Optimization of Precast
Prestressed Concrete
Bridge Girder Systems
Z. Lounis, Ph.D.
Research Assistant
Department of Civil Eng ineering
University of Waterloo
Waterloo, Ontario, Canada
M. Z. Cohn, P. Eng.
Professor
Department of Civil Engineering
University of Waterloo
Waterloo, Ontario, Canada
60
A practical approach to the optimal design of precast,
prestressed concrete highway bridge girder systems is
presented. The approach aims at standardizing the optimal
design of bridge systems, as opposed to standardizing girder
sections. Structural system optimization is shown to be more
relevant than conventional girder optimization for an arbitrarily
chosen structural system. Bridge system optimization is
defined as the optimization of both longitudinal and transverse
bridge configurations (number of spans, number of girders,
girder type, reinforcements and tendon layout). As a result, the
preliminary design process is much simplified by using some
developed design charts from which selection of the optimum
bridge system, number and type of girders, and amounts of
prestressed and non-prestressed reinforcements are easily
obtained for a given bridge length, width and loading type.
review of American bridge construction practice'·2•3 enables the identifica-
A tion of some broad trends in the selection of bridge superstructure systems.
Although this selection reflects some regional and historical variations, it
appears to be primarily affected by the span length, as shown in Fig. 1. The figure
suggests that reinforced and prestressed concrete solid or voided slab and light
girders are suitable for short spans not exceeding 65 ft (20 m). However, for
medium spans up to 130 ft (40 m), prestressed concrete stringers or box girders
are preferred. For spans longer than 130 ft (40 m), prestressed concrete box gird-
ers or other forms of construction become desirable.
Fig. 1 shows that several alternative solutions exist for any given span range.
For example, for spans of 65 to 100ft (20 to 30m), no less than eight candidate
solutions are available. How does a designer decide on the best system selection
for a given project? Bridge systems may be selected either by relying on existing
tradition and accumulated experience, or by minimizing the costs under the spe-
cific conditions of the project. In the first case, the solutions may not be the most
PCI JOURNAL
 mization at any of three possible lev-
SPAN LENGTH (ft) els: (1 ) bridge members, (2) bridge
(longitudinal and transverse) configu-
33 tiJ 99 132 165
rations, or (3) overall bridge system.
RC SOLID SLAB
- Level 1, member optimization of a
predetermined structure, is the sim-
AASHTO TYPE I
- plest and most widely reported opti-
AASHTO TYPE II

PC VOIDED SLAB
- mization procedure. It involves the op-
timization of girder cross sections of
specified structure types such as sim-
ple span prestressed concrete slabs,5
PCI DOUBLE T I I single span steel girders,6·7 single span
AASHTO lYPE Ill
I prestressed concrete girders, 8·9·'0·'' sin-
gle span box girders, 12 multi-span sim-
RC T GIRDER I ple girders, '3·14 continuous prestressed
concrete box girders, '4· ' 5·' 6 continuous
AASHTO TYPE IV &
I reinforced concrete girder and frame
PCI CHANNEL
PRECAST PCI BOX I I bridges.' 7
Level 2, configuration or layout op-
AASHTO TYPE V I timization, is concerned with finding
the best longitudinal and tran sverse
RC BOX GIRDER
I member arrangement within a given
bridge sys tem, i.e., the number of
AASHTO lYPE VI
spans, the position of intermediate
RC SLAB ON PC/ supports, simply supported or continu-
STEEL GIRDER ous members, etc. Much less work is
CAST-IN ·PLACE reported on thi s type of optimiza-
PC BOX GIRDER
I tion'3·16than on member optimization.
Level 3, system optimization, at-
10 20 30 40 so
tempts to identify the overall features
SPAN LENGTH (m) of the structural system, including ma-
terial(s), structural type and configura-
Fig. 1. Short and medium span bridge system selection. '·2 •3 tion, as well as member sizing. This is
the most complex design problem and
only few a ttempt s to solve it are
economic, while in the second case they may be very time known. Among these are noted the studies by McDermott,
consuming, particularly if the number of bridges to be de- Abrams and Cohn'8•19 on tall building systems, Gellatly and
signed is large. Dupree 20 on surface effect vehicles, Kulka and Lin 2' on
If existing practice is to be taken as a starting point in medium two-span continuous prestressed concrete bridge
bridge selection, the exhaustive survey< on the performance systems, and Mafi and West7 on short simple span bridge
of highway bridges in the United States, published in the systems.
May-June 1992 PCI JOURNAL, can be used with great It is apparent that the overall economic impact increases
benefit. The survey indicates that since their introduction in with higher optimization levels; i.e., optimizing the overall
1950, prestressed concrete bridges have captured almost a bridge system is considerably more effective in reducing
quarter of the total bridge market, and since 1985, about 65 total costs than optimizing its components.
percent of the bridges with spans between 60 and 120 ft Whether bridge systems are selected by following current
(18.3 and 36.6 m). The survey also indicates that, during the practice or through formal optimization techniques, the de-
last 40 years, about 35,000 bridges built (or close to half of velopment of some standards for selecting optimal bridge
all pre stres sed concrete bridges) are of the stringer systems for various ranges of geometry would have consid-
(1-girder) type (42.1 percent simple and 6 percent continu- erable practical significance. These standard optimal sys-
ous stringers). tems could serve both as guidelines for preliminary designs
Thus, the stringer system is the most common prestressed and also as yardsticks against which final bridge designs
concrete bridge type in the United States, a feature it retains could be evaluated.
throughout 25 states for spans of 60 to 80ft (18.3 to 24.4 m). A comprehensive investigation on this topic, in progress
It is also noted that from 1985 to 1989, prestressed concrete at the University of Waterloo, aims at developing optimiza-
stringer and box girder types have each covered about one- tion procedures and solutions for short and medium span
third of the prestressed concrete bridge market. length highway bridge systems. The approach consists of
On the other hand, if "custom-tailored" optimal designs first optimizing configurations of major bridge types (pre-
are desired for individual bridges, designers may seek opti- stressed concrete stringers and box girders, steel stringers,

July-August 1993 61
etc.) and then identifying optimal systems from the various
pre-optimized configurations using sieve-search tech-
niques.' 8· '9·20 Results on optimization of simple span bridge
systems have been reported elsewhere.22
The object of this paper is to present further results on the
optimization of continuous 1-girder systems and to compare
them with the earlier single span solutions. More precisely,
the following problem is addressed: What is the optimal pre-
stressed concrete stringer system for given bridge length,
width, traffic loading and standard provisions? Specifically,
what are the longitudinal and transverse configurations that
result in minimum superstructure cost and what are the cor-
responding prestressed and non-prestressed reinforcements?
It is emphasized that "an optimal bridge superstructure"
is understood as one of minimum total cost, using standard-
ized girder sections and traffic loading. In this study, CPCI
I-sections 23 and Ontario Highway Bridge Design Code
(OHBDC) 24 loading, safety and design specifications, as
well as Ontario costs, are used. The sensitivity of optimal
solutions to the changes that occurred in the new edition of
the OHBD Code 25 is investigated. Alternative use of stan-
dard AASHTO-PCI sections 2·26 and AASHT0 27 specifica-
tions is also considered. Worth noting is the recent interest
in standardizing, across the United States, the use of pre-
stressed concrete !-sections for bridge design 28 which could
be achieved through the formal optimization (Level 1) pro-
cedures described here and elsewhere. 22
This paper presents a general approach to optimization at
Levels 1, 2 and 3, and readily usable results of optimal
stringer configurations for precast, prestressed concrete
girders. The paper is organized in three main sections: the
first part details the formulation of the optimization proce-
dure; the second part presents its practical implementation
to bridges having design parameters within the investigation
range; the third part offers some practical recommendations,
extension to AASHTO specifications and design examples.
OPTIMAL BRIDGE DESIGN PROCEDURE
Design Variables
Given a bridge of total length L, width Wand some speci-
fied truck and lane loads, the aim of the design is to deter-
mine the minimum cost of the bridge system. The optimum
bridge is defined by the best structural system type (longitu-
dinal and transverse bridge configurations) and optimum de-
sign of its individual members (slab, girders, diaphragms,
etc.). In this paper, the selected structural system consists of
a reinforced concrete slab on precast, prestressed concrete
!-girders, as it represents about 40 percent of the short and
medium span bridges built in the United States and
Canada.4·2'
The optimum longitudinal configuration is defined by the
number of spans, restraint type (simply supported or contin-
uous) and span length ratios (Fig. 2). The transverse config-
uration is defined by the number of girders (or girder spac-
ing and slab overhang) (Fig. 3). Simply supported bridges
(with one, two and three spans) and continuous bridges with
two and three spans are considered, although the results are
62
conservative and may be used for four or more equal spans.
Finally, the optimum designs of the girder and slab are
defined by the girder cross section dimensions (Figs. 4 and
5a), slab thickness, amounts of prestressed and non-
prestressed flexural reinforcements in the girder and slab,
as well as the tendon eccentricities at midspan and supports
(Figs. 5a and 5b).
Objective Function
The assumed merit function is the unit superstructure cost
and is defined as :
Z = [n C8 L + Ccs Vcs + C5 W, + n' Cpc + C5 W,, + C5 W,p] I W L
(1)
where
n = number of girders
n' = number of positive moment connections (at piers)
C8 = cost of precast girder per length (including cost of
fabrication, prestressing, delivery and erection)
Ccs = concrete cost in slab and diaphragms
C5 = non-prestressed steel unit cost
Cpc = unit cost of positive moment connection at piers
including cost of steel and bending of bars (as-
sumed cpc = 1.7 cs ~c· where vpc =volume of pos-
itive moment reinforcement)
Vcs = volume of concrete in slab and diaphragms and
W,, W.n and W,p are the weights of non-prestressed
steel in slab and diaphragms , negative moment
steel in slab at piers and positive moment steel in
girders, respectively.
Design Constraints
In the design of precast concrete girders, all serviceability
limit states (SLS) and ultimate limit states (ULS) require-
ments of the Ontario Highway Bridge Design Code 24
(OHBDC-1983) are to be satisfied. However, only the flex-
ural constraints at SLS, transfer and ULS are considered, in
addition to the side constraints that reflect specific OHBDC
requirements. Other constraints (camber, deflection, vertical
and interface shears, etc.) could easily be added but, in gen-
eral, they affect only marginally the flexural design and are
left to be checked at the final design stage. Special attention
must be given to the connection design and detailing for the
positive moments at piers induced by the creep and shrink-
age effects.
(a) Transfer Stresses Constraints- Normal stresses in
concrete O'er due to girder self-weight should be within the
allowable limits 24 at midspan and support critical sections:
(2)
where fu and frc are the allowable tensile and compressive
stresses at transfer, respectively. For the girder with a con-
crete having J; = 40 MPa (5800 psi), j,1 = -0.24 {!;; = -1.3
MPa (-190 psi) and frc= 0.6:f;; = 18 MPa (2610 psi).
(b) Service Limit State Constraints - In the design of
prestressing steel for girders and non-prestressed reinforce-
PCI JOURNAL
 1-S~
I• •I
(a)

2-S

CPCI900 CPCI1200 CPCI1400

3-S~
L/.3 ~ L/.3 ~ L/.3 ~
Fig. 4. Standard CPCI 1-girder sections.23 All dimensions in
(c) mm ; 1 mm = 0.0394 in.

2-c ~
I• L/2
~
•I•
L/2 ~
•I
(d)

3-c
~ L/.3 ~
L/.3 ~
L/.3 ~
(e)

Fig. 2. Longitudinal bridge configurations: (a) one simple

span , (b) two simple spans, (c) three simple spans, (d) two
continuous spans, (e) three continuous spans. (a)

C.G.Steel
C. G. Concrete

~:_~:_---- -]:--- -:_~-~

·-·- · ·--·-·

1:s' ~I· s ·I· s

w
~I· s ~l·s' :I !/3
:I
(b)
Fig. 3. Transverse bridge configuration.

Fig. 5. Girder design variables: (a) cross section,

ment for positive moments at piers, the normal stresses in
(b) tendon layout.
concrete (<Yes ), mild steel (as) and jacking prestress ( Oj)
should be:
For concrete: (4)
fst ~ <Yes ~ f sc (3a)
where M 8 , M s and M a are the girder, slab and asphalt pave-
ment self-weight moments, respectively. Also, ML and Mcs
where fsr = -0.48 -fJZ = -3 MPa (-435 psi) and fsc = 0.45f~ are the live load moment and creep and shrinkage moment,
= 18 MPa (2610 psi) are the allowable24 tensile and com- respectively.
pressive stresses at service, respectively. (c) Ultimate Limit State Constraints - The ultimate
For non-prestressed steel: load moment Mu should be less than or equal to the resisting
(3b) moment of the section:

For prestressing steel at jacking: (5)

(3c) where Mn is the nominal resisting moment of the section and

If>= 0.85 is the flexural strength reduction factor. In the new
The SLS moment is defined as:24 OHBD Code/' the single understrength factor (If> = 0.85) has

July-August 1993 63
been replaced by strength reduction factors for concrete
( if>c =0. 75), steel (if>s =0.90) and prestressing steel ( 1/>p =0.95).
(d) Side Constraints - These reflect some specific
OHBDC requirements: I ¥1 i¥
s ~ 15 t (6a) •
.e,-o.e.t •
.t.2•0.4.t •
.t 5 •0.6.t
I• I 1'4 •I. •I

t ~ 225 mrn (8.9 in.) (6b)

S' ~Min [0.6 S, 1.8 m (5 .9 ft)] (6c) Fig. 6. Equivalent spans for live load moment computations
at critical sections. 24·30
(() ~ 0.30 (6d)

c bor ~ 100 mrn (3.9 in.) (6e)

Crop~ 80 mrn (3.1 in.) (6f)

where S, S', t, w, cbor and C10 P are the girder spacing, slab (a)
overhang, slab thickness, total tension reinforcement index continuity
of the section, and bottom and top minimum concrete cov- reinforcement
/ reinforced concrete slab
ers, respectively. The total reinforcement index is defined as:
~- x~ ;g F
(7)

Dead and Live Load Analysis

In order to properly account for the construction sequence
.'•
®
I
-ln+1

of this type of bridge, two analytical models are used to
compute the moments at different critical sections.
(a) Model 1: Simply Supported Beam - Initially, the
precast girders are placed on the substructure and the analy-
sis under the girder and slab self-weights is carried out as
for simnle beams. At this stage, only the girder section is ef-
fective in computing the stresses.
(b) Model 2: Continuous Beam- Subsequently, conti-
nuity is achieved by adding non-prestressed reinforcement
over the piers, and the analysis for the live load and pave-
ment weight is carried out as for a continuous beam. Com-
posite (slab and girder) action develops and must be consid-
ered in evaluating the stresses . However, the live load
moment obtained should be multiplied by the transverse dis-
tribution factor S/Dd and augmented by the impact factor I
(dynamic load effect), conservatively estimated as 0.4 for
truck load and 0.1 for uniform lane load. In OHBDC, 24 Dd
values (given for two-, three- and four-lane bridges) depend
on the torsional and flexural parameters a and 8, respec-
tively, in addition to the design lane width:
Dxy + Dyx + D1 + D2
a= os
2(Dx Dy) .
0.25
()= ~ Dx
u * ( vy )
where Dx> Dy, Dxy, Dyx> D 1 and D 2 are the flexural , torsional
and coupling rigidities in both transverse and longitudinal
directions, respectively. The Dd factors have been derived
by using the orthotropic plate analogy for simple span
bridges. 29 In order to use the same factors for continuous
(b)
Fig. 7. Analytical models for continuous precast girders:
(a) statically determinate system , (b) statically indeterminate
system.
bridges, an effective length £* is determined for all critical
sections to compute the flexural parameter in Eq. (8b) which
may be taken as shown24 ·30 in Fig. 6. In the new edition of
the OHBD Code, 25 the determination of the Dd factors has
been simplified to only depend on the bridge type and span
length.
Creep and Shrinkage Moments Analysis
(a) Creep Moments due to Prestressing and Dead
Loads - Continuity of this bridge type requires that posi-
tive moments developed over piers due to creep of the pre-
stressed girders and the effects of live load on remote spans
be considered. Positive moments due to creep under sus-
tained (prestressing and dead) loads are partially counter-
acted by the negative moments due to differential shrinkage
(8a) between the cast-in-place deck slab and precast girders. Of
the several methods of creep analysis proposed in the litera-
ture ,'1.3 233 the Trost-Bazant age-adjusted elastic modulus
method 33 is used to compute the prestress and dead load re-
(8b) straint moments.
For illustration purposes, consider a continuous girder
with n spans, initially assumed freely supported (Fig. 7a),
and subsequently connected for continuity at an age t 1 (Fig.
7b). Using the age-adjusted elastic modulus method, the
moment M~ (t) induced by creep at a support n is given by:

64 PCI JOURNAL
where Mn is the moment at support n if the structure is con-
tinuous and monolithically cast in a single stage.
The restraint moments at the supports are determined by
using any elastic analysis method (e.g., the three-moment
equation); v(t, t0 ) is the creep coefficient (ratio of creep
strain to initial elastic strain); t0 is the age at loading; t 1 is
the age at which continuity is achieved and X is the aging
coefficient introduced because the creep moment develops
gradually with time (Trost ptoposes X = 0.8). Assuming t 1 =
28 days and t = oo, the following values are obtained from
ACI-209: 33 V (t, t0 ) = 0.96 Vu, V (tl, t0 ) = 0.42 Vu and V (t, t1)
= 0.54 vu, where vu is the ultimate creep factor which may
be conservatively taken as 4.5.
Since v(t 1, t0 ) = 0.42vu (i.e., 42 percent of the ultimate
creep will have occurred at the time the continuity is
achieved), the connections should develop the moments
from the remaining creep of 0.54 Vu· Substitution of parame-
ters in Eq. (9) yields:
M~ (t = oo) = 0.80 Mn (10)
(b) Shrinkage Moments - Differential shrinkage be-
tween the reinforced concrete deck slab and the prestressed
concrete girders induces the moment Msh in the composite
section:
(11)
where
A 5 = area of effective slab deck
£ 5 =Young's modulus of slab concrete
Ys = distance from centroid of composite section to cen-
troid of slab
t:sh = differential shrinkage strain24 = 10_.
The (shrinkage) restraint moments induced by differential
shrinkage can also be obtained using the three-moment
equation; however, the presence of creep reduces shrinkage
restraint moments. In the absence of accurate data, the
OHBDC 24 recommends the use of a 60 percent creep reduc-
tion factor. Therefore, the final moment Mcsn at support n
due to the combined effects of creep and shrinkage is :
where Mpn' MDn and Mshn are the restraint moments at support
n due to prestressing, dead load and shrinkage, respectively.
Formulation of the Optimal Bridge Design Problem
The optimization problem may be stated as follows :
Minimize:
Z = [n CgL + Ccs ~s + Cs W, + n' Cpc + Cs W.n + Cs l-v,p] I W L
(13a)
such that:
(13b)
(13c)
July-August 1993
Mj = 0.85 [0.8 (Mpj +MD).:... 0.4 Msh) + 0.75 MLj
~ Asj (0.6 fy) (d- kd/3) ( 13d)
U = 1, 2, ... np) Muj = 1.5 Maj + 1.4 MLj ~ 0.85 Mnj (13e)
(i = 1, 2, . . . n5 ) Mui = 1.1 Mgi + 1.2 M 5; + 1.5 Mai + 1.4 Mu
~ 0.85 Mni (13f)
s ~ 15 t (13g)
t ~ 225 mm (8.86 in.) (13h)
S' ~Min [0.6 S, 1.8 m (5.91 ft)] (13i)
Q)~ 0.30 (13j)
cbor ~ 100 mm (3.94 in.) (13k)
Crop~ 80 mm (3.1 in.) (131)
Eqs . (13b) and (13c) represent the transfer and service
stresses constraints, respectively. Eq. (13d) represents the
SLS constraint on the steel stress for the design of positive
moment reinforcement at the piers. Eqs. (13e) and (13f) are
the ULS flexural strength constraints for negative moment
reinforcement at the nP piers and positive moment reinforce-
ment at the n5 critical span sections, respectively. Eqs. (13g)
to (131) are the side constraints.
Solution of the Optimization Problem
The optimization problem, defined by Eqs. (13a) to (131),
includes both discrete and continuous design variables; the
constraints are nonlinear functions of the design variables.
Solution of such a problem is very difficult and requires the
use of mixed integer programming algorithms which are not
efficient and may not converge to an optimum. To over-
come this difficulty, a two-stage optimization with continu-
ous variables only is performed by using nonlinear program-
ming methods. This approach has been used for the
optimization of simply supported bridge girders. 22
The two-stage optimization operates as follows:
(a) Maximum Feasible Girder Spacing - For each
simple span and continuous bridge and precast girder, the
maximum feasible girder spacing is determined for span
lengths varying from 10 to 30 m (33 to 98 ft) by solving the
optimization problem defined by Eqs. (13a) to (131), but
maximizing the girder spacing instead of minimizing the
cost as objective function. Once the maximum feasible
girder spacing is determined, for every given bridge width
and span length, the minimum number of girders resulting
in the minimum superstructure cost can be obtained.
(b) Optimum Reinforcements - After the optimum
number of girders has been determined, the total superstruc-
ture cost can be minimized by solving the problem defined
by Eqs. (13a) to (131) with the minimization of the pre-
stressed and non-prestressed reinforcements as objective
function.
These two optimization problems can be solved by any
nonlinear programming technique. In this paper, the
65
GAMS/MINOS program, based on the projected Lagrangian • Optimal prestressing (force and tendon layout)
algorithm, has been used. The description of the algorithm is • Non-prestressed steel reinforcements for girder and slab
given elsewhere. 34•35 • Minimum cost of superstructure per unit deck area
• Identification of active constraints

OPTIMAL DESIGN OF BRIDGE SYSTEMS- Optimal Design Solutions for Simple and
PRACTICAL APPLICATION Two- and Three- Span Continuous CPCI Girders
Optimal solution features for the simple and two- and
Investigation Outline three-span continuous girders are presented in Table 2 and
To achieve the stated objectives of the investigation, Figs. 8 and 9. With the completed optimal designs for the
the above -mentioned optimization procedure and the simple and two- and three-span continuous girders, the op-
GAMS/MINOS program have been used to produce optimal timal superstructure costs for discrete bridge lengths up to
bridge designs for a large number of precast, prestressed con- 100 m (330 ft), three bridge widths and three CPCI girder
crete girder bridges. Some 165 optimal solutions have been types in five longitudinal bridge configurations are summa-
generated for the following variations of the main design rized in Table 3.
data indicated by the dots (Table 1):
(a) Five longitudinal bridge configu-
rations: one, two and three simply sup- Table 1. Specimen data for parametric investigation for bridge widths
ported spans, and two and three con- W= 8, 12 and 16m (26, 39 and 52ft).
tinuous equal spans Longitudinal bridge configuration and CPCI girder type
(b) Three girder types : CPCI 900, Span
length 1-S 2-S 3-S 2-C 3-C
1200 and 1400 (m) 900 1200 1400 900 1200 1400 900 1200 1400 900 1200 1400 900 1200 1400
(c) Three bridge widths: W = 8, 12 10 • • • • • • • • • • • • • • •
and 16 m (26, 39 and 52 ft) (i.e., two-, 15 • • • • • • • • • • • • • • •
three- and four-lane bridges) 20 • • • • • • • • • •
(d) Five span lengths: 10, 15, 20, 25 25 • • • • • • • • • •
and 30 m (33, 49, 66, 82 and 98 ft) 30 • • • • •
(whenever feasible)
Metric (SI) conversion factor: I m = 3.28 ft.
For all designs, the following values
are adopted as pre-assigned parameters:
• Class A highway
Table 2. Maximum feasib le girder spacing , optimum girder spacing and optimum
• Slab thickness: t = 225 mm (8.9 in.) number of girders for single-, two- and three-span continuous CPCI girder bridges.
(OHBDC minimum)
• Minimum OHBDC slab reinforce- Girder Girder type and span length
ment (i.e., p = 0.3 percent top and Bridge spacing
width and number CPCI900 CPCI 1200 CPCI 1400
bottom reinforcements in both
W(m) of girders 10m ISm 10m 15m 20m 25m 10m 15m 20m 25m 30m
directions)
• Concrete grades: f~ = 40 MPa (5800 Maximum
All
feasible 3.37 2.52 3.37 3.37 2.95 1.95 3.37 3.37 3.37 3.10 2. 19
psi); f~; = 30 MPa (4350 psi) for widths spacing (m)
"'
girder and f~ = 30 MPa (4350 psi) " OJ)

~
for slab S(m) 2.5 2.5 2.5 2.5 2.5 1.9 2.5 2.5 2.5 2.5 1.9
... 8m
..0

• Steel grades: ipu = 1860 MPa (270 "

'0 n 3 3 3 3 3 4 3 3 3 3 4
ksi) for prestressing steel and !y = 400
·61
~ S(m) 2.9 2.3 2.9 2.9 2.9 1.95 2.9 2.9 2.9 2.9 2.0
MPa (60 ksi) for mild steel ~ 12 m
n 4 5 4 4 4 6 4 4 4 4 6
• Unit costs based on average Ontario
costs always expressed in Canadian
Oil
c:
i:i)
" S (m) 3.1 2.2 3. 1 3.1 2.6 1.95 3.1 3.1 3.1 3.1 2.2
16m
dollars: cg = $514, 486, 580 perm n 5 7 5 5 6 8 5 5 5 5 7
for CPCI 900, 1200 and 1400, re-
spectively; Ccs = $61 7 per m3 and C5 ""' All Maximum
OJ)

~ feasible 3.37 2.79 3.37 3.37 3. 10 2.19 3.37 3.37 3.37 3.30 2.33
= $1400 pert
..0 widths spacing (m)
il
For simple span bridges, Eq . (1) 'E
·en S(m) 2.5 2.5 2.5 2.5 2.5 1.9 2.5 2.5 2.5 2.5 1.9
with cpc = WSII = 0 applies, but the "'0::J 8m n 3 3 3 3 3 4 3 3 3 3 4
additional cost of deck joints, c dj = ::J
c:
'.;:l
$1410 perm, must be added. c:
0 S(m) 2.9 2.9 2.9 2.9 2.9 2.0 2.9 2.9 2.9 2.9 2.3
C,) 12m
Solutions of the 165 bridge designs c: n 4 4 4 4 4 6 4 4 4 4 5
include: '"ff
!"'\ S(m) 3.1 2.6 3. 1 3.1 3. 1 2.0 3.1 3.1 3.1 3. 1 2.2
• Optimal girder spacing and the cor- o(l 16m
M n 5 6 5 5 5 8 5 5 5 5 7
responding minimum number of
girders Metric (S I) conversion factor: I m = 3.28 ft.

66 PCI JOURNAL
Table 3. Minimum superstructure cost per deck area ($/m 2 ) for investigated bridge systems.

Bridge Bridge 1-S 2-S 2-C 3-S 3-C

length width
L (m) W(m) 900 1200 1400 900 1200 1400 900 1200 1400 900 1200 1400 900 1200 1400
8 363 352 388
10 12 341 332 363
16 331 322 351
8 363 352 388
15 12 384 332 363
16 395 322 351
8 352 388 441 433 470 371 365 403
20 12 332 363 420 413 446 349 344 378
16 352 351 409 403 434 339 333 366
8 413 388 426 417 454 370 363 401
25 12 413 363 404 397 430 348 342 376
16 443 351 426 387 417 353 332 364
8 460 415 406 443 368 361 399 467 460 498 374 369 376
30 12 460 437 386 419 346 340 374 446 440 473 353 348 383
16 424 447 376 406 367 330 361 435 430 461 342 337 371
8 393 429 358 396 441 433 470 372 366 379
40 12 373 405 338 371 463 413 446 351 346 381
16 393 393 328 359 473 403 434 360 335 369
8 449 424 387 395 432 424 461 371 364 381
45 12 409 400 377 371 454 404 437 349 344 380
16 449 424 372 359 465 394 425 371 333 368
8 445 421 416 394 426 417 453 370 363 382
50 12 445 397 416 370 447 398 429 348 343 379
16 476 384 416 358 458 387 417 370 332 366
8 488 464 406 443 362 383
60 12 488 416 386 418 342 376
16 451 428 406 406 331 363
8 456 432 421 385
75 12 456 407 426 374
16 487 395 382 362
8 497 457
90 12 497 422
16 460 433
Metric (Sl) conversion factor: 1m' = 10.76 sq ft.

Finally, the optimization results are synthesized in Fig.
10, which identifies the optimum precast girder bridge sys-
tem (i.e., the best longitudinal configuration, as well as the
optimum number and type of girders) for a specified bridge
length and width.
(a) Optimum Girder Spacing (Minimum Number of
Girders) - Optimization results show that the optimum
girder spacing is the same for two- and three-span continu-
ous girders. The maximum feasible and optimum girder
spacing and corresponding number of girders are shown in
Table 2 for the three bridge widths [8, 12 and 16m (26, 39
and 52 ft)] investigated. From Table 2, it is seen that the
maximum feasible girder spacing depends on the span
length, girder type and slab thickness. However, the opti-
mum girder spacing depends on the bridge width in addition
to the above-mentioned parameters.
It is also noted that, for a given precast girder section,
bridge span and width, the optimal girder spacing is less
than the maximum feasible value and always corresponds
to the maximum feasible slab overhang. For a given precast
girder section and bridge width, the optimum girder spac-
ing (or number of girders) is a constant that minimizes su-
perstructure cost up to a certain span length. Beyond this
span length, an extra girder is required, at an increased su-
perstructure cost. For CPCI 1200 and 1400 girders, these
span lengths are .e = 20 m (65.6 ft) and .e = 25 m (82.0 ft),
respectively.
(b) Optimum Prestressing Force and Tendon Layout
- Optimization results show that the optimum prestressing
force and total reinforcement index for the three-span con-
tinuous girders are up to 2 percent higher than for those cor-
responding to the two-span continuous girders. This minor
difference is due to the fact that a large part of the load
(girder and slab self-weights) is carried in the same way
(simple beam) by both configurations.
Further, since both resist the same dead load moments and
only marginally different live load moments, the variations
of the optimum prestressing force with span length for vari-
ous CPCI girders and bridge widths are illustrated in Fig. 8b
for the three-span continuous girders only. However, this
plot can be conservatively used for continuous beams with
any number of equal spans. The optimum prestressing

July-August 1993 67
 4000
Walm
4000

~ w =am z
~ 3000 CPCI1400
J
Oil
3000 CPCI 1400

CPCI 1200
i CPCI1200
.5 Oil
c: 2000

I
e::I
e
2000

1000
CPCI 000~
"ii
!!

l§ 1000
e
8 !
0 0
0 10 20 30 40 0 10 20 30 40
Span length (m) Span length (m)

4000 4000
~ W=Um ~ WaUm

JOil
3000 CPCI 1400 ]
Oil
3000
c:

1§
2000
CPCI 1200 li
i
a
e
::I
e
2000
CPCI1208 -

! 1000 1000
8
0 0
0 10 20 30 40 0 10 20 30 40
Span length (m) .Span length (m)

4000 4000
W =16m Wa16m
~ ~
] 3000 3000

.5
Oil i Oil CPCIUOO

I~ 2000

1000
1 e::I
2000

1000
8 .5
c!
0 0
0 10 20 30 40 0 10 20 30 40
Span length (m) Span length (m)

(a) (b)

Fig. 8. Optimum prestressing force for CPCI girder bridges: (a) simple span bridges, (b) continuous span bridges.
Note: 1 kN = 0.225 kips; 1 m = 3.28 ft.

68 PCI JOURNAL
 0.07 0.07
:s .8
~ 0.06 CPCI
.s 0.06
i!!
ll
e 0.05 CPCI 5E 0.05 CPCI

j 0.04 CPCI j 0.04

CPCI 12001--__.
.5
e 0.03 -~ 0.03
5
9
sg
0.02 0.02
§ §
-~ 0.01 E
·~
0.01
8 8'
0.00 0.00
0 10 20 30 40 0 10 20 30 40
Span length (m) Span length (m)

(a) (b)

Fig. 9. Optimum total reinforcement index for CPCI girder bridges: (a) simple span bridges, (b) continuous span bridges
(all bridge widths) . Note: 1 m = 3.28 ft.

forces for simply supported girders Table 4. Optimum number and type of girders for two- and three-span continuous
shown in Fig. Sa are seen to be 10 to CPCI girder bridges and various bridge widths.
40 percent higher than for the corre- Girder type and span length, m
Bridge
sponding continuous girders. width CPCI900 CPCI1200 CPCI 1400
As expected, it is seen that for a W(m) 10 15 10 15 20 25 10 15 20 25 30
given span length, the amount of pre- 8 3 3 3 3 3 4 3 3 3 3 4
stressing decreases for larger girder
9 3 4 3 3 3 4 3 3 3 3 4
depths. The lowest values always cor-
10 3 4 3 3 3 5 3 3 3 3 5
respond to the CPCI 1400 section,
which requires very limited prestress- 11 4 4 4 4 4 6 4 4 4 4 5
ing for spans smaller than 15 m (49 12 4 4 4 4 4 6 4 4 4 4 5
ft). This suggests that for short spans, 13 4 5 4 4 4 6 4 4 4 4 6
a reinforced concrete girder with the 14 4 5 4 4 5 7 4 4 4 5 6
same depth will be able to sustain the
15 5 6 5 5 5 7 5 5 5 5 7
same loads.
16 5 6 5 5 5 8 5 5 5 5 7
The optimum prestressing force al-
ways corresponds to the maximum Metrtc (SI) converston factor: l m = 3.28 ft.

midspan tendon eccentricity (i.e., min-

imum concrete cover), with the tensile service stress con-
straint being active. However, for the CPCI 900 section and
f = 15 m (49ft), the tendon eccentricity is not maximal be-
cause the transfer tensile stress constraint also becomes ac-
tive and thus limits the feasible eccentricity. For the end
support eccentricity, any feasible solution can be used since
its effect on the optimum design is negligible. However, tak-
ing the smallest feasible value may reduce the amount of
positive moment reinforcement at the piers.
(c) Optimum Total Reinforcement Index- The total
reinforcement index is found to be rather low (comax = 0.06)
for the three investigated girder sections. This suggests that
the ductility constraint [Eq. (6d)] is redundant for this type
of girder. Furthermore, the identical values obtained for all
bridge widths indicate that the number of lanes does not in-
fluence the total reinforcement index.
Some additional non-prestressed reinforcement is re-
quired in compliance with the ULS flexural strength crite-
rion. This reinforcement percentage varies between 0.85 and
0 percent, with higher percentages needed for shorter spans.
Variations of co with span length for the three precast girder
types are shown in Fig. 9. Since the amount of prestressing
is known from Fig. 8, and the total reinforcement index is
known from Fig. 9, the amount of non-prestressed rein-
forcement is easily determined from Eq. (7) .
(d) Optimum Unit Cost of Superstructure- Table 3
shows unit superstructure costs (i.e., cost per deck area) for
bridge lengths varying from 10 to 90 m (33 to 295 ft), three
bridge widths [8, 12 and 16m (26, 39 and 52ft)], three CPCI
girder types and the five longitudinal bridge configurations
investigated in this paper. From Table 3 it is noted that:
• Configurations 2-S and 3-S are never competitive because
of the added cost of the deck joints and are reported here
for the sake of completeness only.
• For the competitive configurations (1-S, 2-C and 3-C),
costs are almost constant up to a certain span length be-
yond which they increase sharply, due to an increase in
the required number of girders. As an example, the cost of
CPCI 1200 at f =25m (82ft) increases on average by 17
percent vs. the cost at f = 20 m (66 ft).
• In general, unit cost decreases for increasing number of
lanes. Thus, the superstructure cost for CPCI 1400 and

July-August 1993 69
 72 ft and f > 72 ft), the CPCI 1200
and 1400 are optimal, respectively.
cs~~----------------------------~
W•8m • For total bridge lengths L ~ 66 m
4&) (2-lane bridge) (217 ft), the optimal system always
consists of CPCI 1200 sections,
with three, four or five girders for
W = 8, 12 or 16m (26, 39 or 52ft),
respectively.
• For L > 66 m (217 ft), the optimal
system always consists of CPCI
1400 sections, with four to seven
girders depending on W and L (Fig.

-
N
E
10 and Table 4).
(e) Governing Criteria (Active
Constraints) - In all cases investi-
4&l~w~.~12~m----------------------------------~ gated, the service tensile stress [ <Yes =
fsv Eq. (13c)] is the only active SLS

~
0
~
0
()
425 (3-lane bridge)

2-c
~ L/3 ot 3-C1.(3 * .rL/3

~I
. ~--
I 5-l.(X)
constraint. Therefore, all other SLS
constraints may be disregarded, except
for some special cases where transfer
stress constraints may also be active
? 'f
-
(])
"0 . ~L/2 L/2 J and need to be considered [for f = 25
375 l·S .... '·• 4-l.(X) r-
~ ·~l.(X) and 30 m (82 and 98 ft) , <Yet = fcc, Eq.
~1l-!--4'"' : . .
+- 4-l.(X)
I
0 (13 b) governs for CPCI 1200 and
(.) 350
...4-9{1J\
~-=:.
325 •• •• • •• •
: -
4-laD
- - _J
4-l:m
1400, respectively, while <Yet = fu,
Eq. (13b) governs for CPCI 900 and
4-l:lD f = 15m (49ft)].

&QL-__.___.___.___~--~--~--~--~--~~
Optimum Bridge
4&l~w~.~l6~m----------------------------------~ Superstructure System
425 (4-lane l:xldge)
3-C f7-1'4D
At this point, the problem to be
~ L/3 .'fi L/3 4!. L/3~ 1
solved is: What is the optimum system
2-<: ..J
400
, L/2 1./2 ~ ? \Jf6.14XJ for some given bridge length and
width; i.e., which of the five longitudi-
375 · --~·· ~ · , \ ·• 'I5-14D I nal configurations and transverse con-
\
6: 1~5-l.(X)
I ,... 00
figurations result in the most economi-
3eO ~~···
~
.s:mr..
. j
I
____
cal superstructure?
The answer is provided by Fig. 10,
325 • • • •• • 5-l:m 5-12CO
which is a synthesis of optimal design
5-l:m solutions in Tables 2 and 3. Fig. 10 il-
lustrates the variations of optimal
bridge costs per unit deck area (Z)
with the total bridge length (L), for
vario us longitudinal configurations
Total bridge length L(m)
and girder sections. Its study justifies
the following remarks:
Fig. 10. Optimum bri dge systems for W = 8, 12 and 16 m (26, 39 and 52 ft).
• Sol utions 2-S and 3-S (two and
Note: $1/m2 = 0.093 $/fF.
three simple spans) are not plotted
because they are not competitive
.f = 25 m (82 ft) of a fo ur-lane bridge is 10 percent (high additional costs of the required deck joints).
cheaper than that of a two-lane bridge. • The optimum longitudinal configuration depends only on
• For short span bridges [ f ~ 13 m (43 ft)], both CPCI the total bridge length and is independent of its width.
1200 and 900 may be considered equally competitive for • Bridge configurations 1-S, 2-C and 3-C are optimal for
all configurations, because CPCI 1200 is a little cheaper short [L ~ 27 m (89 ft)], short to medium [27 < L ~ 44 m
than CPCI 900. However, if the unit costs of the two gird- (89 < L ~ 144ft)] and medium [55 < L ~ 100m (180 < L
ers are equal, the CPCI 900 girder will be optimal for this ~328ft)] length bridges, respectively.
span range. For 13 < .f ~22m and f >22m (43 < f ~ • For 44 < L ~55 m (144 < L ~ 180ft), the three-span con-

70 PCI JOURNAL
 tinuous (Solution 3-C) bridge is not necessarily optimal
although its superstructure cost is about 7.5 percent
cheaper than that of the two-span continuous (Solution
2-C) bridge. Determination of the optimum configuration
should be based on the minimum total bridge cost (super-
structure plus substructure), as discussed in the next
section.
• For 27 < L ~ 44 m (89 < L ~ 144ft), the two-span contin-
uous (Solution 2-C) bridge is optimal with CPCI 900 and
1200 as girder sections. The optimum number of girders
is three, four and five for W = 8, 12 and 16 m (26, 39 and
52ft), respectively, as shown in Fig. 10.
• For 55< L ~ 100m (180 < L ~328ft) , Solution 3-C is the
only configuration that remains feasible. The CPCI 1200
and 1400 are the optimal girder sections for bridge
lengths smaller or larger than 65 m (213ft), respectively.
• For L > 65 m (213 ft), the three-span continuous bridge
with CPCI 1400 girders should be compared to the four-
span continuous bridge with CPCI 1200 or 1400 girders
and other systems - such as voided slab or box girder
bridges - which may be competitive for medium to long
span bridges.
• For short span bridges with f ~13m (43ft) [i.e., L ~ 13m
(43 ft) for Solution 1-S, L ~ 26 m (85 ft) for Solution
2-C and L ~ 39 m (128 ft) for Solution 3-C], CPCI 900
and 1200 sections are competitive because their unit costs
are almost equal. In such cases, a final selection may be
based on some alternative objective function (minimum
weight or minimum reinforcement) or response criterion
(minimum deflection).
Solution Sensitivity Considerations
The optimal solutions presented in this study are based on
a set of parameters consistent with current design practice.
The parameters to be considered in concrete bridge design
are longitudinal configuration, bridge length and width, traf-
fic loading, slab thickness, girder (standard or non-standard)
section type, concrete grade, superstructure and substructure
cost parameters.
In this section, the validity and possible application of the
optimal solutions obtained outside the range of adopted pa-
rameters are examined.
(a) Longitudinal Configuration - Optimal bridge sys-
tems of 1-S, 2-S, 3-S, 2-C and 3-C type (Fig. 2) have been
identified. The optimization procedure can be applied to
any number of simple or continuous spans. For any span
length, results of the present study of 2-C and 3-C configu-
rations are still valid and conservative for 4-C, 5-C, etc.,
configurations.
(b) Bridge Length (L) -With the three standard girder
sections and five longitudinal configurations considered, op-
timal solutions for bridges not exceeding 100 m (328 ft)
have been determined. Longer bridges may be designed
with other than adopted standard girders (e.g., CPCI 1900,
2300 or AASHTO-PCI Type V, VI, etc.) or longitudinal
configurations (4-C, 5-C, etc.). It also must be noted that for
bridges with spans in excess of 40 m (131 ft), other struc-
tural systems than reinforced concrete slab on precast, pre-
July-August 1993
stressed concrete !-girders may represent competitive alter-
natives (prestressed concrete box girders or other forms of
construction).
(c) Bridge Width (W)- Although optimization has been
conducted for bridge widths W = 8, 12 and 16 m (26, 39
and 52 ft) , additional data are provided in Table 4 which
lists the optimal number of girders for bridge widths varying
between 8 and 16 m (26 and 52 ft). Determination of the
prestressing force is possible by interpolating in Fig. 8 or by
using Eq. (14).
(d) Traffic Loading- The study is based on the OHBD
Code24 loading specifications for Class A highways (average
daily truck traffic higher than 1000 or traffic higher than
4000).
In the new edition of the OHBD Code/ 5 the two 140 kN
(31.5 kips) axles have been increased to 160 kN (36 kips),
which will induce an increase in the live load moment of
smaller than 14 percent. However, the corresponding impact
factor (function of the number of axles governing the de-
sign) is equal to 0.25 for the span lengths investigated in this
paper. Since an impact factor of 0.4 has been assumed in
this study, which is 60 percent higher than the new 0.25
value, these optimal solutions remain feasible within the
new OHBD Code requirements.
It will also be shown in the next section that the design re-
sults are comparable to those based on AASHTO HS20
loading/ 7 and may also be used for lighter loadings (e.g.,
AASHTO HS15 and H20 loadings).
(e) Slab Thickness (t) - Optimal solutions assume a slab
thickness t = 225 mm (8 .9 in.), the minimum required by
OHBDC. Optimal designs will remain valid for slab thick-
nesses of at least 200 mm (8 in.).
(f) Girder Section Type - The optimization program
can readily include section variables in order to determine
the optimal girder sections (Fig. 5a) 22 in addition to system
optimization variables. Consideration of general economy
(formwork, fabrication and transportation costs) suggests
that using standardized girder sections is more economical,
although not structurally optimal. This is why the CPCI 900,
1200 and 1400 girders have been used in the present investi-
gation. The CPCI 1900 and 2300 girders are not included
because related costs are unavailable.*
(g) Concrete Grade (j~) - Girder optimization has as-
sumed J; = 40 MPa (5800 psi). Results remain valid on the
conservative side for higher concrete grades.
(h) Superstructure and Substructure Costs - Opti-
mization of precast, prestressed concrete girder bridges has
been investigated for the minimum cost of superstructure,
assuming relatively moderate or low substructure costs. The
effect of substructure costs are briefly discussed in the next
section.
Although absolute costs of bridge systems will vary with
the specific unit costs, the optimal design solutions will be
insensitive to variations of these unit costs as long as their
relative magnitudes remain unchanged.
* Private communication with Kris Bassi, Ministry of Transponation, Ontario,
Canada, September 1990.
71
 Table 5. Comparison of CPCI 23 and AASHTO-PCI 2•25 precast concrete girder sections.
Moment of Top section Bottom section
Girder types Depth,H Area, A inertia,/g modulus,S1 modulus,Sb
and ratios (rnrn) (10' rnrn') (10' rnrn') (10' rnrn') (10' rnrn')

CPCI 900 900 218 19.30 38.40 48.60

AASHTO-PCI II 900 238 21.20 41.43 52.77
Ratio 1.000 0.916 0.910 0.927 0.921

CPCI 1200 1200 320 53.9 80.00 102.30

AASHTO-PCI III 1143 361 52.2 83.08 101.37
Ratio 1.050 0.886 1.032 0.960 1.010

CPCI1400 1400 413 102.6 134.2 161.4

AASHTO-PCI IV 1372 509 108.5 146.0 172.8
Ratio 1.020 0.811 0.946 0.920 0.934
Metric (SI) conversion factors: I mm =0.0394 in.; I mm' =0.00155 in.' ; I mm' =0.000061 in.'; I mm• =0.0000024 in.'

PRACTICAL RECOMMENDATIONS

Optimality Criterion for Prestressing Force

The service tensile stress constraint, which is always ac-
tive for this type of bridge, may be adopted as an optimality
criterion for the determination of the prestressing force:

P = [fst + (M8 + M5 )/ Sb + (Ma + 0.75ML)/ Sbc] I (l/A + ejSb)

(14)
TypeD Type m Type IV
where sb and sbc are the section moduli with regard to bot-
tom fibers of the girder and composite (girder plus slab) sec-
Fig. 11. AASHTO-PCI standard girder sections 2•26 Types II, Ill
tions, respectively. and IV. All dimensions in inches; 1 in.= 25.4 mm.

Substructure Cost Consideration

sub 2_c
Fig. 10 shows that the optimization of the superstructure 1+-- -
yields unique optimum bridge designs, except at two spe- 8 - SUP2-C
I - sup3-c sub3-c (15b)
cific length ranges where two configurations are competitive --- +---
and the final choice is to be made by comparing their total SUP2-C SUP2 -C
(superstructure and substructure) costs.
It is noted from Fig. 10 that the superstructure cost ratio
• For 27 < L s;; 32 m (89 < L s;; 105 ft), the superstructure
sup 3 _c/sup 2_c = 0.93 (constant) for all three bridge widths
cost for the 1-S configuration (feasible for the CPCI 1400
considered. Hence, the ratio 8 1 depends on two parameters:
girder only) is on average 24 percent (rather high) more
expensive than the two-span continuous (2-C) configura- ~ = sub2_c /sup 2_c (15c)
tion, which may be considered as the optimum (Fig. 10). and
However, if the cost of the intermediate pier in the 2-C ~ = sub3_c/ sub2_c (15d)
configuration is very high (e.g., very difficult access site),
the 1-S configuration may be preferable. 8l -
- 1 + 82
(15e)
• For 44 < L s;; 55 m (144 < L s;; 180 ft), the 2-C and 3-C 0.93+8283
configurations may be competitive, and the. final choice
will depend on their substructure cost ratio and the super- From Eq. (15e), it can be concluded that:
structure to substructure costs ratio, as is shown in the fol- • 3-C is optimal for ~"' 1.
lowing. Let 81 be the ratio of the total bridge costs for the • 2-C is optimal for~> 1.15.
2-C and 3-C configurations: • 3-C and 2-C are competitive for~< 1.15 and~ s;; 0.7.
= (sup+ sub ) 2_c
81 (15a) AASHTO Implementation
(sup+ sub ) 3_c
(a) AASHTO-PCP·25 Girders vs. CPCP3 Girders -
where "sup" and "sub" are the costs of the superstructure From Figs. 4 and 11, it is seen that the precast CPCI girders
and substructure, respectively. Division of both numerator 900, 1200 and 1400 have very similar geometries with the
and denominator by sup2_c yields: AASHTO-PCI girders Types II, III and IV, respectively.

72 PCI JOURNAL
(' .
 Table 6. Comparison of OHBDC 24 and AASHT0 27 HS20 live load moments.

Code Simple span (1-S) Two-span continuous (2-C)

moments Span section Span section Support section
and ratios 10 20 30 10 20 30 10 20 30

OHBDC
658 1872 3426 526 1500 2892 468 1298 2394
(kN-m)

AASHTO
439 1229 2018 332 970 1676 290 775 1512
(kN-m)

Rl 1.499 1.523 1.698 1.584 1.546 1.720 1.614 1.675 1.583

R2 0.941 0.956 1.066 0.994 0.970 1.080 1.013 1.052 0.994

Metric (SI) conversion factor: I kN-m = 8.850 kip-in.

Table 5, which compares the geometrical and mechanical
characteristics of the two sets of girders, indicates that they
have virtually identical section moduli S1 and Sb. Differences
between CPCI and AASHTO-PCI section moduli are infe-
rior to 8 percent for Types II and IV vs. CPCI 900 and 1400,
and less than 4 percent for Type III vs. CPCI 1200. This sug-
gests that the AASHTO-PCI Types II, III and IV are struc-
turally equivalent to CPCI 900, 1200 and 1400, respectively.
(b) AASHT0 26 vs. OHBDC24 Design Requirements-
SLS Constraints
Allowable Stresses: OHBDC 24 and AASHT0 26 adopt the
same allowable tensile and compressive stresses at SLS, i.e.,
-0.5 .fJZ and OAf~, respectively.
comparison purposes, one may conservatively adopt Dd =
1.9. The ratio of OHBDC to AASHTO (HS20) maximum
truck (lane) load moments, R 1, and the ratio of their corre-
sponding effective design moments, R 2 , are related by:
(16e)
For the simple and two-span continuous girders, Table 6
shows the maximum truck (lane) load moments by the two
codes, their ratios R 1 and ratios R2 of the effective design
moments. The real differences between the AASHTO and
OHBDC design service live load moments, illustrated by
ratio R2 , remain inferior to 6 percent, and, therefore, for all
practical purposes, identical designs will satisfy SLS re-
quirements of both codes.

Effective Service Moments:

OHBDC: M=MD+0.75 ML
AASHTO:
where MD and ML are the moments due to dead and live
loads, respectively.
Although the OHBDC truck (lane) load moments are
higher than the AASHTO (HS20) truck (lane) load moments
(see Fig. 12), the OHBD Code considers only 75 percent of
these moments at SLS. Moreover, transverse distribution
factors in the OHBDC are larger than the AASHTO values
(Dd = 1.8 to 2.15 vs. 1.67). Hence, OHBDC and AASHTO
live load moments should be compared on the basis of their
effective truck (or lane) load design moments, which in-
clude the transverse distribution factor S/Dd, SLS load com-
bination factor and impact factor / . These effective design
live load moments M~ may be expressed as:
OHBDC: M~ =0.75 (1 + /) (SIDd) Mr~/2
AASHTO: M~ = (1 + l) (S/1.67) M 1;/2
where Mtll is the moment due to the governing truck or lane
load. In the OHBD Code, 24 Dd is not a constant as in
AASHTO, but a function of the number of lanes, torsional
and flexural parameters of the bridge. In the new OHBD
Code/ 6 the determination of Dd has been simplified to de-
pend on the span length only (for a given bridge type). For
ULS Constraints
(16a) The ultimate limit state (ULS) flexural requirements in
OHBDC and AASHTO codes are:
(16b) OHBDC:
Mu = 1.1 Mg + 1.2 Ms + 1.5 Ma + 1.4 ML :S: 0.85 M 11 (5)
AASHTO:
In AASHTO, a constant overload factor is applied on the
total dead load, while in OHBDC this factor varies from 1.1
(for girder weight moment) to 1.5 (for asphalt pavement
weight moment). From Table 6, it can be assumed on aver-
age that the OHBDC live load moment is 60 percent higher
than the AASHTO moment R 1 = 1.6. Taking the AASHTO
live load moment as a reference, and including the trans-
verse distribution factors, the OHBDC live load moment is
40 percent higher than the AASHTO moment (1.6 x
(16c) 1.67/1.9 = 1.4). In addition, since the main dead load mo-
ment is due to the girder and slab weight moments, the vari-
ous dead load factors in Eq. (5) are substituted by a single
(16d)
factor, approximately 1.2, and the OHBDC ULS require-
ment becomes:
In the new OHBD Code, 25 the single understrength factor
(0.85) has been substituted by strength reduction factors for
concrete (0.75), steel (0.90) and prestressing steel (0.95).

July-August 1993 73
 60 140 !«) 200 160 oale load kN
30 70 70 100 80 wlleellood kN

~'~~+~i~~~~--'~~~*
:
3.6. ~21 • 6.0 ·'·
1~·-~----~~~~·xO ________~:
7.2 L 412 9p98 140
Cllle L.Dad kN
112

f1~ II !Tfl ~I I I II II 'I

,, (typ.)
I
1 I II I II I' ltjq
~ 1:3.6.~21• 6t~.o·'· 1.2 :1 1
~ ! ms J!:avel ~m\ ~r- Hiqhway Class

t_ I.
~ .. m
"'[
e =
e~Oil
~
Uniform!~ Distributed Lood
A
10
B Ct or Cz
9 8

r..~-ffi-iit-~~-o-a4:~! i~j
q =(kN/m)

l~.-
0.25
mm
Plan
m
(AI dhntnsions 111 •)
Q.25~

OHBDC truck load OHBDC lane load

IIIJIIIO .. . , k N I I a f -
j...- ~ 111111 ""1 2UDO. 1111 kNIIaf-
~ / UnifO<m laldl40 pff 18.4 kN/ml of laid lono

I
H$20-44 fMS181 1000 lb 138 kNI 32JIIIO ft.t 1144 kNI 32JIIIO 1b 1144 kNI
H$110. . fMS13.S 1000 lb 127 kNI :MJIIIO lb fUJI kNI :MJIIIO 1b 1108 kNI

AASHTO HS lane load

W • Combined -ight on tht lim two uloa wllid> io tht ,._ • for 1M---'""
H fM! trudt .
V • Var-spoo:int-14 It to 30ft 14.2&7108.144 ml indlllift.S.Ocinvto•-

---T
10 ft f3.0.S ml
Clooronooand

mil~"'
k~'j r~on
2ttl I Itt I 12ttt
111.110 "'' 11.130 .., 11.110 "''

AASHfO HS ttuck load

Fig. 12. OHBDC24 and AASHT0 27 live loadings: (a) OHBDC truck and lane loadings [Axles 2 and 3 have been changed to
160 kN (36 kips) in the new OHBDC 25] , (b) AASHTO HS20 truck and lane loadings.

74 PCI JOURNAL
However, since the optimal girder designs obtained are al-
ways under-reinforced with low reinforcement indices, the
corresponding equivalent understrength factor will be higher
than 0.85. Hence, the original assumption (cf> = 0.85) is more
conservative than the new code safety requirements.
Let A0 be an overall safety factor defined as the ratio of
the ultimate resisting moment to the service moment. The
overall safety factors A0 for the two codes are:
OHBDC:
A0 =M/lf>MD+L = 1.41 + 0.953 ML/ MD+L (17c)
AASHTO:
A0 = M/cf> MD+L = 1.44 + 0.967 ML/ MD+L (17d)
where Mv+L is the service dead and live load moment. The
difference between the two safety factors does not exceed
2 percent for all practical values of MLIMD+V and it is con-
cluded that the same design will be feasible for both the
AASHTO and OHBDC codes.
Optimal Design Procedure
With the design aids resulting from the optimization study
of precast, prestressed !-girder systems, the preliminary op-
timal design of a bridge system consists of four simple steps
to determine: (a) system selection, (b) optimum prestressing
force and tendon layout, (c) non-prestressed steel for ULS
flexural strength (if needed), and (d) bottom and top rein-
forcements at piers.
(a) System Selection- Given the total bridge length L,
width W and OHBDC or AASHTO (HS20) loadings, use
Fig. 10, Table 2 and Table 4 to determine the optimal con-
figuration, number and type of girders.
(b) Optimum Prestressing and Tendon Layout-
Given the span length £, the optimum prestressing is ob-
tained from Figs. 8(a) and 8(b) for simple and continuous
span bridges, respectively. For bridge widths between the
values investigated in this paper, interpolate linearly from
Fig. 8 or use Eq. (14).
The tendon layout adopted for the type of girders investi-
gated has two depressed points at the third span from each
support (Fig. 5b). The eccentricity at the middle third span
(ec) should be taken as the maximum feasible value (i.e., for
minimum concrete cover), while the eccentricity at the sup-
port (ee) could have any value that satisfies the transfer
stress constraints. Let "Ptim be a certain limit value for the ef-
fective prestressing force:
where f3 is the ratio of the initial to effective prestressing
forces. Thus:
(18b)
(18c)
(c) Non-Prestressed Steel for ULS Flexural Strength
- Determine m from Fig. 9 as a function of the span length.
Find the non-prestressed reinforcement from Eq. (7).
July-August 1993
(d) Bottom and Top Reinforcements at Piers- Deter-
mine the bottom and top reinforcements at piers from Eqs.
(13d) and (13e), respectively.
Of course, a final design and thorough check will follow
and possibly result in appropriate adjustments of the opti-
mum preliminary design.
Design Example 1:
Member vs. System Optimization
The relevance and benefits of system optimization vs.
member optimization are illustrated by a cost comparison
of optimal designs for a reinforced concrete slab on a pre-
cast, prestressed !-girder bridge with a total length L = 60 m
(197 ft) and width W = 16 m (52 ft) (four lanes) for two
cases: (a) specified longitudinal configuration and (b) vari-
able longitudinal configuration.
(a) Design 1: Specified Configuration
• Consider a two-span simply supported girder bridge (2-S)
with £ = 30 m (98 ft).
• Optimum girder section is CPCI 1400, the only feasible
alternative for £ = 30 m (98 ft).
• Design requires seven CPCI 1400 girders'' with P = 3700
kN (832 kips) and ec = 535 mm (21 in.) (maximum feasi-
ble eccentricity).
• From Table 3, unit superstructure cost per deck area Z =
$451 per m' (Z = $42 per sq ft), yielding a total super-
structure cost of about $433,000.
(b) Design 2: System Optimization
• From Fig. 10, the optimum solution for L = 60 m (197ft)
and W = 16 m (52 ft) is the three-span continuous (3-C)
bridge system.
• Optimum solution requires five CPCI 1200 girders.
• From Fig. 8b, the optimum prestressing force is P = 2300
kN (517 kips) and ec = 427 mm (17 in.) (maximum feasi-
ble eccentricity).
• From Fig. 10 or Table 3, unit superstructure cost per deck
area Z = $335 per m' (Z = $31 per sq ft), yielding a total
superstructure cost of about $322,000.
The 25.7 percent cost reduction of Design 2 vs. Design 1
demonstrates the economic importance of system optimiza-
tion vs. member optimization.
Design Example 2:
AASHTO-PCI Application
A reinforced concrete slab on precast, prestressed
AASHTO-PCI girders bridge of total length L = 130 ft
(39.7 m) and width W =40ft (12.2 m) is to be designed ac-
cording to AASHTO specifications for the HS20 loading.
The slab thickness is 8.5 in. (216 mm) with an asphalt
pavement of 3.5 in. (89 mm). For the girders, f~ = 6000 psi
(41 MPa) and f~; = 4500 psi (31 MPa), and for the slab,
f~ = 4500 psi (31 MPa). Also, fpu = 270 ksi (1860 MPa)
and fy = 60 ksi (414 MPa).
(a) Optimal Bridge System Selection
From Fig. 10, for L =130ft("' 40 m) and W =40ft ("'12m),
the best structural system consists of four CPCI 1200 two-
span continuous [span length £ = 65 ft (20 m)] girders,
75
equivalent to four AASHTO-PCI Type III girders and S =
9.5 ft (2.9 m) girder spacing.
(b) Optimum Prestressing and Tendon Layout
From Fig. 8b and for a span length of 65 ft (""20 m),
the optimum prestressing for the AASHTO-PCI Type III is
483 kips (2150 kN) with eccentricity at midspan of 17 in.
(427 mm). The tendon layout is depressed at the third span
points [21.6 ft (6.58 m)] from each support, as shown in Fig.
5b. The end eccentricity is easily determined from the trans-
fer stress constraints. ' piers is determined from Eqs. (13d) and (13e).
(c) Non-Prestressed Steel for ULS Flexural Strength
From Fig. 9b, for £ = 65 ft (20 m) (and AASHTO-PCI
Type III), the total reinforcement index is m = 0.03. The
amount of non-prestressed reinforcement is obtained from
Eq. (7):
As= (Wbd f~ -Aps fps)lfy = 0.37 in. 2 (239 mm2)
(d) Bottom and Top Reinforcements at Piers
Eqs. (17d) and (17e) yield the positive (bottom) and nega-
tive (top) moment reinforcements at the piers as 3.2 in. 2
(2065 mm2) and 6.9 in. 2 (4452 mm2), respectively.
CONCLUSIONS
1. The general optimization approach presented in the
paper is ideally suited for determining the optimal designs
of precast, prestressed concrete sections, members or bridge
systems.
2. Standardization of optimal bridge systems through opti-
mization of longitudinal and transverse bridge configura-
tions is the main objective of the paper. The overall design
economy may rationally be improved by optimizing bridge
systems rather than their (standardized) components.
3. Optimal longitudinal configurations are simply sup-
ported girders (1-S) for bridge lengths not exceeding 27m
(89 ft); two-span continuous girders (2-C) for bridge lengths
varying between 28 and 44 m (92 and 144 ft); and three-
span continuous girders (3-C) for bridge lengths between 55
and 100m (180 and 328ft). For bridge lengths of 44 to 55 m
(144 to 180ft), both optimal superstructure systems 3-C and
2-C are competitive as long as the substructure cost of the
former remains moderate; otherwise, the two-span continu-
ous system is more economic.
4. Optimal girder sections: For total bridge lengths L :::; 66 m
(217 m), the optimal system always consists of CPCI 1200
sections [with three, four or five girders for W = 8, 12 or 16m
(26, 39 or 52ft), respectively]. For L > 66 m (217 ft), the
optimal system always consists of CPCI 1400 sections (with
four to seven girders depending on Wand L) (Fig. 10 and
Table 4).
5. Optimal system selection may be obtained directly
from Fig. 10 by specifying the bridge length (L) and width
(W). Although Fig. 10 identifies optimal bridge systems
under OHBDC traffic loading, it may conservatively be
used for lighter loading schemes.
6. Maximum feasible girder spacing is insensitive to
bridge width and depends only on the bridge configuration
(simple or continuous), span length, girder type and slab
76
thickness. However, the optimal girder spacing depends on
the bridge width in addition to all other parameters.
7. Prestressing reinforcement of !-girders may be found
directly from Fig. 8, which corresponds to maximum feasi-
ble eccentricity (i.e., minimum concrete cover) at midspan.
The support eccentricity may be determined from Eqs. (18b)
and (18c). Additional non-prestressed reinforcement re-
quired at midspan may be determined from Fig. 9 and
Eq. (7). The non-prestressed continuity reinforcement at the
The total reinforcement index of precast girders is rather
low and is marginally affected by the bridge width. It in-
creases with span length and decreases with girder depth.
8. Although the superstructure costs of various bridge sys-
tems (Table 3) depend on the adopted unit costs, optimal
bridge systems are unaffected by absolute values of the lat-
ter as long as their ratios remain comparable. Hence, the
system effectiveness will conserve the trends found in this
investigation, while real costs may vary with the specified
economic conditions of a particular project.
9. Design aids provided in the paper (Figs. 8, 9 and 10
and Tables 2 and 4) may be used to generate preliminary
designs, to be detailed in compliance with any additional
economic or structural requirements (shear strength, deflec-
tion and other parameters). Alternatively, basic design pa-
rameters may be determined by using the prestressing force
optimality criterion.
10. Sensitivity considerations of optimal solutions suggest
that results obtained in this study are applicable to most
practical values of design parameters. Also, these results are
representative of general trends in optimizing bridge config-
urations and may be used for preliminary designs with
bridge standard specifications other than OHBDC.
11. AASHTO implementation of bridge system optimiza-
tion is direct and only requires substitution of standard sec-
tions AASHTO-PCI Type II, III and IV for CPCI 900, 1200
and 1400 sections, respectively.
12. Extensions of bridge system optimization procedures
are necessary and will be considered separately. In particu-
lar, the following topics deserve further investigation: opti-
mization of precast, prestressed concrete !-girder sections to
be standardized, other competitive transverse configurations
(prestressed concrete voided slab or box girder, composite
steel girder-concrete deck, etc.) and comprehensive sensitiv-
ity studies.
ACKNOWLEDGMENT
The work reported in this paper is part of a Ph.D. thesis
prepared by the first author under the direction of the second
author, in partial fulfillment of the degree requirements in
the department of civil engineering, University of Water-
loo, Waterloo, Ontario, Canada.
The financial support of the Natural Sciences and Engi-
neering Research Council (NSERC) of Canada under Grant
A-4789 is gratefully acknowledged. The authors are in-
debted to the Ontario Ministry of Transportation for provid-
ing valuable bridge design data and to the PCI JOURNAL
reviewers for their constructive comments.
PCI JOURNAL
 REFERENCES
1. ACI-ASCE Committee 343, "Analysis and Design of Rein-
forced Concrete Bridge Structures," ACI Manual of Concrete
Practice, Part 4, American Concrete Institute, Detroit, MI,
1989.
2. Precast Prestressed Concrete Short Span Bridges - Spans to
100 feet, Second Edition, Precast/Prestressed Concrete Insti-
tute, Chicago, IL, 1981.
3. Heins, C. P., and Lawrie, R. A., Design of Modem Concrete
Highway Bridges, John Wiley & Sons, New York, NY, 1984.
4. Dunker, K. F., and Rabbat, B. G., "Performance of Prestressed
Concrete Highway Bridges in the United States - the First
40 Years," PCI JOURNAL, V. 37, No. 3, May-June 1992,
pp. 48-64.
5. Kirsch, U., "Optimum Design of Prestressed Plates," ASCE
Journal of the Structural Division, V. 99, No. ST6, June 1973,
pp. 1075-1090.
6. Wills, J., "A Mathematical Optimization Procedure and Its
Applications to the Design of Bridge Structures," Transport
and Road Research Laboratory, Report LR 555, Crowtorne,
Berkshire, England, 1973, pp. 1-28.
7. Mafi, M., and West, H. H., "Cost-Effective Short Span Bridge
Systems: A Selection Concept and an Optimization Procedure,
Second International Conference on Short and Medium Span
Bridges, V. 2, Ottawa, Ontario, Canada, 1986, pp. 117-131.
8. Naaman, Antoine, "Computer Program for Selection and Design
of Simple-Span Prestressed Concrete Highway Girders," PCI
JOURNAL, V.l7,No.l,January-February 1972,pp. 73-81.
9. Torres, G. Guzman Baron, Brotchie, J. F., and Cornell, C. A.,
"A Program for the Optimum Design of Prestressed Concrete
Highway Bridges," PCI JOURNAL, V. 11, No. 3, June 1966,
pp. 63-71.
10. Rabbat, B. G., and Russell, H. G., "Optimized Sections for
Pretensioned Concrete Bridge Girders in the United States,"
First International Conference on Short and Medium Span
Bridges, V. 2, Toronto, Ontario, Canada, 1982, pp. 97-111.
11. Cohn, M. Z., and Mac Rae, A. J., "Prestressing Optimization
and Design Practice," PCI JOURNAL, V. 29, No. 4, July-
August 1984, pp. 68-83.
12. Templeman, A. B., and Winterbottom, S. K., "Optimum De-
sign of Concrete Cellular Spine Beam Bridge Decks," Pro-
ceedings, Institution of Civil Engineers, Part 2, V. 59, 1975,
pp. 669-697.
13. Aguilar, R. J., Movassaghi, K., and Brewer, J. A., "Computer-
ized Optimization of Bridge Structures," Computers & Struc-
tures, V. 3, 1973, pp. 429-442.
14. Bond, D., "An Examination of the Automated Design of Pre-
stressed Concrete Bridge Decks by Computer," Proceedings,
Institution of Civil Engineers, Part 2, V. 59, 1975, pp. 669-697.
15. Bond, D., "Optimal Design of Continuous Prestressed Con-
crete Bridges," International Symposium on Nonlinearity and
Continuity in Prestressed Concrete (M. Z. Cohn, editor),
V. 3, University of Waterloo, Waterloo, Ontario, Canada,
1983, pp. 201-234.
16. Aziz, E. M., and Edwards, A. D., "Some Aspects of the
Economics of Continuous Prestressed Concrete Bridge Gird-
ers," The Structural Engineer, V. 44, No. 2, February 1966,
pp. 49-54.
17. Adin, M., Cohn, M. Z., and Pinto, M., "Comprehensive Opti-
mal Limit Design of Reinforced Concrete Bridges," Inter-
July-August 1993
national Symposium on Nonlinear Design of Concrete Struc-
tures, University of Waterloo, Waterloo, Ontario, Canada,
1979, pp. 349-377.
18. McDermott, J. F., Abrams, J. I., and Cohn, M. Z., "Some Re-
sults in the Optimization of Tall Building Systems," IABSE
Ninth Congress - Preliminary Report, Amsterdam, Nether-
lands, 1972, pp. 855-861.
19. McDermott, J. F., Abrams, J. I., and Cohn, M. Z., "Computer
Program for Selecting Structural Systems," ASCE Annual
Meeting, Houston, TX, Preprint 1863, October 1972.
20. Gellatly, R. A., and Dupree, D. M., "Examples of Computer-
Aided Optimal Design of Structures," IABSE Tenth Congress
-Introductory Report, Tokyo, 1976, pp. 77-105.
21. Kulka, F., and Lin, T.Y., "Comparative Study of Medium-
Span Box Girder Bridges With Other Precast Systems," First
International Conference on Short and Medium Span Bridges,
V. 1, Toronto, Ontario, Canada, 1982, pp. 81-94.
22. Lounis, Z., and Cohn, M. Z., "Optimal Design of Prestressed
Concrete Highway Bridge Girders," Proceedings, Third Inter-
national Symposium on Concrete Bridge Design, Washington,
D.C., March 1992 (to be published).
23. CPCI, Metric Design Manual, Second Edition, Canadian Pre-
stressed Concrete Institute, Ottawa, Ontario, Canada, 1987.
24. OHBDC-1983, Ontario Highway Bridge Design Code and
Commentary, Second Edition, Ministry of Transportation and
Communications, Downsview, Ontario, Canada, 1983.
25. OHBDC-1992, Ontario Highway Bridge Design Code Draft,
Third Edition, Ministry of Transportation and Communica-
tions, Downsview, Ontario, Canada, 1992.
26. PC/ Design Handbook - Precast and Prestressed Concrete,
Fourth Edition, Precast/Prestressed Concrete Institute,
Chicago, IL, 1992.
27. AASHTO, Standard Specifications for Highway Bridges, Fif-
teenth Edition, American Association of State Highway and
Transportation Officials, Washington, D.C., 1992.
28. Geren, K. L., Abdelkarim, A. M., and Tadros, M. K., "Precast/
Prestressed Concrete Bridge !-Girders: The Next Generation,"
Concrete International, V. 14, No.6, June 1992, pp. 25-28.
29. Bakht, B., Cheung, M. S., and Aziz, T. S., "Application of a
Simplified Method of Calculating Longitudinal Moments to
the Ontario Bridge Design Code," Canadian Journal of Civil
Engineering, March 1979, pp. 36-50.
30. Bakht, B., and Jaeger, L. G., Bridge Analysis Simplified,
McGraw-Hill Book Company, New York, NY, 1985.
31. Neville, A. M., Dilger, W. H., and Brooks, J. J., Creep of
Plain and Structural Concrete, Construction Press, 1983.
32. Freyermuth, C. L., "Design of Continuous Highway Bridges
With Precast, Prestressed Concrete Girders," PCI JOURNAL,
V. 14, No.2, Aprill969, pp. 14-39.
33. ACI Committee 209, "Prediction of Creep, Shrinkage and
Temperature Effects in Concrete Structures," ACI Manual of
Concrete Practice, Part 1, American Concrete Institute, De-
troit, MI, 1989.
341 Murtagh, B. A., and Saunders, M.A., "MINOS/Augmented,"
User's Manual, Department of Operations Research, Stanford
University, Stanford, CA, Technical Report SOL 80-9, 1980.
35. Brook, A., Kendruck, D., and Meeraus, A., "GAMS- Gen-
eral Algebraic Modeling System," User's Guide, Scientific
Press, 1988.
77
 APPENDIX A- NOTATION
The following symbols are used in this paper in addition n. = number of critical span sections
to the standard ACI and PCI notation: R1 = ratio of OHBDC and AASHTO HS20 truck
A = girder cross-sectional area (lane) load moments
Ccs = concrete cost in slab and diaphragms R2 = ratio of OHBDC and AASHTO HS20 effective
cdj = cost of deck joints design moments
Cg = cost of precast girder per length S = girder spacing
Cpc = unit cost of positive moment connection at S' = slab overhang
piers Sb, Sbc = section moduli with regard to bottom fibers of
c. = non-prestressed steel unit cost girder and composite sections
cbot = bottom concrete cover sup = superstructure cost
crop = top concrete cover sub = substructure cost
D d = design load distribution factor t = slab thickness
Dv Dy = bridge flexural rigidity in transverse and longi- t0 = concrete age at loading
tudinal directions, respectively t1 = concrete age at which continuity is achieved
Dxy, Dyx = bridge torsional rigidity in transverse and lon- Vcs = volume of concrete in slab and diaphragms
gitudinal directions, respectively Vpc = volume of positive moment reinforcement
D 1, D 2 = bridge coupling rigidity in transverse and lon- W = bridge width
gitudinal directions, respectively w. = weight of non-prestressed steel in slab and
fsP fsc = allowable tensile and compressive stresses at diaphragms
service, respectively Wsn = weight of negative moment steel in slab at
ftp frc = allowable tensile and compressive stresses at piers
transfer, respectively wsp = weight of positive moment steel in girders
ee, ec = tendon eccentricities at end and midspan sec- y = distance between centroids of girder and com-
tions, repectively posite sections
I = impact factor Ybc = distance from centroid of composite section to
lg, lgc = moments of inertia of girder and composite bottom fiber
sections, respectively y1 = distance from centroid of girder section to top
L = total bridge length fiber
c = span length Z = superstructure cost per deck area
£* = equivalent span length for live load moment a = bridge torsional parameter
computation in continuous girders f3 = ratio of initial and effective prestressing forces
M, Mv+L = total service moment 81 = ratio of total bridge costs of 2-C to 3-C config-
Ma = asphalt pavement weight moment urations
Mcsn = moment at support n due to combined effects 8z = ratio of substructure to superstructure costs for
of creep and shrinkage 2-C configuration
Mvn = support n restraint moment due to dead load ~ = ratio of substructure costs of 3-C to 2-C config-
Mg = girder weight moment urations
ML = live load moment ¢ = flexural strength reduction factor
Mn = nominal resisting moment; moment at support X = aging coefficient
n if structure is continuous and monolithically .A.o = ratio of ultimate resisting moment to service
cast in a single stage moment
Mpn = support n restraint moment due to prestressing v (t, t0 ) = creep coefficient (ratio of creep strain to initial
M 8 = slab weight moment elastic strain)
Mshn = support n restraint moment due to shrinkage vu = ultimate creep factor
Mu = ultimate load moment m = total reinforcement index
M~ = moment induced by creep at support n acP acs = concrete stresses at transfer and service, re-
n = number of girders spectively
nP = number of piers 8 = bridge flexural parameter

78 PCI JOURNAL