Bubiyan Bridge
and at a time when the Syllans and Glacires viaducts were to be built, Pierre Richard preferred for the
deck an expensive three-dimensional prestressed concrete truss the cost of which eliminated the solution.
The deck was totally suspended from the pylons to adapt to longitudinal length variations - concrete creep
and shrinkage, elastic shortening produced by prestressing tendons installed after span closure and effects
of temperature - in complete opposition with the solution proposed for Millau and Geneva. Traffic loads
were perfectly balanced by the large flexural inertia of the three dimensional truss which constituted the
deck, and very classical, almost slender pylons could be designed in this situation.
7.2 This is why, when Jacques Combault asked for our opinion on the design of the Rion-Antirion
Bridge, we suggested to have a continuous deck, totally suspended from the four pylons. The concept has
been immediately adopted and developed with many advantages as compared to the initial design :
continuity, a regular distribution of cable-stays in the spans to perfectly balance loads... Rigidity this time
comes from the pylons, made of four legs with an inverted V-shape in both directions ; the composite
deck is rather flexible. The final project, now being detailed by GTM and Ingerop, has a continuous deck
with five spans, 286 - 3 x 560 and 286 metres long; and pylons are rigidly connected to the piers, a much
more comfortable situation than installing a cantilever on sliding bearings and dampers (figure 11).
Figure 12 The final design of the Rion-Antirion Bridge
8.
Conclusion
As evidenced by this survey, cable-stayed bridges with multiple spans might develop rapidly in the
coming years, specially if the Millau viaduct and the Rion-Antirion bridge are erected as expected,
evidencing the enormous capacities of this new structural type.
Literature
[1]
[2]
[31
[4]
The bridge spanning Lake Maracaibo in Venezuela. Bauverlag. Berlin. 1963

Boaga G. and G. Boni. The concrete architecture of Ricardo Morandi. Alec Tiranti. London. 1965.
Wittfoht H. Triumph der Spannweite. BetonVerlag. Dsseldorf. 1972.
Podolny W. and J. Scalzi. Construction and design of cable-stayed bridges. John Wiley and Sons.
New York. 1976.
[5] Gimsing N.J. Cable-supported bridges. Concept and design. John Wiley and Sons.
Chichester. 1983.
[6] Wittfoht H. Bridges. BetonVerlag. Dsseldorf. 1984.
[7] Walther R. and als. Ponts haubans. Presses polytechniques romandes. 1985.
[8] Leonhardt F. Ponts. Puentes. Presses polytechniques romandes. 1986.
[91 Troitsky M.S. Cable-stayed bridges (second edition). BSP Professional Books. Oxford. 1988.
[10] Ricardo Morandi. Innovazione, tecnologia, proggetto. Gangemi. Roma. 1991.
[11] Freyssinet. Cable-stayed bridges. 1994.
[121 The new Macau-Taija Bridge. The friendship bridge. Port and Bridge office. 1994.
The Development of Composite Cable-Stayed Bridges

Holger Svensson, born 1945 received
his Diplom-Ingenieur (M.Sc.) degree
in 1969
He specialised in all aspects of the
design and construction of long-span,
mainly cable-stayed bridges all over
the world.
Holger S. SVENSSON
Managing Director
Leonhardt, Andr und Partner
Stuttgart, Germany
Summary
Although the first modern cable-stayed bridge (Strmsund / Sweden, 1955) used a composite
deck, this bridge system really became successful in the mid-eighties. Currently 44 major
composite cable-stayed bridges are completed or under construction, including a twin bridge
(Baytown) and two double deck bridges (Kap Shui Mun, resund).
Composite cable-stayed bridges will remain the dominant system for medium and longer spans.
1.
Introduction
For the last few years, composite cable-stayed bridges have been much more common worldwide than all-concrete or all-steel ones. The main reasons are economy in materials and ease of
construction. By using concrete rather than steel in compression, and by using a concrete
roadway slab rather than an orthotropic deck, substantial savings against all-steel bridges are
realised.
A composite deck can use small parts - main girders, floor beams, precast slabs - which can
easily be lifted. They can be joined simply by bolting the steel girders together and connecting
the precast slabs with cast-in-place joints. Thus, smaller lifting equipment and the absence of
match-cast joints together with savings in cable steel favour composite decks against all-concrete
ones. This trend was already pointed out by us in 1984 [44].
In order to distinguish the different types of composite decks we split them into four groups, see
Tables 1 to 4:
-
Main girders having a concrete roadway slab on top of a steel grid or deck.
Main all-steel (or composite) girders in the centre span combined with all-concrete side
spans.
Concrete main girders with a concrete floor slab supported by steel floor beams.
Composite roadway slabs are orthotropic steel decks stiffened by a substantial layer of
concrete.
Composite bridges are currently not only more numerous, but they include the last three record
span holders: the Yang Pu Bridge [22] with 602 m in 1993, the Normandy Bridge [37] with
856 m in 1995 and the Tatara Bridge [38] with 890 m in 1999, see Fig. 1.
Main Span
Length [m]
1000
[38]
[37]
900
symmetric
configuration
800
Composite
main girders
700
[22]
600
500
400
300
(not built)
[4]
Design
completed
[28]
[29]
[9]
Steel main and
concrete side
pans
200 [1]
(not built)
100
0
1950
1960
1970 1980 1990 2000

Year of Completion
2010
Figure 1. Development of span lengths, from Tables 1 and 2
2. Historical Development
2.1
Composite main girders (Table 1)
The first modern cable-stayed bridge - the Strmsund Bridge [1] - has a concrete slab on top of a
steel grid. The concrete roadway, however, distributes only the wheel loads and does not
ostensibly act composite for primary forces, unlike the other bridges listed in Tables 1 to 3.
The first and - curiously - the only composite cable-stayed bridge in Germany is the one in
Bchenau [2], built in 1956. The reason is that the German codes at that time did not permit
tensile stresses in the concrete slab [40]. This would have required post-tensioning the slab
(together with the steel deck) which would have been uneconomical.
A major step forward took place with the design of the Hooghly River Bridge [4] with a
457 m (1500 ft) main span. In order to suit local Indian fabrication methods a concrete slab was
chosen to avoid the welded orthotropic deck which would have been chosen elsewhere. All steel
connections are actually riveted, and a design requirement was that the concrete deck be without
stress under permanent loads. Had it been completed in 1980 as originally planned, the bridge
would have held the span record (Figure 1). However, design changes and construction problems
delayed the bridge opening until 1992, by which time a longer span had been constructed
elsewhere.
Ref. Name
No. Location
Slab Type
Tower Type
Post-Tension Cable Planes
CIP, 0,20 m
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Completed Deck
Main Span Width
Deck Depth
Strmsund Bridge
1955
14,30 m
Sweden (not composite) 183 m
3,00 m
Bchenauer Bridge
1956
20,80 m
Bruchsal, Germany
58,80 m
1,40 m
Pont des Iles
1967
28,65 m
Expo 67, Canada
2 x 105 m 2,82 m
2nd Hooghly River Br.* (1980)1992 25,00 m
India
457 m
2,33 m
Sitka Harbor Bridge
1972
11,00 m
Alaska, USA
157 m
1,80 m
Heer-Agimont Bridge
1975
14,50 m
Belgium
123 m
2,05 m
Steyregger Donau Br.
1979
24,86 m
Linz, Austria
161,2 m
4,07 m
Sunshine Skyway Br.*
(1982)
27,50 m
Florida, USA: Design
366 m
2,34 m
Annacis Bridge
1986
28,00 m
Vancouver, Canada
465 m
2,10 m
Saint Maurice
1986
2 x 11,75 m
Switzerland
105 m
1,01 m
Quincy Bridge
1987
13,80 m
Mississippi River, USA 274 m
2,10 m
Kemjoki Bridge
1989
25,50 m
Finland
126 m
Weirton-Steubenville Br 1990
28,00 m
Ohio River, USA
250 m
2,74 m
Nan Pu Bridge
1991
25,00 m
Shanghai, China
423 m
2,10 m
Burlington Bridge*
1993
25,70 m
Ohio River, USA
195 m
1,85 m
Thtiniemi (Heinola)Br.* 1993
22,00 m
Finland
165 m
3,20 m
Utsjoki Bridge
1993
12,00 m
Finland
155 m
1,76 m
Karnali River Bridge
1993
11,30 m
Nepal
325 m
3,00 m
Mezcala Bridge
1993
18,10 m
Mexico
300/311 m 2,79 m
El Canon
1993
21,00 m
Mexico
166
2,11 m
El Zapote
1993
21,00 m
Mexico
176
2,11 m
Yang Pu Bridge**
1993
32,50 m
Shanghai, China
602 m
3,00 m
Clark Bridge
1994
30,50 m
Mississippi River, USA 230 m
1,90 m
Baytown Bridge*
1995
2 x 23,83 m
Texas, USA
381 m
1,83 m
2nd Severn Bridge
1996
34,60 m
UK
456 m
2,70 m
Kap Shui Mun Br.*
1996
35,20 m
2 H inclined
2
CIP, 0,25 m
2 H w/o struts
longitudinal
2
CIP, 0,19 m
1 H w/ struts
none
2
CIP, 0,23 m
2 H inclined
none
2
CIP, 0,20 m
2 H w/o struts
none
2
CIP, 0,19 m
2 H w/ 2 struts
none
2
CIP, 0,20 m
2 A, unsym.
longitudinal
2
PC, 0,23 m
2 Diamond
none
2
PC, 0,27 m
2 H inclined
none
2
CIP, 0,22 m
1 A incl. longit.
none
2
PC, 0,23 m
2 H inclined
longitudinal
2
CIP, varies
1 centre mast
transverse
2
CIP, 0,22 m
2 A, unsym.
none
2
PC, 0,26 m
2 H inclined
longitudinal
2
PC, 0,25 m
2 H inclined
longitudinal
2
CIP, varies
2 H, 1 strut
transverse
2
PC, 0,26 m
2 H, inclined
none
2
PC, 0,23 m
1 H, unsym.
none
2
CIP, 0,20 m
3 H towers
none
2
0,20 m
1 H inclined
2
0,20 m
1 H inclined
2
PC, 0,26-0,40 2 inverted Y
long. + transv. 2
PC, 0,27 m
2 centre masts
longitudinal
2
PC, 0,20 m
2 twin diamond
none
4
PC onto grid, 2 H w/ 2 struts
0,25
2
none
PC onto grid, 2 H inclined
Ref. Name
No. Location
Hong Kong
27
28
29
30
31
Ting Kau Bridge

Hong Kong
Raippaluoto Bridge,
Finland
resund Bridge **
Sweden / Denmark
Sunningesund Bridge **
Uddevalla, Sweden
Kolbcksbro **
Ume lv, Sweden
Completed Deck
Slab Type
Main Span Width
Post-Tension
Deck Depth
430 m
7,46 m
0,25
none
1997
43,00 m
PC, 0,23 m
448/475 m 1,75 m
none
1998
15 m
PC, 0,27 m
250 m
2,8 m
none
2000
30,50 m
CIP
490 m
10,20 m
transverse
2000
26,5 m
PC, 024 m
414 m
2,20 m
none
2001
17,65 m
CIP, 0,27 m
130 m
2,59 m
none
Tower Type
Cable Planes
2
3 centre masts
4
2 diamond
2
2 H w/o strut.
2
2 diamond
2
1 H inclined
2
Table 1: Cable-stayed bridges with composite main girders

* Co-designed by Leonhardt, Andr und Partner GmbH
** Reviewed by Leonhardt, Andr und Partner GmbH
The next important step forward came with the design of the Sunshine Skyway Bridge [8] in
1982. It has the main characteristics of modern composite stayed decks:
- open steel grid from I-girders as shown in Fig. 2
- outside main girders to which the stay cables are directly connected
- precast concrete slab elements, spanning longitudinally between floor beams, connected by
lap spliced reinforcement in CIP joints.
- erection in small elements possible: main girders, cross-girders, precast slabs.
- crack control in the slab by rebar only, without post-tensioning.
Although this composite alternate lost by a small margin in competitive bidding against the
concrete alternate, similar design principles were successfully applied to the Annacis Bridge [9],
which captured the span record in 1986. This initiated a series of cable-stayed bridges with
composite decks all over the world. In the US five cable-stayed bridges of composite
construction [11, 13, 15, 23, 24, 28] have been completed since 1987. In Finland 4 cable-stayed
bridges [12, 16, 17] were built differently. Their steel decks were launched into position and
supported by temporary piers. The concrete slabs for the Kemjoki [12] and Heinola Bridge [16]
were then cast-in-place. They span transversely with cantilevers and are, consequently, posttensioned transversely. The Utsjoki Bridge deck [17], however, uses the principles outlined for
the Skyway design. The Raippaluoto Bridge [28] was built by free cantilevering.
The single-tower Karnali River Bridge [18] is remarkable as it would have a span of more than
600 m, if doubled to a symmetric configuration, and because it was built in a very remote
location at the foothills of the Himalayas.
Since from 1975 China has built more than 30 cable-stayed bridges [46], most of them from
concrete. With the composite Nan Pu [14] and Yang Pu [22] bridges (once a record holder)
China became a front-runner in this type of construction.
The Baytown Bridge [24] across the Houston Ship Channel is a twin bridge with two
independent decks supported by 4 cable planes from double diamond towers.
4
The Kap Shui Mun Bridge [26] has the first composite double deck with highway traffic running
on the top slab and road + rail traffic on the bottom slab, similar for the resund Bridge [29].
The Ting Kau Bridge [27] has two main spans and 3 towers. The 43 m wide deck is also
supported by 4 cable planes like the Baytown Bridge, but the two decks are connected by crossgirders.
2.2
Steel or composite main span and concrete side spans (Table 2)
In order to use a concrete deck as counterweight for a steel main span it must be possible to place
piers in the sidespan. Under permanent load conditions these piers support the additional weight
of the concrete required for safety against uplift. Savings can be achieved against a balanced steel
side span.
This system was first used for the Mannheim North Bridge ,[32] in 1972, and again for the Flehe
Bridge [33] in 1979. If doubled to a symmetric configuration, the Flehe span would come to
about 700 m, Fig. 1.
The former record holder for all types of cable-stayed bridges, the Normandy Bridge [37], has a
steel main span and concrete side spans. An additional economic advantage was realised by
incrementally launching the concrete approach bridges. In order to stiffen the 856 m main span
for aerodynamic stability, the concrete approach span is extended into the main span by 116 m on
each side, so that only the remaining 624 m utilise an all-steel cross-section.
In Section 4 it is shown that such a reduced steel deck length is economically advantageous. The
double-deck Kap Shui Mun Bridge [26] not only has a composite deck in the centre span as
mentioned under Section 2.1, but has also concrete side spans. It is currently the only bridge of
that type. The concrete approach spans were also incrementally launched to cantilever 21,5 m
each side into the main span. The first three steel sections of the main spans were used as a
launching nose with some modifications because the water underneath was too shallow for
barges.
The current record holder, the Tatara Bridge [38] in Japan has an all-steel deck which extends far
into the side spans and uses concrete counterweight beams only for the backstay cables.
Ref. Name
No. Location
Tower Type
Cable Planes
32
1 A, unsym.
(2)
1 inverted Y
1
2 H vertical
2
2 centre masts
1
2 diamond
2
33
34
35
36
Completed Deck Width Steel Deck

Main Span Deck Depth Length
% of Main Span
Mannheim North*
1972
36,90 m
287 m
Rhine River,Germany 287 m
4,50 m
100 %
Flehe Bridge*
1979
41,70 m
368 m
3,80 m
100 %
Tjrn Bridge
1982
15,75 m
386 m
Sweden
366 m
3,0/3,0 m
105 %
Emscher Bridge
1990
41,00 m
310 m
3,68 m
100 %
Ikuchi Bridge
1991
24,10 m
490 m
Japan
490
2,48 m
100 %
37
26
38
Normandy Bridge
Seine River, France
Kap Shui Mun Br.*
Hong Kong
Tatara Bridge
Inland Sea, Japan
1995
856 m
1997
430 m
1999
890 m
22,30 m
3,05 m
35,20 m
7,46 m
30,8 m
2,7 m
624 m
73 %
387 m composite
90 %
1312 m
147 %
2 inverted Y
2
2 H inclined
2
2 diamond
2
Table 2: Cable-stayed bridges with steel composite main and concrete side spans
2.3 Composite cross girders (Table 3)
The foundations for the East Huntington Bridge [39] were already built for a steel superstructure
when it was decided to consider a concrete alternate for competition. We thus had the task of
designing a rather light, "mostly concrete" alternate. We did that by using high-strength concrete
and steel floor beams. This combination was bid considerably lower than the original steel
design. The Vasco da Gama Bridge [40] also uses this deck design.
Ref. Name
No. Location
Cross-Gird.,h
Distance
Tower Type
Cable
Planes
39
I-deck 0,91 m
2,73 m
I-Section, 2,0 m
4,41 m
1 A unsym.
2
2 H inclined
2
40
Completed Deck
Main Span Width
Deck
Depth
East Huntington Bridge* 1985
12,20 m
Ohio River, USA
274 m
1,52 m
Vasco da Gama Bridge 1998
31,20 m
Lisboa, Portugal
420 m
2,50 m
Table 3: Cable-stayed bridges with composite cross girders

2.4
Composite roadway slab (Table 4)
Orthotropic steel decks have two disadvantages (besides high cost): the asphaltic wearing surface
may not adhere well to the steel plate, and the steel deck may permit the forming of ice faster
than a concrete deck slab. These potential problems may be mitigated by using a concrete
wearing surface.
In 1969 a 0,22 m thick layer of concrete was cast onto the orthotropic deck of the Massen
Bridge [41]. Because this arrangement gave a very satisfactory service, it was used again with a
0,12 m thick concrete layer for the Dartford Bridge [43] in 1991.
The two Zrate Bridges [42] use a 0,10 m thick concrete layer. Its stiffening effect permitted the
distance between the longitudinal ribs to be increased from the usual 300 mm to 400 mm. Due to
the possibility of extensive repairs, the concrete wearing surface was not considered to carry the
global deck forces. On the other hand, sufficient shear studs and reinforcement were provided to
assure the integrity of the steel-concrete roadway slab under all load conditions.
Ref. Name
No. Location
41
Massen Bridge
Paris, France
42
2 Zrate Bridges*
Rio Paran,Argentina
43
Dartford Bridge
London, UK
Completed
Main Span
1969
161 m
1976/77
330 m
1991
450 m
Deck Width
Deck Depth
36,30 m
4,35 m
22,60 m
2,60 m
19,00 m
2,00 m
Concrete Slab
Thickness
0,22 m
0,10 m
0,12 m
Tower Type
Cable Planes
2 Centre masts
1
2 H with crosses
2
2 H w/o struts
2
Table 4: Cable-stayed bridges with composite roadway slab

3. Roadway Slab
For composite main girders it is of overriding concern to keep tensile stresses away from the
deck slab. Transversely this is achieved by using two external cable planes which provide
compression in the slab which acts as the top flange of a simply supported girder. Table 1 - Cable
Planes - thus shows no exception to this rule.
In the longitudinal direction, the slab receives little bending stress because the neutral plane is
located close to its underside, Fig. 2. Negative moments caused by load positions away from the
section under consideration are relatively small, and the corresponding tensile stresses are
generally overcome by the normal forces from the inclined cables. In the centre of the bridge
where these normal forces decrease to zero, sufficient compression in the roadway slab can be
created by cambering the deck, i.e. by introducing a positive moment.
Figure 2. Typical composite cross-section [15]
4. Economic Comparison
For this investigation we have consistently used the following average unit prices: concrete:
500 Euro/m, structural steel: 1.750 Euro/t, orthotropic deck steel: 3.000 Euro/t, stay cable steel:
6.000 Euro/t.
To resist a normal force thus costs about 65 % more if steel is used instead of concrete.
Therefore, towers are generally concrete. An orthotropic deck (180 kg/m) costs about four times
as much as a 0,25 m thick concrete slab. Therefore, orthotropic decks are used in relatively few
cases.
The bridge costs per m as a function of main span length are shown in Fig. 4. When neglecting
the steel main and concrete side span decks, concrete decks are most economical up to a main
span of 400 m; above that composite decks govern up to 1000 m where steel takes over.
Steel main concrete and sidespan decks reach their minimum costs if the concrete approach
bridges protrude about 100 m constant into the main span. Fig. 4 indicates that a combination of
concrete approach with composite centre deck then governs from 350 m to 600 m. Above that, a
combination of concrete approaches and all-steel main spans becomes more economical. The
results confirm the concept of the latest steel concrete bridges: Kap Shui Mun Bridge [26] with a
composite centre span of 430 m and Normandy Bridge [37] with an all-steel centre part for a
main span of 856 m.
The individual unit prices may, however, vary considerably for different locations and competition conditions, so that the intersections of the various curves may shift.
8000
7000
6000
Cost DM/m
All-concrete deck
All-steel deck
Composite deck
5000
Longit. conc/steel
Longit. conc/comp
4000
3000
2000
200
300
400
500
600
700
800
900
1000
Mainspan Length [m]

Figure 3. Unit costs for different types of cable-stayed bridges
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
Wenk, H.: Die Strmsundbrcke - Ein Grobrcken-Auftrag der Knigl. Wege- und
Wasserbauverwaltung in Stockholm. Der Stahlbau 23 (1954), p. 73 - 76.
Kunz, R., Trappmann, H., and Trndle, E.: Die Bchenauer Brcke, eine neue Schrgkabelbrcke der Bundesstrae 35 in Bruchsal. Der Stahlbau 26 (1957), p. 98 - 102.
N.N.: Stayed Girder in the Dry. Engineering News-Record, August 5, 1965.
Schlaich, J.: On the Detailing of Cable-Stayed Bridges. Cable-Stayed Bridges - Recent
Developments and their Future. M. Ito et al. 1991 Elsevier Science Publishers B.V., p. 57 76.
Gute, W.: First Vehicular Cable-Stayed Bridge in the U.S. Civil Engineering-ASCE,
November 1973, p. 50 - 55.
Mahieu, L. and Wouters, M.: Pont Hauban sur la Meuse a Heer-Agimont.
Burgholzer, L.: Die Steyregger Donaubrcke in Linz. IABSE Structures C-12/80, p. 24 25.
Svensson, H., Christopher, B., and Saul, R.: Design of a Cable-Stayed Steel Composite
Bridge. Journal of Structural Engineering, Vol. 112, No. 3, March, 1986, p. 489 - 504.
Taylor, P.: Hybrid Design for the Worlds Longest Span Cable-Stayed Bridge. Proceedings
of the IABSE Conference 1984, p. 319 - 324.
Ponts sur le Rhne Saint-Maurice. Tunnels de lArzilier et Ponts sur le Rhne.
Department des Travaux Publics du Canton de Vaud, p. 14 - 23.
Kulicki, J., Waldner, E., and Prickett, J.: Design of the Cable-Stayed Mississippi River
Bridge at Quincy, Illinois. Transportation Research Record 950, Second Bridge
Engineering Conference, Volume 2, p. 34 - 50.
Kettunen, A., and Jrvenp, E.: Kemijoki River Bridge at Rovaniemi (Finland). IABSE
Structures C-44/88, p. 6 - 7.
Kozy, W., and Kolmus III, R.: Design of the Cable-Stayed Girder Weirton-Steubenville
Bridge. Transportation Research Record 950, Second Bridge Engineering Conference,
Volume 2, p. 23 - 33.
Quinghua, Z.: Shanghai Builds Worlds Second Largest Cable-Stayed Bridge. Stahlbau 61
(1992), p. 26 - 27.
Svensson, H., and Petzold, E.: The Cable-Stayed Bridge over the Mississippi at Burlington,
USA. Proceedings of the Third Symposium on Strait Crossing, lesund, Norway, June
1994. Balkema, Rotterdam (1994) p. 239 - 246.
Pulkkinen, P., Saul, R., and Kettunen, A.: Thtiniemi Bridge. Proceedings of the 1994
International Symposium on Cable-Stayed Bridges, Shanghai, p. 153 - 157.
Jrvenp, E., Lunabba, T., and Kettunen, A.: Utsjoki Bridge - The Northernmost CableStayed Bridge in the World. Proceedings of the 1994 International Symposium on CableStayed Bridges, Shanghai, p. 148 - 152.
Arzoumanidis, S., and Kunihiro, M.: The Karnali River Bridge. Proceedings of the
IABSE/FIP Conference on Cable-Stayed and Suspension Bridges, Deauville, October
1994, p. 395 - 403.
King Revelo, C., Alvarez Guillen, C., Chavin, A .et al: The Four CableStayed Bridges of the Mexico-Acapulco Highway. Proceedings of the IABSE/FIP
Conference on Cable-Stayed and Suspension Bridges, Deauville, October 1994, Vol. 1,
p. 351 - 360.
Lin Yuan-pei et al.: Yangpu Bridge. Shanghai Scientific and Technological Education
Publishing House, 1994.
McManamy, R.: Still Tall in the Saddle, Project Survives Flood - Mississippi Span Bends
Cables over Single Pylons. ENR, November 8, 1993, p. 35.
9
[24] Svensson, H., and Lovett, T.: The Twin Cable-Stayed Baytown Bridge. Proceedings of the
IABSE/FIP Conference on Cable-Stayed and Suspension Bridges, Deauville, October
1994, Volume 1, p. 361 - 368.
[25] Maury, Y. et al: Some Aspects of the Design of Second Severn Crossing Cable-Stayed
Bridge. p. 369 - 279.
[26] Saul, R., Lovett, T., and Hopf, S.: The Kap Shui Mun Bridge at Hong Kong. Proceedings
of the IABSE/FIP Conference on Cable-Stayed and Suspension Bridges, Deauville,
October 1994, Volume 1, p. 405 - 412.
[27] Highways Department Hong Kong Government, Ting Kau Contractors Joint Venture:
Project Information, Issue 1 September 1995
[28] Esko Jrvenp, Pekka Pulkkinen: Raippaluoto Bridge. Finnish Civil Engineering, 1/97, p.
7-10.
[29] ASO Group: The resund Bridge, May 1996
[30] Vgverket: Sunningeleden the loveliest short cut!
[31] Aas, Jakobsen: Tender Documents for Kolbk Bridge, 1998
[32] Volke, E., and Rademacher, C.-H.,: Nordbrcke Mannheim-Ludwigshafen (Kurt-Schumacher-Brcke). Der Stahlbau 42 (1973), p. 97 - 105, 138 - 152, 161 - 172.
[33] Modemann, J., and Thnnissen, K.: Die neue Rheinbrcke Dsseldorf-Flehe / NeussUedesheim. Bauingenieur 54 (1979), p. 1 - 12.
[34] Kahmann, R., Koger, E.: Die neue Tjrnbrcke - Konstruktion, Statik und Montage der
Stahlkonstruktion. Bauingenieur 57 (1982), p. 379 - 388.
[35] Schmackpfeffer, H.: Rheinbrcke Emscherschnellweg - Bauausfhrung der Strombrcke in
Stahl. Bauingenieur 65 (1990) p. 453 - 462.
[36] Ohasi, M. et al: Design of Complex Cable-Stayed Bridge. International Conference on
Cable-Stayed Bridges, Bangkok, November 1987.
[37] Virlogeux, M. et al: Design of the Normandie Bridge.Proceedings of the IABSE/FIP
Conference on Cable-Stayed and Suspension Bridges, Deauville, October 1994, Volume 1,
p. 605 - 630.
[38] Housku-Shikohu Bridge Authority: Tatara Bridge. Structural Engineering International
1/98, p. 12, 13.
[39] Grant, A.: The East Huntington Bridge - Design and Construction Highlights. PCI Journal,
vol. 32, no. 1 (1987), p. 20 - 29.
[40] Prime Piling. New Civil Engineer 17 November 1994, p. 38 - 42.
[41] Dussart, R.: Le Pont Massen Paris. TRAVAUX - Mars/Septembre 1970
[42] Leonhardt, F., Zellner, W., and Saul, R.: Zwei Schrgkabelbrcken fr Eisenbahn- und
Straenverkehr ber den Rio Paran (Argentinien). Der Stahlbau 48 (1979), p. 225 - 236,
272 - 277.
[43] Knox, H.S.G., and Walther, F.: The Open Steel Deck Cable-Stayed Bridge. Strait
Crossings 94, Krokeborg (ed.), 1994 Balkema, Rotterdam, p. 123 - 131.
[44] Zellner, W., Saul, R., and Svensson, H.: Recent Trend in the Design of Cable-Stayed
Bridges. Proceedings of the 12th IABSE Congress in Vancouver, B.C., September 1984,
p. 279 - 284.
[45] Saul, R.: Schrgseilbrcken im Verbund fr das Ausland. Stahlverbund-Brckenbau.
Tagungsband des Stahl-Informations-Zentrums, p. 75 - 84.
[46] Bridges in China. Editor: Xiang Haifan, Tongji University Press, A & U Publication (HK)
Ltd., 1994.
10
Comparison of Slab Participation: Assumed for Design vs. FEA

David D. BYERS
Steven T. HAGUE
Project Engineer,
HNTB Corporation
Project Engineer,
HNTB Corporation
Steven L. McCABE
David M. ROGOWSKI.
Manager, Highway Bridge Design
HNTB Corporation
Professor
The University of Kansas,
Lawrence, KS, USA
Summary
Results of a study in slab participation and resulting stress distribution in the concrete deck of
composite cable-stayed bridge systems are presented. Analytical models developed using the
ANSYS finite element analysis package have been investigated for typical span
arrangements, similar to those being designed and constructed in the United States. Particular
attention is given to the longitudinal stress distribution across the deck section and the
resulting effective slab width. Recommendations for the implementation of a modified
effective slab width procedure are referenced. Finally, stress results are compared for a fullscale cable-stayed bridge model using both the current method of practice as well as the
proposed modified method.
Problem Description
Bridge design has developed through the centuries in a fashion that continues to improve upon
the types of materials being used, as well as to use existing materials in a more efficient manner.
Thus, as new materials, analysis methods, design concepts and construction methods are
developed, they are frequently employed in bridges because of society's need for longer, more
durable spans that can be built within ever-tightening public budgets. However, with new
technologies such as the cable-stayed bridge, the rush to implement the concept frequently does
not permit the answering of all the important engineering questions prior to implementation. To
date, no information has been recorded in the literature that sheds light on the actual
longitudinal stress distribution in the concrete deck portion of any of the composite steel and
concrete cable-stayed bridges that have been constructed. Without this information,
determining the extent to which the concrete deck is participating in the resistance of external
force is unknown.
The use of Finite Element Methods (FEM) for design in civil-structural applications has been
slow in evolving, primarily due to the cost associated with engineering design time and the
general simplicity of most of the models encountered. In addition, proper modeling of
structures as large as a typical cable-stayed bridge structure requires modern software to be
pushed to its maximum capacity for operation. Consequently, modeling of structures of this
nature for design is generally performed using a two-dimensional (2-D) or three-dimensional
(3-D) direct stiffness model, making use of three degree of freedom (DOF) or six DOF nodes
respectively. Cable-stayed bridge design is no exception and makes almost exclusive use of
these less complex, direct stiffness analysis methods.
The focus of this paper is to investigate the issue of slab participation in composite steel and
concrete cable-stayed bridges and to determine the proper assumptions and guidelines that
should be adhered to by designers of these structures. Specifically, a comparison is presented
between the current method of practice for determining effective flange width and a proposed
modified method developed by the author.
Current Method for Determining Effective Slab Width

The analysis of any deck and beam system where a flanged compression region acts as a "wide"
compression element requires assumptions for design purposes as to the actual width of the
flange that can be assumed as acting together with the beam or girder. The problem for
composite concrete deck-steel girder systems is the notable difference in elastic moduli between
the two materials. For example, to obtain equilibrium when subjected to positive bending, the
steel girder resists tensile stress that must be reacted in compression by the relatively flexible
concrete. Additionally, as you moves away from the girder the compressive stress distribution
in the slab drops quickly due to shearing deformation of the flange elements near the web.
There is a great deal of data in the literature regarding the performance of conventional
flanged concrete systems under flexural load on rigid supports. These studies have been utilized
in developing guidelines for bridge structures through AASHTO (1994) as well as foreign
bridge codes such as the German Deutsche Norm (DIN 1075), (1981). The basic concept is one
of identifying the effective portion of the deck slab that can be realistically considered to resist
external force. Attempting to apply these code provisions to a real design process for a
cable-stayed bridge creates a great deal of latitude for interpretation on the part of the
designer. The current state of the art method of implementation involves applying the above
methods for determining effective slab width for both axial and bending forces separately,
thereby making use of two distinct values of effective width for use in stress computations on
the composite member.
Axial Force Effective Slab Width

One important issue that is not adequately addressed in the literature or design codes, but is
directly related to the total involvement of the slab, is the participation of the slab in resisting
axial forces that are present in the composite cable-stayed deck system. Effective width or
"cooperating slab width" for concentrated axial forces, similar to those introduced into the deck
system through the cables, is addressed as a separate issue in the codes. The AASHTO LRFD
(1994) Design Specifications, in dealing with segmental concrete box girder construction with
normal forces present, recommends an effective width of flange generated by intersecting 30
degree lines drawn from the edge of the concrete girder stem. This is shown in Fig. 4.6.2.6.2-4
of AASHTO (1994). The DIN 1075 specification offers a similar recommendation, however
with an angle of only 26 degrees.
Note that each of the longitudinal edge girder members will contain an effective width that
varies linearly with the distance from the point of application of the load. In design, it is
generally assumed that the portions of the deck located at a sufficient distance from the
anchorage point of the outer-most cable possesses an effective slab width equal to one-half
the total bridge deck width. That is to say; if the assumed linear stress distribution given by
Fig. 4.6.2.6.2-4 of AASHTO (1994) exceeds one-half the bridge deck, the entire deck is
assumed to be effective in resisting axial force.
Variations on this simplifying assumption require that the force influence from individual
cables near the point of interest be continuously monitored so that the proper portion of the
total axial force be applied to a different cross-sectional property. Modeling of this nature is
impractical, especially when considering moving loads such as those encountered for typical
truck loading.
Bending Effective Slab Width

Effective slab width for bending of beam sections with wide flanges presents the designer
with a greater difficulty. Although the code provisions outlined in both the AASHTO and
German DIN 1075 are nearly identical, the extent to which they can be accurately applied to
structures of this type is more uncertain. Using the abbreviations given by the AASHTO
LRFD code provisions, the effective slab width for bending, bmf or bms is computed as
follows.
From the geometry of the preliminary cross-section, the constant, b can be determined by the
use of Fig. 4.6.2.6.2-3 of AASHTO (1994). Depending on the width of overhang, engineering
judgement is required to determine to what extent the overhang portion of the slab should be
considered. Generally speaking, the difficulty of engaging this small portion of the deck,
which is often physically separated from the main deck by the cable connection plates, and
the relative size of the overhang in proportion to the overall bridge deck excludes it from
consideration as part of the effective slab.
Having established the available flange width, b, the notional span length, Li, is required
for each member. This parameter is not clearly defined in the code for beam structures
supported on elastic supports, such as cable-stayed bridges. However the AASHTO LRFD
code, section 4.6.2.6.1, offers the following guidance for beams supported on rigid supports:
The effective span length used in calculating effective
flange width may be taken as the actual span for
simply supported spans and the distance between
points of permanent load inflection for continuous
spans, as appropriate for either positive or negative
moments.
Taking this statement to its logical conclusion, the points of moment inflection due to a unit
load located at the point of interest can be used to establish the notional length, Li for all
points of interest along the length of the deck. Initially an estimate is made with regard to the
relative stiffness of the deck members in the stiffness model. Vertical unit nodal loads are
applied in sequence over the entire length of the bridge structure. After application of the unit
load at each location, the moment diagram is analyzed and the length of positive moment
adjacent to the point of interest is established, thus giving Li.
Having established the notional length, Li, for each of the edge girder members, Fig. 4.6.2.6.24 of AASHTO (1994) can be used to determine the effective flange coefficients, bf and bs,
depending on the location being considered. The coefficient bf is to be used for all interior
portions of the span as defined by Fig. 4.6.2.6.2-1 of AASHTO (1994)6 while bs is used at or
near rigid vertical support locations such as at the tower or anchor pier location. Again, the
codes fall short in determining the boundaries of such regions and engineering judgement is
required to complete the task.
Once the initial effective width is established, the notional length should be re-computed to
verify the original assumptions of the relative stiffness of the girder section. Generally, a
single iteration is all that is required to converge on a satisfactory value for effective width
that will not change the notional length significantly.
Now, the effective width for both bending and axial force has been computed and the model
is ready for analysis for all the various load combinations required in the code. Once the
member end forces have been determined for each of the required load combinations, stresses
are computed by applying the member end forces separately to both the axial composite
member and the bending composite member. Stresses caused by both axial and bending
forces are superimposed and compared to the code allowable values for strength and
serviceability.
These assumptions combine a great deal of information that is borrowed from various related
analysis methods. For example, the original material contained in section 4.6.2 of the
AASHTO LRFD code for Approximate Methods of Analysis appeared in the Guide
Specifications for Design and Construction of Segmental Concrete Bridges, (1989) as does
the German DIN specification. It is of particular interest to note that structures supported on
elastic restraints with combined axial and bending forces present are not addressed explicitly
anywhere in any of these design codes. Addressing this issue was one of the primary focuses
of this research
Modified Method for Effective Flange Width

Finite elements are used to investigate the interaction of the slab and steel girder system and its
influence on the deck behavior under immediate loading. Studies using the Finite Element
Method have been conducted by Byers (1999) to establish an effective representation of the
composite girder and slab system that is consistent with the known behavior. This study
includes three-dimensional modeling of the major components within the composite deck
system including the slab, edge girders and floor beams shown in Figs. 1 and 2.
Figure 1 Floor Beams and Edge Girder
Figure 2 Model with Tower and Back Span
Particular attention is given in the finite element analysis to the longitudinal stress distribution
across the section and the resulting effective slab width. Linear assumptions in typical
composite member design, using transformed sections, assumes that plane sections remain
plane during bending. Near the edge girder, analysis has indicated that a nonlinear strain
distribution exists across the deck section. A typical example is shown in Fig. 3.
Figure 3 Sample Longitudinal Slab Stress from Finite Element Analysis

18 full-scale models were developed to encompass the desired span and deck widths for
structures similar in size to those being constructed in the United States. Specifically, main
span lengths of 240 m, 300 m and 360 m were evaluated, each having accompanying deck
widths of 8 m, 12 m and 15 m. Along with these variable dimensions, all of the models
contain the following constant properties:
Slab Thickness
Floor Beam Spacing
Cable Spacing
Edge Girder Web
250 mm
5000 mm
15 000 mm
PL 28 x 1800 mm
Edge Girder Top Flange
PL 25 x 600 mm
Edge Girder Bottom Flange
PL 55 x 800 mm
Floor Beam Web
PL 12 x 1800 mm
Floor Beam Top Flange
PL 25 x 600 mm
Floor Beam Bottom Flange
PL 38 x 600 mm
Cable Modulus
200 000 Mpa
Steel Elastic Modulus
200 000 Mpa
Concrete Elastic Modulus
29 914 Mpa
Poissons Ratio,
0.30
The shear modulus for each material was computed using the usual equation for
homogeneous, isotropic materials given in Eq. (1).
G=
E
2(1 + )

For each of the models indicated, a series of calculations was performed. First of all, each
model was analyzed using the ANSYS finite element software. In the post-processing
portion of the analysis, sections were cut through the slab elements at each floor beam
location and 1/3 points between edges of floor beams. Stresses were recorded at the top and
bottom of the slab and imported into an Excel spreadsheet developed by the author. Here, the
stresses across the section, such as those shown in Fig. 3, were summed per Eq. (2) to give
the total force resisted by the slab at each section. Next, the effective slab width is computed
in accordance with Eq. (3).
n
Q = i t s bi
i =1
b eff =
max t s

!
Effective slab width was then plotted over the span length for each of the models.
From the onset, the goal of this research was to establish what effective width of slab should
be used to accurately determine stress and deflection. In an attempt to locate any general
trends or consistencies between the various spans and slab widths being investigated, the
computed effective slab widths were normalized with respect to span length (x/L) and with
respect to deck width (beff/b). Good correlation between the models appeared to exist which
allowed for a beginning point in establishing a modified effective width formulation.
Next, the task of determining if a generalized or modified solution could be found and what
form that solution might take was undertaken. Various shapes were attempted for use in both
the back span and main span models. Each attempt consisted of the following steps:
1. An effective slab width curve was assumed for the deck system
2. A two-dimensional plane frame model was assembled using 6-DOF composite beam
elements
3. The stick model was loaded and analyzed and the stress and deflection values stripped
and compared with those obtained in the finite element solution
This process was repeated until satisfactory correlation of the results between the modified
method and the finite element solution was obtained. Good correlation of the results was
obtained for all of the models analyzed by using the proposed modified effective slab width
shown in Fig. 4.
Figure 4 Proposed Modified Effective Slab Width
Comparison of Methods
The largest of the models analyzed in the finite element portion of this research was singled
out to make a comparison between the current method described above and the proposed
modified method. The center span is made up of symmetrical 12 cable spans of 15 000 mm
each and one additional 15 000-mm span between the outer-most cables (375 m tower to
tower).
The laborious task of computing effective slab width using the current method was
performed. It should be emphasized that this step required over 100 separate analyses of the
same structure using an assumed value for the effective slab width along with computation of
the effective span length for each member. Effective slab width for bending was then
computed using spreadsheets that make use of curve-fit approximations for values in Fig.
4.6.2.6.2-2 of AASHTO (1994). This process takes several days to complete. Effective slab
widths for axial load between the cables using the current method were computed as
described above.
Effective slab width was then computed using the proposed modified method indicated in Fig.
4. Again, the available width of slab, b, is equal to 15 000 mm and the thickness of the slab,
ts, is equal to 250 mm. This process, ignoring the two years of research, took less that ten
minutes to complete. Combined results of the effective slab widths used in the comparison
are shown in Fig. 5 showing the proper orientation with respect to the structure.
Figure 5 Effective Slab Width for Comparison Model

Findings from this comparison show that the methodology currently being employed gives
good results throughout the middle portion of both the back and main span of the structure.
The greatest area of inconsistency with the finite element model occurs in the slab stress near
the anchor pier support of the back span and the center portion of the main span near the
closure area. These variations are largely due to the manner in which the axial portion of the
load and the effective slab width for axial load are dealt with in the current method of design.
In the region near the center of the main span, a reduced effective slab width is used for both
axial force and bending by the current method as can be seen in Fig. 5. A comparison of the
top of steel stress for the main span is shown in Fig. 6. Note that the results labeled T187Modified b_eff represent the stresses obtained using the proposed modified method in a
conventional direct stiffness stick model.
It should be noted that in both of these regions, the stress values obtained by the current
method are conservative in nature, but raise questions as to the appropriateness of the
methods used for an economic design solution.
Figure 6 Top Flange Stress Comparison
Conclusions
In an age of ever-increasing complexity for the field of structural engineering, designers are
asked to perform more rigorous detailed investigations on atypical structures such as the
cable-stayed bridge. Therefore, the need to simplify broad categories of assumptions into a
more direct solution method is greater than ever.
In the early development of composite design methods for simple structures, generalizing
mathematical models were developed to explain such phenomena as shear lag and effective
breadth. These theoretical solutions, though exact, were dependent on multiple variables that
change with virtually every problem such as beam depth, flange width, span length, slab
thickness etc Given the intricacy of the solution methods presented, simplifying
assumptions were needed to allow engineers to perform repetitive design tasks on typical
structure types without being burdened by the complexity of the intermediate solutions.
Thus, design codes were established that provided a vehicle for the engineer to safely
establish intermediate computations, such as effective flange width, so that more time and
effort could be given to the more important big-picture items in the design.
Because of the high degree of indeterminacy presented by complex bridge structures like
cable-stayed bridges, direct mathematical theory becomes too complex to offer any practical
consideration for solution. Finite elements of an elastic continuum, such as the ones used in
this research, offer approximate solutions that can be used in a fashion similar to the early
mathematical models mentioned above. This research has provided a first attempt at
establishing a simplified method of analysis for determining the effective slab width to be
used in the design of composite cable-stayed bridges comprised of steel and concrete sections.
At the present time, no research of this kind has been previously recorded in the literature.
Using the modified effective slab width described above, good correlation of the results are
obtained when comparing both stress and displacement between the simplified direct stiffness
models and the full finite element models. In addition, little difference exists in the results
when comparing the current method to that of the proposed modified method except as noted
above. Therefore, it appears the modified effective slab width method proposed provides a
suitable tool that can be expediently used by designers to predict both stress and deflection of
composite cable-stayed bridges.
Based on the results of this research, it is recommended that the effective slab width for
composite cable-stayed bridges meeting the parameters and limitations described by Byers
(1999) may be accurately estimated using the guidelines presented in Fig. 4.
Further research into localized areas near the cable to deck connection and the portion of the
deck between the cables at the center of the main span will provide additional insight into
design parameters that can be implemented by engineers.
References
[1].
AASHTO (1994), LRFD Bridge Design Specification, American Association of State

Highway and Transportation Officials, First Edition.
[2].
D.D. Byers, Evaluation of Effective Slab Width for Composite Cable-Stayed Bridge
Design, Ph.D. Dissertation, The University of Kansas, 1999
[3].
DEUTSCHE NORM, German DIN 1075, April, 1981, DIN 1072, Din 1076, Current.
Yamuna Cable Stayed Bridge at Allahabad/Naini, India

Ejgil VEJE
M.Sc. Civil Engineer
COWI AS
Lyngby, Denmark
Poul Mller NIELSEN

COWI AS
Lyngby, Denmark
Flemming PEDERSEN
COWI AS
Lyngby, Denmark
Kent FUGLSANG
COWI AS
Lyngby, Denmark
1 Introduction
Yamuna is a tributary of Ganga River. Very near the holy site at the confluence of Yamuna and
Ganga a new bridge will be built across Yamuna between the cities Allahabad and Naini.
The new river crossing has to cater for a large volume of traffic between Allahabad and Naini,
due to the industrial growth of Naini. It will also provide an important intra-state link to the cities
of Mirzapur and Varanasi.
The existing road which links Allahabad to National Highway 27 crosses Yamuna on a combined
railway and road bridge, built about 100 years ago. This old bridge has inadequate capacity in
regard to traffic and structural strength.
Initially the project preparation was carried out by Consulting Engineering Services (India) Ltd.,
New Delhi (CES) based on a solution with haunched box girders of 120m spans constructed by
free cantilevering for the deep channel portion of the river.
Subsequently, the Ministry of Surface Transport (MOST) received a loan from OECF, Japan and
engaged COWI-SPAN JV for a supplementary feasibility study and detailed design.
2 Feasibility Study
The objective of the supplementary feasibility study was to investigate larger cable stayed span
alternatives in concrete or steel. Since the crossing is near a historic site, aesthetic qualities of the
bridge were also considered to be important. Moreover a large span alternative could be a guide
for future bridges crossing wide rivers in India.
2.1 Bridge Alternatives

The river is notionally divided
into a deep channel portion (approximately 600m wide) and a
shallow channel portion to include
flood plains.
Concrete Alternatives: Span Arrangement: 60-115-260-116-60m
a) Deck section A with different
arrangements of the pylons, on
one or two wells under each pylon with double plane semifan or
harp cable stay arrangement
b) Deck section B or C with pylons
on one or two wells with double
plane semifan cable stay arrangement
c) Deck section D with twin pylons
on single wells with four planes
semifan cable stay arrangement
d) Deck section E with two vertical
cable stay planes
e) Deck section F with one leg pylons on single wells and with a
single central plane harp cable
stay arrangement
Figure 1 Alternative Cross Sections, Feasibility Study
The requirement of 260m as minimum of the central span arose to accommodate the horizontal navigational clearance of 240m corresponding with two 120m spans of the initial haunched
box girder solution. Two 60m end spans are introduced, because the anchor stays are distributed on both sides of the anchor piers to reduce bending moments in the critical parts of the
115m spans. Furthermore, the weight of the 60m spans results in avoiding upward forces in
the anchor piers. Finally, the southern 60m span gives a good transition to the following approach spans (the part of the bridge crossing the shallow channel portion of the river).
Concrete Alternative: Span Arrangement: 145-320-145m
f)
Deck section C with double plane semifan cable stay arrangement.
The larger span alternative proved to be approximately 20% more expensive than the smaller
span alternative b with deck section C.
Steel alternative: Span Arrangement 60-115-260-115-60m

A solution with orthotropic steel deck was considered. This solution proved to be approximately
25% more expensive than the concrete alternative b, with deck section C.
Composite Deck alternative: Span Arrangement: 60-115-260-115-60m
A composite solution with concrete deck and longitudinal girders and cross girders of steel was
considered. This solution proved to be approximately 20% more expensive than the concrete,
alternative b, with deck section C.
Figure 2. Alternative Pylon Solutions

2.2 Optional Solution
The harp cable stay system and the semifan cable stay system were compared and the semifan
cable stay arrangement proved to be 15% cheaper than the harp cable stay arrangement.
The weight of the bridge deck was a deciding factor for selection of the optimum cross section,
because increase in weight would increase the cable stay cost, which was estimated to be in the
order of 40% of the total costs. Cross section C was most advantageous. The ratio of the permanent load of the different deck solutions as compared to deck solution C was found to be 1.38,
1.17, 1.30, 1.22 and 1.38 for deck solution A, B, D, E and F, respectively.
Also the pylons for deck solution C have better proportions than the pylons for solutions with a
larger distance between the cable stay planes. The pylon legs below the deck level are inclined to
minimise dimensions of the well foundations. A single well for each pylon was preferred, as one
well below each pylon leg could result in extra stresses in the pylon frame in case of differential
settlements. Therefore, this type of pylon with semifan arrangement of cable stays in two vertical
planes with deck section C was recommended to be adopted for the detailed design.
For feasibility investigations, a cost comparison for the two bridge options, cable stayed versus
haunched girders, has been worked out considering unit prices from similar works recently executed in India.
The costs for the bridge modules exclusive contingencies etc. were (April 1996) estimated to:
Cost of the cable stayed solution
Cost of the haunched concrete girder solution:
792 Million Rupies

869 Million Rupies
The main reason for the lower cost of the cable stayed solution was that the length of the viaduct
on the Allahabad side could be reduced, because the small depth of the cable stayed deck above
the navigation channel allowed for a lowering of the longitudinal road profile.
The lower longitudinal profile also results in lower running cost for vehicles passing the bridge and less pollution.
Figure 3 shows the general arrangement of the two above-mentioned alternatives.
Figure 3. General Arrangement of Cable Stayed Solution and Box Girder Solution
3 Design Criteria
The design of the concrete structures has been based on the CEB-FIP Model Code 1990, while
for foundation design Indian Standards and Codes have been used.
3.1 Loads
Traffic loads are in accordance with current Indian Codes, i.e. 2 lanes of Class 70R or 4 lanes of
class A according to IRC:6-1966.
As a special case an accidental crowd loading of 5 kN/m2 has been considered to act on the entire
bridge deck due to religious festivals where millions of pilgrims gather at the confluence of Yamuna and Ganga river near the bridge site.
Cable stays are examined for fatigue load of 0.5 x class A traffic load with 2 x 106 stress cycles.
Wind loads are based on IS 875 (Part 3) - 1987. The basic wind speed (peak gust velocity for 100
years return period averaged over about 3 sec in 10m height) is 47m/sec. Flutter is checked for a
wind speed of 1.45 times the hourly mean wind speed at the girder level.
Water current forces are based on a maximum velocity of 3.0 m/s. Ship impact forces of 10 MN
and 5 MN perpendicular to and parallel to the bridge alignment, respectively have been considered for the design of he pylon foundations.
Horizontal forces from earthquake of
4.5% of gravity have been considered
in accordance with the seismic zone
defined in IRC:6-1966.
3.2 Concrete Structures
In accordance with CEB-FIP Model
Code 1990, exposure class 2a, the
maximum allowed crack widths are
0.30mm for reinforced concrete and
0.20mm for pre-stressed concrete. The
loads combinations in SLS for frequent situations are used for check of
crack widths, see figure 4.
The design is based on a characteristic
compression strength of fck = 40 Mpa
(cylinder strength) for the pylons and
the bridge girder.
The partial material factors for ULS
(Accidental load cases) m are:
Concrete: m,c = 1.5 (1.2)
Reinforcement: m,s = 1.15 (1.0)
Concrete compression stress limitations in serviceability state are:
c 0.6 f ck , rare combinations

c 0.4 f ck , quasi permanent situation
The load combinations considered are
shown in Figure 4.
Figure 4. Load Combinations
3.3 Cable Stays
The cable stay force for permanent load + full live load 0.45 times the minimum breaking load.
It is possible to remove the cable stay one at a time for replacement without restrictions for the
traffic. During cable replacement, all normal requirements for ULS and SLS are fulfilled, except
that 25% increase in cable stay forces in stays next to the stay being replaced is allowed.
A sudden rupture of a cable stay is assumed to act in combination with full live load. For this
situation 50% increase in cable stay forces are allowed.
4 Final Design
The bridge is divided into tree modules, see figure 3.
Module 1:
From pier P15 to pier P20 (the cable stayed module): Span arrangement: 60-115-260-115-60m
Module 2:
From pier P20 to abutment A2 on the Naini side: Span arrangement: 9x60+45m
Module3:
From abutment A1 on the Allahabad side to pier P15: Span arrangement: 30+13x25+20m
4.1 Superstructure
The 260m main span and the 115m side spans have two longitudinal girders (18.2m apart),
1.37m deep and 1.4m wide supported by cable stays per 10m.
The side span girder in module 2 has the same principle arrangement as the girder in the cable
stayed module except that the girder depth is 3.5m instead of 1.37m.
The 26m wide and 250mm thick deck slab is supported on cross beams per 5m as shown in figure 5.
The cross beams are pre-stressed by two tendons having 12 15.7mm strands tensioned from one
end.
In the central part of the main span each longitudinal girder is provided with 12 tendons consisting of 19 15.7mm strands. They are tensioned after the closure joint at mid span is cast.
In the side spans near the anchor piers each longitudinal girder is provided with 10 tendons consisting of 19 15.7mm strands. They are tensioned after the closure joint at the anchorage piers
are cast.
Figure 5. Section in Main Span, Module 1
4.2 Pylons
The pylons have slender solid legs
above the deck with dimensions as
shown in figure 6.
In the longitudinal direction of the
bridge the thickness of the legs varies
from 2.50m at the top of the pylon to
4.00m just above the lower cross beam.
The lower part of the pylon legs is hexagonal in shape.
The upper cross beam of the pylon is
solid while the lower cross beam is
hollow with a provision for access.
Both cross beams are post tensioned.
The upper cross beam is provided with
6 tendons, and the lower with 20 tendons of 19 No. 15.7mm strands.
4.3 Cable Stays
The cable stays are galvanised locked
coil ropes with diameters between
76mm and 116mm. The minimum
breaking load varies between 5.77 MN
and 13.60 MN.
The wire material has u = 1570 MPa
and E-modulus = 160000 MPa.
The outer 3 layers of the wirer in the
cable stays have a Z-shaped cross section.
Figure 6. Section Through Pylon
At the lower anchorage, the cable stays have sockets with thread and nut. The cable stay forces
are transferred to the longitudinal girders through steel plates. The cable stays pass through the
girders in steel pipes projecting 1.8m above the girders. At the top of the steel pipes neoprene
dampers and neoprene covers are arranged.
At the top of the pylons, the stay cables
are anchored by fork sockets to thick steel
plates protruding from the pylon legs, see
figure 7.
4.4 Foundation
The pylon foundations are double Dshaped open wells of 10x20m size which
shall be sunk approximately 40m below
the river bed by dredging inside the well.
After sinking to the required depth an underwater concrete bottom plug will be
cast.
The foundation for the piers in module 1
and module 2 are circular wells with 7.5m
outer diameter.
Bridge module 3 piers are founded on
1.2m diameter bored piles.
5 Construction Aspects
The anticipated construction method is
illustrated in figure 8.
In stage 1, the spans P15-P16 and P19P29 are cast and posttensioned.
15m bridge deck at each pylon (supported
on the lower pylon cross beam), is cast
and the first stays are erected and partly
tensioned.
Figure 7. Steel Anchorage at Pylon Top

In stage 2, cantilever equipment for construction of 10m sections of the deck is erected. Part of
the deck sections are anticipated to be pre-cast elements, which are transported by barge to the
site, lifted in position and supported on the cantilever equipment.
A stabilising system consisting of all four permanent cable stays 13' (see figure 8) is established
and the cable stays partly tensioned. The sliding bearings on all 4 pier shafts of P19 and P20 are
locked, so that longitudinal forces can be transferred.
In stage 3 the erection/casting of the 10m sections and the erection and tensioning of cable stays
are continued (first a part in the main span and then a part in each of the side spans).
One section on each side of a pylon will be erected/cast per 14 days. After casting the closure
joints at P16 an P19 side spans will be pre-stressed and the temporary fixing of the girder at the
pylons removed.
Figure 8. Construction Method for the Cable Stayed Bridge Deck of Module 1.
In stage 4, the erection/casting of 10m sections in the main span and erection and tensioning of
cable stays will be continued.
Before casting of the closure joint at mid main span, the cantilever equipment is removed, the
two ends of the cantilevers are temporarily connected, and bearings on P19 and P20 are made
movable. After casting of the closure joint, the main span is pre-stressed. Then cantilever slabs,
footways, edge beams and verges are cast, railings, safety barriers and expansion joint structures
installed, and surfacing executed.
The total period of construction is assumed to be 40 months.
6 Conclusion
The feasibility study carried out for a new major bridge crossing of the Yamuna River at Allahabad, India has revealed that the introduction of larger cable stayed spans are more cost effective than smaller spans of typically 120m for cantilever concrete box girders traditionally used
for major river bridge in India.
Design criteria for the cable stayed bridge has been based on CEB-FIP Model Code 1990, as the
Indian Codes are not suitable for design of the this type of structure.
The design of the bridge has focused on simplicity and consideration to local conditions and
technology. At the same time considerable effort to achieve a light and elegant impression of the
bridge has been aimed at. It is expected that the developed design may form the basis for several
new cable stayed bridges across the numerous rivers in India in he coming years.
Probabilistic FE analysis of a cable stayed composite bridge

A. de BOER
P.H. WAARTS
M.Sc.
Civ. Eng. Div of RWS,
Utrecht, The Netherlands
M.Sc.
TU Delft & TNO,
Delft, The Netherlands
Ane de Boer, born in 1951,

graduated as a Civil
Engineer from TU Delft in
1988. At present he is senior
structural research engineer
in the civil engineering
division of the Dutch
Ministry of Transport,
Public Works and Water
Management.
Paul Waarts, born in 1961,

got his civil engineering
degree at TU Delft in
1988. At present he is a
senior project engineer in
the department of
structural dynamics and
reliability engineering at
TNO and research assistant
at TU Delft, faculty of civil
engineering.
Summary
This paper describes the design of a new cable stayed composite bridge near Kampen in the
Netherlands. In the design process, the safety of bridges is insured by means of partial safety
factors for both strength and load parameters. As a result it is generally accepted that the
structure as a whole matches the desired probability of failure. In this paper another method is
followed. A full probabilistic analysis on the complete composite structure is performed using
FE analysis. The paper described the design of the Kampen bridge and the full probabilistic
study. It is concluded that the computed safety of the bridge is well above the required safety.
1.
Introduction
In 1998 a new cable stayed bridge was designed near the city of Kampen in the Netherlands.
Figure 1 shows an artist impression of the cable stayed bridge to be built. The bridge has been
designed by the civil engineering division of the Dutch Ministry of Transport, Public Works
and Water Management. The cross-section of the main span has a composite character. The
main span is built up out of a beam grid of steel. The concrete slab on top of the beam grid is
initially used as compression zone in the total cross section of the bridge deck.
In the design process, the safety of bridges is insured by means of partial safety factors for
both strength and load parameters. As a result it is generally accepted that the structure as a
whole matches the desired probability of failure. The safety factor was often based on
common practice developed over a long term of experience of successful design. For new
designs and new materials or loads, the experience only partly exist.
Figure 1: Artist impression of the Kampen bridge

The combination of materials (composite structures) is an example of such a new design.
Johnson and Zhang [1,2] for example have analysed this behaviour. They derived safety
factors for the composite structure. For structures different from the ones analysed by Johnson
and Zhang, other safety factors have to be derived to insure the bridge safety. In this paper
another method is followed. A full probabilistic analysis on the complete composite structure
is performed using FE analysis.
A full probabilistic analysis used to be a too much computational effort. As computers
become faster and faster and alternative computational methods of probabilistic analysis are
used, it becomes feasible to carry out probabilistic computations within a design office. This
makes it possible to directly analyse (bridge) structures in a probabilistic way. In cooperation
with TNO, Delft University of Technology has written a computer code to calculate the failure
probability of structures. The civil engineering division of the Ministry of Public Works is a
sponsor of this project, within the framework of safety aspects of structures. Probabilistic
methods have been implemented in a pilot version of the existing TNO Finite Element code
DIANA [3], which already includes physically and geometrically non-linear behaviour of the
structure.
In the probabilistic finite element code, all variables like material properties, geometrical
dimensions and loads can be treated as random variables. The probabilistic method used in
this case is based on the principle of adaptive conditional directional sampling [4].
2. Cable stayed bridge structure

A side view of the cable-stayed structure is given in Figure 2. The bridge was designed in
1998 with use of partial safety factors. The figure shows a main span of 148.4 m, a side span
of 92.5 m, while the height of the pylon reaches 70 m above the bridge deck. The netto width
of the bridge deck is 17 m, four traffic lanes of 3.25 m each and two maintenance lanes with a
width of 2 m each. The bridge is supported at the both ends, the pylon and in the middle of the
2
left side span. All supports are assumed to be springs, representing the soils stiffness. This
paper concentrates on the main span and its probabilistic behaviour of the composite structure.
70 m
92.5 m
148.4 m
Figure 2: Side view of the cable-stayed bridge.

Side span
The concrete side span has a post-tensioned pre-stressed character. The cross-section of the
side span is given in Figure 3 and shows a middle rectangular part with 8 holes. Each hole has
a diameter of 0.9 m. The side parts have a trapezoid shape and do not contain holes. The
middle rectangular part of the cross section accommodates 9 groups, with in total 30 posttensioned pre-stressed cables.
0.22 m
1.25 m
2.6 m
2.6 m
13.5 m
Figure 3: Cross section of the concrete side span

Main span
The cross-section of the main span has a composite character. The 148.4 m main span is build
up with a beam grid of steel and is subdivided in sections with a length of 14.5 meter. Each
bridge deck section contains two main girders and four cross beam girders (see Figure 4). On
top of the beam grid lies a concrete slab with a thickness of 0.25 m. The concrete slab is
initially used as compression zone in the total cross section of the bridge deck. The main
girders have a centre to centre distance of 13 meter. The cross beam girders on both sides of
the connection between the stay cable and the bridge deck are heavier then the other two cross
beam girders of the section.
The connection of the cable stay centre is 1 m outside the centre of the main girder. At this
point also a rather rigid grid has been used to distribute the stay cable force into the bridge
deck. A top view of the FE mesh of a connection section is shown in Figure 4.
14.5 m
2m
main girder
cross beam
stay cables
13 m
main girder
2m
5 * 3.625 m
Figure 4: Top view steel section of the beam grid

The main girders of the steel beam grid have a rigid connection with the side span. The
connection between the concrete bridge deck and the beam grid is designed with a so-called
stud connection (see Figure 6 for FE mesh) which is typical for steel concrete composite
structures. Studs are small pins welded on the upper flanges of the main girder and the cross
beam girders of the grid of steel. The main and cross girders of the beam grid are welded
profiles. Near the connection of the stay cable and the deck the lower flange of the cross girder
on both sides are heavier then the other cross girders (see Figure 7). The measures are given in
Table 1.
0.25 m
2m
13 m
2m
Figure 5: Cross section main span

Profile
Height
main girder
1.5
light cross girder 1.1
heavy cross girder 1.1
Upper
flange
2.2
0.5
0.5
Width
Lower
flange
1.0
0.5
0.75
Web
0.020
0.014
0.014
Table 1: Measures of the beam grid under main span [m]
Thickness
Upper
Lower flange
flange
0.020
0.040
0.020
0.030
0.020
0.040
concrete
interface plane
with studs
cross beam girder
Z
Y
X
main girder
Figure 6 : FE mesh of half steel section main span

Concrete deck
stay cable
cross beam girder
Z
Y
main girder
Figure 7: FE mesh near the connection with the stay cables

Pylon and anchor block
The pylon is made from concrete and has an oval cross-section. The top of the pylon measures
about 8.6 times 13.0 m2 and anchors the 2 times 9 stay cables in the top of the pylon. The base
of both pylon near the foundation has a width of 25 m centre to centre.
The anchor block support on the side span has rough dimensions of 17 times 13 m2, with a
height of 5 m. The length of the support block (17 m) is divided into 5 rooms, each room has a
length of 2.8 m.
Stay cables
Typical in the process of the cable stayed bridge design, is the finding of the prestress of the
24 stay cables. Within the bridge module of the FE code, these prestress of all different stay
cables is easily found by balancing the vertical deflections of the bridge deck and the
horizontal deflection of the top of the pylon at the connection point of the stay cables to these
structure parts. Within this probabilistic study, only the linear static analysis option of
balancing is used, however a nonlinear can be used as well. The balancing load case is selfweight of the bridge structure. A view of the total 3-D model is presented in Figure 8.
Y
X
Figure 8: Iso view 3-D model cable stay bridge
3.
Probabilistic method
Reliability methods compute the probability of failure given a limit state model and stochastic
parameters. Limit states might be for instance exceedance of yield stress in a member,
exceedance of maximum deformation or global collapse. Well-known methods for computing
the reliability are Monte Carlo simulation (MC) [7] and the First Order Reliability Method
(FORM) [5]. In this paper an unusual method is applied: Directional Sampling [6].
Most structural reliability problems have many stochastic variables. Up to 100 or 1000
stochastic variables is common place. For standard directional sampling about 20000 samples
are necessary [4]. For every sample several FE computations have to be carried out. The
adaptive conditional direction sampling (ACDS) is introduced to speed up the computations.
In short the improvement to the standard directional sampling lies in the use of FE for the
important directions and a response surface for less important directions. In practise this
means that after the response surface is constructed, only a few FE computations have to be
carried out.
For the construction of the response surface all variables are varied individually and increased
or decreased until failure. A FE model with n stochastic variables gives 2n (directional)
samples in the principal directions. Consequently a quadratic response surface is fitted to
6
these results. Following this starting procedure the random directional sampling takes place.
The response surface is used in case of a large distance from the origin to the response
surface. FE computations are used to calculate the real distance in case of a small distance
from the origin to the response surface. In that case the response surface is updated (adapted).
Influence factors give insight on the importance of stochastic variables on the limit state. After
finishing the directional sampling procedure, the influence factors are computed by means
of a FORM analysis on the response surface.
4.
Stochastic properties of the 2-D and 3-D model
The probabilistic analysis is first performed on a 2-D bridge model. The analysis on such a
model is much faster then on the full 3-D model. The results of the 2-D model are checked by
means of an analysis on a 3-D model. The (stochastic) properties of the 3-D model are the
same as the 2-D model. Additional to the 2-D model, the probabilistic analysis of the 3-D
model gives more information over the width of the structure. Also the torsion behaviour of
the bridge deck is taken into account in this analysis.
Variable
Deck side span
Econcrete
A
I
q (load)
Deck main span
Econcrete
Esteel
Thickness
q (load)
Footing pylon
Econcrete
Pylon
Econcrete
Stay cables
E
Prestress
u
Supports (springs)
Vertical translation side span
Rotation under pylon foot
lower bound
Unit
38500
0.2252
0.1991
45.
0.48
0.002252
0.01991
4
1
0
0.01
0
N/mm2
m2
m4
kN/m
38500
2.1 105
0.25
45
0.48
8.4 103
0.005
4
1
1
0.10
0
N/mm2
N/mm2
m
kN/m
38500
0.48
N/mm2
38500
0.48
N/mm2
1.95 105 7.8 103

nom. val. 0.1*nom. val.
1800
100
0.1
0.
700
N/mm2
N/mm2
N/mm2
nom. val. 0.1* nom. val.

nom. val. 0.1* nom. val.
0.5 * nom val.

0.5 * nom.val
N/mm2
N/mm2
Table 2: Input of probabilistic properties

The traffic loads in the bridge are derived from [8]. The lifetime of the structure has been set
equal to 100 years. For this period a reliability index = 3.6 has been chosen according to the
Dutch Building regulations (NEN 6700). In [8], according to ISO 2394, a design value of the
load is found at a probability of exceedance of (-) =(-0.7 *3.6) = (-2.5)=0.0062 in 100
year. If the load process is ergodic, this corresponds to a load effect with a return period of
7
100/0.0062 = 16000 year. For a span length of 100 m and 4 lanes, [8] gives a distributed load
of 45 kN/m for a return period of 100 years. For a return period of 16000 years the distributed
load is equal to 55 kN/m. Assuming a normal distribution this means a mean load of 45 kN/m
and a standard deviation of 4 kN/m for the 100 year maximum.
The thickness of the concrete deck and Youngs moduli of steel and concrete are random
variables. Stochastic parameters are the prestress force of the cables and material properties of
the concrete pylon and pylon footing. Stochastic properties are summarised in table 2. Most
geometric and material parameters are derived from [9].
The ultimate limit state is overstressing of either the cables or concrete-steel deck. The
ultimate strength of the cables is supposed to be random as well. The ultimate strength in the
24 cables is supposed to be uncorrelated.
All variables presented in table 1 have a truncated normal distribution. The 2-D and 3-D
models have in total respectively 27 and 58 independent stochastic variables.
5.
Results of the 2-D and 3-D analysis.
The probability of failure of the bridge structure is computed by means of the ACDS
procedure. The 2-D model results is an expected reliability index = 5.7, with 95 %
confidence intervals: 5.5 < < 6.0. The reliability index corresponds with a probability of
failure Pf = 6 10-9. This reliability index is well above the required reliability according to the
Dutch building code (=3.6). The main influence to the probability of failure is found in the
thickness of the concrete deck (an influence factor =0.8 is found). Next the prestress of the
cables connected to the bridge deck (=0.4) is of importance. The ultimate stress of the cables
connected to the bridge deck is very important as well (=0.3). The cumulative influence of
the prestress of all cables is equal to = 0.55. Most other variables have an influence factor
< 0.10. The randomness of traffic load has little influence on the reliability index.
The 3-D model results in a reliability index = 5.05 (intervals 4.8 < < 5.4). The reliability
index corresponds with a probability of failure Pf = 4 10-7. The probability of failure of the 3D model is higher compared to the 2-D model because of the fact that more variables are
taken into account and more limit states can occur in the 3-D model. The main influence
factors are found in the prestress of the cables, = 0.7 and ultimate stress of some cables
(=0.6). Next the thickness of the concrete deck ( = 0.5) and traffic load (=0.2) are of
importance.
For the 2-D and 3-D model 476 and 274 FE computations were carried out respectively. The
total computational time for the 2-D and 3-D model on a HP 9000 130 MHz UNIX machine is
27 minutes and 4:10 hours respectively.
6.
Conclusions
The paper shows the possibility to perform a probabilistic analysis in the design environment.
A 2-D model takes half an hour computational time, which is acceptable in this environment.
In the definite design process, the elapse time of a 3-D model (4 hours) is also acceptable.
8
A 3-D model results in a lower reliability index compared to the 2-D analysis. The 3-D
model shows that there is an additional safety compared to the safety required by the building
codes. It may be concluded that this bridge, designed by using partial safety factors, is much
safer then required, at least for the limit states considered. The partial safety factors are mostly
based on an assumption that the influence factors on the loading are = 0.8 and =0.7 for
strength parameters. The results show that the traffic load has only an influence factor 0.2.
This explains some of the difference between the required and computed safety. Another
influence may be a conservative assumption of the systems behaviour by the designers.
Acknowledgement
We like to thank the members of the steel and concrete design offices of the civil engineering
division of the Ministry of Public Works, Transport and Water Management for the support at
the several FE models used in this paper.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
Johnson, R.P., D.J. Huang, Partial safety factors for composite beams in bending found
from test data, University of Warwick, Research report CE38, 1991
Johnson, R.P., D.J. Huang, Calibration of safety factors for composite steel and
concrete beams in bending, Proc. Engrs Structs & Bldgs, Aug. 1995
Diana Users Manuals, release 7.1, 1998
Waarts, P.H., Directional sampling and FORM in structural reliability by means of an
adapted response surface, TNOReport 98-CON-R0483, 1998
Hasofer, A.M., N.C. Lind, An exact and invariant first order reliability format, J. Eng.
Mech. Div., ASCE, Vol. 100, 1974, p. 111-121.
Bjerager, P., Probability integration by directional simulation, J. of Eng. Mech., Vol.
114, No. 8, 1988
Rubinstein, Simulation and the Monte Carlo method, John Wiley and Sons, New
York, 1981.
Vrouwenvelder, A.C.W.M, P.H. Waarts, Traffic loads on bridges, Structural
Engineering International, no. 3, IABSE, 1993
Vrouwenvelder, A.C.W.M, A.J.M. Siemes, Probabilistic calibration procedure for the
derivation of partial safety factors for the Netherlands building codes, Heron, vol. 32.
A Method For The Creep Analysis Of Composite Cable-Stayed Bridges
Gan XIA
Civil Engineer
Ruhr-Universitt
Bochum, Germany
Rolf KINDMANN
Professor Dr. -Ing.
Ruhr-Universitt
Bochum, Germany
Gan Xia, born 1963, received his Master

degree in civil engineering from Southeast
Univ., Nanjing, P.R. China. He is currently
a research assistant at Ruhr-University
Bochum. His research interests include the
time-dependent behaviour of composite
structures.
Rolf Kindmann, born 1947, worked for

one of the greatest German companies for
10 years. He was head of the technical
departments for the design and
construction of structures. Since 1990 he
is professor for steel and composite
structures at the University in Bochum.
Summary
This paper proposes a method for the time-dependent analysis of composite cable-stayed bridges.
Using this method creep analysis can be performed by applying finite element method in only one
time step and therefore it is useful for practical design. A fictitious transformed section for creep
analysis is introduced and equivalent loading vectors are derived from the newly obtained results of
creep analysis of generic composite sections. A numerical example of application is given for a
composite cable-stayed bridge.
1.
Introduction
Due to creep of concrete, redistribution of stresses and/or stress resultants, loss of prestress and
additional displacement in a composite structure made of steel and concrete may be occurred.
Therefore, the time-dependent analysis is unavoidable for designing composite structures.
For the creep analysis of complex composite structures by applying finite element method there are
two types of analysis methods with regard to time discretization, namely step-by-step [1] and onestep method [2,3]. As an internally heterogenous structure with external elastic restraints the exact
creep analysis of composite cable-stayed bridges has to be performed by using numerical step-bystep techniques with more than 100 calculation steps [1]. One-step method is apparently applicable
for most practical designs. Klement [2] proposed an approach which is based on some solutions by
using dischinger method (or rate of creep method) to analyse composite structures, while the
method proposed by Schade [3] is based on the age-adjusted effective modulus (AAEM). It should
be noted that the solutions for composite cross sections obtained by using dischinger method and
the usual AAEM method inevitably have errors, which in some cases are negligible, but in other
cases unnegligible. A new method has been in [4] developed for the creep analysis of generic
composite sections, which was direct based on the integral-type creep law (i.e. the superposition
equation) of concrete and mathematically exact on the usual hypotheses. Therefore, the accuracy of
the results by using the new method is better than ones by using the AAEM method or the
dischinger method.
Based on the new results, this paper proposes a one-step method for the creep analysis of complex
composite structures including composite cable-stayed bridges. The basic assumptions of this
paper are:
(1). Instantaneous strain and creep of concrete are linearly proportional to the applied stress, so
that the principle of superposition is applicable;
(2). Plane sections remain plane so that the total strain distribution is linear;
(3). Time development of internal forces due to creep and shrinkage in interval (t0,t) in a
statically indeterminate composite structure depends linearly on the creep coefficient (t,t0);
(4). Internal forces after creep are distributed longitudinally linearly along a finite beam element.
2.
Section properties
2.1
Properties of transformed section for elastic analysis
Elastic analysis of a composite section at the time t0 can be easily performed by using so-called
transformed cross section properties, which is illustrated in Fig. 1. The properties of the
transformed section can be calculated by the following formular:
Ec,Ac,Ic
Es,Ac,Ic
Es,Acr,Icr
Gc
ac
as
ac
Gi0
a
Es,As,Is
Gs
Es,As,Is
a)
b)
Gi
Es,As,Is
c)
Fig. 1 a) Original cross section b) Fictitious cross section for elastic analysis
c) Fictitious cross section for creep analysis
Ac
I
;
I cr = c
n0
n0
Ai 0 = As + Acr
I i 0 = I s + I cr + S i 0 a
Acr =
Si 0 = As a s = Acr a c =
(1)
As Acr a
Ai 0
where
n0 =
Es
E c0
= ratio of elasitic moduli,
Es, Ec0
As, Is
Ac, Acr
Ic, Icr
Ai0, Ii0
Si0
a, ac, as
2.2
= Youngs modulus of steel and of concrete at time t0,

= area and inertia moment of steel parts,
= area and reduced area of concrete part,
= inertia moment and reduced inertia moment of concrete part,
= area and inertia moment of transformed cross section,
= static moment for elastic analysis,
= distance between the centroids of steel, concrete part and transformed
cross section according to Fig. 1.
Properties of fictitious transformed section for creep analysis
Differing from the famous aging coefficients of the AAEM method, two corrected aging
coefficients N and M for a composite section were in [4] derived, which are the aging
coefficients with respect to the axial force Nc and bending moment Mc of concrete respectively
and make an exact description of the creep behaviour of concrete in composite sections possible.
The creep analysis of composite sections by using the method with the two aging coefficients N
and M developed in [4] is similar to that of the famous AAEM method, but the results are exact
under usual assumptions. For the purpose of the creep analysis of complex composite structures,
we can define the following two corrected age-adjusted effective moduli for concrete in a
composite section:
E c0
1 + N (t , t 0 )
E c0
=
1 + M (t , t 0 )
E cN =
E cM
(2)
By means of the corrected age-adjusted effective modulus EcN and EcM, a fictitious transformed
section can also be introduced for creep analysis, which is similar to the above mentioned
transformed section for elastic analysis and illustrated in Fig. 1c). The properties of the fictitious
transformed section for creep analysis may be similarly determined with:
Ac
I
;
I c = c
nA
nI
Ai = As + Ac
Ac =
I i = I s + I c + S i a
S i =
a = (
(3)
As Ac a
Ai
1
1
) As a
Ai Ai 0
and
Es
= n0 (1 + N t )
E cN
nA =
Es
nI =
= n0 (1 + M t )
E cM
(4)
In equation (3) and (4)

t=(t,t0)
nA, nI
Ac, Ic
Ai, Ii
Si
a
= creep coefficient,
= ratio of moduli for reduced area and reduced inertia moment of
concrete part, respectively,
= reduced area and reduced inertia moment of concrete part,
= area and inertia moment of transformed cross section for creep
analysis,
= static moment for creep analysis,
= distance between the centroids of the transformed cross sections for
elastic analysis and for creep analysis (see Fig. 1c).
The aging coefficients N and M have to be previously found out in order to determine the
properties of the transformed section for creep analysis. It should be explained that N and M for
a generic composite section depend not only on the material properties of steel and concrete on
the one hand and on the section geometrical properties on the other hand, but also on the type of
loads [4]. For the purpose of determining the change of internal forces in a statically
indeterminate structure due to creep and shrinkage, only the aging coefficients N and M for the
loads which increase from 0 to its end value in interval (t0,t) linearly with respect to the creep
coefficient t (basic assumption 3) are of interest. For the determination of the fictitious section
properties it is assumed that the aging coefficients for the varying axial force and varying bending
moment have the same value. The calculation of N and M for the varying bending moment will
be therefore discussed in the following.
There are two ways to determine the corrected aging coefficients N and M. The first way is
based on exact creep analysis of composite sections in which the integral-type creep law (i.e. the
superposition equation) of concrete and a step-by-step procedure are used [4]. The extensive results
according to the CEB MC 90 model [5] were reported in [4]. In this case, exact results for the
creep analysis of composite sections are given by adopting the corrected aging coefficients N
and M. The another way to determine N and M is approximate, nevertheless, in most cases
excellent results can be obtained by using the approximate N and M if the formula proposed by
Trost in [6] is used and the variation of elastic modulus of concrete can be neglected. The
expressions for the corrected aging coefficients N and M in this case are as follows [4]:
N =
1 I i 0 Acr a 1 ( 1 1 ) M c1 ( 1 2 )(1 M )c2 1
]
[
t Is
Si 0
M c1 + (1 M )c2
Ii0
Si 0 a
1 I cr Si 0 a
1
=
[
]
M
M
t Is
I i 0 ( 1 1 ) c1 ( 1 2 )(1 )c2 M
M c1 + (1 M )c2
(5)
where
As I s
Is
M =
;
Ai 0 ( I i 0 I cr )
I s + I cr
I
( 1 cr ) 2
Ii0
( I + I cr ) As
1 = s
=
;
1 2
I i 0 Ai 0
1,2 =
I i 0 Ai 0 + I s As I cr Acr
I A + I s As I cr Acr 2 I s As
( i0 i0
)
2 I i 0 Ai 0
2 I i 0 Ai 0
I i 0 Ai 0
c1 =
1 t
1+ 0 , 4 k v 1
1 e
1 t
c2 =
(6)
2 t
1+ 0 , 4 k v 2
1 e
2 t
1
1
+t
k v = +
with
2
4
k =1
v
fr t < 2
fr t 2
It is to be noted that the use of the Eqs. (5) and (6) is not restricted to the CEB MC 90 model and
these expressions are also applicable to other creep models, e.g. ACI 82 model.
3.
Creep analysis of statically indeterminate structures
The change of the internal forces due to creep in a statically indeterminate structure consisting of
heterogeneous sections may be calculated by using usual finite element program for elastic analysis.
But the matrix of stiffness has to be determined by using properties of the fictitious transformed
section for creep analysis (Eq.(3)) and for a usual beam element we have the following matrix of
stiffness:
[k ]
ij
E s Ai
l
0
0
=
E s Ai
12 E s I i
6 E s I i
l3
6 E s I i
l2
4 E s I i
l2
12 E s I i
l
12 E s I i
E s Ai
6 E s I i
l
E s Ai
6 E s I i
2 E s I i
l2
l
0
0
l3
6 E s I i
l2
0
12 E s I i
l3
6 E s I i
l2
6 E s I i
l2
2 E s I i
6 E s I i
l2
4 E s I i
l
0
(7)
where l is the length of the element ij.
The equivalent loading vectors can be derived from the results which were obtained in [4]. For
constant sustained load, for example, the axial force and bending moment increment of steel part
of every section Ns(t) and Ms(t) can be obtained, namely
N s ( t ) 1
=
M s ( t ) a s
0 1 0 T
1 a 1 T
B
nn
B
mn
Acr
T Ai 0
T 0
Si 0 As
I i 0 Ai 0
I cr
0
I i 0
B
nm
B
mm
Si 0
I i 0 N (t 0 )

I s M (t 0 )
I i 0
(8)
where
1 t
2 t
1
T =
[( 1 2 )e 1+ 0,4 kv 1 ( 1 1 )e 1+ 0 ,4 kv 2 ]
1 2
B
nn
B
nm
1 t
2 t
Si 0 I i 0
1+ 0 , 4 k v 1
1+ 0 , 4 k v 2
(e
)
=
e
1 2
TmnB =
( 1 1 )( 1 2 )
(e
( 1 2 ) Si 0 I i 0
2 t
1+ 0 , 4 k v 2
1 t
1+ 0 , 4 k v 1
(9)
)
2 t
B
mm
1 t
1
[( 1 2 )e 1+ 0,4 k v 2 ( 1 1 )e 1+ 0,4 kv 1 ]
=
1 2
The change of the strain and curvature (t) and (t) at the centroid G of the transformed cross
sections for creep analysis can be easily found out because of the linear elastic property of steel
(see Fig. 2):
1
(t ) E s As
=
( t ) 0
a a s
E s I s N s ( t )
1 M s (t )
E s I s
The equivalent loading vector for constant

sustained load can be then determined by
{P}e
i + j
E s Ai
2
6E I
s i ( )
i
j
l
E s I i (4 i 2 j )
=
+ j
E s Ai i
2
6 E s I i
( i j )
E I l ( 4 2 )
j
i
s i
(10)
Gc
G
G
(11)
Gs
x
as
Fig. 2 Strain distribution in a cross section
where
i, i
j, j
4.
= strain and curvature increment of the left node i of a beam element,

= strain and curvature increment of the right node j of a beam element.
Calculation procedure and numerical example
The creep analysis of statically indeterminate composite structures using the here presented method
is divided into the following steps:
1. Calculation of the properties of the fictitious transformed section for elastic analysis and for
creep analysis according to Eqs. (1) and (3);
2. Elastic analysis of the structure for various loads by using usual finite element program, in
which the properties of the fictitious transformed section for elastic analysis (Ai0, Ii0) are used;
3. Creep analysis of the structure for sustained loads using the same finite element program but
by means of the fictitious matrix of stiffness Eq.(7) and the equivalent loading vectors e.g. for
constant sustained load according to Eq. (11), in which the internal forces N(t0) and M(t0) are
the results of the elastic analysis (step 2). It should be noted that the structure for creep
analysis has a system line which is different from that for elastic analysis(i.e. a remove of a);
4. The results from step 2 together with that from step 3 as basis of the proof and design.
To show its application, redistribution of
internal forces due to creep in a cablestayed bridge (see Fig. 4) with composite
cross section shown in Fig. 3 is evaluated
by using the presented method. The
properties of the cross section, the fictitious
transformed section for elastic analysis and
for creep analysis are shown in Table 1.
The elastic internal forces N(t0), M(t0) and
the change of internal forces N, M due
to creep for sustained load g=300 kN/m,
which are the results of first order analysis
by using a finite element program, are
illustrated in Fig. 4.
4000mm
250mm
40020
200014
60080
Fig. 3 Cross section
It should be said that the influence of geometric non-linearity in the stiffening girder (second
order effect) has to be taken into account if higher precision is required.
Ac
cm2
10,000
Ic
cm4
5.21105
As
Is
cm2
cm4
840 4.86106
a
cm
164.3
Ai0
cm2
2745
Ii0
cm4
2.07107
Si0
cm3
9.58104
as
cm
114.0
Ai
cm2
1626
Ii
cm4
1.59107
a
cm
50.3
Tabel 1. Properties of cross section and fictitious transformed sections
5.
Concluding Remarks
In order to perform the creep analysis of statically indeterminate composite structures by applying
the presented method, only the properties of the fictitious transformed sections and equivalent
loading vectors have to be implemented in a usual finite element program for the elastic analysis of
structures. The creep effects can be then evaluated in only one time step and therefore the presented
method is useful for practical design.
a=0 for elastic analysis

a=34.6 cm for creep analysis
24.0 m
a
20.0 m
20.0 m
20.0 m
M(t 0)
48037 kNm
19463 kNm
-3622 kN
-19427 kNm
-9004 kN
N(t 0)
-19104 kNm -27287 kNm -26772 kNm

M
-934 kN
-2425 kN
N
Fig. 4 Example
References
[1]
Jurkiewiez, B. ; Destrebecq, J.-F.: A Global Approach to Account for Time Effects

in Composite Structures. Composite Construction-Conventional and Innovative
(Conference Report), Innsbruck, pp 229-234, 1997.
[2]
Klement, P.: Die Berechnung komplizierter Verbundstabwerke unter Verwendung

blicher Programme. Der Bauingenieur 60 (1985).
[3]
Schade, D.: Zur Berechnung der Schnittkraftumlagerungen infolge von Kriechen und
Schwinden des Betons bei statisch unbestimmten Stabwerken mit
Verbundquerschnitten. Der Stahlbau 49 (1980).
8
[4]
Xia, G.: Zur wirklichkeitsnahen Berechnung von Verbundkonstruktionen unter

Bercksichtigung des zeitabhngigen Betonverhaltens. Dissertation, Ruhr-Universitt
Bochum, Oktober 1998.
[5]
CEB: CEB-FIP Model Code 1990. Final Draft, Bulletin DInformation No. 203,
Lausanne, Juli 1991.
[6]
Trost, H.: Dischingers grundlegende Arbeiten und neuere Erkenntnisse ber die
Auswirkungen des zeitabhngigen Werkstoffverhaltens in vorgespannten und nicht
vorgespannten Stahlbetonkonstruktionen. Spannweite der Gedanken, Springer-Verlag,
Berlin, 1987.
Composite Structures in the resund Bridge
Henrik CHRISTENSEN
Civil Engineer MSc
Design Manager
resundskonsortiet
Malm, Sweden
Henrik Christensen received his
degree from the Technical
University of Denmark in 1987.
He joined resundskonsortiet
in 1994 and is Design Manager
for the bridge.
1.
Kaj MADSEN
Civil Engineer MSc
Director
Gimsing & Madsen A/S
Horsens, Denmark
Kaj Madsen received his degree
from the Technical University of
Denmark in 1959. He is a technical
advisor to the ASO Group,
the bridge consultant
to resundskonsortiet.
Christian Riis PETERSEN

Civil Engineer MSc
Chief Engineer
ISC Consulting Engineers A/S
Copenhagen, Denmark
Christian Riis Petersen received his
University of Denmark in 1978. He
is technical advisor to the ASO
Group, the bridge consultant
to resundskonsortiet..
Introduction
The resund Bridge was developed for a design competition in 1993 by the ASO Group (Ove Arup &
Partners, SETEC TPI, Gimsing & Madsen and ISC). The design was subsequently modified to a
conceptual design and the ASO Group was appointed as the Owner's bridge consultant for carrying the
concept to the tender stage. The bridge was tendered in 1995 as a "Detailed Design and Construct"
contract, where the main visible dimensions were fixed by the Owner through contractual Definition
Drawings. The bridge contract was in November 1995 awarded to Sundlink Contractors HB (Skanska,
Hochtief, Monberg & Thorsen and Hjgaard & Schultz). The detailed design for the Contractor is by
CV Joint Venture (COWI and VBB).
The bridge is 7.8 km long. The 1.1 km long cable-stayed High Bridge consists of a central navigation
span with two side spans each side. Minimum vertical navigational clearance of the main span is 57 m.
The High Bridge is connected to the artificial island, Pepperholm, and the Swedish coast at Lernacken
via in all 6.7 km long Approach Bridges. The bridge deck is in two levels with a dual two-lane
motorway at the top and a two-track railway at the bottom. The two levels are connected by two
parallel Warren type steel trusses. The 20 m bay length of the truss is constant along the bridge.
The conceptual design included both an upper road deck in concrete and a lower railway deck in
concrete, but the tenderers were free to choose steel decks instead of concrete decks. The final solution
was, however, that the concrete decks were preferred for the Approach Bridges and a lower steel deck
was chosen for the High Bridge, where the weight saving was found more important than the otherwise
cost effective composite solution. Road and railway bridges are to an increasing extent constructed as
composite structures. In some of the most recent composite bridges a double composite concept has
been introduced, as the bottom chord has been strengthened by a concrete slab in regions at the inner
supports. In this context the Approach Bridges of the resund Bridge are double composite over the
entire length.
As extensive use of the composite principle was expected, a number of studies and investigations for
the composite solution were carried out in the pre-tender stage along with the development of the
design criteria to be used in the detailed design of the bridge structure.
2.
The Concrete Decks
2.1
The Road Deck
The upper concrete road deck is made both in the conceptual design and in the final design with
bonded transverse prestressing tendons and normal reinforcement in the longitudinal direction. This
solution can be classified as well known technology, which will lead to satisfactory behaviour of the
concrete deck with regard to strength, rigidity and durability. The Contractor preferred to omit the
prestressing and increase the thickness of the deck slab in regions adjacent to the in-situ stitches above
the piers.
Cross section approach spans
20m truss module
Permanently unbonded tendons as an alternative to bonded tendons are widely used in building
structures, but has only recently been introduced in bridge building. Long time experience with this
solution under fatigue loading is thus not available. The use of unbonded tendons at the resund
Bridge was restricted by the requirement, that the design of members should accord with a recognised
code of practice, subject to the Owner's approval, but it turned out that this technology was not brought
into play.
The connection between the concrete deck and the steel top chord transfers a longitudinal shear force
through a great number of relatively short 22 mm diameter headed studs. The transverse bending
moments in the deck are partly taken up by the torsionally stiff steel chords. This transfer of stresses is
secured by a smaller number of long studs over the webs of the chords, loaded in tension. The shear
and bending connectors are placed along the entire length of the steel chords with a concentration at the
7 m length of the truss girder nodes. Full scale laboratory tests, carried out at the Danish Technical
University, with a similar connection combining short shear studs and long tension studs designed for a
non-realised proposal for the Great Belt West Bridge have shown a satisfactory behaviour.
2.2
The Railway Deck
The lower railway deck on the Approach Bridges is a containment structure comprising two parallel
trough sections supported on cross beams spanning between the lower truss girder nodes. The trough
sections are prefabricated and stitched above the cross beams and fixed at the same place to the steel
structure. According to the ASO investigations it would be possible to apply longitudinal posttensioning after connecting the concrete deck to the steel structure. This solution was, however, found
to be uneconomic, as a major part of the prestress force would be transferred to the steel structure. The
lower deck is consequently made of reinforced concrete. ASO Group carried out analyses of alternative
solutions, among others a separation of the concrete troughs from the steel structure. This would
eliminate tensile stresses due to global sagging moments. As a large number of bearings and joints
would be involved the solution was, however, not considered feasible due to maintenance costs.
Headed shear studs are used in a similar way as at the upper deck for the connection between the lower
concrete deck and the steel cross beams. Short studs are mainly placed at the outer trough webs to
transfer the longitudinal shear between concrete and steel, while long tension studs are placed at all
webs of the troughs to transfer part of the local hogging moments in the troughs to torsion in the cross
beams. The transfer of longitudinal shear leads to bending of the cross beams in a horizontal plane and
thereby a certain rotation of the top plate of the cross beams relative to the bottom slab of the troughs
and a non-uniform distribution of the stud shear forces in this area. A thorough calculation turned out
to be necessary to demonstrate, that the stud forces were suitably limited in the serviceability limit
state.
Cross section cable-stayed spans
3.
Structural Analysis
Elastic behaviour of the structure without moment re-distribution shall be assumed according to the
Design Requirements for global static analysis at both the ultimate limit state and the serviceability
state. The distribution of the internal forces between steel and concrete parts of the structure depends
then on the stiffness of the concrete, which will be influenced essentially by formation of cracks in the
concrete. Cracked sections are defined as sections where the maximum tensile concrete stress has
reached the mean tensile strength fctm in the serviceability limit state. The stiffness of concrete cross
sections may be based on gross-sections, and the modulus of elasticity shall be taken as Ecm for
uncracked sections and 0.6Ecm for cracked sections.The use of simple rules of this kind is clearly
convenient in large calculations, but it can be claimed, that the effect of a complicated process like
crack formation, including tension stiffening, hardly can be covered by a single factor. In the present
case it was realised during the design process, that the factor 0.6 may be unrealistic high in some
situations.
4.
Shear Connections
Headed studs are used as already mentioned to connect the concrete deck structures with the steel truss
girder. The basic codes of practice for the design of the resund Bridge were the Eurocodes, including
EC4: Design of composite steel and concrete structures. Only Part 1-1 of this code comprising general
rules and rules for buildings was available, and a comprehensive literature study of shear connections
between concrete and steel was therefore carried out by the ASO Group, partly to assess the viability of
alternatives to headed studs and partly to determine the fatigue characteristics of shear connections.
4.1
Type of Shear Connection
The result of the investigation regarding different types of shear connections was that the only realistic
alternative to headed studs seemed to be the so-called PERFOBOND strip, developed by H.-P. Andr,
to be used especially in fatigue sensitive composite structures, eg. railway bridges. The basic idea
behind this solution is to limit the contact pressure between concrete and steel and thus avoid
premature crushing of the concrete and slip between concrete and steel in the state of serviceability.
The practical solution is to introduce a roughness by welding a simple perforated steel plate to the
flange plate of the beam. The perforated plate acts as an anchor as well, if it is placed in an upright
position. Other measures to prevent uplift of the concrete plate are then superfluous. It has been
demonstrated that it is possible to maintain a high static strength and plasticity during the complete
fatigue life of this connection, even if it is set in a fully cracked concrete. Another point is that the
perforated plate is connected to the steel flange by longitudinal fillets welds, which belong to a higher
fatigue strength category than the weld of a shear stud or the transverse welds of other types of shear
connections.
The PERFOBOND strip shear connection has been applied in the road and railway bridge over the
river Caroni, Venezuela. It is allowed at the resund Bridge according to the Design Requirements if a
suitable test programme is set up and supported by a conceptual model. A test programme is not
required for the headed stud solution, which is commonly used in modern road and railway bridges and
thereby can be classified as well known technology. Headed shear studs were selected finally for the
resund Bridge.
4.2
Static Strength of Headed Shear Studs
The ASO investigation showed that a proper model for the behaviour of headed studs is missing. The
static resistance according to EC4 and several other codes of practice is given by two empirical
expressions, one comprising the yield stress or the tensile strength of the stud material and another
comprising the concrete strength and stiffness. Using high concrete strengths as in bridges, the
expression based on the characteristics of the steel is governing.
The failure mechanism of studs is normally described in the following manner: The shear effect leads
to a concentrated pressure against the concrete at the root of the stud; this causes local failure (crushing
or softening) of the concrete. The compression moves subsequently away from the root of the stud,
which leads to the stud being gradually more and more subject to bending and tension. Based on this
description it may be questioned, if a better value of the carrying capacity of the stud can be obtained
by an expression involving the characteristics of concrete and steel at the same time. Research is made
at present at the Danish Technical University in order to establish a theoretical clarification of possible
failure mechanisms for headed studs and thereby an improved basis for predicting the resistance of
such shear connections.
The current formulas for the resistance of shear stud connections are mainly based on tests of the pushout type. Such tests have been criticised for being over-simplified compared to the conditions in
composite beams. The push-out test specimen consist of a short length of a rolled steel I-section
connected to two parallel, reinforced concrete slabs by means of the shear connection to be tested. The
specimen is supported at the ends of the concrete slabs and a load is applied to the steel section. F.
Leonhardt states that the studs are subject to compression and shear in the push-out tests, whereas the
studs are subject to shear and tension in real structures. The push-out tests lead thus to an
overestimated resistance.
Research has been made to reveal if the shear capacity is influenced by crack formation in the concrete.
The test results are contradictory, but it is believed that a 20% reduction of the stud shear capacity in
concrete subject to tension is a very conservative estimate.
The unfactored shear resistance of headed studs is according to EC4 equal to 0.8Asfu. As is the nominal
stud shank section area and fu is the specified ultimate tensile strength of the stud material, limited to
500 N/mm2. The ASO investigation showed that the resulting safety level using this resistance is
considerably lower than according to the corresponding Danish and German rules. Especially the latter
have been extensively used in major bridges with a satisfying result.
Considering the above mentioned situation regarding the understanding of the behaviour of studs, it
was decided to change the EC4 formula to 0.7Asfu and to introduce a further reduction by a factor 0.8
for studs set in concrete in tension. The rules are valid for studs of diameter 22 mm or less.
4.3
Fatigue Resistance Of Headed Shear Studs
The Eurocode requirements regarding fatigue resistance of headed studs are found in Eurocode 3,
Design of Steel Structures, Part 1-1. The verification is based on nominal stresses corresponding to two
failure modes:
failure in the weld between stud and steel flange or in the heat affected zone under the weld; the
detail is classified as a welded joint with load carrying weld; the nominal stress is the shear stress
on the cross section of the stud, and the fatigue strength at 2106 cycles (equal to the detail
category) is C = 80 MPa,
failure in the tension flange caused by the effect of attaching the

stud; the detail is classified as a welded attachment with non-load carrying weld; the nominal
stress is the normal stress in the flange, and the fatigue strength at 2106 cycles is C = 80
MPa.
A study of recent literature indicated that an interaction between failure of the stud and failure of the
flange is possible if the steel flange is subject to tension. As for the static strength it is questioned by
some authors if the push-out test specimen are sufficiently representative for composite beams. The
outcome of the ASO investigation was, that the following verification procedure for the fatigue
resistance of headed shear studs was decided:
the fatigue strength of headed studs in shear welded to a steel flange in compression shall be
assessed in accordance with EC3, but using a lower detail category, 60 in stead of 80,
the fatigue strength of headed studs welded to a steel flange in tension shall be assessed according
to the following requirement for combined shear and flange tension when using constant
amplitude stress ranges:
E .2
3
2
E .2
+
3
2
where C = 60 MPa and C = 80 MPa are the fatigue strengths, while E.2 and E.2 are the
equivalent constant amplitude stress ranges corresponding to 2106 cycles, calculated from the
basic fatigue strength curves for and according to the rules given by EC3. Partial safety
factors shall be applied to the load effects and the materials strengths according to the Design
Requirements.
The interaction criterion is based on experimental evidence considering available beam test results but
avoiding push-out tests in order to obtain a conservative rule.
Build a Link - Goals, principles, strategies and results

Peter LUNDHUS
Civil Engineer, M. Sc
Technical Director
resundskonsortiet
Copenhagen
Denmark
Peter Lundhus graduated in 1965 from the

Technical University of Denmark as M.Sc in
civil engineering. From 1965 to 1988 he
worked for an international contractor, covering
all aspects of design, bidding and construction
of civil engineering works. From 1988 he
joined the owner organisation responsible for
the construction of the Great Belt Link. In 1992
he was appointed Technical Director for the
resund Link
Introduction
Scandinavians have a tradition for building bridges which, of course, is quite natural
for countries surrounded by water or characterised by large lakes and fjords.
But a bridge is not just a bridge. In the case of the resund Link, we're talking about
one of the largest infrastructure projects in European history. Its dimensions and
position are both unique, as are the technical challenges confronting the builders. In
addition, there were significant psychological barriers to be overcome before
Denmark and Sweden decided to build the link in 1991.
Building a bridge between two nations, Denmark and Sweden in an area of 3
million inhabitants - is an enormous undertaking. Yes, our two countries have a great
deal in common in terms of their past history and future interest. Yet we have
different traditions, different systems and, in many ways, a different approach to
things. The bridge, therefore, will have large impact on the surrounding region
culturally and economically as well as in terms of development. Technically, too, we
have had to find common ground. In one country for instance, the railway runs on the
left, in the other on the right.
The resund Fixed Link is a combined rail and motorway link extending 16 km
between Kastrup on the Danish side to Lernacken on the Swedish side.
From the project's earliest days, it has attracted enormous interest from political
circles, from the authorities, from future users and from the local residents. Where
should the link be sited? What should it look like? How will it affect the environment?
These are just some of the many questions raised in the media questions which have
added to the pressure on decision-makers, authorities and the project's owners on both
sides of resund.
The populations demanded to be heard on issues relating to traffic, the environment
and finance. The environment played a particularly major role in the debate in
advance of the decision to build the resund link a relatively new item on the
agenda. As late as the mid 1970s, Danish and Swedish authorities seriously
considered building a new airport on the island of Saltholm in resund. 10-15 years
later, as the 80s drew to a close, Saltholm had become a protected area. The effect of
the Fixed Link on the environment, therefore, became a subject for heated debate at
all levels.
In Scandinavia, debates of this type are unlikely to be quiet affairs and, true to form,
our democratic traditions gave rise to extensive discussion among all interested
parties. The project was taken very seriously indeed - by organisations and interest
groups, by future users and by its neighbours. By tradition, Scandinavians are inclined
to be critical. They put questions to political leaders and they demand answers. And
they're not satisfied with easy solutions.
In this respect, too, resundskonsortiet has earned the title of "bridge builder."
What distinguishes this project from many other projects, however, is its surrounding
community and the many factors that needed to be taken into account:
-
The project is politically sensitive and has been highly affected by the political
decision processes relating to other infrastructure projects (resund's
landworks, the City Tunnel, etc.
Although its framework was laid down from the outset, the project had to
define its own standards and ascertain whether they worked.
In view of the attention given to local as well as global environmental issues,

the environmental impact of the construction process, as well as that of the
completed traffic facility, has had to be considered very carefully.
The professional demands on those involved have been very high. A

contributory factor has been the strong emphasis on quality and transparency
in recent years.
The Decision to build the Bridge

The decision to build the resund Fixed link was taken by the Danish and Swedish
governments in 1991, when the two parties each agreed to establish a state-owned
limited company for the purpose of forming resundskonsortiet which would be
responsible for constructing the resund link.
In addition, the government agreement assigned resundskonsortiet the task of
operating and maintaining the link following its completion in the year 2000 partly
by levying toll fees on the link's users.
Some considerable time elapsed, however, before the detailed planning, let alone the
construction phase, could commence.
On the Danish side, a number of hearings on the assessment of the effects on the
environment as well as on the control and monitoring programme for the coast-coast
section and near-lying areas were held.
Final permission
In Sweden, the establishment of the resund Fixed Link was tested under the terms of
the Natural Resources Act, the Water Act and the Environmental Protection Act. On
June 16, 1994 the Swedish Government approved the construction of the resund
Link under the terms of the Natural Resources Act and the Water Act and in 1995, the
Licensing Board for Environmental Protection gave its permission to build and
operate the fixed link on Swedish territory under the terms of the Environmental
Protection Act. Then, in a verdict in July, the Swedish Water Court gave permission
to construct the Fixed Link under the terms of the Water Act. At the same time, the
Water Court set out the environmental requirements which resundskonsortiet had to
meet in connection with construction activities in Swedish territorial waters.
On July 8, 1994, the Danish Government approved the project. In March 1995, the
Danish Minister of Transport approved the resund project's environmental quality
objectives as well as the criteria and requirements for the control and monitoring
programme for the link's Danish sections.
The period prior to the decision
The extensive investigations and verifications generated considerable uncertainty
among the populations of the two countries. In particular, the environmental impact of
the resund Fixed Link was, as I said, the subject of enormous public interest and
debate. Economic and technical issues as well as future expectations were widely
discussed.
These debates, however, did not deter resundskonsortiet. The time was spent laying
down the strategy how best to proceed? The dual-nationality of the project, for
instance, necessitated a detailed examination of the different legislation in the two
countries. In fact, for each decision concerning the facility's design and construction,
resundskonsortiet had to examine the legislation of both countries a task which
was not made any easier by the fact that, at this stage, Sweden had not yet become a
member of the EU. Consequently, a number of EU laws and regulations had to be
considered.
In each individual case, resundskonsortiet had to decide whether a joint platform
could be established (the lowest common denominator), whether EU regulations
applied or whether resundskonsortiet was able to set its own standards and norms.
With regard to norms, it became clear that a number of problems were bound to arise.
Since Denmark and Sweden do not apply the same norms and regulations,
resundskonsortiet was faced with the choice of either co-ordinating the existing
norms of the two countries or coming up with an alternative solution. Consequently,
the consortium decided to apply the future EU regulations or, in cases where norms
had not yet been defined, to work actively towards making them applicable to the
project. In this way, resundskonsortiet solved both the project's own technical norm
problems and acted as a catalyst for the definition of a series of technical EU norms.
In fact in several instances, this marked the initial introduction of a number of EU
norms. Since then, such norms have been applied at several other projects, including
the Metro project in Copenhagen.
Within the field of the working environment, resundskonsortiet was also faced with
certain choices. Here too, both countries have different norms but in accordance
with resundskonsortiet's objective of keeping the number of work-related accidents
to less than half the national average (in Sweden and Denmark as well as in Spain
where the bridge girders are built), the working environment requirements were
written into the contracts which stipulate that all planning must incorporate the
environment.. In addition, an extensive safety-at-work campaign has been focused on
the employees throughout the construction period. One aspect of the safety-at-work
campaign is that it has continuously emphasised the individual employee's personal
responsibility as far as safety issues is concerned.
In the light of the political requirements from the two countries, resundskonsortiet
adjusted the original alignment as well as the link's technical design. The objective
was to optimise the link - technically, economically and environmentally - in order to
reduce any harmful effects on the environment.
The new project
Following international tender, resundskonsortiet invited resund Link Consultants
and the ASO Group to put forward proposals for the link's design. The results from
these proposals formed part of the ongoing planning of the project for, for instance,
the purpose of ensuring that the facility's blocking effect on the water flow in resund
was reduced.
resundskonsortiet also decided to split the coast-coast section into five large
projects:
-
Dredging & reclamation work (6 contractors tendered)

The immersed tunnel under the Drogden Channel (5 tenders)
The low and approach bridges (6 contractors took part in the competition)
The high bridge across the Flinte Channel (5 contractors tendered)
The railway engineering system and installations
Organisation
resundskonsortiet's main aims are:
-
The design and construction of the Fixed Link to Sweden

Ownership
Operation and maintenance of the Fixed Link
the organisation has two key functions to establish the link and to operate
and maintain it
Organisation diagram
Management
Joint functions
Staff
Technical
Department
Operations
Management principles
resundskonsortiet's management principles are not static. Flexibility is an important
component in the whole process and resundskonsortiet is conscious of its
responsibility to initiate and motivate change within its own organisation as well as
within its partners and the world beyond.
For all this to happen, two pre-conditions must be met:
-
Everyone should know the objectives

Everyone should know how we achieve them
The vision
resundskonsortiet's basic management principles can be summarised as management
by objectives, delegation and pro-activity. Consequently, from the earliest stage,
resundskonsortiet defined clear goals within the vision and business concepts.
As such goals should be measurable, a strategy for their achievement was developed.
Once the goals and strategy are in the place, resundskonsortiet determines the
organisational structure which will eventually realise the goals and business
concepts.
At regular intervals (approximately every six months), the objectives and strategies
are reviewed. Wherever necessary, resundskonsortiet adjusts the organisation to
ensure that sufficient resources are available for the tasks in hand and that those in
charge of individual tasks have the authority to act accordingly.
This "controlled process" serves to emphasise the concept of pro-activity that
resundskonsortiet is ready to act before events force us to and that any change of
direction has been carefully considered.
The successful implementation of a project the size of the resund Link and within
such a short time span requires the existence of a company culture which
continually and naturally adapts to new demands and requirements, internally as well
as externally. No less importantly, it also requires an acceptance of openness, a
willingness to listen to suggestions and to adapt to ideas from partners as well as from
the world at large. Only in this way will resundskonsortiet be able to bring the
project to its successful conclusion and thereby lay the foundation for strengthened
links between Sweden and Denmark.
The bridge as the resund region's focal point
The resund Fixed Link is intended to bring Denmark and Sweden together two
countries which are each other's largest trading partner. It will also draw Europe
closer together. Once the link is completed, there will be fixed rail and motorway
connections from the North Cape in Northern Norway to Southern Italy for the first
time ever.
In the process, Copenhagen and Malm will become one city a new metropolis with
a highly skilled workforce of nearly 2 million offering the best from the two nations,
economically, technologically and culturally.
Linking two of Europe's most developed societies, the link represents a huge advance
in terms of public transport. Both Denmark and Sweden are, of course, members of
the European Union and both countries enjoy long traditions for political stability,
stable economies and high educational standards. It should also be noted that this
huge investment in public transport is a consequence of the region's endeavours to
create environmentally sound growth. This philosophy for growth is a natural
extension of the whole concept of the Scandinavian model.
The resund region will become Northern Europe's political and economic centre a
centre which will attract large numbers of businesses, organisations, scientists and
researchers as well as tourists.
With 3.2 million inhabitants, the region will, in fact, be Northern Europe's largest
domestic market, equalling Berlin, Hamburg and Amsterdam.
The region will also become a natural centre for inward and outward economic
activity in the Baltic area and parts of central Europe. Our biggest ports already
handle very large quantities of freight and Copenhagen Airport in Kastrup is Northern
Europe's largest and most modern. Significant investments are also being channelled
into railways and motorways on both sides of the Fixed Link.
In a number of contexts, the region will be at the cutting edge of international
developments not least within the field of biochemistry, pharmaceutical production,
medical equipment, IT, environmental technology and food processing. In these areas,
a number of companies and research institutions rank among the foremost in the
world, at least partly because of the exemplary collaboration between the public and
private sectors in creating the optimum conditions for development environments.
Design and industrial production also have a high profile, with a number of world
class designers and architects living and working in the area. As a conference
destination, the region is the foremost in Northern Europe.
The level of education, too, is second to none. The population of the resund region
is open, socially responsible and internationally-oriented. To most Scandinavians,
English is the second language and many are familiar with German, French, Spanish
or Russian from their senior school classes.
The quality of life in the region is among the highest in the world and its inhabitants
benefit from unparalleled personal safety, freedom and excellent public systems
including free education and an extensive and free health system.
Vision and strategy
In 1994, resundskonsortiet invited tenders for the link's main contracts the tunnel,
the dredging works, the high bridge and the approach bridges - based on the principles
in the EU Directive 93/37/EEC concerning public agreements. This was despite the
fact that the directive did not yet apply to the resund Link.
The tender process itself did not follow traditional methods either. Due to the dualnational aspect, resundskonsortiet was able to lay down its own tendering strategy.
In a traditional tendering process, the client undertakes the detailed planning and then
invites tenders for a number of contracts in international tendering. You could say that
the architect designs the house and the contractor prices and builds it. In the
background, the client manages the entire process with a firm hand.
resundskonsortiet wished to handle the process differently.
For instance, the consortium wanted to avoid the discussions, confrontations and
arbitrations which are often inherent in conventional client-contractor relationships.
The aim, of course, was to build the facility at the agreed price and at the agreed time
and also to create a working relationship which ensured fair and constructive
collaboration between the many parties involved.
The basic concept was "partnership." A survey of resundskonsortiet's "stakeholders"
revealed that no fewer than 150 organisations were involved in the project, all of
which all exercised a degree of influence. Unless this was tackled in a new way, it
could easily lead to a large number of confrontations.
Stakeholder diagram
Companies
Organizations
Media
Citizens
Parliament
Municipalities
Neighbors
Government
resundskonsortiet
Transport authorities
Expropriated
Fixed link
Road-and-rail-users
Contractors
Consultants
resundskonsortiet's basic concept, therefore, was to be a client that valued a

constructive partnership with all parties involved, including, of course, the
contractors. In contrast to the conventional methods, the consortium did not wish to
manage in detail, but to delegate responsibility and power while, at the same time,
maintaining an constant overall view.
The contract philosophy
One consequence of this was the introduction of the Design+Construct philosophy
which provided the contractors with strong draft proposals and, not least, a free
hand to improve the project.
The concept is that responsibility for detailed planning and execution should be
clearly assigned. This limits the conventional working relationship's potential for
confrontation and provides the individual contractor with greater opportunity for
solving his task efficiently. This again creates committed contractors who become
involved in the detailed planning, initiating improvements and who feel total
responsibility for the execution.
This involves considerable freedom for the contractor a freedom which the client
must be continually aware of and respect.
The basic premise for Design+Construct is the illustrative design where the client has
developed one or more designs which meet all specifications set out by, for instance,
the authorities and the client. You could say that it's the client's "homework" that
provides the contractor with the parameters for the execution of the task: this is how it
can be if the regulations are adhered to. Since, however, the contractor possesses
considerable knowledge, the project will inevitably be improved once the contractors
have examined it.
With the illustrative designs, resundskonsortiet provided constructive groundwork
which was made available to the bidding contractors, guiding them in the right
direction while not tying them down. At this stage, a number of requirements from the
authorities, approval of worksites as well as a series of obvious questions which
always crop up at the detailed planning stage, had been taken care of.
Quality
All contracts comprise descriptions of the fundamental quality management
philosophy of "own control". This means that the contractors have been assigned
responsibility for the design and administration of the quality management systems
and the quality control of their own work. The contractors must therefore
establish/implement and maintain an extensive quality management programme based
on the EN 29001 standards. The programme must comprise the construction works
(temporary as well as permanent), the external environment and the working
environment (health & safety).
KKSURR
As part of the consortium's preparatory work on safety, accidents, rescue and
clearance, close working relationships have been established with a number of
individuals representing the authorities (police and emergency authorities) on both
sides of resund. This aimed at involving the authorities in an advisory capacity in
relation to resundskonsortiet's draft proposal to the contractors, thus ensuring that
the regulations laid down by the authorities of both countries as well as their
professional and technical knowledge contributed to the quality of the completed
facility. These joint efforts have meant, for instance, that general agreement on the
facility's safety systems was achieved - for example on possible sprinklers in the
tunnel. In this case, the KKSURR members agreed that the disadvantages outweighed
the advantages of sprinklers because they would make the smoke wet, heavy and
more difficult for the rescue teams to work in. The involvement of these
representatives, however, in no way undermined the authorities' role as the final
arbiter of the facility. In the example mentioned above, the authorities subsequently
approved the KKSURR members' recommendation concerning the sprinkler system.
Assessment of risks
A natural consequence of this model is that the contractor has full responsibility for
risks over which he has control, ie. detailed planning, the workforce, permissions
from the authorities. On his part, the client is responsible for other risks: currencies,
political aspects, the overall granting of approvals by the authorities, i.e. areas in
which contractors traditionally calculate a risk premium and which often become
subject to dispute following the completion of a project.
Dispute Review Board
The Design+Construct working relationship also reduces the number of potential
conflicts compared to conventional methods. By establishing a Dispute Review
Board, resundskonsortiet has further reduced the likelihood of a dispute.
The principle is that a Dispute Review Board a panel of three experienced,
independent and internationally respected engineers is set up for all contracts.
They take no nonsense
In the event of a dispute, the Dispute Review Board functions as chief arbiter. The
Board convenes every second month when contractors give a briefing on the status of
the contracts and bring up possible problems. The aim is to solve such problems as
quickly as possible preferably before they affect the running of the construction
sites. The experience with the Dispute Review Board shows that problems are rarely
allowed to come this far if ever. Firstly, there are few potential areas for conflict
(these have been pre-empted in the contracts). Secondly, the preparatory work for
Dispute Review Board meetings reveals problems at an early stage, thus enabling
them to be resolved before the matter is brought before the Board.
The Board's decisions are not binding, but can be referred to arbitration. In the
contracts, however, the parties have agreed to accept the Board's decisions until the
arbitration process has been completed.
The facility
The aesthetic aspects of the resund Link have been subject to extensive
considerations. The intention throughout has been for the link to appear as a whole
with an architectural and aesthetic expression which reflects Nordic architectural
traditions. The link should express a cohesive, rythmic harmony in which the
individual components fit effortlessly into each other and where the details are
subordinate to the project as a whole. The link should also complement the open
landscape on each side of resund.
The aesthetics of the resund Link have been comprehensively discussed at an
international reference group with participation from international experts.
The environment
The governments of the two countries and their environmental authorities have laid
down rigorous regulations for the acceptable environmental impact of the link. In
many areas, resundskonsortiet has, therefore, had to develop new strategies and
methods in order to limit the effects on the marine environment in resund and the
Baltic Sea. The fact that the project extends across national borders has played a
special role, too. In many instances, the project has had to obtain double approval
from the authorities. Consequently, due to the judicial, traditional and political
differences between the two countries, variations in the definitions of the prescribed
environmental requirements have occurred.
A key element in the national hearing procedures has been the need to optimise
environmental planning. The resund Fixed Link may not impact on the Baltic Sea
(the physical/chemical marine environment and the biological marine environment in
the Baltic). The construction works must only temporarily affect the environment
surrounding the facility.
This resulted in a number of changes, including:
-
The blocking effect of the resund Fixed Link on the water flow may not
exceed 0.5%. Compensation dredgings in the Drogden Channel will even
reduce the blocking effect to 0%.
Sediment spill from dredging works must be limited to 5% of the dredged

material. The spill must be limited in intensity, time and space.
Total dredging works for the entire resund Fixed Link must be limited to m3
7 million.
The changed conditions necessitated a change to the facility's design. A more northern
alignment was chosen and the Danish land facility at the Kastrup peninsula was
modified. Compensation dredgings were carried out in the Drogden Channel while the
immersed tunnel was extended and its alignment altered. The design of the artificial
island was also changed and realignment and dredging of the Flinte Channel were
carried out.
The stipulated requirements also incurred further costs, particularly as a result of the
new tunnel solution and the compensation dredgings.
Control and monitoring programme an example of good co-operation
The Design+Construct partnership has also resulted in a novel solution within the
field of the environment. The authorities' demand for a control and monitoring
programme resulted in the setting up of the On-line system, EAGLE, which enables
the authorities, the client and the contractor to monitor the environmental impact of
the construction works on an ongoing basis.
The EAGLE system is operated by the contractor who, as part of the contract, has
assumed responsibility for meeting the authorities' requirements. The point is that the
system makes it possible for both authorities and client to follow developments. In
practice this means that only one environmental monitoring vessel the contractor's is permitted to sail in the area. It is not necessary for the authorities or the client to
operate their own vessels as both are kept informed via the EAGLE system. This is
one example of resundskonsortiet's delegation of responsibility to the contractor and
its confidence in the fact that the contractor will fulfil his contractual obligations.
In order to check that the prescribed environmental requirements and conditions are
met, a number of control and monitoring programmes have been established and
implemented. These must ensure that no unexpected short or long-term impact on the
environment occurs during the construction period. A further aim is to document that
the stipulated environmental requirments and conditions are maintained.
Execution and economy
The project's economy has developed steadily and as expected. The extensive cooperation with the contractors has meant that the Consortium has maintained a clear
overview of all activities. Only in one instance has it been necessary to revise the
budget.
The only major item to impact on the budget has been the so-called zero-solution
designed to ensure an unchanged water flow through resund. This meant that the
budget had to be increased by approximately DKK 1.5 billion.
The experiences
The resund Fixed Link must be ready by the summer of the year 2000. This is the
overall objective which resundskonsortiet and the contractors are working towards.
The secret behind the new time schedule is a "parallel works" approach on various
sections of the project. The purpose of "parallel works" is improved utilization of
time. Thus, any surplus time on one project is transferred to another contractor who
lacks time. It's obvious, for instance, that the bridge has to be welded together before
rail tracks are laid. But the rail contractor can begin preparations before the high
bridge is ready.
resundskonsortiet could not have operated the Parallel Works method over the last
two years unless co-operation procedures had allowed for planning across the
contracts. Conventionally, each contractor has completed the task assigned to him as
quickly as possible irrespective of whether his neighbour might benefit from using,
for instance, some of his scaffolding overnight in order to finish more quickly, too.
With the parallel works method, we created a common understanding of the project's
time schedule and an openness and knowledge of time schedules etc., which has
made it possible for everyone to "pull together" in the final months.
Such co-operation means that the individual contractor doesn't have to accelerate his
pace of work in order to be finished on time rather, that he has more time and scope
to complete on time and thus achieve the bonus offered by us. Of course, the client
doesn't give a bonus to the contractor just because he's completed his task, but
because he has co-operated with others so that everyone can finish on time.
The resund Bridge:

Project Development from Competition to Construction
Klaus FALBE-HANSEN
Civil Engineer MSc
Director
Ove Arup & Partners DK
Copenhagen, Denmark
rjan LARSSON
Civil Engineer MSc
Contract Director
resundskonsortiet
Malm, Sweden
Klaus Falbe-Hansen is Project

Director of the ASO Group, the bridge
consultant to resundskonsortiet.
rjan Larsson is Contract Director

responsible for the resund Bridge
Contract.
Summary
resundskonsortiet, formed January 1992, is owned jointly by the two states Denmark and Sweden
and is the Owner of the resund Link. resundskonsortiet is responsible for planning, designing,
financing, constructing, and after completion of the road and rail Link of its operation and
maintenance. ASO Group is the bridge consultant and was engaged by resundskonsortiet in 1993
after an international design competition. ASO Group is responsible for the bridge concept and is
presently integrated in the Owner organisation with particular responsibility for monitoring the
construction works of the bridge. The paper describes the project development that took place for
the bridge from the design competition to the construction. Also described are the Owners
construction contract strategy, the quality management policy and the Owners active role through
cooperation with the Contractor in achieving the earliest possible opening of the Link.
1.
Introduction
resundskonsortiet invited, towards the end of 1992, engineers and architects to take part in a
competition for the design of the Link. In response to the invitation the ASO Group was formed by
Ove Arup & Partners of the UK with SETEC of France and Gimsing & Madsen and ISC of
Denmark. Georg Rotne of Denmark is architect to the group. The group was selected for the
competition together with five other international groups. The competition led to two teams, ASO
and LC, being chosen to develop their designs further. As a result of this ASO later became
consultant for the bridge and LC became consultant for the tunnel and reclamation works.
The basic principle of the construction contract is the design & build concept. It means that the
Contract specifies a number of requirements that the finished product shall fulfil. The Contractor
has undertaken to design and construct works that fulfil the requirements. The Owner has
undertaken to pay the contract price. The undertaking of the Contractor includes everything required
for the total completion of the Portion of the Link to a state in which it is fit for its intended
purpose. Excluded from the Contractors undertaking are only those items that are expressly
excluded by the Contract. This runs contrary to the principles of a traditional construction contract.
Basically the Owner specifies what the Contractor shall achieve, and the Contractor determines how
to achieve it. As a logical consequence of the design & build concept the Owner has adopted the
basic quality assurance principle of self-control. The Owner is monitoring the Contractors
compliance with his Quality System and the requirements of the Contract.
Although only Denmark was a member of the EEC at the time of signing the Contract, the Council
Directive 93/37 EEC, restricted procedure including prequalification, has been applied in the
procurement process throughout.
2.
From Competition to Tender
The design competition took place over a few months in the early part of 1993 and the result was
announced in July that year when two groups proposals were chosen as joint winners. They
included two very different bridge designs: ASO Groups two-level predominantly steel structure
with the road placed above the railway and LCs proposal for a single-level concrete bridge with
the railway placed in between the two road carriageways. resundskonsortiet decided to develop
both designs further in parallel before choosing between them. In the following only the two-level
solution is described in any detail as that in the end proved the successful concept.
The two-level bridge

The detailed concepts were developed during the later half of 1993. The two-level bridge concept as
developed during the short competition phase proved a robust design and fundamental changes to
the concept were not required. Early 1994 it was decided that both bridge solutions should be taken
forward to bidding stage on a common alignment.
A number of alternatives for the layout of the whole Link were prepared and evaluated at this stage.
The main issues were the environmental impact of the Link and the economic consequences of
alleviating the impact. The critical factor was the blocking of the water flow through resund due to
the physical obstruction of the Link. The main parameters in these investigations were, the extent of
the artificial peninsula at Kastrup, the length of the tunnel and artificial island, and also whether the
island should be split in two with the opening being bridged by a low level structure. Number of and
shape of bridge piers were also considered in these studies.
A value of the blocking effect of the Link at around 2 to 3% had been assumed acceptable by the
governments at the time of the resund Treaty in 1991. However, it soon became apparent that a
much lower value had to be achieved if the project were to gain the necessary environmental
permits. The environmental investigations carried out by resundskonsortiet and their consultants
proved that a zero solution could be achieved without serious economic consequences, and on this
basis the two governments gave their approval in the summer of 1994 to constructing the Link.
However, the Swedish environmental legislation requires that the Swedish Water Court, which is
independent of the Swedish government, must rule on the effect on the water regime of constructing
the Link as a prerequisite of giving the permission to construct the Swedish part of the Link.
In July 1995 the Water Court gave its ruling on

the construction of the Swedish part of the Link,
which constitutes approx. 80% of the bridge.
The Link itself should be designed so as to limit
the blocking effect to 0.5%. To achieve the
zero solution compensation dredging should
be carried out. A maximum of 7 million m3
could be excavated in connection with the
construction of the Link of which the excavation
in Swedish water was limited to 2 million m3.
Another important part of the ruling was that the
sediment spill during dredging activities for the
whole Link must not exceed an average
maximum of 5% of the total dredged volume.
The Concession Board, another independent
Swedish authority, at the same time ruled that it
would not put any restrictions on the traffic on
the bridge, thus giving resundskonsortiet the
right to construct and operate the Swedish part
of the Link.
The resund and the Baltic
With these approvals the main legal obstacles for the construction of the Link had been cleared. The
timing and content of the rulings were critical considering that the tender documents, the so-called
Enquiry Documents, had been issued to the prequalified bidders in December 1994 and their bids
received in June 1995 i.e. before the ruling of the Water Court.
2.1
Contract Strategy
At an early stage resundskonsortiet had decided to let the works in the form of design & build
contracts based on the following principles:
Owners defined requirements on function, aesthetics, safety and environmental protection,

everything included in Contractors scope of work unless expressly excluded,
only specified duties of the Owner,
specified division of risks attributable to geotechnology, weather and permits,
Contractors self-control within QA system,
Owner is monitoring the Contractors performance and
review/inspection/approval does not relieve the Contractor.
resundskonsortiet also decided not to follow any standard general conditions. In stead a tailor
made document, General Conditions of Contract containing all general and legal provisions, was
produced for the project.
Under the form of contract chosen the Contractor has considerable freedom regarding the means and
methods for carrying out the works. The Contractor is responsible for the detailed design as well as
the physical works. He is supervising his own work and is responsible for providing documentation
to prove that he is doing so, and that as a result the work he is doing is of the quality required by the
Contract. This means that the Contractor is approving his own work. However, the Owner is
monitoring the Contractors performance.
The contract strategy led to the inclusion of a number of particular documents within the Enquiry
Documents issued to the bidders:
Definition Drawings and Illustrative Design

Reference Conditions
Quality System Requirements
2.1.1 Definition Drawings and Illustrative Design

The design & build type of contract has some obvious advantages, but clients have little direct
influence on designs as long as the functional requirements formulated in tender documents are met
and the design conforms to the design basis.
In the case of the resund Bridge, extensive consultations had been held with authorities in the two
countries on matters like aesthetics, environment, road and railway operation, navigation, safety,
etc. The designs prepared by the Owners consultants were markedly influenced by these activities
and included many features over which the Owner wished to maintain control. Some of these could
be expressed as functional requirements but some important design features could not. The Enquiry
Documents therefore included not only the usual design and construction requirements but also
Definition Drawings, which described the design features, geometry, and materials that should be
retained in the Contractor's design. The Definition Drawings were only concerned with the visible
geometry, foundation types and sizes were not shown.
The Illustrative Design drawings were included as an example of a design, which fulfilled the
Owner's requirements. The Illustrative Design was included in the Enquiry Documents for
information only and would not become part of the Contract, while the Definition Drawings were
contractual documents, which the Contractor was obliged to follow in order to fulfil the
requirements of the Contract.
2.1.2 Reference Conditions
Reference Conditions refer to geotechnical conditions and weather conditions, and define
benchmark values above or below which as the case may be the Contractor is entitled to
compensation. The purpose is to avoid the Contractor having to unnecessarily add contingencies to
his lump sum price.
Regarding the geotechnical reference conditions the Owner had, prior to issuing the Enquiry
Documents, carried out his own site investigations in the bridge line and he had also carried out
detailed testing on shore of in particular the predominant Copenhagen Limestone. The geotechnical
conditions contained the ground stratigraphy and a summary of the strength and deformation
properties for the ground. Bands of uncertainty were provided within the Reference Conditions and
the Contractor must accept these as foreseen ground conditions. Where the ground conditions
were outside the Reference Conditions, the Owner accepted the risk, and the Contractor would be
compensated for proven extra cost. However, he was not required to compensate the Owner if
conditions were better than assumed.
2.1.3 Quality System Requirements
The basis for the Owners monitoring of the Contractors work is the Contractors Quality System
and the Owners requirements to this system were laid down in the document, Quality System
Requirements.
The Contractor was required to establish, maintain and adhere to a Project Quality Programme,
specifically adapted to the Contract. The PQP may adopt routines and procedures developed for the
Contractors internal quality system, if applicable to the Contract. The PQP had to be based upon
the contract document, Quality System Requirements, which in turn is based on the EN ISO 9001
standard, 1st edition.
The PQP should be documented by a Quality Manual including General Procedures and a number
of Quality Plans. The Quality Manual and the General Procedures (typical administrative
procedures) are meant to set out the overall systems and principles governing for all activities under
the Contract. The Quality Plans with method statements and work procedures are intended to be the
Contractors operative instruments for his planning, execution and control of the numerous work
activities.
2.2
Tender
resundskonsortiet decided to split the bridge

into two contracts, one called the High Bridge,
containing the cable-stayed main span and the
flanking side spans and one called the Approach
Bridges containing the eastern and western
approach bridges leading to the High Bridge. By
tendering the bridge in two separate contracts
the Owner could combine the lowest bid for
each package, which could lead to a lower total
price than one price for the whole project.
The Enquiry Documents were issued to the
prequalified contractor groups in December
1994. In all six groups competed for the
Approach Bridges contract and five groups for
the High Bridge contract. Both the one-level and
the two-level bridge concepts were offered to
the bidders.
Approach Bridges and High Bridge
Tender Evaluation and Award
The tenderers bids were returned on 1st of June 1995. All tenderers had put in bids for the twolevel solution. Two tenderers also put in bids for the one-level bridge. The lowest bids for both the
Approach Bridges and the High Bridge contracts were for the two-level concept.
The tender evaluation was carried out strictly on the basis of the submitted bids, the competition had
ceased. The actual process of tender evaluation followed a number of procedures prepared by the
Owner. The purpose of the evaluation was to identify the economically most advantageous tender to
the Owner.
The evaluation was in two stages, a preliminary evaluation shortlisting two to three bidders
followed by a detailed evaluation of the shortlisted bids. In both stages the evaluation was carried
out under separate headings: technical, financial, quality & planning systems and contractual. No
cross communication was allowed between the groups.
The technical group was again subdivided into a number of subgroups concerned with: design,
construction, environment/authorities and aesthetics (this subgroup was only established at the
detailed evaluation stage)
In case clarification was required at the detailed evaluation stage the necessary communication was
strictly controlled by the Owner with clearly defined and specific questions with no possibility for
the bidder to improve their bid or attempt to negotiate. Deviations were identified and the assessed
costs were added to the tender sum.
The evaluation led to the award of a single combined contract for the High Bridge and the Approach
Bridges to an international group of contractors, Sundlink Contractors HB on 27th November 1995.
Sundlink consists of Skanska of Sweden, Hochtief of Germany and Monberg & Thorsen and
Hjgaard & Schultz of Denmark. The Contractors designer is a Joint Venture of COWI of
Denmark and VBB of Sweden.
Sundlinks successful tender design followed ASO Groups two-level concept without any deviation
from the requirements of the Definition Drawings.
4.
The Construction Phase
The Contractors bid was based on the principle of large scale onshore prefabrication, with finished
elements being installed by heavy lifting equipment. The most important centres of production have
been: Malm and Karlskrona in southern Sweden and Cadiz in southern Spain. The two pylons, and
the onshore structures are the only major elements constructed in situ.
Malm North Harbour: substructure
Cadiz: approach spans
Karlskrona: high bridge steel
Bridge Site: installation by Svanen
4.1
The Owners Monitoring during the Construction Phase
The Contract is based on the principle of self-control. In general the Owner does not need to nor
does he want to approve materials, which are to be incorporated in the works. The Owner does not
get involved in the day-to-day inspection and approval of construction work. It is tempting to do so,
and the tradition of detailed inspection by the Owners representatives dies hard. In fact it often suits
the Contractor to have inspections by the Owner; it makes his work easier, can reduce his staffing
level, and may reduce his responsibility for his work. The Owner is of course present at the sites and
may from time to time test materials or inspect the works. However, it is vitally important that the
Contractor sees the Owners monitoring as a supplement to and not a replacement of his own
supervision and QC function. The Contractor must know not only what the Owner will do but also
what he will not do as part of his monitoring.
The Owner will
approve the basic design assumptions, the Project Quality Programme documentation, and the
basic planning for the works,
visit, observe, meet, discuss, witness, review quality records, audit, comment (all hands-off
activities) and may carry out random sampling,
review and if required comment/approve actions in connection with site questions,
review and approve non-conformity reports,
effectively approve the physical work when Payment Milestones are authorised for payment.
The Owner will NOT
get involved in day-to-day supervision, inspection and approval of construction,

inspect setting-out,
inspect construction joints, re-bars, formwork etc.,
inspect every radiograph of welds,
watch every batch of concrete being placed,
be present all day, every day at each construction site,
produce quality records.
An important part of the Owners strategy has been to stay a comparatively small and proactive
organisation. The Consultant ASO Group has since the start of the construction phase been totally
integrated in the Owners organisation. This has created a non-bureaucratic organisation with fast
and direct lines of command and communication. Only a total of 12 to 15 people have been
involved in the Owners monitoring activities on the many and geographically widely spread
production sites.
4.1.1 Review and Approval of Documentation
Owners approval is only applied to certain forms of basic documentation and never to the actual
physical works produced by the Contractor. Areas requiring approval are limited in order to
maintain a clear division of responsibility.
The Contractors design assumptions, his so-called basic design is submitted to the Owner for
approval and has been checked for adherence with the design requirements including the Definition
Drawings. The detailed design on the other hand is not subject to the Owners approval, it has been
submitted for the Owners review and possible comments. The detailed design is part of the
permanent works and therefore the Contractors responsibility.
The most important approval by the Owner is the approval of the Contractors Project Quality
Programme, which forms the basis for the Owners monitoring. The Quality Plans are crucial for the
planning, execution and control of the works and are therefore subject to scrutiny by the Owner. No
work activity is allowed to commence until the Quality Plan covering that particular activity has
been approved by the Owner. The Quality Plans are the proof that the Contractor has understood the
specified requirements and knows how he will achieve them.
The Owners approval is also required for the Contractors basic planning and for plan revisions.
The Owner may require revisions if delays on any activity affect the critical path.
4.2
Parallel Works
In addition to the major civil contracts that physically make up the Link: the tunnel, the reclamation
works and the bridge, there are a number of coast-to-coast contracts. These contracts cover the
installation of the railway, the communications and SCADA and traffic control systems. The coastto-coast contracts basically have continuous interfaces with each other and with the civil contracts,
Although a large number of interface milestones for providing access are identified in the various
contract documents, this is not a guarantee for the Link to be completed on time as planned. A
Contractor might be delayed in a certain area denying access to the Contractor following on and
thus delaying the completion. In order to alleviate the effects of unforeseen delays the Owner has
taken an active role in the coordination and planning optimisation of the so-called Parallel Works.
The clear objectives are to have the Link structurally completed, comprehensively tested and opened
for traffic as early as possible and within budget.
The Parallel Works procedure was established, in the autumn of 1997, by the Owner preparing an
integrated timetable where all the coast-to-coast contractors planning was incorporated in the civil
contractors planning. This became the Reference Timetable and was the basis for an optimisation
towards a Target Timetable. The Target Timetable required rescheduling of different activities
within the timetables but it also opened up for several activities to happen in parallel despite being
performed by different contractors. Prior to implementation, the Target Timetable was discussed
and refined with the respective contractors; with the Owner acting as the coordinator between the
contracts.
The system would not have worked without the cooperation of the contractors, and in order to
achieve this an award was identified for finishing on the target date. The Owners award being the
extra revenue for opening the Link early.
The cable-stayed bridge nearing completion
5. Conclusions
The bridge is not yet complete but it is not too early to state that the Owners strategy of
cooperation, trust and openness has been a success. The bridge Contract is almost 80% complete
and is as the other Link contracts on time and on budget. The quality of the permanent works is to
the Owners satisfaction. There have been no disputes so far and therefore no significant claims
against the Owner, and none are expected either. This is not usual for a project of this size and
complexity and can to a large degree be attributed to the spirit of partnership between Contractor
and Owner, which has been allowed to develop throughout this truly international project soon to
connect the two countries Denmark and Sweden.
Getting the Balance Right

The resund Bridge - Design Concept
Jrgen NISSEN
Civil Engineer, MSc
Director
Ove Arup & Partners
London, United Kingdom
Georg ROTNE
Architect MAA
Associate Professor
Royal Academy of Fine Arts
Copenhagen, Denmark
Jrgen Nissen is Director of the

ASO Group, the bridge consultant
to resundskonsortiet.
Georg Rotne is the ASO Groups

architect.
1. Introduction
Vitruvius set the agenda: good architecture is about the proper balance of Firmitas, Utilitas and
Venustas - or firmness, commodity and delight. And so during the Renaissance designs for
bridges, as for buildings, aimed to meet this ideal.
With the invention of industrialised processes and new materials and technology, a new
profession was created: the engineer. The new materials and technology were used in the
building of bridges and other structures. Firmness in particular was the domain of the engineer;
commodity was interpreted as function but more interest was shown in inventiveness and
efficiency than in delight.
The first engineers built remarkable bridges of iron, and later of steel and reinforced concrete.
They used a rational and economic design approach form follows function which worked
hand in hand with the current belief in progress.
Following the end of the Second World War, an urgent need for new infrastructure meant that
bridges, like other structures, became more and more mass-produced. Efficiency dominated;
which left little room for inventiveness and experimentation.
Sometimes architects were involved in the design of the more prestigious bridges, but mostly in a
secondary role. This has now changed. Architectural competitions are increasingly being held for
bridges. Delight is firmly back on the agenda.
Antonio da Ponte: Rialto Bridge, Venice 1597
Gustave Eiffel: Viaduc de Garabit, Truyre, 1889
2. The resund Bridge

The resund Bridge is very long. Technical, functional and economic issues are critical. It
includes approaches, where shorter spans are suitable and a much longer span over the navigation
channel. The structure should appear as a whole and not a collection of parts; it should be
economic for the approach spans as well as for the main span.
A small team of designers developed the design concept. A larger team, which also included the
client and various public authorities, developed the tender design, and the contractors team
carried out the detailed design within a design and construct contract. And because it is very
large, the client had divided the project into a number of contracts to be handled by different
contractors, and different teams might therefore handle the final design. The continuity of the
design process would thus be broken.
In these circumstances, how is it possible to ensure unity and the proper balance of firmness,
commodity and delight?
The right design would ideally be functional, economic and beautiful. Given the huge scale of the
project, emphasis should be put on developing a consistent and robust design befitting the unique
site.
It is often this relationship to its site that gives a bridge its special character. This bridge would
stand in a seascape without any dramatic natural forms to set it against. The landscape on both
sides is gentle and friendly, with small and rolling hills and curved coastlines where the land
merges with the sea. The bridge rises gradually from the tunnel at the artificial island like from a
hole in the sea. The Link would often be seen at a distance: from the shores, the sea and the air.
Those travelling on the bridge would mostly see it at speed, but they could, given a suitable
design, have exceptional views of the sea, the islands, the coastlines and the cities of Copenhagen
and Malm.
View from the Danish coast at Dragr south of Copenhagen Airport.

In the distance the artificial island Pepperholm and the bridge
3. Design Objectives
The design concept was developed in a design competition for the whole Link: an artificial
peninsula, a tunnel, an artificial island and the bridge. The two Governments had decided on the
overall functional and technical requirements and described these in a Reference Project that was
part of the Agreement between them. Under the Treaty the Link should be designed and
constructed with due consideration of what is ecologically motivated, technically feasible and
economically reasonable in order to prevent any detrimental impact on the environment. The
competition brief was simply to improve on the reference project if possible; to propose the best
scheme - the most beautiful, the best technical and the best environmental solution - within the
budget costs that has been agreed by the two Governments. Three independent panels would
judge the competition: an aesthetic, an environmental, and a technical and cost panel, which
indicated that the client would attach equal importance to these factors: the client clearly wanted
a balanced design.
Our design strategy was to use a simple and straightforward design: to express function and
structure in a direct way without unnecessary detail. We used well proven design concepts,
materials and construction methods to create a strong and robust form, capable of safely being
divided into smaller parts, which could be detailed by different contractors and still make a
harmonious and coherent whole.
As it would be for a design and construct contract, the design should allow the contractor some
choice of construction methods and materials to suit his particular skills and techniques. The
design, the materials and the methods of construction should be well proven to minimise the risk
of changes being necessary during the detailed design and construction and, given the difficult
site conditions, allow for repetition and for minimum on-site construction for better quality and
reduced risks.
Aerial view of the Link. Copenhagen is to the left and Malm to the right. The 4km tunnel
under the Drogden channel is adjacent to Copenhagen Airport, the 4km artificial island Pepperholm is south
of the bird protection area on Saltholm island and the 8km bridge arrives at Lernacken south of Malm.
4. Design Concept
Just two months were granted for the competition, which made the concept stage even more
crucial than usual. The main features of the design had to be settled very early, since there would
be little time to make fundamental changes later on as the design was developed and refined, if
the basic concept was then found to be unsuitable.
A small team of engineers and an architect, working alongside each other, created the
competition design. They shared an approach to design that had been formed over some years as
they worked together on a number of other bridge projects. They had learnt to understand what
each of the team members could contribute to the whole and knew how to combine the
contributions of their many different skills and experience. No solution was taken for granted,
every suggestion was examined and challenged, and nothing was accepted until the proposal was
not only technically and functionally sound and efficient but also delightful.
The form of the alignment is important for the experience of the journey across the bridge. We
chose to alter the reference projects straight alignment to a curved alignment to give a more
interesting journey across the resund.
We also decided to separate the road and the railway traffic on the whole Link, a decision that
meant that the bridge would have two levels. The Link should provide a dual two-lane motorway
and two tracks for high-speed passenger trains and heavy goods trains. In the reference project
the road and railway were placed side by side on the bridge, but separating the traffic gives
obvious operational advantages and flexibility. During lane closures, whether caused by
accidents or maintenance, road traffic can be directed onto the other carriageway, and on the
railway, crossovers can be placed freely. With the motorway carried at the upper level and the
railway below it, users are also given ample comfort and security. Cars are separated from the
high-speed trains and travellers can have free and excellent views of the resund.
We decided finally that the high bridge would be cable-stayed. For a two-level bridge, the most
economical structure is to use steel trusses with diagonals connecting the upper and lower decks.
The deep girders naturally lead to longer approach spans, which has environmental advantages
and gives a lighter and more elegant appearance. They are also rigid enough for a cable-stayed
bridge to comfortably span longer than the reference projects 330m and 290m spans over
Flintrnnan and Trindelrnnan. We therefore proposed a single navigation span of 490m over
Flintrnnan alone.
5. The Journey across the resund

The main attraction of the journey across the resund will be the unique views from the bridge: views of the ever-changing sky and sea with a remarkable play of light including every shade of
grey and blue, and views towards the gently curved coastlines, to the island of Saltholm and to
the cities of Copenhagen and Malm.
With the curved alignment, a drive across the bridge will be a continually changing experience
and the cable-stayed bridge, which is the visual peak of the journey, will be seen from changing
angles, whether one is driving from Sweden or from Denmark.
We proposed an S-curve for the competition design, as
this would give the best views of the high bridge
whether it is approached from east or from west. It
would also give a nearly perpendicular crossing of the
Flintrnnan navigation channel, and the straight cablestayed bridge could be accommodated between the two
curves of the S. Later, however, the alignment was
altered from an S-curve to a C-curve when changes
were made to the alignment at Copenhagen Airport in
order to simplify construction through the airport and
also improve the water flow through the Link.
Although the changes were made 8 km from the start
of the bridge they altered the bearing of the alignment
at the artificial island and this, in combination with
bird sanctuary restrictions south of Saltholm, made the
S-curve impractical.
The motorway is on the upper level and the cars are
separated from the high-speed trains, which means that
motorists have a free view of the resund. The railway
runs inside the structure. To give railway passengers a
clearer view, the trusses have been given a more open
bracing than is usual. This in turn helps to make the
bridge appear lighter and more transparent.
Changing views from the S-curve
Roads - and railways in particular come with a great deal of furniture: barriers, signs, signals,
power supplies, surveillance cameras, and lights, windshields etc. All are necessary but all
distract and obstruct the views from the bridge, and more so if all traffic is at the same level. On
a two-level bridge, the railway furniture can be attached to the inside of the structure, and the
road furniture alone obstructs the view from the road.
S-curve (viewed towards Sweden)
C-curve
6. The Girder as a unifying Element

The resund Bridge is not just a structure
that extends from bank to bank with a
distinct beginning and a distinct end. It is
part of a much larger link; - a road and a
railway that wind their way as continuous
bands across the water; rising from the
tunnel, passing over the island and onto the
bridge and up and over the navigation
channel.
In these circumstances, and given that the
journey across the bridge is made at the
high speed, monumental end pieces or
abutments would seem inappropriate.
Unity and continuity are fundamental, and
the approach bridges and the high bridge
should merge together rather than conflict
with each other.
Most of the competition entries for the
high bridge were cable-stayed, but two
were arch bridges. The arch is a famous structural form with great symbolic value and an easily
understood flow of forces. The cable-stayed bridge has not yet acquired the same standing. It
looks more rigid and the flow of forces is often less clear. At resund, however, where the high
bridge is part of a longer sequence, a cable-stayed high bridge joins readily with the approach
bridges; - a strong continuous bridge girder of uniform and consistent design ties the whole
length together.
The bridge is continuous from Sweden to Pepperholm, the artificial island. The simple but strong
horizontal bridge girder is the principal element that gives the design a sense of unity. The
overall effect is a clear statement of structural purpose: the strong horizontal girder is supported
on concrete piers in the approaches and, at the main span by cables which continue the line of the
truss diagonals to the two high free-standing towers.
7. Beacons in the Sea

The towers stand out like beacons, rising from the sea, aspiring towards the sky, marking the
summit of the journey and the border between Denmark and Sweden and announcing that here is
the crossing of the navigation channel. All rising lines are straight without kinks. All visible
planes reduce in size from the sea towards the sky, expressing strength and stability. The inner
faces of the towers are slightly inclined outwards, sufficient for the towers not to appear to be
leaning inwards.
The stiffness of the truss deck was a factor in choosing the harp pattern for the cables. The
repeating geometry of the truss has a natural affinity with the harp, which was highlighted by
matching the slopes of diagonals and cables. The cables are arranged in completely vertical
planes, with the cables symmetric about the pylons. All cables have the same inclination.
Whereas the cables under certain light and weather conditions will seem to vanish, the towers
will always be a prominent part of the Link.
The cable planes on cable-stayed bridges are often slightly twisted. This becomes particularly
apparent when the cables are viewed under a very acute angle, such as from the bridge, and tends
to lead to an awkward and uncomfortable appearance.
The visual formality of the harp system is particularly impressive when the cable planes are
vertical. Straight towers together with outrigger brackets at the deck make it possible to arrange
the cables in vertical planes. The centres of gravity at all cross sections of the towers are on a
vertical axis, which is also the cable plane and vertical loads induce only axial forces in the
towers. The centroid is at the third point of the cross sections so as to minimise the separation of
the two cable planes and reduce the reach required for the outrigger brackets. However, the
cables are at a safe distance from the roadway and well protected from accidental damage.
The towers are high to give a good inclination of even the longest cables, and therefore smaller
cable forces and an increased stiffness of the whole system, which is useful as the bridge carries a
high-speed railway. The towers are connected by a substantial crossbeam immediately below the
deck and by their shared foundation below water level, but they are free-standing above the deck.
The pylons are slightly higher than is strictly necessary. The extra height accommodates a
working platform and railings on top of the towers are avoided.
8. A Structure tied tightly together

The trusses are uniform throughout the bridge with
the diagonals on a constant 20m module. The
regular pattern in the approaches is modified at the
cable-stayed main span so that every other
diagonal has the same direction as the cables.
At the deck, the cables are anchored to large
sloping outrigger brackets that transfer the cable
forces from the anchorages to the bottom chords
and through the upper deck. The outrigger
brackets lie in the same inclined plane as the long
diagonals and continue the line of the cables in
elevation, in an easily understood structurally
efficient way.
The modified truss pattern, the parallel cables, the
outrigger brackets and the vertical cable planes
bind the structure so closely together that it is
difficult to alter. Changing one part would lead to
changes everywhere else. Because all the main
components of the cable-stayed bridge are
interconnected the design has an overall
coherence.
9. Some Details
In developing the design we intended the details to support and enhance the overall design
concept. The outrigger brackets are a good example. They are an harmonious part of the
continuous girder but at the same time stand out in their own right demonstrating their special
function.
All angles are rounded in the steel girder to stress the unity and the continuity of the girder. And
the structural steel is painted black. Since the girder is the unifying element in the bridge it needs
to stand out from the other parts of the bridge, which are predominantly in grey concrete. Black
also suits the natural setting as it acts a neutral background to the subtle shades of grey and blue
in the sky and the sea.
All the concrete parts of the bridge; the pylons, piers and decks have simple and strong forms.
The finishes are fair faced and horizontal lines between formwork panels have been made as
small as is technically reasonable so as not to conflict with the verticality of pylons and piers.
The two abutments, at Lernacken where the railway carries through in a tunnel and on
Pepperholm where it continues within an open viaduct structure, are designed to express the
continuity of the journey rather than emphasise the two ends of the bridge.
Much of the equipment added to the bridge; barriers, stairs, supports for lights and signs etc. are
made from grey galvanised steel showing their subordinate role to the main structure. Some of
the furniture such as the safety barriers on the motorway is in harmony with the straightforward
design of the main structure. Other, such as the outside emergency stairs, drainpipes and the
railings on the lower chord of the girder looks awkwardly out of place at close quarters.
However, the main structure with its the huge scale and simple and robust design seems to accept
and cope with this.
The road is lit. The lights are carried by a row of 12m high masts placed in the central reserve,
which will produce an evenly lit road surface and a continuous string of light in the air like a
luminous pearl necklace.
Navigation lights are attached to the towers and the main span girder and the towers are floodlit.
Spotlights are placed at the outrigger brackets nearest the towers. The floodlighting is aimed at
the cable anchorages at the tower and the shadows of the cables will form a distinct vertical line
on the upper part of the towers.
10. Getting the Balance Right

Bridges are structures just as buildings are, but engineering considerations have a much stronger
influence on bridge design. Engineering is presumed to be rational and objective and some claim
that it provides a unique solution to all problems. Everything else being equal, all engineering
works would then be much the same with little variety or room for surprise. And if one also
believes that pure rational forms are inherently beautiful, what role is then left for imagination let
alone for involving architects?
Everything is never equal. Design has a rational and an irrational part. Understanding is as
important as knowledge.
When knowledge does not suffice, the designer must use intuition and synthesis to reach a
balanced design. Technology is necessary, as help and a tool but also as an inspiration. Ove Arup
put it well: "Engineering is not a science. Science studies particular events to find general laws.
Engineering design makes use of these laws to solve particular problems. In this it is more
closely related to art or craft; as in art, its problems are under-defined, there are many solutions,
good, bad or indifferent. The art is, by a synthesis of ends and means, to arrive at a good solution.
This is a creative activity, involving imagination, intuition and deliberate choice.
The resund Bridge: The Tender Project

Jrgen Gimsing is
Technical Director of the
ASO Group, the bridge
consultant to
resundskonsortiet
Jrgen GIMSING
Director
Horsens, Denmark
Summary
In 1991 the Danish and Swedish Governments signed a Treaty for establishing a fixed Link across
the resund. The Treaty laid down the main principles for the Link: A combined 4-lane motorway
and a 2-track railway for both passenger trains and freight trains, a tunnel under Drogden, an
artificial island south of the natural island Saltholm and a bridge from here to Sweden. In 1993 ASO
Group presented a solution for the bridge in an international design competition. This solution was
refined to a Tender Design, which formed the basis for the bridge Contract with the successful
tenderer. The paper describes the design process and the Tender Design. The emphasis is placed on
demonstrating the robustness of the Tender Design by highlighting the very few and minor
modifications, both compared to the original competition design and to the Bridge actually being
constructed.
1.
Introduction
The design process for the resund Bridge has consisted in a larger number of steps than usual,
mainly due to the selected form of tendering as Design and Construction contracts for two
separate bridge contracts, one for the High Bridge and one for the Approach Bridges.
The main steps in the process were:
Competition Design
Conceptual Design
Tender Design
Contractors Tender designs
Design Evaluation of Contractors Tender designs
Basic Design
Detailed Design
Changes to Basic and Detailed Design during construction and due to changed wishes from the
Owner.
During the conceptual and tender design phases, the Design and Construction Requirements were
worked out. The Design Requirements are based on the Eurocode system, as far as known for the
first time for a major bridge project.
2.
The Design Process
2.1
The Competition Design
In November 1992 the Owner, resundskonsortiet, invited consulting engineers to seek prequalification for a design competition for the resund Link.
Each group had to present a design (Reference Project) conforming to the agreement between the
Danish and the Swedish governments and could present one or more designs (Open Proposals)
which deviated in specific instances from this agreement.
The main restrictions imposed on the conforming design were:
a fixed alignment
fixed navigation spans of 330 m and 230 m for the bridge
a one level bridge with the railway at the centre in the cross section.
The outcome of the competition was that the Owner selected two Open Proposals, a two level
bridge from ASO Group and a one level bridge from LC, for further investigations.
The Competition Design

2.2
During the first half year after the announcement of the result of the competition the two consultants
refined their designs for the total link in order to arrive at realistic cost estimates and time schedules
for construction.
At the end of this period the Owner decided to continue with the bridge designs from both LC and
ASO Group and to let LC design the tunnel and dredging/reclamation.
2.3
The Tender Design
The tender design period, from November 1993 to November 1994, involved a large number of
technical investigations both in relation to the bridge designs and to the Design and Construction
Requirements.
Both due to the decision to tender for two separate Bridge Contractors and to the Owners wish to
have a strong influence on the aesthetical design of the Bridge, all visible dimensions in the Bridges
were fixed by so called Definition Drawings. At the same time it had to be ensured that the fixed
dimensions were realistic and even nearly optimal, as deviations proposed by one Tenderer leading
to major savings would complicate the evaluation of the tenders.
Some of the investigations carried out by ASO Group for the two level bridge are described in the
next chapter.
2
2.4
Contractors Tender Designs and Evaluation
All 5 pre-qualified Tenderers for the High Bridge, 6 for the Approach Bridges submitted tenders on
the two level bridge. All tenders were conforming to the Definition Drawings, although some
tenderers indicated that they had investigated alternative pylons, cable stay arrangements etc., which
might lead to some savings in construction.
The evaluation of the tenderers designs concentrated on checking their fulfilment of the Design
Requirements including Definition Drawings and identifying qualifications hereto. A direct
comparison of the design proposals was to some degree hampered by wide differences in their
detailing, which was not specified in the Enquiry Documents.
At the end of the tender evaluation, Sundlink Contractors were chosen for both the High Bridge and
the Approach Bridges. This Joint Venture consists of Skanska (Sweden), Hochtief (Germany),
Hjgaard & Schultz (Denmark) and Monberg & Thorsen (Denmark). Their bridge consultant is CV
JV consisting of Cowi (Denmark) and VBB (Sweden).
2.5
Basic and Detailed Design
The Basic Design was defined in the Enquiry Document; it should be sufficiently detailed to give all
main dimensions, prestressing and main reinforcement and it should indicate the analysis
procedures to be employed in the Detailed Design. Both the Basic and Detailed Design were agreed
to be subdivided in more than 60 packages in order to finalise the most time critical design parts as
quickly as possible.
The Basic Design was first submitted to the Owner for commenting and subsequently resubmitted
for approval. After this approval the Basic Design drawings (to some extent) replaced the Definition
Drawings as the Owners expectations to the final product. Changes to the Basic Design found
necessary or desirable in the Detailed Design could only be introduced with the Owners approval.
The Detailed Design packages were also first issued to the Owner for commenting. After
incorporation/clarification hereof the Detailed Design drawings were issued for construction, as the
Owner did not wish to approve the Detailed Design in order to emphasise that the Contractor had
the full responsibility for this.
2.6
Changes during Construction
The Contractors designer is retained during the construction planning, as it is important that he
endorses the Method Statement for construction thus ensuring that this construction method is
covered by the Detailed Design analyses. In exceptional cases the Contractors choice of
construction method has led to a need for re-analysis, but in general the construction methods
indicated in the Basic Design have been followed.
The normal procedure for changing design details during construction is that the Contractor (or one
of his subcontractors) propose the change in the form of a Site Question. If the proposed change is a
change to the Basic Design, it shall be approved by the Owner, whereas changes to the Detailed
Design may be introduced after endorsement by the Contractors designer and subsequently issued
to the Owner for information.
3.
The Tender Design
3.1
Comparison to the Competition Design
In the Competition Design by the ASO Group all main features of the resund Bridge now under
construction were fixed. The investigations leading to the Tender Design therefore concentrated on
ensuring that the main dimensions chosen in the competition were adequate and on introducing
additional degrees of freedom for the tenderers.
The main objective in the competition was to get a reasonable compromise between economy,
aesthetics, traffic safety and environmental considerations. The first choice was a two level bridge
both due to traffic safety and environment. For the traffic safety the two level bridge is the optimum
solution, as the noise and headlights from rapid trains close to the road traffic would have been an
extra hazard. From an environment point of view the blocking of water flow from bridge piers is a
major concern, where the two level bridge leads to long spans due to a large construction depth.
The second choice was a truss girder, which ensures a good view from the lower railway deck and at
the same time is an economic solution. For the 6.6 km long Approach Bridges the optimum truss
solution is two vertical trusses, which provide the most efficient supports for the wide road deck,
when the requirements for railway clearance width and services at the lower deck have been
fulfilled. All steel members in the bridge are box sections, both for aesthetical reasons and to
minimise the extent of paint by use of dehumidification of the interior.
140 m Approach Span Model

The third choice was to minimise the number of diagonals in the trusses, both for aesthetical reasons
and to improve the views for the train passengers. This was achieved by a pure Warren truss with
diagonals under 450 instead of the normally used 600. The additional cost of this solution was found
to be very marginal as the local bending in the 20 m spans between nodal points was taken by a very
efficient composite cross section at the upper deck.
Cross Section Approach Spans

The above considerations combined with a small number of rough calculations fixed the cross
section for the superstructure in the Approach Bridges. Due to the length of the Approach Bridges
(6,6 km) compared to the High Bridge (1,1 km) at least 75% of the construction cost was the
Approach Bridges, and therefore this was chosen as the starting point in the optimisation of the
resund Bridge.
For the High Bridge the main span was not fixed in the competition rules, only that it should be
more than the 330 m stated in the Treaty between the Danish and the Swedish Governments. We
chose a 490 m main span, and subsequently this was found by ship simulation studies in the pretender period to be sufficient and adequate. With this span length a cable stayed High Bridge was
clearly indicated, even though two entries in the competition were based on arch bridges.
The chosen cross section for the High Bridge described above imposed a new problem, as the
vertical trusses positioned just outside the railway area made it impossible to attach the stay cables
directly to the trusses. The chosen solution to this was the unique outriggers, which transfer the
cable forces to the lower nodes in the trusses. This again led to two vertical cable planes with the
cables well protected from car collisions.
Elevation Cable-Stayed High Bridge

In the Competition Design the stay cables were single cables with an ultimate load capacity of
nearly 40 MN. In the pre-tender period it was found necessary to introduce an alternative solution
with smaller stay cable dimensions in order to have a wider selection of potential manufacturers.
The alternative introduced on the Definition Drawings consisted of 4 cables with their centres
forming a square. Based on a question from one of the tenderers a solution with 2 closely spaced
stay cables was also informed to be acceptable. This solution was chosen by the successful tenderer
with the twin cables in the same vertical plane.
Cross Section Cable-Stayed Spans

In the Competition Design the road deck was transversely prestressed concrete, which was also
found by most tenderers to be the optimum solution. In the Tender Design it was, however, decided
to include an orthotropic steel deck design and to state the maximum concrete dimensions
sufficiently large to allow reinforced concrete solutions. The successful tenderer used this latter
option for the in situ cast joints in the road deck above the piers, whereas the precast parts are
transversely prestressed.
The lower railway deck underwent more substantial changes in the pre-tender period. In the
Competition it was an orthotropic steel deck in the High Bridge and a thin concrete slab acting
compositly with a central, trapezoidal steel stringer beam. During the pre-tender development of
risk analyses and Design Requirements, the risk of a derailed train rupturing one diagonal was
identified as one of the major risks for the two level bridge. It was therefore decided to contain the
trains by introducing a double concrete trough with 1.8 m high walls just outside the railway
clearance profile. This solution gave additional benefits by reducing the train wheel noise and by
making the steel stringer redundant. Finally the outside of the concrete walls provided an excellent
support for horizontal cable trays for railway, communication and traffic surveillance cables.
The span lengths for the Approach Bridges were chosen to be 120 m in the competition, but during
the more detailed analyses carried out in the pre-tender period it was found that the capacity of the
cross section was sufficient to introduce 140 m spans as an alternative option for the tenderers. The
tenderers were finally given a free choice between 100, 120 and 140 m spans, and even
combinations hereof provided the shorter spans were used near the shores and either the same or
larger spans towards the High Bridge. The final choice by the Contractor was three and four 120 m
spans from the abutments, whereas the remaining 18 and 24 spans were 140 m.
Pylon: Competition Design and Conceptual Design

In the High Bridge the most visible change during the pre-tender period related to the shape of the
pylons, which became more tapered upwards and their cross section was simplified from a
hexagonal to a pentagonal shape. Finally a visible cross beam above the water was moved
downwards below the water level and incorporated in the foundation structure. The main feature of
the pylons, the free standing legs above the bridge deck, was retained following a desk study, which
demonstrated that this solution with the chosen dimensions was economic.
During the pre-tender period relatively detailed Finite Element analyses of typical nodal points in
the truss were carried out in order to verify that the proposed radii of curvature did not lead to
unacceptable stress concentrations and/or uneconomic plate thicknesses. The outcome hereof was
that the radii chosen in the competition design were found to be adequate.
The separation of the road and railway on the one artificial island comprises a viaduct, which leads
the road to the North of the railway in a gentle continuation of the horizontal curvature on the
Approach Bridge, while the railway is continued in a straight line.
In the period up to the tendering a number of consultations with authorities such as fire brigades,
police etc., led to definition of escape routes:
emergency walkways along the railway

access stairs between upper and lower deck at 600 m intervals
cross-overs in the median on the motorway.
These were introduced on the Definition Drawings.

3.2
The Bridge Contract
The tender design from the successful tenderer fully complied with the Definition Drawings. He had
chosen the steel deck solution for the railway deck in both the High Bridge and the Approach
Bridges, but before signing the Contract he agreed to construct the concrete containment platforms
in the Approach Bridges at the same cost. This solution was preferred by the Owner based on noise
studies at the Swedish coast.
3.3
Changes Introduced by the Owner
The tenderers were asked to price a number of options, which could be included in the works if the
Owner wanted to pay the extra cost. Two of these options were actually included, one being an
extension of the width of the motorway deck from 21.5 m to 23.5 m, the other being an extension by
360 m of the western Approach Bridge.
The first option was intended to be used to widen the 2.0 m service strip at the parapet to a full 3.0
m emergency lane, but it was later decided to only have 2.5 m emergency lanes and use the last 1.0
m to widen the central median from 0.5 m to 1.5 m, which would make it possible to place masts for
motorway lighting in the median.
The second option was introduced to improve the water flow through resund and thereby reduce
the compensation dredging required to obtain the zero blocking solution stated in the Treaty
between the two Governments.
After signing the Contract the major change on the Bridge initiated by the Owner has been the
introduction of a lower walkway at the railway deck. At the bridge ends both at Lernacken and at
Pepparholm, the architect has introduced some aesthetical changes to abutments, earthworks and the
viaduct.
3.4
Changes During Detailed Design
The main change introduced during the design phase was a modification to the pier tops
necessitated by increased dimensions of the bridge bearings. The outer dimensions of the piers were
fixed on the Definition Drawings based on analyses for the 120 m approach spans, but these
dimensions had not been checked for the 140 m spans, when this option was introduced as an
allowable alternative.
The overall dimensions of the pier shafts were determined by large ship impact forces, which are
independent of the span lengths and therefore the pier shafts were adequate for the 140 m spans.
However, on the pier tops a number of aesthetical and functional requirements had limited the
available space for the bearings to such an extent that the increased bearing size due to the 140 m
spans could not be accommodated without visible changes to the shape of the pier tops.
Pier Top: Conceptual Design and as constructed

The chosen solution was to omit the inclined triangle at the top of the outer face of the pier shaft,
and this provided the extra horizontal surface needed for the bearings.
On top of the bearings, the nodal point in the truss had an increased radius of curvature compared to
the other nodal points due to the concentrated forces from the bearings. During the detailed design it
was found that a further increase of this one radius was desirable due to the bearing replacement
situation, where the two spans are lifted on jacks outside the permanent bearing. Following
discussions with the architect, this increase of the difference between the bearing nodal point and
the normal nodal points was not only found acceptable but in fact preferable from an aesthetical
point of view.
4.
Conclusions
The paper has demonstrated that the design concept for the two-level bridge has been very robust in
the sense that the long design process from the first sketches in the design competition to the
Contractors detailed design has led to very few changes.
The majority of the changes have been initiated by the Owner and his consultant. The reason for
change has often been to make a feature more maintenance friendly but in some cases improved
technical and aesthetic solutions have been found.
The use of Definition Drawings as part of the Contract Documents did limit the possibilities for the
bidding contractors to modify the design concept. However, when one contractor was chosen for
both Bridge Contracts there was a risk that he would take the opportunity to introduce a number of
major changes. That this did not happen can among others be attributed to the fact that the
Definition Drawings were based on a design optimisation process and also that they represented the
design the Owner wanted to see constructed. The Owner has throughout the process insisted on a
strict adherence to the Definition Drawings.
From Eurocodes, Special Investigations and Risk Analysis

To
Design Requirements for the resund Coast to Coast Structures
Eilif SVENSSON
Civil Engineer
ES-Consult Ltd.
Vedbk, Denmark
Eilif Svensson, born 1945

Civil engineering degree 1970 and
Ph.D. degree 1973 from The
Technical University of Denmark.
Director ES-Consult from 1990
Summary
Establishing a physical link of the magnitude and importance such as the resund Coast to Coast
calls for early and focused attention towards developing and implementing a coherent strategy
dealing with the Design Requirements for the structural parts of the link.
Design Requirements are instrumental in securing the owner and ultimately the Society, that
issues such as structural safety and durability are addressed for each and every part of the Link,
in a uniform manner and independent of the contract form and contract division.
The owner addressed these issues at an early stage of the present project. This paper briefly gives
a picture of the options available at the time, the strategic decisions taken and the organisation of
the work leading to the final preparation of the Design Requirements.
1.
Early stage issues facing the owner
At the very start of the resund Coast to Coast project, when the most basic features of the Link
had been defined, which in case of the present project given its magnitude, envisaged socioeconomic and cultural impact and bilateral character required governmental decisions in Sweden
and Denmark, an owners organisation had to be established.
Needless to say at this stage the owner agenda is deluged with issues to be addressed, each and
every one representing important strategic elements in the overall process of fulfilling the goals
set out by the policy makers.
One of these issues deals with how to define and implement the desired level of structural safety
and durability. Primary questions in this context are:
-
Which set of structural codes shall be/can be applied?

Can the chosen set of basic structural codes properly fit the project at hand or will
amendments have to be prepared?
Does project specific features call for additional preparatory investigations?
A brief expose of the considerations behind these questions will be given below.
1.1
Structural and civil engineering codes
The decision on which particular set of codes to apply prompted two circumstances to be
considered:
-
The bilateral nature of the project.

The magnitude of the project.
The first evidence has the potential of leading to a mild form of schizophrenic chaos. Given the
co-operative and professional ambience of the combined Swedish-Danish owners organisation
this however never has been a real concern. Any choice between Swedish or Danish codes could
have been decided and accepted however due to the nature of the problem not without an
inherent arbitrariness attached to it.
The shear magnitude of the project itself meant that the pre qualification would attract
contractors and engineers from the international arena and ultimately that the execution of the
project only could be envisaged to be undertaken by contractors with international experience.
Both circumstances led the owner after more thorough contemplation referred to in a later section
to adapt Eurocodes as the underlying set of Codes of Practice.
1.2
Adaptation of Codes of Practice
Regardless of the choice taken with regard to the basic set of codes the magnitude of the project
dictated a thorough examination of the codes in order to reassess the suitability of all provision
concerning loads, structural verification, material specifications and other related matters.
Choosing Eurocodes as explained above introduced an additional concern. At the time of the
decision in the early nineties a final adopted and ratified set of Eurocodes did not exist. In fact all
codes and attached parts existed in premature editions the so-called ENVs.
Consequently a consistent and complete set of amendments therefore has been prepared. These
were named Project Application Documents or in short PADs.
1.3
Additional preparatory investigations
In addition to preparing the PAD, which amend provisions of the codes, supplementary
investigations may be called for to elucidate specific issues of importance not treated satisfactory
in the codes. This may be a load of a special kind and of special influence on the structure or a
special repetitive important structural feature, which one wishes to ensure focused attention in the
design process.
These considerations led the owner to initiate several extensive studies aiming at establishing the
best possible basis for the preparation of the Design Requirements.
The subsequent chapters describe more in detail the process - following the early stage
clarification of the issues mentioned above - of establishing the basis for the Design
Requirements for each main contract of the complete link.
2.
Preparation of Design Requirements
The key elements in the process leading from a clarified early stage strategy to contractual
technical requirements for the main contracts are illustrated in figure 1 below.
Imposed requirements:
Swedish-Danish treaty
Authorities
Environment
Aesthetics
BASIC INFO.
SPEC. STUDIES
RISK
ANALYSIS
CODES
DESIGN BASIS
PAD
DESIGN REQUIREMENTS
D&R
TUNNEL
BRIDGES
OTHER CONTRACTS
DESIGN AND CONSRUCT

Detailed Design
FIGURE 1. Main elements in the process leading to Design Requirements of the various
contracts of the resund Link Coast to Coast.
The Design Basis is a key document, prepared as an internal technical document containing all
technical specifications and serving the purpose of being a common platform for the various
contract specific Design Requirements.
2.1
Codes and PAD
The decision to apply Eurocodes in their preliminary editions as the basic Codes of Practice was
taken after a more thorough investigation.
Working groups consisting of Swedish and Danish experts concluded, that it was possible and
feasible to apply Eurocodes as reference codes provided amendments were prepared and certain
modifications introduced in other words that a Project Application Document - a PAD had
to be prepared.
The conclusions of the working groups, summarised in [1], contained specific priority issues to
be addressed in the preparation of the PADs.
The Eurocodes, more specific the ENVs not mentioning the associated parts were at that
time:
EC1-Structures
EC2-Concrete
EC3-Steel
EC4-Composite
EC5-Timber
EC6-Masonry
EC7-Geotechnics
EC8-Seismic
EC9-Aluminium
Further systematic review of the ENVs finally led to the preparation of the PADs, [3].
2.2
Design Basis
The Design Basis [2] (also called the Technical Design Basis) is an internal document prepared
at an early stage to compile all relevant design provision. It has been divided into the documents
Design Basis General
Design Basis Environmental
Design Basis Civil and Structural
Design Basis Geotechnical
Design Basis Mechanical
Design Basis Railway Works and Installations
Design Basis Safety
The Design Basis obviously has a much wider scope that the Eurocode + PAD which merely are
concerned with civil and structural works:
It compiles the overall and general requirements as defined in the treaty and the subsequent
government act, requirements as defined by authorities, requirements aiming at reducing the
environmental impact and general aesthetic requirements.
Other input to Design Basis has been Basic Information , Special Studies and a separate Risk
Analysis, more detailed account of which are given in the following.
2.2.1
Basic Information and Special Studies
The magnitude and character of the project has motivated the owner to supplement the
underlying Codes Eurocodes+PAD with extensive studies in order to:
Determine with the greatest possible accuracy environmental forces acting on the structure
from:
-
Wind
Waves and current
Ice
The results of these studies have been reported in [5]-[8] and has served other purposes such as
providing information of importance for the execution of the works.
Special studies treat a wide range of subjects of importance of which just a few shall be
mentioned illustrating the wide range of subjects covered:
-
Calibration of partial factors and listing of relevant load cases and load combinations
Wind tunnel tests leading to well defined design rules for aeroelastic forces acting on the
bridge and definition of design and test requirements to be met by the contractor.
Extensive geotechnical investigations improving the level of information available for the
contractor and enhancing the quality level of the Design Requirements.

Ship-Structure impact studies
Studies of the dynamic train-structure interaction
The calibration study [4] was undertaken to implement the general decision that all structural
element of the Link should belong to high safety class corresponding to a safety index = 4.7
briefly and formally defined by an annual collapse frequency lower than 1.310-6 for each
structural element.
Another and very important category of special studies have had a direct influence upon the
predefined i.e. the mandatory features of the project. An example of this is:
The span of the high bridge 490 m has been determined after several simulation studies
undertaken by The Danish Maritime Institute and SSPA Maritime Consulting Sweden. The
outcome of the simulation studies was a determination of the free span of the high bridge in
fact an increase compared to earlier concept studies and a realignment of the navigation
channel. A summary of these studies is given in [9].
2.2.2
Risk analysis
Separate studies have been undertaken by a permanent group with representatives from the
owner, the owners consultants and external experts in order to ensure that the general risk policy
formulated by the owner:
-
The total individual user risk for road or railway users shall not exceed the comparable
level for the user risks on a Swedish or Danish motorway or railway having similar length
and traffic intensity.
User risk comprise events and accident scenarios, which do not necessarily compromise the
structural integrity of the link, but which may imply closure of the link for shorter or longer
periods of time. User risks are composed of
1)
2)
Ordinary risk comparable to accidents occurring at ordinary motorways or railway

Link specific risks
Some contributing factors to ordinary accidents are not present at the link (such as road
intersections and branching) therefore the ordinary risks for motorway users on the link have
been found to be smaller than for similar motorways.
The net result from both contributions as reported in [11] show that the individual risk measured
as number of fatalities per 1 billion passages of the Link is at a satisfactory level compared to
the declared general policy. 21 fatalities per 1 billion passages for road users and 4 per 1 billion
passages for train users. The corresponding figures for motorway and railways are 33
respectively 4 fatalities per 1 billion passages.
The risk studies for the operational phase have been collected in separate reports and technical
notes, an overview of which is given in [13], and annually updated in the Operational Risk
Analysis reports, the last being ORA 98.
Account sheets for each and every contributing scenario quantify the calculated frequency and
the consequences hence giving a risk measure and contribution to the total risk from the hazards
considered. These accounts are subject to review and are enclosed as part of the annual ORA.
Some of the various accidental actions included in the risk analysis are listed below:
-
Train derailment loads

Sunken ship on tunnel roof
Falling/ dragging anchor (tunnel roof)
Fire loads
Seismic loads
Rupture of cable stay
Ship collision
Aircraft collision
As previously mentioned, the Operational Risk Analysis considers scenarios which do not
necessarily lead to structural collapse or compromise of structural integrity.
However an interface directly to the part of the Design Requirements dealing with structural
safety does exist for most accidental loads. An example is ship impact with the bridge structure,
which turned out to be important:
The risk analysis includes assessments of collision frequencies for each relevant part of the
structure, piers for head on - and sideways collisions and superstructure for deckhouse
collisions, for the population of ship envisaged by the prognosis.
Combining the collision frequency model with a mechanical/dynamic model for the impact
scenarios the ship collision design loads for each part of the structure has been defined to meet
the formal high safety class target. The risk analysis of course has been taken further to include
risk contributions not leading to structural collapse but being associated with other user risks,
for example as reported in [14] and [15].
A summary of the various and extensive studies carried out in relation to ship collision is found
in [10].
3.
Design Requirements
The Design Basis has been prepared as an internal background document being a common basis
for technical requirements of the various contract documents.
The technical requirements the Design Requirements for the various contracts, confer figure 1,
of course reflect the different character of the contracts but on the other hand also include
identical requirements for works/ structures of a common nature.
The contracts of course share the same underlying codes and PAD, also they share the same
requirement to loads and load combinations as determined in the supplementary studies and
described in the Design Basis.
4.
Concluding remarks
At this time not long before the opening of the Link it is possible to take a retrospective view
upon the results of the process described above.
It seems to be a fair assessment, that the early preparations including the early strategic
decisions focusing on defining and describing the technical requirements, followed by a
consistent and thorough implementation process headed by the owner will contribute to a
successful final result.
5.
References
[1]
Eurocodes as Reference Codes, resundskonsortiet, ES-Consult 1993-06-14
[2]
Project Technical Design Basis, resundskonsortiet, LC, May 1994
[3]
Project Application Document, resundskonsortiet, LC, 1994-03-16
[4]
Calibration of Partial factors and Load Combinations Phase 0, resundskonsortiet,

LC, 1993-12-20
[5]
Basic Information. Ice Conditions, resundskonsortiet, DHI/LIC, 1993-08-31
[6]
Basic Information. Water Level Conditions, resundskonsortiet, DHI/LIC, 1993-08-31
[7]
Basic Information. Wave Conditions, resundskonsortiet, DHI/LIC, 1993-09-03
[8]
Basic Information. Wind and other Meteorological Conditions, resundskonsortiet,

DHI/LIC, 1993-09-22
[9]
Summary Report of Ship Simulations in the Flinte Channel and the Drogden
Channel, resundskonsortiet, ES-Consult 1998-03-06
[10]
Summary Report - Ship Collision, resundskonsortiet, LC, 1994-12-15
[11]
Operational Risk Analysis ORA-97, resundskonsortiet, LC, April 1998
[12]
Risk Account ORA-97, resundskonsortiet, LC, April 1998
[13]
ORA collection of Memos, Vol. I and II, resundskonsortiet, LC, April 1998
[14]
Ship-induced Derailment on a Railway Bridge, J. Jensen, E. Svensson, H.J

Eiriksson and F. Ennemark, IABSE Struct. Eng.., Vol.6, no.2, 1996
[15]
Train derailment due to ship impact on bridges, E. Svensson, Proc. Int. Symp. Adv. Ship
Collision, Ed. H. Gluver & D. Olsen, Copenhagen, 10-13 may 1998.
Detailed Design of the Cable Stayed Bridge for the resund Link
Lars HAUGE
Dept. Head, Major Bridges
COWI
Denmark
Anton PETERSEN
Dir., Bridges
COWI
Denmark
Mr. Hauge graduated from the

Technical University of Denmark
in 1986. Since 1990, he has been
employed by COWI, where he at
present is head of the department
for design of major bridges. Mr
Hauge was in charge of the
detailed design of the cablestayed bridge for the resund
Link
Mr. Petersen has since his

graduation from the Technical
University of Denmark in
1974, been employed by
COWI, where he currently is
director for bridges. Mr.
Petersen has been the project
manager for the detailed design
of the resund Bridges.
Summary
The 7.7 km long resund bridge is a major part of the resund Link between Denmark and
Sweden. The most significant element of the entire Link is the cable stayed high bridge
spanning the navigation channel. The bridges were tendered on a design-built basis leaving
the responsibility for the design with contractor. Sundlink was awarded the contract to built
the bridges in 1995 and subcontracted the design to CV Joint Venture, comprising COWI
from Denmark and VBB from Sweden. This article describes the detailed design of the High
Bridge.
Fig. 1
The resund Bridges
The bridge will carry a four lane motorway with emergency lanes and dual tracks for a high
speed railway, and will when completed be the longest cable-stayed bridge for high speed
railway. The traffic is arranged in two levels with the roadway on the upper deck and the
railway on the lower deck. The rails are laid in ballast over the entire length of the bridge.
1 Design Requriements
Very early the Client decided the project to be based on the Eurocode system. The problems to
integrate two different national standards, the Danish and the Swedish, were thus avoided.
The applicable Eurocodes comprise:
EC1 Basis of Design and Actions on Structures
EC2 Design of Concrete Structures
EC3 Design of Steel Structures
EC4 Design of Composite Steel and Concrete Structures
EC7 Geotechnical Design
At the time, when the Enquiry Documents were prepared (1993), the Eurocodes were only
available as European Pre-standards (ENV's) or in draft versions. To obtain a fixed
contractual basis for the design, it was decided to select a specific version for each of the
Eurocodes as a reference document.
Project Application Documents (PAD) have been prepared to supplement each of the
Eurocodes. The PADs have the same function as the National Application Documents
(NADs) developed by the CEN member countries, which implements the Eurocodes.
Similar to projects as the Great Belt Link, general design requirements were specified by the
Client in addition to the Eurocodes and the PAD's to cover special features of a major civil
construction work.
The general design requirements cover the areas :
Functional and aesthetical requirements as alignment, gradients, cross-sections and
clearance profiles.
Civil and structural loads, load combinations, and partial safety coefficients. Methods of
structural analysis and design.
Geotechnical requirements to geotechnical design and construction, including soil strength
and deformation parameters.
Mechanical and electrical requirements to tunnel and bridge installations, including
systems for supervision, control and data acquisition (SCADA), power distribution, traffic
control, communication, etc.
In case of discrepancies between the various documents, the hierarchy between the different
elements of the Design Requirements is:
Design Requirements - Volume 1 (Design)
Standards and codes specifically referred to herein with the exception of the Eurocodes
Design Requirements - Volume 3 (PAD)
Eurocodes
Fig. 2
Order of precedence of documents
The resund Link is the first major civil works project, which to a large extent has been
designed according to the Eurocodes. Considering the complexity of the project and the state
of the applicable Eurocodes, the system has proved to an operational but very comprehensive
design basis.
2 High Bridge, General

The high level bridge is outlined as a cable-stayed bridge with a main span of 490 m and a
total length of 1090 m. The cables are arranged in a harp system with an inclination of
approximately 30. The cables are supported by 203.5 m high H-shaped pylons. A tie-down
system is arranged at the piers closest to the pylons to balance live load in the main span.
Expansion joints are provided at the transition to the approach bridges.
Fig. 3
High level bridge, Elevation
The overall performance of the bridge is dictated to secure a safe and comfortable operation of
a railway connection. The comfort to passengers has been verified trough analysis of the
vertical acceleration of passenger coaches from different train configurations.
Considering the navigation clearance special attention has been paid to shrinkage and creep,
of the concrete pylons and the concrete bridge deck, as the requirements to the navigation
clearance have to be fulfilled both when the bridge opens and after 100 years. The vertical
deflection due to shrinkage and creep from year 0 to year 100 in the centre of the main span
has been calculated to 155 mm.
The backbone in the verification of the design has been a 3D finite element model of the
bridge. The computer model includes all truss members as beam elements, and the roadway
deck as shell elements, and allows for individual optimisation of all elements. A total of
approximately 70 construction stages has been modelled to ensure that the correct force
pattern in the completed bridge structure is obtained, and to ensure the correct production
geometry of the elements. The model considers shrinkage and creep in accordance to
Eurocode 2.
Further, time-history analysis has been carried out to determine dynamic amplification factors
for passage of train loads and ship collision
loads.
3 Substructure
3.1 Pylons
Rising 203.5 m above the sea level, the
pylons of the cable-stayed bridge will be
the landmark of the entire link. The pylons
are designed as clean Hs without an upper
cross beam and with the legs disappearing
directly into the sea. The outer shape of the
visible parts of the structures was
determined by the SK and part of the
contract.
The pylons are founded directly on
Copenhagen limestone in level -17 m and
level -18.5 for the east and west pylon,
respectively. The foundation structures are
cellular caissons with a footprint of 35 m x
37 m.
Fig. 4
Pylon, elevation
One of the key issues in the design has
been to keep the dimensions and the weight of the caissons as low as possible because of the
size of the available
dry-dock facilities
and the available
draft during towout. Intensive cooperation between
designers and
contractors was
established
throughout the
project to achieve
these goals. The
result lead to a
design which to an
Fig. 5
Pylon caisson
outsider may look
inefficient, as for instance the ribs, supporting the bottom slab. But a total optimisation
considering design, fabrication and transport lead to the present design. The structure is posttensioned in the bottom slab, in the ribs and wall and in the top slab.
The dominating load for the foundation is
[MN]
ship impact. The caisson is designed to
100
0
withstand a ship collision force of 560 MN
0.00
-100
in the longitudinal direction and 438 MN in
-200
the transverse direction of the bridge
-300
excluding dynamic enhancement. Dynamic
-400
-500
ship collision analyses were carried out on a
-600
computer model of the entire bridge to
-700
determine the dynamic enhancement and the
amount of load being transferred trough the
cable system to the other pylon. The
Fig. 6
calculation showed peak-forces of 638 MN
and 651 MN for the east and the west pylon
caisson, respectively.
Shear forces in pylon leg
0.20
0.40
0.60
0.80
1.00
1.20
1.40
[s]
Shear force
Sideways ship impact

force
Shear force at level -17.0 m during

ship impact
Advanced soil/structure interaction calculations were carried out to demonstrate the bearing
capacity and the requirements to plastic deflections after a ship collision. Prior to tender, the
Client, had carried out extensive field and laboratory tests, including determination of stress
strain curves for 1m x 2m plates which made it possible to determine a stress limit for the
limestone.
This lead to the following verification procedure for ship collision:
1. The dynamic amplification of the impact force was determined, based on a linear elastic
time history analysis of the ship collision on a 3D finite element model of the entire cablestayed bridge with the foundations modelled as
linear elastic springs.
2. For the verification of the soil/structure
interface a non-linear material model, a DrugerPrager cap model, was calibrated, based on
full-scale tests carried out by SK.
3. The force-displacement curve for the
foundation, subjected to the enhanced ship
collision force, was determined using a nonlinear model of the caisson and the surrounding
soil.
Fig. 7 Plastic zone development
4. The linear elastic time history analysis was redone with the spring constant for the foundations, found in the force displacement analysis.
5. The bearing capacity of the foundation was verified.
The pylon legs have an outer pentagonal shape, selected for aesthetical reasons. Having the
centre of gravity dictated by the position of the cables, the cross section is balanced with thick
walls towards the girder and thin walls at the opposite side to avoid long time deflections from
normal forces in the pylon. Below the cross beams, the reinforcement in the legs are to a large
extent governed by ship collision. From level

17.5 (level of main deck for colliding ship) and
down an intermediate wall of structural
concrete has been introduced in the pylon legs
to increase the shear capacity and nontensioned bars (36 mm diameter) with a yieldstress of 1030 MPa were applied in the
transition between legs and caisson to be able
to fulfil the requirements during ship collision.
Above the cross beam at level +55 m, the pylon
legs are designed to resist an accidental impact
load of 8 MN in any direction. Due to
aesthetical considerations the cross beams are
omitted above the bridge girder which results
in almost 150 m free standing legs in the
transverse direction of the bridge alignment.
This leads to heavy vertical reinforcement
above the cross beams of the legs.
The cable stays are anchored in individual steel
boxes, one for each pair of stays. The
horizontal component of the cable force is
transferred directly through the steel between
two opposite cables, and the vertical
component is transferred to the concrete via
Fig. 8
shear studs.
Pylon, stay cable anchorage
Extensive analyses of the interaction between concrete and steel have been carried out to
verify the strength of the steel boxes and the crack widths in the concrete around the anchor
boxes.
The quantities for the pylons (two
pylons) are 35.000 m3 of concrete, 5700
tons of reinforcement and 730 tons of
post-tensioning cables and bars.
3.2 Piers
The sidespan and anchor piers are
designed as the approach bridge piers
which in total comprise 51 piers all
founded on open cellular caisson
structures directly on Copenhagen
limestone To limit the flow resistance, all
caissons are submerged into the seabed.
Depending on the water depth, the piers
must be able to resist different ship
impacts. The fine tuning of the bearing
capacity has been carried out by filling of
the piers. Eclogite ( = 26 kN/m3,
Fig. 9
Approach bridge, piers
saturated) and sand ( = 11 kN/m3, saturated) have been applied.

A goal has been to achieve as much uniformity as possible for a simplified production. In the
design, a parametric 3D finite element model was established, enabling individual
optimisation. COWI's in-house developed computer programme, IBDAS, was used. The
programme is especially suitable for parametric modelling and has the advantage of being
integrated with the CAD system. The drawings for construction were produced automatically
from the 3D finite element model, thus assuring a rational production with a high degree of
"automatic" quality control.
4 Superstructure
4.1 Bridge Girder
The bridge girder for the
high bridge is arranged as a
steel truss girder with an
upper transversely posttensioned concrete roadway
deck and a lower deck for
the railway, designed as a
closed steel box. The
inclination of the truss
members is arranged with
approximately 30 and 60,
respectively, allowing the
flat diagonals to match the
inclination of the cable
stays. The cables are
anchored to the girder on
outriggers with the same
Fig. 10
inclination as for the flat
diagonals.
High Bridge girder, cross section
The lower steel deck has proved to be a robust

structure, with a satisfactory post-accident
performance when the structure is being
subjected to train derailment, which results in
loss of a diagonal or loss of stay-cables and
fire.
Diagonals, chords and the railway deck are in
steel grade S420 (EN10113) except for the
secondary structures inside the railway deck
which are designed in S355 (EN10113). Plate
thickness' vary between 9 mm in the underside
of the railway deck at the centre of the main
span to 50 mm in the webs of the nodes at the
pylons.
Fig. 11
Dehumidification system
The interior of the steel truss is protected from corrosion by dehumidification. By keeping the
interior air below 60 % relative humidity no corrosion will occur.
The design ended up with 15.800 tons of structural steel, 11.000 m3 of concrete and 1900 of
tons post-tensioning cables.
4.2 Cables
The cables are arranged in a harp system with all cable stays in parallel. As for recently built
cable-stayed bridges as the Normandy Bridge in France and the Second Severn Bridge in UK,
PWS (Parallel Wire Strands) cables have proved to be competitive.
The effects from buffeting, vortex shedding, forced vibrations of the supports and rain/wind
induced vibrations have been investigated analytically. A final verification of the rain/wind
induced vibrations has been carried out by wind tunnel testing. The cables are designed with
an outer PE-sheeting with helical ribs as a countermeasure for rain/wind induced vibrations. In
addition, the cables will be prepared for later erection of tie-down ropes if excessive vibration
should occur. The maximum transverse amplitude of the cable vibration must be less than
1/3000 times the cable length and also less than 60 mm for a 10 minute mean wind velocity of
20 m/s at a height of 100 m.
In total 2.700 tons of stay-cables are used.
Design coordination of a design-build project

Ingvar OLOFSSON
Vice President
Skanska Teknik AB
Gteborg, Sweden
Ingvar Olofsson is head

of the Division for Design
of Bridges and Civil
Engineering Structures
within Skanska Teknik
AB, a subsidiary of
Skanska AB.
Summary
The architectural and structural design is in a design-build project developed through all the
different phases of the project. The conceptual design, where the general and functional
requirements for the project are defined, is carried out by the Owner and his consultants. Pretender designs, tender design, basic design and detailed design are performed by the
Contractor and his consultants, during the latter phases in close contact with the Owner.
The paper discusses how the Contractor and his consultants in major design-build projects
perform design coordination and design management. Examples are given from the
organization of the design for the resund bridges.
Professional design management, strictly defined design agreements, devoted and competent
design organizations within each one of the design phases are, together with positive and
open atmosphere in the relations with the Owner and other involved parties, essential
ingredients for the successful outcome of a design-build project.
1.
Design phases
1.1
Pre-tender phase
In order to reach the primary goal (to win the contract) the Contractor and his tender project
group should be prepared to start their work very early, if possible long before the inquiry
documents are released. Even if the particulars of the project requirements are still unknown,
getting familiar with the project and the local conditions is essential. The core of the future
competitive design team is created here and introduced to the project. Hence many of the tasks
mentioned below should be handled during this stage.
Collect as much information as possible of the future project. Soil investigations made for
neighboring structures are often available as well as reports of previous proposals. Even
reports of rejected concepts may be of interest, as they can reveal information regarding
anticipated difficulties.
1.2
Identify critical parts or areas, where special competence or experience will be needed.
Complex or difficult parts of the structure should be carefully evaluated especially when
new methods or inventions are being introduced. In some cases and in major projects,
planning for necessary tests and simulations could be made in this stage.
Allocate best possible design and engineering resources to the group. Individuals with
specialist competence should be identified and engaged. As many winning concepts are
connected with the choice of efficient construction methods (rather than optimizing of
quantities) a close cooperation within a team of experienced designers and construction
engineers is essential when developing the winning concept.
Organize brainstorming meetings where alternative approaches and solutions are invented
and scrutinized. A screening of available and possible construction methods, design
systems and material combinations can be started up during carefully organized
brainstorming sessions. Even if such a session should be unlimited and free of its nature, it
has to be supervised and documented. Subjects and focuses should be chosen and varied.
No options should be disqualified until disadvantages are obvious and proven.
Select alternatives for detailed evaluation. In addition to the selected obvious
alternatives also some other, promising alternatives should be chosen for detailed
evaluation. New possibilities will often appear in connection with discussions around
improvements to an impossible option.
Estimate the need for design resources in the following phases. Be sure that your internal
and external resources are well equipped not only with regard to experience and
competence for a handful of specialists but also in numbers of qualified engineers and in
status of computer facilities and software. Social competence is needed to achieve good
relations and smooth cooperation with authorities and Owner.
Prepare for a possible prequalification together with the chosen internal and external
design consultants. Be aware that some consultants may be disqualified from participating
in late stages of a project if they are challenged by involvement in previous phases.
Prequalification phase
Prequalification is made based on the experiences gained during the pre-tender phase.
Possible exclusivity for the tender phase and the future design phases should be discussed or
formally agreed with the consultants, with regard to confidentiality, strategy or the specific
requirements of the Owner.
1.3
Tender phase
The tender phase will to a major extent contain the same activities as the pre-tender phase.
Based on the knowledge of the Owners requirements some alternatives can now be excluded
from further investigation, while some might be added. The design resources are split up in
different sub-project groups, each group working according to a time schedule. Sufficient time
and resources for evaluation of new ideas and concepts popping up during the course of the
work must be available. A close cooperation between the design team and the Contractors
own engineering staff is essential for a good result.
In order to maintain the intentions given by the Owner special attention should be given to the
esthetical appearance of the structures as well as to environmental issues and of course to
matters related to quality, durability and maintenance.
3
Risk analyses could be carried out for crucial construction phases or for comparison of risks
(and the related costs) in different construction methods. While risk analyses are performed
according to formal methods, the quality of the output always depends on the quality of the
input. Once again, experience is essential, not only from making risk analyses but also
experience gained from real construction works under corresponding conditions.
Design work performed at this stage should aim at providing a winning concept for the
Contractor. It should of course arrive at sufficiently low and correct quantities and costs, to be
maintained in the construction phase. Likewise important is that calculations and other
documentation are prepared for immediate continuation in the construction phase.
At the end of this phase the conditions of the future design contracts should be negotiated with
the different design consultants.
1.4
Basic and Detailed design (Construction phase)
Based on the Contract with the Owner and the time schedule, as defined in the Contract, the
Basic design and the Detailed design is performed in parallel with the construction work.
To save time and make an early start of the construction work possible, the Basic design phase
should be an immediate continuation of the tender design. If possible the same general
calculations and drawings should be used after proper, minor updating.
In a design-build contract all parties must realize, that method statements and design
documents must be successively approved and released for construction before the total
project is known to its details. A suitable submission procedure must be worked out and
agreed upon early. The procedure should define structural parts, which are technically
independent of each other and thus suitable for independent handling and release. It should
also be possible to define different design levels, i.e. design documents to be approved by the
Owner, to be sent for information to the Owner or for internal review by the contractor only.
One of the main advantages with a design-build agreement is that construction methods and
design details from the very beginning can be adapted to the contractors available equipment
and techniques. The synergy effect arising from a close and serious involvement by the
contractor in the design development is important. The responsible site engineers (not only the
managers) should thus contribute in the early deliberations regarding transport and lifting
equipment, crane capacities, construction joints, bar lengths, bar splicing, joining of steel
members etc.
The above implies further that agreements with possible subcontractors should be closed soon
after signing of the main contract, especially in fields where extensive or specific detailed
design influencing other involved parties, has to be performed.
A considerable design effort has normally to be laid down in this phase in connection with
change orders from the Owner and from the Contractors engineering teams. Careful planning
and cost control (follow-up on quantities of construction material as well as design costs) of
the design activities are crucial to avoid delays as well as excessive construction and design
costs.
1.5
Endorsement
The late design phases and the successive finishing of the construction works are
accompanied by endorsement of the construction material, the workmanship and the
completed works. It is recommended that some of the experienced designers involved in the
project take part in these activities as all design requirements and specific details of the project
are well known to the designers.
2.
Design contracts
Design contracts are normally concluded separately for the different phases. Possibilities and
options are given to the consultant to continue into the next phase, if successful. Local
standard agreements (i.e. the Swedish ABK 96 General Rules of Agreement for Architectural
and Engineering Consulting Services) or international standard contracts are commonly used
as the basis for the design agreements. Some general comments to the different types of
agreements are given below.
2.1
Fixed fee agreements
Fixed fee or lump sum agreements tend to be used when conditions and requirements are well
known and design changes are not anticipated or rare. Hence, it is not recommended for a
tender phase, but could be applied in other phases. The risk that the quality of the presented
design might suffer from an underestimated lump sum price, should however not be
disregarded.
2.2
Variable fee agreements
A variable fee agreement is commonly used in combination with agreed hourly rates for
different categories of engineers. Expenses are compensated at verified costs together with an
additional, agreed percentage on the prime cost. Restrictions regarding maximum, total fee or
reduced rates when exceeding a certain fee target level are commonly applied to this type of
agreement.
2.3
Incitement and bonus agreements
To stimulate the consultant to use his resources efficiently, and keep design costs as well as
quantities low, different kind of incitement and bonus agreements are often discussed and also
applied. The following types are common (combinations and variation in details are however
frequent)
Reduced rates in a tender phase combined with a future success fee

Sharing of profit if final design cost is below the target level
Quantity bonus if the final built-in quantities are below the tender quantities
The two latter alternatives involve a considerable effort in follow-up and time-consuming
discussions. A carefully elaborated agreement of this kind can however be rewarding to the
consultant as well as to the contractor.
Another option, so far rarely used, is to form a consortium between the consultant and the
contractor, thus sharing risks as well as possible profits.
3.
Design management
The management of the design work for a major design-build project requires a number of
routines and administrative tools, which should be carefully chosen and developed in an early
stage of each design phase. The most important documents, the quality systems and the
project quality plan, should in detail specify the applicable routines and systems to be used, of
which some are mentioned below:
Organization charts for Client, Contractor, subcontractors and design organization

Definition of contact persons and information paths
Definition of duties and responsibilities
Routines for verification, submission, approval and release of documents
Routines for handling of changes, non-conformances and corrective actions
Distribution lists (including numbers to telephone, telefax, e-mail, etc)
Rules for exchange of information
Information and planning systems, specific software to be used
Design cost management systems
Meeting routines and participation rules, standard agendas, periods for meetings
Design time schedule, design milestones
Routines for interface identification and control
Diaries
Document numbering systems
Document standards and document lists
CAD-manual
Rules for safety and security
Filing systems
It should be mentioned that one of the most essential early planning tasks in a design-build
project is to define safe, simple and efficient communication means and paths. Information
should of course be provided promptly to those needing it. On the other hand strict rules
should prohibit unlimited distribution of documents just for information.
Another important issue is to find solutions for separate handling of technical matters and
issues related to design costs. Cost discussions often tend to be more time-consuming than the
processing of technical questions. Simultaneous handling will have a tendency to slow down
the rate of design progress.
4.
Design coordination for the resund Bridges
For tendering and, later on, structural design of the permanent structures for the resund
bridges the Contractor (Sundlink) have engaged CV joint venture, consisting of highly
6
qualified engineers from the Danish consultant COWI AS and the Swedish consultant VBB
Anlggning AB.
Shop drawings and, to a minor extent also the detail design, have been carried out by the
different subcontractors designers, while design of temporary facilities and auxiliary
equipment has been handled by Sundlinks internal design organizations.
The organization of the work has to a large extent followed the guidelines given above, with
the same key personnel engaged in the project from prequalification and tender phase into
later phases.
As the design work had to be carried out hand in hand with the construction works, the
following proved to be essential in order to guarantee a fast and safe design work providing a
cost efficient and high quality bridge design.
Close cooperation between Contractor and Consultant. A number of experienced design

engineers from the Contractors own staff were on full time basis incorporated into the
design teams to guarantee that the design was adapted to the planned construction
methods.
Frequent design meetings and close contacts also between the Owners and the
Contractors design organizations. Flexibility, fast decisions and willingly accepted ad hoc
meetings (often on short notice) created a friendly and cooperative atmosphere to the
benefit of the project.
An efficient internal review system. For advanced designs also external specialists,
employed by the Consultant, were incorporated or complementary to the review groups.
The agreed submittal procedure allowed for early approval of the Basic design, which
formed the basis for the consecutive Detailed design. Provided that the Basic design
remained unchanged no major comments were to be anticipated to the Detailed design.
The Basic and Detailed design was subdivided in suitable, independent design packages.
Each design package could thus be approved and released for construction more or less
independently.
Efficient communication systems, utilizing modern IT-technique for production and
distribution of documents also over long distances.
The agreement between the Contractor and the Consultant included different incitements
to stimulate and challenge the Consultant, ie success fee for successful tender design and
quantity bonuses in case the Contract quantities were not exceeded in the final design.
Everybody who has been involved a design-build project knows that no project develops
completely as planned. Unexpected obstacles or surprises will occur, always when least
wanted. Hence the successful outcome of the project will depend not only on the technical
skill and experience of the staff but to a major extent also on their creativeness and devotion
even in late evenings or during weekends. The latter is always efficiently promoted in a good
working climate with stimulation and good leadership provided by the management.
The resund Bridge, Erection of the Cable-Stayed Main Span
Lars T. SRENSEN
Sundlink Contractors HB
Malm, Sweden
Niels E. THORSEN
Monberg & Thorsen A/S
Copenhagen, Denmark
M. Sc. in Mech.
Engineering
Employed by Monberg &
Thorsen A/S since 1981
M. Sc. in Civil and

Structural Engineering
Employed by Monberg &
Thorsen A/S since 1978
Summary
The erection of the superstructure for the resund bridge is a challenging task involving
development of erection techniques for girders weighing up to 6900 tonnes.
The erection of the superstructure for the cable-stayed bridge was commenced by the erection
of the first girder in June 1998 and completion of the erection of girders and stays is scheduled
for the summer 1999.
Fig. 1 The resund Bridge
1. The Cable Stayed Bridge.

The resund bridge consists of two approach bridges and a cable-stayed central bridge 1092
meters long. The main span is 490 meters which is a world record for cable-stayed bridges
carrying both highway traffic and trains. The navigation clearance is 57 meters.
The bridge girder is a composite structure with a steel truss and a concrete deck carrying four
lanes of highway traffic on top. Inside the truss girder a steel railway deck carrying two railway
tracks is installed.
2.
Prefabrication of the Girders.
The construction of the resund bridge is to a very high degree based on prefabrication of large
elements on-shore - 8 girders with lengths 120 or 140 meters and weights up to 6200 tonnes
are used for the cable-stayed bridge.
The steel part of the girders for the high bridge are prefabricated in Karlskrona in Sweden and
transported by barge to Malm where they are unloaded at the girder reloading station and the
concrete deck is cast.
Fig. 2 Cross section of the cable-stayed bridge

The girders consist of a railway deck which is a steel box girder, two steel trusses, a concrete
road deck and steel outriggers for the attachment of the cable-stays (see fig. 2).
The girders are prefabricated in 20 meter sections complete with railway deck, trusses and
outriggers.
The girder elements are transported from the shipyard to a girder assembly yard by barge. The
20 meter long elements are painted and assembled to 120 or 140 m girders.
Girder joint
see fig.13
60 m
140 m girder
Fig. 3: Trial assembly of 20 meter section to 140 meter girder.
20 m section
During the assembly of the girders each 20 meter element is supported separately and the
required precamber is built into the structure.
When the girder assembly is complete the supports are changed to hydraulic supports with 60
meter spacing and the geometry of the girders is verified. At the girder ends a trial assembly
with the first 20 meter section for the next girder is done (see fig. 3), the joint is prepared for
welding and all equipment which shall be used for the girder connection at the bridge site is
installed.
Load out of the girders onto a barge is done by skidding and the girders are transported to
Malm.
Skidding line
Additional supports for casting
Skidding line
Fig. 4: Main bridge girder at girder reloading station in Malm.

In Malm the girders are placed on the girder reloading station and moved by skidding to the
casting positions on tracks with 100 meter spacing (see fig. 4). Additional supports are installed
under the girder and used to obtain the correct precamber for casting of the road deck. Due to
the heat development which occurs during curing of the concrete an additional precamber is
required for compensation.
The casting of the entire deck is done in one operation.
Before load out of the completed girder the geometry is checked with 100 meter spacing
between the supports, the girder is loaded with all the equipment to be used for the erection and
with the strands for the cable-stays.
The cable-stays represent a considerable weight which is used to counterbalance uneven weight
distribution in the girders and thereby to obtain a balanced girder lifting with Svanen.
3.
Erection of the Girders.
The main span of the bridge is erected using an innovative method. The girders are placed by
the floating crane Svanen on temporary support towers placed on the seabed.
Temporary
tower 2
.
Fig. 5. Installation
of a 120 meter girder with Svanen
Temporary
tower 1
Svanen was originally designed for the erection of the Great Belt West Bridge in Denmark
afterwards the height and the lifting capacity for Svanen was increased for the erection of the
Prince Edward Island Bridge in Canada. The lifting capacity for Svanen is today 8700 tonnes.
For the erection of the resund Bridge Svanen has been equipped with a special lifting tool
which is used for the erection of all the elements (see fig. 5). The lifting tool has a hinged rear
part which is opened when Svanen moves in for pick up of a girder - the tool is closed and
Svanen is ready to pick up a girder. Inside the lifting tool a spacing of 60 meters between the
girder support points ensures that the tension in the road deck due to bending moments in the
girder does not exceed the capacity.
The lifting tool for Svanen is made from high strength steel and weighs 1800 tonnes. Svanen is
capable of lifting girders weighing up to 6900 tonnes with the tool.
Due to the high girder installation level for the cable-stayed bridge recesses are made in the
concrete deck to avoid conflict with the structure of Svanen. Furthermore due to the geometry
of the lifting tool and to avoid conflict with the structure of Svanen erection of some of the
outriggers is done after placing of the girder.
Temporary tower 2
Temporary
tower 1
Pylon girder
Anchor Pier
Pier 2
Pylon
Fig. 6: Erection of the pylon girder.

The first girder (the 140 meter long pylon girder - see fig. 6) is inserted between the two pylon
towers from the main span and reaches 20 meters into the side span. At the pylon the girder is
placed directly on the permanent bearings. The main span end of the girder is placed on a
temporary tower which is placed on a grounded barge.
Temporary Guide
Arrangement
Fig. 7. Guiding arrangement for the girder at the pylon.
The clearance between the concrete deck of the girder and the pylon towers is approx. 15 cm at
each side during the insertion of the girder therefore a guide arrangement is used to avoid direct
contact with the towers (see fig. 7). The guide arrangement is equipped with hydraulic jacks
which enables the precise positioning of the girder before the placing on the permanent
bearings.
Secondary leg
Primary leg
Load distributor
Brackets for horizontal
adjustment
Hydraulic jacks
Shim plates
Support stack
Fig. 8. Temporary tower 1 on barge

On the temporary tower the girder is placed on a stack with built in hydraulic jacks for the
vertical and horizontal adjustment of the girder (see fig. 8). The stacks are furthermore
equipped with POT bearings enabling node rotations, girder deflections and temperature
expansion. During the load transfer from Svanen to the supports the temporary tower is
transferred from a fixed tower to a pendulum tower by removal of shims at the secondary legs.
Main span
Girder joint
Pylon girder
Fig. 9: Erection of the side span girder.
Side span girder
Anchor pier
Pier 2
The second girder (the 140 meter long side span girder - see fig. 9) is placed on the anchor pier
and attached to the end of the pylon section with the girder connection arrangement. During
welding of the joint with the pylon girder it is jacked up approx. 500 mm at the temporary
tower to obtain the correct geometry and moment distribution in the bridge. After welding the
girder is jacked down.
Main span
Girder joint
Main span girder
Anchor pier
Pier 2
Fig. 10: Erection of the main span girder.

The third girder (the 120 meter long main span girder - see fig. 10) is placed on a central
temporary tower at one end and the other end is attached to the pylon section with the girder
connection arrangement.
During the load transfer from Svanen to the supports the central temporary tower is transferred
from a fixed tower to a pendulum tower by removal of shims at the secondary legs. During
welding of the joint with the pylon girder the main span girder is jacked up approx. 65 mm at
the central temporary tower to obtain the correct geometry and moment distribution in the
bridge. After welding the girder is jacked down.
Main span
Casting of
joint in road
deck
Girder joint
Girder 1
Fig. 11: Erection of girder 1
Anchor pier
Pier 2
The fourth girder (the 140 meter long girder 1 - see fig. 11) is placed on pier 2 at one end and
the other end is attached to the side span girder with the girder connection arrangement. During
welding of the joint with the side span girder the girder 1 is jacked up approx. 600 mm at the
pier 2 to obtain the correct geometry and moment distribution in the bridge. After welding the
girder is jacked down.
Main span
Tower 2 is
turned 180
Anchor pier
Pier 2
Fig. 12: Half the cable-stayed bridge erected including stays.

Simultaneously with the girder erection activities the stay-cables are erected and the girders are
lifted off the supports on the temporary towers furthermore the joints and the recesses in the
road deck is cast (see fig. 12).
The temporary tower 1 is moved to a new position at the other pylon and prepared for the
erection of the other bridge half. Furthermore the temporary tower 2 is turned 180 which
brings the primary legs in position for the erection of the second main span girder.
The main advantage of the erection of the long girders is that the expensive off-shore activities
are cut down to a minimum which leads to time and cost savings compared to the normal
cantilever method.
4.
Girder Connection Arrangement
The girders for the high bridge are during erection connected to the previously erected girders
transferring the weight at one end directly to the previously erected girder.
For this purpose a girder connection arrangement has been developed using the short permanent
diagonal in the truss as a support element. Hinged upper chord connectors transfer longitudinal
forces between the ends of the chords and a contact and guide arrangement, installed on the
upper chords, control the local geometry in the transverse direction at the joint.
Contact and guide arrangement
Upper chord connector
Joint in upper chord
Short diagonal acting as

main support
Contact
plates
Joint in lower chord
Fig. 13. Girder Connection Arrangement
The diagonal is temporarily hinged at both ends to enable node rotations, when the girder
weight is transferred from the Svanen to the support arrangement at the joint and the temporary
suppport on the top of the pier or the temporary towers.
An adjustable connector controls the position of the diagonal in relation to the upper chord
before load transfer is initiated.
The contact and guide arrangement on the upper chords and the contact plates in the diagonal
are installed during trial assembly in order to assure the correct welding gap and geometry of
the joint.
As the short diagonal is inclined, the resulting horizontal drift force in the upper chord is
transferred by hydraulic jacks in the upper chord connector. At the completion of the load
transfer the final compression force in each diagonal is 13 to 16 MN and the tensile force in the
upper chord is 5.0 to 6.5 MN.
During load transfer the girder is moved forward by SVANEN until contact at the guide and
contact arrangement at upper chord and then gradually lowered until the upper chord connector
can be activated by turning the hinged part forward and by increasing the load in each jack to
500 kN and contact is obtained at the contact plates in the diagonal
Prior to the erection of the girder a load schedule has been established describing the local
geometry at the joint in relation to the rate of completion of the load transfer. When the load has
reached 50 % of the final anticipated load a survey is performed. An assessment of the
continued load transfer is made. Load transfer is continued until the total load of the girder is
transferred to its supports
Adjustable push-pull connectors with a capacity of 200 kN control the vertical alignment
between upper chords and lower chords before welding.
5.
Erection of the folded-down outriggers.

Lifting tool for Svanen
Future
position
of outrigger
Hinged rear part of lifting tool
Fig. 14. Cross section of Svanens lifting tool and high bridge girder
Folded-down
outrigger
Due to interference between the outriggers in their final position and the lifting tool or Svanens
structure itself, as shown above, in 42 positions of a total of 80, the high bridge girders had to
be erected with half of the outriggers in a down-folded position. Hinges were installed during
fabrication in Karlskrona at a joint in the upper outrigger and at the joint between the lower
outrigger and the lower node in order to ease the installation of the outrigger at the bridge site.
6.
Connection of the two cantilevers.
Closing the main span is an activity on the critical path of the sequence of operations. In order
to limit its duration, loose members to be fitted during this operation are prefabricated and
prefitted to highest extent possible.
When all the girders and the stays are erected the two cantilevers are joined at the centre of the
main span. While the connection operation takes place the joint is exposed to environmental
loads resulting in bending moments and shear forces and normal forces at the joint. Therefore a
cantilever connection arrangement has been developed to overcome these solicitations during
the welding operation.
Horizontal push-pull connector
Vertical push-pull connector
Vertical lattice triangle
Horizontal push-pull connector
Fig. 15. Connection Arrangement between the two cantilevers

The connection arrangement consist of one horizontal and two vertical lattice triangles which
enable the transfer of shear forces and torsion at the joint. Bending moments from wind load on
the bridge girder is taken by strong push pull connectors installed at the chords. The system is
designed to enable rapid connection of the cantilevers to obtain a quick and safe transfer from
the two free cantilevers to one main bridge span. The anticipated duration of the installation of
the connection arrangement is less than 48 hours at a maximum wind speed of 16 m/s.
Before the closing operation is initiated, a comprehensive survey is done to establish the actual
geometry of the girder corresponding to the actual load and temperature conditions in order to
define the anticipated distance between girder ends at the upper and lower chords. This survey
will determine the length of the infill pieces between the brackets in the longitudinal bracing.
The vertical slope of the girder ends is controlled by a combination of ballasting the road deck
and stay cable adjustment to produce a relative vertical alignment between girder ends, that is in
accordance with the predefined erection geometry for this particular phase.
When correct geometry is achieved the longitudinal bracing is locked and it will now be
capable of resisting the combined forces due to temperature loads and wind load up to 25 m/s.
7.
Erection of the Stay Cables.
The main span of the bridge will be carried by Freyssinet stay-cables. 160 stays each consisting
of approx. 70 strands will be used. The 7-wire strands are individually corrosion protected and
anchored by wedges at the ends. A PEHD casing keeps the bundle of strands in position and
reduces the drag coefficient of the stay-cable.
Galvanized strand (wires)
Wax
High density polyethylene
Fig. 16. Cross section of stay cable

In order to prevent rain/wind induced vibration the casing is equipped with a double helical 2
mm thick spiral. The casing with the spiral has been tested in a climatic wind tunnel and it was
found that the spiral prevented the formation of a regular water rivulet and thereby provided
good protection against rain/wind vibrations.
Helical spiral
Twin cable connector
300
2
Fig. 17. Stay cables with the cable connector

The stays are placed in pairs. Therefor the possibilty of wake galloping has been examined by
wind tunnel tests. It was found that the distance 670 mm between the stays was sufficient to
minimize the risk of this phenomena. The stays are however equipped with connectors at
approximately 100 meter spacing to further reduce the risk of aerodynamic interaction between
the stay cable pairs.
The stay-cables are erected strand by strand. The first strand to be erected in each stay is the
reference strand for which the length has been measured with high precision by the
manufacturer ( 1/10000 of the stay cable length for the short stays and 1/20000 of the stay
cable length for the longest stays), this strand is stressed to the correct length. The first strand is
equipped with a load cell and the following strands are all stressed until the load in the
reference strand is reached.
In order to obtain the correct geometry of the bridge and the correct distribution of forces in the
stay cables a precise prefabrication of all the bridge elements and control of the geometry is
made. This provides knowledge of all offsets of stay cable anchorages in the pylons and in the
bridge deck.
Furthermore the bridge geometry and force distribution will be calculated for the various
erection stages taking into account actual erection loads on the bridge and the measured weight
of the girders.
The above mentioned measurements and calculations enable the calculation of the required
length of each individual stay cable and thereby the required length of the reference strands.
While the stays are erected using the geometry method the development of the stay force is
followed closely enabling the discovery of deviations from the calculated stay cable forces.
The vertical deflection of the bridge girder and the horizontal deflection of the pylons is also
measured at several stages during the erection and compared to the computer calculations.
The close survey makes it possible to compensate for any deviations occurring during the
erection process
Cable-Stayed Bridges with Special Features

Jrg SCHLAICH
Prof. Dr.-Ing.
Structural Engineer
Stuttgart, Germany
Jrg Schlaich, born 1934, received

his civil engineering degrees from
the Universities of Stuttgart and
Berlin, and from Case Tech.,
Cleveland, Ohio. He is professor
and director of the Institute for
Structural Design II, Univ. of
Stuttgart and partner of Schlaich
Bergermann und Partner,
Consulting Engineers, Stuttgart,
Germany
Summary
Usually if we speak of cable-stayed bridge design parameters, we have their cable-arrangement,
pylon-geometry, the cross-sections and the materials of their deck etc. in mind. But the overall
layout is considered to be more or less invariable: a three-span arrangement with two pylons, a
main-span and two holding down side-spans, and occasionally half of that with one pylon.
However, the cable-stayed bridge concept offers more and can adapt to very special boundary
conditions, from local availability of only certain materials or wires to unusual topographical
conditions.
The outcome may be e.g. one out of a large number of feasible multi-span arrangements, or a
combination of cable-stayed and cable-supported. Other situations may call for cable-stayed
bridges, where the deck is not straight in plan but curved, or even for convertible or folding decks.
1.
Introduction
A committed structural designer will case by case strive for the best possible solution,
considering all individual boundary conditions, the functional, environmental, climatic,
technological, human etc.. In certain cases, the cable-stayed approach may be the appropriate
answer.
This needs to be emphasized in times when one sometimes gets the impression that bridge designers
are too hastily restricting the canon of bridge shapes to box-girders for short spans and to cablestayed for long spans. No question, after the development of cables with simple and robust
anchorages, high fatigue strength and reliable corrosion protection in recent years, the cable-stayed
bridge has firmly established itself in the span-range between 200 and 500 meters, and is now even
approaching the 1000 m limit.
If we discuss their design parameters, we have their cable arrangements (one, two or more cableplanes in a fan-, semi-harp-, harp-shape), pylon- or mast geometries (single or double masts, pylons
with H-, A-, double-A-shape etc.), deck cross-sections (slabs, grids, boxes etc.), deck materials
(concrete, steel, composite "new" materials) in mind.
But generally if we speak of a cable-stayed bridge, we mean a three-span-arrangement with two
pylons or masts on either side of the main span and two side-spans, which are somewhat shorter
than half of the main span to counterbalance the cantilevering main span or occasionally half of that
with one pylon or mast [see the postscript].
Though, the author and his team are happy to have received the chance to contribute to this
development
-
with the first composite deck, which was even riveted to allow for an indigenous construction,
by designing the 183 m/457 m/183 m Second Hooghly River Bridge in Calcutta, India [1] and
following Ren Walther's Diepoldsau Bridge and his further research on solid concrete slabs
for cable-stayed bridges [2], [3], by extending this concept to 215 m with a 45 cm solid slab, by
designing, together with S. Stathopoulos, the Evripos Bridge in Greece [4],
the purpose of this paper is to make aware that the cable-stayed concept offers a much larger
variety of general layouts, cable-arrangements and that even its girder must not be straight in
plan. Cable-stayed bridges can in fact adapt to many different and very special boundary conditions
and thus develop a large number of special features.
In order to keep the length of this paper within the given limit, the author will try to exemplify
this statement by describing some own cable-stayed bridge designs with such special features, but
not before naming at least a few other very stimulating ones such as the multi-span Sunniberg
Bridge near Klosters in Switzerland by Chr. Menn [5], the "extradosed" Odawara Blueway Bridge
by A. Ogawa, A. Kasuga and H: Okamoto [6], or the Sancho El Major Bridge in Spain by
C. F. Casado, I. Manterola Armisen and L. F. Troyano.
2.
Combined Cable-Stayed/Cable-Supported "Obere Argen Bridge"

This valley, crossed by a 6-lane Autobahn
is 730 m wide with maximum depth of
only 45 m above the little river "Obere
Argen". Whereas between the eastern
abutment and western bank of the river the
soil conditions are good over a stretch of
about 440 m to allow for a continuous
girder on flat foundations, the slope between the western bank of the river and the
western abutment over a length of 290 m is
continuously sliding downhill at a rate of
some 10 20 cm per year permitting no
foundations. Therefore, the Highway
Authority through a design competition
invited proposals for a bridge with a 260 m
end-span followed by regular spans over
the remaining 440 m, with the aim to
reconcile these two completely different
boundary conditions in a very sensitive
rural environment in the foothills of the
Alps.
Finally from a number of alternatives a
sequence of 86 m cable-supported spans
was proposed, followed by a 3 x 86 m =
258 m combined cable-stayed and cablesupported end-span (Figs. 1 + 2).
**
*
Fig. 1: "Obere Argen Bridge": Alternatives

* proposed; ** built
With that, as against a one-sided pure

cable-stayed layout, the height of the pylon
is only half, without increasing the cable
forces respectively the amount of cables
needed. The cable alignment follows the
topography and the bridge blends in
modestly with its surroundings. In fact,
the main feature of this design, the western
cable-stayed/cable-supported 258 m endspan was built whereas on the eastern side
due to cost reasons unfortunately only
a continuous girder with 56 m span without cable supports was accepted (Fig. 3).
Fig. 2: "Obere Argen Bridge": Proposal
Fig. 3: "Obere Argen Bridge", completed 1990
3.
Multi-Span Cable-Supported Bridges
3.1
Proposals for Ganga Bridges at Allahabad and Patna
Wide, alluvial rivers like the Ganga in India require caissons or well-foundations 60 or more meters
in depth to resist scour (Fig. 4a). Thus for standard girder bridges, requiring piers at frequent
intervals, the foundations are extremely costly. The costs of the foundations decrease almost
proportional to the increase in the span, and this suggests large spans, especially with cable-stayed
decks. In the sixties F. Leonhardt's innovative proposal for Allahabad basically consisted of
individually balanced cantilevers. However, to counteract the cantilevering moment due to onesided traffic load, it avails only of a relatively small lever arm as given by the longitudinal distance
between the tower legs (Fig. 4b). Therefore, for Patna this lever arm was dramatically increased to
about 0.8 of the main span L by arranging crossing back-stays in every second span, resulting a
alternating spans of length L and about 0.8 L (Fig. 4c).
It is interesting to note that basically the "new" extradosed bridges of our days are nothing but a
revival of the Allahabad proposal.
Fig. 4 Very long bridges with deep foundations

3.2
Proposal for the Prince Edward Island Link
The Patna type requires an expansion joint at the center of every main span of length L or better a
dropped girder of say 20 m length to avoid a kink under traffic load and to account for manufacturing inaccuracies between adjacent spans. Thus such a multi-span cable-stayed bridge typically
consists of self-balanced units which are [1.8 L 20 m] long (Fig. 5). For the 12 km long Prince
Edward Island Link it was proposed to choose L = 180 m and to completely prefabricate the 380 m
long deck + pylon + cable-units under shop conditions ashore to overcome the adverse climatic
conditions. During the ice-free period, after first installing the prefabricated foundations, the bridge
units would be lifted into position by use of a floating crane (Fig. 6). This basic concept was further
developed for a crossing of the Strait of Gibraltar. In this case the pylons were to be mounted on
submerged caissons which would serve as pontoons anchored with cables to the seabed [7].
Fig. 5: Prince Edward Island Link, proposal 1988
Fig. 6: Prince Edward Island Link, proposed construction procedure

3.3
The Ting Kau Bridge in Hong Kong
This is one of the few multi-span cable-stayed bridges really built. Its cable-supported deck is
127+448+475+127 = 1177 m long (Fig. 7). The special features of this bridge are its two adjacent
main spans with their stabilizing cables which run as in Fig. 4c diagonally from the top of the
main tower towards the deck at the side towers. With further stabilizing cables in the transverse
direction, the towers of the bridge appear like masts of a sailing boat.
Fig. 7: Ting Kau Bridge

The bridge must span the 900 m wide Rambler Channel. This distance is too short for an
economical suspension bridge and is rather long for a single main span cable-stayed bridge. Further,
as any pier in the busy channel would require costly ship impact protection, the number of piers had
to be kept to a minimum. Therefore, the conventional design of a single main span bridge of 600 to
800 m with two offshore towers was excluded. With load-carrying rock generally at 40 m,
fortunately there exists an underwater hill with a peak at 30 m almost in the middle of the channel
offering the chance to place one central tower at this location and two smaller towers on either side
on-shore (Fig. 8) [8].
Fig. 8: Ting Kau Bridge, Hong Kong, completed 1998
4.
The Hybrid Main Span + Extradosed Side Spans of the Dresden

Waldschlchen Bridge Proposal
Dresden, the Florence of the North, needs a new bridge across the Elbe. A design competition was
organized to find a solution which obstructs as little as possible the famous panoramic view of the
historic city, respectively what is left of it after the air-raids during the Second World War. This
resulted in the proposal of a delicate combined suspension-/cable-stayed main span a modern
paraphrase of the Blue-Wonder-Bridge at a short distance upstream with the cable-stayed units
continuing on either side for the approaches (Fig. 9). The idea of these multiple "extradosed"
relatively short side-spans was to embrace the deck or to signal the car-driver with useful and
decorative elements that he is travelling on a bridge (Fig. 10). However, the jury decided just on the
opposite: An arch for the main span with a plain deck on either side.
Fig. 9: Waldschlchen Bridge in Dresden, proposal, 1997
Fig. 10: Model of the Waldschlchen Bridge in Dresden
5.
Another "Extradosed" Approach for a Railroad Bridge in

Stuttgart-Bad Cannstatt
Imagine the cables of an "extradosed" bridge with several spans are milled into a plane steel sheet
which is cut out where the stresses are small or imagine a haunched bridge, where the haunches are
not below the deck as usual, but above and thus in tension, then you arrive at this 4-track railway
bridge (Figs. 11 +12). The masts and the deck are from concrete, the sails, suspending the deck, are
from 80 mm thick steel sheet. This solution resulted from a design competition and is going to be
built soon.
Fig. 11: Railroad Bridge, Bad Cannstatt, view, cross-section and detail of "sails"
Fig. 12: Model of the Railroad Bridge, Bad Cannstatt (under design)
6.
Three Recent Cable-Stayed Footbridges with Special Features
We usually think of long spans if we consider the choice of cable-supported decks. But why should
we not use this means as well, to give small bridges a human scale? F. Leonhardt did so already in
the fifties with his fine Schillersteg in Stuttgart and many followed thereafter, however, mostly with
the typical three-span-arrangement or half of that. Especially for footbridges more variety is
desirable.
6.1
"Truss" Bridges for the EXPO 2000 in Hannover
For crossing several highways on the EXPO area, a number of bridges of different span and width,
some permanent, some temporary were asked for in a design competition. The successful solution
in a design competition uses a modular system of square 7.5/7.5 m slabs either directly supported
on poles at their four corners or indirectly cable-stayed where longer spans were needed. Stimulated
by the idea of the architect Volkwin Marg, that poles at equal distance and constant height,
projecting vertically beyond the deck shall signal the passage and permit the installation of lights,
flags, canopies etc. a structure was developed which is a typical harp-type cable-stayed bridge, but
may as well be interpreted as a cantilevering truss (Fig. 13). After all a cable-stayed bridge acts like
a cantilever and this remains so if the cables are not directly fixed to the tower but diverted by struts
and ties parallel to the deck at a certain level.
Fig. 13:
EXPO Hannover, footbridge
Fig. 14:
EXPO Hannover, footbridge (presently under construction)
6.2
A Folding Cable-Stayed Bridge in Kiel
This little bridge crosses the Kiel Frde, the end of a fiord of the Baltic Sea, right at the center of
the town. Its middle part with a length of only 25 m has to be opened and closed 10 to 15 times per
day to permit ships to pass. Again together with the architect Volkwin Marg we felt, that this bridge
could become a landmark of Kiel if it reflects ships and cranes and moves in an imaginative
manner. Therefore, hidden hydraulic means were ruled out as against winches and cables. Instead of
going for a simple cable-stayed solution with one mast, one hinge and one back-stay, the deck was
further subdivided to fold with three hinges, two masts and two back-stays in a more stimulating
way (Figs. 15 + 16). For safety reasons and to avoid any restraints due to irregularities in the
movement, the cable-system is statically determinate and is moved by one winch only, rotating at a
constant speed.
Fig. 15: Folding Bridge, Kiel
Fig. 16: Folding Bridge, Kiel, completed 1998

6.3
An Ondulating Cable-Stayed Bridge Proposal for Kassel
A cable-stayed bridge must not be straight in plan, but may be curved without abandoning selfanchorage. Here such an unusual geometry developed from connecting two given footpaths on
either side of the Fulda in a natural way (Fig. 17). A circular girder needs to be supported on one
side only, because it can resist torsion by in plane bending, resulting in axial compression in the
concrete deck slab and tension in the bottom chord of the supporting truss. The radial stay cables
cause ring tension in the circular deck slab which partly counterbalances the axial compression from
the torsional effect.
Fig. 17: Ondulating cable-stayed bridge in Kassel, proposal
Fig. 17: Ondulating cable-stayed bridge in Kassel, proposal (cont'd)
7.
Postscript: Pylons or Towers/Masts?
It has become a habit to call a pylon whatever props up the cables of a roof or a bridge. However,
the word "pylon" goes back to the Egyptian temple gates which clearly consist of two vertical
towers, interconnected by a horizontal beam, leaving a (small) gate beneath (a).
Thus a bridge pylon
has two legs connected in the transversal
direction with a
beam or struts and
ties to form a gate.
An A-support may
also deserve this
name (b). But any
single support is just
a "tower" or a
"mast" (c).
Acknowledgement
On these bridges the author collaborated with his partners and colleagues Rudolf Bergermann,
Andreas Keil, Michael Schlaich, Jan Knippers. Thanks!
References
[1]
Bergermann, R., Bhasin, P.C., Design and Construction of Second Hooghly Bridge in
Calcutta. Cable-stayed Bridge Seminar, 1988, Bangalore, India
[2]
Walther, R.,. et.al, Cable stayed bridges, Telford Publ., London, 1988
[3]
Klein, J.-F., Ponts Haubanes: Comportement et Stabilit des Tabliers Minces, Ph.D-Thesis,
cole Polytechnique Fdrale de Lausanne, Switzerland, 1990
[4]
Bergermann, R., Stathopoulos, S., Design of the Evripos Bridge in Greece. Cable-stayed
Bridge Seminar, 1988, Bangalore, India
[5]
Menn, Chr., Functional Shaping of Piers and Pylons. Structural Engineering International,
4/98, pp. 249 251
[6]
Ogawa, A., Kasuga, A., Okamoto, H., Prestressed Concrete Extradosed Bridge Odawara
Blueway Bridge. Prestressed Concrete in Japan, XIII FIP Congress National Report,
Amsterdam, The Netherlands, 1998
[7]
Holgate, A., The Art of Structural Engineering. Edition Axel Menges, Stuttgart/ London,
1997
[8]
Bergermann, R., Schlaich, M., Monoleg Towers with Transverse Stabilising Cables.
Structural Engineering International 4/98, pp. 252 255
Stay Cable Technology: Overview

Manabu ITO
Professor
Takushoku University
Tokyo, Japan
Manabu Ito, graduated in 1953

and given Dr. Eng. degree in
1959 from the University of
Tokyo, was engaged in teaching
and research of bridge
engineering at his alma mater.
After his retirement in 1991, he
moved to Saitama and then
Takushoku universities. He has
been involved in many large
bridge projects in Japan. He is
former vice-president of IABSE.
Summary
The cables of cable-stayed bridges are required to have superb mechanical properties such as
high tensile strength, high elastic modulus, satisfactory fatigue resistance and sectional
compactness, as well as excellent corrosion resistance, easiness of handling and installation, and
naturally not to be costly. Referring to a brief chronological developments during the past half
century, the paper reviews the technology of bridge stay cables. The fatigue problems and end
fittings are not mentioned herein, while particular emphasis is placed on the corrosion protection
and wind-induced vibrations.
1.
Introduction
The requirements for stay cables are excellent mechanical properties (such as high strength, high
elastic modulus and fatigue resistance), sectional compactness, corrosion resistance, easiness of
handling and installation, and naturally low cost. To find the solution to satisfy these
requirements, the research and development on stay cable technology have been continuously
conducted over the past half century. In the consequence, cables for cable-stayed bridges are
more diverse in their types, corrosion protection methods and end fitting, in contrast to those for
suspension bridges.
In an international survey conducted by Hamilton and Breen [1], durability and fatigue were
most highly rated keywords concerning important aspects of stay cables. Although the keyword
vibration was rated lower in this survey, suppression of wind-induced vibrations of stay cables
has been gaining importance especially on long span cable-stayed bridges since 1980s.
The present article places emphasis on the types, corrosion protection and vibration control of
stay cables, while the installation and end fitting of the cables are not referred herein, mainly
because of the restriction of space. Another excuse is that the content may be a little biased by
the information in Japan where modern cable-stayed bridge construction started rather earlier
next to Germany and fairly large amount of experiences have been accumulated.
More extensive state-of-the-art of stay cables was presented by Ohashi [2] in 1991, which is a
little old though. Gimsing also describes the basic types, corrosion protection and mechanical
properties of structural cables in his text book [3]. As for the stay cables for prestressed concrete
(PC) cable-stayed bridges, the recommendations prepared by the Post Tensioning Institute will
be a good guideline [4]. The present article owes very much to these references.
2.
Mechanical Properties of Cable Materials
Cold-drawn steel wires with round section being used for bridge cables have a diameter of 4 to
7mm and very high tensile strength as compared with structural steels although ductility is lower.
Their guaranteed tensile strength is 1500-1600MPa in spiral ropes and 1570-1765MPa in parallel
wire strands (PWS), whereas the steel wires with deformed cross section composing locked coil
ropes (LCR) and the prestressing bars have more or less lower tensile strength. A steel wire with
minimum tensile strength of 1800Mpa was newly developed for the main cables of Akashi
Kaikyo suspension bridge which has the worlds longest span, by increasing silicon content.
Although this material for parallel wire strands has not yet applied to cable-stayed bridges, longlay seven-wire strands that initiated from prestressed concrete structures display the same level
of tensile strength.
The modulus of elasticity of parallel wire strand or ultra-long lay strand is only slightly less than
that of the component steel material, but the Youngs modulus and tensile strength of multi-wire
helical strands decrease due to the twisting of wires. A typical Youngs modulus of the helical
strands is 170GPa for spiral ropes, 180GPa for LCR, and 190GPa for seven-wire strands.
New composite materials have been tentatively used on stay cables of a few short span cablestayed bridges. Such fiber reinforced plastic materials as Aramid fiber cables may be more
widely employed in the future.
3.
Types of Stay Cables
The types of stay cables used on earlier steel cable-stayed bridges were characterized by the
specific countries: namely, LCR in Germany, spiral ropes in UK, and quick transition from LCR
to PWS in Japan. On the other hand, those in PC cable-stayed bridges have been mostly
associated with the development of the post-tensioning type PC bridge technology. However,
with the development of cable technology and the transition from a few stay system to multi-stay
system, the situation has been changed.
3.1
Helical Wire Ropes
As already mentioned, multi-wire helical strands are inferior to parallel wire strands in
mechanical properties, although they are easier in handling. Therefore, spiral ropes which have
been extensively used as a main cables of short span suspension bridges are not popular in cablestayed bridges except for UK.
LCR is composed of two types of twisted wires: normal round wires in the core layers and the Tand Z-shaped wires in the outer layers. As compared with spiral ropes, LCR has such advantages
as smooth surface, more compact cross section and less sensitivity to side pressure, stiffer to
handle and a little costly though. LCR has been popular particularly in German-style steel cablestayed bridges. The largest ever used has a diameter of 174mm in the Rama IX Bridge in
Bangkok. The multi-strand cables consisting of several LCRs of smaller size that were prevalent
in the early German bridges are hardly used now mainly due to rather complicated anchorage
details and difficulty of replacement, but similar usage is still found in Norwegian suspension
bridges recently built.
3.2
Parallel Strand Cables
The seven-wire strand which has been extensively used as tendons for prestressed concrete is the
simplest and most prevalent in the stay cables of PC cable-stayed bridges. As the pitch of twisted
wires is relatively long, the stiffness of the strand is close to that of straight wire strand and its
breaking strength is even higher. For cables, the strand is normally made from 5mm wires and its
nominal diameter is 12.7mm or 15.2mm, and these strands are arranged in parallel to form a stay
cable. The number of the seven-wire strands varies from 7 to 127 dependent on the required
design force.
There are such a diversity of strand or cable systems using the seven-wire strands according to
the corrosion protection, assembling method and end-fitting techniques, as Freyssinet, Dywidag,
VSL, Stronghold, SEEE and ASP (at-site prefabricated cable system) and so on. The parallel
strand cables are either shop-fabricated or site-fabricated, and sometimes their combination like
SEEE. Cost saving is claimed by the site-fabrication of stay cables with individual strands
pushed through a pre-installed sheath. Examples of the corrosion protection will be illustrated in
the next chapter. This type of cables are also applied recently to steel or steel/PC hybrid cablestayed bridges such as Normandy Bridge in France, Kap Suimun Bridge in Hong Kong and the
new Onomichi Bridge in Japan. In case of the Normandy Bridge [5] which is the worlds second
longest cable-stayed bridge, the individual strand supplied by Fressynet has a 15mm diameter
and protected with extruded high density polyethylene (HDPE) covering. For the new Onomichi
Bridge built in 1999, each strand is also covered by extruded HDPE, and stay cables are semiprefabricated and provided with SEEE type anchorage.
3.3
Parallel Wire Cables
A parallel wire bundle of prestressing wires with a diameter of 6-7mm is incorporated as a stay
cable with a polyethylene pipe filled by cement grout as corrosion protection and with HiAm
anchor sockets as the end fittings. These parallel wire cables have been widely used on both PC
and steel cable-stayed bridges. In Japan, the HiAm-anchor cables were adopted on such large
steel cable-stayed bridges as the Meiko-West (405m span, 5mm wires, 1985), Iwaguro-jima and
Hitsuishi-jima (420m, 7mm wires,1988) bridges. In the latter, polybutadiene resin was grouted
instead of cement mortar.
3.4
Parallel-Wire Strand Cable
The shop-prefabricated parallel wire strands (PPWS) have been extensively used on Japanese
suspension bridges since 1968, even on the worlds largest Akashi Kaikyo Bridge in which the
maximum strand length reaches 4074m. The parallel wire strands to be built into suspension
bridge main cables and earlier steel stay cables comprise 5.0-5.5mm wires, whereas 7mm wires
are more common in recent stay cables aiming at more compact anchorage details and easier
cable erection. A PWS is a bundle of galvanized wires to form a hexagonal cross section, and
in many cases a multiple number of PWSs are formed into one large round cable at the site. For
example, the Yamatogawa Bridge in Japan has 16 squeezed stay cables each consisting of 19
strands of 217 galvanized 5mm wires and a plastic covering. The largest cable strand of this type
is 337 x 7mm wires used in the Parana Bridge in Argentina. In Japan, the use of PWSs started in
the late 1960s, but it was almost switched to the New PWS described in the next section twenty
years later.
3.5
Ultra-Long Lay Stay Cable
The idea of an ultra-long lay cable strand was initiated in the 1980s as the improved variant of
parallel wire strand and parallel wire cables. Twisting the wires up to 3-4o enables the wire
bundle to ease reeling and make the strand self-compacting under axial tension without spoiling
the mechanical properties. These cables are designated as New PWS which was developed in
Japan and HiAm-SPWC in Europe, respectively. The New PWS is also featured by extruding
HDPE cover directly onto the wire bundle so that no void will exists between the wires and the
surrounding cover. New PWS is assembled by 7mm wires and the thickest one comprise 421
wires. The longest stay cable of this type is 460m long with the outer diameter of 165mm used
on the Tatara Bridge, the worlds longest span cable-stayed bridge.
3.6
Bar Stay Cable
Bar stay cable consists of round steel bars with a diameter of 26-36mm, being covered by a steel
pipe, the inside of which is filled with cement grout. The external steel tube is considered in the
cable cross-section when live load is applied. Since the lengths of the bars can not be long,
coupling is normally needed. This type of stay member is scarcely used, particularly for large
cable-stayed bridges.
4.
Corrosion Protection
Quite a few corrosion troubles were reported on the stay cables despite of rather short history of
cable-stayed bridges. In the case of multiple-stay system which is prevalent in these days, the
replacement of stay cables is possible but needs extra cost. Therefore, various means to protect
the main tension elements from corrosion have been tried. Today the corrosion protection system
usually consists of at least two barriers: the internal barrier immediately adjacent to the main
tension element and the external barrier or covering which is exposed to the outside
environment. The current corrosion protection methods for stay cables are well summarized by
Podolny [6].
4.1
Coating Wires
As far as the internal barrier is concerned, the wires to be used in strands for bridge cables are
zinc-galvanized or non-galvanized. Mainly in the North America, epoxy coating of individual
wires or seven-wire strands has been widely used instead of galvanization. As mentioned later,
the combination with zinc-galvanization and cement mortar grouting is to be cautious, while
disturbing observations are reported in the fatigue test with non-intersticially epoxy coated
strands [7]. In addition, the manufacturing and handling of epoxy coating on stay cables should
be executed very carefully [3].
4.2
Wire Ropes
The wires for spiral ropes are zinc-galvanized and the voids are filled with a sealing compound
such as metalcoat which is a suspension of aluminum flakes incorporated into a hydrocarbon
resin carrier suitably diluted with a solvent for ease of application.
LCR in the very early cable-stayed bridges in Germany were manufactured of non-galvanized
wires in fear of hydrogen brittlement, and the voids were filled with red lead during rope closing.
After application of all permanent loads, the rope surface was thoroughly cleaned and two basic
coats of red lead as well as two finishing coats such as iron glimmer were applied.
On the other hand, LCR used on Japanese bridges were manufactured of galvanized wires,
applying a minimum amount of lubricating oil during rope closing to avoid any concern about
future stains of the surface. The outer surfaces were usually painted after the dead load had been
fully applied. In the Onomichi Bridge, for example, the cable repainting has been executed
almost every five years, and the cables seem now to be found sound after more than thirty years
of use.
The recent German practice of corrosion protection for LCR has been modified. In the new
practice, wires are to be galvanized with zinc, the inner voids are filled by polyurethane with
zinc dust or linseed oil with red lead, and outer surface of the rope is to be coated by
polyurethane. Metalcoat mentioned above is also sometimes applied on LCR as the second
barrier during its fabrication.
4.3
Covering or Sheathing
Covering the strand or cable as the external barrier has been common to other stay cables than
helical wire ropes. One of the methods was to wrap a foamed polyethylene tape with glass fiber
reinforced plastic covering over the PWS. In the early 1970s, this was executed by hand-lay-up
method on site in Japan [8]. Several years later, prefabricated-segment method was newly
developed to improve the workability of hand-lay-up method. This is to fabricate FRP segments
in the shop, and just to connect them on site to form the complete covering. But an installation of
catwalk was indispensable for the erection. In addition some expansion joints had to be placed
on the covering with certain intervals to absorb the difference of the expansion and contraction
between the cable and covering. In some cable-stayed bridges after the age of 20 years or so,
such damages as small cracks on the covering and a little deterioration on the expansion joints
were found. Then repairing works were needed for these plastic covering.
The covering by a metal tube made of steel, stainless steel or aluminum alloy has been often
applied to stay cables of PC cable-stayed bridges. Stainless steel and aluminum alloy offer the
advantage of long lasting protection, while steel pipes have to be further coated: for example, the
three-coat system consisting of the prime coat of self-curing inorganic zinc coating at shop and
epoxy intermediate coat as well as polyurethane finish coat at site. Anyhow the installation of
metal pipes should be done at the erection site and their stiffness may cause some difficulty in
handling during erection when a cable is long. In case of ASP cable, two semi-circular covers of
aluminum alloy are assembled into a complete tube around the bundle of parallel seven-wire
strands at the bridge site.
The use of a fiber reinforced plastic or polyethylene tube was initiated rather earlier in 1960s.
HDPE material which is most widely used is selected to resist weathering, high pressure, high
temperature and external injury. At the same time, 2-3% carbon is mixed to protect sheath from
ultra-violet rays. Now HDPE tubes are most popular for both parallel wire cables and parallel
strand cables, and either shop-fabricated or site-fabricated. When heavier corrosion protection is
needed, double layer PE tube is used to prevent the cracks on the outer surface from reaching to
the main tension elements.
In case of New PWS and seven-wire strands mentioned in the previous chapter, the covering is
completely shop-fabricated by a directly extruded HDPE sheath after coating wires with
corrosion protection compound. In the former, any further work for corrosion protection is not
required at the site. Even if the PE covering is injured, the durability of the cable can be kept for
some time because wires are galvanized. The repairing of injured PE envelope can be easily
done. The inspection of the cable can be also possible by tearing off a part of the PE covering
and by watching the cable from the torn PE window.
Although the original color of the PE covering is black due to the mixed carbon, cable coloring
techniques have been developed. One is to extrude a colored thin fluoro-polymer on the black PE
layer, and another method is a paint coating system that consists of an application of primer
made from adhesive components for PE envelope and for the finish coat, and a baking of the
primer with the far infrared ray. The finish coat is usually done by fluoro-olefin paint. The light
color is preferred not only for good looking but also for reducing the temperature effect.
Supplementary wrapping with colored Tedlar tapes is an alternative.
4.4
Blocking Compound
Blocking compound is the corrosion inhibiting or water repelling material used to fill the voids
between the tension elements and the outer sheathing. Cement grout has been most popular
blocking compound for its alkaline properties providing an active corrosion protection to the
steel wires. Cement mortar grout is injected after the stay cables are erected on the bridge when
the cable is under full dead load stress to suppress the formation of cracks in the grout, because
the presence of cracks may be associated with the potential for fretting corrosion of steel wires.
However, cracks may still occur due to shrinkage of cement mortar and stress repetition under
cyclic live loading, and actually have been observed on some cable-stayed bridges.
Another problem of cement grout combined with galvanized wires is a fear of hydrogen
brittlement caused by reaction of zinc and cement milk. In order to avoid it, the non-galvanized
wires or the galvanized wires coated with polyester to isolate zinc from cement milk have been
used in this case. Further, another measure is to substitute normal Portland cement by polymer
cement. The advantages of this material are that it is far more ductile, does not shrink after
grouting nor bleed during placing, and does not require special technique and equipment, and
that it can be used in combination with galvanized wire without a fear of chemical reaction
between zinc layer and cement. On the other hand, its disadvantages are relatively high material
cost and temperature-dependent viscosity and hardening. Cement grout plasticized with
polyurethane was used on some other bridges.
Alternatives to cementitious grout have been sought and used for stay cables. Synthetic resin
material based on polybutadiene was used on the Iwaguro-jima and Hitsuishi-jima bridges of the
Honshu Shikoku linking projects in Japan about ten years ago. This two-component material,
one liquid consisting of polybutadiene polyurethane polyol resin and the other of an isocyanate
hardener, has very low viscosity during pouring, is very flexible after hardening, and has such
low density as about half of that of cement grout. But the material and execution costs were high
and it is highly temperature dependent and flammable. Epoxy resin is specified in the ASTM
provisions as a filling material into the interstices of epoxy-coated seven-wire prestressing
strands.
Other blocking compounds are grease used on PWS or prestressing strands and wax for parallel
strand cables. The petroleum is injected in a liquid state at temperature of 85-105oC and
afterwards it solidifies upon cooling. However, it shrinks during cooling process and cracks may
develop. A soft petroleum base wax that can be applied at ambient temperature on the monostrand system seems promising [7]. It has a melting point over 260oC and displaces any moisture
on the surface of the steel. In any case, grease and wax are to be used in combination with other
corrosion protection measures. It is also noted that the non-grout type PWS which is completely
fabricated in shop is now available.
5.
Vibration Control
Wind-induced vibrations of stay cables have attracted concern of bridge engineers with
increasing span length of cable-stayed bridges [9] and in particular since the introduction of
multi-stay system with thin cables covered by polyethylene-sheath having smooth surface. The
types of the wind-induced vibrations are full of variety and therefore the preventive measures
that are classified into mechanical means and aerodynamic means are also manifold.
5.1
Types of Cable Vibrations under Wind
Vortex Excitation is a cross-wind vibration caused by periodically shedding vortices in the wake
behind the body when their frequency coincides with the natural frequencies of the body. Since
this is the resonance phenomenon in forced vibration, its amplitude is limited and inversely
proportional to structural damping inherent to the body, but may occur at rather low wind
velocity. Higher modes of the vibration can be also observed.
Galloping is a self-excited oscillation in cross-wind direction, being caused by the negative
damping effect of aerodynamic force. Although circular cables can not gallop because of crosssectional symmetry under the wind acting perpendicularly to its member axis, small deviations
from a perfectly circular shape may imply galloping instability. For example, transmission line
cables with iced snow attached to the lower surface have been often suffered from this galloping
.
Wake Galloping is the dynamic phenomenon caused by fluid-elastic interaction between
neighboring cables. When two stay cables are closely placed, leeward cable may be largely
excited due to the presence of windward cable. Like vortex excitation, wake galloping may occur
at rather low wind speed and goes down with increasing wind speed.
Rain(-Wind-Induced) Vibration was first reported on the Meiko-West Bridge built in 1985 in
Japan [10]. Since then many cable-stayed bridges have suffered from this phenomenon. Its cause
is the water rivulets on the upper and/or lower side of cable surface. The vibration may occur at
rather low wind speed and continue in the wide wind speed range. Its direction has been mainly
cross-wind but in-wind vibration can occur according to the positions of water rivulets, and the
phenomenon is largely independent of the natural frequency [11].
Buffeting is a random vibration caused by turbulence in the on-coming air flow. Its amplitude is
limited and suppressed by increase of structural damping.
Attention should also be given to the possibility of cable vibrations that are caused by the
dynamic forces acting on the girder and/or pylon. From the past experiences on stay cable
behavior, however, the phenomena on which special care shall be taken are vortex excitation,
wake galloping and rain vibration. Tentative or approximate stability criteria for typical vibratory
phenomena have been proposed [4]. At this time, the effect of local terrain features on these
wind-induced vibrations should be taken into account. In case of the Chichibu Park Bridge in
Japan, for example, large-amplitude cable vibrations under moderate wind in only one span of a
symmetrical two-span cable-stayed bridge have been observed because the bridge crosses a
curved gorge and the terrain on both sides is strongly non-symmetrical.
5.2
Countermeasures
In general the following preventive measures are considered for wind-induced vibrations of
cables when the responses are not acceptable:
a) increase of structural stiffness and natural frequency
b) increase of structural damping
c) modification of the cable surface.
a) and b) are categorized in the mechanical or structural means, while c) is the aerodynamic
means to weaken the exciting mechanisms by disturbing or reducing wind-induced dynamic
force acting on the cable. However, because the exciting mechanisms of different vibration types
differ, the countermeasures shall fit for the phenomenon concerned.
Occurrence of wake galloping depends on the spacing of neighboring cables. Very small spacing
or quite wide spacing more than six times the cable diameter can remarkably moderate the
response. If these conditions can not be satisfied for other design reasons or when undesirable
wake galloping is observed after erection, the cable vibration can be suppressed by connecting
the both cables by a few spacers or small mechanical dampers (e.g. Yokohama Bay Bridge).
One of the common countermeasures that has often been adopted is to connect the stay cables
with secondary cables which may terminate at a cable or at the deck (tie-down cables). Even
with a few and small stabilizing ropes, stay movements can be restrained. The natural
frequencies of each stay, and thus, the resistance to dynamic excitation can be raised by
shortening the effective free length of the main stay cables by means of transverse connections.
Such elements may be also sources of additional damping. But even if the size of stabilizing
ropes is significantly smaller than that of the primary cables, they may affect the appearance of
the structure to some extent. Furthermore, the rupture of these interconnecting ropes or the
fatigue failure of the connection fittings have been reported in several bridges. Viscoelastic
bushings can reduce fatigue and provide additional damping. In case of the Normandy Bridge,
the stabilizing effect was expected to further increase by installing special clamps with damping
devices in the joint of the secondary ropes and the primary stays.
Structural damping of long stay cables is often so low as the logarithmic damping of the order of
10-3. The increase of the damping to some extent is effective in suppressing the amplitude of
buffeting, vortex excitation and rain vibration, and in raising the critical wind speed for the onset
of galloping. It is first recommended to place such damping material as neoprene ring or highdamping rubber between cable and steel exit pipe at pylon and deck anchorage. Use of damping
material results in additional benefit of reduced bending moment in the cable. Further additional
damping, if necessary, can be provided by mechanical damping devices. Very simple and small
tuned mass dampers (TMD) represented by the classical Stockbridge damper that has been used
on transmission power lines were applied to stay cables or diagonal hangers of some European
cable-supported bridges. But it is not so popular for aesthetic reason.
When rather high additional damping is required, the most prevalent is to install a dash-pot type
viscous damper between the stay and the bridge deck. In the Brotonne Bridge, after experiencing
severe oscillations of the stay cable, shock absorbers similar to those used in automobiles were
attached, while in case of the Sunshine Skyway Bridge later built in the US, hydraulic (oil)
dampers were installed beforehand and it was possible to position them inside the box girder.
The dampers shall not spoil the appearance of the bridge. The height of the oil dampers was
designed not to exceed that of guard rails in the Aratsu Bridge in Japan (Photo 1). The
mechanical dampers utilizing shear-viscous material can be more compact (Photo 2). The
position of the above-mentioned dampers influences the mode shape and the damping effect of
the cables attached by the damper. It is not easy to make compromise between the requirement to
lower the damper position and the efficacy of the damper when the cable is long [9].
Photo 1. Hydraulic damper at the

Aratsu Bridge, Fukuoka
Photo 3. Protuberance on the HDPE

sheath surface at the Higashi Kobe
Bridge
Photo 2 Viscous-shear type damper at the Meiko-Central

Bridge, Nagoya. The elastic seal material is also
inserted under the anchorage cover to reduce bent
of cable.
Photo. 4. Pattern-indented surface of the Tatara

Bridge stay cable. High-damping rubber between
thestay and the exit pipe at the anchorage, which will
be covered later, is also seen.
The aerodynamic countermeasures for the round cables are to modify the cable surface. The idea
of helical fins often used on circular stacks to prevent vortex excitation was applied in the
Normandy Bridge. HDPE duct is composed of two half-elements shaped by small helical filets.
It is reported that these were designed to eliminate also the rain-induced vibration [5]. The first
trial of this kind on stay cables, aiming at reducing the rain vibration, was the axial
protuberances in the form of longitudinal ribs on the HDPE tube surface developed in the
Higashi Kobe Bridge in Japan (Photo 3). Similar idea is later seen in the more simpler HDPE
sheaths with fine grooves, being used on a few other Japanese cable-stayed bridges. Further on
the Tatara Bridge, HDPE sheaths are provided pattern-indented surface with roughness of 1%
applied disorderly in a convex or a concave pattern (Photo 4) [12]. The effect of these surface
modification is linked to influence the behavior of the water rivulets. However, it should be
noted that these means are not necessarily effective in suppressing vortex excitation and that the
drag coefficient may be increased. In case of the cables of the Tatara Bridge, however, the drag
coefficient in the super-critical Reynolds number range could be reduced to 0.6.
6.
Concluding Remarks
The overview of stay cable technology is presented in this paper. The development of cablestayed bridges in the past half century has been very remarkable in both the increase of span
lengths and the number of constructions worldwide for their wide applicability and aesthetic
appeal. Different from suspension bridge cables, however, the cable technology in cable-stayed
bridges is much diverse not only in their types and corrosion protection system but also, although
being unable to mention herein, in the methods of end fitting and anchorage details. The recent
international survey [1] indicates such scattered preference on these items. The vulnerability of
the stay cables to corrosion damage and wind-induced instability is still a keen concern despite
of the continuing evolution of cable technology. Among them, the effectiveness of corrosion
protection systems shall be determined by only time. Therefore, the appropriate methods of nondestructive inspection have to be pursued [13].
Although the choice among available technologies may depend on the features of the structure,
the natural and local conditions of the site, the technology atmosphere of the area, as well as the
judgement of the owners, designers and contractors, taking into account of cost estimation, the
pertinent combination of the available technologies on the above-mentioned items should be
determined by careful considerations for such long lasting structures as bridges. The
practicability of new composite fiber materials shall be also pursued for the future development
of cable-stayed bridges.
References
[1] HAMILTON III, H. R.; BREEN, J. E. International Survey of Current Opinion on Bridge
Stay Cable Systems. IABSE Report 73/2, Symposium on Extending the Life-span of Structures
(San Francisco), August 1995, pp.843-853.
[2] OHASHI, M. Cables for Cable-Stayed Bridges. Cable-Stayed Bridges: Recent
Developments and their Future (ed. M. Ito et al.), Elsevier, 1991, pp.125-149.
[3] GIMSING, N. J. Cable-Supported Bridges - Concept and Design (2nd ed.), Wiley, 1996,
Chapter 2: Cables.
[4] PTI Committee on Cable-Stayed Bridges; Recommendations for Stay Cable Design, Testing
and Installation. (Draft), Post Tensioning Institute, January 1998.
[5] Virlogeux, M. et al. Design of the Normandie Bridge. Proceedings of International
Conference on Cable-Stayed and Suspension Bridges (Deauville), Vol. 1, 1994, pp.605-630.
[6] PODOLNY, Jr. W. Current Corrosion Protection Methods for Cable Stays. op. cit. [1],
pp.855-860
[7] LAPSLEY, R. D.; GANZ, H. R. Experience, Developments and Trends for Improved
Durability of Stay Cables. op. cit. [1], pp.879-884.
[8] ITO, M.; TADA, K.; KITAGAWA, M. Cable-Corrosion-Protection Systems for CableSupported Bridges in Japan. Op. cit. [1], pp.873-878.
[9] YAMADA, H. Control of Wind-Induced Cable Vibrations from a Viewpoint of the Wind
Resistant Design of Cable-Stayed Bridges. Proceedings of the International Seminar on Cable
Dynamics (Tokyo), JAWE, 1997, pp.129-138.
[10] HIKAMI, Y.; SHIRAISHI, N. Rain-Wind-Induced Vibrations of Cables in Cable-Stayed
Bridges. Journal of Wind Engineering and Industrial Aerodynamics, No. 29, 1988, pp.409-418.
[11] VERWIEBE, C. Exciting Mechanisms of Rain-Wind-Induced Vibrations, Structural
Engineering International (IABSE), Vol. 8, No. 2, May 1998, pp.112-117.
[12] MIYATA, T.; YAMADA, H.; HOJO, T. Aerodynamic Responses of PE Stay Cables with
Pattern-Indented Surface. op. cit. [5], Vol. 2, pp.515-522.
[13] ZAHN, F. A.; BITTERLI, B. Developments in Non-destructive Stay Cable Inspection
Methods. op. cit. [1], pp.861-866.
Design of Girder and Cables for Train Loads

Niels BITSCH
MSc Civil Engineer
COWI
Lyngby, Denmark
Mr. Niels Bitsch, born
1958, received his civil
engineering degree from
the Technical University
of Denmark, year 1984.
Has in the past 10 years
been working with the
bridges of the two major
links in Denmark, the
Great Belt Link and the
resund Link
1.
Lars HAUGE
MSc Civil Engineer
COWI
Lyngby, Denmark
Mr. Lars Hauge born
the Technical University
of Denmark, year 1986.
Has through his work
with some of the world`s
largest bridges obtained
considerable expertise in
design of cable stayed
and suspension bridges.
The Cable Stayed Bridge of the resund Link
The High Bridge of the resund Link is a Cable Stayed Bridge with a Main Span of 490 m and
two side spans of 160m and141m respectively. The Pylons, with two single towers each, are
constructed in reinforced concrete and the Bridge girder is a two level composite girder. The two
level composite girder comprises a main carrying steel truss and an upper roadway deck slab in
concrete.
The Cable Stays are arranged as a harp system with 10 Cable Stays in each cable fan. The Cable
Stay inclination with horizontal is 30o , and distance between the anchorages on the Girder are 20
m. For anchorage of the Cable Stays in the Pylon, a cast-in steel box has been designed. See
elevation and cross section of the bridge in Figure 1-1 respectively Figure 1-2.
Figure 1-1 :
resund Link, Cable Stayed Bridge. Side view.
Figure 1-2 :
resund Link, Cable Stayed Bridge. Cross section
The structural design is based on the Eurocodes with an associated Project Application Document
and a Design Requirement document. The Design Requirement document supplement and takes
precedence over the other two documents, with specific loads and other requirements covering
topics which are not considered in the Eurocodes.
During detailed design computer models was established in order to perform the general
documentation of the Bridge, but also to perform the rather complex analyses related to the Train
loads :
Comfort analyses
Dynamic Actions
Fatigue analyses
Cable Stay replacement
Deflections
Other complex effects with considerable design impact, but without connection to the Train loads,
such as shrinkage and creep effects, shear lag and cable stay rupture, was also analysed.
Focus is in this paper put on Comfort analyses, Dynamic Actions and Fatigue analyses, as these
subjects in particular, are difficult analyse in detail, on complicated bridge structures. The analyses
was however performed with success on the cable stayed bridge of the resund Link, using up to
date designs knowledge, extensive computer capacity and specialist software.
2.
Comfort analyses
2.1
Requirements
A number of requirements to the accelerations and deformations of the Bridge Girder has been
specified for the Bridge in order to get an adequate level of safety and comfort during crossing of
the Bridge by train. With span lengths exceeding 90 m it is not possible to use conventional design
rules based on Girder deflections alone in order to verify an adequate level of passenger comfort.
Detailed analyses of the vertical accelerations within a passenger coach need to be performed.
The analyses only comprised the accelerations induced in the trains due to the flexibility of the
Bridge. The effect of irregularities of track and wheels was not taken into account.
2.2
Acceleration limitations
Figure 2-1 shows the requirements to the vertical acceleration limits within the passenger coach,
in term of root-mean-square (rms) accelerations. For vertical accelerations with duration less than
0.5 seconds, the acceleration (peak acceleration) is not allowed to exceed 2 m/s2. The
accelerations of the train is frequency weighted according to ISO2631, which applies to simulate
the human perception of vertical vibrations, e.g. vibrations in the longitudinal direction of the
human body.
Figure 2-1 : Acceleration limits as function of duration.

2.3
Dynamic Train Models

Four types of passenger trains was specified for the dynamic analyses. In
the analyses the loads of the passenger trains was applied on one track,
with or without rail load UIC 71 on the other track, whichever created the
most unfavourable effect for the situation analysed.
The passenger trains consist of a Loco and Passenger Coaches. The
boogie support system of the Passenger Coaches was modelled as an
equivalent single degree of freedom (SDOF) systems (Figure 2-2), used
to simulate the accelerations in the passenger coach.
Figure 2-2 :
SDOF model
The SDOF system was described by a mass M, a spring Ke and a dashpot Ce. Ke and Ce are
equivalent values calculated from detailed knowledge of primary and secondary spring and
dashpot systems of the coach boogies.
Table 2-1 summarises the parameters of the four SDOF systems and the associated natural
frequencies, fe, and relative damping ratios, e.
Passenger Coach
M
(kg)
Ke
(N/m)
Ce
(kg/s)
fe
(Hz)
T1
T2
T3
T4
17000
16800
16200
16100
390000
575000
720000
805000
52000
32000
37000
22000
0.76
0.93
1.06
1.13
0.32
0.16
0.17
0.10
Table 2-1 :
Parameters for simplified SDOF model
The general loads from the passenger trains was modelled as a set of point loads. The point loads
gave a quasi-static deflection corresponding to the deflection of the entire train.
2.3
Bridge Model
The bridge structure was modelled with COWI's in-house developed finite element program,
IBDAS.
The finite element model, Figure 2-3, comprised all main structural elements. The concrete
roadway deck was modelled using shell elements, the cable stays were modelled as cable
elements, and the pylons, piers, railway deck, chords, outriggers and diagonals were modelled
with beam elements.
Figure 2-3 : resund, Cable Stayed Bridge. Finite element model (IBDAS)
2.4
Calculations
The train induced accelerations was determined by simulation of the train response in the time
domain for the four types of passenger trains crossing the bridge. The shape of the deflected
bridge girder, zg , during the train crossing, comprised contributions from :
Vertical alignment of the bridge
Deflection due to crossing of the passenger trains, taking into account dynamic effects of the
moving loads.
The static deflection of the bridge due to service (SLS ) loads, with a dynamic amplification
factor of 1.2. (primarily UIC 71 load on other track and road traffic loads).
The acceleration response in the coach travelling along the deflection shape, zg , was found by
numerical integration of the equation :
Mz + ce ( z z g ) + k e ( z z g ) = 0
where z g = z g (V / t ) and z g = z g (V / t ) with V being the speed of the train and t the time. The
accelerations was derived by numerical differentiation of the displacement response, z.
Finally the accelerations was frequency weighted according to ISO2631 ref. sect.2.1 and compared
with the acceleration limits.
The applied time step of the above integration was 0.009 seconds equivalent to spatial steps of 0.5
meters along the bridge axis (200 km/h).
2.5
Results
For each of the four passenger trains, three train load configurations was analysed :
1. Maximum main Span deflection
2. Maximum rotation of the Girder at the Expansion Joints
3. Maximum local deflections of the Railway Deck
Figure 2-4 illustrate the acceleration response along the bridge for train type T4 crossing the
bridge, considering maximum mid span deflection.
Figure 2-4 :
Acceleration results for T4 with maximum mid span deflection.
Figure 2-5 compares the results with the accelerations limits specified in sect. 2.2 . For the load
configuration maximum rotation of the girder at the expansion joint, the figure shows that the
requirements to the acceleration is satisfied.
Figure 2-5 :
Acceleration results for T4 compared with the design requirements. All duration's.
3.
Dynamic Actions
3.1
Introduction
As part of the ULS In-service and ULS Fatigue verification of the Bridge, analysis of the dynamic
train-bridge interaction were carried out. The analyses were used to determine dynamic load
factors to be applied to global static rail load effects.
The following global rail load effects were considered :
longitudinal load effects in the principal structural elements of the Bridge Girder
load effects in Pylons and Piers
load effects in the Cable Stays
Dynamic analyses were carried out for nine different train types, for which moving point load
models were provided. The influence of the dynamic behaviour of the train was not included in
the analyses.
The analyses were carried out using a 3D-computer model, similar to the model used for static
load effect calculations.
Regarding the dynamic enhancement of local train load effects on the Railway Deck, conventional
theory was used.
3.2
Train types
Nine different train types was analysed. UIC Train type 1, used for the ULS-verification of the
bridge, and Passenger Train types 1-3, Freight Train types 1-3 and Heavy Rail Traffic Train types
1 and 2, used for the Fatigue verification. Figure 3-1 shows an example of the moving point load
model for Passenger Train type 1.
Figure 3-1 : Passenger Train type 1, point load model

The design speed of the UIC train is 120 km/h and for the "fatigue trains" 80 to 200 km/h.
As the length of the UIC load is dependent on the load effect and the section considered, 28
different load models with train length between 75 m and 750 m were defined in order to create
the most unfavourable effects. The length of the "fatigue trains" lies from 97.4 m to 519.2 m.
3.2
Investigated bridge parts
A representative number of sections in the Cable Stayed Bridge were examined ref. Figure 3-2 .
Only dynamic effects of the Bridge Girder and the Cable Stays is described in the following.
In the Bridge Girder dynamic load factors were determined for 3 different load effects, normal
force, bending moment and vertical shear force, in 18 different sections along the bridge axis.
Dynamic load factors determined for normal forces and bending moments, applied to normal
forces and bending moments of the upper chords, lower chords, railway deck and concrete
roadway deck.
Dynamic load factor for vertical shear applied to normal forces and in-plane moments of the
diagonals.
The dynamic load factor of the Cable Stays were determined for normal forces of the cables.
Figure 3-2 :
3.3
Cross sections for which a dynamic load factor was determined.
Dynamic load factor
A dynamic analysis were carried out in order to determine the "dynamic" max/min values of the
examined load effects for each of the selected cross sections described above.
The result of the dynamic analysis is a combined static and dynamic value for the considered load
effect. In order to get the dynamic contribution to the load effect, the static value of the load effect
shall be obtained from a separate static analyses and then be subtracted from the results of the
dynamic analysis.
The eigen-modes and the corresponding eigen-frequencies of the bridge structure was specified as
input for the dynamic analyses. The dynamic analyses were carried out using time steps of 0.1 sec.
A comparison of the eigen-frequencies for the vertical modes, calculated using the simplified
model used for the dynamic analyses, with the corresponding eigen-frequencies calculated with
the full truss model showed very good agreement.
The eigen-frequencies for the first vertical mode calculated with the simple model was equal to
0.363 Hz and the frequency calculated with the full truss model was equal to 0.351 Hz.
The structural damping (Raleigh damping) used in the dynamic analyses was equal to 1%, at a
frequency corresponding to the first vertical mode, f1 = 0.360 Hz, which was considered to be
conservative.
Based on the results of the static and dynamic analyses, the dynamic factor, , was determined as:
= Dynamic result / Static result
3.4
Dynamic load factor results
A total of approximately 13000 dynamic load factors were determined and only a summary of the
calculated dynamic load factors for the Bridge Girder and Cable Stays is presented here.
In Figure 3-3 detailed information regarding the dynamic load factors for various load effects is
informed.
Figure 3-3 : UIC rain type 1.Global dynamic load factors for various load effects.
Table 3-2 summarises the dynamic load factors of the UIC train type 1 and the "fatigue trains",
used in the design. Regarding position and name of the Cable Stays, "S" = side span position and
"M" = Main span position. The cable stay numbers starts with 1 at the Pylon.
UIC train
type 1
Passenger
train I
Passenger
train II
Passenger
train III
Freight
train I - III
Heavy Rail
train I and II
Upper chords
Lower chords
Railway deck
Concrete deck
1.02-1.05
1.10-1.40
1.10-1.15
1.06
1.04
1.02
Diagonals
1.03-1.02
1.10-1.20
1.01-1.08
1.01-1.06
1.02
1.02
Cable Stays
(Tension)
1S, 1M
2S, 2M
other cable stays
1.06
1.02
1.01
2.00
1.20
1.06
1.07
1.05
1.03
1.20
1.05
1.05
1.04
1.04
1.04
1.02
1.02
1.02
Bridge Girder
Table 3-2 : Summary of dynamic load factors.
4.
Fatigue analyses
4.1
Introduction
Assessment of the fatigue capacity of a steel bridge girder subjected to train loads, is a very
important part of the design work. Having a cable stayed bridge makes the assessment even more
complicated.
On the resund cable stayed bridge the fatigue capacity of all bridge elements were analysed, steel
truss, cable stays and not to forget the concrete roadway deck with shear stud connection to the
steel truss. The steel truss with the orthotropic steel deck for support of the two railway tracks
gave many challenges.
During the design of the Cable Stayed Bridge of the resund Link, extensive computer
calculations was made in order to document the required fatigue capacity. Further "best practice"
was used with respect to design against fatigue problems, without making the steel details
unacceptable from a production point of view.
An example of "best practice" in the design, is the "clown mouth" cut out in the transverse
bulkheads, where the troughs of the orthotropic steel panel crosses, see Figure 4-1 .
Figure 4-1 : "Clown mouth" cut out in transverse bulkheads.

4.2
Fatigue loads
Eight types of "fatigue" trains was specified for assessment of the fatigue capacity. Three
passenger trains, three freight trains and two heavy rail trains, each of them provided with
following information's :
Train nos. per track per day
Service hours per day
Max. speed
Train configuration. Axle loads and axle positions.
Max. and mean length.
Max and mean total load.
Based on the information's it was possible to find the number of train crossings during the 100
year life time of the bridge, and the numbers of train crossings with simultaneously loading of
both tracks.
For every girder detail investigated the dynamic effect of the train load was taken into account.
For information's about the global train load effects reference is made to chapter 3.
The dynamic effect of the local train loads is not regarded in this paper, but for information the
dynamic load factor, , was found to be in the range of 1.13 to 1.58 . The local dynamic factors
was only used on local load effects on the railway deck.
4.3
Fatigue stress calculations
The stress calculations was based on results from two computer models, the global model
described in sect 2.3, giving the global load effects and a detailed semilocal FEM-model giving
the local load effects in the railway deck.
The stress spectres for all the details investigated was in general determined by computing an
influence line for a specific effect and then traversing the load across the influence line.
The Figures 4-2 and 4-3 gives the global and local stress spectre for one of the trough splices in
the deck panels of the railway deck.
Figure 4-2 : Freight train 1 stress spectres for global load effects in the trough splice of the
railway deck.
Figure 4-3 : Freight train 1 stress spectre for local load effects in the trough splice of the railway
deck.
4.4
Fatigue assessment
Foe each of the load events (stress spectres) a stress range spectrum was calculated by means of
the Rainflow counting method, using a special developed programme.
The fatigue assessment was based on the calculated nominal stress ranges and a Cumulative
Damage Assessment based on a fatigue strength curve with double slope constants (m=3 and
m=5), and a cut-off limit at N=100 million cycles in accordance with EUROCODE 3.
Construction details was classified according to Swedish codes, giving detail categories.
The damage for each loading event was calculated and based on the occurrence of the loading
events and the design working life of the bridge, 100 years, a total damage was calculated.
5.
Conclusion
The design for the Cable stayed bridge for the resund Link has proved to be both suitable and
adequate for safe operation of High Speed Railway. This has been verified by extensive and
advanced analyses.
Finally it is important to conclude, that it has not been necessary with any special precautions in
order to fulfil the comfort requirements.
An Innovative Technique for Fitting Trackwork Alignments Through the

Railway Envelope of a Cable-stayed Bridge
Robin SHAM
Technical Director
Maunsell Ltd
Beckenham, UK
Robin Sham, born 1954, received his

BSc in 1978 (Birmingham) and PhD
(Structural Engineering) in 1989
(Imperial College). He is Technical
Director responsible for Bridges &
Special Structures and was
Designers Project Manager for the
successful Kap Shui Mun Bridge &
Ma Wan Viaduct Project in Hong
Kong, opened in 1997.
Summary
The paper describes how an innovative technique was successfully applied to fit trackwork
alignments through the railway envelope of a cable-stayed bridge against a background of
difficulties.
1.
Introduction
The Kap Shui Mun Bridge and Ma Wan Viaduct, which symbolise the
vision and energy of Hong Kong, were opened to traffic in May 1997.
The structures form a crucial part of the Lantau Link, the rail-road
infrastructure network which connects the new Hong Kong Airport at
Chek Lap Kok to Kowloon and Hong Kong Island. The two-level
structures carry the Expressway at the top deck and the Airport Railway
in the central region of the lower deck. Emergency carriageways for use
under typhoon conditions are provided on either side of the railway.
The Kap Shui Mun Bridge is the worlds longest cable-stayed bridge
which carries combined rail and road traffic. The design harnesses and
amalgamates technologies in different disciplines, such as long-span sea Fig 1 - Kap Shui Mun
crossings, tunnelling and railway works. The bridge is a product of Hong Bridge & Ma Wan
Kongs first Design-and-Construct Contract for a major sea crossing. In Viaduct
an extremely tight programme of 4 years many innovative ideas were
developed to integrate design with construction.
The design concept targets at
making possible what
effectively is a tunnel in the
air, to be constructed at an
extraordinary rate, on time and
within budget, with heavy
liquidated damages imposed
on a number of critical
Fig 2 - Cross-section of 2-level Cable-stayed Bridge
milestone dates. Table 1 lists
the six critical key dates. While liquidated damages were imposed on all key dates, failure to
meet Key Dates 3 and 4 would incur very substantial penalties. Key Date 3 effectively called for
completion of the bridges to permit the passage of trains.
Key date no
Key Date 1
Key Date 2
Key Date 3
Key Date 4
Date
10.12.1995
7.5.1996
11.8.1996
11.8.1997
Critical stage
Lantau approaches - release of works area to an adjacent contractor
Lantau substation for China Light and Power
Release of the railway envelope to the airport railway operator
Access to Lantau substation and the utilities envelope to other
contractors of the employer
Key Date 5 24.11.1996 Access across the bridge and viaduct to authorised contractors of the
employer
Key Date 6 18.5.1997 Substantial completion
Table 1. Key dates for the fast track construction programme
This paper describes how an innovative technique was
successfully applied to fit trackwork alignments through
the railway envelope, against a background of a tight
construction programme and technical difficulties.
2.
The Cable-stayed Bridge
The cable-stayed structure has a main span of 430m and

a total length of 750m. Structural efficiency is enhanced
Fig 3 - The Railway Envelope in the
by double steel-concrete composite action in the main
Main Span
span in that both the top and bottom flanges of the steel
superstructure are formed in concrete. For aerodynamic stability however, perforations are
introduced into the central region of the flanges. In the transverse direction vierendeel action is
achieved by cross frames; with alternate frames being suspended from the stay cables. Stiffened
steel plates form the inclined outer webs. Longitudinal bracings which coincide with the inner
columns of the vierendeel frames, distribute live loads to other cross frames and reduce trafficinduced vibrations. Further efficiency is obtained by bolting the diagonals subsequent to the
application of dead loads, thereby relieving them of dead load stresses from the main girder.
The design of the main span enabled erection by free cantilevering from both towers. It also
allowed steelwork fabrication in China and segment assembly in the Lantau casting yard prior to
float-out. The simplicity of the design ensured an erection cycle of 2 weeks for each segment
weighing 500 tonnes.
The side spans consist of twin box girders, designed for fast erection by incremental launching.
The H-shaped towers were designed and detailed for fast construction by a self-climbing
formwork system; the architecture being one that combines structural simplicity with
aerodynamic efficiency.
Fig 4 - Free Cantilevering from Both Towers
Fig 5 - Main Span Closure
3.
Trackwork
The trackwork for the airport railway is contained in the central region of the lower deck of these
bridges. The trackform design consists of precast, post-tensioned concrete trackslabs mounted on
resilient bearings which are installed on transverse beams. The trackslabs are also restrained
laterally through resilient bearings fixed to concrete corbels which are positioned on either side
of the trackslab and cast into the bridge superstructure. The design constitutes a non-ballasted
floating trackslab system which isolates the trackform from the main bridge structure thereby
minimising the generation of noise and vibration.
The trackform design enabled many construction activities to be implemented concurrently.
Trackslabs were precast on site while the concrete corbels were constructed in-situ on the bridge
superstructures. The trackslabs were then transported to strategic locations on the ground from
where they were lifted by cranes up to deck level. The slabs were craned into the railway
envelope in the central region of the lower deck. A travelling gantry installed the trackslabs
progressively along the railway envelope. The post-tensioning of the trackslabs trailed closely
behind the construction head.
4.
Fitting Trackwork Alignments
In order to meet Key Date 3, trackwork construction had to be concurrent with bridge
construction. However, trackwork construction within the partially completed bridge
superstructures would only be possible if design techniques could be developed to control the
setting out of the trackwork under the most unusual conditions which existed.
With full co-operation from the contractor, Maunsells Wriggle technique was applied to fitting
the trackwork through the railway envelope. The Wriggle technique originates from tunnel
engineering and involves the determination of track alignments in three dimensions to fit through
the surveyed tunnel. The technique was successfully used to fit a rail profile (vertical) and
alignment (horizontal) through the as-constructed railway envelope in the lower deck. The track
profile must satisfy the railway design criteria as given in Table 2. The alignment had to take
account of the as-constructed shape of the railway furniture and emergency exit walkways. In all
cases the minimum structure gauge must be maintained.
Maximum Design Speed (km/h)
Maximum Horizontal Radius (m)
Maximum Applied Cant (mm)
Maximum Cant Deficiency (mm) (on plain line c.w.r.)
Minimum Cant Gradient in Transition
Maximum Cant Gradient in Transition
Rate of Change of Cant or Cant Deficiency (mm/sec @ 135km/h)
Minimum Vertical Radius (m) (Crest Curve)
Minimum Vertical Radius (m) (Sag Curve)
Maximum Gradient
Table 2 Railway Design Criteria
140
1250
150
75
1 in 1000
1 in 440
37.5 (Desirable) / 55 (Max)
5140
7260
3%
Fig 6 - Chronology of the Trackwork Construction

Unlike a tunnel, a long-span crossing is subject to transient as well as long-term movement. In
particular the cable-stayed main span is susceptible to considerable movement between different
survey operations carried out at different times of day. Due to the very tight programme of work
the survey of the approach spans had to be carried out when temporary propping and falsework
were still in place and the main span closure was yet to be completed. Similarly, as construction
of Ma Wan Viaduct progressed, the Wriggle exercise had to produce trackwork setting out data
for the existing spans without the benefit of any survey results on spans which were yet to be
built. The deflection predictions, on which the Wriggle exercise partly relied, were incrementally
calibrated and adjusted, when it became possible to survey newly constructed spans.
The objective of the Wriggle exercise was to produce a smooth track profile and alignment that
provided the necessary clearances at pinch points and maintained minimum curvature
requirements. The output from this exercise had to be supplied to the trackwork subcontractor at
a staggering rate and in a form that was simple, accurate, and practical to use for setting out.
5.
Overall Methodology
The methodology for fitting a trackwork profile (vertical) through the railway envelope in the
cable-stayed bridge is described as follows. Similar techniques were used for the approach
viaduct.
Step 1
Common reference survey station for the structures was to be established
Step 2
Level survey was to be carried out through the full length of the bridge structure
Step 3
The railway envelope was to be surveyed, determining levels at the points depicted in
Fig 7
Step 4
In acquiring the survey data in Steps 2 & 3, bridge temperatures at time of survey were
to be recorded
Step 5
Acquired data would be analysed by the Wriggle technique, to determine the rail
profiles achievable, allowing for construction tolerances in the trackform
Step 6
The best engineering estimate of the main span deflections between time of survey and
Key Day 6 would be determined, to account for
the prevailing state of construction and cable stressing
temporary loads, including plant and equipment on the bridge
superimposed dead loads to be applied by Key Date 3 & Key Date 6
time-dependent effects
Step 7
Lower and upper bound curves corresponding to states at Key Date 3, Key Date 6 and
time infinity would be determined, allowing for variability in the parameters described
in Step 6, and the effects of live loads and temperature changes
Step 8
Profiles determined in Step 5 would be adjusted by values derived in Step 6, to allow

for expected future structure movements between the time of survey to Key Date 6
Step 9
A new profile (the Desired Rail Profile) would be determined to

maintain the depth of trackform to within specified tolerances
satisfy the criteria for vertical clearance
comply with the railway design criteria in terms of gradient, curvature and other
considerations, at the limits of each envelope at Key Date 3, Key Date 6 and time
infinity
Step 10 Data for the Desired Rail Profile would be used for setting out purposes on site.
Integrated with the re-profiling, the methodology for fitting a horizontal trackwork alignment
involved establishing the centreline of the bridge within the railway envelope and verifying the
potential clearances. In order to ensure compliance with the clearance criteria, a design for the
emergency exit walkways was specially developed. It provided much flexibility in the setting out
and was well-engineered to accommodate construction tolerance of the elements in the railway
envelope.
6.
Assessment of Structure Movements
The determination of the Desired Rail Profile took into consideration all expected structure
movements which would occur subsequent to the time of survey.
The deflected shapes of the railway envelope were predicted for different live load, dead load
and time-dependent effects. In order to tackle the inherent variability and future changes in the
parameters contributing to the movements, the predicted profiles were combined to represent the
expected structure movements at different times in the future. In essence, an envelope was
defined by means of a lower and an upper bound curve. The calculations took into account the
following phenomena:
(a)
State of construction at the time of survey

It was assumed that the closure segment was placed into position, the concrete stitches in
the flanges were cast and the four outer cables nearest to mid-span were stressed. The
design allowed for the cables anchored to the launching nose to be re-stressed after main
span closure. At the time of the main span reprofiling exercise, these cables were yet to be
re-stressed and the jacking forces required were not known precisely. Deflected shape A
was derived in accordance with the design assumptions and a 30% variation in the
predicted change in cable loads was allowed for in determining the envelope.
(b)
Temporary loads
A detailed record was made of the weight and positions of the plant and equipment which
were present on the bridge at the time of survey. A total of 770 tonnes was used for this
construction load, which was used to determine deflected shape B. In order to allow for
the variability in the magnitude or distribution of the construction loads, an envelope was
calculated on the basis of the temporary loads differing by +170 tonnes.
(c)
Superimposed dead loads

At the time of survey, only part of the superimposed dead load had been applied. An
allowance of 15kN/m was made in the calculations involving this load.
The concrete trackslabs, other elements of the trackform and bridge furniture would be
installed prior to Key Date 3. Deflected shape C allowed for 31kN/m per track over the
full length of bridge.
The road surfacing and the remaining components of the superimposed dead load would
be installed between Key Date 3 and Key Date 6. Deflected shape D allowed for
51kN/m per track over the length of the main span and 82kN/m on the side spans.
The envelope allowed for a +15% variability in the superimposed dead loads.
(d)
Structural response
In the development of the movement envelopes, account was taken of the variability in
material properties, approximation in modelling the transverse behaviour of the bridge
superstructure and other phenomena which were not explicitly tackled by the analysis. On
the basis of a comparison of the results given by different analytical models, a correction
of +10% was applied to each of the deflected shapes B, C and D, in addition to the
adjustments which were previously explained
(e)
Creep and shrinkage

The effects of creep and shrinkage were predicted analytically for different times in the
future.
Deflected shape E illustrates the expected creep and shrinkage movements from the
time of survey to Key Date 3 (8mm in 3 months at mid-span).
Deflected shape F represents the expected creep and shrinkage movements from Key
Date 3 to Key Date 6 (22mm in 9 months at mid-span).
In view of the uncertainty in predicting time-dependent effects, an adjustment of +50%

was applied to each of the above deflected shapes. The adjustment aimed to correct for
both the magnitude of the total movement and the rate at which it takes place.
The end result, supplied for use in the Wriggle analysis, was the best engineering estimate
of the main span deflections between the time of survey and Key Date 6. It was given by
the summation of deflected shapes A to F. A series of deflection profiles was also
derived with the inclusion of transient loads, as an assessment of the lower and upper
bound curves at Key Date 3, Key Date 6 and time infinity.
(f)
Live loads and temperature changes

Live load deflections vary from one point on the bridge to another. For an overall check,
the deflected shape was the range of peak values at each point. If the gradient or curvature
criteria were found to be critical at a particular location, a detailed check would be carried
out
Investigations showed that an overall temperature change does not induce significant
vertical movements. Sizeable movements are caused by differential temperature effects
between the stay cables and the superstructure
A temperature difference of +15oC between the stay cables and the rest of the bridge was
assumed for a load combination involving the permanent loads, live loads and
temperature.
Main Span - Cross Section
Main Span - Longitudinal Section
Fig 7 - Sections through the Railway Envelope
Fig 8 - Deflected Shapes of the Cable-stayed Bridge
Fig 9 - Deflected Shapes of Main Span
Fig 10 - Deflected Shapes of Lantau Side Span
Fig 11 - Deflected Shapes of Ma Wan Side Span
7.
Conclusion
The logistics of the work associated with the railway envelope survey and track reprofiling were
complicated by construction activities and the intense pressures from the construction
programme. Despite these constraints, the work enabled on-site adjustments during construction
to achieve compliance and was critical to the successful completion of the railway work.
8.
Acknowledgement
The author would like to thank the Government of Hong Kong Highways Department, the
Kumagai-Maeda-Yokogawa-Hitachi Joint Venture, Maunsell Consultants Asia Ltd and
Leonhardt Andr und Partner for the teamwork which made the project a success.
Comfort Criteria for High Speed Trains on

The resund Bridge
Jrgen GIMSING
Civil Engineer MSc
Horsens, Denmark
Anders THOMSEN
Civil Engineer MSc
ISC
Copenhagen, Denmark
Jrgen Gimsing received his

He is Managing Director of
Gimsing & Madsen A/S and
Technical Manager of the ASO
Group
Anders Thomson received his

He has worked as design
engineer for the ASO Group on
the resund Bridge design for 3
years and joined ISC in 1998.
Summary
The High Bridge of the resund Link is the longest cable stayed bridge carrying both a motorway
and a two-track railway for both passenger trains and freight trains.
Due to this, considerable efforts have been made to establish a comprehensive and coherent set of
design rules for the comfort of passengers in trains travelling at 200 km/h. The paper summarises
the investigations carried out by the Owners bridge consultant to establish these rules. One of the
objects of the investigations was to try to limit the number of analyses to be carried out in the
detailed design. This would have been possible if it could be demonstrated that some of the
requirements were always stricter than other requirements. This part of the study was, however, not
very successful. Only a small number of the original requirements were deleted due to a clear
demonstration that they would automatically be fulfilled.
1.
Introduction
The main features of the resund Bridge were developed by ASO Group (Ove Arup & Partners,
SETEC TPI, Gimsing & Madsen and ISC) for a design competition in 1993. During the following
1 year the design was refined to a tender design suitable for tendering on a Detailed Design and
Construction basis with all visible dimensions fixed by so-called Definition Drawings.
During this pre-tender period a number of special investigations were carried out. Some of these
were related to the aesthetical design, demonstrating that the dimensions on the Definition Drawings
were reasonably optimised, some were intended to provide a consistent basis for estimating cost and
construction time and some were carried out to establish relevant Design and Construction
Requirements. Among the latter were an investigation of comfort criteria to be fulfilled for the
bridge design, especially in relation to the high speed passenger trains travelling at 200 km/h while
the second railway track might be loaded by a heavy freight train. The basis for this investigation
was established by the Danish and the Swedish Railway Authorities based on the six ORE (Office
for Research and Experiments of the International Union of Railways) Reports: Permissible
Deflection of Bridges.
Fig. 1 Elevation Cable-Stayed High Bridge
2.
Comfort Criteria
The requirements from the railway authorities stated limitations on:
vertical deformations
vertical accelerations
horizontal deformations
horizontal accelerations
torsional deformations
deformations at expansion joints.
All these requirements were related to comfort criteria for passenger trains. For freight trains the
only relevant criteria was related to the wheel relief factor, which should be limited to 25%.
The first task in transforming the above criteria to usable Design Requirements was to define the
corresponding load combinations. It was found that the Ultimate Limit State combinations were too
onerous for this purpose, whereas the Serviceability Limit State combinations would be too lenient.
Three new load combinations for investigations of comfort criteria were therefore set up. In all these
the permanent loads including prestressing, creep, shrinkage and expected settlements were factored
by 1.0. The dominant traffic load (in most cases railway load) was factored by 1.0 while the other
important live loads (road traffic, wind, temperature effects) were factored by 0.5. The railway load
consists of the train under investigation on one track and the UIC 71 rail load on the other track.
2.1
Vertical Deformations
For span lengths up to 90 m the deflection at mid span should be limited to 1/2000 of the span
length. This requirement could for longer spans be replaced by requirements to the vertical
accelerations, refer Section 2.2 below. Even for the inherent, high rigidity of the two level bridge
(span to depth ratio in the Approach Bridges of approx. 13) it was found that the limitation on the
deflection would be far more onerous than the limitations on accelerations, and it was decided to
state the limit on deformations to be valid only for spans below 90 m.
2.2
Vertical Accelerations
The limitation on the vertical acceleration depends on the duration of the event in the following
manner:
for events with a duration less than 0.5 sec., the acceleration shall never exceed 2.00 m/sec2.
for events with a duration of min. 10 sec., the frequency weighted rms value of the acceleration
according to ISO 2631 shall be limited to 0.35 m/sec2.
for events with a duration between 0.5 and 10 sec. arms shall be limited to values between 0.5
and 0.35 m/sec2.
In principle all factors affecting the acceleration in the coaches should be taken into account, but
one of the first results of the investigations was that the contribution from allowable track defects
was so small that the designer was allowed to disregard this.
The passenger trains, which should be included in the analysis, are 3 existing Scandinavian trains
and the Euro City, refer the table below. The table also contains 4 freight trains for which only the
wheel relief factor is relevant, refer Section 2.7 below. The 4 freight trains look identical in the
table, the difference between them is the spring and damper system for the investigated freight
wagon.
Type
Speed (km/h)
Weight (kN/m)
Coaches
Loco
Mean train
length (m)
Train No. 1 (T1)
X2000 (S)
200
19.2
42.3/29
140
Train No. 2 (T2)
IR4 (DK)
180
21
21
153
Train No. 3 (T3)
IC3 (DK)
180
21
21
153
Train No. 4 (T4)
Euro City
160
17.6
46.3
270
Train No. 5 (T5)
Freight train
120
28.9
44.3
520
Train No. 6 (T6)
Freight train
120
28.9
44.3
520
Train No. 7 (T7)
Freight train
120
28.9
44.3
520
Train No. 8 (T8)
Freight train
120
28.9
44.3
520
Table 1 Investigated Train Types

The coaches in the modern passenger trains have a complicated system of springs and dampers at
each bogie. This system is simplified to one spring and one damper arranged in parallel. The
simplified effective spring and damper give the correct results assuming that interaction between the
two bogies and resonance within the complete spring-damper system can be disregarded. The
simplified model was stated to be acceptable in the Design Requirements.
The maximum vertical acceleration occurs when passing the angular bend at an expansion joint. A
number of analyses were carried out to study the effect of one sharp bend, two separate bends, each
half the value at varying distances and the total bend distributed over a varying length.
As an example the results for the trains T2-T5 are given in the figure below when passing 2 bends
each 0.002 rad and spaced from 0 to 40 m. The figure shows that the maximum acceleration in the
passenger trains is not significantly influenced by the distance between the two bends, and that the
requirement for max. acceleration below 2 m/sec2 is fulfilled in all cases. For the freight wagon the
influence of the distance between the two bends is marked, but it was found impossible to find a
length, which would be suitable for all 4 freight wagons, both loaded and empty. The simple sharp
bend at one point gives many advantages in the design of the track expansion and was therefore
chosen.
Accelerations
meter
Distance between bends
Fig. 2 Max. Acceleration as a Function of Distance between Bends
2.3
Horizontal Deformations
The Approach Bridges of the resund Link are placed with a horizontal curvature with a radius of
minimum 12800 metres. The minimum radius for the high speed trains is 5000 metres. The
requirements from the railway authorities stated that the minimum radius shall be fulfilled for the
bridge subject to wind load.
It was found that this requirement was superfluous, as the inherent stiffness of the bridge was more
than sufficient to fulfil it. The requirement was therefore not included in the Design Requirements.
2.4
Horizontal Accelerations
The requirement from the railway authorities was that the frequency weighted rms value of the
horizontal accelerations should be less than 0.35 m/sec2.
As above this requirement was not included in the Design Requirements, as it was demonstrated
that it was easily fulfilled with the minimum stiffness of the bridge.
2.5
Torsional Deformations
The requirements:
total torsional deformation shall be limited to 1.5%

torsional deformation per meter shall be limited to 0.06% for trains travelling at 200 km/h,
0.10% in all other cases
were included in the Design Requirements, even though they were found to be fulfilled for the two
level bridge. The reason for keeping the requirements was that an alternative design proposed by a
tenderer could have a reduced torsional stiffness, at least in theory, making the requirements
relevant.
2.6
Deformations at Expansion Joints
The requirements to angular deflection are stated in the table 2 below.

Angular Deflection
Train Type
Horizontal
Vertical
Freight Train v 120 km/h
0.2%
0.6%
Passenger Train v 200 km/h
0.1%
0.4%
Unloaded Bridge*
0.4%
0.9%
Table 2 Permissible Total Angular Deflections at Railway Expansion Joints.

*Refers to maximum permanent deformations, which can be corrected on ballasted tracks.
These requirements were found to be very strict and could lead to substantial extra costs for the 2
spans adjacent to an expansion joint in order to increase the stiffness of these spans. The solution to
this was to allow the Bridge Contractor to assume that the track laying would introduce a precamber
at the expansion joint of up to 0.2%. This allows an angular deflection of 0.6% for the passenger
train in one track, UIC 71 load in the other track plus 0.5 times road load, which was found to be
just fulfilled with the steel and concrete quantities needed in the static design of the end spans.
The fulfilment of the requirement to vertical angular deflection was found automatically to lead to
fulfilment of the requirement for maximum vertical acceleration 2 m/sec2 in the passenger coaches.
The latter requirement was, however, maintained as it is easily checked when the analysis to
determine the rms value of the accelerations is carried out.
The Design Requirements also include limitations on deformations at expansion joints as follows:
vertical displacement across a railway expansion joint max. 2.5 mm, however 5 mm during
bearing replacement
horizontal displacement across a railway expansion joint max. 1 mm.
The first requirement can only be fulfilled when the expansion joint is located at a pier top, which
makes it possible to jack up both bridge spans during a bearing replacement. The second
requirement necessitates that one of the two bearings for each span is transversely fixed, whereas
the second bearing shall allow temperature movements in the transverse direction relative to the pier
top.
This example shows that the comfort criteria have had a considerable influence on the allowable
choices of statical systems for the resund Bridge.
2.7
Wheel Relief Factor
The requirements from the railway authorities stated that the wheel relief for all freight wagons
should be limited to 25%. It was, however, found that this requirement could not be fulfilled for an
empty freight wagon in either of the four freight trains T5-T8 indicated in table 1 above. The wheel
relief factor when passing an expansion joint with the allowable angular deflection was found to be
up to 50%.
A literature study was then carried out to determine the origin of the requirement, which was found
to be Japanese rules for the Shinkansen high speed trains. The only requirement for wheel relief
factors in freight trains found in the literature was a United States requirement of 75%.
On this background it was agreed with the railway authorities to waive this requirement.
3.
Conclusions
The paper summarises some of the investigations carried out to finalise the section on comfort
criteria in the Design Requirements for the resund Bridge.
It is demonstrated that some of the original requirements from the Danish and the Swedish Railway
Authorities would have been nearly impossible to fulfil and that a consistent set of requirements has
been established.
The main conclusion is that the comfort criteria could be fulfilled for the two level bridge design,
that went out to tender in December l995. This conclusion has been confirmed by the more detailed
analyses carried out by Cowi-VBB, the bridge consultant for the successful tenderer, Sundlink
Contractors.
Nonlinear Dynamic Analysis of Cable-Stayed Bridges Excited by

Moving Vehicles
Raid KAROUMI
Researcher, Dr. Techn.
Dept. of Structural Eng.
Royal Inst. of Technology
Stockholm, Sweden
www.struct.kth.se
raid.karoumi@struct.kth.se
1.
Raid Karoumi, born 1964,

received his M.Sc. degree in
Civil Engineering in 1990
and his Dr. Techn. degree in
1999 from the Royal Institute
of Technology in Stockholm.
Between 1990 and 1993 he
worked as a consultant and
was involved in a variety of
special projects.
Introduction
Due to their aesthetic appearance, efficient utilization of structural materials and other notable
advantages, cable-stayed bridges have gained much popularity in recent decades. Bridges of this
type are now entering a new era with main span lengths reaching 1000 m. This fact is due, on
one hand to the relatively small size of the substructures required and on the other hand to the
development of efficient construction techniques and to the rapid progress in the analysis and
design of this type of bridges.
The recent developments in design technology, material qualities, and efficient construction
techniques in bridge engineering enable the construction of not only longer but also lighter and
more slender bridges. Thus nowadays, very long span slender cable-stayed bridges are being
built, and the ambition is to further increase the span length and use shallower and more slender
girders for future bridges. To achieve this, accurate procedures need to be developed that can
lead to a thorough understanding and a realistic prediction of the structural response due to not
only wind and earthquake loading but also traffic loading. It is well known that large deflections
and vibrations caused by dynamic tire forces of heavy vehicles can lead to bridge deterioration
and eventually increasing maintenance costs and decreasing service life of the bridge structure.
Although several long span cable-stayed bridges are being build or proposed for future bridges,
little is known about their dynamic behavior under the action of moving vehicles. The dynamic
response of bridges subjected to moving vehicles is complicated. This is because the dynamic
effects induced by moving vehicles on the bridge are greatly influenced by the interaction
between the vehicles and the bridge structure. To consider dynamic effects due to moving
vehicles on bridges, structural engineers worldwide rely on dynamic amplification factors
specified in bridge design codes. These factors are usually a function of the bridge fundamental
natural frequency or span length and states how many times the static effects must be magnified
in order to cover the additional dynamic loads. This is the traditional method used today for
design purpose and can yield a conservative and expensive design for some bridges but might
underestimate the dynamic effects for others. In addition, design codes disagree on how this
factor should be evaluated and today, when comparing different national codes, a wide range of
variation is found for the dynamic amplification factor. Thus, improved analytical techniques
that consider all the important parameters that influence the dynamic response are required in
order to check the true capacity of existing bridges to heavier traffic and for proper design of
new bridges.
The recent developments in bridge engineering have also affected damping capacity of bridge
structures. Major sources of damping in conventional bridgework have been largely eliminated
in modern bridge designs reducing the damping to undesirably low levels. As an example,
welded joints are extensively used nowadays in modern bridge designs. This has greatly reduced
the hysteresis that was provided in riveted or bolted joints in earlier bridges. For cable supported
bridges and in particular long span cable-stayed bridges, energy dissipation is very low and is
often not enough on its own to suppress vibrations. To increase the overall damping capacity of
the bridge structure, one possible option is to incorporate external dampers (i.e. discrete damping
devices such as viscous dampers and tuned mass dampers) into the system. Such devices are
frequently used today for cable supported bridges. However, it is not believed that this is always
the most effective and the most economic solution. Therefore, a great deal of research is needed
to investigate the damping capacity of modern cable-stayed bridges and to find new alternatives
to increase the overall damping of the bridge structure.
In this paper, the nonlinear dynamic response of a simple two-dimensional cable-stayed bridge
model, subjected to a moving vehicle, is studied. Bridge damping, exact cable behavior, and
nonlinear geometric effects are considered. This study focuses on investigating the influence of
vehicle speed, bridge damping, and a tuned mass damper on the bridge dynamic response.
2.
Bridge and Vehicle Modeling
2.1
Bridge structure
Modern cable-stayed bridges exhibit geometrically nonlinear behavior, they are very flexible and
undergo large displacements before attaining their equilibrium configuration. As an example,
due to this inherently nonlinear behavior, conventional linear dead load analysis, which assumes
small displacements, is often not applicable [1].
Cable-stayed bridges consist of cables, pylons and girders (bridge decks) and are usually
modeled using beam and bar elements for the analysis of the global structural response. To
consider the nonlinear behavior of the cables, each cable is usually replaced by one bar element
with equivalent cable stiffness. This approach is referred to as the equivalent modulus approach
and has been used by several investigators, see e.g. [1, 2, 3]. It has been shown in [4] that the
equivalent modulus approach results in softer cable response as it accounts for the sag effect but
does not account for the stiffening effect due to large displacements. Still, for some cases, e.g.
for short span cable-stayed bridges, linear analysis utilizing the equivalent modulus approach is
often sufficient [3], especially in the feasibility design stage. Whereas, long span cable-stayed
bridges built today or proposed for future bridges are very flexible, they undergo large
displacements, and should therefore be analyzed taking into account all sources of geometric
nonlinearity. Although several investigators studied the nonlinear behavior of cable-stayed
bridges, very few tackled the problem of using cable elements for modeling the cables. See ref.
[5, 6] where different cable modeling techniques are discussed and references to literature
dealing with the analysis and the behavior of cable structures are given.
In this paper, an alternative approach is presented where accurate and efficient elements are
adopted for the modeling. A beam element, which includes geometrically nonlinear effects and is
derived using a consistent mass formulation, is adopted for modeling the girder and the pylons.
Whereas, a two-node cable element derived using exact analytical expressions for the elastic
catenary, is adopted for modeling the cables. The nonlinear finite element method is utilized
considering all sources of geometric nonlinearity, i.e. change of cable geometry under different
tension load levels (cable sag effect), change of the bridge geometry due to large displacements,
and axial force-bending moment interaction in the bridge deck and pylons (P- effect).
The adopted beam element, able to resist bending, shear, and axial forces, is developed following
the total Lagrangian approach and using a linear interpolation scheme for the displacement
components. This element is chosen because it can handle large displacements and shear
deformations and because it is simple to formulate the element matrices. This beam element is of
minor interest and, due to space limitation, not discussed here in more detail. The interested
reader is referred to the authors doctoral thesis, reference [5], where formulation of this beam
element is presented in detail.
In the following subsection, the cable element matrices will be given in the element local
coordinate system. Using this approach, each cable may be represented by a single 2-node finite
element, which accurately consider the curved geometry of the cable. Despite the fact that this
cable modeling technique has been available for many years it has, at least to the authors
knowledge, very seldom been used for analysis of cables in cable-stayed bridges.
2.1.1 Cable element
Consider an elastic cable element, stretched in the vertical plane as shown in Figure 1, with an
unstressed length Lu, modulus of elasticity E, cross section area A, and weight per unit length w
(uniformly distributed along the unstressed length). For the elastic catenary, the exact relations
between the element projections and cable force components at the ends of the element are:
L
1 P4 + T j
Lx = P1 u + ln
EA w Ti P2
Ly =
T j Ti
1
T j2 Ti2 +
2 EAw
w
u 4 , P4
(1a)
node j
u 3 , P3
(1b)
Ly
Lu , E , A , w
u 2 , P2
where Ti and T j are the cable tension forces at

the two nodes of the element. For the above
expressions it is assumed that the cable is
perfectly flexible and Hookes law is applicable
to the cable material.
u 1 , P1
node i
Lx
Figure 1: Catenary cable element
By rewriting the above expressions for Lx and Ly in terms of the end forces P1 and P2 only
using the relationships:
P4 = w Lu P2 ;
P3 = P1 ;
Ti = P12 + P22 ;
T j = P32 + P42
(2a-d)
differentiating the new expressions for Lx and Ly and rewriting the results using matrix notation
gives:
Lx
dLx P1
dL = L
y y
P1
Lx
dP
P2 dP1
= F 1
Ly dP
dP2
2
P2
(3)
where F is the flexibility matrix. The stiffness matrix is given by the inverse of F, i.e. K = F 1 .
The tangent stiffness matrix K t and the corresponding internal force vector p for the cable
element can now be obtained in terms of the four nodal degrees of freedom as:
k1 k 2 k1
k4 k2
Kt =
k1
sym.
k2
k4
;
k2
k4
P1
P

p = 2
P3
P4
(4a,b)
where
k1 =
k4 =
1 Lu 1 P4 P2
;
+
+
det F EA w T j Ti
k2 = k3 =
1 P1 1 1
det F w T j Ti
1 Lx 1 P4 P2
+
+
det F P1 w T j Ti
(5a,b)
(5c)
L
P L
P P 1
1 P
1 P
1
det F = u 4 + 2 x + 4 + 2 1
EA w T j Ti P1 w T j Ti w T j Ti
(5d)
The element tangent stiffness matrix K t relates the incremental element nodal force vector
to the incremental nodal displacement vector { u1 , u2 , u3 , u4 }T . To
evaluate the tangent stiffness matrix K t , the end forces P1 and P2 must be determined first.
Those forces are adopted as the redundant forces and are determined, from given positions of
cable end nodes, using an iterative stiffness procedure. This procedure requires starting values
for the redundant forces. Based on the catenary relationships the following expressions will be
used for the starting values:
{ P1 , P2 , P3 , P4 }T
P1 =
w Lx
2
and
P2 =
w
cosh
+ Lu
Ly
2
sinh
L2u L2y
1
where = 3
L2
(6a-c)
In cases where equation (6c) cannot be used because the unstressed cable length is less than the
chord length, a conservative value of 0.2 for is assumed. Another difficulty arises in equation
(6c) for vertical cables. In that case an arbitrary large value of 106 for is used. Using equations
(2a-d), new cable projections corresponding to the assumed end forces P1 and P2 are now
T
determined directly from equations (1a,b) and the misclosure vector { Lx , Ly} is evaluated
as the positions of the end nodes are given. Corrections to the assumed end forces can now be
made using the computed misclosure vector as:
Lx
P1
= K L ;
P2
y
P1

P2
i +1
P P
= 1 + 1
P2 P2
(7a,b)
where K is the stiffness matrix (the inverse of F in equation (3)) and i is the iteration number.
For the present study, this iteration process continued until Lx and Ly are less than 1 105 . As
will be demonstrated later, this iterative procedure converges very rapidly.
To determine the unstressed cable length, Lu , for cases where the initial cable tension is known
instead, a similar iteration procedure can be adopted. A starting value for the unstressed cable
length is assumed, e.g. equal to the cable chord length, and cable end forces P1 and P2 are
computed using the iterative procedure described above. Using equation (2c,d), cable tension can
now be computed. This is then compared with the given initial tension to obtain a better
approximation for Lu for the next iteration step.
For the dynamic analysis, mass discretization is simply done by static lumping of the element
mass at both ends giving the following lumped mass matrix ( is the mass density of the cable):
M=
2.2
AL
2
1
0
0
1
0
0
0
0
1
0
0
0
(8)
Vehicle model
The vehicle model used in this study is a so-called suspension model that includes both primary
and secondary vehicle suspension systems, see Figure 2. This model is sufficient since the main
concern is to investigate the dynamic response of bridges and not the dynamics of the vehicle
itself and since the spans of cable-stayed bridges are considerably larger than the vehicle axle
base. It is assumed that the vehicle never loses contact with the bridge and the contact between
the bridge and the moving vehicle is assumed to be a point contact. The equation of motion for
the vehicle is coupled to the bridge equation of motion through the interaction force existing at
the contact point of the two systems. To solve these two sets of equations, an iterative procedure
is adopted, as the interaction force is dependent on the motion of both the bridge structure and
the vehicle. Vehicle load modeling and the developed moving load algorithm are described in
detail in reference [5]. The implemented codes fully consider the bridge-vehicle dynamic
interaction and have been verified in [5].
w3(t)
v (t)
m3
ks
w2(t)
(b)
(a)
cs
Vehicle data
m2
kp
w1(t)
cp
m1
node i
node j
xc
e
mode 1
1.5-4 Hz
mode 2
8-15 Hz
m1 = 0 kg
m2 = 4840 kg
m3 = 39160 kg
cp = 1.76 104 Ns/m
cs = 13.2 104 Ns/m
kp = 15.4 106 N/m
ks = 8.8 106 N/m
Figure 2: (a) vehicle model on a bridge element; (b) typical vehicle modes of vibration
3.
Nonlinear Analysis Procedure
The equation of motion for the entire bridge is obtained as:

+ C q + p(q ) = f (q, q , q
, t )
Mq
(9)
are the bridge node displacement, velocity, and acceleration vectors, respectively,
where q, q , q
M the bridge mass matrix, C the bridge damping matrix, p(q) the vector of internal elastic
,t) the external force vector resulting from the dead load, the moving
forces, and f( q, q , q
vehicles, and the tuned mass dampers. As indicated, the external force vector is not only time
dependent but is also dependent on the bridge displacements, velocities and accelerations. This
vector contains the interaction forces existing at the contact points between the vehicles and the
bridge and thereby couples the bridge equation of motion with those of the vehicles.
For this study, an implicit procedure based on Newmarks average acceleration method
combined with a full Newton-Raphson solution procedure (i.e. the Newton-Newmark algorithm)
is adopted for solving the bridge equation of motion. For the interested reader, details concerning
the derivation etc. of this nonlinear dynamic procedure as well as a linear dynamic procedure
based on the utilization of the dead load tangent stiffness matrix can be found in [5]. Nonlinear
dynamic analysis is essential if it is believed that the bridge will not behave linearly during the
application of live loads. If this is the case, the natural frequencies will vary with the amplitude
of response and linear dynamic analysis will consequently be inadequate.
To evaluate the nonlinear static response, an incremental-iterative procedure using full NewtonRaphson iterations is adopted. This procedure is generally expected to give quadratic
convergence.
4.
Numerical Example
A 2D model of the cable-stayed bridge described in [1] was adopted for this investigation. The
bridge geometry is shown in Figure 3 and the properties are given in Table 1.
5x3
45.7
30.5
335.3 m
146.3 m
146.3 m
Figure 3: Geometry of the cable-stayed bridge. The cables are numbered from the left to the
right starting with cable 1
Girder
E (N/m2) A (m2)
I (m4)
w (t/m)
Cable no.
E (N/m2)
A (m2)
Lu (m)
w (t/m)
2.01011
0.26
19.64
1, 24
2.01011
0.0362
158.13
0.398
2.010
11
0.0232
134.66
0.255
2.010
11
0.0204
111.64
0.225
2.010
11
0.0176
89.43
0.194
2.010
11
0.0139
68.80
0.153
2.010
11
0.0113
51.69
0.125
2.010
11
0.0372
158.12
0.409
11
Girder
central part
2.010
Pylons above
deck level
2.81010
Pylons below
deck level
2.81010
0.93
1.11
13.01
18.58
1.29
34.52
86.31
19.64
30.65
43.78
2, 11, 14, 23
3,10, 15, 22
4, 9, 16, 21
5, 8, 17, 20
6, 7, 18, 19
11
Links deck to
2.010
0.56
pylons
Including weight of cross beams.
0.10
4.38
12, 13
Table 1: Parameters for the cable-stayed bridge model defined in Figure 3

For the model, it was assumed that the girder was pinned at the ends, i.e. only rotations were
allowed, and elastically connected to the pylons by vertical links. The pylons were assumed to be
rigidly fixed to the piers, and all cables were assumed fixed to the pylons and to the girder at
their joints of attachment. The model had 119 active degrees of freedom and was composed of
66 elements and 43 nodal points. The CPU time used by the computer (200 MHz Pentium Pro)
to find the tangent stiffness matrix at the dead load deformed state and solve the system
eigenvalue problem determining all 119 modes of vibration, was about 15 seconds. This
indicates high efficiency of the presented elements. The first three bending natural frequencies
obtained utilizing the dead load tangent stiffness matrix are: 0.332, 0.436, 0.692 Hz. Bridge
damping ratios were assumed constant for all modes and equal 0.0056.
This bridge model was then subjected to a 44 ton truck moving from the left to the right on a
smooth road surface at the constant speed v, see Figure 3. The body-bounce and wheel-hop
frequencies, for the truck model, were chosen as 1.89 and 11.35 Hz. The corresponding mode
shapes and vehicle model properties are shown in Figure 2. In the following subsections, the
effect of speed, bridge damping and a tuned mass damper on the bridge response is presented.
4.1
Vehicle speed and bridge damping effect
The vertical displacement of the girder at the center of the bridge, due to traffic load only, is
shown in Figures 4a and 4b for different speeds and damping ratios. For the curves in Figure 4b,
Vert. displacement (mm)
the vehicle speed was v = 90 km/h.

The static traffic load response is
also plotted in this Figure. 1500
increments were required for the
solution of the 50 km/h case, and
1000 increments for the rest.
25
5
-15
-35
v = 50 km/h
v = 90 km/h
v = 110 km/h
(a)
-55
As expected, damping reduces the
0
5
10
15
20
25
30
35
40
45
bridge response. For = 0.0056 in
Time (s)
Figure 4b, the absolute maximum
dynamic displacement is about 20%
25
larger than the static one (dynamic
5
amplification factor of 1.2). It can
be concluded from the results in
-15
Static
Figure 4 that the response increases
=0
-35
with the increase in vehicle speed
= 0.0056
(b)
= 0.03
and that bridge damping has a
-55
significant effect upon the response
0
5
10
15
20
25
and should therefore always be
Time (s)
considered if accurate representation
Figure 4: Vertical displacement at the center of the bridge
of the true dynamic response is
calculated for different vehicle speeds (a) and
required.
bridge damping ratios (b)
The effectiveness of a tuned mass

damper (TMD) in suppressing
vibrations due to a single 44 ton
moving truck is investigated in this
study. The truck was assumed to
move on a smooth road surface at
the constant speed of 110 km/h.
Reasonably converged reliable
solutions were obtained using 1000
increments corresponding to a time
step of 0.025 s. The TMD was
positioned at the center of the bridge
and tuned to the first bending mode
of vibration. The following most
often used optimum tuning
parameters, derived in [7] for an
undamped structure, are adopted:
tmd
20
5
-10
-25
-40
(10a)
3
8 (1 + ) 3
(10b)
with TMD
without TMD
(a)
-55
0
90
60
30
0
-30
-60
-90
-120
10
20
30
40
Time (s)
50
60
70
10
20
30
40
Time (s)
50
60
70
(b)
0
= i
1+
tmd =
Effect of a tuned mass damper
4.2
Figure 5: (a) Vertical displacement of the girder at the center

of the bridge; (b) vertical displacement of the TMD
mass. The dashed vertical line indicate when the
truck leaves the bridge
Axial force (MN)
Bend. moment (MNm)
Hor.displacement (mm)
where tmd and i are the circular

frequencies of the TMD and the
10
dominant bridge mode to be tuned
4
to, tmd the damping ratio of the
-2
TMD, and is the mass ratio which
relates the TMD mass to the modal
with TMD
-8
mass of the dominant bridge mode
without TMD
(a)
-14
to be tuned to, = mtmd / mi . The
0
10
20
30
40
50
60
70
mass ratio was here set to 0.005
Time
(s)
giving a TMD mass of about 15.6
ton. Some results from the nonlinear
10
dynamic analysis showing the
with TMD
(b)
response due to traffic loads only
without TMD
4
are presented in Figures 5 and 6.
Figure 7 (in the abstract) shows a
-2
cross section of a bridge girder with
a tuned mass damper. It was found
-8
from the analysis results that the
0
10
20
30
40
50
60
70
TMD not always is very effective in
Time (s)
reducing the maximum dynamic
response during the forced vibration
0.5 (c)
period (i.e. when the vehicle is on
with TMD
without TMD
the bridge). In fact, due to the
0.2
interaction between the bridgevehicle-TMD systems, the
-0.1
maximum response for certain
elements and nodes can even
-0.4
increase due to the TMD. However,
0
10
20
30
40
50
60
70
it is evident from Figures 5 and 6
Time (s)
that the TMD is very effective in
reducing the vibration level in the Figure 6: (a) horizontal displacement of the right pylon top;
(b) bending moment at the right pylon fixed end; (c)
free vibration period for all elements
axial force in the right anchorage cable. The
and nodes. This is due to the
dashed vertical line indicate when the truck leaves
increase of the overall damping of
the bridge
the bridge by the TMD.
5.
Conclusions
This paper has presented a method for modeling and analysis of cable-stayed bridges subjected
to moving vehicles. Bridge damping, exact cable behavior, and nonlinear geometric effects have
been considered when analyzing the nonlinear dynamic response. The study has only focused on
investigating the influence of vehicle speed, bridge damping, and a tuned mass damper on the
bridge dynamic response.
A two-node catenary cable element was adopted for modeling the cables and it has been found
that the main advantages of this cable element are the simplicity of including the effect of
pretension of the cable and the exact treatment of cable sag and cable weight. Moreover, the
iterative process adopted to find the internal force vector and tangent stiffness matrix for the
cable element is found to converge very rapidly. According to the authors opinion, linear
analysis utilizing the traditional equivalent modulus approach, is not satisfactory for modern
cable-stayed bridges. Modern cable-stayed bridges built today or proposed for future bridges are,
as they are highly flexible, subjected to large displacements. The equivalent modulus approach
however accounts only for the sag effect but not for the stiffening effect due to large
displacements.
From the study of the traffic load response of cable-stayed bridges it is concluded that the
response increases with the increase in vehicle speed and that bridge damping have a significant
effect upon the response and should always be considered in such analysis. Bridge damping
ratios should be carefully estimated to insure more correct and accurate representation of the true
dynamic response. To obtain realistic damping ratios, such estimation should be based on results
from tests on similar bridges. In addition it is concluded that a tuned mass damper is not very
effective in reducing the maximum dynamic response during the forced vibration period (i.e.
when the vehicle is on the bridge). In fact, such a device can even increase the maximum
dynamic response of some nodes and elements. However, the reduction of the vibration level in
the free vibration period is significant as the tuned mass damper increases the overall damping of
the bridge by working as an additional energy dissipater
In reference [5], the influence of other important parameters such as road surface roughness,
bridge-vehicle interaction, and cables vibration (i.e. multi-element cable discretization) is
investigated. In addition, the dynamic response of other bridge models, such as the Great Belt
suspension bridge in Denmark, is also studied.
6.
References
[1]
Nazmy A.S., Abdel-Ghaffar A.M., Three-Dimensional Nonlinear Static Analysis of

Cable-Stayed Bridges, Computers and Structures, 34, 1990, pp. 257-271.
[2]
Karoumi R., Dynamic Response of Cable-Stayed Bridges Subjected to Moving Vehicles,

IABSE 15th Congress, Denmark, 1996, pp. 87-92.
[3]
Kanok-Nukulchai W., Yiu P.K.A., Brotton D.M., Mathamatical Modelling of CableStayed Bridges, Struct. Eng. Int., 2, 1992, pp. 108-113.
[4]
Ali H.M., Abdel-Ghaffar A.M., Modeling the Nonlinear Seismic Behavior of CableStayed Bridges with Passive Control Bearings, Computers and Structures, 54, 1995, pp.
461-492.
[5]
Karoumi R., Response of Cable-Stayed and Suspension Bridges to Moving Vehicles

Analysis methods and practical modeling techniques, Doctoral Thesis, TRITA-BKN
Bulletin 44, Dept. of Struct. Eng., Royal Institute of Technology, Stockholm, 1998.
[6]
Karoumi R., Some Modeling Aspects in the Nonlinear Finite Element Analysis of Cable
Supported Bridges, Computers and Structures, (in printing).
[7]
Den Hartog J.P., Mechanical Vibrations, 4th edition, McGraw-Hill, New York, 1956.
Deformability of Long-Span Cable-Stayed Bridges for Railways

Domenico BRUNO
University of Calabria
Cosenza, Italy
Antonio GRIMALDI
University of Rome
Tor Vergata, Rome, Italy
Angelo LEONARDI
University of Rome
Tor Vergata, Rome, Italy
Summary
In this paper an analysis of the statical and dynamical behaviour of long span cable-stayed bridges
traversed by moving loads is developed. The structural model of the bridge is defined by referring
to the fan-shaped scheme and according to a continuous distribution of stays along the girder .
An analytical and a numerical analysis are developed to study the statical behaviour of the bridge
by using a continuous or a discrete model of the structure; while, a numerical analysis is carried
out, based upon time integration of the motion equations of the discretized structure to model the
dynamical effects of the moving loads.
The main structural nonlinearity arising from the elastic response of stays is accounted for together
with the nonlinear effects related to the dynamical effects of the moving loads.
Some numerical results show the influence of the main geometrical and load parameters on the
elastic response of the bridge.
1. Introduction
Cable-stayed bridges have been of great interest in recent years, particularly with respect to the
fan-shaped scheme as a valid and alternative solution to suspension bridges for long spans.
Troitsky (1) and Gimsing (2) have reviewed the problems and advantages of cable-stayed bridge
solutions and reported on the latest and most interesting projects.
For long-span bridges one of the most important problems is related to the deformability under
live loads. In the case of bridges carrying both road and railway traffic, and for spans greater than
1000 m, this aspect can seriously influence the design and the feasibility of the structure.
In this work an analysis of the static and dynamic behaviour of the bridge is developed, modeling
the train passage as a dynamical action or an equivalent static load.
Como et al. (3) analyzed the static behaviour of long span stayed-cable bridges showing the
prevailing truss behaviour of the bridge. Bruno and Grimaldi (4) investigated the nonlinear static
behaviour of cable-stayed bridges using both a continuous and a discrete model of the bridge, and
showed the strong influence of nonlinearities for long spans. Moreover, the dynamic behaviour of
cable-stayed bridges has been investigated by Bruno (5) analyzed the effects of moving loads, and
by Bruno, Maceri and Leonardi (6) who analyzed aerodynamic instability problems.
In above studies the fan-shaped cable-stayed bridges was studied using both a continuous and a
discrete model of the bridge, and the dominant truss behaviour of the bridge was found. In
particular, the influence of the dynamic properties and geometric nonlinearities of the structure are
included in the analysis.
In this paper the continous model proposed in the previous works is employed to develop static
and dynamic analyses. In addition a discrete model is also applied to give some useful
comparisons between analytical and numerical results.
The aim of the paper is to give same results and conclusions related the main geometric and
mechanical parameters able to influence or control the bridge deformability.
In particular, the geometric aspect ratio L/H between the main span length and the tower heigth,
the loads ratio p/g between live and dead loads, the relative flexural stiffness between girder and
stays, are taken into account.
Obtained results show the strong influence of the rail load on the midspan deflection and on the
girder slopes. As well known this last quantities in many cases represent the main parameter
which have to be considered in deformability control.
2. Bridge scheme and numerical results.

The fan-shaped scheme of cable-stayed bridge of Fig.1 is considered, in which the girder is simply
supported at its ends and is hung to the tops of H-shaped towers by means of two stays curtains.
It is assumed that the stays spacing is a small quantity compared to the central span length Lc.
The aspect ratios r1=Lc/H; r2=Ls/H of span legths to the tower height are usually obtained on the
basis of economy and of the anchor cable stability condition.
The longitudinal vertical plane yz is assumed to be a symmetrical one; in addition, the bridge is
also symmetrical with respect to the midspan cross plane.
According to the usual erection procedures, girder and towers are assumed to be free from bending
under dead load g. Then, the cross sectional areas As and Ao of the couple of diffused stays and of
the anchor stays, respectively, are obtained by referring to the truss scheme of the bridge:
As =
g
;
g sin
A0 =
where
g
;
g = a
p+g
1
gL s
L
L
[1 + ( s ) 2 ] 2 [( c ) 2 1]
2 g 0
2L s
H
g0
2L
p
= a 1 + [1 ( s ) 2 ]1
Lc
g
and where a is the allowable stress, is the angle between a stay and its horizontal projection, p
is the live load and g is the dead load.
We assume that towers and girders axial elongations are negligible, and we apply the beam model
for bending and torsion of the girder.
As far as the stays behaviour is concerned, the Dischinger modulus Es*=E/(1+2l02 E/1203) is
used, where Eis the Young modulus, is the specific weight, l0 is the horizontal projection length
of the stay and 0 is the initial tension. The tower is characterized by the flexural stiffness k and
the torsional stiffness kT, while the girder is characterized by the inertia I and the torsional
stiffness factor Ct.

It is convenient to introduce the following non dimensional quantities:
2 H 2E
a=
3 ;
12 g
4 I g
= 3 ;
4 H g
Ct g
,
2
Eb Hg
that is:
-the bridge span parameter =
-the girder nondimensional flexural stiffness parameter
-the girder nondimensional torsional stiffness parameters
For the H-shaped towers scheme, the deformation of the bridge can be described by the following
displacement parameters:
- the horizontal translation w of the girder;
- the axial displacements L, R and the torsional rotations L, R of the tower tops;
- the vertical deflection v of the girder;
- the torsional rotation of the girder.
While, if the A-shaped towers scheme is considered, the tower tops torsional rotations must be
taken : L=R =0.
E A
0
k,k T
0
E A
s
H
0
I,C t
Ls
Lc
Ls
Fig.1. Cable stayed bridge scheme
It is possible to show (3,4) that the statical behaviour of the bridge can be studied by using a
continuous structural model which gives the main bridge deformation and stress parameters.
In Fig.3 some results relative to the deformation of the bridge are given, where both an analytical
continuous model and a FEM discrete one of the bridge are employed.
The results
refers
the following geometrical and material parameters of the bridge: r1=2.5; r2=5/3;
6
2
E=21x10 t/m ; k/g=50.
In this figure the case of high live loads (p/g=1), like that railway bridges, is considered. It can be
observed that, the transverse deflection of the bridge is practically unaffected by the tower shape,
on the contrary, the torsional deformation is strongly influenced by the tower shape. The results
are given in dimensionless
form, where the quantity 0 represents the nodimensional torsional
couple 0 = H g m/Egb2.
Lp
c
p
ct
Ls
Lc
Ls
Fig.2. Bridge scheme under moving loads

A dynamical analysis of the bridge can also be developed by using both a continuous model and a
discrete one to analyze free vibrations, aerodynamic instability and the dynamical effects of the
moving loads.
Here we use a discrete formulation via finite elements to investigate the dynamical effects of
moving loads on stress and displacement characteristics.
Let us consider the bridge traversed by a strip of load p moving at constant speed c (Fig.2);
moreover, let be the moving load mass and the girder mass, with /=p/g.
3
The aim of the analysis is to determine the stress and displacement increments starting from the
initial equilibrium configuration under dead load g.
To this end a discrete model of the bridge is constructed by using a finite element model of the
structure where a cubic interpolation of the vertical displacements and a linear interpolation for
torsional rotations is applied.
Then, the dynamical equilibrium equations of the bridge can be put in the following discrete form:
..
.
M s + K(s)s= F(s ,s, t )
where M is the mass matrix, K is the stiffness matrix, F is the load vector and s is the generalized
displacement vector. It must be observed that in the previous equation the load vector F accounts
for the dynamical effects of the moving load mass.
The above nonlinear problem was solved numerically by using the Newmark integration method.
Preliminary numerical experiments have been made to choose the time step-size in Newmarks
scheme.
Numerical results are given in Fig.4, with the meaning:
- v : the dynamic amplification factor of the midspan vertical deflection;
- M : the dynamic amplification factor of the midspan bending moment;
- : the dynamic amplification factor of the midspan stay;
The two cases of massless moving loads ( /=0) and moving loads with mass ( / 0) were
examined. It can be observed the influence of the moving load mass on the stress and
displacement amplification, particularly at high speed.
References.
[1].
[2].
[3].
[4].
[5].
[6].
Troitsky, M.S. (1977). Cable - Stayed Bridges . Crosby Lockwood Staples, London.
Gimsing, N.J. (1983). Cable supported bridges. Wiley Interscience.
Como, M., Grimaldi, A., and Maceri, F. (1985). "Statical Behaviour of Long-Span CableStayed Bridges". Int. J. Solids and Struct., 21(8), 831-850.
Bruno, D., and Grimaldi, A. (1985). "Nonlinear Behaviour of Long-Span Cable-Stayed
Bridges". Meccanica, Vol. 20, N. 4, 303-313.
Bruno D. (1985). "On the dynamical behaviour of long-span cablestayed bridges under
moving loads, Costruzioni Metalliche, N. 2.
Bruno, D., Maceri, F., and Leonardi, A. (1987). "On the Nonlinear Dynamics of CableStayed Bridges". Procs. Int. Conf. on Cable-Stayed Bridges, Bangkok, Thailand, 529-544.
M ID S P A N V E R T IC A L D E F L E C T IO N
M ID S P A N T O R S IO N A L R O T A T IO N
10 3
Lc
= 0 .1
= 0 .2
=0
.0
" H " -s h a p e d to w e rs
= 0 .3
=0.
= 0 .0
" A " -s h a p e d to w e rs
( b /H = 0 .1 )
p
g
a n a ly tic a l
n u m e ric a l
Fig.3. Statical analysis :Flexsural and Torsional deformations
p
g
= 0 .1
a n a ly tic a l
n u m e ric a l
1.6
1.6
1.5
1.4
L
2H =2.5
l
=5/3
H
E =7200/2.1x10
K
g =50
1.5
1.4
1.3
1.3
1.2
1.1
L
2H =2.5
l
H =5/3
E =7200/2.1x10
K
g =50
a
1.2
p/g=1
a =0.10
/=1
/=0.5
1.1
1.0
1.0
/=0
10 20
p/g=0.5
a =0.10
30
40
/=0
50
60
70
80
10 20 30 40 50 60 70 80 90 c(m/s)
90 c(m/s)
1.6
1.5
1.4
L
2H =2.5
l
=5/3
H
E =7200/2.1x10
K
g =50
a
1.5
1.3
1.4
/=1
1.2
1.3
1.1
1.2
1.0
0.9
1.0
p/g=1
a =0.10
0.8
0.9
/=0
0.7
0.8
0.1
0.1
1.1
L
=2.5
2H
l
H =5/3
p/g=0.5
a =0.10
=0
10 20 30 40 50 60 70 80 90
10 20 30 40 50 60 70 80 90 c(m/s)
=0.5
1.1
6
a
E =7200/2.1x10
L
=2.5
2H
1.1
K =50
g
l
H =5/3
C (m/s)
6
a
E =7200/2.1x10
K =50
g
=1
=0.5
1.0
1.0
=0
p/g=1
a =0.10
0.9
10
20
=0
30
40
50
60
p/g=0.5
a =0.10
0.9
70
80
90
c (m/s)
10
20
30
40
50
60
70
80
90
c (m/s)
Fig. 4. Dynamical analysis of the bridge under moving loads (/=0 : massless moving loads)
6
Active Tendon Actuators for Cable-Stayed Bridge

Claude DUMOULIN
Civil Engineer
Bouygues Travaux Publics
Summary
Presented in this paper are the most recent numerical and experimental studies performed by the
partners of a European research project, partly funded by the European Commission, which aimed at
the development of Active Control Actuators for Cable-Stayed Bridge. This project is concerned with
the use of active control, and more precisely active tendon control, in order to reduce induced vibrations
of cable-supported structures. Such studies include development of innovative materials and techniques,
numerical models of actuators and structures, manufacturing and testing of prototypes and mock-ups.
At the present stage the performance requirements of the innovative actuators have been defined, the
actuators and the mock-up have been constructed.
Introduction
Improvements in materials have led to the construction of progressively longer, structurally more efficient
and slender bridges. But consequently, structures are more and more flexible. The mitigation of deck
and cable vibrations has become a major issue in cable-stayed bridge design. The increasing span length
of bridges makes them more sensitive to flutter instability as well as to wind and live load induced
vibrations.
This project is concerned with the use of active control, and more precisely active tendon control, in
order to reduce induced vibrations of cable-supported structures. The objective is not to minimise the
transverse vibrations of the cables alone. Many passive damping devices have already been developed
and used, especially dashpot dampers, systems of cable ties and viscous-elastic systems. These devices
will be cheaper than active control systems for a long time. The main objective is to reduce the
vibrations of the whole structure. In other terms, the aim is to increase the equivalent damping of cablesupported structures, which have a poor structural damping. Typically, the damping ratio, which is the
ratio of the given damping to the critical value, is often lower than one percent. Furthermore, the
damping (of the odd modes) of the cables will also be obtained.
The proposed technology uses a control strategy developed by the Active Structures Laboratory of
Universit Libre de Bruxelles, a partner of the consortium. The control strategy is based on a force
sensor co-located with the active tendon actuator. The basic principle of the active control is to force
the dynamic tension of the cable to produce a work by moving an anchorage according to the variation
of the tension measured at the same anchorage. Stability is guaranteed, even at the parametric
resonance. Numerical results have been confirmed in laboratory by experiments. An active damping
ratio has been obtained, which is superior to previously published results. In spite of the strong nonlinearity of the cables, an approximate linear design method has been developed. The technique has a
strong physical support; it allows prediction of the closed-loop poles with a root-locus technique.
Simulations have confirmed the effectiveness of this method.
Figure 1 - Basic principle of active damping

The aim of this project is to implement this laboratory development with complex civil structures. The
key problem is the actuator. The ambition is to limit the number of active devices needed for the active
control of a structure. Magnetostrictive actuators may provide a very innovative but unexplored solution.
An alternative more classical hydraulic actuator will also be studied.
The project requires activity on the following topics:
Definition of performance requirements for selected cable-supported structures and, consequently,
performance needs for the innovative active control system elements.
Development of the computer simulation and the innovative active control system.
Development of improved materials, functional mechanisms, geometry, operational simplicity and life
of the actuators.
Design and manufacture of actuator prototypes, including the actuator itself and the improvement of
the cable needed to cope with the movement induced by the actuator stroke.
Design and manufacture of suitable mock-up structures.
Development of specific test techniques and execution of experimental analysis of the actuator
prototypes.
Development of numerical models of the devices and mock-up and validation of such models based
on numerical analysis of test results or available measurements of existing structures.
Development of numerical models (integrating those previously defined and validated) and execution
of detailed numerical analysis of structures to be built or retrofitted by means of the innovative active
control systems, in order to improve the design of such structures.
Development of active damping and control algorithms.
The partners involved in the project can cover all the relevant aspects required for a fruitful co-operation
because they include R&D performers, designers of structures, manufacturers of devices, testing
laboratories and end-users. They are listed in alphabetical order, with specification of their main
functions relevant to the project:
Bouygues
Defence Evaluation and Research Agency
Johs.Holt
Joint Research Centre of EC Ispra
Mannesmann Rexroth
Newlands Technology
Technische Universitt Dresden
Universit Libre de Bruxelles
VSL
(FR)
(GB)
(NO)
(EC)
(DE)
(GB)
(DE)
(BE)
(FR)
designer of structures & end-user

R&D performer
designer of structures
testing laboratory
manufacturer of devices
R&D performer
R&D performer
The research is partly funded by the European Commission under the Brite-EuRam programme
(Contract NBRPR-CT97-0402, project ACE). The author as co-ordinator of the ACE project,
gratefully acknowledge the contributions of the other partners of the Consortium.
Definition of performance requirements for the active control system

The preliminary requirements concerning typical values for active tendons located on the main cables of
a cable-stayed bridge exhibit high forces and large strokes. Hydraulic actuators could cope with these,
but not magnetostrictive actuators.
Three governing loads have been identified as acting on the main stay cables:
a) static loads, mainly dead loads that could be considered as constant values versus time,
b) quasi-static loads, variable values at low frequency, mainly thermal action or wind average value,
c) and the dynamic loads, due to vibration behaviour.
In order to eliminate the need of high forces induced by active tendon control on the main stay cables,
an alternative solution was proposed based on the use of active tendon control on additional cable ties.
This solution will drastically reduce the required static load, but would perhaps increase the actuator
stroke. From a structural engineering point of view, the initial solution will both mitigate the induced
cable vibrations and increase the deck structural damping while the alternative solution would only
mitigate the cable vibrations. Both solutions have their field of interest in bridge design; the initial
becomes very promising for long span bridges where the need of structural damping is tremendous.
As explained previously, the use of active control to mitigate transverse cable vibrations is not an
economic solution in civil engineering. The aim of using the alternative solution in the project is to be able
to test the magnetostrictive actuators with the same mock-up as the hydraulic actuators. The ambition is
to demonstrate that the proposed magnetostrictive actuator is able to damp a substructure within
industrial constraints. This also allows the use of only one mock-up, but consequently as large and
sophisticated as possible, only limited by the capabilities of the test laboratory. At the end of the tests,
having more information about the real behaviour of the system and better specifications, an application
field has to be found. The proposed magnetostrictive devices will be the biggest units ever built in
Europe. If the estimated stroke is lower than expected (and by security, the specifications have been
pessimistic), applications where static loads are low could be a typical application field.
Cables have to cope mainly with static forces and partly with quasi-static and dynamic loads. The active
control strategy aims at the dynamic loads; the maximum of dynamic load should not exceed one tenth
of the static load. Civil engineers have many difficulties to obtain specific dynamic values needed for
active control requirements, the dynamic approach being very poor from this point of view regarding
classical structure design. Nevertheless, part of real structures composed of cables with boundary
conditions simulating parametric excitations have shown that the dynamic value is lower than one tenth
of the static load. The main experience came from the laboratory experiments carried out by Universit
Libre de Bruxelles. The results were also extrapolated to the large-scale mock-up to be tested.
Initial solutions for the bridge mock-up, i.e. active tendon on the main stay cables:
stroke : 5/10 cm
static force : 150 kN
frequency : 5 Hz
dynamic force : 1/10 of static force
maximum velocity : 30 cm/s
Alternative solution for the bridge mock-up, i.e. active tendon on the tie cable:
stroke : 5 cm
static force : 5 kN
frequency : 5 Hz
maximum velocity : 30 cm/s
With respect to the high share of static loads, a separate passive device should cover this in order to
minimise the power consumption of the actuator system. It was decided to realise a hydraulic
compensation of static load because the force level can be adapted easily to quasi-static changes.
The pressure chambers of the static and the dynamic actuator part can be integrated into one unit.
Active Control Strategy and Actuator Design

The Integral Force Feedback has been clearly identified as a suitable solution for the control strategy of
a cable structure. In addition to a guaranteed stability and a good active damping ratio, this technique
offers the big advantage to require neither a precise definition of the phenomenon inducing the vibrations
nor a precise definition of the structure subjected to the vibrations. The whole information is only
extracted from a good measurement of the tension. This remark is very important. Generally all the
passive damping devices are tuned on theoretical simulation results, which can be partly far from the real
world, and fit only with previous predefined scenarios. In some cases, a part of the filtered energy is
transferred onto other vibration modes (spillover). Finally, these devices do not take into account the
ageing process of the structure components, and consequently the variation in time of the structure
behaviour.
More information could be find in a paper submitted to this conference by the consortium involved in the
project.
Mathematical models of the actuators and local control strategies have been developed with the view to
be incorporated into the structural dynamics code for the complete model of the control system. Control
algorithms have been developed and tested against the computer model and calibrated from
experimental laboratory results.
The demonstrator
The design of the demonstrator has followed several objectives:
To have a mock-up as large as possible: the length about 30 m is only limited by the allowable
space in the test laboratory,
To use industrial cables and components: strand .5 and the smallest industrial anchorage,
To be able to create the worst cases i.e. accordance between cables and structure frequencies (in
some cases, the first torsion frequency very close to the first bending frequency very close to the
first frequency of the longest cable),
To obtain natural frequencies as low as possible in order to be as close as possible of the
frequencies of large real structures,
To obtain during the tests large amplitude vibrations as well for the cables as for the structure.
The deck, about 30 meters long (which is certainly the largest mock-up equipped with active tendons
ever built), is mainly composed of two HEB500 beams whose axis are 3.00 meters spaced. These two
H-beams are transversely linked each 3.50 meters by UAP150 beams, welded on the upper and lower
flanges of the main beams, in order to bring to the whole structure enough transverse stiffness to avoid
all buckling risks and to increase torsion stiffness. A secondary transverse bracing has been added to
increase the transverse stiffness. Each H-beam is fixed on a stiff vertical wall in the laboratory, generally
used as a Reaction Wall for PsD analysis. At the free end of the deck, the two main H-beams are
connected transversely by a HEB400 beam (instead of two UAP 150 beams), which can be used as
support for the excitation source.
Four pairs of parallel stay cables support the deck. Along the longitudinal axis of the mock-up, the
spacing between consecutive anchor heads is 7 meters. Transversely, two stay-cables of a same pair
are 2 meters spaced. Each stay-cable is composed of one .5 strand, with a slope of 1/3 (vertical
versus horizontal). This slope is very close to that of the longest stay cable of modern cable stayed
bridges. The upper anchor head (dead end) is linked on the Reaction Wall. The lower anchor head is
fixed on a strut composed of two UAP150 welded between the web plates of the main H-beams and
allowing enough space (80 mm) for the strand. Each lower anchor head (stressing end) is equipped with
a ring nut allowing a precise adjustment of the length of the stay cable. Subject to assumption that the
initial stressing is carried out with a good precision, an adjustment capability of 50 mm would be more
than enough. The two hydraulic actuators are located behind the lower anchor heads, on the first or the
second pairs of the longest cables.
The first natural frequency of the two longest pairs of cables is higher than 4 Hz if the cables are not
overloaded. In order to obtain a frequency of about 1.2 Hz for the longest cable, it is necessary to
increase the distributed mass of this cable. A computation exhibits a value close to 15 kg/m. The
requested mass will be obtained by fixing additional lead masses regularly spaced on the cables (7kg
per mass). It turns out that, with this dead load weight, the sag of the longest cables will just be lower
than the hundredth of their length. This relative sag value is close to that obtained with the longest stay
cables of Pont de Normandie, before the erection of the secondary tie cables. This was not the initial
objective, but only the consequence of the need to obtain a defined frequency.
It can be frequently read in the literature that the most dangerous parametric excitation for cables is
obtained for an excitation frequency 2f1, where f1 is the first in-plane natural frequency of the cable. In
fact, a detailed analysis shows that the area of instability correspond to frequencies of parametric
excitation close to 2f1/k, where k is any positive integer. Our computations show that, when the static
sag is sufficient the hundredth of the length is large enough oscillation amplitudes at least as large
with an excitation frequency f1 than 2f1 can be obtained. Moreover, the response to an excitation
frequency f1 is very quick while the response to an excitation frequency 2f1 is long to take place.
Figure 2 Demonstrator sketch

Let us call H1 and H2 the shortest cables and H3 and H4 the longest ones. H1 and H2 do not need
additional masses. The mass of the cable is close to 13 kg/m for H3 and close to 15 kg/m for H4. Then
the first three natural frequencies of the structure are: 1.17 Hz for the first vertical bending mode, 1.27
Hz for the first torsion mode and 3.00 Hz for the first transverse mode. The structure has been stiffened
versus the transverse effects.
Taking into account the discriminating tuning devices planned for the anchor heads of H3 and H4, the
following situations can be obtained:
1. Regarding the test with a vertical excitation frequency of 1.17 Hz: the first in-plane natural
frequency equal to 1.17 Hz (k=2) for H4 and equal to 1.75 Hz (k=3) for H3.
2. Regarding the test with a torsion excitation frequency of 1.27 Hz: a first in-plane natural
frequency equal to 1.27 Hz (k=2) for H4. A good coupling between the in-plane and out-ofplane frequencies for H4 could be obtained. Each point of H4 will then describe an ellipse.
Positioning an intermediate support under the deck, between H1 and H2, the first natural frequency of
the whole structure can be doubled. Then a parametric excitation at the frequency 2f1 (k=1) for H4 can
be obtained. Selecting the appropriate position for the intermediate support, one can finally tune the first
vertical natural frequency of the structure in a manner it corresponds to any excitation frequency of the
cables.
Finally as it has been designed, the mock-up offers us a large scope of experiments. More than a scaled
bridge, the mock-up has to be seen as a demonstrator allowing the test of the efficiency of the active
control system in the worst conditions that can be faced by real structures.
The tests on the demonstrator will start on spring 1999. The main objectives are to validate the active
control system on a structure of a larger scale than the initial small-scaled laboratory mock-up and with
industrial components. Then it could be possible to extrapolate to real structures.
More information could be find in a paper submitted to this conference by the consortium involved in the
project.
Numerical analysis
Cable-supported structures become more and more large and flexible. The convenient assumption of a
linear behaviour, as well regarding static or dynamic analysis, becomes less and less acceptable. A linear
elasticity assumption remains applicable to the constitutive materials of the structures, but the
displacements have to be taken into account in the equations of equilibrium, including geometric nonlinearity. Computations are then carried out using iterative methods which, although they present no
theoretical difficulties, are however computation time consuming because the structure is assembled at
each iteration (or at each time step in dynamic analysis). It is therefore very worthwhile to have the use
of finite elements of high accuracy allowing computations with the minimum number of nodes compatible
with the requested accuracy.
Bouygues has developed a cable finite element model of high accuracy. For all static structural analyses,
it is convenient to use only one single innovative cable finite element, whatever its length. A very specific
and unusual method has been developed for dynamic analyses. The accuracy is so good that to obtain
the first k th in-plane or out-of-plane natural frequencies with an accuracy better than 0.3 percent, only
2k innovative finite elements are required to model the cable, which gives a saving higher than six times
that obtained via modelling using classical truss elements. Regarding a step by step integration dynamic
analysis of a cable subjected to support excitation, for instance a parametric excitation, a reduction by
six in the number of finite elements leads to a reduction in computation time by a factor higher than 100.
The objective is to validate this numerical tool with the experimental tests in order to have a predictive
tool for the design of real structures.
Preliminary Results
Numerical simulations carried out on the demonstrator with only one pair of cables equipped with
actuators exhibit an equivalent viscous damping ratio between 15 to 25 percent of critical damping
regarding the first vertical bending and torsion modes of the structure. The stroke of the actuators is
about some ten thousandth of the cable length. If the experiment confirms this preliminary result, the
number of actuators needed for a long span cable-stayed bridge in order to obtain enough damping
(around 5 to 8 percent) will be low versus the total number of stay-cables.
Conclusion
The research on Active Tendon Actuators for Cable-Stayed Bridge nearly reached its mid-way. Even if
there is a lot of work and testing to be completed, the preliminary results are very promising.
There are several interesting aspects that have put in evidence and will be deeper investigated.
Particularly important is the capability to increase the structural damping which is quite mandatory for
future long span bridges.
References
[1].
Preumont A., Helduser S., Foersterling H., Bartlett P., 1999, Active Tendon control of CableStayed Bridges: Control Strategy and Actuator Design, IABSE conference on Cable-Stayed
Bridges Past, Present and Future, Malm
[2].
Magonette G., Bournand Y., Hansvold C., Jenner A., 1999, Experimental Analysis on a Large
Scaled Cable-Stayed Mock-Up, IABSE conference on Cable-Stayed Bridges Past,
Present and Future, Malm
[3].
Preumont A., Achkire Y., 1996, Active tendon control of cable-stayed bridges, Earthquake
Engineering and Structural Dynamics
[4].
Preumont A., Achkire Y., 1996, Active damping of cable structures, ESA Conf. on
Spacecraft Structures and Materials, Noordwijk, 27-29 march 1996.
Stay Adjustment: From Design Perspective to On Site Practice
Michel MARCHETTI
Managing Director
Formule Informatique
Paris, France
Benoit LECINQ
Project Manager
SETRA
Paris, France
Born 1950, graduate from

Ecole Polytechnique and
Ecole Nationale des Ponts
et Chausses, Paris
1.
Born 1970, graduate from

Ecole Polytechnique,
Paris,
Ingnieur des Ponts et
Chausses
Introduction
Stay adjustment is a major topic in cable-stayed bridge construction. As a matter of fact, this
issue, which directly controls the stress distribution in the structure as well as the final geometry,
concerns both analyses during detailed design and tensioning procedures during erection on site.
More precisely, from the designer's perspective, adjusting stays consists in finding the suitable
sequence of tensioning operations, so that stresses in stays and structure remain allowable during
construction stages and during operations. For the site engineer, stay adjustment refers to the
actions to be carried out during the construction process, in order to build a structure, which
fulfills the theoretical requirements specified by the design.
Experience shows that there exist a great variety of approaches to characterize and determine
stay adjustment at the design level and to perform it on site. The purpose of this paper is to revisit the subject of stay adjustment, both from a theoretical and a practical perspective. We
introduce a few innovative concepts, which enable to tackle this problem efficiently with
specialized computer software, while taking into account technological constraints.
2.
Stay adjustment from designer's perspective
2.1. Objectives
The basic design of a cable-stayed bridge deals with the general layout (height and position of
the pylons, other piers, etc.) and yields a tentative cross-section of deck and pylons. Thereafter,
detailed design firstly defines the arrangement and sectional areas of the cable stays, then shall
find out when and how each stay must be stressed.
The sequence of tensioning operations must fulfill several targets:
Practicability of stay installation and construction simplicity. From this standpoint, the ideal
solution would be to stress each stay in a single step, at the time when it is installed.
Structural design objectives (acceptable stresses in structure and stays during erection phases
and after completion). Ideally, re-adjusting stays at different stages would make it possible to
reduce bending stresses in the structure almost to zero, as erection proceeds.
In many cases, the trade-off between these opposite objectives is to tension stays in two steps:
firstly at installation, then just prior to finishing completion. In some projects, a global retensioning operation is required after a few years of bridge operation, in order to compensate for
creep effects.
2.2. Final stage stay tensions
Usually, the search for an adjustment solution starts from the final erection stage, just after
bridge completion. Several solutions have been proposed in the literature, some of them
involving sophisticated numerical analysis techniques, such as multicriterion optimization.
These methods are beyond the scope of this article, but we simply give a hint of a possible one.
As many cable-stayed bridges designed nowadays have a very slender deck, stay tensions are
roughly determined by the so-called pendulum method, i.e. stay tensions balance the deck dead
weight. This rule yields a tentative stay cross-sectional area. Then, the search for an optimal
adjustment of the stay system can be achieved by using an iterative process that considers pylon
and deck deflections, rather than stay tensions:
1. Stays being introduced without any pre-loading, the completed bridge is loaded with its
permanent loads. A significant vertical deflection will happen.
2. Thereafter, stay lengths must be shortened, in order to correct vertical deflection and to
recover the target profile of the deck.
2.3. Search for an adjustment solution
Once the set of tensions at bridge completion has been selected, the tensioning sequence must be
defined. Usually a stage by stage structural software is used to simulate bridge construction. In
the framework of such a program, the installation of a stay is idealized by activating a new bar
or cable finite element. For the design engineer, adjusting the stay at this stage consists in
imposing an appropriate initial condition to the related element.
According to his habits and to the software used, the initial condition modeling stay installation
may be expressed, either by imposing the value T of the tension that must exist in the element
l
of the distortion applied to the element or
just after assembly, or by imposing the value =
l
to assembly. In the same way, retensioning of an already installed stay
may be characterized, either by
B
B
increasing the stay tension with an
increment T to the current tension, or
T
by shortening the current length with an
additional strain .
o
In practice, the search for a solution is

carried out using a trial and error
b) Using imposed distorsion
a) Using tension value T
method: a tensioning sequence of one,
after assemby
prior to assembly
two or more tensioning steps is
Fig. 1 : stay adjustment definition
proposed. Then, an analysis of all the
construction stages is performed in order to verify if stresses remain allowable in the structure.
The tensioning forces to be applied and the number of steps are refined, until a satisfactory
solution is reached.
2
2.4. The dismantling procedure

The previous task requires computing current stay tensions at any erection stage. A very
popular approach often used in practice is the dismantling procedure. The basic idea is very
simple: starting from a well-defined condition - e.g. the completed bridge under permanent
loading - the sequence of elementary operations corresponding to the actual construction stages
is performed step by step backwards. This reverse analysis enables for example to determine the
value of the tension T to be applied to a newly installed stay, in order to reach the desired state at
completion.
However popular, stay adjustment using the dismantling method remains a tedious task, as it
requires the reverse description and analysis of the complete erection sequence, from finished
deck with equipment back to the dismantling of the first segment. Moreover, it is important to
notice that the dismantling procedure cannot take into account time dependent effects such as
creep or shrinkage, as the corresponding models implemented in structural software generally
only consider construction sequences by increasing time.
Hence, a simpler and more efficient procedure to determine the suitable values of the adjustment
parameters shall be used. Before tackling this issue, we will firstly review a few possible means
used to characterize stay adjustment.
3.
Parameters for stay adjustment characterization
3.1. Stay adjustment defined by permanent tension values

In theory, the structural state of the
bridge under permanent loading could
be entirely defined by the set of stay
T1
tensions, as it is possible to derive from
this data internal forces in pylons and
T i (N,V,M)
R1
deck, as well as the related deflection
T1
R2
profiles (Fig. 2). Unfortunately, due to
Ti
accuracy issues, this approach cannot be
used in practice to characterize stay
adjustment at permanent stage.
(N,V,M)
Fig. 2 : forces in deck and pylon deduced from stay

An illustration is given by the analysis
of re-tensioning operations to be carried tension values
out on the Vasco de Gama Bridge in Lisbon. The computations have shown that the deck
central span could be lifted by more than 0.60 meters without inducing a change of more than
3% in the permanent tension values. This result may be surprising, but it is clearly explained by
the fact that stay tensions must balance under any circumstance the deck dead weight, due to the
extreme flexibility of deck over the central span length.
We shall recall that the accuracy of well calibrated jacks or tensioning cells only reaches 2 %.
Hence, a practical consequence is that the actual state of a bridge cannot be experimentally
evaluated by measuring the stay tensions only. Therefore, a geometrical monitoring of the
bridge is required to reliably keep track of the structure evolution with time.
3.2. Stay adjustment defined by stressing force

In this section, a stay is implicitly assimilated to a cable installed and stressed with a force T in a
single step. From this standpoint, the installation of a stay may be split into two elementary
actions:
firstly, application to the anchorages
of two equal forces T with opposite
B
B
T
T
directions,
(EA)
then, assembly to the structure of a
T
T
pin-jointed structural element
displaying an initial tension T, and
A
A
an axial stiffness EA.
a) Forces T applied
b) Assembly of a new element
Such a description is especially suited

to anchorages
with initial tension T
to a conventional linear stage by stage
analysis, in which internal forces are
determined by summing up the effects Fig. 3 : model of a stay as a finite element
of each elementary action occurring from the beginning of the construction.
However, the use of the stressing force T as a parameter to represent stay adjustment suffers a
severe drawback: the stressing force is not an intrinsic characterization of stay pre-loading.
Indeed, stressing a stay with the same force T in presence or in absence of temporary loads, such
as a mobile crane, will not yield the same tension in the stay at bridge completion. This gives
rise to tedious constraints in current practice. For example, any modification in carriage weight
during design requires resuming the computation of all the stressing forces, to reach the same
state at the end of construction.
3.3. Stay adjustment defined by unstressed length l0
The previous two paragraphs have shown that stay adjustment must be characterized by an
intrinsic parameter, i.e. a parameter that is independent on the construction sequence and on
temporary loads. Moreover, a good analysis method should enable one to determine the forces
in each section without having to sum up the effects of each elementary action occurring during
the construction history.
Stay adjustment can be achieved by assembling a cable element with a proper initial length l0
smaller than the distance l between anchorages in the reference geometry described in the
model. In the literature, l0 is referred as the cable unstressed length (or cable neutral length).
The unstressed length of a stay - which can be seen as the length of the cable when it is lain on a
flat support, without any tension - is clearly an intrinsic parameter.
3.4. Stay adjustment defined by a pre-deformation
The value of the unstressed cable length l0 has a valuable meaning only when compared to the
distance l between anchorage nodes in the model theoretical geometry. Therefore, the
dimensionless parameter , referred to as pre-deformation, which is nothing else than a strain
pre-loading, is often preferred:
l l l 0
=
=
l
l
In most cable-stayed bridges, the value of this parameter related to a completed bridge under
permanent loading, ranges between 2.510-3 and 3.510-3.
The unstressed length l0 and are intrinsic parameters. If the mobile carriage weight changes
during the design, these values, that describe the adopted stay adjustment, are not impacted. Of
course, there remains to check that the new weight values still produces allowable stresses.
3.5. Stay adjustment defined by reference tension
The reference tension notion was introduced in order to dispose of a common means for
characterizing stay adjustment both from designer and contractor perspectives. As l0 and , the
reference tension constitutes an intrinsic parameter, which in addition is extensive.
The reference tension Tref of a stay at temperature ref is defined as the tension that would exist
in the stay, if the structure deformations were frozen, i.e. if the displacements of the anchorages
were prevented. We shall see later that, with this definition, the actual stay tension value T may
be seen as the reference tension value modified by the effect of anchorage displacements and
temperature variation.
If linear behavior is assumed, the reference tension value is directly derived from the predeformation value by multiplying it by the axial stiffness EA:
Tref = EA
4.
Adjustment instructions for on site operations: current practice
4.1. Link between designer and site engineer: use of stressing force
Most often, the adjustment instructions supplied by the designer to the site engineer are
expressed in terms of stressing force to be applied to stay anchorage at given construction
stages. If the complete stay is placed in one single stage before being stressed, the theoretical
force value can be used straightforwardly, provided that the actual site conditions are very
similar to those taken into account in the design.
4.2. Limitations due to actual site conditions
However, such an ideal situation is seldom met in practice, because structure and stay
temperatures, as well as site loads, always differ from those taken into account during the
design. A current practice consists in determining the tensioning force value by updating the
computational model with values measured on site just before stressing operations. However,
this method is not very flexible, as it imposes the stringent timing constraints of the construction
site on the design office that runs the computational model.
4.3. Case of a flexible deck: use of vertical deflection
If the deck is very flexible, a slight error T in the tensioning force T produces a significant
parasitic vertical displacement w of the deck. In such a situation stays are adjusted by
controlling the altitude of the deck, rather than the stressing force.
However, the distinguishing between a flexible deck and a rigid one is often uneasy. In most
cases, the deck segments close to the pylon constitute a stiff structure, and the deck becomes
more flexible when cantilever length increases.
4.4. Use of unstressed cable length

An alternative to using stressing force or vertical deflection would be to prepare the cable stays
with their exact unstressed length, and to extend them to attach their two extremities to the
structure.
This method is often used for

prefabricated stays, installed in one
stage only, but also applies to stays
erected by threading strands one by
one, provided each strand is cut or
marked at the right length. However,
two accuracy issues must temper this
optimistic judgment:
f
re
Theoretically at least, basing stay

adjustment operations on unstressed
cable length has several advantages:
actual loading conditions
(temperature, site loads) do not
impact anymore the adjustment
values,
the same procedure applies both for
rigid and flexible decks.
Reference geometry
Structural elements to assembly
( with initial deformation )
Fig. 4 : stay adjustment using unstressed length l0
4.4.1. Accuracy on the cable length itself

If the anchorages were perfectly located at their theoretical positions on deck and pylon, the
accuracy of the adjustment would only depend upon the precision attached to the unstressed
cable length l0. Experience shows that the best accuracy associated with the measurement of the
distance between two marks drawn on a cable is of the order of 10 mm per 100 meters.
Such an uncertainty in length measurement is only a relative error of 0.01 / 100 = 10-4 which
may seem fairly good. However, in cable-stayed bridges, permanent strain usually ranges
between 2.510-3 and 3.510-3. Hence, the elongation of a 100 meter long stay is in the order of
0.3 meters, and the relative error on the actual pre-deformation , i.e. the precision of stay
tension adjustment, is not better than 0.01 / 0.3 = 3.3 %.
4.4.2. Accuracy on the anchorage position
The accuracy related to anchorage location has a similar effect on adjustment precision. For
concrete decks cast in situ, the anchorage position is generally set by the mobile carriage.
Hence, anchorages are generally located with a tolerance reaching several centimeters. In such
cases, using unstressed cable length l0 as the adjustment parameter would result in poor quality
(about 10 to 20%).
Ti
Ti 1
As a consequence, inaccuracy on the

unstressed length of a stay (i) gives rise to
significant variations about nominal values
in the stay tension Ti itself, but also in the
neighboring stay tensions Ti-1 and Ti+1, as
well as bending stresses in the deck.
Ti
Ti 1
Ti
4.4.3. Impact on tension distribution

We have said earlier that cable-stayed
decks are very flexible beams when
considered over the central span length.
This no more the case when a short block
of deck or pylon is observed locally.
Ti
Ti 1
Fig. 5 : stay tensions differ locally, due to deck

When anchorage location cannot be
bending stiffness
controlled with a precision of some
millimeters, unstressed cable length l0 does not constitute a suitable adjustment parameter for
practical purposes. Other methods must be derived, in order to cope more efficiently with the
unavoidable construction tolerances. However, the unstressed length l0 remains a valuable
parameter to crosscheck the adjustment data supplied by other methods.
5.
Site procedure organization
5.1. Global scheme

Stay adjustment is only one topic in the field of the survey procedures carried out during the
construction of a cable-stayed bridge. Before addressing the details of a proposed stay
adjustment method, we shall present the general layout of these procedures. First of all,
adjustment and verification procedures must be clearly distinguished.
SITE PROCEDURES
Adjustement
Stay Pre-loading
Verification
Geometry
Pylon
Deck
5.1.1. Adjustment procedures

Adjustment procedures are tightly connected to the erection operations on site. Their purpose is
to provide the construction team with precise instructions for site operations, in order to build a
structure that complies with design specifications. These procedures split into stay installation
adjustment, on the one hand, and geometrical setting of deck and pylons, which must be built
with the proper shape, on the other hand.
5.1.2. Verification procedures

Verification procedures are meant to check that the actual structural behavior of the bridge, as
erection proceeds, is in accordance with the predictions of the computational model. Therefore,
survey and tension measurements are performed at times corresponding to design stages and the
results are compared with their theoretical counterparts.
5.2. Control of deck geometry
The procedures of geometrical setting of deck and pylon are beyond the scope of this paper. We
shall merely recall that:
During cantilever erection, prefabricated segments or, in the case of a cast in situ deck, the
mobile carriage, are positioned relatively to the part already built, rather than at an absolute
altitude given by the model. This positioning in relative terms makes it possible to obtain an
accurate profile despite various parasitic effects.
Furthermore, this positioning cannot be based on the actual concrete surface, which is
subjected to construction tolerances in the order of centimeters. Consequently, one must
consider an ideal theoretical profile line attached to the deck.
Let i be the vertical offset between the
theoretical profile line attached to the
deck and a given benchmark (i) laid on
the actual surface. The value i is a
constant that must be determineded just
after segment casting, using an
appropriate procedure. Knowing these
offsets , the actual deck deflection
values w at all the benchmarks can be
derived from simple altitude
measurements, with a very good
precision.
6.
Reference line
Z ref
Zc
Theoretical profile
Actual deck surface
Fig. 6 : vertical distance between theoretical deck

profile and actual deck surface: w = Zref - Zc -
Stay adjustment procedure using the reference tension concept
6.1. Relating current tension to reference tension

The reference tension concept is useful, as it is very easy to relate the current tension in a stay to
the reference tension. If we assume that the stay behaves linearly, i.e. that catenary effect can be
neglected, then the behavior of the stay, considered as a sub-system of the complete structure, is
clearly given by the following formula:
T = Tref +
EA &
& &
( u B u A ). i EA
l
with the notation indicated on figure 7. The first term represents the effect of stay adjustment,
the second is the effect of anchorage displacements and the last is the effect of stay temperature
variation from ref.
This formula is a very simple way to compute the current tension T at any construction stage,
knowing anchorage displacements and stay temperature. In fact, considering the stay as a subsystem replaces the computation of the complete structure by simple survey and temperature
measurements.
EA &
& &
( u B u A ). i + EA = Tref
l
EA
()
uB
6.2. Is the reference tension really intrinsic?

As already said, the reference tension is, by definition,
an intrinsic parameter to characterize stay adjustment
(pre-loading) and it remains constant as long as no
retention of the stay is performed. Conversely, the
above formula can be reverted:
uA
and it is interesting to check "experimentally" on a

A
computational model that the first member is actually
constant throughout construction stages, provided that
Fig. 7 : stay cable considered as a
no re-tension operation is performed on the considered
sub-system of the complete structure
&
stay. Knowing T at any stage, the displacements u A
&
and u B , and , one can determine the value of the reference tension Tref.
6.3. Application to stay tension adjustment
For sake of presentation simplicity, a linear behavior is again assumed in this section. The
vertical deflection of deck benchmarks can be determined from altitude measurements, using the
procedure evoked in section 5.2. In the same way, a suitable monitoring of pylon enables to
keep track of its horizontal deflection. This way, anchorage displacements about their position
in reference geometry can be evaluated.
If stay tension as well as stay temperature are measured on site, the actual value of the reference
tension Tref,temp can be easily determined by formula given in section 6.2 and cross-checking is
possible. Conversely, to reach the reference tension value Tref specified by design, the
elongation l to be applied to the end of the cable is derived as follows:
l =
l
T Tref ,temp
EA ref
6.4. From principle to operational procedure

In practice integrated computer software is used to assist the site engineer in each stay
installation and re-tensioning operation. Such programs shall be tailored on a case by case basis
to each project and erection sequence, but they all rely on the simple reference tension concept
presented before, and no computation of the complete structure is necessary for adjustment
purposes.
Second order effects (catenary) are generally not negligible during the stay installation process.
Therefore, a more sophisticated model based on a non-linear behavior of each stay must be used
in practice for evaluating data related to adjustment operations. But the principle of stay
adjustment operations remains the same.
This method can be used for adjusting prefabricated stays placed in a single stage, as well as
stays installed by threading strands one by one. In the latter case, the method is particularly
adapted to the use of iso-tension technique. Then, the procedure comprises 5 operations:
1. evaluate by survey the anchorage displacements and compute the tension t0 to apply to the
first strand in order to impose on the stay a given fraction of the target reference tension, for
example 80 % of Tref,
2. insert the first strand and stress it to the tension t0,
9
3. insert all the other strands and stress them by comparison with the first strand, using isotension technique,
4. in a further step, measure the actual tension of the complete stay, and the related values of the
anchorage displacements ; then deduce the elongation l to apply to the stay to reach the final
target value Tref,
5. perform a check by evaluating the actual value of the reference tension through simultaneous
measurements of stay tension and anchorage displacements.
Using a procedure based on the reference tension concept dramatically improves the flexibility
and reliability of stay adjustment operations. Indeed, discrepancies between the actual site
conditions (site loads, temperature) and their design counterparts have no influence on the final
result. Adjustment accuracy is outstanding, as the method is not disturbed by tolerances on
anchorage location. Finally, the method preserves the flexibility of site operations timing, as the
integrated software used for this procedure is parallel and independent of the complete
computational model, and because survey measurement periods, during which no site loading
variation can occur, is very limited.
7.
Conclusion
The reference tension concept results from continuous progresses achieved by many engineers
during the last decade to improve the way of characterizing stay adjustment in design models, as
well as to simplify the adjustment operations on site.
The method has been successfully used on major construction projects, such as the Elorn Bridge
(France), the Second Severn Crossing (UK), or the Vasco de Gama Bridge (Portugal), thus
clearly demonstrating the pertinence of adjustment procedures based on the reference tension
concept.
8.
Bibliography
[1] AUBLANC P., AUGUSTIN V., BERTOCCHI C., DUFRESNE J.L., ENSELME J.,
MARCHETTI M., PLACIDI M., REDOULEZ P., REGALLET D., VASSORD J., Le nouveau
pont sur l'Elorn Brest. Revue TRAVAUX, n 699, June 1994.
[2] COMBAULT Jacques, HERVET Jean, VESVAL Vincent, Le second franchissement de
l'estuaire de la Severn, Revue TRAVAUX, n 719, April 1996.
[3] GIMSING N. J., Cable Supported Bridges, Concept and Design, 1994, John Wiley & Sons.
[4] MARCHETTI Michel, Dominique REGALET, Charles BERTOCCHI, Grard HOCHET,
Elorn Bridge Stay and Geometry Adjustment, Proc. Of the Int. Conference AIPC - FIP,
Deauville, October 1994.
[5] VIRLOGEUX Michel, Erection of Cable-Stayed Bridges, The Control of the Desired
Geometry, Proc. Of the Int. Conference AIPC - FIP, Deauville, October 1994.
10
Damping Characteristics of Carbon Fiber Composite Cables for Application

in Cable-Stayed Bridges
H.M.EL KADY
Associate Researcher
National Research Center
Egypt
M. AROCKIASAMY
Professor
Florida Atlantic Univesity
USA
S. SAMAAN
Professor
Cairo University
Egypt
Y. BAHIE-ELDEEN
ProfessorAssociate
Cairo University
Egypt
M. M. BAKHOUM
ProfessorAssociate
Cairo University
Egypt
M.A.EL GAMMAL
Professor
National Research Center
Egypt
Summary
The unique properties of fiber composites such as high strength, light weight, flexibility,
corrosion resistance, etc. make carbon FRP as an attractive alternative to conventional steel for
application in many structures, including Cable Stayed Bridges (CSB). The paper presents the
results of experimental and analytical studies on the loss factor of CFRP and prestressing steel
tendons when subjected to out of-plane loading. Analytical approach is used to verify the
experimental results, and extrapolated functions are given for different cable lengths. Finally, the
damping ratio is obtained for a CSB cable including the initial prestress effect, and is expressed
as a function of the loss factor.
Introduction
High performance fiber-reinforced polymers (FRPs) have proved as outstanding engineering
materials over the past few decades. Their cost has, however, limited the applications to the
areas of aeronautics/astronautics. A rapidly changing market with decrease in fiber prices makes
the FRP as an viable alternative to conventional structural steel used in civil engineering
applications. Several structures in Germany are experimentally equipped with glass tendons as a
replacement for steel post-tensioning cables. The combination of desirable mechanical and
electrical properties make aramid guy wires an excellent choice for the staying of tall antennas.
Also, the high resistance to chemicals makes FRPs clearly a superior choice for structures
subjected to aggressive environments.
The innovative use of fiber-reinforced polymer composites can lead to significant economy in
the design of massive structures, since the higher specific strength (strength/mass density) of
FRPs reduces the self-weight of structures; this becomes quite significant with increasing
dimensions such as cable stayed bridges. Cable stayed bridges with main spans about 2000 m are
at present beyond the limit of feasibility for steel. An ideal application of FRPs is in cables for
long span cable-stayed bridges ( Kim and Meier, 1991).
Little information is known on the response characteristics of CFRP cables subjected to dynamic
loads. Further research on the dynamic properties of CFRP is required for application in CSB, as
these CSB cables are usually subjected to wind induced vibrations.The paper presents the results
of experimental and analytical studies on the loss factor of CFRP and prestressing steel tendons
when subjected to out of-plane loading. CFRP tendons with different lengths and diameters were
subjected to forced vibrations using a hydraulic shaker and white noise (up to 20Khz) on the
double cantilever system. Accelerations were measured using "piezoelectric accelerometers" and
the data obtained through a Fast Fourier Transform (FFT) analyzer.
Analysis is performed using both the half power band width, and the resonance dwell technique
to obtain the loss factor of the specimens at different frequencies, the shape function, and the
relation between loss factor and the ratio between acceleration at the cantilever tip to the support
movement The results are presented as graphs expressing the relationship between the loss
factor, length, and the eigenvalues of different mode shapes. Similar tests are being performed
also on steel strands for CSB.
Experimental method
The resonant dwell technique (Gibson, et al. 1982; Gibson and Plunket 1977; Paxson 1975) is
used to determine the dynamic material properties of the CFRP tendons and steel cables. The
technique has proven effective in determining the material damping of a variety of composite
materials. The measured material properties are determined from an energy balance approach
which assumes that the amount of energy supplied due to input acceleration at the base is
subsequently dissipated or damped by the cantilever beam specimen material. Double cantilever
beam specimens are excited with white noise in the frequency range of 0 to 20 kHz.
A schematic of the measurement process and hardware and the details of the experimental setup
are shown in Figs. 1 to 3.
Accelerom eters
Noise
generator
Hydraulic
shaker
Power
am plifiers
Specimen
Power
supply
FFT HP
analyzer
Fig.1 : Schematic of the test apparatus for measurement of response of double cantilever
specimens.
Fig.2: Hydraulic shaker, test specimen, accelerometers, and power supply
Fig. 3. FFT analyzer, noise generator, and power amplifiers

Three different materials were tested in this study: aluminum, prestressing steel and CFRP
tendons. The sectional and material properties are shown in Table 1. Spectrum of resonant
frequencies were obtained for specimens with different lengths.
Material
Aluminum
(T2024)
P.S (Grade 270)
Nominal
Area (mm2)
Diameter (mm)
12.5
126.5
Elastic Modulus
(Mpa)
.9e5
2.056e5
Unit Weight
(N/m)
3.345
12.5
93
CFRP
12.5
49/76*
2.13e5/1.37e5*
1.496
CFRP
9.91
35.6/ 55.7*
2.13e5/1.37e5*
1.092
* Effective / nominal properties
Table 1: Sectional and material properties of specimens.
E(Mpa)
Two control aluminum specimens are tested to check the calibration and accuracy of the
experimental setup, and the equipment. The diameter of both the test specimens was 12.5mm
and two different lengths of 0.5m and 0.4m were considered in the tests. Different mode shapes
are considered and the elastic modulus calculated for both specimens at different frequencies.
Fig.4 shows that the elastic modulus of aluminum specimen is independent of the frequency of
excitation.
1 .0 0 E + 0 5
9 .0 0 E + 0 4
8 .0 0 E + 0 4
7 .0 0 E + 0 4
6 .0 0 E + 0 4
5 .0 0 E + 0 4
4 .0 0 E + 0 4
3 .0 0 E + 0 4
2 .0 0 E + 0 4
1 .0 0 E + 0 4
46
398
1150
2170
3584
F r e q .(H z)
Fig.4: Elastic modulus of aluminum specimen of length 0.4metres at different resonant

frequencies.
CFRP tendons of two different diameters of 12.5mm ,9.9mm and prestressing steel specimens
of 12.5mm diameter are tested with three different lengths of 52 cm, 41 cm, and 31 cm. Modal
frequencies are detected from the spectrum at the point on the tip of the cantilever beam
specimens. Transfer function for the test specimens is recorded and the flexural rigidity (EI)
calculated from different mode shapes for different samples and compared with the values
specified by the manufacturer. The damping factor is calculated for different specimens, at
various frequencies using two different methods: the well known half power band width and
resonance dwell technique Fig.5 shows the loss factor against the parameter bl for 12.5mm
prestressing steel and CFRP specimens with different lengths. The parameter bl is a
dimensionless quantity in which l is the specimen length, and
1
0 ,9
0 ,8
Loss factor
0 ,7
0 ,6
0 ,5
0 ,4
0 ,3
0 ,2
0 ,1
0
0
10
15
20
bl
p .s 5 2 c m
p .s 4 1 c m
p .s 3 1 c m
c fr p 5 2 c m
c fr p 4 1 c m
c fr p 3 1 c m
Fig.5:Loss factor for 12.5mm diameter CFRP and prestressing steel Specimens using half power
band width method.
b4= wn2 m/ EI
where wn = angular resonant frequency of n th mode.
Loss Factor
Fig.6 shows the loss factor at different frequencies for 9.99cm diameter CFRP tendons, with
lengths of 52, and 41 cm.
0,5
0,45
0,4
0,35
0,3
0,25
0,2
0,15
0,1
0,05
0
cfrp 52 cm
cfrp 41 cm
10
15
bl
Fig. 6: Loss factor for .99 cm diameter CFRP Specimens.

Loss factor for longer specimen lengths by extrapolation
The loss factor based on the limited number of tests and specimen lengths can be extrapolated
from Figs.5 and 6 for applications to longer lengths as shown below:
For CFRP tendons:
The following polynomial can be derived:
U=1.548x-1-15.579x-2+52.588x-3-10.294x-4-30.481x-5-27.308x-6-19.479x-7
For prestressing steel tendons:
The polynomial given below can be obtained;
U=1.69x-1-23.21x-2+106.376x-3-58.969x-4-81.984x-5 57.935x-6-34.302x-7
Analytical model
The equation of motion for lateral vibration of the undamped beam is applicable in the present
support motion problem, because the only force acting on any section of the specimen is the
internal elastic shear force associated with bending of the test specimens. A solution using
separation of variables for harmonic motion at a resonant frequency is given by
y(x,t)= Re {A1sinbl+ A2 cosbl +A3 sinhbl +A4 coshbl} eiwrt
Boundary conditions
At x=0:dy/dx(0,t)=0
(a)
md2y/d2t(0,t)=Re[Feiwt]-2V(0,t) (b)
and at x=l :
M(l,t)=0
(c)
V(l,t)=Mad2y/d2t(l,t)
(d)
(1)
Substitution of the boundary conditions (a) to (d) in Eqn.(1) leads to the following frequency
equation:
(m+m cosbl coshbl +bM coshbl sinbl-bM cosbl sinhbl)=0
(2)
where M is the accelerometer's mass, m is the beam mass/unit length.

The loss factor is obtained from the following relationship:
A= atip/asup=y(l,t)/y(0,t)
(3)
and can be derived as

A ={(2(m Cos[b(1-Iu/4) l] +2 M b(1 - I u/4) Cosh[b(1 - I u/4) l] + mCos[2 b(1 - I u/4)l] - 2 b(1 Iu/4) M Cos[b(1-Iu/4) l]Cosh[b(1-Iu/4) l] -mSin[b(1-I u/4) l] +m Cosh[2 b(1-Iu/4) l] Sin[b(1Iu/4) l] +2bMCosh[b(1-Iu/4)l]Sin[2b(1 -I u/4) l] +2 mSinh[b(1- Iu/4) l] -mSin[2 b(1 - Iu/4) l] +
mCos[b(1-I u/4) l] Sinh[2b(1Iu/4) l]}/{(4(m +mCos[b(1-Iu/4) l] Cosh[b(1- Iu/4)l] + b(1 Iu/4)
M Cosh[b(1-Iu/4) l] Sin[b(1 - I u/4)l] -b(1-I u/4) MCos[b(1-Iu/4) l]Sinh[b(1-Iu/4) l] }
(4)
where:
A=tip displacement/base displacement.
u=loss factor.
I=-1^.5
Eqn. 4 is solved numerically for different resonance frequencies to obtain the equivalent loss
factors.
Estimation of damping ratio for steel and CFRP CSB cables including the
effect of initial prestress
Prestressed steel and CFRP cables with different lengths of 465,622,933 meters are considered
having cross sectional area of 0.015 sq.m and are discretized with 100 elements in order to
analyze the effect of initial prestressing forces on the damping ratio, and logarithmic decrement
of one of the a typical individual CSB cable. The steel cables are prestressed with a value of
0.45 times the ultimate stress (1912 Mpa), while CFRP cables are prestressed to 0.3 times the
ultimate stress (2118Mpa).
Applying the energy principle, the total modal potential energy Un can be defined in terms of
two parts: modal strain energy, Vn and potential energy due to initial prestress, Utn. The
following formulae can be derived based on the equations given in Ref.[1]
Un=Vn+Utn
= .5 Vn / (Vn+Utn)
Utn = .5 A (tan2 )s
Where:
A = the cable cross sectional area,
= the initial prestress,
= the element angle of rotation,
s = element length.
6
= damping ratio
Vn = 0.5 E A 2 s
Where
E = Youngs modulus,
= the dynamic axial strain.
= tan2 + 0.5 tan tan
and
is the static angle of the element.
Material
Steel
CFRP
Horizontal
Cable
Sag ratio
Strain energy ratio
Proj.(metres) length
(metres)
330
465
0.0026
0.01
440
622
0.0035
0.02
660
933
0.005
0.05
330
440
660
465
622
933
0.001
0.0015
0.0022
Damping ratio
.005
.01
.025
0.002 .001
0.004 .002
0.01 .005
Table 2: Damping ratio for CFRP and PS cables as a function of loss factor.
Table (2) shows the results obtained from the static and dynamic modal analyses of the cables
with a 45 degree inclination (cross sectional area=0.015 sq.m). The damping ratio in terms of
the loss factor for steel cables is five times that of the CFRP cables. Loss factors for prestress
steel, and CFRP are estimated from curves (Fig.5), and is found to be almost constant for such
long cables with low fundamental frequencies. The loss factor is 0.04, and 0.05 for steel and
CFRP cables respectively, the value for steel is in good agreement to previously published work
[1]. Also the logarithmic decrement for the steel cables under study varied from 0.1% to 0.5%,
which have the same order given in previous literature [1,2].
Results and Discussions

Results obtained from the experiments are in close agreement with the analytical calculations by
both methods: half power bandwidth, and the frequency equation and the difference in results
from the two methods was about 7%. It is worth noticing that the effect of bonding material
between the individual strands is very significant, especially in the low frequency modes. That
leads to obviously higher loss factor at the lowest modes especially for the CFRP strands. The
loss factor tends to remain constant at higher modes for both prestressing and CFRP tendons.
The difference in the loss factor for both materials decreases at higher modes. The loss factor for
the cables considered in this study is almost constant, and the loss factor of CFRP is higher than
prestressing steel by about 25%. Finally, the damping ratio of steel is about four times that of
CFRP.
Potential for future applications

The potentials for successful application of CFRP cables in cable-stayed bridges are great due to
its higher specific strength, and equivalent modulus, excellent corrosion and fatigue resistance.
The CFRP cables are more viable for use in long span CSB due to the above mentioned
characteristics.
Although CFRP seems to have lower damping ratio than that of prestressing steel for very long
cables due to its lighter weight, the over all effect of damping for both materials in long span
cable-stayed bridges is very small, thus the previously mentioned advantages still govern the
superiority of CFRP for long span bridges.
Acknowledgments
The authors wish to thank The Egyptian Government for providing support to the first author
through the Exchange Visiting Scholar Program. They wish to express their appreciation to Dr.
S.E. Dunn, Professor and Chairman, Department of Ocean Engineering and Dr. John Jurewicz,
Dean, College of Engineering, Florida Atlantic University for their continued interest and
encouragement.
References
[1].
[2].
[3].
[4].
[5].
[6].
[7].
[8].
1.H.Yamaguchi, and Y. Fujino, Damping of cables in Cable-stayed bridges with and

without vibration control measures, International conference on Cable stayed and
suspension bridges, Deauville, France, Oct.1994.
2.G. Hirsch, Cable vibration overview, International conference Cable stayed and
suspension bridges, Deauville, France, Oct.1994.
R.F.Gibson, "Damping characteristics of composite materials and structures", Journal of
material engineering and performance, 1992.
R. Greif, B.Herbert, Experimental techniques for dynamic characterization of composite
materials, Advances in experimental mechanics and biometics, ASME 1992.
P. Kim, P. and U. Meier, CFRP cables for large structures, Proc. Advanced Composite
Materials in Civil Engineering Structures, MT Div., ASCE, Las Vegas, Jan. 31, 1991
R.F.Gibson, A.Yau, D.A.Riegner, "An improved forced-vibration technique for
measurement of material damping", Experimental Techniques, vol.6,no.2, April 1982.
R.F.Gibson, R. Plunkett, A forced vibration technique for measurment of material
damping, Experimental mechanics, 11, (8), Aug. 1977, 297-302.
R.F.Gibson, R. Plunkett, Dynamic mechanical behavior of fiber-reinforced composites:
measurement and analysis, Journal of composite materials, 10, Oct. 1976, 325-341.
Development of new stay cable dampers
Yves BOURNAND
Mechanical Engineer
VSL International
France
Summary
Cable vibration on cable stayed bridges is known since several years, and can be considered
today as one of the most critical problems for this type of bridge.
Engineers developed some damping devices that will be reviewed here, with a particular point
on their installation, fatigue and maintenance.
Dampers are submitted to small movements and small loads but with a high number of cycles.
Fatigue and maintenance are perhaps the two important criteria to be considered.
Considering the present experience, engineers analysed the design criteria of a damping
system and developed a new system, the friction damper, that will be installed on the stay
cables of the UDDEVALLA Bridge in Sweden.
Vibration phenomena
Many different phenomena can generate cable vibrations. They are mostly nonlinear, their
analysis is very delicate.
The precise conditions that trigger the vibrations are not fully understood, for example,
conventional statics and dynamics are insufficient to precisely model the behavior of the cable
under the rain vibration phenomenon.
Wind/rain vibration phenomenon on cable stay bridges has only been identified within the
past few years, and can be considered today as the most important cause of cable vibrations.
Many bridges being designed today still dont account for it. Of 28 such bridges in the United
States, at least half have the problem.
Different types of damping systems
Various methods have been tried to mitigate the cable vibrations, such as :
- The installation of mechanical dampers placed near the ends of the stays, at the deck or
tower. In order to damp.
- Cross-cables interconnecting the stay-cables of the bridge. In order to increase cable
frequencies.
- Shapping of the stay pipes. In order to destroy the aerodynamic excitation mechanism.
2.1. Neoprene/rubber damper
It has been customary to install a small neoprene or rubber damper at the end of the steel pipe
where the stay cable passes into the anchorage at the deck.
Figure 1.
The design of this damper is based on expected lateral forces, and to mitigate the transmission
of potentially damaging bending stresses arising from stay vibrations to the cable stay in the
near field region of the anchorages.
In fact, this small, compact damper has an only negligible damping effect. It is not designed
for the high-amplitude, low-frequency oscillations associated with rain-induced vibrations.
2.2
External dampers
At present, the installation of dampers at the cable connection parts on the bridge deck to
increase the structural damping of cables are the most common countermeasure to mitigate
the vibration.
The most classical solution consists in the installation of hydraulic or viscous dampers
connecting the stay cable to the deck, near the lower anchorage.
This installation can be as simple as for Brotonne Bridge, Elorn Bridge or Erasmus Bridge or
it can be more complicated as for Normandy Bridge.
This solution has the advantage to be easily accessible for the maintenance operations, but the
aesthetical appearance could be criticized in some cases.
Some systems are designed so that to be installed without connection to the deck but as a ring
around the cable. The damper is installed in the formwork steel pipe embedded in the deck
concrete or within a steel support pipe extending the formwork pipe, as those that will be
used on UDDEVALLA Bridge.
Generally, the damper is located at a distance
between 0,015 and 0,030 (L) from the lower
theoretical hinge point of the cable. (L) is the
total length of the cable.
The more common types of dampers are :
- The viscous dampers
- The hydraulic dampers
Viscous dampers consist of freely moving plates
in a viscous, silicon-like material, which assures
the dissipation of energy. Many bridges in Japan
have been provided with viscous dampers
because of the low maintenance costs. An
important disadvantage is that the damper
characteristic is strongly depending on
temperature and frequency.
According to recent experiences, its seems that
hydraulic dampers have relatively high
maintenance costs and complex adjustment.
Fig 2. External damper on

Normandie bridge
2.3
For the designer of a damper, the objective is to

produce the desired damping coefficient at the
desired frequencies and temperature. On bridges
it is important to place the damper on a rigid
support and to have a very tight connection (no
play) of the damper to the stay cable.
Cross-cables
These cables are mainly used as a temporary solution during construction or as an interim
solution. Polypropylene ropes proved very effective during the eight months they were in
service on Erasmus Bridge (Netherlands).
Cross-cable can be used to stop or limit cable vibrations after construction of the bridge, by
increasing the natural frequencies of cable above 3 Hertz to be safe.
Before to use this system as a final solution,

the designer has not to forget the problems
observed on some bridges, as for example on
the Fred Hartman Bridge in Texas. On this
bridge, the forces have broken the crosscables that were placed on the stay-cables one
year after their installation in response to
fatigue and fretting. $ 0,6 millions will be
spent to attach new cross-cables, with better
fatigue characteristics, to install shock
absorbers for the stay cable anchors and to
repair broken weld connections. Moreover,
the vibrations have dramatically increased
costs because of the need for inspections,
maintenance and repairs.
On some bridges, the cross-cables are
replaced by stronger ones with higher
tensions.
Fig 3. Cross-cables
If the bridge designer choses cross-cables it is

recommended to follow the following criteria:
The cables must be designed using a flexible
wire rope or similar system (with high
internal damping) and with good fatigue and
wear characteristics.
The initial tension in these cables must be such that to be not detensioned under extreme loads
so that to avoid shocks which could produce some ruptures. Thus this solution is difficult to
apply on existing bridges since, the pre-tension of the cross-cables may change the stay cable
geometry and particularly the angle between deck and cable at the anchorages.
The main reasons cross-cables are not considered as a permanent solution by some bridge
owners is that they have a major impact on the aesthetics of cable stayed bridges and they
present high inspection and maintenance costs.
2.4
Shape of the stay pipes
European and Japanese engineers developed this aerodynamic solution which is used on
several cable-stayed bridges, e.g. for UDDEVALLA Bridge in Sweden. Helical ribs or
protuberances are added to the stay pipe ; this disrupts the flow of water, and destroy the
continuous galloping-promoting rivulet.
Stay pipe of 165 and 250 mm diameters have been tested with these helical ribs in the wind
tunnels of the CSTB (France) and of the Danish Maritime Institute. The tests showed a strong
reduction of the rain-wind induced vibration. However the designer has to be careful because
some vibrations can persist as observed in some worst cases.
Care must be exercised with these solutions so that to keep a small increase of the drag factor,
to limit wind forces on the bridge, and to have a nice aesthetics of the stay cables.
For the designer, the shapping of the stay pipes have been developed to prevent only one,
although the most important, type of vibration (rain and wind induced vibration). If the risk to
have other types of vibrations is high, this solution will have to be combined with other
solutions.
Fig. 4. Stay pipe with helical ribs on Normandie bridge
Design criteria and specifications of a damping system
- The steel elements that constitute the stay cable are generally designed for pure cyclic
tension stresses. But we also have to consider the unintentional bending stresses resulting
from cable vibrations which could significantly reduce their expected fatigue life and
affect integrity, serviceability and durability of the stay cable system.
- It can be very difficult or impossible to inspect for potentially damaging effects within the
near-field region, at the cable-stay anchorages and at the damping devices.
The ability and access facility to inspect the stay cables near the anchorages and the
damping devices will increase the safety and durability.
- In the following years, technologies and equipments will be developed to detect wire
failure. These equipments will be introduced as a part of the usual maintenance equipment
of the bridge.
- For the designer of the damping system the most important criteria to follow could be the
following :
Adjustability
Ease of access
Relative low cost
Easy mounting
Installation on existing bridges

Aesthetics
Maintenance
Reliability
Not temperature dependant
Damping characteristics insensitive to the frequency and the amplitude of vibrations.
- Allowable amplitude for the cable vibration is a very important issue.

Bending angle of 0,5 to 0,6 degree at the cable anchorage can be a criteria of the fatigue
damage. However this 0,5 to 0,6 degree bending angle means quite large amplitude of the
vibration and it will not be the acceptable amplitude for visual uneasiness.
Although often used, 1D or 2D amplitude criteria (D : diameter of the cable) does not
seem to have any concrete reason, actual decision to use this criteria for additional
reduction of the vibration seems acceptable in many cases.
- In the implementation phase, special attention has to be paid to possible stay cable bending
through the damper due to mobile loading of the bridge-deck.
A new development : the friction damper
- We observed on the existing damping systems that the cost of maintenance can be
expensive. The damper can be permanently solicited to small, non-critical vibrations and
very quickly it will have to support a high level of cycles and consequently a rapid
deterioration and replacement.
- To answer to this problem of fatigue and maintenance, D. KOVACS (Dynamik
Consulting) proposed a technical solution : the friction damper. This product has been
developed in collaboration with VSL for a future installation on the UDDEVALLA
Bridge (Sweden).
Fig. 5. Installation of friction damper on UDDEVALLA Bridge

Fig.
The damper will be placed at the top of a rigid steel support pipe bolted to the steel deck
structure supporting the load of the stay cable deck anchorage. The friction damper is
composed of mechanical components assembled together so that to have no play (to keep a
high level of efficiency).
Its composed of two parts. The movable part is tightly fixed to the strands of the stay cable
by a bolted collar and will move in the three directions. The bolted collar has several friction
wings. The fixed part is bolted to the steel support pipe and is composed of several spring ring
blades supporting several friction screws.
The ring blades are deflected so that to have a steady friction contact of the friction screws
against the friction wings whatever the movements of the movable part of the damper.
The main advantages of this friction damper developed for UDDEVALLA bridge are the
following :
Small, non-critical vibration amplitudes remain undamped. Thus, we have reduce wearing
and low maintenance costs.
For UDDEVALLA Bridge, the friction forces will be adjusted so that to have an action of
the damper only when the amplitude of vibration of the longest cable will be beyond 70
mm.
For each cable, the friction force of the damper will be adjusted according to the allowable
amplitude of vibration defined by the designer.
The friction damper is designed so that to be easily installed on existing bridges where
cables are submitted to unexpected vibrations.
All components of the damper are accessible and can be easily inspected and replaced, if
necessary, during the maintenance operations.
The characteristics of the damper can be easily adjusted during the maintenance
operations. This adjustment consists only to turn the four screws supporting the friction
pads.
The friction forces are practically constant and independent from the speed of the point to
be dampened.
The damping characteristics are insensitive to the frequency and amplitude of the
vibrations.
The friction damper is designed so that to have a constant damping of the stay cable when
this cable has a longitudinal movement due to load variations.
Conclusions
Researchers are just beginning to study vibrations on cable-stayed bridges. We observed that
some progresses have been achieved but we have to recognize that today the precise
conditions that trigger the vibrations are not fully understood.
The cables of bridge structures are easily excited by natural wind because of their low
structural damping characteristics, therefore problems such as structural fatigue will arise.
Its important to precisely clarify the vibration mechanism in order to suppress the
vibrations.
At present, the installations of dampers at the connection of the stay cable at the deck to
increase the cable damping are the most common solution to reduce the vibrations. More
investigations should be continued with regard to their durability and maintenance.
The definition of the damper characteristics to suppress the vibration has to be optimized,
because the generation mechanism and the magnitude of aerodynamic exciting forces of
this vibration is not fully clarified. And, particularly, if we have to deal with parametric
excitations where the usual damping systems are not efficient.
For the future :
All the damping systems cited in this paper are designed to reduce the vibration of the cables.
But designers are working on the complex civil structures for the future, and for these cable
supported structures of the future, the objective is not only to reduce the vibrations of the
cable alone, but also to reduce the vibrations of the whole structure. In order terms, the aim
will be to increase the damping of the complete structure which has a poor structural
damping. For these structures the existing passive damping systems have some limitations
that lead designers to imagine a new kind of damping system. Its why VSL is working inside
a BRITE EURAM research program, to the development of the new generation of active
control damping systems that will be used in the future.
Fatigue Reliability Evaluation of Cables in Cable-Stayed Bridges.

Case Study: The Sama de Langreo Bridge.
Jos L. GONZLEZ
Civil Engineer
ENGITEC Eng. Consultants.
Andorra
He received his degree in 1997 at
the Thecnical University of
Catalonia, UPC. At present, he is
working as a consulting engineer at
ENGITEC in the highway and
structural department.
Juan A. SOBRINO
Assistant. Prof.
Technical University of Catalonia
Barcelona, Spain
Mr. Sobrino received his PhD in Civil
Engineering at the Thecnical University
of Catalonia, UPC, in 1994. Currently,
Mr. Sobrino is holding a position as an
assistant Professor at the UPC. His
research is focused on the reliability of
existing structures and both load and
resistance models. Apart from his
research, Mr. Sobrino is running the
engineering consultant company
PEDELTA, S.L. specialized in bridge
design and assessment.
Summary
This paper presents a procedure for the structural evaluation of fatigue damage in cable-stayed
bridges due to traffic loads. The method is based on the probabilistic analysis for the calculation
of the safety margin of the Cable-Stays Fatigue Limit State under traffic loads. Structural
reliability techniques are introduced to obtain the statistical parameters of the resistance and load
variables.
A practical, real-life example, the analysis of the Sama de Langreo Bridge in Spain under real
traffic loads, is presented to illustrate the general procedure. Reliability fatigue analysis of the
Sama de Langreo bridge shows that current design criteria for the Fatigue State limit due to
highway traffic loads (wind effects are not included in this study) leads to a very low failure
probability, even for very extreme traffic load conditions. This example confirms that fatigue
design specifications of Design Codes are very conservative.
1. Introduction
The current methods for the verification of the Cable Fatigue Limit State in cable-stayed
highway bridges are usually conservative. An accurate procedure for the evaluation of the
Fatigue Limit State is presented in this paper in terms of stocastic variables based on structural
reliability theory.
In this document, it is presented a general probabilistic fatigue resistance model for cables
made up of n parallel elements (wires, strands or bars) based on the damage variable D. This
variable ranges from D=0 in the initial state to D=1 when failure is produced. Parameters of
the model are estimated with bayesian techniques from real data given by one of the most
important cable manufacturers [1]. This method, named D METHOD and its generalization
D-Bayesian METHOD are presented in chapter two. In chapter three, some simplifications
are assumed to obtain an easy semiprobabilistic method to evaluate the Fatigue State Limit.
In order to estimate the real traffic load effects in cable-stays, numerical traffic simulations
can be carried out to obtain a representative stress history of cables. For the presented case
study a numerical program has been used to obtain the traffic stress spectra using real traffic
data coming from weigh-in-motion measurements at different highways in Spain [2].
2. Fatigue models
2.1 Load model
Fatigue damage could be obtained from the load history of the element, (t). If fatigue damage
is caused by the stress amplitude and not by the stress value, the element load history could be
resumed by the load spectra (Ni,i) where Ni is the number of load cycles represented by
the stress amplitude i . If we assume, as in the classical Palmgren-Miner theory, that
damage in an element due to ni cycles of i stress amplitude is given by:
n
Di = i
N ui
eq. 1
with:
Nui= C
i-n i>r0
eq. 2
Where C and n represent material properties and r0 is the fatigue limit which will become the
most important parameter in the resistance fatigue models.
We can describe the element damage as:
m
D = Di
eq. 3
i=1
where m is the number of pairs (Ni, i).

It would be helpful to resume the spectra in a single pair of values Neq, eq. Equation 4 may
be used:
m
N eq
n
eq
= n i ni
eq. 4
i= 1
Where:
N eq = N t = n i
eq. 5
i=1
According to the equivalent stress amplitude there are basically two families of load models.
Those that consider the fatigue limit as a constant r0 and those that consider the fatigue limit
as a function of the damage variable D, r(D). Some usual possibilities are:
s=
n
eq
s = r0
f ( s ) ds
eq.6
D= 1
n
eq
eq
( D ) n dD
eq.7
D= 0
s=
n
eq ( D) =
f ( s) ds
s= r ( D )
eq.8
1
n 2
r( D ) = r0 (1 D )
; n>2
eq.9
There is an additional family of models that considers the ultimate number of cycles of
amplitude i as:
Ni= C(
i-r)-n i>r ; r : fatigue limit
eq. 10
The same algebraic operations leads to the models:

s=
n
eq
(s r )
s = r0
f ( s) ds
eq.11
D= 1
n
eq
eq
( D) n dD
eq.12
D= 0
s=
n
eq ( D) =
( s r( D))
s= r ( D )
r( D ) = r0 (1 D )
1
n 2
f ( s ) ds
; n>2
eq.13
eq.14
Using these models it is possible to obtain a simple expression for the element damage
variable when we have a variable amplitude stress process by obtaining a constant amplitude
stress process from the former variable amplitude stress process
D=
N eq neq
C
eq.15
S
C
This is the main expression to describe the resistance model of the element. It becomes
obvious that the load is represented by the product:
m
S = N eq neq = n i ni
eq.16
i= 1
And the fatigue resistance by the Paris-Erdogan parameter C

R=C
eq.17
Of course, failure is produced when S=C or D=1 and the failure probability can be expressed
as:
Pf=P(S
C)=P(D
1)=1-P(D<1)=1-FD(1)
eq.18
Pf=P(D<Dmax,asumed)
eq.19
Or alternatively
Where Dmax,assumed is the maximum assumed damage.

In the next section, a distribution function of the element damage variable is introduced.
2.2 Single element resistance fatigue model:
It is usual to accept a Weibull distribution function for the probabilistic fatigue resistance
model:
L c 0

Fc ( clL ) = 1 exp
L
0
0
c
eq.20
where L is the legth of the element, L0 is the laboratory reference length and 0 ,0 and are
parameters to be estimated in the laboratory [3,4].
This distribution function leads to the S-N-Pf curves:
(Nt-N0)
eqn = C
eq.21
1
where:
L
0
C = + 0 Ln(1 Pf )
L0
eq.22
or alternativaly to the failure probability :
L (N N 0 ) eq
Pf = 1 exp
0
L 0
Some examples of S-N-Pf curves are shown in figure 1.
eq.23
1600
Stress amplitude (MPa)
1400
1200
1000
800
Pfe=0.95
600
Pfe=0.5
400
Pfe=0
r0=247
200
0
0.00E+0
4.00E+5
8.00E+5
N0=42351
1.20E+6
1.60E+6
2.00E+6
number of cycles
Figure 1. S-N-Pf. Curves.( Pfe=0,Pfe=0.5 and Pfe=0.95)

It is shown that in the case of large number of cycles, probability curves are too close to each
other in order to try to design in the non zero probability zone, so we need to design in the
zero probability zone. As this zone is controlled by the fatigue limit, this parameter becomes
the most important in resistance fatigue models.
2.3 Cable fatigue model. D Method.
Using the probabilistic fatigue resistance model for a single element it is possible to obtain an
upper and lower limit for the distribution function of the global damage variable:
ne
Dt =
D
i=1
e( i )
ne
This variable tries to describe the global fatigue damage in a cable made up of n parallel
elements.
eq.24
Of course each of these elements is controlled by the Weibull distribution function introduced
in 2.2. The main problem is to avoid the statistical dependency of the De(i) variables caused by
the stress redistribution process when failure occurs in one of the elements.
Let us introduce the new variables:
e ( i , j ) = D e ( N t , eq ( j ) ) =
n
( N t N 0 ) eq
( j)
eq.25
C
e(i,j) is the i-element damage as all stress cycles had the constant amplitude associated to a j
broken elements situation:
( j ) = ( 0 )
ne
ne j
eq.26
eq(j)= (j) - r0
so, if De(i,j) is the i-element damage when j elements are broken, we have:
eq.27
De(i,j) e(i,j) j=1,i
eq.28
Otherwise, there will not be a stress redistribution process until the first failure is produced,
so:
De(i,0) = e(i,0) i=1,ne.
eq.29
We can obtain the following expressions from previous ones :

De(i,j)De(i,k) j
k
eq.30
De(i,0) = e(i,0) De(i,j) e(i,j)
eq.31
If Dt(i) is the cable damage when i elements are broken, we can write:
m1
m
m
P( D t
) = P( D t ( i )
) P( n failure = i ) =
ne
ne
i=0
m1
ne
i=0
j= i + 1
P( D
e( j ,i )
m i ) P( n failure = i )
eq.32
Lets try to aproximate these expressions:

On one hand we have:
so:
De(j,i) e(j,i) j=i+1,ne
eq.33
ne
ne
j= i + 1
j= i + 1
P( D e ( j , i ) m i ) P ( e ( j , i ) m i )
eq.34
and:
so:
De(j,i) e(j,0)
eq.35
ne
ne
j= i + 1
j= i + 1
P( D e ( j , i ) m i ) P( e ( j , 0 ) m i )
eq.36
On the other hand we have:

P(nfailure=i eq(0)) P(nfailure=i) P(nfailure=i eq(i))
eq.37
It is possible to identify the n failure variable distribution with a binomial with a Pfe(i)
parameter as follows:
P( n failure = il eq( i ) ) = ( ni e ) Pfei ( i ) (1 Pfe ( i ) ) ( n e i )
eq.38
L (N N 0 ) eq( j )
Pfe ( j ) = 1 exp
0
L 0
ne
eq( j ) = ( eq( 0 ) + r0 )
r
ne j 0
eq.39
eq.40
These expressions let us obtain an upper and a lower limit for the Dt distribution function:
P( D t
ne
m1
m
) P( e ( j, 0 ) m i ) P( n failure = il eq( i ) ) = Pf ,sup
ne
i= 0
j= i + 1
eq.41
ne
m 1
m
P(D t
) P( e( j,i) m i) P(n failure = il eq( 0) ) = Pf ,inf
ne
i= 0
j=i +1
eq.42
These limits become closer when m is lower than ne. An example is shown in figure 2. It is
posible to include probabilistic distribution of the model parameters getting into a bayesian
context:
P( D t
m
m
) = P( D t
lp ) dP(p ) Pf (sup) ( mlp ) f p (p ) dp
ne
ne
p
p
eq.43
where p is the parameters vector.
Pf(inf), Pf(sup)
1.0
Pf(sup)
Pf(inf)
0.5
0.0
0.0
10.0
20.0
Dt %
30.0
Figure 2 Upper and lower limit of Dt distribution function: ne =100, N=2106

cycles, =264.5 MPa, L=L0=2m, N0=42351 cycles, r0=247 MPa,n=2.06,
=6.974108, 0=11010, 0=1.45
When N is very high (design zone) the upper limit is more useful:
P( D t >
m
)=
ne
r0 = r0 ( )
P( D t >
r0 = 0
m
lr ) dP( r0 ) <
ne 0
r0 =
r0 =
dP( r ) = f
0
r0 = 0
r0 = 0
r0
( r0 ) dr0 = P( r0 )
eq.44
so the fatigue limit becomes the most important parameter in fatigue resistance models that
can be used for the stay-cables or anchorages.
3. Safety criteria
To evaluate de fatigue safety margin, the following expression may be used:
eq,d adm,t
eq.45
with:
s=
n
eq,d =
s= 0
f ( s) ds
eq.46
If no data is available, a conservative value of n=3 may be used:
adm,t = r 0
eq.47
fat
where r0 is the mean of fatigue limit and fat is the proposed safety factor that depends on [5]:
Coefficient of variation of the equivalent stress amplitude V

Coefficient of variation of the Fatigue Limit Vr0
eq,d)
Reliability Index = 1 (Pf ) with Pf=P(ro<
when normal distribution is supposed for fatigue limit and equivalent amplitude stress, the
safety factor verifies the following implicit equation.
2
r0
fat = 1 + V
2
fat
+V
1 + ( Vr0 fat
2
V
)
+
2 Vr0 fat
eq.48
The above mentioned formulation could be applied to any mechanic system with a fatigue
limit resistance parameter as anchorages, etc.
4.
Case study. Sama de Langreo Bridge.
To illustrate the possibilities of the above proposed methods, in this chapter a case study is
presented. The Sama de Langreo bridge is an asymmetric cable stayed bridge with a single
tower over the Nalon river placed in the north of Spain designed by Fernandez Casado, S.L.
built in 1989. The main span over the river has a length of 130 m. The total length of the
bridge is about 300 m, including side spans. The bridge has two traffic lanes with a total
width, including shoulders, of 14.17 m. The tower is an A-shaped concrete construction,
located on the right riverside and has a height of 57 m. The girder of the superstructure
consists of a reinforced concrete voided slab girder. The cross section height is only 1.3 m.
The main span and side span are supported by two vertical cable fans consisting of 16 steel
cables anchored in the girder each 15 m. Figure 3.
The aim of this investigation is to find the fatigue damage (or the fatigue level of safety) in
the stay cables due to traffic loads on the bridge. Wind effects or other variable loads are
neglected in this study.
Figure 3. Sama de Langreo Bridge elevation.
4.1
Fatigue analysis
The fatigue analysis has been made taking into account the following considerations:
A simplified structural model consisting of a grillage model of the superstructure (bridge

girder) with nodal springs to simulate the stay cables.
Traffic simulations have been developed using a numerical simulation program developed
in [2,6]. The program is based on Monte Carlo simulations, taking into account free or
congested traffic on the bridge. Two different free and congested traffic situations have
been studied with different truck intensities (see Table 1).
Traffic situation n 1
Traffic situation n 2
Car intensity
70 %
50 %
Truck intensity
30 %
50 %
% of vehicles in lane 1
47 %
57 %
% of vehicles in lane 2
53 %
43 %
Table 1. Sama de Langreo Bridge elevation.
Stress spectra has been obtained for different stay-cables. Stress amplitudes vary between
5 and 140 MPa, with equivalent amplitudes around 40 to 80 MPa (Figure 4). These values
are very conservative if compared with those used in the fatigue design. The obtained
values of equivalent amplitudes are about 40-50 % of fatigue design admissible stresses
for traffic loads.
0.80
frecuency
0.60
0.40
0.20
0.00
stress amplitude (MPa)
Figure 4. Stress spectra obtained using numerical simulations in the cable-stay n 8.
Using the stress spectra due to traffic loads, safety margin for fatigue may be obtained for
different resistance models. Figure 5 shows the reliability index (inverse of the Gaussian
probability of failure), for two different fatigue limits (ro) and for two different variations
of these limits in order to take into account the uncertainties in the cable-stay resistance (or
anchorages). The obtained reliability index is always higher than the minimun accepted
in structural codes >3,5 to 5, even for the very conservative traffic conditions assumed
in this study.
20.0
Vr=0.05
18.0
Average r0
16.0
200 MPa
14.0
x
e
d
n
i
250 MPa
12.0
Vr=0.1
10.0
8.0
6.0
4.0
2.0
0.0
0
40
80
120
160
200
equivalent stress amplitude(Mpa)
Figure 5. Reliability Index versus equivalent stress amplitude for two different fatigue
resistance limits (ro) and for two different coeficient of variation of this parameter Vro.
5.
Conclusions
As the main conclusion, in this example it is clear that safety criteria for fatigue design seems
to be very conservative. On the other hand, technical specifications for fatigue of stays and
anchorages should be more realistic, and fatigue tests should be carried out on the range of
real behaviour of these elements under traffic loads or other variable loads (winds, etc). More
data and further investigations are needed to obtain simplified and realistic fatigue design
specifications for stay-cables.
6.
References
[1]
[2]
TYCSA, Data of fatigue strands. Internal report of TYCSA 1997.

Sobrino, J.A.; Evaluation of structural safety and serviceability of existing prestressed
and reinforced concrete bridges (in Spanish). Ph. D. Thesis, Civil Engineering School,
Barcelona. 1993.
[3]
Castillo, E. & Fernndez Canteli, A.;Statistical models for fatigue analysis of long
elements, IABSE periodica. 1992
[4]
Arnold, B.C., Castillo E. & Sarabia, J.M.; Classical and bayesian analyses of fatigue
strength data, IABSE periodica, Madrid 1992
[5]
Thoft-Christensen, P.& Baker, M.; Structural reliability theory and its applications.
Springer-Verlag, 1982.
[6]
Sobrino J.A.; Structural evaluation of an old prestressed concrete bridge. IABSE

Symposium. Extending the lifespan of structures, pp. 817-822. San Francisco, 1995.
7. Acknowledgements
The authors thanks Prof. Javier Manterola of Fernndez Casado, S.L. (author of the Sama de
Langreo Bridge) for the information provided for the case study. The authors also wish to
thank Mr. Jess Pars of TYCSA for the fatigue data used in this paper.
The Super High Damping Rubber Damper on the Stay-Cables of

Meiko East Bridge
Minoru MIZOE
Mgr, Head Office
Japan Highway Public Co.
Tokyo, Japan
Sinji MUROI
Mgr, Civil Eng. Division
Nippon Steel Co.
Tokyo, Japan
Takashi HORII
Civil Eng.
Bridgestone Co.
Yokohama, Japan
Toshiyuki ISOBE
Mgr, Civil Eng.
Bridgestone Co.
Tokyo, Japan
Renji KIYOTA
Mgr, Civil Eng.
Yokogawa Bridge Co.
Funabashi, Japan
Yasuo IMADA
Civil Eng.
Yokogawa Bridge Co.
Funabashi, Japan
Summary
The stay-cables of the Meiko East Bridge have a parallel wire strand coated with
polyethlene tubes. In Japan, a rain vibration has been often observed at the cable
coated with the smooth surfaces, such as a polyethylene. Both damping
countermeasure and consideration foraesthetics point of view, was requiredin Meiko
East Bridge. As the countermeasure, a damping device using SDR(Super Damping
Rubber) that can be installed inside of a waterproof cover, was adopted. As a result of
experiments, the required damping of the cable were obtained.
This paper gives an outline of the damping device and experimental results of thestay
cable in Meiko East Bridge.
1. Introduction
Meiko East Bridge is one of the three cable-stayed bridgesto across the Ise Bay in
Japan. The bridge has 700m in total length and a center span length of 410m. The stay
cables are arranged in two planes of a multi-fan shape. Each stay cables is made of
galvanized wires and coated with polyethylene tubes. The longest stay cable is 209m
in length, and the maximum diameter is 165mm.
In the cables mentioned above using in a cable-stayed bridge,a vortex induced
vibration and rain vibration are easily caused by the wind because of the smooth
surfaces on the cables and low structural damping. The rain vibrations induced by
both wind and rain have often been reported forthe cables having a smooth surface,
such as a polyethylene and 120-200mm in diameter.
In Japan, rain vibration was recognized on the stay-cables inMeiko West Bridge-l for
the first time, which is located at 3km west from Meiko East Bridge and opened in
March 1985.
For the design of Meiko East Bridge it was decided tocontrol rain vibration because
of the reasons mentioned above.Some techniques have been used to control rain
vibration; (1) to install the damping devices on the cable nearbythe anchor point, (2)
to connect neighbor cables with wires, (3) to improve theaerodynamical
characteristics of cables by increasing roughness on their surfaces. At Meiko East
Bridge, both damping countermeasure and consideration foraesthetics point of view,
were required. The cable damping device using SDR(Super Damping Rubber) that
can be installed inside of a waterproof cover at the top of anchor pipes was adopted as
the countermeasure.
700 m
410 m
145 m
145 m
TP+130.2m
TP+130.0m
N.H.H.W.L=TP+1.4m
P1
P2
C30N
C26N
40 m
CL
P3
P4
Fig. 1 General View of Meiko East Bridge
2. Damping required for stay cable

Since 1984, the field measurementsand wind tunnel tests have been done for many
cable-stayed bridge designs. As a result, when the Scrutons number Sc is less than 50
and a natural frequency of the cable is less than 3.0Hz, the vibrations occur. Sc is
given as follows.
m
Sc =
(1)
D2
where,
m; cable mass.
; logarithmic damping decrement of a cable.
; air density.
D; cable diameter.
Therefore, if Sc could be designed greater than 60, which has natural frequency less
than 3.5Hz, we can prevent rain vibration. In order to prevent vortex induced
vibrations of the cable, not as much damping is required. Regarding cable damping
countermeasure, the aesthetics of the bridge was an important consideration in Meiko
East Bridge. For this purpose, we decided to adopt a cable damping device using
SDR(Super Damping Rubber) that can be installed inside of a waterproof cover at the
anchor pipe.
3. Outline of the device

SDR is installed between the cable and the anchor pipe. One side of the
SDR(rectangular section) is connected to an anchor pipe and the other side to a cable.
When the cable is excited, a relative displacement occurs between an anchor pipe and
the cable, and the SDR undergoes a shear distortion. SDR can absorb the vibration
energy of a cable due to this shear distortion. This device is useful for all radial
vibrations of a cable. Figure 2 shows the structure of the device, and photograph1
shows the views before and after installing the waterproof cover.Generally, in a
cable-stayed bridge, the steel pipe is

used for the anchor. The device is
installed at the top of an anchor pipe,
then covered by waterproof cover. The
device needs no additional attachment
points on the bridge deck, andthe
devices are simple to maintain.
Cable
SDR
Steel pipe
Waterproof cover
Fig. 2 Cable Damping Device using SDR
a) Without Waterproof Cover
b) With Waterproof Cable
Photo 1 Cable Damping Device
4. Material properties of SDR
Load P (KN)
We have developed SDR by blendingon SBR(styrene butadiene rubber) polymer,

high damping carbon and some plastics toachieve the high damping properties. The
hysteresis loss at shear distortion for SDR is shown in figure 3. SDR has greater
damping characteristics than the HDR(High
Damping Rubber) used for seismic bridge
3
bearings. However, the shear modulus and
damping coefficient of SDR are dependent on
2
shear distortion, shear velocity and ambient
1
temperature. Figure 4 shows variation of the
shear modulus with and temperature.
0
Damping coefficientdepends almost on
entirely temperature. These material were
-1
developed experimentally andcan be
-2
expressed by some influential functions.
Shear modulus;
-3
-3
-2
-1
0
1
2
3
G(, f, T) = CGf (f) CGT (T) G( )
(2)
Shear Distortion
where,
Fig.3 Hysteresis Loops of SDR
CGf (f); velocity revision coefficient.
CGT(T); temperature revision coefficient.

G(); shear modulus at shear distortion .
Damping coefficient;
(, f, T) = C (f) C (T) ( )
(3)
where,
Chf(f); velocity revision coefficient.
ChT(T); temperature revision coefficient.
h(); damping coefficient at shear distortion.
5
Shear Modulus G(KN/mm2)
Shear Modulus G(KN/mm2)
5
4
3
2
1
0
0.0
0.2
0.4
0.6
0.8
1.0
4
3
2
1
0
-10
Shear Distortion
(a) Variation of Shear Modulus (G) with

Shear Distortion (
) (f=1.0Hz,T=20? )
10
20
30
40
Temperature T()
(b) Variation of Shear Modulus (G) with

Temperature (T) (
=0.2,f=1.0Hz)
Fig.4 Shear Modulus of SDR
5. Calculation method of additional damping decrement

A dynamic modelof a damping device with SDR can beshown in figure 5. The
hysteresis loss of SDR is due to viscous damping. The additional damping decrement
of a cable damping device can be expressed as
follows.
2
{ ( c )}
=
C
( 3)
a
M
where,
a; additional logarithmicdamping
decrement
by the device.
(c); mode amplitude of a cable at damping
device installed position.
C; viscous damping coefficient of a damping
device.
M; equivalent mass of a cable.
= m 2 ( )d
(m; mass per unit length)
Cable
K
C
L
Fig.5 Model of Cable Damping System
; circular frequency.
Mode amplitude (c) can be obtained by mode analysis. The equivalent value C of
viscous damping coefficientis shown (5).
C=
2h
(5)
where,
h; damping coefficient of SDR.
K; stiffness of SDR.
A G ( )
=
(6)
H
A, H; section area of SDR, thickness.
G (); shear modulus of SDR.
; shear distortion of SDR.
6. Confirmation of the cable damping

6.1 Experiment method
Through experimentation we have
confirmed an additional damping
decrement. The cable damping
devices were installedon two
cables(C26N, C30N) at the upper
row of a bridge. We excited a
cable to make free vibration and
got damping decrements of the
cable using the free vibration
wave shape(see photograph 2).
The vibration mode was set up
within 3.0Hz of the known range
for rain vibration. We
experimented from 1st to 5th
mode concerningthe cable
Cable-Number
Cable Construction
Corrosion Protection
Cable Diameter
Cable Length
Cable Unit Weight
Frequency
Photo.2 Experiment Situation
C26N
C30N
283 galvanized wire( 7mm) 265 galvanized wire( 7mm)
Polyethylene coating
Polyethylene coating
0.150 m
0.145 m
198.5 m
143.7 m
91.6 kg/m
85.7 kg/m
0.56 Hz
0.78 Hz
Table 1. Dimensions of Cables

6.2 Calculation of additional damping decrement
Figure 6 shows the relationship betweenstiffness and additional damping decrement
of SDR with depended ontemperature and shear distortion. The value of additional
damping decrements are shown at ? symbols. Table 2 shows additional damping
decrements that can be obtained for each vibration mode of a cable. In case of the
range of 0.05-0.30 of shear distortion and 0-40? of temperature, a logarithmic
0.05
0.04
=0.05
0.03
=0.10
0.02
=0.30
0.01
0.00
1000
2000
3000
4000
5000
Additional Damping Decrement a
Additional Damping Decrement a
damping decrement is 0.020-0.035 for C26N cable and is 0.026-0.041 for C30N
cable. As assuming a structure damping decrement of a cable is 0.005 before damping
device installing, a damping decrement after installing is estimated as 0.025-0.041 for
C26N cable and is estimated as 0.031-0.046 for C30N cable. On the other hand, when
Scrutons number is 60,required damping decrement is0.018 for each cable. The
estimated damping decrement ofthe cable should exceed the required value.
0.05
0.04
0.03
40
0.01
0.00
1000
Stiffness K(KN/m)
(a)
a K (T=20? )
20
0.02
2000
3000
4000
5000
Stiffness K(KN/m)
(b)
=0.20)
a K (
Fig.6 Variation of Additional Damping Decrement (

a) with Stiffness (K) (C26N-Cable,1st mode)
Cable- Mod Frequenc Require Temp Additional Damping

()
Numbe
Dampin
() =0.3 =0.1 =0.0
(Hzj
C26
1
0.56
0.01
0
0.02
0.02
0.02
20
0.03
0.03
0.03
40
0.02
0.03
0.03
2
1.12
0.01
0
0.02
0.02
0.02
20
0.03
0.03
0.03
40
0.02
0.03
0.03
3
1.68
0.01
0
0.02
0.02
0.02
20
0.03
0.03
0.03
40
0.03
0.03
0.03
4
2.24
0.01
0
0.02
0.02
0.02
20
0.03
0.03
0.03
40
0.03
0.03
0.03
5
2.80
0.01
0
0.02
0.02
0.02
20
0.03
0.03
0.03
40
0.03
0.03
0.03
C30
1
0.78
0.01
0
0.03
0.03
0.02
20
0.03
0.03
0.03
40
0.03
0.03
0.04
2
1.56
0.01
0
0.03
0.02
0.02
20
0.03
0.03
0.03
40
0.03
0.03
0.04
3
2.34
0.01
0
0.03
0.02
0.02
20
0.03
0.03
0.03
40
0.03
0.03
0.04
Damping
(a\
s)
0.02
0.03
0.03
0.02
0.03
0.03
0.02
0.03
0.03
0.02
0.03
0.03
0.02
0.03
0.03
0.03
0.04
0.03
0.03
0.04
0.03
0.03
0.04
0.03
` 0.03
` 0.03
` 0.04
` 0.03
` 0.03
` 0.04
` 0.03
` 0.03
` 0.04
` 0.03
` 0.03
` 0.04
` 0.03
` 0.03
` 0.04
` 0.03
` 0.04
` 0.04
` 0.03
` 0.04
` 0.04
` 0.03
` 0.04
` 0.04
Logarithimic Damping Decrement of Cable before setting Damping Device (0.005)

sF
Table 2 Calculations of Additional Damping Decrement
200
150
100
50
0
-50
-100
-150
-200
0
10
20
30
40
50
60
Time (sec)
(a) Time Histories without Damping Device

Amplitude (gal)
Figure 7 shows a typical example of

the experimental results for the
C26N cable. The damping
decrement does not depend on
amplitude. The vibrations of
damping decrement is caused by
coupled vibration of the cable
having a similar natural frequency.
The dispersion at small amplitude
range is caused by wind.
Amplitude (gal)
6.3 Results of experiment
200
150
100
50
0
-50
-100
-150
-200
0
Logarithmic Damping
Decrement ( a + s )
Table 3 shows the complete result

of experiment. The structural
damping decrement ofthe cable
without damping deviceis 0.0050.010. The structural damping
10
20
30
40
50
60
decrement of a cable is greater than
Time (sec)
estimated values because of
(b) Time Histories with Damping Device
absorbers installed on the tower.
0.10
An averaged of logarithmic
{: without Damping Device
0.08
: with Damping Device
damping decrement is 0.033-0.045
0.06
at C26N cable and 0.042-0.046 at
0.04
C30N cable in the range of
amplitude 5-310gal. The damping
0.02
decrements of 1st mode of C26N
0.00
0
20 40 60 80 100 120 140 160 180 200
cable and 2nd mode are smaller
Amplitude (gal)
compared with the other mode due
(c) Logarithimic Damping Decrement
to the influence of coupled
vibration. Experimental damping
Fig. 7 Typical Time Histories and Logarithmic
decrement is in excess the estimated
Damping Decrement of C26N-Cable (3rd Mode)
value by calculation. The damping
decrement has not dependence on vibration mode.
Cable- Mode
Number
Temp.
( )
C26N
1
4.1
4.3
2
4.2
4.1
3
4.4
4.8
4
5.2
4.9
5
5.3
5.4
C30N
1
5.4
5.6
4.8
2
4.7
5.0
5.0
3
5.1
5.1
5.0
Damping Decrement
Additional Damping Decrement ( a)
a
s)
Amplitude
Measured
Average
Calculated
Measured
/
(gal)
Temp.( )
=0.13 Temp.( )
=0.13
5
20
0.020
0.045
0.036
5.0
0.026
4.1
0.030
1.17
5
15
0.020
0.043
0.033
4.3
0.028
1.09
15
75
0.033
0.052
0.040
5.0
0.024
4.2
0.030
1.23
15
75
0.039
0.050
0.042
4.1
0.030
1.23
25
125 0.039
0.050
0.045
5.0
0.024
4.4
0.034
1.43
25
125 0.039
0.050
0.045
4.8
0.034
1.43
40
270 0.026
0.052
0.042
5.0
0.023
5.2
0.037
1.58
40
230 0.030
0.051
0.042
4.9
0.037
1.58
50
200 0.035
0.049
0.042
5.0
0.023
5.3
0.035
1.50
50
310 0.039
0.050
0.045
5.4
0.035
1.50
5
30
0.039
0.050
0.046
5.0
0.029
5.4
0.042
1.46
5
30
0.041
0.051
0.046
5.6
0.040
1.39
5
25
0.038
0.051
0.045
4.8
0.040
1.39
15
70
0.038
0.052
0.045
5.0
0.028
4.7
0.039
1.42
15
75
0.038
0.051
0.046
5.0
0.040
1.45
15
75
0.035
0.050
0.043
5.0
0.036
1.31
30
130 0.022
0.058
0.042
5.0
0.027
5.1
0.036
1.34
30
180 0.038
0.052
0.045
5.1
0.035
1.30
30
180 0.021
0.060
0.044
5.0
0.038
1.41
Table 3 Test Results of Cables
7. Conclusion
In this paper, we described the abstract of a cable damping device using SDR(Super
Damping Rubber) that was adopted to Meiko East Bridge. The damping effects was
estimated by the calculation, and confirmed throughthe field experiments.
Experimental results exceed calculated value. As a result of these experiments,
damping devices were installed at all cables except some lowerrow cables. Since
completion in April, 1998, wind induced vibrationhas not been observed.
The original aesthetics of the bridge were maintained by installing the cable damping
devices within a waterproof cover of the cable anchor pipe.
References
[1].
[2].
[3].
[4].
Public Work Research Center, Report on Study of Wind Resistance of Cables

in Cable-Stayed Bridges, 1993.
S.Kuranishi and T.Takahashi, Vibration Analysis of Beams with Damping at
Discrete Location, Proceedings of JSCE, No.187, 1971.
Y.Imada, R.Kiyota, Y.Suizu and Y.Kasahara, Development of Damping
Device Using Shear Resistance of High Damping Material, Proceedings 48th
annual conference of JSCE, 1, 1993.
Y.Imada, T.Sasaki, R.Kiyota, and K.Goda, Experiment for Control Means On
Cable Vibration Using High Damping Rubber Damper on Cable-Stayed
Bridge, Proceedings 50th annual conference of JSCE, 1-(B), 1995.
Corrosion Protection of Locked Coil Ropes at Road Bridges

Arnold
HEMMERT
HALSWICK
Civil Engineer
Arnold HemmertHalswick, born 1954,
received his diplom degree
in 1981 and his doctor
degree at the RWTH
Aachen in 1986. Since
1987 he works in steel
structures at the BASt.
Siegfried SCZYSLO
Civil Engineer
Fed. Highway Res.. Inst
Siegfried Sczyslo, born
1937, received his
graduate degree in 1959
and worked at consultants
and steel structure
companies to build steel
bridges. Since 1971 he
works in steel structures at
the BASt and since 1986
he is head of the group
steel structures, corrosion
protection at the BASt.
Summary
The subject are corrosion protection measurements at ropes resp. cables of bridges in Germany.
According to the experiences over the years resp. decades standards were drafted and improved
according the experiences which were won in working.
1.
Introduction Rope Bridges in Germany
In the course of federal motorways, major roads and some city roads in Germany there are about
40 big road bridges which have high strength cables. All but one bridges have locked coil ropes,
one bridge has parallel wire strands. In Germany there are no bridges with bundles of strands
how they were used elsewhere last time e.g. in France at the Normandie Bridge near Le Havre.
Most bridges are cable stayed bridges, some bridges are suspension bridges, at two bridges there
are under-guyings. Closely packed bundled single ropes are called as cables, see Fig. 1.
Left: Reconstruction of the suspension

bridge over the river Rhine near
Cologne-Rodenkirchen with a cable still
without hangers on the left
Fig. 1: Cable
Right: Side view of a hexagonal cable
2.
Locked Coil Ropes
2.1
General
Locked coil ropes inside consist of round wires and outside of several layers of Z-wires which
tighten the inner structure by their shape and order, see Fig. 2. They have the advantage that a
wire takes load after two lay lengths if it is broken. The lay length is the distance within a single
wire turns around the longitudinal axis of the rope one time with a spiral. This behaviour is got
by pressing this wire into the structure of the rope through contraction of the rope caused by the
spiral structure. At parallel wire strands a broken wire is lost for the strength over the whole
length of the cable. Furthermore locked coil ropes have the advantage that they can be designed
in a more slender way than parallel wire strands and that by the locked type the inner structure of
the rope is mechanically tightened and therefore much better protected against corrosion attack.
But it seems that they have the disadvantage that they are more expensive than parallel wire
strands having the same strength. Regarding the rope mechanics it has to be considered that they
have a lower modulus of elasticity than parallel wire strands by reasons of the spiral structure
which is about 170 000 N/mm.
Fig. 2: Cross section of a locked coil rope, see [1]

2.2
Galvanizing of Wires
In the beginning (about 1959 1965) the single wires were not galvanized. Later on only the
outer wires were galvanized, and since the end of the seventies all wires are galvanized. The zinc
layer amounts to 280 300 g/m corresponding about 40 m. At some bridges wires were used
which were electrogalvanized with a coating of about 500 g/m corresponding to about 47 m,
but in the meantime only galvanized wires prevailed on the market. The center wires are round,
the outer layers of the wires have a Z-profile. In the beginning trapezoidal wires were used also
between the round and Z-profiled wires, but they were let away by reasons of fatigue since the
middle of the seventies. They had the advantage that they formed as at a stone arch bridge an
arch which additionally defended the entrance of water in the inner of the rope.
3
The galvanizing alone is not a sufficient corrosion protection. Especially at the corners of the
form wires the zinc layer is thinner as at the plane areas. Therefore an outer corrosion protection
is necessary in the form of a coating, consisting of several coats, see Fig. 3.
Fig. 3: Sequence of coats

2.3
Blocking Agent
Beside some exceptions red lead was and is used as blocking agent which is filled in the inner of
the rope during the fabrication of the single layers. The aim is that by the blocking agent an easy
gliding is possible between the single wires (lubricating) and that the single wires are protected
against corrosion. As these characteristics must be available over the whole life time the
blocking agent has to fulfil high requirements of the lubricating, the corrosion protection and the
durability.
Until now in Germany the normal blocking agent was red lead known from the sector corrosion
protection. But there were some attempts to use mixtures of polyoil and zinc dust because of the
galvanized wires, but the experiences so far at bridges show that because of the higher viscosity
the risk exists that the blocking agent comes out of the rope due to the contraction. Also artificial
waxes were used, but in one case the last layer of wires was fabricated without a blocking agent
because of the uncompatibility with the outer corrosion protection coating what seems doubtfully
facing the thoughts to the sense of a blocking agent. The standard blocking red lead agent can
come out of the rope, if during the fabricating mistakes were made.
2.4
Rope anchoring
The anchoring of a rope is normally done by the embedding of the wires in cast corpus
consisting of a zinc alloy and by wedging the wires against each other in the socket. Beside the
stressing of axial loads according the strength of the cable a radial tension strength develops in
the socket consisting of cast steel or rolled steel by the wedging. Therefore a high level of
exemption from cracks is required which is verified by ultrasonic tests. If no clear result are got a
radiographic inspection is necessary. For the outer condition a magnetic particle test must be
carried out. If damages are found it is allowed that they are touched up, but every socket must
fulfil the requirements for its own.
3.
Rules
3.1
General
Standards were drafted and introduced in the responsibility of the German Ministry of Transport
to avoid mistakes at this member sensitive and important for safety and to have a basis for the
invitation to bid for rope works. By this a high level of quality at the fabrication of the rope and
the capability for inspection under traffic is to be ensured. The maintenance is also dealt with in
detail. These additional technical contract conditions mention the special things of locked coil
ropes if they are used at bridges. Additional to the regulations in the standards requirements to
the rope or cable (rope bundle) and to the corrosion protection are formulated to fulfil the
expectations to the life time: It is counted on that the existing bridges reach 80 to 100 years
supposed that the heaviness of the traffic does not grow strongly and that no other life time
diminishing influences occur which are not determined by the structure itself. For the structure
itself a regular test of the stability is necessary in any case. The rope is not understood as a
working part, but modern structures allow the changing of single ropes in an accidental case.
3.2
TL Seile
Technical Delivery Conditions are carried out for Locked Coil Ropes, short TL Seile (Ropes)
[1], where requirements for the rope itself are laid down. The requirements are widespread from
the single components over the fabrication to the delivery on site.
The bridge rope consists of the locked coil rope with the in-locating blocking agent and the
sockets at both ends of the rope, which are fixed at the ends by casting. For the blocking agent a
number of characteristic values has to be guaranteed. For the rope cast the rope is put through the
openings of the sockets. The ends of the rope are opened to a broom of wires and cast in the
socket with molten metal. An alloy of zinc is used as cast metal.
In the course of the fabrication of the ropes a number of tests has to be carried out, ending with a
tensile test at the rope for the verification of the ultimate strength with the destruction of the
specimen.
The TL Seile contains a technical delivery condition for the standard blocking agent red lead at
its own within characteristic values are fixed. The aim is that these values deviate as little as
possible as this stuff can be seen as well-tried.
The TL Seile contains also an instruction for the carrying out of the standard test for fatigue of
the rope. It is important that the specimen comes out of the delivery for the structure.
3.3
ZTV-KOR-Seile
The Technical Delivery Conditions for the Corrosion Protection of Bridge Ropes, short ZTVKOR-Seile [2] (in preparation) will replace in a short time the still existing Guidelines for the
Corrosion Protection of Bridge Ropes, short RKS [3].
Die ZTV-KOR-Seile concern all measurements for the corrosion protection at bridge ropes and
cables. They are subdivided as follows:
General
5
-
Design considering corrosion protection

Planning of the corrosion protection measurements
Surface preparation
Protective paint materials, sealing material, injection material and protective systems
Execution of the corrosion protection work
Health/safety/environment protection requirements and disposal of blast-cleaning rubble
Acceptance
Guarantee
Inspection and testing of the ropes and cables
Reference standards
There are three supplements;

Requirements to the qualification of the execution personal
Protocol for the corrosion protection
Example for working instructions in drawings and texts for the different member areas at
ropes and cables
Basis among other things is the EN ISO 12944 Parts 1 to 8 [5]. The sense of these detailed
instructions is to be not obliged to invent everything new at any structure. It cannot be expected
that every time the same personal comes to site and bring along experiences. Therefore
recommendations for the execution of repair works are given for the special conditions at rope
and cable bridges, especially in the drawings of supplement 3. Many times it is referred to the
new DIN EN ISO 12944 [5]. The surface preparation according Part 4 of DIN EN ISO 12944
has great importance as on the one side blocking agent is still on the surface of the rope from the
fabrication and can have effects as a separation layer, on the the other side the new coating
should be resistant and durable. Very often it is worked with rotating brush at galvanized ropes,
see Fig. 4.
Fig. 4: Surface preparation with rotating brush

The corrosions protection works itself can be done from a cabin where a protection against
weather influences must be given, see Fig. 5, which protects at the same time the environment
against damaging effects e.g. by dust and paint. If a further surface preparation than with rotating
brushs is necessary a housing in on the complete length of the rope or cable from the bridge deck
up to the pylon is necessary.
Fig. 5: Cabin for the corrosion protection

The use of sealings is required, e.g. by bellows of synthetic material, see Fig. 6, so that no
moisture enters the rope anchoring.
Fig. 6: Bellows for the sealing of a rope
4.
Maintenance of the ropes
4.1
Provided tests
At the construction works in the responsibility of the road administration in Germany the
German standard DIN 1076 [4] is used. A main inspection has to be done every 6 years and
between after 3 years a simple inspection. Beside this a currant observation takes place normally
every 3 months and a yearly inspection where no bigger auxiliary measures are used as e.g.
bridge inspection trolley. If a test seems to be necessary after an inspection or if exceptional
events could have damaged the bridge a test for a special reason is executed.
The former usual test loading is not carried out in Germany for decades as the benefit is to less
opposite the damage which eventually occurs by the normally big loading of a loading test.
4.2
Bridge rope inspection machine
For the main inspection at bridges with ropes and cables in Germany a Bridge rope inspection
machine was developed. It is a kind of cable railway with whom a driving of the ropes and
cables is possible without loading the tensile elements themselves. A carrying rope bearing the
7
cabin is passed over the pylon/pylons. For the operating a number of machine components are
necessary, see Fig. 7.
Fig. 7: Bridge rope inspection machine

1
hasp
2
3
poller station
deflection unit for the
carrying rope
deflection unit for the
traction rope
6
7
deflection unit at the pylon

shaft for carrying and
traction rope
deflection spreader
traction rope
traction rope winch
4.3
working platform with

lift winch
10
11
carrying rope
Chassis with hangings
Magnetic inductive test
For the inspection of ropes to detect wire fractures the magnetic inductive test is used. To do this
a magnetic sensor is pulled along the single rope and the sensor reacts to changes of mass. By the
different intensity of the signals of the magnets located around the rope a localisation of new
wire fractures is possible by comparison with the test 6 years before. By comparison of the
summed up wire fractures with the condition just after completion of the structure during the first
test an estimation of the bearing behaviour is possible. It is only necessary to sum up these wire
fractures which lay within four lay lengths because every broken wire bears the full load after
two lay lengths. As the magnetic sensor has a big load and stresses the rope surface by wheels
directly caution is necessary at the corrosion protection coatings. Investigations are desirable to
develop less sensor loads.
5.
Investigations of Ropes
By the bast investigation were carried out with the aim to get information about the mechanical
situation for bridges ropes in the structure. Beside that investigations were done about the
climatic effects by exposure tests. At a chemical laboratory the chemistry of the blocking agent
itself was investigated. In particular the water absorption was important. Also the outer corrosion
protection was investigated especially in covered areas later not accessible as clamps, saddles,
anchorages e.g., and long time experiences were evaluated.
At long time measures at a cable stayed bridge the load changes at ropes were registered [6]. It
was shown that the big rope elongations did not result from rope vibrations but from
superstructure vibrations. However the sums of the elongations are nearly the same for both
reasons due to the more frequent occurence of the rope vibrations e.g. as a result of wind.
Temperature measures were also carried out at ropes where the temperatures were measured on
and under the corrosion protection coating and the sensors, see Fig. 8, located in one cross
section up and down, upstream, downstream each 90 progressively. There were big differences
on and under the coating, where in Fig. 9 the given curves are the mean values of the four
temperatures. For the conditions in the rope this means that also at intensive sun shine the
temperatures in the inner rope are not as high as on the surface of the rope.
Fig. 8: Temperature sensors
Fig. 9: Temperatures at the rope on and under the corrosion protection coating
6.
Conclusion
The locked coil rope in bridge structures presents itself as a robust, durable and easy-care
member if the requirements of the above mentioned regulations are fulfilled which can be
inspected easily, too. Single wire breaks of the outer layer possibly appearing are not critical and
can be seen at the surface.
7.
Literature
[1]
TL Seile Technische Lieferbedingungen fr vollverschlossene Brckenseile.

Bundesministerium fr Verkehr, Abteilung Straenbau. Verkehrsblatt Verlag Dortmund
1994
[2]
ZTV-KOR-Seile Zustzliche Technische Vertragsbedingungen und Richtlinien fr den

Korrosionsschutz von Seilen und Kabeln im Brckenbau. Bundesanstalt fr
Straenwesen Entwurf 1998
[3]
Richtlinien fr den Korrosionsschutz von Seilen und Kabeln im Brckenbau RKSSeile. Der Bundesminister fr Verkehr, Abteilung Straenbau, Deutsche Bundesbahn
Ausgabe 1983
10
[4]
DIN 1076: Ingenieurbauwerke im Zuge von Straen und Wegen, berwachung und
Prfung. Normenausschu Bauwesen (NABau) im DIN Deutsches Institut fr Normung
e.V. Mrz 1983
[5]
DIN EN ISO 12944 Parts 1 - 8: Paints and varnishes Corrosion protection of steel
structures by protective paint systems 1998
[6]
Eilers, M., Hemmert-Halswick, A.: Seilverfllmittel Mechanische Randbedingungen

fr Brckenseile. Bericht der Bundesanstalt fr Straenwesen, Brcken- und
Ingenieurbau Heft B 16. Wirtschaftsverlag NW Bremerhaven 1997
[7]
Vermerk BSG des LVR Kln 1988
Experimental Analysis of the Active Tendon Control of a Large-Scale

Cable-Stayed Bridge Mock-up.
Georges MAGONETTE
ELSA Laboratory,
Ispra (VA), Italy.
Carl HANSVOLD
JOHS.HOLT A.S.,
Oslo, Norway.
Vito RENDA
Yves. BOURNAND
Alan G. JENNER
Heino FSTERLING
ELSA Laboratory,
Ispra (VA), Italy.
VSL France,
Massy, France.
Newlands Technology Ltd, , Mannesmann Rexroth AG, VT Hull, GB.

Lohr, Germany.
Summary
A large-scale cable-stayed bridge mock-up has been constructed and installed at the ELSA
Laboratory of the JRC-Ispra. This mock-up has been equipped with active tendon actuators
to increase the structural damping in order to mitigate vibrations. The testing campaign in
preparation at ELSA aims to validate the control strategy for active damping developed at the
Universit Libre de Bruxelles and to analyse the behaviour of the most critical element of the
active damping system: the actuators. This paper is part of a trilogy [1, 2] submitted to this
conference by the consortium involved in the Brite-Euram project ACE [3]. It is focused on
the mock-up description and on the preparation of the experimental verification of the active
system installed on the bridge mock-up. The testing activity is currently in course and the
experimental results will be presented at the conference.
1.
Introduction
Improvements in materials and computational technology have led to progressively longer

and more slender cable-stayed bridges (Fig. 1 and 2). Long inclined cables are characterised
by low structural damping and low natural frequencies. They are flexible and light and prone
to vibrate under dynamic disturbances from wind and traffic. Stay cables are designed to be
tension elements that can support loading up to about 50% of the ultimate strength of the steel
elements used for the stay. From about 1977 several events of cable vibrations have been
observed in various types of cable-stayed bridges. During stay vibrations some stays can
reach peak amplitudes of oscillation of more than five times the stay diameter. During these
vibrations, the resulting stresses in the stay are unknown. The state of stresses in the steel
element is critical in the anchorage region due to bending stresses with effect of fatigue and
fretting in this area. Damage of anchor details and cable ducts has been reported.
Much progress has been made in bridge dynamics during the last 10 to 20 years. In several
cases the exciting mechanisms have been identified, and engineers and scientists have
progressively got a better understanding of the problems. A frequent reason for cable
vibrations appears to wind in combination with rain: the so-called rain-wind induced
vibrations. But also wind and traffic loads may give parametric excitation of the cables
resulting in unacceptable oscillations.
To prevent such vibrations, passive measures like cross-ties interconnecting the stay cables or
dampers installation at the bridge deck have been widely used. But some problems have
occurred with these systems. The initial tension of the cross-ties must be selected with care in
order to avoid detensioning and shock effects in the cable system. Viscous dampers located
near the cable anchorage at bridge deck have a limited damping effect, in particular in the
case of parametric excitation.
For long cables, the active damping strategy may be applied. The aim of the active control
system is to upgrade the damping of the structure and consequently to mitigate the induced
vibration of the stay cables. The methodology considered here is based on an active tendon
consisting of an actuator collocated with a force sensor. The active damping is based on the
control of the displacement of the cable anchor point. This technique developed at the
Universit Libre de Bruxelles [4, 5] has a strong physical support and its effectiveness has
been confirmed experimentally by tests performed on small-scale mock-ups.
Figure 1: The Skarnsundet cable-stayed

bridge in Norway (Trondheim).
Figure 2: The Normandie cable-stayed

bridge in France (Le Havre).
The computer simulation software developed in the ACE project takes into account the cable
dynamics, its interaction with the structure and the possibility of active control of some of the
cables. The main objectives of the experimental analysis of the cable-stayed mock-up are to
provide benchmarks for checking the validity of the dynamic computer model and to study
the effect of active control applied to a representative model of a bridge structure. The
potential interaction between cables and structure will be analyzed by the tests achieved on
the mock-up. The validation of the active control system prototype on a large-scale mock-up
will give us a better knowledge of the non-linear dynamic behavior of the cables and of the
real loads in the cables and the anchorage.
The proposed design and testing of the large-scale bridge mock-up are outlined here. After
the description of the mock-up and of the control actuators, details are given on the
measurement system and on the testing campaign in preparation to assess the performances of
the active tendon system. The experimental results not yet available at the moment will be
presented at the conference.
2.
Mock-up description
The bridge mock-up has been designed by Bouygues. It is a cable stayed cantilever beam that
will basically represent a cable-stayed bridge under construction. The deck, about 30-m long
(which is the maximum dimension allowed in the ELSA laboratory), is mainly composed of
two H-beams whose axis are spaced 3.0-m apart (Fig. 3). They are appropriately linked each
3.5-m with transversal H-beams to provide to the whole structure sufficient transverse and
torsion stiffness. Each H-beam is fixed to the Reaction Wall. The vibration excitation source
is anchored at the free end of the deck. Four pairs of parallel stay cables support the deck.
Each2 stay-cable is composed of one T13 strand (7 non-circular wires, total section of 112mm ) with a slope of 1/3. This slope is very closed to that of the longest stay cable of modern
bridges. A couple of secondary tie-cables will be inserted in order to study the control of
transverse vibrations of the stay-cables. The static tension being about 70 kN, the first free
frequency of the longest stay cables (29.5 m long) is around 5.7 Hz, which can be rarely met
on modern cable stayed bridges. To give the stay cables enough sag and consequently reduce
their free vibration frequencies, they are heavily overloaded with split lead cylinders. This
increases their average mass to an amount of about 15 kg/m. In this way, the sag of the
longest stay cables is about 0.8 % of their length and their first free frequency in the vertical
plane is closed to 1.2 Hz.
By positioning the intermediate

support under the deck, the vibration
frequencies of the whole structure
can be varied.
Figure 3: large scale cable-stayed bridge

mock-up in construction at the ELSA
Laboratory of the JRC-Ispra.
It is possible to adjust the stay cables length in order to obtain a very critical situation for the
structure: the first vertical free frequency of the complete structure (flexion) can be very close
to that of the longest stay cable. By positioning an intermediate support under the deck, the
vibration frequencies of the whole structure can be varied and eventually adjusted to be very
near to the double value of the first vertical free frequency of the longest stay cable. Similarly
by modifying very slightly the length of the stay cables or by reducing the amount of
additional masses gripped on the stay cables, it is possible to merge or to separate modal
frequencies and to create critical situations for the structural behaviour. As designed, the
mock-up will allow the complete analysis of numerous particular situations. Without the
intermediate support under the deck, the first modal frequencies of the structure are estimated
to 1.10 Hz for the first flexional vertical mode, 1.17 Hz for the first torsional mode and 3 Hz
for the first flexional horizontal mode.
3.
Actuator description and tests
3.1
Hydraulic actuator
The main function of the hydraulic actuator is to track the displacement command required by
the damping law. As the actuator is mounted between the bridge deck and a stay cable it has
to carry both the static and the quasi-static loads (bridge deck and live loads). Energy-saving
concerns led to the design of an actuator integrating two functions within one device: an
asymmetric large cylinder (pressurised by an accumulator) compensates the static load while
a smaller double rod cylinder provides the displacement necessary to satisfy the damping law
(Fig. 4).
Figure 4: Hydraulic Actuator (Principle)

The manufacturer carried out the preliminary tests. A special set-up was designed to analyse
the dynamic performances of the hydraulic actuator and to fit its numerical model.
Nevertheless the response of the actuator depends of the external loads and disturbances.
Therefore it is necessary to incorporate the actuator into the structure (bridge mock-up) in
order to:
assess the functioning of the actuator under real load conditions including the spring-mass
behaviour of the cable and the lateral forces acting on the actuator head.
optimise the setting of the control loops.
test the auxiliary functions (soft start/stop, monitoring functions, ...) and the fail-safe
situations.
measure the energy consumption and establish an energy balance.
achieve the measurements necessary for the validation of the actuator/ bridge numerical
model.
investigate alternative electro-hydraulic designs.
During the experimental tests all important state values describing the dynamic behaviour of
the electro-hydraulic system will be measured and recorded for documentation. This includes
the command signal, the cylinder load cell signal, the servo-valve spool position, the cylinder
displacement, the oil pressure in the cylinder chambers and in the accumulator, the oil supply
pressure and the oil flow. The electro-hydraulic control sensors provide most of these
measurements. The frequency filter designed to separate the static and dynamic components
of the force acting on the cylinder will be fine-tuned to an optimum transfer function.
Moreover, as we assume that the performances of the prototype actuators in terms of dynamic
requirements are overestimated, the experimental analysis should lead to optimised final
specifications and alternative hydraulic concepts should be established to reduce the energy
consumption to a minimum.
Figure 5: Mannesmann Rexroth double-function electro-hydraulic actuator for stay cable

actuation.
3.2
Magnetostrictive Actuator
In engineering applications, the term magnetostriction is used to describe the magnetic-field

induced strain exhibited by magnetic materials. For actuator and transducer applications, e.g.
sonar and active vibration control, the material Terfenol-D, (typically Tb0.3Dy0.7Fe1.95) is of
significant technological importance. This material is characterised by large magnetostrains,
typically 1000-1600 ppm, with a high intrinsic force capability and is available commercially.
A variety of devices, resonant and non-resonant, have been designed and constructed to
harness the properties of Terfenol-D [6, 7]. Previous work on a non-resonant actuator with the
highest possible bandwidth resulted in Europes largest actuator to date. This was capable of
generating a net ac force of ~ 9 kN up to 400 Hz with a 20 mm diameter active element for 6
kW of input electrical power [8]. Each application usually has a set of unique specifications.
These may include large displacements, minimum power consumption, etc. Compliance of
the load has an effect on actuator forces and displacements. The design has to be chosen
according to the actuators ultimate use.
Within the remit of the ACE project two magnetostrictive transducer systems have been
designed and are in the process of being fabricated for incorporation into a mock-up bridge
test facility. These are intended for active damping trials of unwanted vibrations in the mockup. This aspect of the project was deemed to be highly innovative, both in terms of the
technology to be employed and the scale of the actuators to be produced. These represent the
largest such actuators to be constructed in Europe, to date.
Each transducer consists of two Terfenol-D rods (30-mm diameter, 254 mm long) and a
surrounding ac drive coil (with a limited dc biasing capability). Surrounding each drive coil is
a ring-magnet stack supplying a static biasing field, which enables each transducer to operate
within a linear region of the magnetic field strain loop. Without lever system amplification,
the transducers would be expected to achieve ~ + 0.1mm of strain with an associated force
~13kN for an associated drive current of + 4 A. By using a lever system this is to be altered to
achieve a maximum strain of ~ + 2 mm with an associated force of ~ + 0.64kN (for a gain of
20:1). Prestress biasing, necessary for optimum magnetomechanical performance, is through
the use of externally located disc springs which can be easily interchanged to alter
compliance and maximum pressure loading (up to ~40MPa) (Fig. 6).
After fabrication, each device will be characterised to enable actuator integration into a local
control-structure prior to inclusion in the bridge mock-up (Fig. 7). A Instron
tensile/compressive test rig has been modified to allow such a characterisation (Fig. 8). An
example of the tests envisaged are as follows; conduct impedance and strain analyses under
various internal prestresses and drive frequencies with and without external mechanical to
determine optimal prestress. Characterise the performance of device for transient response.
0.5 m
Figure 6: Schematic of magnetostrictive actuator
Figure 7: System configuration and attachment
Instron crossbeam
Load cell
Laser probe
positions
Variable
k lever
arm adapter
~80%
extant
Ability to change
transducers
location required
Instron base
Transducer
support base
Figure 8: Transducer characterisation set-up
4.
Exciter
The mock-up will be excited by a shaking devices located at the free end of the deck. The
forced vibrations will be obtained by means of electro-hydraulic device operating in a
frequency sweep excitation manner. This kind of excitation, where the input force can be
perfectly monitored and measured, is the most suitable to perform experimental modal
analysis. The exciter will produce a unidirectional force with a continuously adjustable
amplitude and frequency in the range of 1 to 50 Hz. The total mass of the structure being
about 10,000 kg, the maximal exciting force value has been estimated in the range of 1.5 to
2.5 kN. The objective is to get, in all critical situations listed in chapters 2 and 5, a vertical
vibration of the deck with an amplitude close to 25 cm peak to peak without any active
tendon control system; this could not bring any part of the mock-up in a dangerous situation.
The electro-hydraulic exciter, especially designed for this application, is equipped with
modular inertial masses. Loaded with a 450-kg mass it is able to generate a sinusoidal inertial
force of 2.5 kN at a frequency of 1 Hz. For a given frequency, changing the displacement
amplitude or the inertial masses can vary the exciting force.
Three positions have been selected for the installation of the exciter on the transversal beam
located at the free end of the deck:
a central vertical position to excite the vertical flexional modes,
a decentralised vertical position to excite both the flexional and torsional modes,
5.
a central horizontal position to excite the torsional modes. In this last case there will be
also a quasi-static excitation on the flexional transversal modes with a low amplitude
response.
Testing campaign
The specific objectives of the testing campaign are:

to improve the understanding of induced vibrations (the mock-up will be subjected to
forcing functions),
to validate the numerical tools for prediction of dynamic behavior of cables,
to verify the capability of the active system to mitigate the effects of induced vibrations,
to evaluate in detail the performances and the reliability of the whole implementation.
The testing campaign will begin with modal analysis of the structure equipped with the cables
without additional masses. Performing frequency sweep excitation and free vibration tests the
first vertical mode ( 1.1 Hz), the first torsional mode ( 1.17Hz) and the modal damping
coefficients will be determined accurately. These values will be used to adjust the numerical
model and to compute the masses of the lead cylinders to fix on the longest cables.
Afterwards, following the same procedure, the structure equipped with the overloaded cables
will be characterised. In addition, the tension of each of the four longest stay cables will be
accurately adjusted to obtain the same vibration frequency for the two cables of each pair.
The vibration frequency and the damping coefficient of each cable will be determined by
applying an excitation at half-length with the help of a small electro-dynamic exciter. The
measurements will be carried out with accelerometers or laser scanning systems.
Successively, by adjusting the structural parameters of the mock-up, modifying some
coefficients of the control law or changing the force loading conditions, a large spectrum of
critical configurations will be investigated. In a very close frequency interval, with the
adopted adjustments, the following cases can be found (assuming that stay cables #4 are the
longest ones): The first vertical free frequency of the complete structure (flexion behaviour)
can be made very close to that of the longest stay cable. By modifying very slightly the length
of the stay cables, it is possible to clearly separate the frequencies of the first vertical mode of
stay cables #4 and the first vertical mode of the structure. Similarly, it is possible to nearly
merge the frequencies of the first vertical modes of both stay cables #3 and #4 and the
complete structure. Positioning the intermediate support under the deck, between the lower
anchor heads of stay cables #2 and #3, the first free frequency of the whole structure can be
very near from the double value of the first vertical free frequency of stay cable #4; this is an
other situation considered as critical for the structure behaviour. By reducing the amount of
additional masses gripped on the stay cables, the first free frequency of the stay cables #4 can
be very near from two times the first vertical free frequency of the structure, which is an other
critical situation (sub-parametric excitation). Finally, the efficiency of secondary tie cables
with or without active tendon control systems will also be tested.
As designed, the mock-up will allow the complete analysis of numerous critical situations.
This mock-up is unique to improve our knowledge in stay cable dynamics. While substantial
progress has been made in the study of components of active damping systems, little attention
has been paid to the overall performance of the system applied to a realistic structure. The
structural control system consists of a number of important components such as sensors,
controllers, actuators, and power generators that must be part of an integrated system.
Moreover, a number of implementation-aspects must be addressed such as intermittent and
fail-safe operations, integrated safety, reliability and maintenance. These issues require
experimental verification under realistic conditions.
6.
Instrumentation
Special care has been assigned to the selection of the most appropriate dynamic testing
techniques and to the selection of the transducers, including their conditioning electronics.
Measurement equipment will include conventional instrumentation consisting of temposonic
and inductive displacement transducers, accelerometers, and strain gauges. Load cells will be
installed in series with the stay cables and the secondary cables. Laser displacement sensors
will be used to monitor the lateral displacements of the deck. To measure the tendon
vibrations other techniques will be considered such as line-scan camera and a laser scanning
systems currently in development at the JRC.
Depending of its frequency spectrum, each signal will be acquired with a constant sample
frequency included between 50 and 1000 Hz. Each test has duration from 6 to 10 minutes.
All the measured values will be recorded continuously on hard disc and will be processed
subsequently. To help the users to interact with the experiment, some measures such as the
cables tension will be displayed in real-time.
7.
Conclusion
A large-scale cable-stayed bridge mock-up has been constructed and installed at the ELSA
Laboratory of the JRC-Ispra. This mock-up has been equipped with active tendon actuators to
increase the structural damping in order to mitigate vibrations.
The bridge mock-up and the exciter system have been designed to allow an independent
adjustment of all the parameters of the controlled system:
the excitation force in amplitude, frequency or direction (flexion and/or compression),
the vibration modes of the structure (by moving an auxiliary support or changing the
cables tension),
the number of active devices (1 or 2),
the gain of the control law,
the delay between the activation of the control and the start of the exciter.
The experimental tests will be repeated with different loading conditions to provide the
reliable data necessary for the validation of a numerical model that includes the structural
dynamics, the control system and the actuator dynamics. The deliverables will help the
various industrial involved in cable-supported structures to better understand the behaviour of
the structures when exposed to vibrations induced by wind, live load, or seismic phenomena.
Acknowledgements
Funding under the EC Brite-EuRam programme (Contract No. BRPR-CT97-0402) is
gratefully acknowledged. The authors acknowledge the contributions of their colleagues at
DERA, Farnborough, UK; BOUYGUES BTP, St Quentin en Yvelines, France; Technische
Universitt Dresden, Germany; and Universit Libre de Bruxelles, Belgium.
Reference
[1]
A. Preumont, S. Helduser, H. Foersterling, N. Galloway, "Active Tendon Control of

Cable-Stayed Bridges: Control Strategy and Actuator Design", IABSE Conference on:
Cable-Stayed Bridges - Past, Present and Future. Malm, June 2-4, 1999.
[2]
C. Dumoulin, "Active Tendon Actuators for Cable-Stayed Bridge", IABSE

Conference on: Cable-Stayed Bridges - Past, Present and Future. Malm, June 2-4,
1999.
[3]
ACE Project funded by the European Community under the Industrial & Materials
Technologies, Programme Brite Euram 3, Proposal N.BE96-3334, Contract N.BRPRCT97-0402, ACE: Active Control in Civil Engineering (1998).
[4]
F. Bossens & A. Preumont, "Experiments of active control of cable structures",

MOVIC-4, Zurich, August 1998.
[5]
Y. Achkire, "Active Tendon Control of Cable-Stayed Bridges", Universit Libre de

Bruxelles, Faculty of Applied Sciences, Department of Mechanical Engineering and
Robotics, Thesis for the degree of Doctor in Applied Sciences (1997).
[6]
R.D. Greenough, I.M. Reed and M.P. Schulze, Magnetostrictive Actuators in

Advances in Actuators (Institute of Physics, London) Ed. A.P. Dorey and J.H.
Moore, chapter 8, pp179-211 (1995).
[7]
F. Claeyssen, D. Colombari, A. Tessereau and B. Ducros, Magnetostrictive

Actuators , IEEE Trans. Mag 27(6), p5343 (1991).
[8]
M.G. Aston, R.D. Greenough, A. G. Jenner, W.J. Metheringham and K. Prajapati,

Controlled High Power Actuation Utilising Terfenol-D , Proc. the 3rd International
Conference on Giant-Magnetostrictive Materials, Honolulu, Hawaii, USA, November,
1996. J. Alloys and Compounds 258 (1-2), pp.97-100 (1997).
Vibration Control of Stay Cables

J. STUBLER
Technical Director
Freyssinet International
Vlizy (France)
P. LADRET
Principal Engineer
Vlizy (France)
J.B. DOMAGE
Principal Engineer
Freysssinet International
Vlizy (France)
Summary
During the recent past years several analyses have been conducted dealing with the vibration
characteristics of stay cables. The fundamental theories as well as the fundamental behaviour
of various types of cables have been developed. At the same time, various remedies and
vibration control devices were proposed by contractors and suppliers.
This paper reviews the previously used systems and presents the up-to-date technology which
is available today. It covers the streamline sheath coping with the rain and wind vibration
phenomena, the internal or external hydraulic dampers, the visco-elastic dampers and the
damping cross ropes. Calculations of the damping system characteristics, prediction and
measurement of the damping ratio are presented.
1. Introduction
Cable vibrations can be excited by dynamic wind forces acting directly on the cable itself or
by the movements of the cable attachments on the pylon or on the deck due to the action of
traffic loads or of the wind itself.
Four different sources of vibrations are considered in the analysis :
- parametric excitation by the movements of the pylons and the deck ;
- rain and wind vibration ;
- low wind dry vortex ;
- galloping.
2. Damping definition
a
The logarithmic decrement is defined as = ln n where a n+1 and a n are two
an +1
consecutives amplitudes of the vibration of a system left free to vibrate.
Figure 1
Usually the stay cables have a high fatigue resistance but, consequently, low natural damping
characteristics.
3. Damping analysis
3.1 Excitation due to movement of pylon and deck
Calculations of the amplitude of vibrations have to be done for each wind speed and each mode
of the structure. The root mean square amplitude is then calculated.
Two criteria should be considered :
1. Appearance and comfort requirements (at 15 m/s wind speed).
a
Amplitude a should be limited to
L = cable length
< 0,0006 .
L
2. Admissible deviation angle at the anchorage : depending on the cable system, is the
admissible deviation angle at the anchorage.
a
Amplitude a should be limited to < .
L
3.2 Rain and wind vibrations
The presence of a double helical rib will reduce the amplitude of stays. This has been
demonstrated by the two wind tunnel tests carried out at the CSTB laboratory (Nantes) in 1992
for the Normandie bridge and at the Danish Maritime Institute in 1997 for the resund bridge.
Figure 2
In case of no helical ribs, a stability criteria could be deducted from SAITO [1]
Sc =
m
> 10
2D 2
m = cable mass per unit length
= logarithmic decrement
= air density
D = cable diameter
Sc is the SCRUTON number or the mass damping parameter.

For typical stay cable (mass and diameter) the requested logarithmic decrement should be
higher than 3 to 6 %.
3.3 Low wind dry vortex
This kind of vibration, induced by vortices around the cable profile, appears at a certain wind
speed V such that
ND
V=
0,22 is the STROUHAL number
0,22
N = frequency of cable vibration.
This phenomenon excites only the high frequencies in low wind conditions (small amplitudes
of vibrations). This is not a major concern in most cases and a logarithmic decrement > 3 %
should be recommended.
3.4 Galloping
This phenomenon has been observed apparently in Japan for high wind condition [1].
If it has never been clearly identified on a bridge, one reason could be the fact that at high
wind speed, the value of the aerodynamic damping is high. It is interesting to look at some
recent experiences :
a) Second Severn bridge
There could be two interpretations of the phenomenon which occurs on January 1995. One of
them is that it was galloping , the other is that it was excitation of the deck from the wind
which was exciting the stay cable (the Second Severn stay cables have a low frequency as the
stay cables have a low permanent stress). The vibration amplitudes of the stay were huge.
This level of vibration is a first indication that it could be a deck induced excitation. In
addition to that, the mode of the deck was a frequency similar to the stay cable one.
After the erection of the cross ties on that bridge, a second vibration event occurs ; at that time
the cables were stable but the deck has some higher vibration.
So it really seems to be that it was an excitation of the deck. When the cross ties were
installed the energy of the deck was no more shared with the stay cable.
b) Normandy bridge
m
is coming from an extrapolation of an article by
D 2
SAITO [1] published at the Normandy Symposium (October 1994) with C = 35 to 40. If this
criteria is applied on the existing bridge we should have experienced huge galloping.
A proposed criteria Vcrit = CND
damping ratio - to - critical

2
During construction longest stay : Vcrit = 40 0,3 0,2
70 1,6 10 3
= 3,62 m/s.
1,23 0,2 2
c) Elorn bridge
. before placing dampers : Vcrit = 40 0,63 0,2
92 1,27 10 3
= 7,76 m / s
1,23 0,2 2
. after placing dampers : Vcrit = 19,42 m/s.

No galloping has been observed. These observations tend to prove that the SAITO results are
questionable. No doubt that more research is needed on that subject of galloping.
4. Damping technologies
4.1 Damping ropes
The natural frequency of the stays can be modified by means of transversal cables connected
to them. This solution which is effective although expensive and delicate to install has been
used for some large bridges. It is recommended when the vibration frequencies of the deck or
pylon are close to the frequencies of the stay cables. For the Normandie Bridge, the main
vibration period for vertical bending would have been of the same magnitude as the vibration
period of the longer cables i.e. 4.5 seconds. The cross ropes reduce the periods of the cables to
1.25 second and less (ref. [2]). Four ropes connect all cables in each plane of stays. Their
tension was selected to avoid slackening effects. They consist of a bundle of stainless steel
wire ropes fully embedded in HDPE and other dampening plastic material. They went through
a testing programme for fatigue and damping capacity. A steel collar provides the fixation of
the ropes on each stay and facilities for any orientation.
4.2 External hydraulic damper
This damper is specifically designed to each project. The damping capacity can be tuned to
obtain the required logarithmic decrement. However it requires a regular maintenance and it is
not always meeting the aesthetics objective of the designer.
4.3 Internal visco-elastic damper (IED) and Internal hydraulic damper (IHD)
This damper is completely invisible from the deck since it is located inside the steel guide
pipe of the stay cable.
Figure 3
5. Vibration control
Calculation models have been developed to evaluate the logarithmic decrement provided by
the various types of damping systems. A universal damping surface has been established
allowing an accurate tuning of the damper. On the other hand damping measurement can be
carried out on the cables after installation. Examples of Tagus bridge in Lisbon (Portugal) and
Charles River bridge in Boston (USA) are shown here below :
Charles River bridge (Boston)
Tagus bridge Lisbon (Portugal)
REFERENCES
[1] Saito T., Matsumoto M. and Kitazawa M. (1994) : Rain-wind excitation of cables on
cable-stayed Higashi-Kobe bridge and cable vibration control, Proceedings of
Conference on Cable-stayed and Suspension bridges / Deauville (France), 12-15 October
1994, pp. 507-514.
[2] Virlogeux M. (1995) : Design of cables for cable-stayed bridges : the example of the
Normandie bridge, Proceedings of the International Symposium on Cable Dynamics /
Lige (Belgium), October 19-21.
[3] Virlogeux M. (1998) : Cable vibrations in cable-stayed bridges, Proceedings of Bridge
Aerodynamics Symposium / Copenhagen (Denmark), 10-13 May 1998.
[4] Matsumoto M. (1998) : Observed behavior of prototype cable vibration and its
generation mechanism, Proceedings of Bridge Aerodynamics Symposium / Copenhagen
(Denmark), 10-13 May 1998.
Active Tendon Control of Cable-Stayed Bridges:

Control strategy and Actuator Design
A. PREUMONT
Bruxelles, Belgium
S. HELDUSER
Technische Universitt
Dresden, Germany
F. BOSSENS
Bruxelles, Belgium
R. BONNEFELD
Technische Universitt
Dresden, Germany
H. FRSTERLING
Mannesmann Rexroth
Germany
Summary
This paper is part of a trilogy describing the Brite-Euram project ACE. The first part of the
paper describes the control strategy for active damping of cable structures with an active tendon
collocated with a force sensor. The main analytical results for predicting the closed-loop poles
are summarized and the procedure for selecting the number and the location of the active tendons
is outlined. The second part of the paper describes a laboratory experiment with a small size
mock-up representative of a cable-stayed bridge during its construction phase. The control of the
parametric vibration of passive cables due to deck vibration is demonstrated. Finally, the third
part of the paper outlines the conceptual design of an hydraulic actuator for industrial
applications.
1. Introduction
In recent years, cable structures have had spectacular applications in large cable-stayed bridges.
However, cable and deck vibrations have become a major design issue, because the ever
increasing span of the bridges makes them more sensitive to flutter instability as well as to wind
and traffic induced vibrations. An overview of the stay cable vibration and control is available in
[1]. The difficulty of damping the stay cable vibration justifies the development of active devices
for future large bridges.
This paper is part of a trilogy describing the current status of the Brite-Euram project ACE
[2,3,4]; it is a follow-up to several papers where the theory was gradually developed
[5,6,7,8,9,10]. Its goals are to confirm and extend previous results obtained with a laboratory size
mock-up, and to outline the design of an hydraulic actuator as required for a large scale
application. It is organized as follows: section 2 summarizes the control strategy for active
damping of cable structures, section 3 states without proof approximate analytical results for the
closed loop poles. The experimental results obtained with the laboratory mock-up are described
in section 4 and the conceptual design of the hydraulic actuator is outlined in section 5. Section 6
gives some concluding remarks.
2. Control of Cable Structures

2.1. Tendon control of strings and cables
The mechanism by which an active tendon can extract energy from a string or a cable is
explained in Fig.1 with a simplified model assuming only one mode (Rayleigh-Ritz), for
situations of increasing complexity. The simplest case is that of a linear string with constant
tension To (Fig.1.a); the equation becomes nonlinear when the effect of stretching is added (cubic
nonlinearity). In Fig.1.b, a moving support is added; the input u of this active tendon produces a
parametric excitation, which is the only way one can control a string with this type of actuator.
The difference between a string and a cable is the effect of gravity which produces sag (Fig.1.c).
Fig.1: Mechanism of active tendon control of strings and cables

In this case, the equations of motion in the gravity plane and in the plane orthogonal to it are no
longer the same, and they are coupled. In the gravity plane, the active tendon control u still
appears explicitely as a parametric excitation, but also as an inertia term c u whose coefficient
c depends on the sag of the cable; even for cables with moderate sag (say sag to length ratio of
1% or more), this contribution becomes significant and constitutes the dominant control term of
the equation. On the contrary, in the out-of-plane equation (y coordinate), the tendon control u
appears explicitely only through the parametric excitation, as for the string.
The use of the parametric excitation to damp the transverse vibration of a string was explored by
J.C. Chen [11]; the same strategy was used to control the out-of-plane vibration of cables by Y.
Fujino [12] who also investigated the use of the inertia term for active damping of cables in the
gravity plane [13]. These attempts used non-collocated feedback of the transverse amplitude;
they worked well when the interaction of the cable with the supporting structure was weak, but
they became unstable when the interaction was strong. On the contrary, the approach followed in
the present study, which is based on collocated actuator/sensor pairs, is guaranteed to stabilize all
the states which are controllable and observable.
2.2. Control strategy

It is widely accepted that the active damping of linear structures is much simplified if one uses
collocated actuator-sensor pairs [7]; for nonlinear systems, this configuration is still quite
attractive, because there exists control laws that are guaranteed to remove energy from the
structure. The direct velocity feedback is an example of such "energy absorbing" control. When
using a displacement actuator (active tendon) and a force sensor, the (positive) Integral Force
Feedback
(1)
u = g T dt
Fig.2: Active damping of cable structures

(refer to Fig.2.a for notations) also belongs to this class, because the power flow from the control
system is W = T u = gT 2 . This control law applies to nonlinear structures; all the states that
are controllable and observable are asymptotically stable for any value of g (infinite gain
margin).
2.3. Experiment
The foregoing theoretical results have been confirmed experimentally with a laboratory scale
cable structure similar to that represented schematically in Fig.2.a, where the active tendon
consisted of a piezoelectric actuator [6]. Figure 2.b shows the experimental frequency response
between a force applied to the structure and its acceleration; also shown in the figure is the free
response of the structure with and without control. We see that the control system brings a
substantial amount of damping in the system, without destabilizing the cable (theoretically, the
control system does indeed bring a small amount of damping to the cable, which depends on the
sag, as we have seen in the previous section); this behaviour is maintained at the parametric
resonance, when the natural frequency of the structure is twice that of the cable.
2.4. Decentralized control

The foregoing approach can readily be extended to the decentralized control of a structure with
several active cables, each tendon working for itself with a local feedback following Equ.(1).
This statement was verified experimentally on a T structure controlled with two cables [8]. It is
important to point out that the concept of active tendon control of cable structures does not
require that all the cables be active; on the contrary, the control system would normally involve
only a small set of cables judiciously selected. Next section summarizes the main results of an
approximate linear theory to predict the performance of the control system and provides design
guidelines to select the active cables.
3. Closed-loop Poles
If we assume that the dynamics of the active cables can be neglected and that their interaction
with the structure is restricted to the tension in the cables, it is possible to develop an
approximate linear theory of the closed-loop system.
Fig. 3: Root locus of the closed-loop poles

For a decentralized feedback control law
g
= K c1T
(2)
s
where T is the local force measurement, is the active tendon displacement, Kc is the stiffness of
the active cable (Kc-1T represents the elastic extension of the active cable) and g is the control
gain (the same for all control elements), the following results have been established in earlier
studies [6]:
1. If we assume no structural damping, the open-loop zeros are ji where i are the natural
frequencies of the structure where the active cables have been removed.
2. The open-loop poles are ji where i are the natural frequencies of the structure including
the active cables.
3. As g goes from 0 to , the closed-loop poles follow the root locus corresponding to the openloop transfer function (Fig.3)
s 2 + i2
G (s) = g
(3)
s s 2 + i2
Thus, the closed-loop poles go from the open-loop poles at ji for g=0 to the open-loop
zeros at ji for g .
4. The depth of the loop in the left half plane depends on the frequency difference ii and the
maximum damping, obtained for g = i i / i , is
(
(
)
)
i i
(4)
2 i
5. For small gains, the modal damping ratio resulting from the active tendon control is given by
g i
i
(5)
2 i
where i=(2i-2i )/2i is the modal fraction of strain energy in the active cables.
imax =
Equations (4) and (5) can be used very conveniently in the design of actively controlled cable
structures.
4. Experiment
The test structure is a laboratory model of a cable-stayed bridge during its construction phase,
which is amongst the most critical from the point of view of the wind response. The structure
consists of two half decks mounted symmetrically with respect to a central column of about 2 m
high (Fig.4); each side is supported by 4 cables, two of which are equipped with active
piezoelectric tendons (Fig.5). The maximum stroke of the active tendon is about 100m.
Fig.4: Experimental set-up for the cable-stayed bridge (the small picture shows the Skarnsund
bridge Norway in its construction phase)
Fig.5: Design of the active tendon

4.1. Closed-loop poles
Figure 6 shows the evolution of the first bending and torsion closed-loop poles of the deck when
the control gain increases (these poles have been obtained with the MATLAB Frequency Domain
Identification Toolbox from frequency response functions). Also shown on the figure are the
predictions of Equ.(3); the agreement is good for moderate values of the gain. For larger gains,
when the modal damping exceeds 20%, the discrepancy between experimental and analytical
results increases, but this seems to be essentially related to the identification algorithm which
cannot distinguish oscillating poles from highly damped frequency responses (Fig.7). Note that
damping ratios significantly larger than 20% can be obtained. Figure 7 shows typical frequency
responses between the voltage applied to one of the active tendons and its collocated force
sensor, for several values of the gains (respectively g1, 3g1, 10g1). We note that for large gains
the resonant peaks of the structure have totally disappeared and that the one of the cable is also
considerably reduced.
Fig.6: Evolution of the first bending and torsion poles of the deck with the control gain
Fig.7: Frequency response between an active tendon and the collocated force sensor
4.2. Control of parametric resonance
In this experiment, the bridge deck is excited with an electrodynamic shaker and the tension in
the two passive cables on one side is chosen in such a way that the 1st in plane mode of one of
them is tuned on the global bending mode of the deck (f), while its neighbour is tuned on f/2, to
experience the parametric resonance when the deck vibrates. This tuning is achieved by
monitoring the cable vibration with a specially developed optical measurement system described
in [14].
Fig.8: Vibration amplitude of the bridge deck and the two passive cables at f and f/2
Fig.9: Detail of Fig.8 between t=10 and t=14 sec showing the transition from the forced
response at f to the parametric resonance at f/2
Figure 8 shows the vibration amplitude of the deck and the transverse amplitude of the in-plane
mode of the two passive cables when the deck is excited at resonance; the excitation starts at
t=5sec and the control is turned on after t=30 sec.
We note that :
1. The amplitude of the cable vibration are hundred times larger than the deck vibration.
2. The parametric resonance is established after some transient period in which the cable
vibration changes from frequency f to f/2. The detail of the transition to parametric resonance
is shown in Fig. 9.
3. The control brings a rapid reduction of the deck amplitude (due to active damping) and a
slower reduction of the amplitude of the cable at resonance f (due to the reduced excitation
from the deck).
4. The control suppresses entirely the parametric resonance at f/2. This confirms that a
minimum deck amplitude is necessary to trigger the parametric resonance.
5. Hydraulic actuator design

Although appropriate to demonstrate control concepts in labs, the piezoelectric actuators are
inadequate for large scale applications. For cable stayed bridges, the active tendon control must
simultaneously sustain the high static loads (up to 400 t) and produce the dynamic loads which
are one order of magnitude lower than the static ones (< 10 %). This has led to an active tendon
consisting of two cylinders working together: one cylinder pressurized by an accumulator
compensates the static load, and a smaller double rod cylinder drives the cable dynamically to
achieve the control law. The two functions are integrated in a single cylinder, as illustrated in
Fig.10; the double rod part of the cylinder is achieved by a rod in rod design.
Fig.10: Conceptual design of the hydraulic actuator

This solution saves hydraulic energy and reduces the size of the hydraulic components. The
double rod part of the cylinder is position controlled. The long term changes of the static loads as
well as the temperature differences require adaptation of the hydraulic conditions of the
accumulator.
A medium size actuator based on the foregoing concept will shortly be tested in the large mockup described in a companion paper [3].
6. Concluding remarks
The use of tension cables for active damping of cable structures has been investigated
theoretically and experimentally. The decentralized control approach proposed is simple, robust
and easy to implement ; in case of a sensor or actuator failure in an active tendon, the
corresponding control loop simply returns to its passive state. Simple formulae have been
developed for predicting the closed-loop poles; these formulae have been confirmed
experimentally and they can be used very conveniently in the design, for selecting the number
and location of the active tendons. The active control of a cable parametrically excited by the
deck has been demonstrated experimentally. The conceptual design of an hydraulic actuator for
large scale applications has been outlined; it will be used in the large scale experiment described
in [4].
7. Acknowledgments
This study was partly supported by the Brite-Euram project n BE96-3334: ACE (Active Control
in Civil Engineering) and the national program IUAP IV-24 on Intelligent Mechatronic Systems.
References
[1] Yamaguchi, H. and Fujino, Y., Stayed cable dynamics and its vibration control,
International Symposium on Advances in Bridge Aerodynamics, Copenhagen, May 98.
[2] ACE, Active Control in Civil Engineering, Brite Euram project n BE96-3334.
[3] Dumoulin, C., Active Tendon Actuators for Cable-Stayed Bridge, IABSE conference on
Cable-Stayed Bridges Past, Present and Future, Malm, June 1999.
[4] Magonette, G., Bournand, Y., Hansvold, C., Jenner, A., Experimental Analysis on a Large
Scale Cable-Stayed Mock-up, IABSE conference on Cable-Stayed Bridges Past, Present
and Future, Malm, June 1999.
[5] Ackire, Y. and Preumont, A., Active tendon control of cable-stayed bridges, Earthquake
Engineering and Structural Dynamics, Vol. 25, N6, June 1996, pp. 585-597.
[6] Preumont, A. and Achkire, Y., Active Damping of Structures with Guy Cables , AIAA, J. of
Guidance, Control, and Dynamics, Vol. 20, N2, March-April 1997, pp. 320-326.
[7] Preumont, A., Vibration Control of Active Structures: An Introduction, Kluwer Academic
Publishers, 1997.
[8] Achkire, Y., Active tendon Control of Cable-Stayed Bridges, Ph.D. dissertation, Active
Structures Laboratory, Universit Libre de Bruxelles, Belgium, May 1997.
[9] Preumont, A., Achkire, Y., and Bossens, F., Active tendon control of large trusses, 39th
SDM Conference, Long Beach, April 1998.
[10] Bossens, F. & Preumont, A., Experiments of active control of cable structures, MOVIC-4,
Zurich, August 1998.
[11] Chen, J., Response of Large Space Structures with Stiffness Control, AIAA, J. Spacecraft,
Vol. 21, N5, September-October 1984, pp. 463-467.
[12] Fujino, Y., Warnitchai, P., and Pacheco, B., Active Stiffness Control of Cable Vibration,
ASME, J. of Applied Mechanics, Vol. 60, December 1993, pp.948-953.
[13] Fujino, Y. and Susumpow, T., An experimental study on active control of planar cable
vibration by axial support motion, Earthquake Engineering and Structural Dynamics, Vol.
23, 1994, pp. 1283-1297.
[14] Achkire, Y. & Preumont, A., Optical measurement of cable and string vibration, Shock
and Vibration, Vol. 5, 1998, pp. 171-179.
CFRP-Tendons - Development and Testing

Dipl.-Ing. Frank ROOS
Lehrstuhl fr Massivbau
Technische Universitt Mnchen
Munich. Germany
Dr. J.F. NOISTERNIG

Dywidag Systems
International GmbH (DSI)
Munich. Germany
Frank Roos, born 1969, received his civil

engineering degree in 1997. He is now a
research assistant at the Lehrstuhl fr
Massivbau at the Technische Universitt
Mnchen conducted by Univ.-Prof. Dr.Ing. Konrad Zilch.
Johannes F. Noisternig, born 1966,

received his doctoral degree in
mechanical engineering in 1995. 1991 1995 he was a research assistant at the
Institute for Composite Materials (IVW)
at the University of Kaiserslautern. Since
1996 he is the head of the polymer group
in the technical department of DSI.
Summary
Carbon fibre reinforced plastic (CFRP) in form of wires is a material with very interesting
properties for stay cables or tendons like high tensile strength, high fatigue resistance as well as
low weight and excellent chemical resistance. The anisotropic composition is the main
disadvantage which makes it difficult to anchor the wires.
This paper gives a short overview of the properties of CFRP-wires, the requirements for stay
cables or tendons and the development of such a system by DSI as well as testing of stay cables
at the Lehrstuhl fr Massivbau of the Technische Universitt Mnchen.
Introduction
The use of carbon fibre reinforced plastics in the past was restricted to the aerospace and defence
industry. Besides the advantages of the strength / weight ratio and a high degree of chemical
inertness in most civil engineering environments this new building material has very high
material and manufacturing costs. Due to this fact the replacement of conventional civil
engineering materials like steel and concrete succeeded very slowly in the past. This has changed
since the high priced defence industry reduced its demand of these materials, advanced methods
of manufacturing were introduced and the prospect for large volume applications in the
construction industry appeared.
First considerations on CFRP-tendons were made in the early eighties, when the possibility of a
stay cable bridge over the Strait of Gibraltar, which is not possible if constructed in steel, was
discussed. On closer examination the anchorage of CFRP-tendons was found to be the most
important problem of this new building material was found [1]. The first applications have been
realized in Japan and the USA [2]. In the last years also in Europe intensive developments have
been started and lead to a first application of two CFRP-stay cables in the Stork Bridge in
Winthertur, Switzerland in 1996 [3]. In Germany the DSI in co-operation with the Lehrstuhl fr
Massivbau of the Technische Universitt Mnchen started a research project on CFRP-stay
cables or tendons with a suitable anchorage in 1996. Today the developed CFRP-system
DYWICARB is ready for applications as stay cables in bridges or as tendons in other civil
engineering constructions.
Carbon Fibre Reinforced Plastics
These materials consist of carbon fibres and a matrix. As carbon itself is very brittle the
composition combine the excellent mechanical properties of carbon fibres with the ductile
behaviour of the matrix, which is in most cases an epoxy resin. The reason for this change of
brittleness of carbon fibres is on one hand due to structural and on the other hand due to static
reasons. If a homogenous component consisting of carbon has a small notch in the surface or a
defect of micro millimetre dimension in the inside it may fail suddenly. If an individual fibre of
many fractures, the bundle will not break. When the fibres additionally are embedded in a matrix,
the fibre takes up loading again at both sides of the fracture point. So the composition of two
materials which on the first glance seems to be very complicated shows its advantages. These are
mainly a very low density, a high strength and stiffness, a strong resistance against aggressive
media and very good fatigue properties.
Carbon fibres which have a diameter between 7 and 9 m are mainly produced from
Polyacrylnitril (PAN) pre-fibres in a multiple process of heating and stressing. By using several
methods of manufacturing it is possible to achieve different properties of carbon fibres. The
tensile strength ranges from 2000 N/mm2 up to 4500 N/mm2 and the modulus of elasticity from
200 000 N/mm2 to 650 000 N/mm2. To produce CFRP-wires the fibres are embedded in a matrix
by pultrusion. Table 1 lists the properties of the CFRP-wire Carbon-Stress used for the
DYWICARB system and of conventional tensioning steel.
Diameter
Fibre volume content
Tensile strength
Modulus of elasticity
Elongation at rupture
Density
Thermal coefficient of expansion
Carbon-Stress CFRP-wire
5 mm
65 Vol.-%
2700 N/mm
160 000 N/mm2
1.6 %
1600 kg/m
0.2 x 10-6 K-1
tensioning steel
5 mm
->1670 N/mm2
205 000 N/mm2
6.0 %
7850 kg/m3
1.2 x 10-5 K-1
Table 1: Properties of CFRP-wires and tensioning steel
Requirements to CFRP-Tendons
CFRP-systems have in no way been standardised nationally or internationally up to now.

Therefore, it is very difficult to work out a valid table of requirements for CFRP-stay cables or
tendons. However, working groups in Japan (JSCE), USA (ACI 440) and Canada (CSA S806) as
well as Europe (fib task group 9.3) are striving to standardise materials, applications and
calculation methods. The fib task group 9.3 is preparing progress reports for 1999 for a public
discussion of such requirements.
Standardisation and characterisation of the material of CFRP-elements can roughly be based on
recommendations known from steel. However, it has to be clear that in contrast to steel, CFRP is
no homogenous material and thus different CFRP-elements also show different properties. For a
general characterisation of the material, not only the mechanical properties under static and
dynamic loading, but also durability as well as the behaviour under influences of different media
are of importance.
As the behaviour of a CFRP-stay cable can not be described exactly by calculation, tests have to
be performed for each project. The testing can be done according to the PTI recommendations for
Stay Cable Design, Testing and Installation [4]. With regard to the application of CFRP-tendons
generally a high static and dynamic capacity of the anchorage must be achieved to exploit the
material as far as possible. For the dynamic test the upper load range is 0.45 of the theoretical
ultimate load with an amplitude of 160 MPa. Normally the ultimate strength of a single CFRPwire can not be reached in stay cables with a large number of wires due to the anchorage
problem. The typical failure of CFRP-elements is a wire fracture in the first third of the
anchorage length. So the acceptance criteria of the PTI should be changed. For the ultimate
strength test of the stay cable after two million load cycles not the ultimate strength of one wire
but of the whole stay cable should be the reference for the acceptance. The failure load of a stay
cable after the dynamic test has to be 0.9 (this value has to be discussed in the task groups) of the
static ultimate load of a similar stay cable specimen. There should be no wire fracture during the
fatigue test.
For the evaluation of the durability of CFRP-systems, especially of anchorages, the lifetime
under permanent load as well as the behaviour under chemical influence have to be determined.
In pilot projects CFRP-systems have to be controlled by optic sensors or similar devices. The
installation should only be done by well educated stuff and companies with quality assurance as
well as supervision.
Development of the CFRP-Tendon
The key problem in the application of CFRP-tendons or stay cables is to anchor them. The best
way to do this is to anchor the CFRP-wires in conical steel hulls, filled with potting material (see
figure 1). The secret of a high load capacity is the composition of the potting material, and the
geometric form of the steel hull as well as the CFRP-wires themselves. The first step to a CFRPstay cable was the development of an anchorage for a single CFRP-wire [5].
Figure 1: Sketch of the anchorage

After the first successful tests the number of wires was increased up to seven. For this anchorage
numerical simulations were carried out [6]. With a FE-program a model was developed and
calibrated as well as verified through comparison with static load tests carried out at the
Technische Universitt Mnchen. The model was used to conduct parametric investigations with
the aim of reducing the critical transverse stresses in the anchorage. The efficiency factor for a
tendon with 7 wires rose from 69 % in the beginning up to 99 % in the end. Subsequently the
number of CFRP-wires of the tendons was increased to 13 and 19 wires and finally stay cables
with 91 and 103 CFRP-wires were manufactured. As potting material DYWIPOX CBV which is
a two-component epoxy resin system was used. Dynamic tests followed by static tests were done
on tendons with 7 and 19 wires and stay cables with 91 (as described in chapter 5.2) and 103
wires. The type of failure of these tendons is always the same. Several secondary failures in the
free cable length caused by shear fracture of the CFRP-wires in the first third of the potting hull
leading to the total collapse. While the tendons with 7 and 19 wires showed a sudden fracture of
the whole specimen the stay cable with 91 wires broke wire by wire.
After the problem of anchoring the wires had been solved the main task was to improve the
manufacturing process and the installation of the tendons and stay cables. Up to this time all
anchorages were potted in a vertical position which is very difficult to do on a building site. With
the new technique developed by DSI it is now possible to inject the potting material into the hulls
in a horizontal position. After successful testing on tendons with 7 and 19 wires a test on a stay
cable with 103 wires has been finished in January 1999.
The stay cables developed by DSI are parallel arranged wires over the whole length. To protect
them against ultraviolet radiation and wind erosion they are guided in a polyethylene or
polypropylene sheathing in their free length.
Testing of the CFRP-Tendon
Most of the testing on the CFRP-tendons and all tests on the CFRP-stay cables were done at the
Technische Universitt Mnchen. This chapter deals with the testing of a stay cable with 91
CFRP-wires.
5.1
Testing Equipment
For the testing of stay cables the Lehrstuhl fr Massivbau of the Technische Universitt
Mnchen, is equipped with a testing arrangement for dynamic and static loading up to an
ultimate load of 19 000 kN(see figure 2).
Figure 2: Sketch of the stay cable testing equipment
The testing machine consists of two reinforced concrete abutments with a centre hole through
which the cable is guided. The tensile force applied on the specimen is measured by three load
cells at the fixed abutment. Two sets of jacks each consisting of three equal jacks have to be
installed between the two abutments. One set of jacks is used for static loading of the specimen,
the other for the fatigue loading. Depending on the size of the cable specimen different sets of
dynamic jacks can be used for the fatigue loading. So all together there are three sets of jacks
available for the testing machine.
Shocks caused by wire fractures during the test are recorded by shock meters. The piston strokes
of the dynamic jacks are controlled by sine impulses. Due to the feasibility of individual
controlling of each dynamic jack, a rotation of the adjustable abutment is possible. This rotation
induces an eccentric loading of the cable specimen. An additional dynamic jack can be installed
to apply a transverse deviation at mid span position of the cable specimen.
The technical data of the testing machine are as follows:
static load capacity
at a maximum piston stroke of
dynamic load capacity
- big dynamic jacks
maximum frequency
(depending on the elongation of the specimen)
- small dynamic jacks
maximum frequency
(depending on the elongation of the specimen)
length of the cable specimen
inclining angle for grouting
rotation angle of the adjustable abutment
usual measurements during a stay cable test
- force applied on the cable specimen
- oil pressure of the jacks
- stroke of the pistons
- elongation of the cable specimen
- displacement of selected wedges and wires,
deflection of the wedge plates
additional measurements if requested
- strains in the wedge plates, trumpets
- displacement of clamps, tubes.
19 000 kN
240 mm
10 500 kN
9 mm
0.5 - 0.8 Hz
3 200 kN
20 mm
1.0 - 1.8 Hz
5.1 - 5.5 m
0 - 65 degrees
0 or 1 degree
0.5
0.1
0.01
0.01
%
%
mm
mm
0.01 mm
0.01 mm/m
0.01 mm
5.2
Testing
The cable specimen was manufactured by DSI in the laboratory i.m.b. in Utting, Germany, as
follows:
First the 91 wires were inserted into four spacers in a horizontal position and the ends were
cleaned properly. After that the cable specimen wires were put up in a vertical position and the
potting material was injected into the steel hulls where the wires run in parallel. First the northern
anchorage was filled. Then the cable was turned around and the southern anchorage was also
filled. To protect the cable against bending and damage throughout the whole process of
fabrication, transport and installation into the testing machine, the two steel hulls were connected
and fixed with two steel L-sections. The cable specimen was mounted in the testing machine at
the Technische Universitt Mnchen.
Figure 3:
Steel hull
Figure 4:
Potting material
Figure 5:
Manufacturing
A shock meter was installed in both anchorages which should record possible shocks caused by
wire fractures during the fatigue test. Besides these measurements, the load, the oil pressure of
the three jacks, the stroke of the pistons and the displacement of the adjustable abutment were
recorded during the whole test. After installation the fatigue test was started and proceeded
continuously until 2 million load cycles were reached. The maximum load was 45 % of
theoretical failure load. The aimed stress range was = 160 N/mm2. The average frequency of
the test was about 1.2 Hz. During the fatigue test no wire fracture occurred in the free length or in
the anchorage of the cable specimen. The displacement of the adjustable abutment (anchor block)
at maximum load increased due to the application of two million load cycles by 1.10 mm. This
elongation was mainly caused by the pull-out of the potting material from both steel hulls. The
stay cable showed very good fatigue properties as expected from CFRP.
After the fatigue test was completed the static load test was performed. The cable was loaded up
to a load of 3 600 kN. The ultimate load of a single CFRP-wire was 50 kN. Corresponding to this
value the ultimate load of 3 600 kN of the cable with 91 wires is equivalent to 78% of the
theoretical ultimate load. The specific elongation of the whole cable specimen corresponding to
this load was about 1,9 %. The test was stopped after six wire fractures where the first occurred
at a load of 3 500 kN. Due to the type of material the wires showed linear elastic behaviour up to
the failure. The cable specimen broke wire by wire not as suspected with a sudden fracture of the
whole cable specimen. Figure 6 shows the specimen at the anchorage before the ultimate strength
test. The wire fractures are shown in figure 7 and 8.
Figure 6:
Anchorage before the
ultimate strength test
Figure 7:
Wire fracture at the
anchorage
Figure 8:
part of the free length of the
specimen
After the cable was fully unloaded and dismounted the condition of the cable specimen was
investigated by a visual inspection. The components of the cable specimen showed no
extraordinary deformation besides the wire fractures at the entrance of the potting material and
the spacers.
Conclusion and Outlook
CFRP is a very interesting building material. Using this composition the advantages of the
carbon like low weight, high stiffness and resistance against aggressive media can be activated
without having the disadvantage of brittleness. The key problem how to anchor the wires which
prevented the use in the past is solved. Working groups in Europe, USA, Canada and Japan are
working on the standardisation of CFRP-systems.
Testing is even more important for CFRP- stay cables than it is for steel. With the available
equipment and the experience made in the last two years at the Technische Universitt Mnchen
testing can be performed without any problem.
The developed CFRP-stay cables as well as CFRP-tendon system DYWICARB are now
available and can be used for first applications. It can be hoped that the use will rise as fast as the
knowledge resulting from latest research work.
References
[1]
Meier, U.: Proposal for a carbon fibre reinforced composite bridge across the Strait of
Gibraltar at its narrowest site. Proc. Inst. Mech. Eng., Vol.201, No. B2 : 1987 (pp.73-78)
Saadatmanesh, H. ; Ehsani, M.R.: Fibre Composites in Infrastructure. 2nd International
Conference ICCI Tucson, Arizona : 1998
Schurter, U. ; Meier, B.: Storchenbrcke Winthertur. In: Schweizer Ingenieur und
Architekt. Nr. 44 : 1996 (pp. 976-979)
Recommendations for Stay Cable Design, Testing and Installation. Post-Tensioning
Institute. Phoenix, Arizona, August 1993.
Noisternig, J.F.: Zum Tragverhalten von Verankerungssystemen fr CFK-Litzen im
Spannbetonbau. Fortschritt-Berichte VDI-Reihe 4 Nr. 133. Dsseldorf : VDI-Verlag,
1996.
Noisternig, J.F. ; Dotzler, F. ; Roos, F. ; Jungwirth, D. ; Zilch, K.: Entwicklung eines
Zug-/Spanngliedes aus CFK fr das Bauwesen. 2. Symposium Neue Werkstoffe in
Bayern. Bayreuth : 1998
[2]
[3]
[4]
[5]
[6]
Bridge Consolidation by Using Cable - Stayed Method
Victor POPA
Michael M. STANCIU
Dr. Eng.
IPTANA-SEARCH Co.
Bucharest, ROMANIA
Bridge Engineer
IPTANA-SEARCH Co.
Bucharest, ROMANIA
Victor Popa, born 1942,

received his civil
Technical Construction
University of Bucharest in
1966. He is currently Head
of the Bridge Design
Department at IPTANASEARCH Co. and
associate professor at
TCUB.
Michael M. Stanciu, born

the University of
Oklahoma in 1983 and his
Master of Science in 1987.
He is president of
IPTANA-SEARCH, an
American-Romanian civil
engineering company,
located in Bucharest,
Romania.
1.
Generalities
The increase of the road traffic and vehicle load during the last decades imposed the necessity to
consolidate some of the existing older bridges.
Many of these bridges require both the carriage-way widening and structure consolidation. The
consolidation of these bridges generally require the consolidation both of the superstructure and
of the infrastructure. Sometimes the consolidation of the infrastructure and mainly of the
foundations is very difficult and expensive, mostly because of the reduced space under the bridge
and the existence of the crossed obstacle. Under these circumstances, is very difficult or
impossible to use suitable equipment. These inconvenient may be eliminated using cable-stayed
method for the consolidation of the existing bridges.
The method consists in supporting the existent superstructure deck from the pylons or towers by
straight inclined cables. The pylons or towers are built in different solutions, according to the
structure of the bridge requiring the consolidation. By this method the consolidation of the
existent infrastructures can be avoided and replaced with the construction of new pylons which
can be built in better construction conditions.
The method can be successfully used also when is required only the consolidation of
superstructure.
Some possible bridge consolidation solutions and two examples of bridge consolidation
presently under construction in Romania using this method are presented in the following.
Method description
The bridge consolidation by using cable-stayed method consists in building towers or pylons that
will support the existing superstructure deck by straight cables.The number of superstructure
deck supports will be increased, thus diminishing the stresses in the strengthening structure.
The pylons or towers usually are built in the existent bridge pier axes but, depending on the
designer creativity, they can be placed in any other favorably locations.
This method can be used mainly for bridges with simple supported or continuous beams but also
for other structure types such as frames, vaults and arches. The method can be used for
reinforced or prestressed concrete structures as well as for metal or composite structures.
The bridge consolidation by cable-stayed method is frequently accompanied by additional
structure prestresses made with external prestressing tendons. Also, to accomplish the new
structures it is necessary to build transverse prestressed beams with cantilevers which will be
anchored from the pylons by cable stayes.
For bridges with simple supported beams, the continuity of the beams on the piers will be
ensured by additional prestressing tendons.
Depending on the construction of the cable-stayes supports, bridge consolidation by cable-stayed
method can be divided in two categories as follows:
1st - Bridge consolidation when the existing foundation consolidation is not required or is easy to
be carried out;
2nd - Bridge consolidation when the existing foundation consolidation is an impossible or very
difficult and expensive work.
2.1
Presentation of the first category of bridge consolidation
If foundations of the existing piers are in good condition and dont require consolidation (e.g.:
piers based on rock or very hard soils) or can be easily consolidated, then the method consists
just in construction of towers for anchorage of the cable-stayes.
Fig. 1 shows the example of a bridge having the superstructure consisting of three spans
continuous prestressed concrete girders requiring only the consolidation of the superstructure.
Figure. 1 Elevation of a bridge with three-span continuous girders

a) existing bridge, b) proposed solution
The superstructure having a box cross section (fig. 2a ) can be consolidated by cable-stayed
method by building two towers. The cable anchorage towers may be supported either by the
existing superstructure (fig.2b) or by the existing piers (fig.2c).
Figure 2 Bridge cross section

a) existing bridge, b) bridge consolidated with towers supported by the superstructure,
c) bridge consolidated with towers supported by the piers
The towers will be fastened on the superstructure or on the piers with transverse prestressing
tendons.
The towers may have steel structure (to decrease the load on piers and the construction time
period) or composite structure. If the foundations bear higher loads, the towers may be built of
reinforced concrete with special steel elements only in the connection areas of the towers with
the supporting elements.
To fix the cables on the superstructure, special cross beams with cantilevers will be built of
reinforced concrete with prestressing tendons. The cantilevers may be built entirely or partially
of steel or composite elements.
For steel bridges, the cantilevers and the towers may be entirely of steel.
Fig.3 shows the elevation of a reinforced concrete bridge requiring consolidation. The piers are
based on rock and allow an important increase of the loads. However, the superstructure is
calculated for an inferior class and requires consolidation in order to bear the new loads. In the
mean time should be analyzed the solution for the carriage-way widening from two to four lanes.
To bear increased loads, the existing superstructure (fig. 4a ) can be consolidated using the cablestayed method, building towers supported by the superstructure over the existing pier (fig.4b).
If the carriage-way should be doubled, the towers supporting the cable-stayes may be built on the
carriage-way axis (fig. 4c ).
Figure 3 Elevation of a bridge with two span continuous girders

a) existing bridge, b) proposed solution for consolidation

a) existing bridge, b)proposed solution for consolidation
2.2.
Presentation of the second category of bridge consolidation
Considering the bridges for which the consolidation of the existing foundation is impossible or
very difficult and expensive for various reasons, the following consolidation solution can be
used.
We consider the case of the bridge in fig. 5 for which the consolidation of the existing
foundations is impossible due to the small space existing under the bridge.
Figure 5 Elevation of a bridge with three-span continuous girders

The superstructure of the bridge consists in a box prestressed concrete girder as shown in fig. 6a.
In this case, pylons, having their own foundations, may be built to take over the additional load
transferred through the cable-stayes (fig. 6b).

a) existing bridge, b) proposed solution for consolidation with two new pylons
3.
Examples of Romanian bridge consolidation projects using the cablestayed method
In the following are presented two projects elaborated in Romania for old bridge consolidation
by the cable-stayed method.
3.1.
Bridge over Jijia River at Carniceni
The bridge was built during 1915-1920. The bridge superstructure was destroyed during the
Second World War and was reconstructed in 1954 in accordance with the original project.
Figure 7 Bridge over Jijia river at Carniceni. Elevation

The bridge superstructure consists of Gerber reinforced concrete beams. The existing bridge has
seven spans with an average length of 15.00m each (fig.7a). The bridge has a narrow carriageway of only 5.00 m wide (fig. 8a). Both the infrastructure, mainly the foundation, and the
superstructure suffered important damages in time. The increase of the road traffic on the road
upon which this bridge is located imposes the widening of the carriage-way and the appropriate
consolidation of the strengthening structure. The bridge crosses Jijia river at its confluence with a
big river, namely Siret. This big river has an important impact on the water level under the
bridge in case of floods.
The technical-economical study has determined that the foundation consolidation has a high
investment cost. Under these circumstances, the solution of the consolidation by the cable-stayed
method proved to be more profitable, thus avoiding the consolidation of four infrastructures
foundations (fig. 7b). The pylons of the cable-stayed structure can be built, without any special
problems, near the existing bridge. For the bridge carriage-way widening is necessary to build
two new side beams that complete the existing superstructure (fig. 8b). The superstructure
becomes a three span continuous structure supported by cable-stayes as shown in fig.7b. The
elevations of the P1, P3, P4 and P6 piers are demolished down to the soil level, in this manner
also increasing the water drainage under the bridge by eliminating some obstacles. The P2 and
P5 piers of the old bridge are incorporated in the new P1 and P2 pylons of the consolidated
bridge.
Figure 8 Bridge over Jijia river at Carniceni. Cross section

3.2. Bridge over Dambovita River at Gemenea
The bridge was built in 1951 and has six spans of 20.45 m each. The total length of the bridge
superstructure is 123.00 m (fig. 9a).
Figure 9 Bridge over Dambovita river at Gemenea. Elevation

The bridge superstructure is made of simple supported pre-cast beams. There are six main beams
in the cross section. The bridge carriage-way is 7.00 m wide (fig.10a).
The bridge infrastructure is based directly on an superficial alluvial bed. After the floods of 1970
and 1975, the bridge suffered important damages, when P4 pier was unequally plunged down
with about 1.00 m.
Thus the pier was also rotated, and the superstructure of the adjacent spans were displaced
together with the pier.
Figure 10 Bridge over Dambovita river at Gemenea. Cross section

a) existing bridge, b) proposed rehabilitation solution
To restore the traffic, P4 pier was consolidated with drilling piles and the displaced
superstructure was lifted. The bridge rehabilitation project was elaborated for the consolidation
of all piers, but the high costs for the consolidation stopped the works on the other piers.
Presently the traffic is restricted on this bridge.
The unsuitable condition of the foundations of this bridge and also the deterioration of the
superstructure determined this bridge to be included again into the consolidation programme.
From the technical-economical analysis of the consolidation solution, the cable-stayed method
resulted to be the best.
For this bridge, four piers can be eliminated if a pylon will be built at the middle of the bridge,
near the existing pier P3 (fig.9b).
The foundations and the towers of the pylon will be built separately of the existing bridge (fig.
10b). After the pylon construction, the existing superstructure will be consolidated and widened
10
building an extra slab of reinforced concrete over the existing one. The bridge superstructure will
become a continuous structure by building cross beams with pre-stressed cantilevers placed in
the superstructure joints. The cantilevers will be used for the cable-stayes anchorage. The
superstructure consolidation and its continuity will be achieved by longitudinal prestress with
external prestressing tendons. The damaged concrete structure will be also repaired at the same
time with the new pylon execution.
4.
Calculation particularities for bridge structures consolidated by the

cable-stayed method
Using the cable-stayed method for bridge consolidation, the original static scheme of these
structures will be radically modified. Thus, the original statically determined structure ( simple
supported beams or Gerber beams) become a statically undetermined cable-stayed structure. The
calculation should consider the effective stresses in the original existing structure, which remain
included in the new structure, and the stresses appearing in different phases of the loads applied
on the new statical structure modified accordingly. Should also be considered the physicalgeometrical characteristics of the structure corresponding to each load phase and construction
element of the structure. Finally it is concluded that the calculus is similar to the calculation of
the composite structures.
5.
Conclusion
Bridge consolidation by cable -stayed method can be an efficient alternative solution to solve the
problems of the old bridges.
Use of this consolidation method may ensure some important technical - economical advantages
as follows :
- lower investment cost;
- better drainage of water under the bridge;
- improvement of the bridge aesthetics.
Ingenuity and creativity of designers may ensure the achievement of very interesting and special
new bridge structure.
The calculus has to consider the effective stresses both in the original and the new statical
structure.
Application of Simultaneous Identification

of Tension and Flexural Rigidity at once to the Bridge Cables
Ichiro YAMAGIWA
Researcher
Kobe Steel, Ltd
Kobe, Japan
Hideo UTSUNO
Senior Researcher
Kobe Steel, Ltd
Kobe, Japan
Koji ENDO
Researcher
Kobelco Res. Inst. Inc.
Kobe, Japan
Kenichi SUGII
General Manager
Kobe Steel, Ltd
Kobe, Japan
Summary
A new vibration method to estimate the flexural rigidity and the tension of cable is examined.
Periodicity of natural frequencies are derived analytically from the frequency equation of the
cables bending vibration with the tension. Utilizing the coefficient of the polynomial equation
which derived this periodicity, the flexural rigidity EI and the tension T are simultaneously
identified. Experiments using spiral rope for a cable-stayed bridge were carried out to verify the
present method. There was good agreement between measured and calculated data.
1. Introduction
When cable-stayed bridges are erected the tension of the cables are adjusted to the designed
specification. The tension of the cable is directly measured by a load cell or hydraulic jack, but
recently practical formulas derived by the vibration method are often used1). In these practical
formulas the first or second natural frequency of the cable vibration is used and it is assumed that
the flexural rigidity is known. The flexural rigidity is frequently investigated in the preliminary
test using a test piece of cable. However it is difficult to investigate the flexural rigidity for all
boundary conditions and all tensions during the erection because the flexural rigidity varies
according to the boundary condition and the tension2). Actually, the flexural rigidity is measured
for only the typical condition.
For this problem we proposed a new vibration method in which the tension and the flexural
rigidity was identified by using plural natural frequencies3),4). In this method the tension and the
flexural rigidity are identified by using the analytical equation derived from the frequency
equation of the cable vibration. Additionally, the precision of tension is studied by a numerical
simulation and an experiment using steel rod. As a result it was established that this method is
practically useful.
In this paper we apply the proposed method to the actual cables of a cable-stayed bridge and
study the precision of tension. The convenience of this method is confirmed in the actual
erection.
2. New vibration method

2.1 Theoretical equation
The cable vibration is considered to be a one-dimensional beams bending vibration with the
tension. The equation of motion for one-dimensional beams bending vibration with the tension
is given by:
EI
4w
2w
2w
T
+ A
=0
4
2
x
x
t 2
(1)
where w is bending displacement, EI is flexural rigidity, T is tension, is mass density, A is

cross-sectional area. Bending displacement w is shown by the product of the function of the
shape of the natural mode and the time function:
(2)
w ( x , t ) = W ( x ) e x p ( j t )
Substitution of Eq. (2) in Eq. (1) leads to the next principal function.
(3)
W ( x ) = C 1 co s( x ) + C 2 sin ( x ) + C 3 co sh ( x ) + C 4 sin h ( x )
where , are defined as
2
A 2
T
+
2 EI
EI
T
2 EI
(4)
A 2
T
T
+
+
2 EI
EI
2EI
(5)
where is the natural angular frequency. In the first place solving Eq. (3) on the simply suppo
rted boundary condition of both ends, the following frequency equation is obtained:
(6)
sin(L) = 0
where L is the length of the beam. The solution of Eq. (6) has periodicity and is given by:
(7)
L = i, i = 0,1,2 ,
Substituting Eq. (4) in Eq. (7), we can derive the next equation connected between the mode
number i and the natural frequency fi.:
T
2 EI 4
(8)
i +
i2
f 2 =
i
4 A L4
4 A L2
Secondary solving Eq. (3) on the fixed boundary condition of both ends, the following frequency
equation is obtained:
(9)
2 {1 c o s ( L ) c o s h ( L ) } + { 2 2 } s in ( L ) s in h ( L ) = 0
Rearranging the expression using approximation, Eq.(9) is simplified:
( 2 + 2 ) sin(L + ) = 0
tan =
2
2
=
2
T

2
(10)
(11)
AEI
The solution of Eq. (10) has periodicity and is given by:

(12)
Substituting Eq. (4) in Eq. (10), we can derive the next equation connected between the mode
number i and the natural frequency fi.
L + = i , i = 0,1,2,
fi 2 =
2 EI
T
i
i +
4AL4
4AL2
(13)
Eq. (8) is considered to be Eq.(13) with the term / omitted.

Eq. (8) and (13) express the relation between the square of the mode number and the square of
the natural frequency. Using that equation we can calculate the tension and the flexural rigidity
from the coefficient of the term of the second power or the fourth power of the mode i.
2.2 Simultaneous identification method of tension and flexural rigidity

In this section the simultaneous identification method of tension and flexural rigidity is shown in
two cases of boundary condition.
2.2.1 Simply supported boundary condition
The squares of the measured natural frequencies are plotted against the squares of the mode
number in Fig.1 (Filled circle). Least-squares fitting using Eq. (8) is applied to these data (Solid
line in Fig.1). As a result of curve fitting we can calculate the tension and the flexural rigidity
from the coefficient of the second power term or the forth power term of the mode i.
2.2.2 Fixed boundary condition
In the case of fixed boundary condition Eq. (13) is applied. In Eq. (13) the term of / is the
function of EI and T. Therefore Eq. (13) is a non-linear equation and needs iterative calculation.
Fig.2 shows the flow chart of this calculation. First initial T and EI is calculated using leastsquares fitting of Eq. (8). Secondary substituting that value in Eq. (11) and (13) and using leastsquares fitting, new T and EI is calculated. This calculation using Eq. (11) and (13) repeats until
convergence of T and EI.
Initial T,EI Eq.8 least-squares fitting

Calculate Eq. 11
Calculate T,EI Eq.13 least-squares fitting
T,EI Converge?
No
Yes
T,EI identified
Fig 1
Least squares fitting for natural

frequencies
Fig 2
Flow chart in fixed boundary condition
2.3 Precision of tension and flexural rigidity

2.3.1 Clear boundary condition
If the boundary conditions of both ends are known like simply support or fix, T and EI can be
calculated accurately using Eq. (8) or (13). We made an experiment using a test piece of cable for
the case of fixed boundary condition. The experimental condition is shown in Table 1 and the
result of the experiment is shown in Fig.3. In Fig.3 the vertical axis is the ratio of the estimated
tension by this method to the measured tension by the load cell. That ratio expresses the
precision of the identification in this method. The horizontal axis is parameter and will be
explained later. In this result this method can identify the precision of the tension to less than 1%.
Material
Spiral Rope
Linear Density 1.113 kg/m
Theoretical EI 295 Nm2
Length (m) Given tension (kN)
2
48.8 99.6 148.4
5
49.6 99.3 144.9
Table 1: Experimental condition
Figure 3: Precision of estimated T
However in the actual cable the boundary condition is known in only a few cases. Therefore we
simulated the precision of the tension and the flexural rigidity by assuming the boundary
condition was not known and investigated the precision limit in 2.3.2.
2.3.2 Order number of using eigenvalue data
In Eq. (8) and (13) it is considered that the first term is the contribution of the flexural rigidity EI
and the second term is the contribution of the tension T for the natural frequencies. The ratio of
the first term and the second term of Eq. (13) is written as:
2
4AL2
2
K=
=
4 2 i / 2
2 EI
(
)
i
4
4AL
T
(14)
where
=
T
L
EI
(15)
is the non dimensional parameter1). When is large the behavior of the cable is close to the
vibration of a string, when is small the behavior of the cable is close to the vibration of a beam.
According to Eq. (14), using a higher mode number i is better for calculating EI and using a
lower mode number i is better for calculating T. Therefore, in this method we use from the first
to the fifth mode of natural frequencies for calculating T and use the five largest measured
natural frequencies for calculating EI
2.3.3 Numerical simulation for unknown boundary condition

The method of simulation is as follows. First natural frequencies were calculated by Eq. (9) (B.C.
simply supported) using given T and EI. Next T and EI were calculated by Eq. (8) (B.C. fixed)
using calculated natural frequencies. Finally calculated T and EI were compared with given T
and EI. Additionally the opposite pattern was simulated. That is, T and EI were calculated by Eq.
(13) using natural frequencies calculated by Eq. (6).
Table 2 shows calculation parameters.
Density
: 7.8 x 103 kg/m3
Youngs Modulus : 2.06 x 1011 N/m2
Length
Diameter
Linear Density
L(m)
D (mm)
A (kg/m)
1, 3.95, 11
6
0.22054
1, 3.95, 11
8
0.39207
Table 2: Calculation Parameters
EI(Nm2)
13.10515
41.41876
Given Tension
(kgf)
75, 125, 250, 375, 500
125, 250, 500, 750, 1000
The mode number used for calculation was from the first to the fifth for T and the largest five for
EI. Fig.4 and Fig.5 show the result of the precision of T, Fig.6 shows the result of the precision
of EI, and Table 3 shows the lower limit of the precision of T and EI. In Fig.4 - Fig.6 the vertical
axes are the ratio of the estimated value to the given value. That is, the precision of the
identification of T and EI. The horizontal axes is parameter. In Fig.4 and Fig.5 the explanatory
notes show the equations used for calculation.
In Fig.4 the tension is calculated for natural frequencies with the fixed boundary condition. In the
case of Eq.(13), T is estimated accurately. This is because the assumed boundary condition of Eq.
(13) and the data boundary condition are the same. In the case of Eq. (8) as is smaller, the
precision of T is lower. On the other hand, as is larger, the precision of T is higher. In the case
of larger the influence of boundary condition for natural frequencies is small, so that the
difference of the precision of T according to the two equations is small. Fig.5 shows the same
result for Fig.4.
The shift value of the precision from 1 against is same in Fig.4 and Fig.5. In short, the precision
of T depends on parameter . This shift value is the precision of T in the largest difference of
boundary condition between equation and data. The precision of T in Fig.4 and Fig.5 is the lower

(B.C.Fixed)
limit of the precision using this method.

(B.C. Simply supported)
Fig.6 shows the result of the estimation of EI using Eq. (8) (B.C. simply supported) for natural
frequencies with the fixed boundary condition. In Fig.6 the parameter is the mode number used
for calculation of EI, and it shows that the measured highest mode number changes. According to
Fig.6, as the mode number used for calculation is higher, the precision of EI is higher. As is
larger, the precision of EI is higher and the difference of the precision of EI according to the used
mode number is smaller.
Table 3 shows the precision of T and EI from Fig.4 - Fig.6. The precision of EI is classified by
the mode number used. The precision of T and EI is decided by parameter and their precision is
higher when is larger. In the following examination of an actual bridge cable is more than 50.
10
30
50
100
6-10
11.5%
9.6%
7.5 %
4%
16-20
5.5%
5.2%
4.8 %
3.5 %
26-30
3.55%
3.5%
3.3%
2.8 %
70 %
15 %
8%
4%
EI
Figure 6: Precision
estimated EI (B.C.Fixed)
of
T (1-5)
Table 3: Precision of T and EI
3. Experiment Using Actual Cable

We carried out an experiment for the actual cable during the erection of a bridge to confirm the
simulated precision by this method. The situation of the experiment is shown in Picture 1. The
specification of the cable and the experimental condition are shown in Table 4 and the outline of
the cable is shown in Fig.7.
Specification of cable
Material
Coating spiral rope
Diameter
22 mm
Linear density
2.9263 kg/m
Experimental condition
Cable number
BRI-GI-SI
BRI-GI-S2
Given
T
kg f
902
Length
m
1768
21.04
3189
21.05
5535
21.06
1169
24.33
1831
24.33
3297
24.33
5535
24.33
21.05
Table 4: Experimental condition
Figure 7: Cable of the cable-stayed bridge

The following is the control method of the cable tension in this experiment. In Fig.7 one end of
cable was installed on the main tower through pin. The other end was set in the stand fixed to the
girder by a sleeve screw and double nuts. We carried out the experiment with cable floating from
the stand using a hydraulic jack. The tension was monitored by the load cell in the hydraulic jack.
The cable length was measured as the length from the upper set point to the contact point of nut
and the hydraulic jack cylinder. The vibration of cable was raised by a hammer striking at a point
1.5m from the contact point of the hydraulic jack and measured by an accelerometer at points 1m
and 2m from the contact point of the hydraulic jack. Natural frequencies were analyzed from
measured data by FFT. Vibration was measured at 2 points in order to avoid overlooking natural
frequencies by agreement between the node position of the vibration mode and the sensor
position.
One of the results of FFT analysis of the cable vibration is shown in Fig.8. The vertical axis
shows relative vibration velocity and its value is not important. The numbers in this graph are the
mode number of natural frequencies. This graph shows that natural frequencies from low to high
are raised regularly.
The result of the estimation of T and EI are
shown respectively in Fig.9 and Fig.10. We
applied Eq. (13) to natural frequencies from
the first to the fifth mode for calculating T and
to the five largest natural frequencies for
calculating EI. In Fig.9 the vertical axis is the
ratio of the estimated T to the given T (load
Figure8
cell). Assuming the given T is correct, the
precision of T is less than 8% and its result
agree with the result of section 2. This
Figures 8: Frequency analysis of cable
precision is sufficient in practical use for the
vibration
measurement of the cable tension. In Fig.10
the vertical axis is the ratio of EI (estimated
cable EI divided by EI of the steel bar with the same diameter as cable). The horizontal axis is the
given T (load cell). The ratio of EI is nearly constant in spite of the tension. In the spiral rope the
flexural rigidity EI increases as the tension increases, but the EI of this cable is constant. We
consider that this is because cable is covered with polymer and the restrained condition among
the wires composing the cable is not changed.
Additionally, in order to confirm the propriety of the identified T and EI we compared the natural
frequencies calculated by Eq. (9) using the identified T and EI with the measured natural
frequencies. One of the results of the comparison is shown in Fig.11. The result show that there
is good agreement between calculated and measured frequencies. We expressed this difference
between calculated and measured value as a percentage and calculated the average for all
measured modes. This average error in the case of Fig.11 is about 1.8%. Fig.12 shows the
average error in various experimental conditions. It was found that the average error is
sufficiently low and the identified T and EI is correct.
Moreover, this method can measure the tension for one cable in about 5-10 minutes including the
installation of the sensor and is very simple and speedy. For reasons of simple measurement we
paid attention to the plural higher mode frequencies which are usually ignored. The advantages
of this method compared with the usual method are as follows.
(1) In the usual method, a preliminary test using a test piece of cable is necessary because the
value of flexural rigidity EI is necessary for calculation of tension T. However, in this new
method, a preliminary test is not necessary because the tension T and the flexural rigidity EI are
calculated simultaneously during the erection.
(2) If special order natural frequency(e.g. first order, second order) is not raised, the tension can
be estimated due to the use of plural natural frequencies. Moreover, the effect of sag in the first
natural frequency is avoided.
(3) Discrimination between the cable's natural frequency and noise peak frequency is easy
because of the regulation of plural natural frequencies.
4. Conclusions
Our conclusions are as follows.
1) We proposed new method for simultaneously identification of the tension and the flexural
rigidity using plural natural frequencies of the cable vibration.
2) In a clear boundary condition the tension and the flexural rigidity can be identified accurately.
In an unknown boundary condition, like an actual bridge cable, the lower limit of the precision is
obtained quantitatively by numerical simulation.
3) Applying this method to the actual bridge cable, the precision of the identified tension agrees
with the simulated precision.
4)By applying this method to the erection of a cable-stayed bridge, it is proved that this method is
simple and speedy.
5) This method can be applied to a very extensive field using the bar expressed one-dimensional
beam apart from cable, because the use of an equation of motion for the one-dimensional beams
bending vibration with tension T.
References
[1].
[2].
[3].
[4].
T.Shinke, K.Hironaka, H.Zui and H.Nishimura : Practical formulas for estimation of cable
tension by vibration method, Jour.of JSCE, No.287,pp.26-32,1979 (in Japanese)
T.Shimada and A.Nishimura : Effect of flexural rigidity on cable tension estimated by
vibration method, Jour of JSCE, No.398/I-10,pp.409-412,1988 (in Japanese)
I.Yamagiwa, H.Utsuno, N.Sugano and K.Sugii : Vibration method to estimate flexural
rigidity of cable and cable tension at one time, Jour.of Structural Engineering, JSCE,
Vol.42A,pp.547-554,1996 (in Japanese)
I.Yamagiwa, H.Utsuno, K.Endo, K.Sugii : Identification of flexural rigidity and tension of
the one-dimensional structure by measuring eigenvalues in higher order, D&D Conf.,
JSME, No.97-10,pp.411-414,1997 (in Japanese)
Dynamic characteristics of two newly constructed curved cable-stayed bridges

Carmelo GENTILE
Assistant Professor
Dept. Struct. Eng.
Politecnico of Milan
Born in 1958, Carmelo

Gentile received his civil
engineering degree in
1984 and his Ph.D. in
1990, both from
Politecnico of Milan. He
has published widely in
the fields of earthquake
engineering, system
identification and fullscale testing of bridges.
Francesco MARTINEZ
y CABRERA
Professor
Dept. Struct. Eng.
Politecnico of Milan
Born in 1929, Francesco

Martinez y Cabrera received
his civil engineering degree in
1956 from University of
Naples. He has published
widely in a variety of research
fields including r.c. and p.c.
structures, bridge engineering
and cable-stayed bridges.
Summary
This paper describes the experimental program of field tests conducted on a couple of curved
cable-stayed bridges and the successive identification of theoretical (finite element) models
of the structures. The main objectives of the research program were: (a) to identify vibration
modes of the structures using ambient and free vibration tests; (b) to correlate computer models
with experimental results; (c) to compare the modal and structural behaviour of the bridges.
1.
Introduction
Theoretical and experimental investigation of two cable-stayed bridges is described in the paper.
The analysed bridges (Fig. 1) were recently erected in the neighbourhood of the new air terminal
of the Malpensa 2000 airport (Milan, Italy). Specifically, the bridges are placed in the north- and
south-side of the air terminal, respectively. The structures are in principle perfectly equal and are
characterised by a curved deck with longitudinal and transverse slope. This geometric layout is
very unusual for roadway cable-stayed bridges and only one dynamic investigation of a curved
cable-stayed bridge was performed before (Deger et al. 1996).
The experimental program of dynamic tests was conducted in three days (for each bridge)
and included extensive measurements of ambient vibrations induced by traffic to determine the
dynamic characteristics of the bridges. The most significant mode shapes and associated natural
frequencies were evaluated at 30 different locations of the deck and tower.
A total of 11 vibration modes were identified in the frequency range of 010 Hz for both bridges.
Due to the curvature of the bridge deck, special attention was devoted to the evaluation of the
degree of coupling between vertical and transverse vibration of the deck. Furthermore, the mode
shapes of the two bridges were compared by using standard techniques such as MAC (Allemang &
Brown, 1983), NMD (Waters, 1995) and COMAC (Lieven & Ewins, 1988). Basing on these global
indices, the bridges show very similar behaviour, being the natural frequencies of the north-side
bridge slightly lower than the corresponding ones of the south-bridge.
The reason of the above discrepancy in the modal behaviour was investigated by using threedimensional finite element models. Once the models were established, the structural
parameters were refined in order to enhance the match between theoretical and experimental
modal parameters. The basic difference between the two bridges was found to be related to
the deck Young modulus. The optimal value of this parameter turned out to be about 30000
N/mm2 for the north-side bridge and 34000 N/mm2 for the south-side bridge.
2.
Description of the bridges
The tested bridges carries two lanes of traffic from the A26 highway to the Terminal 1 of
Malpensa 2000 airport. The two bridges curve with a radius of 100 m and 6% longitudinal
slope so that the air terminal seems to be embraced from the bridges which, at the same time,
provide a limit to the airport area. Thus, the layout of the bridges gives an unique appearance
to the airport and architectural concerns played a determining role in the design.
Bridge construction began in 1994 and was completed in 1997. Fig. 1a shows a view of the
north-side bridge. The curved bridge girders have a centreline length of 140 m with two equal
side spans and 4 cables supporting each side span. The ultimate capacity of the individual stays
ranges from 8745 to 20405 kN.
The deck is a five-cell box concrete girder (Fig. 1b), 11.75 m wide and 1.35 m high, which was
cast in place and post-tensioned. The curved decks have an average transverse slope of 4%.
The cast-in-place concrete tower (Fig. 1c) is about 37 m high and consists of two legs
connected by a lower concrete cross-beam supporting the deck and an upper steel strut
providing the anchorage for the stay cables.
3.
Full-scale testing
The dynamic tests included extensive measurements of ambient vibrations induced by traffic. A
microcomputer-based measurement system was used on-site to measure and record the dynamic
response of the bridge. Ambient vibration response was acquired in about 50 minute records per
channel and data were sampled at 200 points per second per channel. The response of the bridges
a)
c)
b)
Figure 1. a) View of the north-side bridge; b) Cross-section; c) View of the tower
Figure 2. Location and numbering of sensors for the test of bridges

at selected points was measured using PCB (model 393C) accelerometers, each with a battery
power unit. Two-conductor cables connected the accelerometers to a computer workstation with
a data acquisition board for A/D and D/A conversion of the transducer signals and storage of
digital data. A schematic of the sensor layout is shown in Fig. 2.
In addition to the ambient vibration survey, free vibration tests were conducted to verify natural
frequencies and mode shapes determined from the ambient vibration tests. Free vibration was
induced by the passage of one axle of a truck over a standardised RILEM plank (which was
located at selected locations) and sudden braking. Space limitations in this paper preclude a
complete analysis of all results; however, it can be said that ambient and free vibration methods
were found to give virtually the same results in term of natural frequencies and mode shapes
for both bridges and a greater number of normal modes was identified during ambient
vibration tests (Gentile et al. 1998).
4.
Data processing and analysis
As it is usual for bridge structures (see e.g. Abdel-Ghaffar & Housner 1978, Muri-Vila et al.
1991, Gentile & Martinez y Cabrera 1997) the analysis of ambient vibration data was based
on the classical spectral techniques described by Bendat & Piersol (1993). Modal frequencies
were identified by the locations of peaks in the auto-spectra (ASD) and in the amplitude of
cross-spectra (CSD) with a resolution of 0.025 Hz. Furthermore, the coherence functions
were computed to assess the quality of data and to investigate non-linear response at each
mode. The experimental mode shapes are obtained from the amplitude of square-root ASD
curves and cross-spectral phases were used to determine directions of relative motion.
To investigate the correlation between the corresponding mode shapes of the two bridges, the
Modal Assurance Criterion (MAC, Allemang & Brown 1983), the Normalised Modal Difference
(NMD, Waters 1995) and the Coordinate Modal Assurance Criterion (COMAC, Lieven & Ewins
1988) were calculated.
The MAC indicates the degree of correlation between the two mode shape vectors A,k, B,j and is
defined as:
( AT,k B , j ) 2
(1)
MAC( A ,k , B , j ) = T
( A, k A, k )( BT, j B , j )
MAC values vary from 0 to 1; a value of 1 implies perfect correlation of the two mode shape
vector (one vector is a multiple of the other) while a value close to 0 indicates uncorrelated
(orthogonal) vectors.
The NMD is related to the MAC as follows:
1 MAC( A ,k , B , j )
NMD( A ,k , B , j ) =
(2)
MAC( A ,k , B , j )
Practically, NMD is representative of the fraction, on average, by which each component differs
between the two vectors A,k, B,j. The NMD is more sensitive to mode shape differences than the
MAC (Maia & Silva 1997) but it is not bounded by unity; thus the comparison is more difficult
for weakly correlated modes but is more discriminating when two modes are highly correlated.
In the COMAC, correlated modes from the two sets A and B are paired at a definite measurement
station r. The COMAC at point r is defined as:
COMAC (r ) =
A,rk B ,rk
k =1
(
k =1
A, rk
B , rk
(3)
)
k =1
where A,rk and B,rk denote the r-th components of vectors A,k and B,k, respectively. The value of
COMAC(r) (ranging from 0 to 1) is a measure of the correlation of corresponding modes at the
location r. This index is particularly feasible to indicate the location where the two sets mainly differ.
Thus, a value of COMAC(r) close to 1 implies that no significant changes of all modal deflections
occurred at r while COMAC(r) should be minimum where the main differences take place.
5.
Dynamic behaviour of the bridges and comparison of modal parameters
For both bridges, 11 vibration modes were identified by spectral analysis in the frequency
range of 010 Hz. A selected number of normal modes of the north-side bridge is shown in
Fig. 3. The identified modal behaviour was strongly dominated by the vertical components,
either pure bending or torsion. In fact, the ratio of the maximum vertical and transverse
modal amplitude of the deck was found to be always greater than 5.0. Generally, the
maximum amplitudes of the vertical and transverse components occur at different stations
a) Mode DV1+
f = 0.781 Hz
b) Mode DV2+
f = 1.221 Hz
c) Mode DV3+
f = 2.344 Hz
d) Mode DV4+
f = 2.808 Hz
e) Mode DV1
f = 3.687 Hz
f) Mode DV2
f = 3.809 Hz
Figure. 3. Examples of vertical and torsional modes of the north-side bridge deck
along the deck. Thus, the observed modes can be basically arranged as follows:
1. vertical bending modes of the deck (DV+) with anti-symmetric deck modes generally
involving a greater longitudinal participation of the tower than the symmetric ones;
2. torsional modes of the deck (DV) which are usually coupled with longitudinal motion of the
pylon and slight transverse displacement of the deck.
An illustrative example of the spectral analysis in Fig. 4 shows the acceleration spectra (square
root ASD) of the deck at locations 15 and 20 (see Fig. 2) for the two bridges. In the spectral plots
of Fig. 4, the thicker line refers to the north-side bridge while the thinner one refers to the southside bridge. The inspection of the acceleration spectra in Fig. 4 first reveals a remarkable
consistency of occurrence of spectral peaks for each bridge. This information and the coherence
values (which were always very close to 1 in the frequency range where spectral peaks occur)
suggest both a good quality of data and the linearity of the dynamic response. Furthermore, Fig.
4 clearly shows that spectral plots from the two bridges exhibit peaks of similar shape which are
slightly shifted along the frequency axis. Specifically, all spectral peaks of the south-side bridge
responses are located at greater frequencies than the corresponding ones of the north-side bridge.
As it had to be expected, the two bridges exhibit very similar mode shapes. The correlation
of the corresponding mode shapes of the two structures is illustrated by:
a) Fig. 5, which directly compares the corresponding mode shapes of the two bridges for a
selected number of vibration modes at the deck measurement stations;
b) Table 1, which compares the corresponding mode shapes and scaled modal vectors of the two
bridges through frequency discrepancy , the MAC and the NMD. It should be noticed that the
values of MAC and NMD in Table 1 were computed from vertical and longitudinal modal
displacements; furthermore, the frequency discrepancy between the corresponding natural
frequencies of the north- and south-side bridge is defined as:
a)
f NORTH f SOUTH
f NORTH
Amplitude (cm/s2)
1.2
Ch. 15 - North-side bridge
0.9
Ch. 15 - South-side bridge
0.6
0.3
0.0
0
10
frequency (Hz)
b)
Amplitude (cm/s2)
2.0
Ch. 20 - North-side bridge
1.5
Ch. 20 - South-side bridge
1.0
0.5
0.0
0
10
frequency (Hz)
Figure 4. Comparison of acceleration spectra from the two bridges at different locations
Mode
Identifier
f NORTH
f SOUTH
(Hz)
(1)
(2)
MAC
(Hz)
(%)
NMD
(%)
(3)
(4)
(5)
(6)
DV1+
0.781
0.825
5.63
0.9995
2.18
DV2+
1.221
1.270
4.01
0.9992
2.89
DV3+
2.344
2.441
4.14
0.9886
10.74
DV4+
2.808
2.905
3.45
0.9931
8.35
DV1
3.687
3.882
5.29
0.9935
8.07
DV2
3.809
4.004
5.12
0.9904
9.85
DV5+
4.541
4.810
5.92
0.9877
11.17
DV6+
5.371
5.664
5.46
0.9926
8.62
DV3
7.300
7.666
5.01
0.9843
12.61
DV4
7.471
7.886
5.55
0.9732
16.61
DV7+
8.789
9.058
3.06
0.9889
10.61
Table 1. Comparison of natural frequencies and mode shapes of the two bridges
North-side bridge
South-side bridge
b) Mode DV2+
0.5
dd
0.0
-0.5
-1.0
1
1.0
0.5
0.0
dd
1.0
Normalised mode shape
a) Mode DV1+
-0.5
-1.0
9 10 11 12 13 14 15 16 17 18 19 20
Measurement points
-0.5
-1.0
4
1.0
0.5
0.0
dd
dd
0.0
0.5
-0.5
-1.0
9 10 11 12 13 14 15 16 17 18 19 20
Measurement points
-0.5
-1.0
4
9 10 11 12 13 14 15 16 17 18 19 20
9 10 11 12 13 14 15 16 17 18 19 20
1.0
0.5
0.0
dd
dd
0.0
0.5
f) Mode DV2
1.0
Measurement points
e) Mode DV1
9 10 11 12 13 14 15 16 17 18 19 20
d) Mode DV4+
1.0
Measurement points
c) Mode DV3+
-0.5
-1.0
1
Measurement points
Figure 5. Comparison of selected mode shapes of the two bridges
9 10 11 12 13 14 15 16 17 18 19 20
Measurement points
By comparing the natural frequencies and mode shapes of the two bridges, the following
comments can be made:
1. there is a one-to-one correspondence between the observed vibration modes of the bridges,
being the natural frequencies of the south-side bridge slightly higher than the corresponding
ones of the north-side bridge. The frequency discrepancy (see Table 1) ranges from 3.06% to
5.92%, with a medium value of 4.79%. Since the environmental conditions during the test of
the bridges were controlled and nearly equal (ranging the temperature from about 11C to
15C), the above difference in natural frequencies is possibly related to structural behaviour;
2. the first two mode shapes are practically equal (being the NMD less than 3%) while the other
modes generally exhibit an average difference of about 10%;
3. by examining the local correlation of mode shapes through the COMAC (Lieven & Ewins,
1988) in order to highlight the locations where the two sets of mode shapes mainly differ, a
nearly uniform distribution of values was found, ranging the COMAC between 0.980 (location
6 in Fig. 2) and 0.996 (location 16 in Fig. 2).
The nearly uniform distribution of discrepancies in both natural frequencies and mode shapes
suggests that the differences of modal behaviour could be related to different elastic
properties of the concrete in deck and tower of the two bridges. The reliability of this
hypothesis was investigated and verified by using three-dimensional finite element models.
6.
Structural modelling, theoretical behaviour and model optimization
The theoretical analysis was divided in two parts. First, a simplified model (described by Gentile
et al. 1998) was formulated using a reduced number of 3D beam finite elements (107 elements,
102 nodes). The results from a dynamic analysis of this model were used to determine the
location of the sensors in the field tests and to evaluate the importance of geometrical nonlinearity. Specifically, the model development began by accounting for the overall non-linear
behaviour due to geometry change by using in the linear dynamic analysis the stiffness as
obtained from a non-linear static analysis corresponding to the dead load deformed state of the
bridge. Once this analysis was performed, it was found to produce practically the same results of
a linear elastic analysis without special consideration being given to the load state or deformed
geometry of the structure, as it happened in other theoretical and experimental studies in the
literature (see e.g. Wilson & Liu 1991, Casas & Aparicio 1993).
After the tests, a much more detailed finite element mesh was carried out with the arrangement
of nodal points and elements shown in Fig. 6. The model involved 4366 degrees of freedom and
used a large number of finite elements for the deck structure, so that a regular distribution of
masses can be obtained. The models was developed using the following assumptions:
a) the box girder deck was modelled by using 904 four-noded shell elements with 6 degrees of
freedom per node;
Figure 6. Three-dimensional finite element model: general arrangement
b) the tower was modelled using 30 3D beam elements while linear elastic truss elements were
used to represent the cable-stays;
c) the tower footing was fixed;
d) a Poisson ratio of 0.15 was assumed for both tower and deck;
e) the possibility of free sliding in the direction of the curved centreline at the abutments was
simulated by means of radial pinned-pinned rod elements.
A preliminary dynamic analysis was performed assuming for the concrete Young modulus of
both deck and tower a base value of 35000 N/mm2 (which was established from design data) in
order to check the similarity between experimental and theoretical modal parameters. The results
from the base model analysis confirm that:
1. a great portion of the bridge dynamic response is associated with vertical motions of the deck,
either in pure bending (DV+) or torsion (DV) with or without significant participation of the
tower. Specifically, within the 18 vibration modes involved in the frequency range up to 10
Hz, 12 are vertical modes of the deck, 1 is a transverse mode of the deck, 4 are transverse or
longitudinal modes of the tower while strong coupling in the three orthogonal directions
occurs in only one vibration mode;
2. slight coupling exists between transverse and vertical motions of the deck.
The columns (3)-(6) of Tables 2-3 compare the experimental natural frequencies and mode
shapes of the north- (Table 2) and south-side (Table 3) bridge and the corresponding modal
parameters of the base model through frequency discrepancy, the MAC and the NMD.
Experimental modal data were then used to adjust some parameters of the base model. Basing
on the results of the previous section, the updating of models involved the Young moduli of deck
(ED) and tower (ET). Since the corresponding theoretical and experimental mode shapes were
quite similar in the investigated variation range of the parameters, the model updating for both
bridges was carried out to correlate the analytical frequencies as closely as possible to the
identified ones. Thus, the optimal estimates of ED and ET are defined to be the values which
minimize the following:
Nm
fi M fiC
=
(4)
J
fM
i =1
i
being Nm the number of identified modes and fiM , fiC the i-th measured and computed natural
frequency, respectively. A plot of error function (4) in the neighbourhood of its minimum is
shown in Fig. 7 for the north-side bridge. Fig. 7 clearly highlights a different rate of change
8
7
6
5
4
3
2
34000
36000
38000
40000
42000
ET (N/m 2
m)
44000
46000
Figure 7. Error function plot for north-side bridge
36000
34000
32000
30000
2
m)
28000
N/m
(
26000
E
D
of J with respect to ED and ET , being greater the sensitivity to ED; a similar behaviour holds
for south-side bridge as well and it has to be expected since all identified modes involved
significant vertical motion of the deck. However, although with different sensitivities, the
error function J gives a clear indication about the optimal values of both ED and ET.
At the end of the optimization procedure, the optimal estimates of ED and ET were approximately:
1. ED = 30000 N/mm2 , ET = 40000 N/mm2 for the north-side bridge;
2. ED = 34000 N/mm2 , ET = 40000 N/mm2 for the south-side bridge.
Thus, the basic difference between the two bridges seems to be related to the values of ED while
the elastic modulus of the pylon is practically the same for both bridges.
The modal parameters of the updated models are compared to experimental data through
frequency discrepancy , the MAC and the NMD in columns (8)-(10) of Tables 2-3. Tables 2 and
3 refer to the north- and south-side bridge, respectively. In general, the updated models shows
excellent agreement in both frequencies and mode shapes (at the measurement locations) for all
identified modes; specifically, it is noted that the models exhibit slightly lower frequencies than
the measured ones with the maximum discrepancy being lower than 5%.
Exp.
f NORTH
Mode
Identifier (Hz)
(1)
(2)
Base Model (ED =ET =35000 N/mm2)
Refined Model
f
(Hz)
(%)
MAC
NMD
(%)
f
(Hz)
(%)
MAC
NMD
(%)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
DV1+
DV2+
DV3+
DV4+
DV1
DV2
DV5+
DV6+
DV3
DV4
DV7+
0.781
0.761
2.56
0.9976
4.86
0.766
1.04
0.9980
4.47
1.221
1.306
6.96
0.9985
3.92
1.270
4.01
0.9982
4.27
2.344
2.369
1.07
0.9847 12.45
2.243
4.31
0.9839
7.85
2.808
2.972
5.84
0.9700 17.57
2.778
1.07
0.9914
9.34
3.687
4.053
9.93
0.9613 20.06
3.829
3.85
0.9765 15.52
3.809
4.107
7.82
0.9615 20.01
3.872
1.65
0.9717 17.07
4.541
4.969
9.43
0.9886 10.76
4.616
1.65
0.9899 10.10
5.371
5.627
4.77
0.9962
6.16
5.234
2.55
0.9961
6.28
7.300
7.857
7.63
0.9412 24.99
7.322
0.30
0.9506 22.81
7.471
7.995
6.48
0.9224 29.00
7.390
1.08
0.9280 27.86
8.789
8.829
0.46
0.9703 17.49
8.814
0.28
0.9528 22.26
Table 2. Comparison of theoretical and experimental modal parameters for the north-side bridge
Exp.
f SOUTH
Mode
Identifier (Hz)
(1)
DV1+
DV2+
DV3+
DV4+
DV1
DV2
DV5+
DV6+
DV3
DV4
DV7+
(2)
Base Model (ED =ET =35000 N/mm2)
Refined Model
f
(Hz)
(%)
MAC
NMD
(%)
f
(Hz)
(%)
MAC
NMD
(%)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
0.825
0.761
7.76 0.99776
4.79
0.785
4.85
0.9979
4.58
1.270
1.306
2.83
0.9989
3.39
1.303
2.60
0.9987
3.63
2.441
2.369
2.95
0.9766 15.46
2.353
3.61
0.9837 12.88
2.905
2.972
2.31
0.9849 12.38
2.928
0.79
0.9741 16.30
3.882
4.053
4.40
0.9720 16.98
4.029
3.79
0.9806 14.07
4.004
4.107
2.57
0.9486 23.27
4.070
1.65
0.9812 13.83
4.810
4.969
3.31
0.9828 13.24
4.904
1.95
0.9835 12.95
5.664
5.627
0.65
0.9905
9.81
5.556
1.91
0.9916
9.21
7.666
7.857
2.49
0.9624 19.75
7.767
1.31
0.9786 14.79
7.886
7.995
0.87
0.9620 19.89
7.834
0.66
0.9703 17.49
9.058
8.829
2.53
0.9562 21.39
8.956
1.13
0.9484 23.34
Table 3. Comparison of theoretical and experimental modal parameters for the south-side bridge
7.
Conclusions
Ambient vibration testing and identification of finite element model of two curved cable-stayed
bridges was presented and discussed. The following conclusions can be drawn:
1. no experimental evidence was found that would suggest either the existence of non-linear
behaviour of the bridges during the tests or a strong coupling of vertical and transverse
vibration of the deck of bridges;
2. 11 vertical modes were identified for both bridges in the frequency range 010 Hz, being
a one-to-one correspondence between the identified modes of the two bridges;
3. the mode shapes of the bridges were compared by using standard techniques such as MAC,
NMD and COMAC. This correlation analysis clearly indicates that the bridges show very
similar mode shapes, being the natural frequencies of the south-side bridge slightly higher
than the corresponding ones of the south-side bridge;
4. the discrepancies of modal parameters was found to be nearly uniform; this suggests that
the differences of modal behaviour could be related to different elastic properties of the
concrete in deck and tower of the two bridges.
5. The above hypothesis was confirmed by using experimental data to evaluate the optimal
value of Young moduli of three-dimensional finite element models. Once the models were
established, the structural parameters were refined in order to enhance the match between
theoretical and experimental natural frequencies. The optimised finite element models
showed very good agreement with the experimental results and the main difference
between the two bridges was estimated to be a difference of about 13% in the value of the
Young modulus of the deck.
Acknowledgements
The cooperation of V. Imparato, G. Paracchini and their staff from the S.E.A. society is gratefully
acknowledged. The authors would like to thank A. Gennari Santori for the assistance in during the
field tests and A. Saisi for the valuable help in the investigation of the north-side bridge.
References
[1].
Abdel-Ghaffar, A.M. & G.W. Housner 1978. Ambient vibration tests of suspension
bridge, J. Engg. Mech.Div. ASCE 104: 983-999.
[2]. Allemang, R.J. & D.L. Brown 1983. Correlation coefficient for modal vector analysis,
Proc. 1st Int. Conf. Modal Anal.: 110-116.
[3]. Bendat, J.S. & A.G. Piersol 1993. Engineering applications of correlation and spectral
analysis. 2nd Ed., Wiley Interscience.
[4]. Casas, J.R. & A.C. Aparicio 1993. Theoretical and experimental dynamic behaviour of
the Alamillo cable-stayed bridge in Sevilla, Spain, Proc. Eurodyn '93: 995-1002.
[5]. Deger, Y., R. Cantieni, C.A.M. deSmet & A.J. Felber 1996. Finite element model
optimization of the new Rhine Bridge based on ambient vibration testing. Proc. Eurodyn
'96: 817-822.
[6]. Gentile, C. & F. Martinez y Cabrera, 1997. Dynamic investigation of a repaired cablestayed bridge. Earthq. Engg. Struct. Dynam., 26: 41-59.
[7]. Gentile, C., F. Martinez y Cabrera. & A. Saisi 1998. Dynamic testing and seismic
response of a cable-stayed bridge. Proc. XI European Conf. Earthq. Engg., CD-ROM,
Rotterdam: Balkema.
[8]. Lieven, N.A.J & D.J. Ewins 1988. Spatial correlation of mode shapes, the coordinate
modal assurance criterion (COMAC), Proc. 6st Int. Conf. Modal Anal.: 690-695.
[9]. Maya, N.M.M. & J.M.M. Silva (Eds.) 1997. Theoretical and experimental modal analysis.
Research Studies Press Ltd.
[10]. Muri-Vila, D., R. Gomez, & C. King 1991. Dynamic structural properties of cablestayed Tampico bridge. J. Struct. Engg. ASCE 117: 3396-3416.
[11]. Waters, T.P. 1995. Finite element model updating using measured frequency response
functions, Ph.D. Thesis, Department of Aerospace Engg., University of Bristol.
[12]. Wilson, J.C. & T. Liu 1991. Ambient vibration measurements on a cable-stayed bridge,
Earthq. Engg. Struct. Dynam., 20: 723-747.
Field Observation of Aerodynamic Response of Meiko West Bridge

Yuji SUZUKI
Kazuyuki MIZUGUCHI
Japan Highway Public Corp. Japan Highway Public Corp.
Nagoya, Japan
Nagoya, Japan
Toshio UEDA
Hitachi Zosen Corp.
Osaka, Japan
Satoru SAKUMA
Toshiaki MAEKAWA Yoshikazu KOBAYASHI
Japan Highway Public Corp. Japan Highway Public Corp.
Hitachi Zosen Corp.
Nagoya, Japan
Nagoya, Japan
Osaka, Japan
Summary
The Meiko West Bridge is composed of two adjacent cable-stayed bridges whose interval is
50m, after the construction of the phase II line bridge in 1998. This paper describes the
aerodynamic behavior of those parallel bridges measured in the site wind in order to verify the
phenomena derived from the 3-dimensional wind tunnel test in a boundary layer turbulent
flow. And also the characteristics of the rain-wind-induced oscillation measured on the staycables of both bridges is referred simultaniously.
1. Introduction
The Meiko West Bridge is composed of two cable-stayed bridges with flat box girders as
shown in Fig.1. The interval between parallel cable-stayed bridges is 50m. The phase I line
bridge was completed in 1985, and this existing bridge has been put in service. The phase II
line bridge was finished in 1998, and that is the new bridge. Now both bridges become a part
of the Ise-Bay Highway that will link the New Tomei and Meishin Expressways. The
aerodynamic stability of these bridges in tandem arrangement was investigated by the the
wind tunnel tests using the 3-dimensional aeroelastic models with the scale of 1/100 in
Hitachi Zosens wind tunnel facility as shown in Fig.2. From the results, it is concluded that
the vortex-induced oscillation occurs in a smooth flow, but it vanishes in a turbulent flow. The
field observation was performed in order to certify those phenomena.
It is well known that the rain-wind-induced vibration was discovered for the first time under
erection of the phase I line bridge of the Meiko West Bridge in Japan. By this time all staycables of this bridge have been connected mutually by the wire ropes. Therefore it will be
necessary for any controlling device to be installed if the same phenomenon occurs in the
stay-cables of the phase II line bridge. In order to obtain the materials for design
judgement, the field observation of the stay-cables motion was also carried out
simultaneously.
2. Field observation of girders motion

The characteristics of the wind on the site were measured by the ultlasonic anemometer set on
the deck of the phase I line bridge. The flat box girders motion was picked up by the servotype accelerometers. Those data were recorded on the opto-magnetic type disk using the
automatic measuring system through the variable triger level of the wind velocity and the
girders motion. Each data unit among 10 minutes was delivered with 0.05 sec sampling time.
Many precious data including the monsoon and three times typhoons which are the 7th to 9th
one in 1997 were obtained. The wind direction was predominantly northwest on the land side
and southeast on the seaside as shown in Fig.3. There is no difference between this time data
and the data measured 10 years ago with the same ultrasonic anemometer. Accordingly each
wind
Fig.1 Outline of Meiko West Bridge
Fig.2 3-dimensional wind tunnel test in boundary layer turbulent flow
Fig.3 Histogram of wind direction
Fig.4 Relationship between turbulent intensity

and wind velocity
Fig.5 Histogram of turbulent intensity
direction corresponds diagonaly to the axis of the bridge girder respectively.

Figure 4 shows the relationship between the turbulent intensity and the mean wind velocity on
the landside and seaside wind. And the histograms of turbulent intensity on both landside and
seaside wind are shown in Fig.5. From Fig.4 the turbulent intensity is nearly over 10% at
about 10 m/s where the vortex-induced oscillation may occur in a smooth flow. Such wind
situation certificates that the 10% turbulent intensity selected in a boundary layer turbulent
flow of wind tunnel test investigating the effect on the vortex-induced vibration was
appropriate. From Fig.5, the mean turbulent intensity is about 15% on both side winds, and
the estimation of the gust response occurred in a boundary layer turbulent flow with 15%
intensity will be suitable.
From the characteristics of some wind data obtained, the longitudinal turbulent scale is
estimated 30m to 340m. The spectrum of wind velocity is good agreement with the Karmantype spectrum as shown in Fig.6.
Fig.6 Spectrum of wind velocity

The vertical bending motions of the girder measured before and after the center connecting
stage are shown in Fig.7 and Fig.8, respectively, comparing with the wind tunnel test results
in a smooth or a turbulent flow. Every data show the standard deviation of dynamic responses
at the center of each main span. Each relationship between the vertical bending motion and
the wind velocity resembles the tendency of the gusty response generated in the boundary
layer turbulent flow of the wind tunnel. The vortex-induced vibration didnt occur in the
natural wind with turbulence. The amplitude of gusty response is less than one derived from
the wind tunnel test in a boundary layer turbulent flow with 10% or 15% turbulent intensity.
Therefore it is concluded that there is no problem in the fatigue strength, referring to the
design assessment based on the wind tunnel test results.
3. Field observation of stay-cables motion

The measured stay-cables are C9 and C10 cable at the side span and C21 and C22 cable at the
main span, whose are selected according to the experience during observation of the phase I
line bridge as shown in Fig.9. During erection of the phase II line bridge, C11, C12 and C13,
C14, C15 and C16 cable are measured by turns instead of above mentioned cables. The
rainfall level was picked up by the rainfall strength meter, and the in-plane motions of staycables were measured by the servo-type accelerometers.
Fig.7 Vertical bending motion

before center connecting stage
stage
Fig.8 Vertical bending motion

after center connecting
Fig.9 Objected stay-cable

Ten times rain-wind-induced vibrations data were obtained. Those rain-wind-induced
vibrations were apt to occur in the main span side cables against the northwest direction wind
and in the side span side ones against the southeast direction wind. Both cables coincide with
the called positive posture side ones against the wind direction respectively.
Fig.10 Occurrence frequencies
Fig.11 Spectrum of cable motion in rain-wind-induced vibration
Fig.12 Amplitude of cable motion
The distinguished frequencies contained in the rain-wind-induced vibrations are shown in

Fig.10. The aspect of the rain-wind-induced oscillations of two bridges are apparently
different each other. Each example of cable motion spectra are shown respectively in Fig.11.
The oscillation in the phase I line bridges cable is composed of many vibration modes, i.e.
from 1st to 20th mode, and on the other hand the oscillation of the phase II line bridges cable
includes only a few modes like 3rd to 7th mode. This different appearance was caused by the
reason why the hard rubbers for suppression of bending at the end of cables and the rubber
covers for sealing the cable-anchorage pipes were set on the phase I line bridges cables, but
there were no attachments on the cables of the phase II line bridge under construction. Such
behavior in the phase II line bridge is apparently similar to the one measured in the phase I
line bridge under construction about 10 years ago.
The maximum total value derived from the addition of each modes displacement whose are
separated by FFT analysis are about 30cm on the phase I line bridge as shown in Fig.12 .
Fig.11(a) shows the rain-wind-induced vibration of the center-span cable C21 occurred in the
northwest wind with a weak rain, and Fig.11(b) shows the one of the side-span cable C9
occurred in the northeast wind with a comparatively strong rain.
The inspection on the fatigue strength of stay-cables was performed, and it is concluded that
there is no problem on the fatigue strength.
The vortex-induced vibration often occurred in the stay-cables as shown in Fig.13. In this
figure the data denoted as the symbol + indicates the relationship between the vortexshedding frequency and the mean wind velocity. From this relationship the Strouhal number
St=fd/V is estimated about 0.15, whose value is a little lower than St=0.18, one of the circular
cylinder in a turbulent flow. In Fig.13 the data denoted as the symbol or are derived from
the rain-wind-induced vibration, and those frequencies are nearly constant. The vortexinduced vibration denoted as the symbol, and the rain-wind-induced vibration denoted as the
symbol measured in the phase I line bridge 10 years ago are similar to the data obtained this
time in the phase II line bridges. Anyway the observated amplitude of the vortex-induced
vibration was very small.
According to the above-mentioned results, the controlling devices with high viscous damping
rubbers have been installed on both bridges from the view point of serviciability, taking the
aesthetic design into consideration as shown in Fig.14. The capacity of the installed dampers
was decided based on the Scruton Number Sc=m symbol 100 \f "Symbol" \s 12/symbol 114
\f "Symbol" \s 12 d2 = 60 (m : mass of cable per unit length, symbol 114 \f "Symbol" \s 12 :
air density, d : diameter of cable, symbol 100 \f "Symbol" \s 12: structural damping of cable
in logarithmic decrement). symbol 100 \f "Symbol" \s 12 equals 0.025 to 0.029 in the phase I
line bridges cables, and about 0.02 in the phase II line bridges ones. Of course, the wire
ropes connecting mutually are removed in all cables of the phase I line bridge.
Fig. 13 Relationship between occured frequencies and wind velocity
Fig.14 High viscous damping rubber installed at the end of stay-cable
4. Conclusion
The wind tunnel test results using the aeroelastic model in a boundary layer turbulent flow on
the Meiko West Bridge composed of 2 cable-stayed bridges with the flat box girders in
tandem arrangement were certificated by the field observation. It is confirmed that the
assessment on the safety and the serviciability based on the wind tunnel test results is
adequate.
From the results of the field observation on both rain-wind-induced vibration and vortexinduced oscillation of the stay-cables, it is cleared that there is no problem in the fatigue
strength. But, from the view point of the serviciability, the controlling devices with the highviscous damping rubbers are installed.
Acknowledgement
This study was done under the technical guidance by the committee on the design and
construction of the Meiko West Bridge, whose chairman is Prof. Emeritus of Tokyo
University Manabu ITO. The authors express their sincere appreciations to the members of
the committee for their valuable advice. And also the authors wish to thank Prof. Masaru
MATSUMOTO of Kyoto University for his helpful advice on the field observation of the
stay-cables.
References
[1] Tadashi NAGAI, Satoru SAKUMA, Toshiaki MAEKAWA and Toshio UEDA :
Aerodynamic response on cable-stayed bridges with flat box girders in tandem
arrangement, The 14th wind engineering symposium, Dec. 1996, in Japanese.
[2] Yuichi HIGAMI : Rain-wind-induced vibration on cable-stayed bridge, Japanese
Association of Wind Engineering, Vol.27 Mar. 1986, in Japanese.
[3] Civil Engineering Research Center Corp : The report of aerodynamic study on cables in
cable-stayed bridges, 1993, in Japanese.
Rehabilitation of the Luangwa Bridge

Peter REINHOLDT
Department Head
COWI AS
Lyngby, Denmark
Peter Reinholdt, born 1953,
received his civil
Engineering Academy of
Aalborg Denmark in 1976.
He is head of Department of
Design and Supervision of
Bridges.
Ejgil VEJE
Department Head
COWI AS
Lyngby, Denmark
Jimmy KALVSLUND
Project Lead Engineer
COWI AS
Lyngby, Denmark
Ejgil Veje, born 1954,

Univ. of Denmark in 1981.
He is head of Department of
Maintenance and
Rehabilitation of Major
Bridges.
Jimmy Kalvslund, born 1963,

Univ. of Denmark in 1989.
He is Project Lead Engineer
in the Department of
Operation Management and
Systems.
Summary
The Luangwa Bridge is a cable stayed bridge situated in Zambia. The bridge was built between
1966 and 1968. Since its early days the traffic on the bridge has been restricted to the crossing of
one vehicle at a time with a maximum gross weight of 50 tonnes travelling with a maximum speed
of 15 km/h. A rehabilitation of the bridge was carried out in 1997 to strengthen the bridge and
enhance its load bearing capacity. The rehabilitation included replacement of all cables and
strengthening of the bridge girder and the pylons. Construction work had to be completed within a
time slot between two rainy seasons, which represented a serious constraint on the project.
1.
Introduction
1.1
Location of the bridge
The Luangwa Bridge is situated on the Lusaka-Chipata Road (the Great East Road) connecting
Zambia and Malawi. The bridge is located approximately 232 km east of Lusaka.
1.2
Technical description
The Luangwa Bridge is a cable stayed bridge build in 1966 - 1968 with a total length of 303.5
metres with a main span of 222.5 metres. The height of the pylons are 33.4 metres measured from
the bridge deck. At the main piers the bridge superstructure including the pylons is supported by
steel columns with a height of 13.9 metres. The steel columns are supported on pier foundations
made of reinforced concrete.
Fig. 1 Elevation of the Luangwa Bridge

The superstructure consists of two main girders made as rectangular steel box girders
(height/width = 2.0/0.6 metres). The steel box girders are bolted together in 10 metres sections
with friction grip bolts in splice connections. The bridge deck is made as a reinforced concrete
slab with a thickness of 0.15 metres. It is supported by steel cross girders (I-beams) with a spacing
of 2.6 metres. The main girders, the cross girders and the concrete deck behave as a composite
structure.
The superstructure is supported by a double cable system.
The original Design Criteria were according to British Standard BS 153 with HA-loading.
1.3
Experienced problems
Shortly after the opening of the bridge it became evident that it was not behaving as intended. Two
factors were especially of concern.
The vertical profile of the bridge girder differed greatly from the theoretical profile of the bridge.
One reason for this problem was that the stay cables were too long. They were designed with a
fixed length and there was no possibility of adjusting the length.
Fig. 2 Profile of the bridge deck
Another factor was the failing of the high friction grip bolts in the splice connections of the steel
box girders. During the years after the opening of the bridge, a number of these bolts started
failing, with severe traffic restrictions as a consequence. Only one vehicle at a time was allowed to
cross the bridge with a maximum speed of 15 km/h and a maximum gross weight of 50 tonnes.
1.4
Remedial works
In 1972-73 remedial works were carried out:
The cable stays were shortened by applying

clamps, reducing their length by approx. 135
mm in order to improve the vertical alignment
of the bridge
Inside the main steel girders horizontal
compression steel tubes were mounted near
the towers and horizontal tension cables were
installed near the middle of the bridge
Failed friction grip bolts were replaced
The steel towers were filled with concrete
below deck level.
2.
Fig. 3 Cable Clamps
Inspection and Rehabilitation Study
The department of international development assistance of the Danish Ministry of Foreign Affairs,
Danida, offered in 1993 to finance a condition assessment and rehabilitation study for the
Luangwa Bridge.
2.1
Inspection
The inspection of the bridge was carried out in 1993 covering both visual inspection, nondestructive and destructive testing.
The visual inspection revealed large cracks in the concrete in the top of one of the main piers and
in one of the abutments. The cracks in the top of the main pier were clearly caused by splitting
forces from the steel towers on top of the main pier. The cracks in the abutment were located
behind the main anchorages for the bridge girders indicating insufficient reinforcement.
In spite of the remedial works carried out in 1972-73 with the intention of rectifying the
longitudinal profile there was still a considerable sag in the main span. The clamps applied to the
cables had reduced the sag but not eliminated it. Furthermore, a more local sag in the alignment
was measured in the middle of the main span probably due to overloading of the friction joints of
the main steel girders.
The reason for the failure of high friction grip
bolts in the splice connections of the main steel
boxes was examined and found to be caused
primarily by intergranular cracking in the bolt shaft
caused by hydrogen. These cracks had developed
after the tightening. Ultrasonic tests on the bolts
still in position showed that the elongation of the
bolts varied several millimetres indicating a
tensioning procedure out of control.
Severe pitting corrosion was found on the cable
stays. The painting of the cable stays was cracked
and there was virtually no adhesion to the cable
surface any more.
Fig. 4 Severe pitting of cables
2.2
Rehabilitation Study
The inspection of the bridge was followed by a comprehensive rehabilitation study. Structural
calculations were based on a 3D FEM-model of the bridge established in the in-house developed
programme IBDAS (Integrated Bridge Design and Analysis System). The model included
information about the original length of cables, remedial works etc. The behaviour of the model
was cross checked by comparison of deflections with measured deflections from a loading test
performed on site with a 30 tonnes truck. This showed a good correlation as the deviation was
within the accuracy of the measurements.
Structural calculations showed that many of the elements of the bridge were heavily overloaded.
The rehabilitation study concluded that the bridge had to be thoroughly strengthened in order to
provide a satisfactory safety level for the traffic passing the bridge.
3.
Rehabilitation Design and Rehabilitation Works
Based on the recommendations of the rehabilitation study Danida decided in 1996 to offer
financing of a rehabilitation of the Luangwa Bridge.
3.1
Modelling for the Rehabilitation Design and Rehabilitation Works
For the analysis and design of strengthening measures the IBDAS FEM-model was further refined
to take into account the full history of construction of the bridge. This included back tracking the
construction phases for the main girders, the casting of the concrete deck in sequences, shortening
of the stay cables and other remedial measures carried out in 1972-73.
As the rehabilitation included installation of new bottom plates, splice plates and bolts in the main
steel girders of the bridge deck, all bolted connections had to be opened. To do so the connections
had to be either in a virtually stress free state or had to be temporarily fixed by clamps or
equivalent. The chosen method of rehabilitation involved bringing the deck into a stress free
state.
An achievement of a "stress free" condition is influenced by many factors and several scenarios
had to be envisaged. Therefore an envelope of occurrences were considered and these were
considered to be described by the following two assumptions:
1. Model the construction sequence from when the bridge was built, reinforcement measures
applied in 1972-73 and assume joints to be opened with zero moment.
2. Ignore all constructions phases and assume all loads to be applied simultaneously on the
rehabilitated structure.
3.2
Rehabilitation Works and Method
The rehabilitation design prescribed the following measures:
Removal of clamps and tension cables from previous attempts to strengthen the bridge
Replacement of existing cable stays with new and stronger stays
Strengthening of the anchorage plates for the cable stays
Replacement of all friction grip bolts in the joints between the steel sections of the main bridge
girders
Replacement and installation of splice plates
Installation of additional bottom plates to the main bridge girders
Installation of additional crossbeams under the concrete deck
Filling of the steel towers above deck level with concrete
Replacement of the main anchorages of the bridge at the abutments
Increase of the weight of one of the abutments
Strengthening of the steel columns below deck level
Application of a compression ring at the top of the main pier to prevent splitting
To replace the existing cable stays, the bridge deck had to be supported by temporary towers at
each cable anchorage point. This was only possible during a period with low water level in the
river due to the risk of the temporary towers being washed away. The water depth of the Luangwa
river varies from approx. 2 metres during the dry season from April to November up to approx. 910 metres during the rainy season from December to March.
The constraints to the duration of the rehabilitation were severe as a time extension beyond the
given time slot would have meant closing down the site and securing the bridge during the rainy
season at an intermediate stage of the rehabilitation works.
The de-stressing of the existing cables had to be performed in two steps. First the cable clamps
installed during the remedial works in 1972 in order to shorten the stays had to be removed. No
records of the forces used to tighten the clamps had been traceable. The forces applied by the
clamps had to be determined by measuring the geometry of the compressed clamps.
After removal of all cable clamps, the stays still had a considerable force left. As the cables were
fabricated with no possibility of shortening or extension, the de-stressing of the remaining force
was difficult. A new clamp system was made which could span a short distance of a set of cables.
By pushing the clamps together the section of cable in between was de-stressed. It was then cut by
a cutting torch and the clamp system was de-stressed.
Now the bridge was in fact with its temporary towers a traditional girder bridge.
The chosen method of rehabilitation by bringing the bridge girder into a stress free condition
was achieved by use of a travelling
girder.
The travelling girder was designed
to span neighbouring support
points lifting the deck section
undergoing rehabilitation. The
travelling girder was in practice a
bridge in itself and in position it
still allowed for traffic passing the
bridge. Spanning a maximum of 56
m and weighing 90 tonnes, it was
designed to carry both the weight
of 55 m bridge girder and the
traffic loads.
The travelling girder was
successively moved on top of the
bridge deck on temporary rails into
positions between two temporary
towers. Once in position the weight Fig. 5 Travelling Girder
of the bridge girder underneath was
transferred to the travelling girder.
This was done by use of macalloy bars connecting from the travelling girder through drilled holes
in the bridge deck to temporary crossbeams underneath the girder. The macalloy bars were
stressed and controlled using jacks and load cells. The virtual stress free condition was achieved
in this way and the required opening of bolted joints could be performed to allow fitting of the
new splice plates and new bolts.
When opening the first joints it was in some places observed that the bottom flanges of the two
neighbouring sections were in close contact, and because of composite effect with the concrete
deck at the upper top flange the joint still possessed moment capacity. In other cases a small slip
occurred and bottom flanges were seen to move towards each other before a near stress free state
was achieved. This information was used to further adjust the FEM-model and by applying the
right combination of forces in the macalloy bars it was possible to achieve near stress free states in
the joints before opening them in the remaining sections of the bridge.
The new cables were locked coil cables with adjustable length. The anchorage system was made
to fit the existing anchorage plates in the girder and the towers.
Thanks to a very intense and close contact during construction between the contractor on site, the
supervision team and the design team in Denmark the contractor succeeded in finishing all critical
operations and removing the temporary towers from the river bed before the heavy rains made the
water level in the Luangwa River rise dramatically by mid December 1997.
4.
Operation and Maintenance
With the chosen level of rehabilitation the bridge is now able to carry HA loading and HB loading
up to 25 units. The HA loading is a formula loading representing normal traffic. The HB loading
is an abnormal vehicle unit loading.
For the Zambian authorities to be able to control the daily traffic on the bridge and administrate
permits for heavy transports a set of restrictions was defined for a range of realistic heavy
vehicles.
A Maintenance Manual for the rehabilitated bridge has been prepared. The manual included a
suggestion for a setup of inspections to be performed as part of the maintenance.
Design of Structural Monitoring Systems
J. LAIGAARD JENSEN
Ph.D
COWI
Lyngby, Denmark
Lars PEDERSEN
M.Sc.
COWI
Lyngby, Denmark
Jakob Laigaard, born 1961,

degree from the Aalborg
University, Denmark in 1986.
He has specialised in structural
monitoring and structural
dynamics.
Lars Pedersen, born 1966,

received his civil
Aalborg University, in
1991. He has specialised
in structural monitoring
and instrumentation of
large structures.
Summary
This paper reviews and discusses approaches and processes involved in the design of structural
monitoring systems. Within the civil and offshore engineering industry, monitoring systems are
used either as permanent or as ad hoc systems (testing) providing information with objectives of
obtaining information maximising revenue with respect to design, construction, operation and
maintenance, and repair of structures.
The design of such systems may simply be carried out based on a pragmatic basis resting on
compromises within what may be called good engineering judgement, or the design decisions may
be made on a more rigor basis applying rational cost-benefit analyses.
Through a discussion of principles and examples, this paper discusses these aspects of the design
of structural monitoring systems. It is argued that there are very good reasons for forcing the
design process into a more rigor framework based on rational decision approaches, well known
from experimental design in general.
1.
Introduction
In the civil and offshore engineering industry, since the mid seventies, there has been extensive
use of structural monitoring systems and structural testing based on ad hoc monitoring systems
applied on large structures.
Within civil engineering industry, new large span suspension bridges such as the Great Belt
Bridge, [1,2], and the Tsing Ma Bridge, [3], both have been designed with extensive monitoring
systems. Similarly, new cable-stayed bridges, ranging from the resund Link Bridge (being build)
to bridges such as those of the Lantau Fixed crossing, [3] and the Far Bridge, [4] comprise also
quite extensive installations of instrumentation. Also, tall buildings subjected to wind or
earthquake loads are quite normally extensively instrumented, [5].
-1-
Within the offshore industry, uncertainty in severity of wave climate as well as difficult inspection
conditions have made structural monitoring a standard tool for enhancing safety of offshore
structures. The standard objectives have been monitoring of wave and wind climate together with
platform displacement response, and fatigue life in terms of stress cycles, [6]. The layouts of such
systems have varied a great deal and this suggest that the design approach on structural monitoring
system may have been based on different approaches and criteria. In the offshore industry, some
platforms have hardly a wave height sensor whereas others have an extensive number of
accelerometers and strain gauges.
With respect to the civil engineering industry, table 1 shows an example of difference scales of
structural monitoring. The table compares the number of sensors between the permanent structural
monitoring systems of the Great Belt and the Tsing Ma Suspension Bridges. The table shows a
significant difference in the amount of instrumentation. One reason is of course different
conditions with respect to site, environment and operation, another reason may be differences in
the priorities in the design of the structural monitoring system. A detailed assessment of the actual
causes for the different scale of instrumentation is not possible in this context but the table
indicate together with other similar cases, that it may be worthwhile considering how structural
monitoring systems are designed.
Sensor Type
Temperature Sensors
Settlement Sensors
Displacement Sensors
Strain Gauges
Wind Sensors
Accelerometers
Corrosion Sensors
Table 1
2.
Tsing Ma
Suspension Bridge
115
9
2
110
6
17
0
Great Belt
Suspension Bridge
50
50
20
0
2
0
42
Comparison of structural monitoring systems for Tsing Ma and Great Belt

Suspension Bridges (estimated number of sensors on primary structure).
Structural Monitoring Objectives
In reviewing the design of structural monitoring systems, it is important to focus on the fact that
the product being sold is information. Sometimes information obtained through monitoring proves
needed, at other times information may be obtained by other means, e.g. by experience or
theoretical ways, and finally quite frequently there is no need for further information at all. When
use of structural monitoring systems are considered, it is important to be aware that such system
can and should be tailored to the information needed. That is, for some information it may be
more plausible to obtain the information through small ad hoc monitoring systems being used for
a short period of testing. In other cases, the need for information is extensive and of long term
nature, meaning that it may be worthwhile installing a permanent system.
In short, the objectives of a structural monitoring system must be clarified at an early project stage
of the design or modification of a structure. This is important as the potential information from
structural monitoring systems interacts with decisions to be made with respect to design,
construction, operation and maintenance of the structure. Also, it is important because the
implementation of the structural monitoring system interacts significantly with construction work
-2-
and hence probably also the critical path of the project. Hence in the design of structural
monitoring systems, the objectives and the monitoring concepts must be discussed and determined
as a holistic exercise before irreversible decisions are taken.
Consequently, such design exercise should ideally take place as a cost-benefit study where the pay
off of each structural monitoring item is considered. Here, the pay off is defined by the benefits
(additional revenue or saved costs) compared to the costs of the implementation and operation of
the structural monitoring system. In total for a given structure, the cost-benefit study must
consider how the structural monitoring systems may interact with structural aspects such as:
2.1
design
construction
operation
commissioning
maintenance
requalification
verification for R&D
Design
In the design process, the consultant should consider how a structural monitoring system may be
used as a tool to obtain a sufficient high level of safety with respect to:
new structural concepts

deviation from existing codes and standard
deviation from the imposed design basis
To use structural monitoring as a tool, the approach will typical be to carry out ad hoc testing on
another structure (full-scale or model), or to implement a structural monitoring system allowing
for later design or operational modifications, if the acquired information supports such a decision.
2.2
Construction
In the construction phase, the contractor should together with the owner consider how a structural
monitoring system may be used to increase the efficiency of the applied construction methods.
This could be related to:
enhancement of personal safety

enhancement of structural safety
allowance for working methods deviating from tender conditions
decrease sensitivity of working method with respect to weather and other ambient conditions.
In such cases, structural monitoring is used as a control tool for on-line quality assurances which
allows for increased efficiency in construction methods.
2.3
Commissioning
At the hand over of a structure, the owner, operator or contractor may apply structural monitoring
as a tool for documentation of the fulfilment of design requirements. In this case, the acquired
information may be used to ensure quality as well as to provide an important element in handling
any dispute effectively.
-3-
2.4
Operation
In the operational phase, the consultant or operator should consider how a structural monitoring
system may be used by the operator to control:
2.5
operational risks
efficiency in operation
structural integrity
Maintenance
With respect to maintenance which actually is a part of the structural routine operation, the
consultant and operator should consider how structural monitoring may be used by the operator as
an efficient inspection tool for planning of maintenance and repair tasks. This may comprise
control of:
2.6
durability of the structural

integrity of the structure
efficiency of repairs and maintenance
Requalification
In connection with changes in operation or lifetime extension, the consultant may use structural
monitoring as a tool for requalifying the structure by increasing information about the structural
performance. Also changes in codes may call for requalification of a structure and thus require
further information about the actual structural performance.
2.7
Verification for R&D
Sometimes structural monitoring systems are justified by the need for verification. In such cases
verification is quite frequently used as an objective where the actual information acquired cannot
be used for changes in decisions for the given structure but rather serves a more broad banded
purpose within research and development. Thus such structural monitoring systems will typical be
more to the benefit of the joint industry rather than the contractor, owner or operator of the given
structure. This point should be realised and in cases where structural monitoring systems are setup for research orientated purposes, the most logical project set-up would in general be joint
industrial projects. This approach is reasonable well-known within the offshore industry (e.g.
Ocean Test Structure in U.S.A, and the Tern Platform in the British North Sea ) whereas it is less
well-known within the civil engineering industry.
3.
Design and Implementation of Structural Monitoring Systems
Structural monitoring systems provide a central tool for obtaining information assisting in making
the right decision regarding the objectives outlined in Section 2. The right decision will be a
decision which is right at the time of decision with the information available at the time. Hence,
the need for a structural monitoring system should be seen in the light that it is expected that the
system will provide information which will have an impact on decisions to be made. This implies
answering the question:
Should structural monitoring be carried out at all?

-4-
In next phase, the following questions must be considered:
What would the allowable monitoring costs be?

Of a number of alternative structural monitoring alternatives which should be chosen for
implementation?
Answering these questions may be carried out on a quite pragmatic basis or it may be attempted to
go through a more rigor and rational decision process.
3.1
Design in a Pragmatic Framework
In practice, design of structural monitoring systems is quite frequently carried out on a fairly
pragmatic basis leading to decisions based on compromises between interests involved. The
interests involved are represented by a number of players who are engaged in the design decisions,
cf. Fig.1. The players will represent economical interests as well as safety interests related to users
and authorities.
Consultant
Operator
Contractor
Users
(Customers)
Owner
(Investors)
Authorities
Fig. 1
Players in the decision making on structural monitoring systems
However all players points of views may not only rest on strict rational arguments with respect to
economy and safety issues but also on less rational arguments which may be related to personal,
political or secondary commercial interests. Hence, a complicated mixture of interest may rule the
discussions on the design of a structural monitoring system. This is further complicated by the
multidisciplinary character of the design of structural monitoring systems comprising conventional
structural engineering, construction engineering, instrumentation, operational risk analyses,
probabilistic studies and experimental design. In total, this complexity means that a pragmatic
decision process may very well depend somewhat randomly on the power balance of interest
involved and thus lead to less efficient structural monitoring systems where benefits are not
balanced by the costs.
3.2
Design in a Rational Framework
To avoid ineffective structural monitoring systems, the best approach is to seek to put the decision
process into a rational framework with the objective of identifying the expected benefits and costs
of the structural monitoring items. In the rational approach, a cost-benefit study is carried out
based on a probabilistic assessment of the expected value of gathering information by a structural
-5-
monitoring system compared to the expected costs of obtaining this information.

An initial exercise to be carried out in the early conceptual design of structural monitoring systems
is therefore to consider the value of an ideal structural monitoring system providing complete
information. This value of complete information defines the maximum acceptable costs of
structural monitoring systems to be considered.
A subsequent detailed cost-benefit analysis should of course consider costs and revenue together
with the uncertainties associated with consequences of the decisions as well as uncertainties of the
information obtainable. Hence, a decision on carrying out a structural monitoring task should
therefore consider:
1.
2.
3.
4.
5.
6.
The alternative decisions to be made on the basis of the acquired information

The consequences (economical and other) of a given choice of decision
The degree of perceived uncertainty with respect to consequences of each choice
The further information which appears necessary to chose an alternative.
The uncertainty associated with the information obtainable by structural monitoring
The costs of structural monitoring
1 to 5 relate to the benefits of information obtainable from the structural monitoring system
whereas 6 represent the costs. In the case of ranking different structural monitoring alternatives,
the benefit-cost ratio must be considered, and the most favourable alternative chosen. A rational
decision analysis may be carried out on by a decision tree analysis with the objective of
identifying the most optimal structural monitoring design. Figure 2 shows a simple example for
illustration only, involving a number of sub exercises:
A number of decision strategies are chosen: For instance structural monitoring versus no
structural monitoring.
Discrete outcomes of information, Ii obtained by the structural monitoring are identified: For
instance positive result versus negative result.
A number of project decisions Di are identified: For instance new design vs. conventional
design.
The probability, P(iIi) of the real outcome i given the obtained information, Ii, is obtained
through the well known theorem of Bayes based on apriory probabilities, the information
acquired from the monitoring system, and the uncertainty of the acquired information.
Based on the decision tree set-up:
The economical consequences, Vi of each outcome, i are identified

The expected revenue, Ri of each branch is derived
Non-optimal branches are cut off
The expected revenue by implementation of the structural monitoring system is calculated and
compared by the decision of no implementation of a structural monitoring system.
The decision tree analysis is in itself quite simple, and the largest difficulty arise from
discretizising the decision process into a finite number of alternatives. The problem of associating
costs and probability may seem difficult but it is a problem which exists and has to be solved
independently on how the decision process is organised. Hence, the advantage is that the decision
tree analysis forces the decision maker to face and discuss the consequences and uncertainties
associated with the decision process. This will assist in promoting the real and important costs and
-6-
benefits and thus result in better and more efficient structural monitoring systems. It is therefore
emphasised that this type of rational decision approach are going to become a more important and
frequently used tool in the design of structural monitoring systems.
Outcome
1
Decision
No Monitoring
Decision
D1
Consequence
V1
P(1)
Outcome
Consequence
V2
2
Decision
D2
P(2)
Outcome
Stop
Consequence
0
Outcome
1
Decision
Monitoring
Information
I1
Decision
D1
P(1|I1)
Outcome
2
Decision
D2
Information
I2
Outcome
Stop
1
P(1|I2)
Outcome
2
Decision
D2
Fig. 2
4.
Consequence
V2
P(2|I1)
Outcome
Decision
D1
Consequence
V1
Consequence
0
Consequence
V1
Consequence
V2
P(2|I2)
Outcome
Stop
Consequence
0
Example of Decision Tree for analysis of cost benefits.
Examples: Structural Monitoring in Practice
Currently the design practice may not in general be claimed to be carried out in strict agreement
with a rational decision approach. However, a number of examples from Danish bridges suggest
that rational aspects in structural monitoring can be identified, and there is a sound basis for
maturing the design practice into a rational framework for design of structural monitoring systems.
4.1
Handling of Girder Vibrations in Design
In the late construction phase of the Great Belt Suspension Bridge, excessive vibration response
was observed for the girder of the suspension bridge, [7]. On this basis, an ad hoc monitoring
system was installed and response measurements carried out. The objective was to assess whether
installation of pre-designed guide vanes would lead to acceptable vibrations. This implied
quantification of wind response mechanisms and determination of structural damping for updating design response predictions and wind tunnel tests (with and without guide vanes mounted).
The measurements showed that vortex shedding frequently occurred in connection with the 5th
vertical girder mode for wind directions approximately orthogonal to the bridge axis. The
structural damping was identified to be very low (0.25%) which was the cause for the
unacceptable level of vibrations. On this basis, it was decided to mount guide vanes on the bridge.
-7-
After the guide vanes have been installed, there has been no excessive vibrations.
Fig. 3
4.2
Example of measured build-up of wind induced girder vibration before installation

of guide vanes on Great Belt Suspension Bridge.
Construction, Monitoring of Critical Phases
The Far Bridge in Denmark, inaugurated in 1985, have been subject to a number of full-scale
measurements during construction, [5]. The central span of the cable-stayed bridge was
cantilevered out from the bridge pylons and one of the critical phases was just before the main
span was made continuous at the center. In this phase the wind speeds at the site and the vibration
response of the cantilevered section were closely monitored to assess the risk of severe windinduced vibrations and thus to control the safety of the structure.
4.3
Commissioning of Railway Track
By structural testing, the track quality of the Great Belt, West Bridge (railway bridge) was tested
in terms of ballast stiffness, and a number of deformation and load measurements were carried out
at the bridge expansion joints. The testing was carried out in connection with hand over from the
railway track contractor to Great Belt Ldt., and the subsequent hand over of the railway track to
the railway operator.
For a number of runs with IC3 trains several
sections of the railway track and the track at
all expansion joints were tested.
Fig. 4 Tests with IC3 trains running 140 km/h.
-8-
The test results suggested that sleeper

displacements were larger at the bridge
expansion joints than of the track in general.
Test with a breaking freight train showed that
the loads in hydraulic buffers of the
expansion joints and girder displacements
were acceptable. The results were used as
hand over documentation of the quality and
state of the railway track and structural
elements interfacing with the railway track.
4.4
Operational Safety Monitoring System
As the Great Belt Bridges span waters with high intensity of heavy vessel traffic, a monitoring
system has been installed to control the operational risks associated with ship traffic. Due to the
risk of large economical consequences and possible loss lives, the objective of the system is
primarily to avoid ship impact and secondly to detect an impact. A dedicated surveillance system
monitors ship traffic in the entire Great Belt region. In case of risk of imminent collision with one
of the bridges, alarms will be issued to the authorities responsible for follow-up actions possibly in
the form of requests for immediate closure of the bridges to traffic.
As back-up monitoring system, the bridges are equipped with tilt-meters and displacements
sensors with the purpose of instant detection of a ship impact. Due to the need for immediate
action these structural monitoring systems are continuously monitored by fail-safe SCADA
systems with pre-set alarm thresholds, and the status of the alarms can be surveyed by all parties
involved in emergency tasks.
4.5
Maintenance through Durability Monitoring
For the concrete structures of the Great Belt Bridges and the resund Bridge durability monitoring
systems have been installed. The objective has been to optimise the planning of future
maintenance of concrete structures. Throughout operation, the concrete structures will be exposed
to chlorides contained in air and sea water and the chlorides will eventually penetrate the concrete
cover to the reinforcement and result in corrosion with the risk of excessive maintenance and
repair costs.
For the Great Belt and resund Bridges, the permanent monitoring systems cover about 300 stateof-the art corrosion sensors. Fundamentally, the embedment sensor consists of six dummy
reinforcement bars, and through periodic surveillance of the corrosion state of these bars, the
sensor assembly will monitor the arrival times of chlorides at different depths below the concrete
surface. The sensors monitor the:
penetration rate of chlorides and thus

time to corrosion of the actual
reinforcement
hidden accumulation of maintenance
needs
efficiency of existing or modified
maintenance.
The essential system features are that the

measurements detect on-coming durability
problems at early stages of degradation, and
Fig. 5 Corrosion sensor.
that they provide the operator with
information on the present and future
condition of different structural elements.
Hence, maintenance programmes can
be reviewed on a regular basis and maintenance efforts can be allocated to the structures and
elements hereof as and when needed. The investment in the monitoring system is expected to
result in considerable cost savings during operation of the structures which are designed to last for
100years.
-9-
The durability monitoring system described in this section is believed to be the monitoring system
for the next generation of concrete structures. Since the instrumentation of the Great Belt West
Bridge in 1992-93, similar type of systems, though of minor scale, have today been installed in
about 15 countries world wide.
5.
Conclusion
The paper has reviewed and discussed the design of structural monitoring systems. The paper
suggests that the use of decision tree analyses may be used to ensure efficient structural
monitoring systems through systematic cost-benefit analyses.
Large differences in the layouts of structural monitoring systems, suggest that existing design
practice does not always result in optimum system design due to a pragmatic approach in the
design. However, a number of examples of structural monitoring systems shows that the rational
aspect of system in general is quite obvious. There appears therefore to exist a sound basis for
introducing more rational design methods. Consequently, it is foreseen that in the future, design
practice will mature and to much larger extent comprise the use of rational methods for the design
of structural monitoring system.
6.
References
[1]
Andersen, E.Y. and L. Pedersen, Structural Monitoring of the Great Belt Bridge,
Proceedings of the 3rd Symposium of Strait Crossings, lesund 1994.
[2]
Andersen, E.Y., Monitoring for Structural Health, International Workshop, "Infrastructure

Development", Hong Kong, dec. 1991.
[3]
Lau, C.K., Wong, K.Y. and K.S. Ho, The Wind and Structural Health Monitoring System
(WASHMS) for the three long-span cable-supported bridges in Hong Kong, IABSE,
International Association for Bridge and Structural Engineering, 15th Congress, June 16 20 1996, Copenhagen.
[5]
Ostenfeld, K.H. and H.E. Langs, Full Scale Measurements and Monitoring of Major
Cable-Stayed Bridge, International Conference on Cable-Stayed Bridges, Bangkok, 1987.
[6]
Jensen, J. Laigaard, System Identification of Offshore Platforms, Ph.D. Thesis, Aalborg

University, Department of Building Technology and Structural Engineering, April 1990.
[7]
Jensen, J. Laigaard et al, Estimation of Structural Damping of the Great Belt Suspension
Bridge, to be presented at EuroDyn Conference, Prague, 1999
- 10 -
The Faroe Cable-Stayed Bridge - Maintenance Experience with Major

Components
Matthew L.
BLOOMSTINE
Project Manager
COWI AS
Lyngby, Denmark
Matthew L. Bloomstine, born
1955, received his engineering
Univ. of Denmark in 1987. He
has specialised in inspection,
maintenance and
rehabilitation of major
bridges.
Erik STOLTZNER
Civil Engineer, Danish
Road Directorate,
Copenhagen, Denmark
Erik Stoltzner, born 1944,
Univ. of Denmark in 1969. He
is Area Manager for
Operation and Maintenance
in the Bridge Department.
Summary
A cable-stayed bridge such as the Faroe Bridge is a unique and complicated structure, which requires a
systematic and technically correct maintenance. This paper presents experience from operation,
inspections and maintenance works over the first 14 years of service, with focus on certain major
components. Overall costs for operation and maintenance are also presented, as well as costs for some
particular operation and maintenance activities.
Based on experience from the first 14 years, the estimated annual total cost of operation and
maintenance over the next ten years corresponds to approximately 0.6% of the present value of the cost
of construction. This is quite a low level, which indicates the success of maintenance considerations in
the design and an effective operation and maintenance program.
1.
Introduction
The Faroe Bridges are two bridges with a total length of approximately 3.3 km, which connect the
Danish islands of Zealand and Falster and carry the southern motorway, which connects Copenhagen
with Germany and the rest of Europe. The 1,596 m long northern bridge crosses the sea between
Zealand and the small Faroe Island. The southern bridge, which is the subject of this paper, is a cablestayed bridge between Faroe and Falster with a length of 1,726 m. Construction of the bridges
commenced in 1980 and they were opened for traffic in 1985.
Figure 1 An overview of the Faroe

Cable-Stayed Bridge
1
The superstructure of both bridges is a steel box girder, which is continuous over the entire length with
expansion joints located at each end. The cable-stayed section of the southern bridge has a navigation
span of 290 m with a vertical clearance of 26 m and two side spans, each with a length of 120 m. The
cables are arranged as a fan system of single cables, which are placed in a central plane symmetrically
around the pylons. The cables are of the parallel wire type and protection is provided by PE ducts,
which are grouted with cement mortar. The general arrangement of the bridge is shown in figure 2.
The superstructure was designed and construction supervision carried out by COWI. The
superstructure was built by the contractor Monberg & Thorsen. The Ministry of Traffic under the
Danish State is the owner. The Danish Road Directorate is responsible for operation and maintenance
with assistance from COWI.
Figure 2 General Arrangement of the Faroe Cable-Stayed Bridge

During the design work special attention was given to:
Aesthetics - concerning the bridge itself and the alignment with respect to the existing landscape.
Economy - concerning construction and maintenance costs.
This paper presents operation experience, inspection routines and experience, maintenance works and
costs from the first 14 years of operation concerning the following topics:
2.
General routines for inspections.

Corrosion protection of cables.
Wind induced cable oscillations and the system introduced to minimise these.
Corrosion protection of the steel box girder by means of dehumidification.
Water intrusion - problems and remedies.
Expansion joints which each have a total movement capacity in the range of 1 m. An unexpectedly
early replacement.
Access equipment.
General routines for inspections
Inspections of the bridge are generally carried out according the Danish Road Directorate's manual
"Inspection of Structures, November 1994". For general use concerning structures the manual specifies
the following three types of inspections:
Running/Continuous Inspection, a quick visual inspection with focus on obvious damages that can
impede the structure in fulfilling its purpose with regards to traffic safety.
General Inspection, a thorough and systematic inspection of all components of a structure,
performed at regular intervals between one and six years.
Special Inspection, normally prescribed when a potentially serious damage is observed during a
General Inspection. The inspection is composed of two main activities: a technical inspection to
determine the cause and extent of damages and an economic inspection, which investigates different
repair strategies in order to determine the most economical solution.
2
The above mentioned inspections are implemented for the Faroe Cable-Stayed Bridge in the following
manner.
Running/Continuous Inspection
Running inspection is carried out on a daily basis by the Road Directorate's own bridge personnel. The
bridge master, who originally worked for the contractor during the construction phase, makes at least
one round trip by car over the bridge each weekday. During this trip special attention is paid to major
components such as stay cables. Any obvious damages will be detected and reported at an early stage.
Approximately once a month the bridge master performs a walking inspection through the entire bridge
girder and along the carriageway (on the emergency walkway) with stops at major components.
General Inspection
Due to the size and complexity of the bridge structure as well as access and budget limitations it is not
feasible to conduct a general inspection of all components in the course of one year. Therefore,
individual components or groups of components are inspected each year, in such a manner that all
components are subjected to a general inspection about every fifth year.
Special Inspection
Special inspections are carried out as deemed necessary in the general inspection reports.
3.
Corrosion protection of cables
The stay-cables are parallel wire cables, manufactured by Stahlton AG. Each cable is composed of
between 145 and 277 individual 7 mm diameter wires. The cables are protected by heavy walled
watertight polyethylene ducts (thickness 9.1 to 11.4 mm) and cement mortar, see figure 3. The PE ducts
have the purpose of keeping water and contaminants away from the cables. The cement grout is thought
to protect the cable wires by providing a passive atmosphere in the same manner as concrete does for
steel reinforcement. The PE ducts are continuous from the anchorage at the top of the pylon to the
anchorage in the bridge girder. Both anchorages are located in areas which are protected by
dehumidification (cf. section 5), which prevents intrusion of moisture in the ends of the PE ducts. Hence
the only possible entrance of water and contaminants to the cables is through possible defects in the PE
ducts.
The latest inspection of the cables was carried out in 1996, eleven years after the bridge opening. The
inspection was performed from 72 m boom, which was parked on the carriageway. With this
boom it was possible to access the entire length of
all 36 stay cables and approximately 50% of the
total cable length was visually inspected at close
hand. The general condition of the PE ductswas
evaluated as good. There were no signs of
cracking, discolouring or other signs of ageing.
The only damages which were found were
identified as originating from the construction
period. These were primarily superficial scrape
marks. There were also three patch repairs which
had begun to deteriorate. These will be further
inspected and re-evaluated during the next
inspection, which is planned to be carried out in
Figure 3 Cross section of cable
2001.
4.
Wind induced cable oscillations and minimisation
As the result of the use of different analysis methods including a special computer program considerable
wind induced cable oscillations were expected. Therefore, on-site investigations and measurements
3
were carried out during construction and after completion of the bridge. The investigations generally
confirmed the calculated values and a system to reduce the oscillations was devised. The individual
cables were interconnected by an upper and lower pre-stressed wire system, as shown in figure 4.
The original wire system was composed of steel
brackets with neoprene lining on the cables and
stainless steel wire connecting these. The wires
were wrapped around a thick washer in the
bracket and secured by two wire locks. This
system effectively reduced the oscillations to a
minimal level for all cables with the exception of
the first cable on each side of each pylon. These
cables have special wind conditions, as they are
nearly vertical and are placed so close to the
pylons. Under certain conditions wind and rain
induced oscillations with an amplitude in the range
of approximately 0.5 m occurred.
Figure
4 Interconnection
of stays
After about
four years of service
the first wire in the interconnecting system ruptured. Upon closer
examination it was discovered that all the wires showed signs of wear from abrasion between the wire
the washer. A more robust design for connection of the wires was devised, with 10 mm marine grade
stainless steel wire, a fork and stud terminal allowing easy adjustment of the pre-stressing force on one
end (cf. figure 5) and a fork terminal on the other end. This system has proven to be much more robust.
After seven years of service the first wire ruptured and a close inspection of all the wires was carried
out. One more wire was found to be seriously damaged and two others had signs that damages might
soon occur. All damages occurred in the vicinity of the terminals and all four wires were replaced. The
total number of damages was only a fraction of the number of damages in the original system and these
first occurred after a much longer period.
A solution for the oscillations of the first
cables by the pylons was applied at the same
time that the wire system was improved. The
cables were spirally wrapped with a gradient
of one rotation per meter with a 16 mm
polyethylene duct. Inside the duct is a 4 mm
steel wire which is pre-stressed to
approximately 50 N and fastened at the
bottom end to a spring. This system has
effectively reduced the oscillations to an
acceptable level.
Figure 5 Detail of improved interconnecting

wire system and original pre-stressing method
During the last inspection in 1998 a tensiometer, as shown in

figure 6, was purchased to measure the pre-stressing force in the
individual wires. This device measures the force required to
deform the wire a certain amount in a perpendicular direction
over a fixed length. This force is automatically transformed to the
axial force. The instrument was calibrated by the manufacturer to
the dimension of the wires. The original method required
unloading of each wire by straps and a winch and measuring
with a dynameter as illustrated in figure 5. This method was
cumbersome and time consuming and therefore only a spot
check of the pre-stressing forces were measured. Use of the
tensiometer is quick and easy with a digital display of the prestressing force and allows an accurate measurement. A complete
check of all the pre-stressing forces was quickly performed
during the inspection and maintenance work.
Figure 6 Measuring of pre-stressing force by means of
tensiometer
5.
Corrosion protection of the steel box girder
The internal surfaces of the steel bridge box girder are protected from corrosion by means of
dehumidification. This was the first bridge to be protected internally by dehumidification alone.
The bridge box girder, with a total length of 1,726 m, is divided into 3 dehumidification sections, each
with approximately a third of the total length. A dehumidification plant is installed in each section. Each
plant is composed of a sorption dehumidification unit with fresh air intake and moisture laden air
discharge and a fan unit for circulation. The fan unit is connected to two trapezoidal stiffeners for
forward air flow. The return air circulates through the box girder and through the manholes in the
bulkheads. The system is illustrated in figure 7.
Figure 7 Principle layout of dehumidification system.

5
The dehumidification system is set to generally maintain a relative humidity of approximately 40%, which
is well below the 60% level where corrosion can start. This difference allows for a margin of safety and
sudden falls in the temperature, which correspondingly give sudden rises in the relative humidity. The
dehumidification system has proven to be completely effective, as no signs of condensation or corrosion
have been observed during the regular inspections. Sensors which constantly measure the relative
humidity are connected to the bridge monitoring system.
The control system for the dehumidification system has been adjusted during the operation period in
order ensure a cost effective operation. To take advantage of the low night time price for electricity, the
control intervals have been modified to the following values:
Night:
Day:
Start at RH 40%.
Start at RH 50%.
Stop at RH 35%.
Stop at RH 48%.
With these control levels the dehumidification plants run almost exclusively at night and takes full
advantage of the low rate. The bridge monitoring system has an alarm which is activated when the
relative humidity reaches a level of 50%. This occurs about 10 times a year and the level is automatically
brought down within a couple of hours.
The average electrical consumption per year is approximately 60,000 kWh, which with Danish prices
cost about US$ 7,000 per year. The individual dehumidification plants are serviced once a year for an
average total cost of approximately US$ 1,500. The total average cost for operation and maintenance
of the dehumidification system for corrosion protection of the 160,000 m2 steel surface area is about
US$ 8,500 per year. This corresponds to approximately US$ 0.04 per m2 steel surface per year.
6.
Water intrusion - problems and remedies
Wind conditions on a major cable-stayed bridge are much more extreme than at the local ground level
because of the height of the structure and turbulence caused around pylons and other components. This
can result in water being pressed in through joints and connections which are believed to be watertight.
Water intrusion has occurred at the following locations:
The steel anchorage boxes at the top of the pylons

The neoprene cuffs at the upper cable anchorages
The neoprene cuffs at the lower cable anchorages
During routine inspections it was found that water accumulated in the anchorage boxes during stormy
weather. The tightness of the anchorage boxes was tested by applying a slight overpressure on the inside
and brushing soap water on the outer surfaces. Leaks were identified by the bubbling action of air
seeping out. The joint filler in the joints between the 3 box sections were leaky and the bolt connections
were leaky around the bolt heads, even though these were painted after erection. The joint filler in the
section joints was replaced with a new elastic joint filler and the joints were covered by stainless steel
sheets to protect them from the environment. Elastic joint filler was also applied around the bolt heads.
No further water intrusion has been observed since these improvements were made, however, future
maintenance of the joint filler is foreseen.
Neoprene cuffs overlap the cables and the anchorage pipes where the cables enter the upper
anchorage. These cuffs were originally fastened with two steel clamps on each end and this was thought
to be a watertight connection. Inspections revealed the opposite, as water accumulations were found in
the cuffs. A twofold solution was devised. A strip of silicone filler was applied under the cuff before
tightening the clamps. In order to be able to easily check for water intrusion in the future, plastic hoses
were placed with one end in the bottom of the cuff and one end on the inside of the anchorage box,
which is easily accessible. It is now possible to easily inspect for water and remove it if necessary. A
pulled back cuff and an inspection hose are shown in figure 8. Follow up inspections have proven that
the connections are generally tight now.
6
Figure 8 Opened cuff at upper anchorage and inspection hose and inside the anchorage box
Similar neoprene cuffs overlap the connection where the cables enter the bridge girder. The upper edge
of the cuffs have been sealed with elastic joint filler. However, the cuffs have a tendency to slide down
here, do to cable vibrations, and leave the joint filler behind. Water can then intrude through the
connection. The solution is regular inspection and maintenance of these connections, which are easily
accessible.
As noted in section 3, the stay cables are covered by PE ducts. These ducts cover the cables in the
areas under the cuffs and all the way to the anchorage, such that the water intrusion in these areas has
not been in contact with the cables.
The lessons from this experience are:
7.
Special attention should be paid to making connections watertight, although permanent

watertightness is not possible for some connections.
The areas near the connections should be easily accessible for inspection and means to prevent
water accumulation should provided.
Regular inspections of potential areas for water accumulation should be performed to prevent
damages.
Expansion joints
The 1,726 m long bridge girder was at the time of opening, to the best of our knowledge, the longest
continuos bridge girder in the world and it still ranks among the world's longest. The main reason for
designing such a long girder was to reduce the number of expansion joints to a minimum and thereby
reduce the amount of maintenance associated with these, improve driving comfort and reduce
construction costs. As the bridge girder is so long the requirements to the design movement capacity of
the expansion joints was very high. These requirements led to a total movement capacity for the
southern joint of 1,040 mm and 880 mm for the northern joint. The joints are of the lamella type with
lamellas perpendicular to the bridge axis with sealing strips in between and supporting joists underneath,
which slide in and out of joist boxes.
After only 4 years of service life fatigue damages began to appear. There were two typical types of
damages, cracks running up through the lamella in the vicinity of the support and cracks in the welded
support connection. Damages were regularly registered and repaired. The damages first appeared in the
most sensitive areas, but continued to spread and increase in number. It was concluded that it was not
worthwhile to repair the joints, as there was an inherent problem in the manufacturer's design. After only
9 years of service and many repairs, it was decided to replace the expansion joints.
In order to minimise replacement costs certain main design features were incorporated in the technical
specifications for the new joints. The necessary movement capacity was re-evaluated on the basis of
recorded data for movement of the end bearings from the bridge monitoring system. It was possible to
reduce the original requirement to movement capacity and still maintain an appropriate margin of safety.
The other means of reducing costs was to reuse the existing joist boxes and thereby save manufacturing
7
costs and even more importantly minimise site work.

The expansion joints were replaced in 1996 at an average
cost of approximately US$ 160,000 each. Aside from
some minor adjustments carried out under the guarantee
during the first year, the new expansion joints have been
performing satisfactorily.
Figure 9 Replacement of expansion joints
8.
Access equipment
In order to perform inspections and maintenance works a variety of access equipment is necessary.
Equipment which can be effectively used on the Faroe Bridges and other bridges is owned by the
Danish Road Directorate. Other equipment, which has a limited use and a high cost of procurement, is
rented for certain inspections and maintenance works.
The following access equipment is owned by the Danish Road Directorate:
Skyclimber wire supported platform
A special Skyclimber platform system was designed for use on the Faroe Bridges. It can be assembled
and used for inspection and maintenance of the pylon legs above the roadway, the cable anchorage
boxes at the top of the pylons and the piers. For use on the pylon legs there are permanent fixtures on
top of the pylons for attaching the wires. The platform can be assembled with either 3 sides to cover the
upper regions of the pylons or with 4 sides, encompassing all 4 surfaces of the pylon leg, as shown in
figure 10. For use on the cable anchorage box a smaller platform for two persons can be assembled.
For use on the piers there are movable fixtures which are fastened to the top of the pier. The platform is
assembled with 3 sections corresponding to half of the hexagonal cross section of the piers.
Bucket snooper boom
The bucket snooper boom is truck borne, as shown in figure 10. It has a bucket for two persons and
reaches out over the edge of the bridge girder and can reach about 10 meters in under the bridge girder,
corresponding to half the width of the girder. This boom has mainly been used for spot check
inspections and minor repairs of painting on the bottom of the bridge girder. It requires a great deal of
maneuvering and has limited room and load capacities and is therefore not suitable for thorough
inspections and major repairs. It is used on many different bridges throughout Denmark.
Wire supported platform
A similar system to the Skyclimber, though somewhat simpler, has been used for inspection of the areas
of the pylons below the bridge girder. The platform has one straight section with an adjustable length
and it has been used on a number of major bridges in Denmark. Support girders are erected along both
edges of the bridge girder and wires for suspension of the platform are fastened on these girders.
Fixed local platform
A small fixed platform has been retained from the construction period. It is lifted into position by a truck
borne crane and fastened to the guard rails along the edge of the bridge girder. It is useful for local
repairs on the outer edge of the bridge girder and especially for maintenance of the horizontal bearings
at the northern pylon, for which it has been specially modified.
Figure 10 Access equipment, Bucket snooper boom and Skyclimber platform

The following equipment has been rented for inspections:
Frg 72 m truck borne boom
The Frg 72 m truck borne boom has been used for inspection and maintenance of the upper cable
anchorages and the stay cables. With the boom parked on the roadway it is possible to reach up slightly
higher than the top of the pylons.
Truck borne underdeck platform
A truck borne underdeck platform has been used for inspection of the surface treatment of the entire
bottom side of the bridge girder. The platform is long enough to cover the entire width of the bridge
girder, making possible an entire inspection travelling in one direction.
9.
Overall maintenance economy
After 14 years of operation it is possible to give some figures for the costs of operation and maintenance
of such a structure. Operation activities are here defined as the annual recurrent activities as well as
minor repairs. Operation costs include the owner's administration, running and general inspections,
minor repairs, such as local repairs of pavement and painting, and service, such as cleaning, service of
electrical and mechanical installations and service of the monitoring system. Annual operation costs are
shown below in table 1.
Administration
Running inspections
Substructure
Superstructure
Pavement
Electric and computer installations
Mechanical installations
Electricity consumption and the like
Vehicles
Cleaning
Miscellaneous
Total
20
33
15
13
13
12
40
39
13
25
10
233
Table 1: Annual operational costs in 1,000 US$ (excl. VAT)
At current prices it would cost approximately US$ 115 million excl. VAT to build the Faroe cablestayed bridge. The annual cost of operation is hence approximately 0.2% of the present value of the
cost of construction.
Since the bridge opened it has been necessary to carry out major repairs amounting to approximately
US$ 1,000,000. This amount covers the following works:
Replacement of interconnecting wires on stay cables

Replacement of expansion joints
Replacement of 3. km bitumen joint filler at edges and surface treatment of steel edge plates
Sealing of anchor boxes at pylon tops
Modernization of monitoring and control system
Repair of horizontal bearings at pylon
On the basis of general inspections, necessary repairs over the next ten years are estimated to cost
about US$ 5,000,000. This corresponds to an average annual expenditure of US$ 500,000.Hence, the
estimated average annual repair costs correspond to approximately 0.4% of the present value of the
cost of construction. Estimated repair costs are shown below in table 2.
Items
Substructure
Pylons
Pier
Cable-stays
Erosion protection
Superstructure
Surface treatment
Pavement
Safety barrier
Mechanical equipment
Total repair costs
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
15
80
30
80
30
50
80
300
2100
800
800
200
30
45
160
460
910
50
800
0 2470
110
200
Table 2: Estimated repair costs in 1,000 US$ (excl. VAT)

The total estimated cost of operation and maintenance per year for the next ten years is approximately
0.6% of the present value of the cost of construction. This is quite a bit below the normal figure for
operation and maintenance of bridges in Europe, which is generally accepted to be in the range of 1% to
1.5% of the cost of construction. This indicates that the maintenance considerations in the design work
have succeeded and that the maintenance works are carried out in an effective manner.
10. Conclusion
Despite a high level of consideration to maintenance aspects in the design of the Faroe cable-stayed
bridge and other cable-stayed bridges, there are still a number of challenging problems to be solved
during the service life. There is still room for improvement concerning accessibility and maintainability.
The lessons learned from maintenance should be incorporated into future design work in order to obtain
even better bridges with lower maintenance costs and a long service lives.
10
Emergency Rehabilitation of the Zrate-Brazo Largo Bridges, Argentina

Henrik ANDERSEN
Civ. Eng.
COWI
Lyngby, Denmark
Dietrich L. HOMMEL
Diplom-Ingenieur
COWI
Lyngby, Denmark
Ejgil M. VEJE
Civ. Eng.
COWI
Lyngby, Denmark
Henrik Andersen, born

Technical University of
Denmark in 1989. He has
specialised in design and
rehabilitation of cable
supported bridges.
Dietrich L. Hommel, born

Braunschweig, Germany, in
1966. He has specialised in
Project Management of large
bridge projects.
Ejgil M. Veje, born 1954,

received his civil
Denmark in 1981. He is
Head of Department for
Rehabilitation of Major
Bridges.
Summary
In November 1996 a cable stay ruptured on the Guaz Bridge. The bridge is one of the two almost
identical Zrate-Brazo Largo Bridges across the Paran river between Zrate and Brazo Largo in
Argentina. The two cable stayed bridges carry combined highway and railway traffic and are 550
m long with a main span of 330 m. The bridges carry a 4 lane highway and a single railway track
placed eccentrically and the bridges were opened to roadway traffic in 1977 and railway traffic in
1978. During the service life of the bridge there had been no prior indication of the critical
situation of the cables. The cables consist of high-strength, non-galvanised, parallel wires
protected by cement grout and a PE-pipe. The cable anchorage's are of the HiAm type.
COWI was immediately after the failure retained as consultant by the bridge owner Direccon
Nacional de Vialidad in order to ensure and document the safety of the bridges and to investigate
the causes of the cable failure.
Emergency rehabilitation has been carried out with as limited as possible restrictions to traffic.
The required traffic restrictions has been evaluated on the basis of reliability methods, which have
been applied to establish a Rehabilitation Design Basis.
Possible causes for the severe deterioration of the less than 20 years old cables have been
evaluated. A combination of corrosion and fatigue has been found to be the cause. Whether or not
one of the two by itself would have been sufficient to cause the rupture of the cable has not been
verified.
The corrosion has taken place due to insufficient corrosion protection of the non-galvanised wires.
The likely cause is that the cement grout, which was supposed to be the main corrosion protection,
was insufficient in the anchorage zone due to the presence of a non-protecting epoxy tar.
The fatigue has taken place due to larger traffic loads than accounted for in the original design, but
not least due to large amplitude cable vibrations. These vibrations have theoretically caused stress
ranges well above the endurance limit of the wires. The corrosion has increased the fatigue
stresses locally due to stress concentration.
-1-
1.
Introduction
A large number of cable stayed bridges have been constructed around the world as this bridge type
often is the most economical structure for medium to large spans.
The stay cables are critical structural elements of these
bridges. Unfortunately, the cables are perhaps also the
less robust of the various structural elements.
Larger partial safety factors are for a number of reasons
applied for the design of stay cables than for structural
steel. Cables are designed to the ultimate breaking load
instead of the load causing first yield and account of the
wire bundle effect and the length effect is taken.
Sufficient fatigue resistance, including a certain
allowance for the deterioration of the cables, is
furthermore accounted for.
Fig. 1 Guaz Bridge
Despite this, experience has shown that deterioration and subsequent ruptures of the cables
actually do occur. It has been necessary to replace the cables on even fairly new bridges, such as
Chaco-Corrientes in Argentina and Khlbrand in Germany.
The recent rupture of a cable on a cable stayed bridge in Argentina has confirmed the importance
of a careful design of details, including corrosion protection and of continuous inspection and
maintenance of cables and other critical structural elements.
In November 1996, a cable ruptured on the Guaz Bridge in Argentina without any prior
indication of a possible critical state of the cables. The cable failed close to the bottom socket
which plunged into the river. The socket was recovered by divers later on. The visual inspection
revealed that the failure had taken place about 200 mm from the bottom socket and severe
corrosion and fatigue like rupture of the wires were observed.
COWI Consulting Engineers and Planners was entrusted with the emergency rehabilitation of the
bridges by Direccon Nacional de Vialidad (DNV). The safety of the bridges and their users
should be ensured and documented in both the short and the medium term. Furthermore, COWI
was asked to investigate the possible causes for the cable failure.
The continued operation of the bridge during rehabilitation is often a critical issue for both the
users and the bridge owner/operator, and at the same time the safety of the bridge users is
essential. The bridge might provide the only connection between two destinations or in case of
closure a large detour maybe the only alternative. This is the case for the Zrate - Brazo Largo
Bridges, where a detour of several hundred of kilometres, apart from ferry services, is the only
alternative.
Today a cable stayed bridge is typically designed such that cables can be replaced with more or
less unrestricted traffic allowed on the bridge. Previously, this situation was not always allowed
for in the design. In case of major deterioration of the cables, such as for the present bridges,
traffic restrictions are even more necessary.
A rational planning is required in order to minimise the necessary traffic restrictions. For the
present project, reliability methods have been used for the establishment of a Rehabilitation
Design Basis (RDB). The RDB has utilised the available information, such that the operation of
the bridges could continue with the largest possible traffic amount on the bridges.
The present paper describes the emergency rehabilitation of the bridges and the evaluation of the
possible causes for the deterioration.
-2-
2.
Zrate - Brazo Largo Bridges
The Zrate - Brazo Largo Bridges are important parts of the infrastructure in Argentina providing
a crossing of the national road No 12 over the two main branches of the Paran River - Las Palmas
and Guaz approximately 100 km north west of Buenos Aires.
Fig. 2 Elevation of cable stayed bridges (all units in m)

The two bridges constitute two identical cable stayed steel girder bridges with main spans of 330
m and side spans of 110 m and a total of 16 km concrete approach spans. The bridges carry 4 lanes
of roadway traffic and a single railway track. The present paper only deals with the main cable
stayed bridges.
In Fig. 2 and Fig. 3, respectively, an elevation of the main bridges and a cross section of the girder
are shown.
Fig. 3 Cross section of girder (all units in mm)
The bridges were constructed during the years 1972-77 and opened to roadway traffic in 1977 and
railway traffic in 1978.
The cables consist of high-strength, non-galvanised, parallel 7 mm diameter wires protected by
cement grout and a PE-pipe. The cable anchorages are of the HiAm type. The number of wires in
the cables varies between 103 and 337. These bridges were the first highway bridges constructed
using these cables with this corrosion protection. Previously, only pedestrian bridges had been
-3-
built with this system applied.
3.
Emergency rehabilitation
The emergency rehabilitation of the bridges has included:

Evaluation of the present condition of the bridges through inspection,
non-destructive and destructive testing
Evaluation of the present load conditions by measuring the permanent
cable forces by a vibration method and establish the characteristic
traffic load (rail and road) on the basis of information on the present
traffic
Evaluation of the effect of large amplitude vibrations of the cables
Evaluation of partial safety factors by means of reliability based
methods
Evaluation of the required temporary strengthening and of the most
urgently required cable replacements
Evaluation of required traffic restrictions in order to ensure adequate
safety at all times
The various investigations and analyses are described in more detail in
the following chapters.
4.
Fig. 4 UT-inspection of
original cable.
Temporary
strengthening also
shown.
Initial Investigations
Immediately after the rupture of the cable on the Guaz Bridge, both bridges were closed to all
traffic. The contractor Albano - DyCASA -Freyssinet was working on another bridge
rehabilitation project in Argentina for the same bridge owner and was immediately asked to
replace the failed cable.
In order to establish detailed information on the actual state of the cable stays, it was decided to
carry out various non-destructive testing of the cables. Initially, a visual inspection was carried out
of all cables, including cutting of windows into the PE-pipes and removal of cement grout as
close as possible to the anchorages in order to observe the degree of corrosion of the wires.
The cable had failed just in front of the anchorage at the lower end. Ultrasonic inspection was
hence an appropriate method for identification of similar problems and this was initiated.
Furthermore, it was decided to carry out a measurements of the permanent load in the cable by
means of the vibration method both in order to receive a confirmation of the permanent load and
to reveal any major changes in stiffness caused by significant wire breaks.
The investigations revealed that a number of cables had corroded wires and that a large number of
wires were broken on some cables. The cable with the largest damage had 62% damaged wires.
On top of this, its neighbouring cables had damages between 41% and 20%. This was indeed
extremely critical and could have caused a collapse of the entire bridge possibly within a
reasonable short period of time.
The ultrasonic inspections were carried out at top and bottom sockets and both cables in the
roadway side and in the railway side were investigated. Only cables in the roadway side turned out
-4-
to be damaged, and the damages were significantly larger at the bottom socket than at the top
socket.
The cable force measurements revealed that the permanent forces in the cables had changed with
up to 20% since similar measurements were carried out just before the inauguration of the bridges.
The largest reduction in permanent force was found for the cable which also had the largest
damage according to the ultrasonic inspection.
In order to ensure the safety of the bridges, it was decided to introduce a temporary strengthening
by means of Freyssinet strands, typically 8 Nos. T15 strands each with a breaking strength of 265
kN. This was applied both to the damaged cables and to the neighbouring cables. This temporary
strengthening served two purposes. Firstly, the strands took over a part of the permanent load in
the original cables, and secondly they would carry part of the variable load. An example of the
strengthening is shown in Fig. 4.
Initial tensile and fatigue tests were carried out on test specimens from the failed cable. The tensile
test revealed a reduction in tensile strength for test specimens taken from the corroded part of the
wires, whereas test specimens from a non-corroded part of the wires still were complying with the
original specification.
5.
Rehabilitation Design Basis
Framework
The design of new structures are typically carried out according to existing Codes. With respect to
reassessment of existing structures and especially unique structures such as cable stayed bridges,
the existing codes are not always fully covering or they are too conservative. A specific design
basis is hence established in such cases. For the Zrate-Brazo Largo Bridges, it was decided to
establish a Rehabilitation Design Basis (RDB) using reliability methods in order to have a basis
for rational planning. All available information on materials, the structure and traffic has been
used as input for the RDB.
The Load and Resistance Factor Design (LFRD) approach has been utilised. Therefore, it has been
necessary to establish the following parameters on the basis of the available information:
Characteristic loads and resistance variables
Partial safety factors for load and resistance
Load combination factors
For a detailed description of the Rehabilitation Design Basis reference is made to [1] and [2].
By means of the RDB, it has been possible to evaluate the allowable traffic on the bridge in a
number of situations during the emergency rehabilitation.
Material properties
Initially tensile and fatigue tests were carried out on test specimens from the failed cable. Later on,
as more cables were replaced, a larger number of tests were carried out on specimens from 2
additional cables.
The tested wires were categorised as corroded or non-corroded and both tensile strength and
fatigue characteristics were established for the two groups of wires. In both cases a reduction of
the characteristic tensile strength was found.
It shall be noted that only cables with large damages at the sockets have been replaced and hence
tested so far. The test specimens described as non-corroded were, however, obtained some metres
from the failure or corrosion zone, and the degree of conservatism is expected to be limited. The
tests revealed that the characteristic breaking strength was 1,525 MPa for non-corroded wires and
-5-
1,460 MPa for corroded wires. This could be compared to the value of 1,670 MPa in the
specification.
Fatigue test were carried out in order to establish both SN-curves and a Smith-diagram for the
wires.
Traffic load
The bridges carry both highway and railway traffic, and on the basis of the available information
characteristic loads for both highway and railway traffic have been established. The characteristic
loads have been developed for the most critical stay cable. This value has then conservatively been
used for all cables as only limited savings could be obtained by utilising individual characteristic
loads for the various cables.
The complete information required for the establishment of the characteristic load has not been
available and some parameters have hence been estimated on the basis of data from Denmark and
Europe in general.
The highway traffic load for unrestricted traffic is shown in Table 1. This traffic load shall be
compared to the original design load which was 7 kN/m in each of the four lanes. The highway
traffic load has hence increased significantly. Besides, the characteristic traffic load has been
established for a number of traffic restrictions too.
The characteristic railway load was determined to 70.2 kN/m compared to the original design
value of 62.8 kN/m in general with an increase to 96.4 kN/m over 35 m length to cover the larger
intensity of the locomotive.
The traffic load which has been allowed on the bridge during the emergency rehabilitation has
varied. During the various emergency phases and dependent on the level of knowledge of the
existing structure, the traffic has been restricted to such an extent that sufficient safety of both the
bridges and for their users were ensured.
Location
UDL (kN/m)
Axle loads (kN)
Lane 1
28.9
2 x 300
Lane 2
9.4
2 x 200
Lane 3
9.4
2 x 100
Lane 4
9.4
Table 1
Characteristic highway traffic load
Immediately after the rupture of the cable the bridges were closed to all traffic as the state of the
remaining cables were unknown. As more information was established and strengthening of the
bridges initiated, it was possible to open the bridges to restricted traffic. The general allowable
traffic (both roadway and railway, including the possibility for simultaneous roadway and railway
traffic) was evaluated a number of times as the rehabilitation progressed.
During each of the cable replacements the necessary additional restrictions to the allowable traffic
were evaluated in addition.
Partial safety factors and load combination factors
The general derivation of partial safety factors for the cables is described in [1] and [2]. Only a
brief description is included in the following.
The partial safety factors for the cables depend, among others, on the deterioration of the cable
due to fatigue. On the basis of the fatigue tests of wires from 3 cables and the expected fatigue
loading derived on the basis of the traffic information and the vibrations of the cables, the partial
-6-
safety factor has been determined for the cables. By this method, it has been possible to evaluate
the partial safety factors for various periods of time. Ultrasonic inspections has been utilised in
order to carry out a monitoring of any possible development of the wires breaks and to allow for
an update of the safety.
Partial safety factors for highway and railway loads and load combination factors was established
on the basis of recommendations in JCSS [3].
6.
Cable vibrations
Large amplitude vibrations of a number of the cables have taken place. Visual observations of
amplitudes of approximately 1 m have been made. Furthermore, pressure marks in the PE-pipes at
locations, where these pass out of the anchorage boxes at the girder level indicate vibrations of a
similar magnitude.
In the present scope of work no detailed investigations of the causes for the cable vibrations have
been carried out. It shall, however, be noted that the vibrations do not only occur during rainy
conditions. The vibrations are hence not believed to be rain-wind induced vibrations. Preliminary
investigations have pointed out that the vibrations might be caused by vibrations of the deck as
there are common eigenfrequencies of some cables and the deck. The wind tunnel testing, carried
out in connection with the original design, found that the deck could vibrate heavily in case of
winds with a large vertical component (attack angles 2.5o - 5o) at wind speeds between 10 and 20
m/s [4].
The theoretical peak stresses caused by these vibrations are even larger than the permanent tensile
stresses in the wires. This was determined under the conservative assumption that the wires are
fully restrained at the entrance to the socket. No guide deviators were included in the original
design of the bridges in order to limit the bending stresses.
There is no doubt that vibrations of this magnitude shall be prevented on cable stayed bridges as it
only is a matter of time before deterioration due to fatigue will be initiated.
For the cables replaced during the emergency rehabilitation, it has been recommended to install
"wind ropes" connecting the cables in order to limit the amplitude of the vibrations. Furthermore,
it has been recommended to install guide deviators in order to decrease the bending stresses at the
socket.
7.
Cable replacements
A total of 13 cables have been replaced during the

emergency rehabilitation corresponding to all cables
with more than 6% damage. The damaged cables have
been concentrated on the Guaz Bridge where in total
10 cables have been replaced, including the one that
ruptured in November 1996.
The existing cables have been removed by an
untraditional method. The state of the cables were
uncertain and a brittle failure could not be ruled out. It
was therefore decided not to use a hydraulic jack for the Fig. 5 Removal of existing cable by
de-stressing and removal of the cables and an alternative
cutting of individual wires
method was developed by the contractor. The PE-pipe
and the cement grout was removed along almost the
-7-
entire length of the cable manually by personnel in a lift. The cable was then secured by a steel
rope connected to the tower head and the girder in order to control it during the following
operations, where each individual wire of the cable was cut as shown in Fig. 5.
Initially, all wires of the original cable were cut before the temporary strengthening (Freyssinet
strands) was removed and the remaining top and bottom part of the cable could be removed. Later
on the temporary strengthening was removed before the wires were cut.
Before, during and after the removal of a cable, levelling of the bridge deck and measurement of
permanent forces in the cables were carried out in order to ensure that the bridge behaved as
foreseen. In all cases a good agreement between the predicted and the measured movements and
forces was found.
The new cables installed consist of a number of Freyssinet strands. They are designed to a higher
service load than the original cables as noted in a previous chapter.
A complete rehabilitation of the bridges is expected to be carried out during 1999/2000.
8.
Cable failure causes
Possible causes for the severe deterioration of the less

than 20 years old cables have been evaluated. The
evaluation has revealed that a number of factors have
contributed to the deterioration of the cables. A
combination of corrosion and fatigue damage have
caused the failure of one cable and large damages to a
number of other cables. The cross section of the cable
that failed in November 1996 is shown in Fig. 6.
The corrosion system applied for the cables were stateof-the-art at the time of the design and construction of
the bridges. In this particular case the system has,
however, not provided sufficient corrosion protection of Fig. 6 Ruptured cable
the non-galvanised wires. The cement grout, which was
supposed to be the main active corrosion protection, was insufficient in the anchorage zone due to
the presence of a non-protecting epoxy tar. Following intrusion of water through defects in the PEpipe or due to condensation of water inside the PE-pipe, corrosion has been initiated.
The fatigue damage has taken place due to larger traffic loads than accounted for in the original
design, but not least due to large amplitude cable vibrations. These vibrations have theoretically
caused stress ranges well above the endurance limit of the wires. The corrosion has furthermore
increased the fatigue stresses locally due to stress concentration.
7.
Conclusion
A combination of corrosion and fatigue have caused severe deterioration of a number of cables on
the Zrate - Brazo Largo Bridges. The corrosion protection of the high strength wires have proven
insufficient as the grouting of the non-galvanised wires have been inadequate close to the bottom
cable socket.
The traffic load has been larger than accounted for in the original design and this has increased the
fatigue loading. A significant contribution to the fatigue load is, however, stemming from large
amplitude vibrations of the cables. The bending stresses in the parallel wires have not been limited
-8-
by deviator guides, which could have reduced the stresses significantly.

All available information on the structure, the materials and the traffic has been used to establish
a Rehabilitation Design Basis and by use of reliability methods, it has been possible to minimise
the required traffic restrictions during the various phases of the emergency rehabilitation.
The most severe deteriorated cables have been replaced. Final rehabilitation of the bridges is
expected to be carried out during 1999/2000.
8.
References
[1]
Hommel, D., Faber, M.H., Maglie, R., Aspects of safety and operation of bridges during
rehabilitation. Proceedings of the international symposium on advances in operation
and maintenance of large infrastructure projects. Copenhagen, Denmark, 1998.
[2]
Hommel, D., Veje, E., Engelund, S., Zrate-Brazo Largo Bridges, Rehabilitation Design
Basis. IABSE Conference Rio de Janeiro, 1999.
[3]
Joint Committee on Structural Safety (JCSS), Background Documentation, Eurocode 1

(ENV 1991, Part 1: Basis of Design, March 1996.
[4]
Atkins, An aerodynamic investigation of the Zrate-Brazo Largo Bridge, Argentina, 1971.

Prepared for Techint-Albano SA
-9-
Field Observation and Vibration Test of the Tatara Bridge

Kazunori YAMAGUCHI
Civil Engineer
Honshu-Shikoku Bridge Auth.
Onomichi, Japan
Yasuhiro MANABE
Civil Engineer
Honshu-Shikoku Bridge Auth.
Onomichi, Japan
Nobuyuki SASAKI
Civil Engineer
Mitsubishi Heavy Ind., Ltd
Hiroshima, Japan
Kunihiro MORISHITA
Civil Engineer
Mitsubishi Heavy Ind., Ltd
Hiroshima, Japan
1. Introduction
The Tatara Bridge ,the longest span cable-stayed bridge in the world, has a total length of 1,480
meters and a center span length of 890 meters and is located at the Onomichi-Imabari Route. The
bridge consists of the girder (height is 2.7 meters), the reverse Y-shaped tower (overall height is
220 meters) and 168 cables (21 rows8 planes), and has a steel-concrete composite structural
system. Pre-stressed concrete girders are placed at the ends of each side spans.
As for the dynamic structural properties, this bridge is tend to appear the sway easily because of
a long-span and the behavior of numbers of long cables(maximum length is 460 meters). The
dynamic stability against wind and earthquakes required to be evaluated and in advance the
further investigations such as a wind tunnel tests and natural vibration analyses with a model of
this bridge had been carried out. Moreover, several field observations and vibration tests have
been performed since the beginning of the construction.
This paper introduces the outlines of following two terms out of these field vibration
investigations by comparing with the results of previous analyses .
(1)Field vibration test
The field vibration test had been performed from November to December 1998 when the
construction of pavement almost completed. The exciters were used and vibrated the girder in
the vertical directions to investigate three vertical bending and two torsinal vibration modes, and
in the horizontal directions to investigate three bending vibration modes in the normal direction
of the bridge axis. The horizontal vibration in large amplitude is the first trial on such a longspan bridge.
The aims of field vibration tests are to confirm the dynamic properties of important vibration
modes by comparing with the results of analyses, especially structural damping, and to verify the
vibration design of this bridge, furthermore, to make these observed data useful for the future
design of the long-span cable-stayed bridge by investigating the phenomena of a coupled
vibration consisting of girder and cables and the causes of structural damping.
(2)Response observation in the strong wind
The field observations in the strong wind have been performed for about two years, from the
beginning to the completion of the construction in order to keep safety under construction, to
evaluate the efficiency of the countermeasure to prevent cable vibration and to investigate the
response properties against strong natural wind.
The construction and above field investigations has been completed and now these data are being
analyzed in detail. In this paper, the basic data analyzed already until now are reported.
2. Field Vibration Test

2.1 Testing procedure
The exciters used in the field test were newly developed especially for the excitation in such a
long period of this bridge. The specifications are shown in Table-1. These are a pair of large
exciters and enable to work in both vertical and horizontal directions. The exciting direction is
changed by using each steel frames supporting motors and weights, respectively corresponding
to the tests for the vertical and horizontal vibrations.
In the vertical vibration test, two exciters were placed at both ends of the cross section of the
bridge and they were operated in the same phase and the opposite phase to excite vertical
bending and torsional vibration modes respectively. In the horizontal test, two exciters were
operated in the same phase.
The exciters were installed at the point in distance of three eighth of the mid-span from the end
of girder in the longitudinal direction. The location of transducers are shown in Fig.-1. The
servo-type accelerometers were used in order to measure vibrations of girder, tower and cables.
Two sensors were installed on the principal cables shown in Fig.-1, and accelerations in both
vertical (in-plane of cable) and horizontal (out-of-plane cable) were respectively measured.
These sensors were properly replaced to cables that swayed greatly during field test. As regards
all bearings, the strain gauge type displacement meters were installed and slip of bearings in the
horizontal direction were measured. Furthermore, the wind direction and wind speed were
measured with Ace Vane type anemometer, and outside temperature was measured for reference.
The situation of field vibration test is shown in Photo.-1.
Photo.-1 Situation of field vibration test
Fig.-1 Subject bridge and measuring points
Table-1 Specification of the exciter
2.2 Theoretical analysis

Previously theoretical analysis were carried out to
compare with field test results. The natural
vibration analysis was performed using three
dimensional frame model of a whole structure
system shown in Fig.-2. The cable models had 50
nodal points with mass that were linked by spring
elements each other, and cable sag was taken into
account. As the above cable models, coupled
system consisting of girder and cable, have too
many nodal points and analysis must be performed
using large dimensional matrix, it is difficult to
analyze directly using a whole structure system
model. Therefore, the mass condensation method 1)
and the partial structural combined method 2) were
used as the efficient and accuracy analysis method.
Fig.-2 Analytical model
2.3 Test results and discussion

(1)Natural frequency and vibration mode
Examples of amplitude response curves are shown
in Fig.-3. These indicate the amplitude-excitation
frequency relations and phase-excitation frequency
corresponding to the vertical and horizontal
excitations. In the lower frequency range, there are
single peaks, respectively corresponding to the 1st
symm. vertical bending and 1st symm. horizontal
bending mode, however, the shapes of curves in
the higher frequency range are complicated. There
are several peaks in the amplitude response curve
of each modes and similarly on the phase response
curves. The cause would be considered as coupled
vibration between girder and cable. As the
minimum frequency of cables is 0.25 Hz on the top
cable and another cables have higher frequency, the
coupled vibration is tend to be excited in higher
frequency range than 0.25 Hz. Such phenomena of
coupled vibration appeared in the sine excitation
test.
Fig.-3 Amplitude and phase response curve
In this paper, the peak frequency and amplitude distribution, on which the maximum response of
girder was obtained, were selected respectively as the natural frequency and the mode shape of
each modes out of many data. The comparisons of natural frequency and mode shapes between
test results and analysis results are shown in Table-2 and Fig.-4. As the results, there is a less
difference than 10% on the natural frequency and only a few differences on the mode shapes.
(2)Structural damping
Structural damping in logarithmic decrement was calculated with free vibration data
corresponding to each modes obtained by suddenly stopping excitation in the resonant condition.
Examples of free vibration data are shown in Fig.-5 and it indicates that the damping changes
according to the change of amplitude of the girder and damping tends to increase proportional to
the amplitude of girder. The structural damping at each modes was calculated as the average
damping in the range from the maximum amplitude to the amplitude reduced to 70% of the
maximum one. The results are shown in Table-2. It is important that the structural damping
change greatly according to each vibration directions.
*)
*) : reference data, under investigation
Table-2 Test results
Fig.-4 Comparison of mode shapes

Regarding to the vertical bending and
torsional vibration modes in the vertical
excitation, the damping is =0.01 0.05, and
on the horizontal bending vibration mode in
the horizontal excitation =0.10 0.20, that
are much larger than ones of vertical
vibration. As the reason, it is considered that
the attachments such as fairings installed at
the ends of the cross section of girder have
influence of these phenomena. There are
several attachments such as fairings for wind
resistance and handrails at the ends of the
cross section. These are separated at regular
interval 10 meters in the longitudinal of the
girder not to propagate stress, and linked to
each members with loose-tightened bolts. On
the horizontal vibration, it is predicted that as
the amplitude become larger, the more bolts
Fig.-5 Free vibration and structural damping
would slip and the friction force would have
more influences of the damping. In order to
verify this prediction, the slip of bolts were measured with displacement meters installed on the
joint of fairings at the mid-point of the mid-span. As the results, it was confirmed that the above
prediction might be right.
Sudden stop of exciter
(3)Cable vibration
This chapter reports about the phenomena of
coupled vibration between girder and cables
obtained in the field vibration test.
Girder(3/8L)
a)Linear coupled vibration

There happened several coupled vibrations between
girder and cable on numbers of vibration modes
mentioned in 2.3. General coupled vibrations were
excited when the excitation frequency of the girder
became near one of cable. In this case, this
phenomenon is considered as the linear coupled
vibration. The mode in this phenomenon is almost
same as one of the result of natural vibration
analysis mentioned in 2.2.
Cable(C53)
Fig.-6 Non-linear coupled vibration
b)Non-linear coupled vibration (parametric excitation)

On the other hand, non-linear coupled vibration was excited, when the frequency of the cable
was nearly equal to half of the excitation frequency of the girder. This is considered as the nonlinear coupled vibration excited due to fluctuation of cable tension. This phenomenon had been
confirmed in another field vibration test 3) .
The non-linear coupled vibration was excited at resonance of 1st asymm. tosional vibration mode
in this test. C53 cable suddenly began to oscillate in half frequency of exciting frequency during
the girder was being excited. The comparison of the amplitude between girder and cable is
shown in Fig.-6. After non-linear coupled vibration was excited, exciters were stopped due to a
sudden increase of the cable amplitude. The 1st asymm. torsional vibration mode would be
excited only in case that an asymm. wind blow in the normal direction of the bridge axis in a
stormy condition, therefore it is considered that non-linear coupled vibration would be rarely
excited. However, the confirmation of the possibility that such a vibration will be excited
including reinvestigation of data in wind tunnel test will be carried out quantitatively.
3.
Response Observation in the Strong Wind
3.1 Observation procedure

The response observations of dynamic
properties against strong wind have been
performed for about two years from the
beginning of the construction until the
completion. The observation points are
represented in Table-3. The meteorological
data are measured with anemometers and
rain gauge, and the response observations
are performed on the girder and cables. The
number and points of observations were
changing properly according to the
construction progress. These observations
were performed automatically using the
developed automatic system.
Table-3 Observation points
3.2 Summary of observation data

A vast amount of observation data has been stored up
since the beginning of the observation, and these are
now in progress. Here, next two kinds of data are
shown as the typical observation data.
(1)Response of the girder against typhoon
During the observation, several typhoon had hit the
bridge. The typhoon which carried the strongest wind
was the 98-10th typhoon which passed on 17th and 18th
October. At this time, the construction was almost
completed including the pavement construction. The
properties of wind and response of the girder at this
time are shown in Fig.-7.
The maximum instantaneous wind speed was recorded
up to 30.2m/sec and the average wind speed of ten
minutes was 22.3m/sec. The wind direction was in the
normal direction of the bridge axis (270 degrees from
the bridge axis) at this time. The girder vibrated the
most greatly in both vertical and horizontal directions
at the same time when the maximum wind speed was
observed, and the maximum amplitude was 6.9cm and
8.5cm respectively on the horizontal and vertical
directions. The girder vibration and results of its
Fourier analysis are shown in Fig.-8.
It is confirmed that the
amplitude in 0.1Hz prevailed on
the horizontal vibration and ones
in 0.23Hz and 0.27Hz on the
vertical, and these results are
almost same as the field
vibration test results mentioned
in the section 2.
Fig.-7 Time histories of wind

and responses
(2)Cable vibration
The dimple processed type
polyethylene pipes were used on
the cables as the aerodynamic
countermeasure
for
cable
vibration. These were developed
in order to prevent rainvibration.
In
the
field
observation under construction,
Fig.-8 Responses of girder
the amplitude of the cable was
measured in order to confirm
the efficiency of this countermeasure. A vast amount of observation data on the cable vibration
in the strong-windy and rainy conditions such as a seasonal wind and a typhoon has been stored
up since the beginning of the observation, however there have happened no injurious rainvibration until now and the effect of the countermeasure has been confirmed. On the other hand,
there often happened the high frequency vibrations assumed to be a vortex-induced vibration on
the cable under construction. The properties of observed vortex-induced vibration are explained
below.
The relations between wind speed

and prevailing cable amplitude, and
between wind speed and frequency
of cable are respectively shown in
Fig.-9(a) and (b). In Fig.-9(a), it is
confirmed that the cable amplitude
increases from the lower wind speed
and the maximum amplitude
happens at about 5.0m/sec and in the
higher wind speed range the
amplitude gradually decreases. In
Fig.-9(b), the prevailing frequency of
the cable increases proportional to
wind speed. As the Strouhal number
calculated using these observation
data distributes in the range from
0.15 to 0.18 on the all cables, above
mentioned prevailing frequency is
considered as one of a vortexinduced vibration.
(a)
(b)
Fig.-9 Relation between wind speed and
cable response
On the past happened accidents of a

vortex-induced vibration, several
Fig.-10 Effect of filling rubber seal
examples have been reported that a
vortex-induced
vibration
was
prevented by filling rubber seals (additional damping to a cable is about 0.005) into the
entrance of cable in the girder. On the Tatara Bridge, rubber seals were filled into several the
entrance of cable and the effect was confirmed on basis of field observation data. The
comparison of cable vibration between unfilled cable and filled in almost same wind conditions
is shown in Fig.-10. It is confirmed that the amplitude is reduced because of filled rubber seal
and this countermeasure for a vortex-induced vibration is efficient to such long cables.
4. Summary
Firstly, the results of field vibration test are summarized as follows;
(1)The frequencies in filed tests and ones in analyses are almost same on the important 8 modes
taken into account, furthermore, each modes are very similar.
(2)The structural damping on the vertical bending and torsional modes are relatively small by
comparing with the test results of another long-span cable-stayed bridges, however these satisfy
the design value =0.02. This cause is considered that this bridge is very long-span bridge and
easy to be influenced by cable vibration. On the other hand, the structural damping on the
horizontal bending mode is large and this cause is presumed that bolts installed on the fairings
and so on would slip according to the horizontal girder vibration.
(3)Linear coupled vibration between girder and cable was excited in the condition that the
frequency of the cable vibration was equal to exciting frequency of the girder. On the other hand,
non-linear coupled vibration (parametric excitation) was excited at resonance of 1st asymm.
torsional vibration mode and at that time the frequency of the cable has become the one second
of exciting frequency of the girder. It is considered that non-linear coupled vibration would be
rarely excited, however, the confirmation of this phenomenon will be performed.
Next, the results in the response observation of strong wind are summarized as follows;
(1)The maximum gust response of the girder was 7 8 cm on both vertical and horizontal
vibration against a typhoon blowing at about 30m/sec (the maximum instantaneous wind speed)
and these are small relatively.
(2)It was confirmed that there was no injurious rain-vibration and the aerodynamic
countermeasure (using dimple processed type polyethylene pipes) was efficient. In addition, it
was confirmed that the countermeasure by filling rubber seal into the entrance of the cable in the
girder enabled to prevent a vortex-induced vibration often happened.
These data are under analysis still now and the new results would be presented in the near future.
Acknowledgements:
The authors would like to thank Professor YAMAGUCHI Hiroki, Department of Construction
Engineering, Saitama University in carrying out the experiment.
Refference
[1] K. Morishita et al. : Non-linear Seismic Response Analysis of Long Span Bridge, Proc. of
2nd World Conference of Structural Control, June-July 1998.
[2] K. Washizu et al. : Handbook of Finite Element Method II, Application Part, Baifukan, pp.57,1983(in Japanese).
[3] I. Okauchi et al. : Field Vibration Test of a Long-Span Cable-Stayed Bridge Using Large
Exciters, Journal of Struct. Engrg., JSCE, Vol.14, No.1, pp.83s-93s, April 1997.
Second Monitoring and Surveillance of the Response of a Cable-stayed Bridge

Roberto GOMEZ
Dr. Eng.
Institute of Engineering, UNAM
Mexico, Mexico
David MURIA-VILA
Dr. Eng.
Mexico, Mexico
Roberto SANCHEZ-RAMIREZ
M. Eng.
Mexico, Mexico
J. Alberto ESCOBAR
Dr. Eng.
Mexico, Mexico
Summary
Experimental data obtained during a second non destructive testing program carried out on the
Tampico bridge is presented. A comparison of ambient vibrations of the deck is performed with
the results of a similar set of tests carried out in 1988; some dynamic parameters and modal
assurance factors are calculated and compared. From the new testing program, parameters of
pylons and cable forces are derived from ambient vibrations of the piers and stays, respectively.
Results from static and dynamic loading are presented as well. Information from the
experimental testing program is used to calibrate a numerical model that will be used to evaluate
the structural safety of the bridge.
1. Introduction
The Tampico bridge was the second cable-stayed bridge built in Mexico. It is located in the
northeast state of Tamaulipas on a highway along the Gulf of Mexico. The bridge crosses the
Panuco River and carries four lanes of traffic with sidewalks and a central barrier. Its total
length is 1543 m distributed in three sections: a main cable-stayed span of 878 m length and two
bridge viaducts. The viaduct on the left shore is 476 m long and the one on the right shore is 189
m. A hybrid type of deck was constructed. Steel box girders were used on most of the central
span (360 m) and pre-fabricated prestressed concrete box girders for the remaining part of the
deck. The cable-stayed system comprises 44 cables arranged in a semi-fan layout.
Regarding the importance of this bridge, just before its opening in 1988, the Bridge Department
of the Ministry of Communication and Transportation decided to carry out an experimental
program in order to determine the dynamic properties of the superstructure. Natural frequencies
and mode shapes were calculated from acceleration time histories recorded during ambient
vibration testing. From the results of a pullback test, damping characteristics were obtained as
well (Muria-Vila et al, 1991; AEIC, 1988).
Currently, new live loads, wind conditions, temperature and mass changes, corrosion effects,
relaxation of cable forces and prestress losses may have modified the structural properties of the
bridge. A surveillance of its structural integrity can be accomplished by evaluating its current
dynamic response.
In this paper the results of a new and extended experimental program are presented. The same
locations of the recording points, used in the 1988 field-testing program, are used in this study.
Ambient vibrations are used to calculate natural frequencies of the superstructure and to derive
frequency functions (transfer, coherence and phase angle functions). From the analysis of these
functions modal shapes are derived.
Based on modal assurance criteria such as MAC and COMAC factors, the mode shapes obtained
during the 1988 monitoring program are compared to the ones derived in 1998.
Time histories of accelerations produced by dynamic loads were also recorded. The response of
the bridge under different arrays of trucks, of known weight, running at different velocities was
studied. Comparison of the ambient and dynamic responses is presented and evaluated in terms
of changes in natural frequencies of the deck.
In addition to the measurement of accelerations on the deck, acceleration records were registered
on the whole set of stays. Based on the theory of cable vibrations, this information was used to
derive their natural frequencies and to calculate the magnitude of their tension forces. These
values are compared to those obtained by means of hydraulic jacks, devices regularly used for
this type of measurements.
The results of this study will contribute to a safer operation and optimal maintenance programs
of the bridge.
2. Instrumentation
During the 1988 field testing program only acceleration records were obtained. For this new
evaluation of the bridge, in addition to accelerometers, strain gauges, LVDTs and anemometers
were used. These sensors were located along the length of the main central steel deck (360 m
length) and on some locations along the height of pier 13. Figure 1 shows the location and
numbering of the acceleration recording points, two sections (C and R) where 10 strain gauges
were placed and the location of two photo electric cells. All of them located between pylons 13
and 14. Figure 2 is used to display the position of the strain gauges inside the box section of the
deck.
Figure 1. Recording points along the deck.
Figure 2. Location of strain gauges, sections C-C and R-R.
With respect to the LVDTs, two of these were placed on the top surface of the foundation
cylinder (caisson) of pier 13 (Figure 3): one on each side of the column of this pier. The purpose
of these sensors was to register the vertical movements induced during the loading of the bridge,
and consequently calculate the rotation of the pier with respect to a vertical axis perpendicular to
the road surface.
As mentioned, in addition of the recording points of accelerations located on the deck, four
points were selected along the height of piers 13 and 14: one at the top, one at the bottom, one at
the bifurcation of the pylon (to form an inverted Y shape) and one on the deck.
Wind velocities and directions were monitored using two anemometers, one located at the top of
pier 13 and the other one located on the deck (see Figure 3). During the dynamic tests velocity of
the trucks was calculated with the information provided by the photo electric cells (Figure 1).
All sensors were connected through amplifiers, filters and signal conditioners to the personal
computer where the measurements under static load and time histories were recorded. A more
complete description of the registration scheme can be found in the correspondent technical
report (Gmez et al. 1998).
Figure 3. Location of sensors along pylon 13.
3. Field Testing
Static loading was produced by five trucks, six axles each, positioned on different arrays along
the length of the main central deck, between pylons 13 and 14. The average weight of the trucks
was 65 t and the maximum static load applied to the bridge was 326 t. A simultaneous recording
of vertical displacements and longitudinal strains was carried out. During this part of the
experimental work traffic was not allowed on the bridge.
Ambient vibration of the bridge was produced by the regular traffic and two trucks passing in
opposite lanes and in opposite directions. Wind velocity was measured; the maximum value
never exceeded 9 m/s.
Dynamic excitation of the superstructure was produced by the same trucks used for the static
loading program, but running along the bridges.
Vibrations produced during the ambient and dynamic tests were registered using several arrays
of accelerometers oriented in different directions. Time history data was recorded for each event
of the instrumentation program (Gomez et al, 1998). However for the dynamic testing, time
histories of strains and displacements were also recorded.
A well known random signals analysis (Bendat and Piersol, 1986) was used to process ambient
vibration records. An average of different number of readings and a suitable windowing
process" was taken to calculate frequency responses: power spectrum, transfer, coherence and
phase angle functions. Emphasis was placed on the determination of natural frequencies of the
superstructure and pylons and mode shapes of the superstructure.
On the other hand, ambient vibrations (traffic and wind) and vibrations produced by pulling a
rope tied to the cable, were registered for each cable. These vibrations were processed to derive
natural frequencies, which were then used to calculate tension forces of the stays.
4. Results
4.1 Results of Field Tests Using Static Loads
For different load cases, Table 1 shows some values of longitudinal strains registered with the
strain gauges. The location of points C1 to C5 and R1 to R5 can be seen in Figure 2. For the first
test (PE1), only one truck, on the inner lane, was used. Its third axle was placed coincident with
the middle line of the central span. For the second test (PE2), a second truck was placed on the
outer lane and parallel to the first one. The fourth test (PE4) included two more trucks but on the
opposite side of the road with similar positions to the ones used for test PE2. Table 1 shows the
different kind of strains produced (tension and compression) and their variation along the deck. It
is also observed that when the number of trucks increases, the values of strains increase as well,
and that values of strains at stations C are greater than the ones registered at section R. Because
the high torsional stiffness of the box section, the effect of loading only one side of the bridge is
slightly observed in sensors C5 and R5.
When using five trucks in convoy on the inner lane, the results obtained are shown on the row
denoted as PE5. This pattern of strains shows significantly better the deformed configuration of
the deck, this is, zones of positive and negative flexure. The maximum values of strains
registered were produced during test PE4.
On the other hand, values of vertical displacements obtained with the LVDTs were used to
calculate the rotation of the pier. Results found were of very low magnitude, which confirmed
the high stiffness of the pylon.
TEST C1
C2
PE1
-39
-68
PE2
-74
-100
PE4
-181 -198
PE5
-127 -129
TEST R1
R2
PE1
-2
10
PE2
25
29
PE4
19
11
PE5
14
9
+
tension; - compresion
C3
95
176
361
257
R3
-40
-70
-146
-169
C4
86
174
351
256
R4
-43
-69
-134
-163
C5
108
202
382
270
R5
-46
-67
-129
-176
Table 1. Microstrains registered.

4.2 Results of Ambient Vibration Tests
Vertical, transverse and longitudinal movements were recorded simultaneously at different
points along the deck. Typical results of power spectrum and transfer functions obtained at
different pairs of points along the deck are shown in Figure 4. Because of the location of these
points, this recording scheme allows the observation of different levels of vibration along the
deck, the analysis of the degree of symmetry of the response and the identification of frequency
values associated to high contents of energy. In spite of the difference in magnitudes, this figure
shows how the peaks of the transfer functions oscillate around the same frequencies.
Figure 5 shows another kind of frequency functions for another set of pairs of points located on
opposite sides of the roadway. Besides the transfer function, coherence and phase angle
functions are also used to identify possible natural frequencies of the superstructure. The
analysis of vibrations recorded on opposite sides of the deck allows the identification of torsional
natural frequencies, which are associated to high coherence values and opposite phase angles.
For example, frequency functions of points 11 and 12 in Figure 5, show that the value of 0.98 Hz
has an associated value of 180 degrees, and for the same value, the phase angle of points 15 and
16 is of the same magnitude but opposite sign. On the other hand, the coherence is close to 1.0
for points 11 and 12 and close to 0.8 for points 15 and 16.
The above facts demonstrate the similarity of vertical vibrations on both sides of the deck and
the torsional movements around an imaginary longitudinal axis of the superstructure. The same
evidence was found comparing the frequency functions at pairs of points on different sections
along the deck (Gmez et al. 1998).
Transverse vibrations were recorded as well. Some results are presented in Figure 6 where
power spectrum functions reveal that the functions a more similar and that is easier to perceive
the coincidence of peaks. When comparing these functions with those presented above, a slight
coincidence of peaks can be observed in the range of high frequencies. This fact shows that the
vertical and transverse responses of the superstructure might be slightly coupled.
Simultaneous recording of different type of vibrations at the same transverse sections were also
considered. Figure 7 is shows a complete set of frequency functions obtained from the signals
recorded, simultaneously, in the vertical (V) and transversal (T) directions, on location number
27.
Figure 4. Power spectrum and transfer function of vertical movements.

Based on the whole set of results obtained, the 3D vibratory response of the superstructure was
studied. Thus, some significant natural frequencies calculated for some modes of vibration of the
central span are shown in Table 2. In the same table the values obtained during the 1988 testing
are reported as well. It is worth to mention that although the same recording scheme was used
during the two experimental programs (1988 and 1998), the goals of the first one were less
ambitious and consequently a minor number of frequencies and modes were identified,
particularly for asymmetric and transverse modes. The values reported in Table 2 show that the
differences are very small for the first three vertical and the first transverse and torsion natural
frequencies. Differences are higher when comparing the rotational modes. However, a trend to
lower values of frequencies can be observed.
Figure 5. Power spectrum, phase angle and coherence functions of vertical movements.
Figure 6. Power spectrum and transfer functions of transverse movements.
Figure 7. Frequency functions for vertical and of transverse movements recorded

simultaneously.
Mode
V
TORSION
(year)
1988
1998
1988
1998
1S
0.40-0.42 0.39
1.06-1.11 0.98
1A
0.54 -0.59 1.95-2.05*
2S
0.90-0.94 0.88
3.10-3.16* 2.49-2.98*
2A
1.17
3.37
3S
1.44-1.49 1.42
4.48-4.52 4.80-4.90*
3A
1.76
5.32-5.38*
4S
2.73-2.80 2.20
4A
2.25
5S
3.75-3.85 3.52-3.61 5A
4.15-4.40 * coupled with T direction; A: asymmetric; S: symmetric
T
1988
0.45
-
1998
0.44
0.73
1.71
1.86-1.90
4.20
4.80-4.90
5.33
5.77-5.86
6.25
6.64-6.60
Table 2. Natural frequencies of the deck (in Hz).

From the spectral analysis of signals recorded along the deck vertical and transverse modal
configurations for the superstructure were identified; some of them are presented in Figure 8 in
plan or elevation views; the longitudinal axis is used to define the length along the instrumented
section of the deck. Furthermore, the modal shapes for the two testing programs carried out
(1988 and 1998) can be used to evaluate qualitatively their dynamic correlation base on the
Modal Assurance Criterion (MAC) and the Coordinate Modal Assurance Criterion (COMAC)
(Allemang and Brown 1982; Lieven and Ewins 1988).
Because of the lack of information generated during the 1988 testing, only the first three vertical
modes were used for the calculation of these criterion. The obatined magnitudes for the MAC
factors were 0.997, 0.970 and 0.967 for the first, second and third mode, respectively. These
results show that the correlation is better when comparing the first mode and the maximum
differences are obtained for the third modal shape.
Figure 8. Modal shapes identified.

On the other hand, the COMAC values calculated using the first three vertical mode shapes are
presented in Table 3. Location numbers are referred to Figure 1. The values show a very good
agreement for points near the center of the span. The major differences are obtained when
comparing points near the piers. This might be produced by a change of section (steel to
concrete) in these zones.
Location 7
Value
11
15
19
23
25
28
29
30
0.947 0.975 0.997 1.000 0.973 0.961 0.985 0.997 0.997 0.670
Table 3. COMAC factors calculated with the vertical modal shapes.

With respect to the recorded data obtained along the height of pylons 13 and 14, the analysis
revealed a severe coupling effect of the longitudinal movements with the vibrations of the deck
in the vertical direction, particularly for the first and third modes. Some results associated to the
most relevant vibrations are presented in Table 4.
Mode
1
2
3
Direction T
0.83
1.17-1.22
1.37-1.47
Direction L
2.30-2.40
2.78-3.03
-
Table 4. Significant natural frequencies for pylons (in Hz).

4.3 Cable Vibration Tests
Ambient and forced vibration of cables produced records of movements that were utilized to
derive frequency functions and dynamic parameters. Possible coupling of the vibration of the
stays with the deck was considered identifying frequencies associated to higher modes and
assuming a linear relationship of their magnitudes and the correspondent mode number (Robert
et al, 1991). Calculated values were in the range of 0.59 to 2.17 Hz. With these values and based
on the classical theory of cables, tension forces in the stays were calculated. These tension were
compared with the ones measured by means of hydraulic jacks, a task that is carried out in the
bridge every three or four years. In spite of some uncertainties in the weight and length of the
cables, results showed differences in the range of 4 to 19%.
4.4 Results of Dynamic Vibration Tests

As mentioned, additional dynamic tests were performed on the deck. Serial and parallel arrays of
trucks were used and different crossing velocities were defined. Figure 9 shows a typical set of
segments of time histories obtained during the crossing of a convoy of 5 trucks on the low
velocity lane. These graphs correspond to the passing of the first truck. On the right side of each
history a number is printed to identify the recording location. The similarity of acceleration
records at points 23 and 24, on opposites sides of the bridge, and the change of sign of the
longitudinal strains records, points C4 and R4, provided a broader insight of the high torsional
stiffness of the superstructure and the induced flexure along its deck. Though Figure 9 only
shows the time histories produced by the first truck, the remainder are very similar.
Figure 9. Selected time histories of strains and accelerations.

The knowledge of the response of a bridge under static and dynamic loads prompted the
calculation of a dynamic amplification coefficient for the strains. The average velocity of the
trucks was in the range of 11 to 19 km/h. The average ratio of strains produced under the same
dynamic and static loads was of 1.06.
A pull back test was also carried out. A steel cable tied to the central span of the superstructure
was released from a towing boat producing the free vibration of the bridge. Acceleration records
were registered. With this information, the fundamental frequencies for the vertical and torsional
component of the deck were identified again; their values were 0.38 and 1.01 Hz, respectively.
The critical damping ratio of the deck was calculated using the logarithmic decrement. For the
vertical vibration the estimated damping was 0.0028 and the correspondent one obtained in 1988
was 0.0045. For the torsional vibration these values are 0.0036 and 0.0052.
5. Numerical modelling
Using a computer program (SAP90, 1995), a finite element model of the bridge was constructed.
Cables and piers were modeled with truss and beam elements, respectively; shell elements were
used for the box girder deck. The model comprised the superstructure between two hinges,
including four piers and pylons 13 and 14. Static load conditions studied were self-weight and
live loads produced by the trucks used during the testing program. Fundamental frequencies and
modal shapes were also calculated using the mathematical models. The results of the
experimental program were used to calibrate the finite element model.
6. Conclusions
Results of two ambient vibration testing programs of a cable-stayed bridge were presented and
compared. In spite of some differences observed, the whole sets of the two testing agree fairly
well, although a general trend in the reduction of frequency values was observed. On the other
hand, the results calculated numerically showed a fair agreement with the values experimentally
obtained.
The work presented is part of an entire structural safety evaluation program of the Tampico
bridge. Results of this study and the numerical model will be used to propose maintenance and
corrective actions in order to enhance the behavior of the bridge.
Acknowledgements
This study was funded by CAPUFE. The authors acknowledge the support and encouragement
of Juan Tellez and Miguel A. de la Cruz. Assistance of the technical staff of CAPUFE at
Tampico is appreciated. G. Rodrguez, M. A. Mendoza also collaborated.
References
[1].
[2].
[3].
[4].
[5].
[6].
[7].
[8].
[9].
[10].
[11].
[12].
[13].
[14].
[15].
[16].
[17].
[18].
[19].
[20].
AEIC, S.C. 1988. Ambient vibration testing of the Tampico bridge, Technical report in
spanish, Mexico
Allemang R.J. and D.L. Brown 1982. A correlation coefficient for modal vector analysis.
Proceedings, 1st International Modal Analysis Conference, Society for Experimental
Mechanics,
110-116. Bethel, Conn.
Bendat, J. S. and A. G. Piersol 1986. Random Data : Analysis and Measurements
Procedures,
Wiley Interscience, New York, N.Y.
Gmez, E., D., M.A. Mendoza, D. Muri-Vila and Badillo E. 1998. Design and
Implementation of an Acquisition and Signal processing System, Technical report in
spanish, Instituto de Ingenieria, UNAM, Mexico.
Gmez, R., D. Muri-Vila, R. Snchez and J. A. Escobar 1998. Static and dynamic
loading and
mathematical modeling of the main central span of the Tampico bridge, Technical report
in
spanish, Instituto de Ingenieria, UNAM, Mexico.
Gmez, R. , C. King, D. Muri-Vila and C. Montoya 1991. Numerical and experimental
determination of dynamic properties of the Tampico bridge. Proceedings, International
Symposium for Innovation in Cable-Stayed Bridges, 111-121, Fukuoka, Japan.
Lieven, N. A. J. and D. J. Ewins 1988. Spatial Correlation of Mode Shapes, the
Coordinate
Modal Assurance Criterion (COMAC). Proceedings, 6th International Modal Analysis
Conference, Society for Experimental Mechanics : 690-695, Bethel, Conn.
Muri-Vila, D., R. Gmez and C. King 1991. Dynamic structural properties of the
cable-stayed
Tampico bridge, Journal of Structural Engineering, ASCE, 117 (11): 3396-3416
Robert, J.L., D. Bruhat and J. Gervais 1991. Mesure de la tension des cables par methode
vibratoire, Bulletin de liaison de Laboratoire des Ponts et Chausses, 173, mai-juin: 109114
SAP90 Structural Analysis Programs 1995, Computers and Structures Inc., Berkeley,
California.
Dynamic Tests on Vasco da Gama Cable-Stayed Bridge

lvaro CUNHA
Assistant Professor
University of Porto
Portugal
Elsa CAETANO
Assistant
University of Porto
Portugal
Rui CALADA
Assistant
University of Porto
Portugal
Raimundo DELGADO
Associate Professor
University of Porto
Portugal
Summary
This paper describes the dynamic tests performed on Vasco da Gama cable-stayed bridge so as to
accurately identify its most significant modal parameters from the aerodynamic and seismic point
of view. The main results obtained on the basis of ambient vibration and free vibration tests are
presented and correlated with the corresponding parameters provided by the finite element model
developed at the design stage. Moreover, the usefulness of application of a laser Doppler sensor
to perform dynamic measurements on a large number of stay cables is also stressed, as well as
the feasibility of experimental evaluation of dynamic amplification factors associated to the
passage of heavy traffic.
1. The Vasco da Gama cable-stayed bridge
Figure 1: View of Vasco da Gama cable-stayed bridge

The Vasco da Gama Bridge is the new Tagus River crossing in Portugal, 17300m long, including
three interchanges, a 5km long section on land and a continuous 12300m long bridge, recently
constructed close to the area of EXPO-98 international exhibition. It includes a cable-stayed
component (Figure 1) over the main navigation channel with 420m central span and three lateral
spans (62+70.6+72m) on each side, corresponding to a total length of 829.2m between transition
piers. The bridge deck is 31m wide and is formed by two lateral prestressed girders, 2.6m high,
connected by a slab and by transverse steel I girders. It is continuous along its total length and it
is suspended at level 52.5m by two plans of 48 stays connected to each tower. The two towers are
H shaped and 147m high above a massive zone at their base as protection against ship collision.
2. Objectives and tasks of the dynamic tests

Due to the high proneness of long span bridges to be affected by aerodynamic instability
problems, as well as to the high seismic risk of the Southern part of Portugal, the dynamic
behaviour of Vasco da Gama cable-stayed bridge has been extensively studied using both
experimental and numerical approaches[1,2]. In particular, dynamic tests have been performed
by the University of Porto[3] in order to experimentally identify the most relevant modal
parameters of the cable-stayed bridge from the aerodynamic and seismic behaviour point of view,
and correlate them with the corresponding parameters provided by the 3-D numerical model
developed by EEG (Europe tudes Gecti, Villeurbanne, France), using the finite element
program Hercules. These dynamic tests, described in this paper, involved the following main
tasks: (i) preliminary measurements for evaluation of the levels of acceleration signals and
identification of an appropriate reference section; (ii) development of an ambient vibration test
for identification of natural frequencies and mode shapes, involving tri-directional measurements
at 58 distinct points along the deck and towers; (iii) performance of response measurements
under the passage of heavy trucks, passing over a hood plank, to increase the vertical
accelerations; (iv) development of a free vibration test by sudden release of a mass of 60t
suspended from the deck, in order to accurately indentify modal damping factors; (v)
performance of dynamic measurements on some of the longest stay cables so as to identify global
and local natural frequencies, both using conventional piezoelectric accelerometers and an
interferometry laser sensor; (vi) experimental evaluation of dynamic amplification factors
(DAFs) associated to the passage of heavy traffic at different speeds and along several lanes.
3. Ambient vibration test

29U
29D
28U
20U
19U 28D
18U
15U
27U
26U
LISBOA
(North)
1U
2U
3U
1D 2D
4U
3D
5U
4D
6U
5D
7U
8U
26D
16U
12,13,14U
27D
9U
8D
10U
9D
11U
15D
19D
17U
16D
17D
21U
20D
22U
21D
P4
18D
PS
23U
22D
24U
23D
25U
24D
25D
P6
P5
SETBAL
(South)
12,13,14D
10D
11D
z
x
7D
6D
PN
P3
P2
P1
Figure 2: Schematic representation of the bridge with indication of the measurement

sections used in the ambient vibration test
This test was developed performing vibration measurements with two triaxial accelerographs at a
given reference cross section (section 10, 1/3 span North) and with two other successively placed
at 28 different mobile sections along the deck and towers (Figure 2). In all sections, the pairs of
sensors were located laterally, upstream and downstream, always oriented according to the
orthogonal referential xx (longitudinal direction), yy (transversal) and zz (vertical). Due to the
relatively low level of signal captured, appropriate amplification factors were used, leading to a
precision of at least 0.015mg (1g/216). The measurements were conducted in order to create time
records with a sampling rate of 50Hz and enabling average spectral estimates with a frequency
resolution inferior to 0.01Hz. Beyond these sensors, previously programmed and synchronised
using a portable PC, an anemometer was still used to regularly measure the wind speed. With the
purpose of increasing the signal level in the vertical component, vibration measurements were
also carried out during the passage of a heavy truck, with a mass of 30t, over a hood plank, 4cm
high, placed at 1/3 span. Inspection of the time records obtained showed a significant variation of
the structural response during the ambient vibration test, which was essentially due to oscillations
of the wind speed. The wind speed measured at midspan changed between 1 and 22m/s, leading
to a ratio between maximum and minimum r.m.s. values of the vertical acceleration at the
reference section of 28.
The identification of natural frequencies was based on the peak values of averages of normalised
acceleration power spectra (NPSD) corresponding to each section (downstream, upstream, halfsum and half-difference signals), as well as on the coherence values associated to the
simultaneous measurements at the several pairs of points[4]. The frequency resolution of the
average spectral estimates obtained, on the basis of time records of 16 minutes, was 0.006Hz.
Figure 3 shows average normalised auto spectra (ANPSD) and cross-spectra (ANCPSD),
corresponding to vertical (Z) and transversal (Y) acceleration components, obtained as average of
NPSD and NCPSD spectra associated to the measurement in 23 different sections, taking into
account both the half-sum and the half-difference signals (upstream-downstream). Figure 4 also
shows the amplitudes of NCPSD spectra and the corresponding coherence functions associated to
simultaneous measurements at sections 10 and 16. Inspection of all the average spectral estimates
obtained[3] permitted to identify the values of natural frequencies summarised in Table 1, in the
range 0-1.15Hz, in correspondence with natural frequencies provided by the numerical model.
1.00E+00
1.00E+00
Half-sum
Half-diff.
1.00E-01
ANPSD (Y)
ANCPSD (Z)
1.00E-01
1.00E-02
1.00E-03
1.00E-02
1.00E-03
1.00E-04
1.00E-05
1.00E-04
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
Frequency (Hz)
0.4
0.5
0.6
0.7
0.8
0.9
Frequency (Hz)
(a)
(b)
Figure 3: Average normalised spectra associated to: (a) vertical acceleration (half-sum and
half-difference signals, upstream-downstream); (b) transversal acceleration (half-sum
signal)
1.00E-01
0.4
1.00E-04
0.2
1.00E-05
1.00E-06
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Frequency (Hz)
(a)
0.7
0.8
0.9
NCPSD
0.6
1.00E-03
1.00E-01
0.8
1.00E-02
Coherence
NCPSD
1.00E+00
0.8
1.00E-02
0.6
1.00E-03
0.4
1.00E-04
Coherence
1.00E+00
0.2
1.00E-05
1.00E-06
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Frequency (Hz)
(b)
Figure 4: NCPSD spectra (amplitude) of the half-sum signal of (a) vertical acceleration and of
(b) transversal acceleration at sections 10 and 16, and corresponding coherences
The identification of modes of vibration with frequencies in the range 0-1.15Hz was based on the
estimates of transfer functions (using estimator H1), and corresponding coherences, relating the
ambient response at the reference section (half-sum and half-difference signals, upstreamdownstream) with the response at the other measurement sections along the deck and towers. The
ratios between the values of those transfer functions related to each natural frequency (linear
magnitude) associated to the several sections led to the absolute value of the modal components,
the corresponding signal having been evaluated on the basis of the phase evolution. Figure 5
shows some of the identified modal shapes of the deck, also presenting the corresponding
numerical modes, as well as some modal components identified using the free vibration test,
described in the next section.
Calculated
frequencies (Hz)
0.2624
0.3185
0.4287
0.4386
0.6268
0.6077
0.6268
0.7600
Identified frequencies (Hz)
Type of mode of vibration
0.298
0.341
0.437
0.471
0.572*/0.590*/0.599*/0.619*/0.624*
0.651
0.693*/0.707*/0.718*/0.755*
0.817*
0.895*/0.917*
0.985
1.129*
1st transversal bending

1st vertical bending
2nd vertical bending
st
1 torsion + transversal bending
2nd torsion + transversal bending
3rd vertical bending
nd
2 torsion + transversal bending
4th vertical bending
3rd torsion
th
5 vertical bending
(*) multiple modes, low signal level
Table 1: Identified and calculated natural frequencies

Freq.=0.341Hz- 1st vertical bending mode
0.001
0.001
Vertical modal component
Transversal modal component
Freq.=0.298Hz- 1st transversal bending mode
0.0005
0
Numerical
-0.0005
Ambient vib.
-0.001
Free vib.
-0.0015
-0.002
-0.0025
-0.003
10950
11050
11150
11250
11350
11450
0
-0.001
-0.002
-0.003
Numerical
Ambient vib.
Free vib.
-0.004
-0.005
10950
11550
11050
11150
Point
Ambient vib.
Free vib.
0.002
Vertical / transversal
modal component
Numerical
0.003
0.001
0
-0.001
-0.002
-0.003
-0.004
11050
11150
11250
11350
11450
0
-0.001
-0.002
-0.004
-0.005
10950
11550
Numerical, Z
Ambient vib., Z
Ambient vib., Y
Free vib., Z
Free vib., Y
-0.003
11050
11150
Freq.=0.572Hz- 2nd torsion mode
11350
11450
11550
Freq.=0.619Hz- 2nd torsion + transversal bending mode
0.005
0.01
0.004
Numerical
Ambient vib.
Free vib.
0.003
0.002
0.001
0
-0.001
-0.002
-0.003
Numerical, Z
Ambient vib., Y
Numerical, Y
Ambient vib., Z
0.008
modal component
11250
Point
Point
-0.004
0.006
0.004
0.002
0
-0.002
-0.004
-0.006
-0.008
11050
11150
11250
11350
11450
-0.01
10950
11550
11050
11150
11250
11350
11450
11550
Point
Point
Freq.=0.651Hz- 3rd vertical bending mode
Freq.=0.817Hz- 4th vertical bending mode

0.01
0.008
0.01
11550
0.001
0.004
Ambient vib., Z
Numerical, Z
Free vib., Z
0.006
0.004
0.002
0
-0.002
-0.004
-0.006
-0.008
-0.01
10950
11450
Freq.=0.471Hz- 1st torsion + transversal bending mode
0.005
-0.005
10950
11350
Point
Freq.=0.437Hz- 2nd vertical bending moment
-0.005
10950
11250
11050
11150
11250
11350
11450
Point
Figure 5(a): Some modal shapes of the deck
11550
0.008
Numerical
0.006
Ambient vib.
0.004
0.002
0
-0.002
-0.004
-0.006
-0.008
-0.01
10950
11050
11150
11250
Point
11350
11450
11550
Freq.=0.985Hz- 5th vertical bending mode
Freq.=0.895Hz- 3rd torsion mode

0.01
0.005
modal component
0.003
0.002
0.001
Ambient vib., Y
Numerical, Y
Ambient vib., Z
Numerical, Z
0.004
0
-0.001
-0.002
-0.003
-0.004
-0.005
10950
11050
11150
11250
11350
11450
0.008
Experimental, Z
Free vib., Z
0.006
0.004
0.002
0
-0.002
-0.004
-0.006
-0.008
-0.01
10950
11550
11050
11150
11250
11350
11450
11550
Point
Point
Figure 5(b): Some modal shapes of the deck
4. Free vibration test

The free vibration test was performed not only to check the main results of the ambient vibration
test previously developed, but essentially to permit an accurate identification of the damping
factors associated to the modes of vibration with a more significant contribution to the dynamic
response of the bridge, particularly under wind loading. For that purpose, a mass of 60t was
suspended from one point of the deck close to the section 1/3 span North, near the upstream
border, and was subsequently released, originating a vibratory phenomenon recorded during 16
minutes by 6 triaxial accelerographs, located at the sections 1/3 and 1/2 span (upstream and
downstream). In order to verify that the wind would not affect the accuracy of evaluation of the
structural modal damping factors, introducing a component of aerodynamic damping, the wind
speed was permanently measured at midspan, connecting the anemometer to a spectral analyzer,
the maximum wind velocity not exceeding 2.5m/s.
The identification of natural frequencies was then made on the basis of the peaks of the FFTs of
the acceleration time series (Figures 6-7). Each one of these series was formed by 32768 points
sampled at 50Hz, corresponding to a time of acquisition of 655.36s, which led to a frequency
resolution of 0.0015Hz.
With regard to the mode shapes, these were identified applying a band-pass 10 poles Butterworth
digital filter around each one of the natural frequencies identified, and comparing the amplitudes
and phases of the filtered signals at different points of measurement. Figure 5 shows the modal
components identified by this procedure, which are clearly in good agreement both with the
mode shapes obtained by the ambient vibration test and with the modal configurations calculated
numerically.
1 .0 0 E + 0 0
F F T a m p litu d e (m g
F F T a m p litu d e (m g )
1 .0 0 E + 0 0
1 .0 0 E -0 1
1 .0 0 E -0 2
1 .0 0 E -0 3
1 .0 0 E -0 4
1 .0 0 E -0 1
1 .0 0 E -0 2
1 .0 0 E -0 3
1 .0 0 E -0 4
0 .1
0 .2
0 .3
0 .4
0 .5
0 .6
F re q u e n c y (H z )
0 .7
0 .8
0 .9
0 .1
0 .2
0 .3
0 .4
0 .5
0 .6
0 .7
0 .8
Frequency (Hz )
(a)
(b)
Figure 6: Amplitude of the FFT of the half-sum signal of vertical acceleration (upstreamdownstream) at (a) 1/3 span North and (b) 1/2 span
0 .9
1 .0 0 E + 0 0
F F T amplitude (mg
F F T am plitude (m g)
1 .0 0 E + 0 0
1 .0 0 E -0 1
1 .0 0 E -0 2
1 .0 0 E -0 3
1 .0 0 E -0 4
1 .0 0 E -0 1
1 .0 0 E -0 2
1 .0 0 E -0 3
1 .0 0 E -0 4
0 .1
0 .2
0 .3
0 .4
0 .5
0 .6
0 .7
0 .8
0 .9
0 .1
0 .2
0 .3
0 .4
0 .5
0 .6
0 .7
0 .8
0 .9
(a)
(b)
Figure 7: Amplitude of the FFT of the half-difference signal of vertical acceleration (upstreamdownstream) at (a) 1/3 span North and (b) 1/2 span
At last, the identification of the modal damping factors was done on the basis of the decay of the
envelope of the filtered signals obtained applying also a band-pass 10 poles Butterworth digital
filter around each natural frequency in the range 0-1.0Hz (Figure 8). As the estimates of the
modal damping factors depend on the level of vibration, different exponential regressions were
performed in correspondence with different time intervals of the free response. Table 2 shows the
average estimates obtained, as well as the respective intervals of variation.
1/2 s pan, ups tream, v e rtica l component
B utte rw orth filter, 10 poles , F inf=0.4Hz ; F s up=0.5Hz
1/2 s pan, half-s um ups tream-dow ns tream, trans v. component

Fitting of the 1s t tors ion damping modal coef. (logarithmic dec.)
2
1 0 .0 0
E n v e lo p e
8 .0 0
A c c e le ra tio n (m g )
A cce leration (mg)
6 .0 0
4 .0 0
2 .0 0
0 .0 0
-2 .0 0
-4 .0 0
F itte d c u rv e 2 0 0 -5 0 0 s
1 .5
1
=0.24%
0 .5
-6 .0 0
-8 .0 0
-1 0 .0 0
100
100
150
200
250
300
T ime (s )
350
400
200
450
300
400
500
600
700
T im e (s )
(a)
(b)
Figure 8: Identification of the modal damping factor associated to the natural frequency
0.467Hz. Analysis based on the measured response at 1/2 span upstream
Identified
Modal damping factor (%)
natural
frequency (Hz)
Mean value
Int. of variation
0.295
1.23
0.87-1.73
0.338
0.21
0.16-0.40
0.456
0.23
0.19-0.27
0.467
0.24
0.14-0.36
0.591
0.34
0.30-0.39
0.647*
0.37
0.653*
0.20
0.707
0.78
0.71-1.12
0.814
0.48
0.45-0.54
0.982
0.74
0.67-1.24
(*) Multiple modes; identification based on the half-power bandwith method
due to the difficulty of application of digital filters
Table 2: Identified natural frequencies and modal damping factors
Type of mode
1st transversal bending
1st vertical bending
2nd vertical bending
1st torsion + transversal bending
3rd vertical bending
Its worth mentioning that, beyond the application of the logarithmic decrement method, the
frequency domain MDOF identification algorithm RFP (Rational Fraction Polynomial
Method[5]) was still used, based on transfer functions relating the response measured at each
point with the excitation. As the excitation was not actually measured when the mass of 60t was
suddenly released from the deck, the transfer functions were evaluated assuming the input as an
impulsive load, with unit magnitude, applied during a very short period of time, leading to a
spectral content of the excitation with an almost constant intensity over the frequency range of
interest.
1/3 span North, half -sum upstream / dow nstream, vertical component
1/3 span North, half -dif . upstream / dow nstream, vertical component
10000
10000
Measured
1000
Measured
1000
Identified
100
100
10
10
Identified
1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.2
0.3
0.4
Frequency (Hz)
0.5
0.6
0.7
0.8
0.9
Frequency (Hz)
Figure 9: Identification of modal parameters using the RFP method.

Measured and synthesised transfer functions.
Figure 9 shows two of the transfer functions obtained following this procedure, as well as the
corresponding synthesised transfer functions, evaluated on the basis of the modal parameters
identified using the RFP algorithm.
5. Dynamic measurements on stay cables

5.1 Identification of natural frequencies
The performance of dynamic measurements on stay cables of cable-stayed bridges is often
required to assess different problems of great interest in the context of the design, construction
and maintenance of this type of structures, such as: (i) the evaluation of cable tensions, whose
knowledge is critical to the correct alignment and distribution of internal forces in the finished
bridge, and whose temporal evolution can provide interesting indications concerning the
structural health; (ii) the assessment of fatigue problems in stay cables caused by long-term
traffic loads; (iii) the evaluation of the level of importance of cable vibrations, that can occur due
to vortex-shedding phenomena, parametric or rain-wind excitation; (iv) the experimental
identification of local and global natural frequencies, contributing to validate and update finite
element numerical models used to simulate the dynamic behaviour of the bridge under wind or
seismic loads.
The most common way of performing dynamic measurements in stay cables is based on the use
of accelerometers conveniently attached to the external cable surface, which involves a rather
hard and tedious set-up preparation when dealing with a large number of stay cables, as it is more
and more common in modern cable-stayed bridges. Therefore, in the present study, the
measurement of vibrations in stay cables was developed not only on the basis of conventional
measurement equipment, but also applying an interferometry laser system that can play an
interesting role in this context, avoiding the direct contact with the structure, and providing an
excellent accuracy.
The laser sensor used in this work[6] is an industrially engineered Doppler-based interferometer,
which functions as a non-contacting velocity transducer capable of remote measurement of the
velocity of a solid surface. The basic principle behind the laser Doppler technique used is that
when a beam of coherent light is reflected from a moving surface, its frequency changes
according to the well-known Doppler effect. Although the fractional change of the frequency of
the light wave is very small, it can be measured very accurately using optical interferometry in
conjunction with electronic frequency measurement equipment, the velocity of the moving
surface being directly derived from the frequency changes.
In order to measure vibrations in some of the longest stay cables of Vasco da Gama cable-stayed
bridge using conventional piezoelectric accelerometers, these were screwed on small metallic
cubes, conveniently attached to the external surface of the stay cables with the help of metallic
belts strongly tightened. This relatively boring preparatory operation, only possible as the bridge
was not open to the normal road traffic yet, was systematically repeated in all the stay cables
observed, placing the accelerometers 5m above the deck by means of a crane, and measuring
vibrations in the vertical plane. The use of the laser transducer became however uncomparably
easier, the only operation needed being the control of the position of the laser head, placed on the
deck under each cable, in order to produce a laser beam hitting the cable surface.
Figure 10 shows, for instance, average power spectra associated to the ambient response of one
of the longest stay cables of the bridge, obtained with simultaneous measurements at the same
point on the basis of the two types of sensors mentioned, using 16 averages and a frequency
resolution of 0.0078Hz. Although those spectra are associated to different mechanical quantities
measured (acceleration and velocity), they clearly evidence an excellent agreement in terms of
identification of local natural frequencies of the cable, characterised by equally spaced well
pronounced peaks. Moreover, some global natural frequencies of the bridge, corresponding to
main peaks of the spectra in the range 0-1Hz, are also apparent, though not so clearly in the case
of the laser sensor, as this transducer measures the relative velocity between the deck and the stay
cable. The same conclusion can also be drawn when comparing the natural frequencies identified
using the laser sensor with those obtained with conventional equipment in the free vibration test
(Figure 10), the contribution of the global modes being naturally more significant in this case.
1.00E-04
Acceleration PSD (m^2/s^4)
Velocity PSD (m^2/s^2)
1.00E+00
1.00E-01
1.00E-02
1.00E-03
1.00E-04
1.00E-05
0
0.5
1.5
2.5
3.5
1.00E-05
1.00E-06
1.00E-07
1.00E-08
1.00E-09
0
0.5
1.5
2.5
3.5
Frequency (Hz)
Frequency (Hz)
(a)
(b)
Figure 10: Average power spectra of the ambient response of a stay cable:
(a) using the laser sensor; (b) using the accelerometer
40
1.00E+00
Acceleration (mg)
10
1.5
0
1
-10
-20
0.5
-30
Wind speed (m/s)
20
Acceleration FFT (mg)
2.5
30
1.00E-01
1.00E-02
1.00E-03
1.00E-04
1.00E-05
-40
0
100
200
300
400
500
Time (s)
600
700
800
0
900
0.5
1.5
2.5
Frequency (Hz)
(a)
(b)
Figure 11: Response of the stay cable during the free vibration test of the bridge:
(a) Cable response and wind speed at 1/2 span; (b) FFT of the cable response
3.5
The values of the first 5 natural frequencies of this stay cable, identified on the basis of these
spectra using the two measurement systems referred, are virtually coincident (0.594, 1.180,
1.766, 2.367, 2.953Hz), the only difference observed in one of the natural frequencies being
equal to the frequency resolution (0.0078Hz).
5.2 Evaluation of cable tensions
Several techniques can be employed to evaluate cable forces, namely measurement of the force in
a tensioning jack, application of a ring load-cell, topographic measurements, elongation of the
cables during tension and installation of strain gauges in the strands. As referred by Casas[7], in
spite of their simple theoretical bases, each of these methods is complex in its practical
application and, in some cases, the level of accuracy is insufficient.
A relatively simpler and less expensive method to estimate cable tensions in cable-stayed bridges
is based on the vibrating chord theory, taking into consideration the identified values of natural
frequencies of the stay cables, which leads to the following relation:
T=
4mf n2 L2
n2
(1)
where T is the cable tension, f n is the n-th natural frequency, L is the cable length and m
represents the mass of the cable per unit length. Application of this approach, taking L = 214.97m
and m = 96.9kg / m , leads to an average value of cable tension, for the stay cable referred, of
6266kN using the laser sensor, whereas the accelerometer led to 6256kN.
6. Experimental evaluation of dynamic amplification factors

The dynamic tests described herein still involved the experimental evaluation of dynamic
amplification factors associated to the passage of heavy traffic. For that purpose, a truck with a
mass of 30t has crossed the bridge several times at different speeds (15, 30, 45 and 60 km/h),
both using a central and a lateral lane, the structural response being measured at sections 10, 13
and 16, upstream and downstream, in terms of vertical acceleration. Comparing the peak values
of the response obtained at different speeds with the corresponding values associated to the lower
velocity (15km/h), it was possible to estimate DAFs, which show a clear tendency to increase
with the vehicle speed, as it is reported with more detail in [3]. Further investigation on this topic
is being proceeded by the authors.
7. Conclusions
The development of the dynamic tests described in this paper permitted to extract the following
main conclusions:
The measurement system used in the ambient and free vibration tests, based on the use of
independent triaxial accelerographs conveniently programmed and synchronised by a
portable PC, revealed to be a very efficient and comfortable solution, avoiding the use of
several hundred meters of electric cables and permitting the integral data acquisition in a
relatively short period of time;
The ambient vibration test provided a very accurate estimate of natural frequencies and mode
shapes, despite the rather low level of signal captured, the low range of natural frequencies of
interest (0-1Hz) and the relatively high number of different modes of vibration in that range;
The free vibration test, based on the sudden release of a mass of 60t suspended from the
deck, seemed to be quite useful as a complementary test that permitted not only to check the
previous identification of natural frequencies and mode shapes, but essentially the very
accurate identification of modal damping factors, whose knowledge is particularly relevant in
terms of the study of the aerodynamic stability of the bridge;
There is, in general, an excellent correlation between modal parameters identified and the
corresponding parameters calculated on the basis of the 3D finite element model developed at
the design stage, though some small differences can be found, as it is the case of the multiple
modes identified associated to the 2nd torsion + transversal bending numerical mode, related
with local stay cable frequencies, or of the 3rd torsion mode, in which no transversal bending
component was experimentally detected;
Although no force measurement has been performed during the free vibration test, a standard
MDOF identification algorithm (RFP Rational Fraction Polynomial method) in the
frequency domain could be applied with success, assuming the input as an impulsive load,
with unit magnitude, acting during a very short period of time, leading to a spectral content of
the excitation with an almost constant intensity over the frequency range of interest;
The application of a laser Doppler velocity transducer reveals to be a rather accurate and easy
to use non-contact vibration measurement technique, particularly appropriate to perform
dynamic measurements in stay cables of cable-stayed bridges, providing a powerful form of
systhematic and accurate evaluation of natural frequencies and cable tensions, which is a
factor of significant importance, in particular, in terms of the long term health monitoring of
this type of bridges;
The experimental evaluation of dynamic amplification factors (DAFs) is an interesting field
of research that deserves further investigation with the aim of testing the appropriateness of
general design criteria defined in national or international codes or regulations concerning the
evaluation of dynamic effects due to traffic loads on bridges.
8. Acknowledgements
The authors wish to acknowledge the Portuguese Foundation for Science and Technology (FCT)
and NOVAPONTE for all the support provided to the development of the present investigation,
carried out in the context of the Research Project no. PBIC/CEG/2349/95, on the theme
Experimental and Numerical Analysis of the Dynamic Behaviour of Cable-Stayed Bridges.
9. References
[1] Grillaud G., Bourcier P., Barr C. and Flamand O., Wind action on the Vasco da Gama
cable stayed bridge, Proc. of the 2nd European and African Conference on Wind
Engineering, Genova, Italy, pp.1449-1456, 1997.
[2] Branco F., Mendes P. and Guerreiro L., Research studies for the Vasco da Gama Project,
IST Science & Technology, No.2, pp.3-7, April 1998.
[3] Delgado R., Cunha A., Caetano E. and Calada R., Dynamic Tests of Vasco da Gama
Bridge (in Portuguese), Report under contract with NOVAPONTE, Faculty of Engineering
of the University of Porto, 1998.
[4] Felber A.J., Development of a Hybrid Bridge Evaluation System, Ph.D. Thesis, University
of British Columbia, Canada, 1993.
[5] Han M-C and Wicks A.L., On the application of Forsythe polynomials for global modal
estimation, Proc. 7th Int. Modal Analysis Conference, pp.625-630, 1989.
[6] Cunha A., Caetano E., Laje A. and Gomes A. A laser system for the identification of
dynamic parameters of civil engineering structures, Proc. Int. Conf. Earthquake Resistant
Construction and Design, pp.985-992, Berlin, 1994.
[7] Casas J.R., A combined method for measuring cable forces: the cable-stayed Alamillo
Bridge, Spain, Structural Engineering International, Journal of the IABSE, Vol.4, No.4,
pp.235-240, November 1994.
The resund stay cables : design for fatigue resistance

and easy maintenance
J.P. FUZIER
Scientific Director
Vlizy, France
J.STUBLER
Technical Director
Vlizy, France
D. GRATTEPANCHE
resund Site Engineer
Vlizy, France
Summary
The 16 km resund fixed link across the shallow channel between Denmark and Sweden
includes a 7-8 km long bridge. The double-deck cable-stayed bridge with its 490 m main span
appears to be the masterpiece of this viaduct.
It is hence of primary importance to use a stay cable technology which provides high fatigue
resistance and durable corrosion protection and allows easy maintenance without any
interruption of the service. This paper deals with the installation specific procedure and the
associated technology of these long cables. Such a technology offers the advantage of avoiding
any limit to the sizes of the anchorages due to weight or to the dimensions of the jack. It allows
the placing of the cables, even the longer ones, with very light equipments and provides the
engineer and the owner a fully documented record of all the tensioning operations thanks to a
microprocessor piloted robot.
All necessary adjustments can then be carried out with great accuracy.
1. Introduction
The resund bridge, like the other recent and major crossings (Normandie, Tagus, Second
Severn, Ting Kau, Tsing Ma), requests a long life span. Very often, 100 or 120 years are
considered for this type of structure ; the only way to meet such a demand is to provide a better
quality product from the very beginning of the construction and to develop a surveillance
concept since there are no material suppliers able to give a 120 years guarantee. This explains
why the stay cable technology has been renovated during the last ten years. The cables supplied
and installed on the resund bridge meet such requirements :
- high fatigue resistance
- high stiffness and mechanical strength
- excellent corrosion protection
- simplicity of installation
- easy maintenance and replacement without any traffic disruption.
2. Technology description
2.1 General
The stay cable system consists of parallel individually protected seven wire strands with wedge
anchorages and additional corrosion protection system consisting of an outer HDPE pipe. The
stay cable design is such that the replacement of any cable can be done, if required, strand by
strand, in order to reduce to a minimum any traffic disruption. The anchorages are filled with
wax. Cement grout filling is rejected because of the unability of cement grout to resist without
cracking to stress variations produced by live loads. This cracking may induce fretting
corrosion of the wires.
The strands are individually protected as follows :
- hot dip galvanization before wire drawing ;
- extrusion around the strand of a high density polyethylene sheath (i.e. 1.5 mm thick
minimum) after coating the wires with wax.
Fig 1.The Freyssinet monostrand

2.2 Independence of strands
The high fatigue performances are obtained thanks to the absence of any steel to steel contact.
All the strands are parallel in the anchorages. They are distributed according to a triangular
network and they form a compact bundle in the typical section outside of the anchorage zone,
thanks to deviators and clamping rings placed at a specified distance from each other.
The transition zone between the bundle of strands and the individual anchorage of each strand
consists of a guide-deviator which groups together the strands which are spaced apart in the
anchorage (deviator function) and prevents any transverse movement at deck level or at pylon
level in order to avoid bending in the anchorage because of the angular variations of the stay
under service loads and wind effect (guide function).
The anchorage itself is made so that each strand is individually anchored by means of jaws
blocked in a conical hole. There is no possibility of steel to steel contact inside the anchorage
which avoids all risks of fretting corrosion.
In the anchorage there is also a protection ensuring watertightness between the area where the
strands are removed.
Beyond the jaws, there is an over length of strands which enables tensioning or detensioning at
any time.
Fig 2. The Freyssinet cable anchorage

2.3 Vibration limiting devices
Helical ribs
The streamlined sheaths covering the stays expose to the wind a cylindrical surface covered
with two criss-crossed helical ribs. This arrangement, developed during wind tunnel tests
carried out at the CSTB in NANTES for the Normandie bridge, prevents the vibration
phenomena produced by vortex-shedding on rainy days, due to, under certain combinations of
force and direction of the wind, the apparition of a water rivalet. This is the rain and wind
phenomenon. Similar tests were performed at the Danish Maritime Institute for the resund
bridge.
Dampers
Provisions have been taken to install, at a later stage if required, visco-elastic internal dampers.
This installation can be carried out easily during regular maintenance operations.
3. Cable installation
3.1 General
The 160 stays are distributed according to a regular harp layout with a constant 30 inclination.
Each plane of stays consists of 10 x 2 twin-stays made of 73 strands each. The spacing between
the twin-stays is equal to 670 mm.
This cable concept based on independence of the strands leads to a simple and easy installation
method which is installation at the site, strand by strand.
3.2 Installation
Because of a tight schedule requesting that 2 x 73 HD15 stays are installed on a 6 days cycle,
the following installation method has been selected :
-
supply and installation, on each side of the deck, of a 40 m long self launching access
platform. This platform provides access to two successive stay cables allowing to
carry out, at the same time, erection of cable n+1 and finition of cable n ;
lifting of HDPE sheath with the referenced strand ;
threading of strands two by two in order to increase the productivity and to reduce
the risk of delay in case of bad weather conditions ;
the strand uncoilers are equipped with an hydraulic braking system permitting to
adjust the tension of the threaded strand ;
stressing strand by strand thanks to the patented Isotension system. A computer
software allows the installator to provide a complete history of the cables.
Fig 3. Self launching access platform
3.3 Corrosion protection

The strands in the free zone are galvanized, waxed with Injectelf and individually protected
with an HDPE sheath.
The top anchorage is injected with petroleum Injectelf wax. The bottom anchorages use the
same dehumidification system already set up for the internal corrosion protection of the steel
truss.
Fig 4. Self launching access plaform
IABSE REPORTS
Available in June 1999
Volume 1
Symposium - New York 1968
Volume 2
Symposium London 1969
RAPPORTS AIPC
Disponibles en juin 1999
IVBH BERICHTE
Erhltlich im Juni 1999
Structural Precast Concrete

Constructions prfabriques en bton
Vorfabrizierte Betonbauten
On Wearing Surfaces for Steel Bridge Decks of
Lightweight Construction
Sur les revtements pour les platelages
mtalliques lgers
ber Fahrbahnbelge von sthlernen LeichtFahrbahnen
Final report Rapport final Schlussbericht
On Concepts of Safety of Structures and
Methods of Design
Sur les notions de scurit des structures et
mthodes dlaboration des projets
ber neue Aspekte der Tragwerksicherheit und
ihre Bercksichtigung in der Bemessung
Volume 3
Preliminary publ. Publ. Prliminaire Vorbericht
Volume 4
Symposium Madrid 1970
Design of Concrete Structures for Creep,

Shrinkage and Temperature Changes
Linfluence de fluage et du retrait, leffet des
Changements de temprature sur les constructions
en bton
Der Einfluss des Kriechens, Schwindens und
Der Temperaturnderungen in Stahlbetonkonstruktionen
Volume 5
Volume 6
Volume 7
Report on an Enquiry Regarding Corrosion of

the Reinforcement in Reinforced and
Prestressed Concrete Structures
Rapport denqute sur la corrosion des
armatures des ouvrages en bton arm
et en bton prcontraint
Bericht ber eine Umfrage betreffend Korrosion
Der Bewehrung in Stahl- und Spannbetonbauten
Symposium Praha 1971
On Mass-produced Steel Structures

Fabrication en srie des constructions mtalliques
In Serien gefertigte Stahlbauten
Volume 8
Introd. Report Rapport introd. Einfhrungsbericht
Volume 9
Volume 10
Seminar London 1971
Volume 11
Symposium Lisboa 1973
Design of Plate and Box Girders for Ultimate

Strength
Etude la rsistance post-critique des poutres
me pleine et des poutres-caissons
Bemessung von Blech- und Kastentrgern im
berkritischen Tragbereich
Proceedings Rapport Bericht
Resistance and Ultimate Deformability of
Structures Acted on by Well Defined Repeated
Loads
La rsistance des structures et leur dformabilit ultime sous laction de charges dfinies,
rptes our priodiquement variables
Tragfhigkeit und Grenzverformung von
Bauwerken unter Einwirkung gegebener
Wiederholter Belastungen
Volume 12
Volume 13
Volume 14
Symposium Qubec 1974
Design and Safety of Reinforced Concrete

Compression Members
Dimensionnement et scurit des pices
comprims en bton arm
Bemessung und Sicherheit von StahlbetonDruckgliedem
Volume 15
Volume 16
Volume 17
Volume 18
Three Contributions on Safety of Structures

Trois contributions sur la scurit des structures
Drei Beitrga zur Sicherheit von Tragwerken
Symposium Dresden 1975
Volume 20
Symposium Mnchen 1977
Volume 26
Steel and Composite Structures for User Needs

Structures en acier et mixtes concues pour lusage
Nutzungsgerechtes Bauen im Stahl- und StahlVerbundbau
Problems Associated with Design and Construction
in Developing Countries
Conception, projet et excution de constructions dans
les rgions en dveloppement
Planen und Bauen in Entwicklungsgebieten
Symposium Moscow 1978
Volume 27
Workshop Bergamo 1982
Volume 40
Colloquium Copenhagen
1983
Main Trends in the Development of Steel

Structures and Modern Methods of their
Fabrication
Tendances principales dans le dveloppement
des structures en acier et mthodes modernes
de fabrication
Wichtigste Entwicklungstendenzen im Stahlbau
und moderne Herstellungsverfahren von Stahlkonstruktionen
Informatics in Structural Engineering
Proceedings
Ship Collision
Problems with Bridges and Offshore Structures
Volume 41
Volume 42
Volume 43
Workshop Tokyo 1982

Volume 44
Workshop Rigi 1983
Volume 47
Symposium Luxembourg
1984
Volume 48
Health and Safety in Construction

Proceedings
Quality Assurance within the Building Process
Proceedings
Steel in Buildings
Lacier dans le btiment
Stahl im Hochbau
Symposium Report Rapport du Symposium Symposiumbericht
Colloquium Stockholm 1986 Thin-Walled Metal Structures in Buildings

Volume 49
Symposium Tokyo 1986
Proceedings
Safety and Quality Assurance of Civil Engineering
Structures
Scurit et assurance de la qualit des
Structures de gnie civil
Sicherheit und Qualittssicherung von
Ingenieurbauwerken
Volume 50
Volume 52
Workshop Zrich 1986
Volume 53
Organization of the Design Process

Organisaiton du processus de projet
Organisation des Planungsprozesses
Proceedings
Colloquium Delft 1987
Volume 54
Symposium Paris 1987
Volume 55
Colloquium Bergamo 1987
Volume 56
Symposium Lisbon 1989
Volumes 57/1 & 57/2

Colloquium Bergamo 1989
Volume 58
Workshop Lausanne 1990
Volume 59
Symposium Brussels 1990
Volume 60
Short Course Brussels 1990
Volume 61
Colloquium Stuttgart 1991
Volume 62
Colloquium Nyborg, DK
1991
Volume 63
Computational Mechanics of Concrete

Structures Advances and Applications
Mecanique des structures en bton arm
Progrs et applications dans les modles
analytiques
Rechnerische Mechanik von Konstruktionen aus
Stahlbeton Fortschritte and Anwendungen
Report Rapport Bericht
Concrete Structures for the Future
Structures en bton de lavenir
Betonbauwerke fr die Zukunft
Monitoring of Large Structures and Assessment
of their Safety
Durability of Structures
Durabilit des constructions
Dauerhaftigkeit von Bauwerken
Expert Systems in Civil Engineering
Remaining Fatigue Life of Steel Structures
Mixed Structures, including New Materials
Construction mixte acier, bton et nouveaux
Matriaux
Misch- und Verbundbauweisen, einschliesslich
neuer Baustoffe
Composite Steel-Concrete Construction and
Eurocode 4
Notes
Structural Concrete
Report
The Interaction between Major Engineering
Structures and the Marine Environment
Report
Symposium Leningrad 1991 Bridges: Interaction between Construction,

Technology and Design
Volume 64
Conference Davos 1992
Volume 65
Workshop Madrid 1992
Volume 66
Colloquium Copenhagen
1993
volume 67
Colloquium Beijing 1993
Volume 68
Colloquium Gteborg 1993
Volume 69
IABSE Symposium Rome
1993
Volume 70
IABSE Symposium Birmingham 1994
Volume 71
Report
Structural Eurocodes
Report
Length Effect on Fatigue of Wires and Strands
Report
Remaining Structural Capacity
Report
Knowledge-Based Systems in Civil Engineering
Report
Structural Serviceability of Buildings
Report
Structure Preservation of the
Architectural Heritage
Report
Places of Assembly and Long-Span
Building Structures
Report
IABSE Colloquium Bergamo Knowledge Support Systems in

1995
Civil Engineering
Volume 72
IABSE Symposium San
Francisco 1995
Volumes 73/1 & 73/2
Report
Extending the Lifespan of Structures
Report
IABSE Workshop Lausanne Evaluation of Existing Steel and

1996
Composite Bridges
Volume 76
IABSE Colloquium Berlin
1998
Volume 77
IABSE Colloquium Stockholm 1998
Volume 78
Report
Saving Buildings in Central and
Eastern Europ
Report
Tunnel Structures
Proceedings
IABSE Symposium
Kobe 1998
Volume 79
IABSE Colloquium New
Delhi 1999
Volume 80
IABSE Colloquium Phuket
1999
Volume 81
IABSE Conference Malm

1999
Volume 82
Long-Span and High-Rise Structures
Report
Foundations for Major Bridges:
Design and Construction
Report
Concrete Model Code for Asia
Structural Concrete: Design, Materials and
Construction, and Maintenance
Report
Cable-Stayed Bridges past, present and future
Proceedings (Abstract report and CD-rom)

Cable Stayed Bridge

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cable Stayed Bridge

Uploaded by

Copyright:

Available Formats

International Association

for Bridge and Structural

The resund Construction Site, October 1998

Malm, Sweden 2-4 June, 1999

Jointly organized by the Danish and Swedish Groups of IABSE

Produced by Congrex Sweden AB

International Association for Bridge and Structural Engineering

Address by the President of IABSE

SESSION 1 - Design and Construction

SESSION 2 Composite Structures

Co-chairman: Henrik Christensen, Sweden

Co-chairman: Karl-Otto Sicking, Sweden

SESSION 3 Cable-Stayed Bridges for Railways

Co-chairman: Ole Damgaard-Larsen, Denmark

SESSION 4 Stay Cable Technology

Co-chairman: Ole Damgaard Larsen, Denmark

SESSION 5 Observation, Maintenance and Repair, followed by Closing

Co-chairman: Hans Ingvarsson, Sweden

History of cable-stayed bridges

Niels J Gimsing, born 1935, is professor

Fig.1 Bridge systems investigated by

In the early stayed bridges it proved very difficult to

Fig.2 The Albert Bridge across the Thames in London.

Fig.4 The Strmsund Bridge.

The start of a new era for cablestayed bridges was to a large

Fig.5 The Theodor Heuss Bridge.

Fig.6 The Severins Bridge in Kln.

Fig.9 The Knie Bridge in Dsseldorf.

In 1969 a notable cable-stayed

First parallel-wire strands

From the same period is another

Fig.13 The Indiano Bridge across the Arno at Firenze.

Fig.14 The Maracaibo Bridge.

The twin bridges across the Parana

Fig.17 The Parana Bridge.

The Tjrn Bridge belongs to the

During the 1980s the activity

Fig.19 The Barrios de Luna Bridge.

Fig.20 The Far Bridge.

Fig.21 The Papineau Bridge in Montreal.

Apart from the Papineau Bridge

Fig.23 The Alex Fraser Bridge.

Fig.24 The Rokko Bridge in Kobe.

Fig.25 The Hitsuishijima and Iwagurojima Bridges of the Seto Ohashi.

Fig.26 The Meiko Nishi Bridge in Nagoya.

Fig.27 The S-shaped Katsuhika Harp Bridge in Tokyo.

The double deck configuration was

Fig.28 The Yokohama Bay Bridge

In Japan the parallel-wire strands

Qingzhou Minjiang Br.

Meiko Chuo Bridge

Tsurumi Fairway Bridge

Troitsky, M.S., Cable-Stayed Bridges, BSP Professional Books, London 1988

Retrospect and Prospect of Cable-stayed Bridges in China

Haifan Xiang, born 1935,

1.Introduction and Historical Review

2nd Nanjing Bridge over Yangtse

17 Dafoushi Bridge over Yangtze River Chongqing

18 Zhenyang Bridge over Yangtze River

Chongming Bridge over Yangtze

Table 1. Major Cable-stayed Bridges in China(L>400m)

Location Main Span

Table 2. Single Pylon Cable-stayed Bridges in China ( L>200m)

2 Long-span Cable-stayed Bridges for 21st Century in China