Professional Documents
Culture Documents
J.T.F.M. TNNISSEN
Joep Tnnissen Dynamic Engineering Consultancy
Veghel - The Netherlands
ABSTRACT
The structures and track system which are part of the Dutch High-Speed Railway connection
between Amsterdam and Paris, have all been analyzed for their dynamic behavior with regard
to structural integrity and passenger comfort. This paper describes the general approach of
these dynamic analyses as performed in the High-Speed Line-Zuid project, and goes into more
detail with regard to bridge-track interactions at the bridge across the Hollandsch Diep, one of
the most eye-catching and largest structures in this project. Relevant issues are the optimization
of the level of passenger comfort by introduction of a pre-camber in the alignment of the track
system and the dynamic behavior of the steel transition slabs, which as a special structure in
the track system allow for the horizontal expansion and contraction of the bridge.
1. INTRODUCTION
October 2007, the Dutch part of the High-Speed Railway connection between Amsterdam and
Paris nears its completion and thereby its goal to reduce the travelling time between both cities
to just 3 hours. In the 125 kilometers from Amsterdam, via Rotterdam to the Belgium border,
of which 85 kilometers have been destined for high-speed railway, trains will pass 170
different structures, designed to cope with train velocities up to 300 km/h.
In the 2nd half of the 90s the HSL-Zuid Project-organization, representing the Dutch State
Department of Transportation, developed the visualization and engineering tools for the
architectural and structural elements in the High-Speed Line-Zuid project. Due to its magnitude
and complexity it was decided to divide the contract into 7 subcontracts. 6 of these
subcontracts dealt with the design and construction of the railway structures, each covering a
separate part of the track. They were granted to different joint ventures between contractors in
the year 2000. The 7th subcontract covered the design and construction of the track system as
well as the future maintenance of the entire railway system and was granted to Infraspeed in
2002. The author has been a member of the design team in 4 out of these 7 subcontracts, as an
advisor on dynamic aspects and as the responsible party for the dynamic analyses performed on
the primary (main structures and track system) and secondary (architectural elements and noise
barriers) structures.
The Dutch terrain has required many adaptations to the high-speed railway structures and track
system as the track encounters highways, rivers, canals, ditches and environmentally important
areas on its way. Together with the necessity for piled foundations, due to the soft clay type of
soil in the western part of the Netherlands, the track has become a chain of different structures
with dilatations generally every 20 to 30 m. Dealing with the design criteria concerning the
dynamic behavior of these structures and the implementation of a ballastless track system, has
made the HSL-Zuid project a challenging engineering experience.
2. HIGH-SPEED RAILWAY STRUCTURES IN HSL-ZUID
The natural and man-made obstacles in the path of the track, the need for pile foundations and
the ever developing engineering and architectural insight during the design phase, have led to a
variety of structures. Some of these structures are highly visible, such as the elevated longviaducts near Hoofddorp and Bleiswijk (see figure 1), and the bridge crossing the Hollandsch
Diep, which with a total span of 1192 meters is the longest HSR-bridge between Amsterdam
and Paris (see figure 2). Others are hardly noticeable, such as the settlement-free slabs, which
cover about 33 km of the track and with their typical length of 30 m per slab often provide the
link with other types of structures. Except for the train passengers, some major achievements
are not visible at all, such as the Ringvaart aqueduct and the tunnels underneath the Green
Heart, with a diameter of almost 15 meters, one of the largest drilled tunnel in the world, and
the rivers Oude Maas and Dordtsche Kil.
Due to the soft clay type of soil in the western part of the Netherlands most of these structures
are supported by pile foundations, reaching into the Pleistocene sand bed. Only in the last 3.5
kilometers towards the Belgium border the sand bed reaches the surface and provides a solid
foundation. At this location the transition into a ballast track is established in order to connect
to the Belgium part of the high-speed railway track.
slabs, viaducts and bridges with dilatations approximately every 15 to 30 meters, but also on
the bridge across the Hollandsch Diep which is categorized as being sensitive to
settlement/vertical deflection.
In the 163 km of Rheda 2000 slab track system, the rails are the only continuous elements
connecting adjacent structures. They are held in place by a Vossloh type IOARV 300 railfastening system which is provided with a highly elastic intermediate layer. These layers are
responsible for the transfer of horizontal and vertical forces into the Pfleiderer type B355
W60M concrete bi-block sleepers. The Rheda 2000 slab track system is completed by casting
the prefab concrete sleepers into a reinforced concrete slab on site. The reinforcement consists
of lattice trusses which provide stable dimensions and assures the required gauge of the track.
The concrete slab grade B35, as used in the HSL-Zuid project, has a standard height, without
cant, of 240 mm and a width of 2600 or 2800 mm, depending on the location. The slab is
reinforced throughout its entire length for systematic prevention of cracks. See figure 3.
Fastening IOARV 300
Sleeper B355 (c.t.c. 650 mm typical)
3000
2600
Dowel
232
240
Intermediate layer
Settlement-Free Plate
4. DYNAMIC ANALYSES
4.1 Design criteria
To aid the design teams of the contractors during the engineering phase of the structures,
guidelines have been prepared and provided by the HSL-Zuid Project-organization. These
guidelines are based on national and European standards anno 1999, such as ENV 19913:1995, supplemented with available research data and experiences with high-speed railways
abroad. In guideline HSL600E, titled Loads and Deformations of Structures, the criteria with
regard to the dynamic behavior of structures are stipulated. These can be summarized as
follows:
1. The vertical accelerations in the structure, as calculated in the center of the track, shall
not exceed 0.50g in case of ballastless track, for frequencies ranging from 0 to 20 Hz.
This is a way to ensure that the structural accelerations will not negatively affect the
passing train (rebound) and is introduced as a safety measure against derailment. It
does not mean in any way that acceleration signals outside the mentioned frequency
range do not matter or may be ignored in the structural analyses.
2. To account for the dynamic response of a structure a dynamic coefficient 2 is
introduced based on the determinant length (L) of the structure at hand. The value of
this coefficient ranges from 1.00 to 1.67, as applicable for carefully maintained track.
By means of dynamic analysis a second coefficient r has to be established, equal to
the quotient of the dynamic and static bending moments at any governing location in
the structure. The maximum of these coefficient 2 and r is then used as the general
multiplier applicable to the load models (LM 71, SW/0 & /2) used for static
analyses.
3. In order to ensure a proper level of passenger comfort, the vertical acceleration of the
train is limited to a maximum of 1.0 m/s2 (classification: very good). However, the
natural frequency of the bridge across the Hollandsch Diep lies close to 1.0 Hz, which
is near the range of natural rigid car body frequencies of most trains (0.6 to 2.0 Hz).
Therefore an additional criterion has been introduced in the form of a weighted level
of incomfort of harmonic vibration (LIh) which may not exceed the value of
45 [ms 5/3].
The limiting value of 45 is time dependent with a maximum duration of passage of
15 s.
Derived from ERRI D190 (Permissible deflection of steel and composite bridges for
velocities V > 160 km/h, December 1995) this criterion is based on empirical data, as
shown in the following figure.
100
100% Satisfied
100% Unsatisfied
the in-betweens
LIh limit value (45)
90
80
70
60
50
40
30
20
10
0
0
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
LI h value
Figure 7: Empirical data, source for the LIh comfort criterion as per ERRI D190
Taking into account a weighting factor of 0.40, the formulae for the LIh-value reads:
LI h = 43.1
1
3
av (t ) dt 45
(1)
) (
2
Wd
NVA = 4 aWZPb95 + 2 aYA
+ aWZAb95
95
) (
2
+ 4 aWXDc 95 2
(2)
(3)
Wd
NVD = 3 16 aWXPd 50 + 4 aYP
+ aWZPd50
50
Wd
+ 5 aYP
95 2
In order to predict the field measurements dynamic analyses have been performed before the
actual installation of the Rheda 2000 slab track system, and with the train types specified in
paragraph 4.2. Assumptions have been made for the accelerations in the horizontal, lateral and
transverse direction based on the ratio (1:2:3) (source: Prof. C. Esveld, TU Delft). By
collecting data every 417 m the chance that relevant data on smaller structures are included
tends to be small. During the passage of the 1192 m long bridge across the Hollandsch Diep,
the governing structure with regard to level of passenger comfort, only 3 field measurements
are collected out of the 60 in 1 period. By assuming that structures in front of and following the
bridge will show an equal or better comfort performance, a ratio of the weighted 50% and 95%
train accelerations (aWZ50 / aWZ95) of 1.0 is derived. This leads to a maximum allowable
weighted RMS-value of 0.210 m/s2. Applying a weighting factor of 0.40 this results in a nonweighted RMS-value of 0.525 m/s2. Due to the fact that structures before and after the bridge
will most probably show better comfort performances, this approach is considered to be on the
conservative side.
4.2 Real train types
The dynamic analyses required by guideline HSL600E have been performed with the, at that
moment in time, relevant or real high-speed train types such as the French THALYS and
German ICE3M plus ICMAT. Their characteristics, axle loadings and positions, are specified
in attachment 1 of this guideline and are shown in figure 8. For the THALYS and IC3M trains,
single and double configurations are considered, identified by index 1 and 2. The THALYS2
characteristic is equivalent to train type B in Annex F of the European standard EN 19912:2003. Hence, the dynamic analyses have been performed with 5 real train types.
To obtain the vertical body accelerations from these train types, they have been modeled as
SDOF spring-mass systems with a mass equal to the axle-forces divided by the g-force. The
spring frequencies range from 0.80 to 1.20 Hz, while the damping ratio is set to be 0.10, both
as per guideline. Hence: Mass M = Axle Loading / g-force and Spring Constant K = M (2f)2.
In general, the structures and track system have been analyzed for the passage of the THALYS
and ICE3M trains at velocities ranging from 160 to 360 km/h being 1.2 times the maximum
design velocity or line speed of 300 km/h, and from 160 to 264 km/h for the ICMAT train as
its maximum permitted vehicle speed lies at 220 km/h. An exception for the upper bound
velocity is made for the bridge across the Hollandsch Diep by reducing the factor of 1.2 to 1.1,
resulting in a maximum velocity of 330 km/h.
4 x 170 kN
2 x 163 kN
14 x 170 kN
4 x 170 kN
2 x 163 kN
C
uvs(t), vvs(t), avs(t)
3.00
3.00
15.70
14.00
17.00
20.28
3.00
3.00
THALYS1 (THALYS2 = THALYS1 + 7.04 m + THALYS1) (Vmax = 1.1 or 1.2 x 300 km/h) 3.28
3.00
7 x 18.70
193.15
18.70
8 x 160 kN
2.50
17.38
19.88
24.78
4 x 150 kN
2.50
14.88
2.50
24.78
8 x 140 kN
4 x 170 kN
ICE3M1 (ICE3M2 = ICE3M1 + 6.70 m + ICE3M1) (Vmax = 1.1 or 1.2 x 300 km/h)
193.30
2 x 225 kN
2.80
4 x 160 kN
5 x 24.78
11.00
3.00
2 x 150 kN
2 x 160 kN
2.50
14.88
2.50
48 x 130 kN
9.70
12.50
17.00
2.70
13.70
2.70
23.10
4.00
ICMAT (Vmax = 1.2 x 220 = 264 km/h)
10 x 23.10
290.20
19.10
In 1999, during the tender phase of the structures in the HSL-Zuid project, the need arose for
software able to perform dynamic analyses and quick parameter studies. At that moment in
time the existing mainframe software packages were found to be too time-consuming with
regard to the number of analyses to be performed. Therefore, a new computer program was
developed by the author based on software readily available and used for seismic and shock
analyses, focused on the queries at hand. This program called DRS, which stands for Dynamic
Response of Structures has been validated and was approved for usage in this project by the
HSL-Zuid Project-organization. In the 8 years since 1999 the program has been updated
continuously and has proven to be a versatile tool in getting insight in the dynamic behavior of
structures, track system, architectural elements and noise barriers.
The heart of the program is based on a Householder and QR-algorithm eigenvalue solution and
direct numerical integration using the Newmark Hilber method. Depending on the
structures geometry there is a choice to construct 2-dimensional beam or 3-dimensional grid
models, using 1-dimensional elements. The program contains a graphical interface and is able
to directly transfer data to MS-Excel for further processing. Special modules are included,
dealing with shear deformation, rail level geometry and signal filtering by means of the Fast
10
Fourier Transform method, which is relevant for the criterion with regard to the maximum
allowable, filtered structural accelerations.
5. PASSENGER COMFORT ON THE BRIDGE ACROSS THE HOLLANDSCH DIEP
5.1 Dynamic model
The structure of the bridge consists of relative torsion stiff structural elements such as steel
box-girders at the V-shaped part of the hammer-pieces and steel U-shaped sections topped by a
concrete deck. The average load of the real trains THALYS and ICE3M is 21 to 22 kN/m
which is about 8% of bridges total dead load of 275 kN/m. The centerlines of the tracks are
5.50 m apart and therefore 2.75 m off center with respect to the centerline of the bridge. The
center of the tracks coincides with the support locations of the steel U-shaped sections. These
considerations have led to the conclusion that for this bridge a 2-dimensional model will be
sufficiently accurate.
45.00 m (hammer-piece)
70.00 m
42.50 m (section 8)
12.70 m (section 2)
50.00 m (section 1)
22.50 m
4.80 m (section 3)
20.00 m (section 4)
track system
8.00 m
section 2
track system
section 3
13,05 m (section 5)
section 4
8,69 m (section 6)
Abutment
Pier 01/12
Pier 02/11
transition-piece
section 7
1.17 m
0.20 m (nts)
4.60 m
11
Structural damping is an important factor in the dynamic analyses. For the bridge structure a
damping ratio of 0.010 was applied as per guideline HSL600E. For the rails and rail-fastenings
damping ratios of 0.005 and 0.100 respectively were taken into account.
5.2 Vertical rail level geometry considerations
The dynamic analyses for the bridge across the Hollandsch Diep were performed twice. The
purpose of the first round of analyses (2001) was to establish the structural integrity of the
bridge and the level of passenger comfort with regard to the applicable design criteria. Static
analyses revealed that temperature fluctuations and creep of the concrete deck will cause a
deformation of the vertical alignment of the bridge (deck) and consequently the rail level
geometry. As a result the level of passenger comfort will be compromised. This systematic
deformation [SD] in great lines follows the bridges natural deflection pattern under dead load,
causing disturbing frequencies of 0.79 Hz at a train velocity of 300 km/h to 0.87 Hz at 330
km/h (main span = 105.0 m). Not only do these frequencies come close to the bridges natural
frequency of 1.10 Hz, they also fall within the range of frequencies, 0.80 to 1.20 Hz, to be
considered for the real trains. Therefore, resonant effects are likely to occur. The maximum
differential systematic deformation was established at 5 mm comprising of 3 mm at mid-span
and 2 mm at the center of the hammer-pieces. In the case the deformation at mid-span is
upward directed (positive) the most favorable situation [SDR, with R = Reversed] is derived as
the train is flattening its way across the bridge deck. Unfortunately, the opposite situation
may also occur. As a solution to this problem a corrective deflection [CD] or pre-camber of the
rail level was suggested by the design team of the bridge structure and incorporated by Rheda
2000 vof. See figures 10 and 11. This pre-camber is defined as follows:
2a
(a ) = ini 1 cos
L
i
where ini = + 1.5 mm and Li = 70.0 m for the end-spans and = 105.0 m between piers and
a = the distance along the bridge deck/track
105
V = 330 km/h
V = 300 km/h
V = 270 km/h
LIh limit value (45)
90
LI h value
75
60
45
30
15
0
-10.0
-7.5
-5.0
-2.5
0.0
2.5
5.0
7.5
10.0
12.5
15.0
Figure 10: The effect of systematic deformation of the rails on passenger comfort
12
Abutment
Pier
Pier
Pier
70.0 m
+ 2 mm
- 3 mm
Systematic Deformation
47.5 m
Abutment
45.0 m
Pier
+ 3 mm
60.0 m
Pier
Pier
70.0 m
- 2 mm
Systematic Deformation (Reversed)
Abutment
+ 3 mm
Pier
70.0 m
Pier
Pier
where L = the span of the bridge with a minimum of 42.5 m (at both ends)
In order to comply, the scaling factor for the bridge across the Hollandsch Diep had to be
reduced to 0.55 (see figure 12). For all other HSL-structures, analyzed for their interaction with
the Rheda 2000 track system, scaling factors were found to be within a range of 0.92 to 1.12.
This meant that the preparation of the vertical rail level on the bridge would require more
attention and accuracy than at other structures, which was to be expected as this bridge
structure was and still is the most comfort sensitive structure in the entire HSL-Zuid project.
13
10.0
9.0
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
-1.0
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
200.0
220.0
240.0
-2.0
-3.0
-4.0
-5.0
-6.0
-7.0
Location [m]
Figure 12: Vertical rail level geometry as applicable for the first 245 m on the bridge
5.3 Results dynamic analyses
For the 5 real train types used in the dynamic analyses, type ICE3M2 proved to be responsible
for the governing results in all of the dynamic criteria considered. For all 4 ICE3M and
THALYS trains the maximum results were obtained at a natural train frequency of 0.89 Hz.
Figure 13 : Graphical display of the bridge and train vertical displacements by program DRS
14
The effect of rail roughness on the structural acceleration of the bridge deck remains small. At
peak values an increase of approximately 10% is found compared to a situation with a perfect
alignment of the rail level. See figure 14. With a maximum value of 0.40 m/s2 or 0.04g,
obtained for the ICE3M2 train at 330 km/h, the criterion not to exceed 0.50g is easily fulfilled.
Maximum results for the THALYS2 and ICMAT trains were established at 0.26 and 0.20 m/s2
at corresponding velocities of 240 and 220 km/h. The impact of train types THALYS1 and
ICE3M1 proved to be smaller than for their double-sized companions.
Due to the low natural frequency of the bridge structure (1.10 Hz), filtering the structural
acceleration signal for frequencies between 0 and 20 Hz did not result in any reduction of these
accelerations. With regard to the magnitude of the structural accelerations it could also be
concluded that the dynamic coefficient r, part of the 2nd dynamic criterion (see paragraph
4.1), would not exceed the value of 1.00 and would therefore not become governing. This can
be verified considering the following. The distributed load of the real train ICE3M is 21.9
kN/m. During its presence on the bridge the structure shows a response of maximum 0.04
times its dead load which results to 11.0 kN/m (= 0.04 275 kN/m). Train load and structural
response combined amounts to 32.9 kN/m, which is considerably less than the uniform
distribution load of 80.0 kN/m belonging to Load Model 71 which is used in the static
analyses. In general terms, the dynamic coefficient r results to 0.41 ( = 32.9 / 80.0), thus less
than 1.00. This approach however is only valid for structures with large spans and a high ratio
of dead load versus live load. For smaller structures, such as viaducts the dynamic analyses
showed higher r-values, however hardly ever exceeding 1.67, the upper bound value for
coefficient 2.
0.450
0.400
0.350
0.300
0.250
0.200
0.150
0.100
0.050
0.000
-0.050
0.0
100.0
200.0
300.0
400.0
500.0
600.0
700.0
800.0
900.0
1000.0
1100.0
-0.100
-0.150
-0.200
-0.250
-0.300
-0.350
-0.400
-0.450
Location [m]
15
corresponding RMS-value results to 111, i.e. for a RMS-value of 0.525 m/s2 the corresponding
LIh-value = 58.3. See also figure 15.
In the worst case scenario, combining systematic deformation and corrective deflection or precamber [SDCD] a maximum LIh-value of 52 is derived. According to figure 7, the number of
100% satisfied passengers would reduce from 84% to 78%, while the number of 100%
unsatisfied passengers almost remains steady at 10%. Due to the fact that the conditions for
which the maximum calculated systematic deformations occur are rare, the level of passenger
comfort is only compromised for short periods of time and then only in certain areas of the
train as can be seen in figure 16. Most of these areas coincide with the axle positions of the
locomotive sections. As a consequence of these arguments, the increase of the LIh-value from
45 to 52 has been acknowledged by the HSL-Zuid Project-organization.
When both approaches for the level of passenger comfort are being compared, it could be
concluded that the criteria for the structure are more strict than those applicable to the track
system. However, the effects of vertical rail imperfections or rail roughness (short and long
wavelengths) have not been taken into account during the first dynamic analyses concerning
the structure as they were nullified by the definition of the roughness factor. The RMS-range
from 0.470 m/s2, corresponding with a LIh-value of 52, to 0.525 m/s2 is used to allow for a,
however reduced (scaling factor = 0.55), realistic vertical rail level geometry in the Rheda
2000 track system analyses.
As a result of the first dynamic analyses performed on the bridge across the Hollandsch Diep
an alternative for the LIh-criterion was introduced in the way of a maximum allowable vertical
train acceleration of 0.70 m/s2 valid for structures with 3 or more repetitive spans in a row. The
value of 0.70 m/s2 coincides with a LIh-value of 45 ms-5/3 and takes out the dependency on the
duration of passage in the LIh approach.
0.700
[CD] [RMS]
[SDCD] [RMS]
[CDRGF055] [RMS]
[SDCDRGF055] [RMS]
[PA] [LIh]
[CD] [LIh]
60.0
0.500
[SDCD] [LIh]
[CDRGF055] [LIh]
55.0
0.450
[SDCDRGF055] [LIh]
50.0
0.550
70.0
65.0
45.0
0.400
40.0
0.350
35.0
0.300
LIh values
0.600
75.0
[PA] [RMS]
0.650
30.0
0.250
25.0
0.200
20.0
0.150
15.0
0.100
10.0
0.050
5.0
0.0
0.000
160.0 170.0 180.0 190.0 200.0 210.0 220.0 230.0 240.0 250.0 260.0 270.0 280.0 290.0 300.0 310.0 320.0 330.0
Legend:
[PA]
[SD]
[CD]
[RGF055]
:
:
:
:
Figure 15: Maximum RMS train accelerations and LIh-values for governing train type ICE3M2
16
0.700
75.0
70.0
0.600
ICE3M2 [CD]
ICE3M2 [SDCD]
ICE3M2 [SDRCD]
0.550
THALYS2 [CD]
THALYS2 [SDCD]
THALYS2 [SDRCD]
65.0
60.0
0.500
55.0
0.450
50.0
0.400
45.0
40.0
0.350
35.0
0.300
LI h value
0.650
30.0
0.250
25.0
0.200
20.0
0.150
15.0
0.100
10.0
0.050
5.0
0.000
0.0
50.0
100.0
150.0
200.0
250.0
300.0
350.0
0.0
400.0
Figure 16: Maximum train acceleration for each axle-position types THALYS2 and ICE3M2
Another important issue is the impact of the vertical rail level geometry on the axial forces in
the rail-fastenings and the bending moments in the rails. In general, the forces in the railfastenings are of a higher importance as the fastenings are the weaker link in the system.
During a quasi-static passage of train ICE3M the benchmark values are established at 3.0 kN
for the tensile (upward) forces and -36.3 kN for the compression (downward) forces in the
fastenings. See figure 17. The THALYS and ICMAT trains show similar results, all in
proportion to the maximum axle loading of 170 kN for ICE3M and THALYS and 225 kN for
ICMAT (see figure 8), and in combination with a center-to-center sleeper distance of 600 mm
to 650 mm. Hence, a pair of rail-fastenings or a sleeper location is subjected to approximately
45% of the axle loading. This is a recurring result for all HSL-structures analyzed and shows
that the distribution of a single axle loading covers at least 3 but most probably 5 sleeper
locations.
The systematic deformation [SD] of the vertical rail level has a minor effect which is only
noticeable for the compression forces in the rail-fastenings (see figure 17). More relevant is the
rail roughness effect [RGF055] which results in an amplification of both tensile and
compression forces. However, the component of the long wavelengths in this signal, has a
minor to negligible impact on the results, which seem to be solely dictated by the short
wavelengths. A plausible reason for this fact is that the frequency range of these short waves,
especially at high train velocities, lies much closer to the natural frequencies of the track
system.
In the design of the rail-fastening system, limitations to the allowable tensile forces have been
formulated and a minimum dynamic coefficient 2 of 1.67 has been incorporated. In the worst
case scenario, combining rail level situations [SD] and [RFG055], the dynamic coefficient r
17
for the rail fastenings in compression amounts to 1.62. However, due to the application of a
pre-camber in the rail level this value is reduced to 1.50. In those areas where the tensile forces
in the fastening systems may exceed the allowable values, special fastenings with a higher
capacity for tensile forces have been applied. This is especially valid at the dilatations between
the smaller structures in the HSL-track, where differential vertical displacements and rotations
are likely to occur. At the bridge across the Hollandsch Diep, these effects have been reduced
by the implementation of steel transition slabs to keep the dynamic amplification of the forces
and bending moments in rail-fastenings and rails within the required limits.
40.0
30.0
20.0
10.0
0.0
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
200.0
220.0
240.0
-10.0
-20.0
-30.0
-40.0
-50.0
Benchmark [Tension]
[SD] [Tension]
[RGF - Factor 0.55] [Tension]
Benchmark [Compression]
[SD] [Compression]
[RGF - Factor 0.55] [Compression]
-60.0
-70.0
-80.0
-90.0
-100.0
-110.0
-120.0
Location [m]
Figure 17: Rail fasting forces in first 240 m of BHD during the passage of train ICE3M(2)
6. STEEL TRANSITION SLAB
6.1 Transition slab design
With a total, non-dilated span of 1192 m the hybrid bridge-structure across the Hollandsch
Diep needs to be able to react freely to temperature fluctuations. Therefore areas for expansion
and contraction are required at both ends of the bridge at its connection with the approach
ramps. Furthermore, the differences in rotations and vertical displacements between the
connecting ends of the bridge and the approach ramps, need to be restricted in order to limit the
bending moments and forces and the rails and rail-fastenings.
To comply with both requirements and ensure the continuity of the track system, 4 so-called
transition slabs have been designed and constructed to cover the expansion/contraction areas.
18
Consisting of a steel plate structure with dimensions 2700 x 2380 x 200 mm, they each
support 2 rails and 4 pairs of rail-fastenings. The expansion and contraction of the bridge has
been determined to be plus and minus 300 mm at both ends. The average distance between the
first pair of rail-fastenings on the transition slab and the first pair of rail-fastenings on either the
bridge or the approach ramps is 500 mm. By allowing the distance between these railfastenings to vary from 350 mm to 650 mm, together with the application of special fastenings
which allow the rails to slide through, the required expansion and contraction can be assured.
Each transition slab is resting on 4 vertically adjustable supports, which are fixed in pairs to the
approach ramps and bridge ends, and are located directly above the supports/rubber bearings of
these structures. PE-elements are used as an intermediate layer between the supports and the
steel transition slab, to provide sliding, damping and spring characteristics. The final shape and
dimensions of the transition slabs have been the result of the dynamic analyses performed on
these structures. See figures 5 and 18.
Special provisions are taken with regard to the horizontal fixation of the slab, perpendicular to
the track and the proper guidance of the slab in longitudinal direction. A steel U-shaped section
is bolted down to the slab in the center of track, in line with the concrete derailment provision
on the bridge and approach ramps.
200
2380
232
2550
100
500
PLAN VIEW
650
650
650
500
2700
40
500
Approach Ramp
Bridge HD
Expansion/Contraction Area
SIDE VIEW
With program DRS a 3-dimensional grid model has been constructed to incorporate the spring
and damping characteristics of the transition slab supports, the steel plate structure of the
transition slab, the spring and damping characteristics of 4 pairs of rail-fastenings on the slab
plus 10 pairs on both sides of the slab, and 2 rails type UIC60. See figures 19 and 20.
The steel plate structure and rails provide only little damping to the system. A ratio of 0.005 is
taken into account. A considerably higher damping ratio of 0.100 is provided by the railfastening as they incorporate a highly elastic intermediate layer. Another source of damping
19
lies within the PE-elements which are part of the special supports of the transition slab. Tests,
as performed by the Technical University of Munich on the stiffness and damping
characteristics of these supports, show that damping ratios ranging from 0.050 to 0.140 may be
expected. In the dynamic model and analyses the lower bound value of 0.050 has been
assumed.
train body mass
train spring and damping characteristics
rail(s)
1500 mm
rail-fastening
spring and damping characteristics
transition slab
mass, flexural and damping characteristics
transition slab supports
stiffness and damping characteristics
CLtrack
Figure 19: Cross section DRS model of transition slab
Figure 20: 1st and 2nd mode shape of transition slab at 37 and 59 Hz respectively
6.3 Results dynamic analyses
The dynamic analyses performed on the transition slab have led to a change of the original
design, which showed a configuration of H and U-shaped beams. The response of this type of
slab was such that the filtered, structural accelerations would not comply with the design
criteria, while in upward motion the dynamic forces at the supports would overcome the static
forces due to dead load, resulting in loss of contact. This is an undesirable effect with regard to
the supports capacity to withstand fatigue. By introducing a design with a higher stiffness-mass
ratio the natural frequency of the combined system (rails, fastenings, transition slab and
20
supports) has been increased. In general, the filtered vertical structural accelerations will
decrease with an increase of the systems natural frequency. At the same time, the increase of
mass works out positively as a measurement to reduce uplift.
From test results the vertical (secant) stiffness of the supports was established which ranged
between 380 and 1125 MN/m, depending on the frequency of the applied load and the number
of load cycles. The natural frequency of the system largely depends on the support stiffness and
ranges from 37 to 59 Hz (1st mode shape, see Figure 20) within the established range of
support stiffness. These results have been obtained for the situation where the supports are
located in the outer position (see also Figure 18), hence when the bridge-structures is fully
contracted. This is considered to be the governing situation, creating the largest span and
consequently the lowest natural frequency possible.
Due to the increase of natural frequency, the filtered vertical accelerations have been reduced
to a maximum value of 0.24g. The unfiltered results however, show values exceeding 1.00g.
Although these values occur at the center of the track and show a decrease towards the
supports, the possibility of a temporary uplift, has been thoroughly investigated.
40.0
Due to DW of transition slab (30 kN/support)
37.5
35.0
30.0
27.5
22.5
20.0
17.5
15.0
12.5
10.0
7.5
5.0
2.5
0.0
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
Figure 21: Maximum uplift support forces versus train velocity, during passage of ICE3M(2)
train
In figure 21, the impact of the support stiffness (selected values: 250, 500 and 1000 MN/m)
and damping ratio (selected values: 0.005, 0.020 and 0.050) is shown for the uplift support
forces during the passage of the governing train type ICE3M(2). With the increase of the
support stiffness, the position of the peak values for the uplift forces shift to higher train
velocities, without showing a severe reduction nor increase in value. As to be expected, the
support damping provides a considerable reduction of these peak values. Within the stiffness
range of 250 to 1000 MN/m, a damping ratio of 0.050 (5% of the critical damping) reduces the
21
uplifting support forces to 50-60% of the support forces due to the dead load of the system,
which is 30 kN per support. This results in a minimum safety factor against uplift of 1.67.
Having adapted the design of the transition slab with the restriction of uplifting forces at the
supports in mind has also been beneficial to the requirements with regard to the dynamic
coefficients, the level of passenger comfort and the forces in the rail-fastenings as they do not
reveal anything out of the ordinary. The dynamic bending moments in the transition slab
remain below the static bending moments derived during the passage of static load models
LM71 and SW2. This results in a dynamic coefficient r below unity. Due to its short
determinant length [L] coefficient 2 is governing with a value of 1.67. The upward dynamic
response of the steel plate is compensated by its dead weight. In a conservative approach of the
passenger comfort, the governing RMS-value of the vertical train accelerations is established at
0.290 m/s2, hence below the maximum allowable value of 0.525 m/s2. The dynamic
amplification factor of the forces in the rail fastenings does not exceed 1.40.
In figure 22, the time-history results for the support forces are shown during the passage of
train type ICE3M2 at a velocity of 290 km/h, with a support stiffness of 500 MN/m and a
damping ratio of 0.050. The characteristics of the train, axle positions and loading effects (2
times 8 wagons with 4 axles each) are clearly recognizable.
30.0
20.0
10.0
0.0
0.000
-10.0
0.500
1.000
1.500
2.000
2.500
3.000
3.500
4.000
4.500
5.000
5.500
-20.0
-30.0
-40.0
-50.0
-60.0
-70.0
Time [s]
Figure 22: Time-history of support forces during passage of ICE3M(2) at 290 km/h
Out of the ordinary structures, especially small sized structures such as these transition slabs,
combining a lot of different functions on a concentrated area, often require more attention than
usual and may create more questions than answers.
The basic design principle of these transition slabs had already been applied and proven itself
before in HSR tracks abroad. New, however were the operational train velocity, the train
characteristics and the depth of the dynamic analyses. Coming up with a workable solution
proved to be an interesting engineering challenge.
22
The structures in the HSL-Zuid project generally show natural frequencies in the range of 3 to
10 Hz. Usually, the governing dynamic criteria for this type of structures are the (filtered)
structural accelerations, which depend on span length, type and velocity of passing train and
the presence of cantilevered ends. The THALYS train for instance shows governing results for
spans between 16 and 20 m, while the ICE3M train shows governing results for spans just over
20 m. Passenger comfort becomes an issue for structures with considerably larger spans,
certainly in situations where there are 3 or more similar spans in a row.
One of the largest structures in the HSL-Zuid project, the bridge across the Hollandsch Diep
and one of the smallest structures, the transition slab covering the expansion areas of this
bridge, have in common that their dynamic behavior has played a considerable factor in their
final design. Considering the many dynamic analyses performed on other HSL- structures,
these structures have shown features which were out of the ordinary.
The low natural frequency of the bridge across the Hollandsch Diep of 1.10 Hz makes this
structure relative insensitive to the dynamic impact of real train loads, resulting in low
structural accelerations. However, trains with natural body frequencies ranging between 0.8 to
1.6 Hz are easily exited, herewith compromising the required level of passenger comfort.
Several measures were taken to keep the passenger comfort within and acceptable level, of
which one has been the application of a pre-camber in the vertical rail level during the
installation of the Rheda 2000 slab track system. The bridge-structure and the track system are
subject to different comfort criteria. The more severe criterion for the bridge structure allows
23
for a better approach of the rail roughness impact in the design of the track system. However,
for the sake of clarity both criteria should merge into one.
The transition slab is a structural element with a relative high natural frequency. Sliding over
the 4 supports but not vertically fixated the main problem here lies in the possible loss of
contact which is not beneficial to the fatigue strength of the supports. By increasing the
stiffness over mass ratio and establishing the damping properties of the sliding PE-elements in
the supports, the (filtered) structural accelerations of the transition slab and the uplift forces in
the supports are kept within acceptable limits.
Figure 24: Rheda 2000 slab track system on bridge across the Hollandsch Diep
(Photo, courtesy of Mr. P. Meijvis, DMC bv, The Netherlands)