LONG-TERM BEHAVIOR OF CONTINUOUS PRECAST CONCRETE GIRDER

BRIDGE MODEL
By Antonio R. Mar,1 Member, ASCE, and Manuel Valdes2

ABSTRACT: Continuous concrete box girder bridges composed of precast reinforced and prestressed concrete
beams with a U cross section and a cast-in-place top slab are frequently used for medium spans due to their
competitiveness. The service behavior of such bridges is very much influenced by their segmental construction,
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

due to time-dependent materials behavior that makes it difficult to accurately predict the stresses, strains, and
deflections at long term. A 1:2 scale model of a two-span continuous bridge was tested in order to study its
behavior during the construction process and under permanent loads. Time-dependent concrete properties, as
well as support reactions, deflections, and strains in concrete and steel, were measured for 500 days. Important
time-dependent redistributions of stresses and internal forces throughout the bridge were also measured. The
test results were compared with analytical predictions obtained by means of a numerical model developed for
the nonlinear and time-dependent analysis of segmentally erected, reinforced and prestressed concrete structures.
Generally good agreement was obtained, showing the adequacy of the model to reproduce the structural effects
of complex interactive time-dependent phenomena.

INTRODUCTION deflections, stresses, and crack width, as required to satisfy the

The use of high-performance building materials and the de- serviceability limit states, and therefore creating uncertainties
velopment of powerful means for lifting and transportation, at the design stage.
together with an adequate combination of precast and cast-in-
place, pretensioning and posttensioning techniques, have im- OBJECTIVES OF PRESENT INVESTIGATION
pelled the development of continuous precast concrete box Although much progress has been made in todays bridge
girder bridges, specially suitable for medium spans. engineering, segmentally erected, prestressed concrete bridges
A typical bridge consists of precast reinforced, pretensioned often exhibit excessive long-term deflections, as reported by
or posttensioned-in-factory beams with a U cross section over Vitek in 1997. In addition, very little experimental work has
which a cast-in-place, continuous slab is poured in several been reported on the time-dependent behavior of precast con-
steps. The beams are supported over the bridge piers and/or tinuous concrete bridges such as those described above. The
the end cantilevers of other previously placed beams, the joint use of such bridges is increasing very quickly; therefore, a
between them being initially a temporary hinge. In addition, better understanding of their complex behavior is needed to
longitudinal continuity is provided to the structure by means avoid similar problems in the future. Furthermore, reliable an-
of posttensioning bars or tendons, after sealing the joint by alytical models, design, and construction recommendations are
injecting a high-strength mortar between the end diaphragms desirable. For this purpose, analytical and experimental re-
of the beams. Depending on the span length, the precast beams search work is currently under development at the Universitat
can be either reinforced, pretensioned, or posttensioned in fac- Politecnica de Catalunya in Barcelona, Spain.
tory. This solution allows building, without important addi- Many analytical models have been developed to predict the
tional cost, of bridge decks with multiple box sections, vari- nonlinear and time-dependent structural response of concrete
able depth, or curved-in-plan beams, which are aesthetically
comparable to those constructed in situ, as shown in Fig. 1.
Creep and shrinkage of concrete and relaxation of pre-
stressed steel, combined with the segmental construction pro-
cess, lead to time-dependent deflections and redistributions of
stresses and internal forces. Also, if the beam or slab is not
prestressed over the piers, cracking occurs due to negative
bending moments, affecting also the state of stresses, strains,
and deflections. In fact, in this kind of structure, the time-
dependent redistributions of internal forces can be very dif-
ferent from those expected in a typical continuous postten-
sioned concrete bridge, due to cracking of the reinforced
concrete beams. Cracks can also appear at the top slab, even
at center span, due to restrained shrinkage of the concrete slab
when it is poured after the beam is hardened. All these phe-
nomena interact, making it difficult to accurately estimate the
1
Prof., Dr. of Civ. Engrg., Universitat Politecnica de Catalunya, School
of Civ. Engrg., Jordi Girona 1-3, 08034 Barcelona, Spain. E-mail:
[email protected]
2
Lect., Dr. of Civ. Engrg., Universitat Politecnica de Catalunya, School
of Civ. Engrg., Jordi Girona 1-3, 08034 Barcelona, Spain. E-mail: dvaldes
@etseccpb.upc.es
Note. Discussion open until July 1, 2000. To extend the closing date
one month, a written request must be filed with the ASCE Manager of
Journals. The manuscript for this paper was submitted for review and
possible publication on June 2, 1998. This paper is part of the Journal
of Bridge Engineering, Vol. 5, No. 1, February, 2000. ASCE, ISSN FIG. 1. Continuous Precast Concrete Girder Bridge, Curved in
1084-0702/00/0001-00220030/$8.00 $.50 per page. Paper No. 18436. Plan

22 / JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000

J. Bridge Eng., 2000, 5(1): 22-30

 frames (Mar 1984; Elbadry 1988; Kang 1989; Abbas and uniformly distributed load of 4.0 kN/m2; and a truck load,
Scordelis 1993). However, less effort has been made to sim- consisting of three-point loads of 25 kN each separated 1.5 m
ulate the effects on the structural response of segmental con- longitudinally. Such a load has been obtained by reducing, for
struction in both longitudinal and transverse directions (Cruz reasons of scale, to 1/4 the load defined by the Spanish code
et al. 1998; Murcia and Herkenhoff 1994; Mar et al. 1996). of practice for bridge design. A 20% increase in the positive
In the present investigation, an analytical model has been de- bending moment at both spans was considered to account for
veloped that takes into account most of the changes that take the effects of cracking over the support and of behavior of
place during construction of bridge structures, as well as their time-dependent materials. Prestressing losses due to creep,
interaction with the nonlinear and time-dependent behavior of shrinkage, and relaxation of steel were also considered at the
concrete and steel. design stage, as well as the evolution of the longitudinal and
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

Much experimental research has been done, both at the ma- transverse structural schemes during construction.
terial and structural levels, to characterize the physical phe-
nomena that take place and to calibrate the analytical models Geometry, Materials, and Construction Process
developed. However, laboratory tests are usually made on sim-
ple specimens at a reduced scale. Although their results pro- The model consisted of a 24 m long continuous box girder
vide reliable information on the behavior of the tested struc- bridge of two spans of 9 and 15 m, respectively. It was com-
ture, sometimes they do not truthfully represent the behavior posed of two precast concrete beams with a U cross section
of actual structures. This is due to scale effects, differences in of 56 cm in depth, a bottom slab 16 cm thick and 1.7 m wide,
the construction procedures or environmental conditions, or and 1/12.5 inclined webs 15 cm thick. The smallest beam, of
the specificity of the test carried out. 10.5 m in length and made of reinforced concrete, covered the
Measurements made on actual constructed bridges (Vitek shorter span and had a 1.5 m cantilever. The other beam, 13.5
1997) provide very valuable but limited information that usu- m in length, was posttensioned with six tendons of five strands
ally is not enough to characterize the behavior of the whole of 15 mm (0.6 in.) diameter each, placed at the bottom slab
structure. It is usual, for example, to measure displacements with a straight layout. This beam was supported at the end of
and strains at a reduced number of points, but not reactions at the reinforced concrete beam and at the other bridge end. At
the supports or prestressing forces along the tendon. Accurate each of the three supports of the bridge, elastomeric pads were
information about the time-dependent concrete properties is placed on top of load cells.
usually not available. In addition, the randomness of the results A 13 cm deep and 4.0 m wide reinforced concrete cast-in-
can be considerable, due to the randomness in material prop- place slab, connected to the precast beam by means of rein-
erties, geometry, and distribution of loads. forcement crossing the joint, was poured in three steps, gen-
For all these reasons, a large-scale laboratory model of a erating the box girder section of 68 cm total depth, as shown
bridge was designed, constructed, and tested (Valdes 1997) in Fig. 2. The amount and position of the mild reinforcement
using the same materials and construction process as in actual of the beams and the slab are shown in Table 1.
bridges, but carrying out exhaustive control of the most rele- The characteristic compressive strength of the concrete at
vant parameters influencing the structural behavior. The ana- 28 days was 50 MPa for the beams and 30 MPa for the slab.
lytical models developed were calibrated using the test results Reinforcing and prestressing steel strengths were 500 and
and used to study the influence of several parameters in order 1,860 MPa, respectively. The most important steps in the con-
to provide design and construction recommendations. In this struction process are shown in Table 2 and illustrated in Fig.
paper, the laboratory tests are described and their main results 3. Figs. 4 and 5 show the bridge during construction (at step
compared with those obtained numerically. 5) and under permanent loads, respectively.

LABORATORY TEST Instrumentation

Design Criteria of Model The bridge was instrumented from the very beginning of
The total length of the model was limited by the length of the fabrication of each element, so that the whole construction
the laboratory test floor (24 m). The structural scheme, cross- process was documented. About 120 measure points were con-
section type, material properties, and construction process trolled to know the reactions, displacements, and strains. In
were chosen to accurately reproduce those of a typical bridge. addition, exhaustive control of the geometry, amount, and po-
In addition, the following considerations were taken into ac- sition of the reinforcement, posttensioning forces, tendon lay-
count: outs, and concrete thickness was carried out. Consequently,
the load distribution due to self-weight and the mechanical
1. The dimensions of the cross section, cover, and separa- properties of the cross sections were known with accuracy.
tion between reinforcing bars should be large enough to Reactions and subsequent bending moments were known at
allow the use of normal concrete (maximum aggregate any time, thanks to load cells placed at each of the six support
size, 18 mm). Moreover, the smaller cross-section di- points of the bridge. The instantaneous and permanent loads
mensions of the model should not modify the time-de-
pendent deformations, due to creep and shrinkage of the
concrete, with respect to the prototype. For that purpose,
1/2-scale factor was chosen for the model.
2. The stresses and strains produced by the permanent loads
in the model should be similar to those in a prototype to
real scale. In order to satisfy such a condition, the value
of the load per unit of surface should be maintained. For
that purpose, sand was placed inside the U beam, and
hollow-core concrete slabs were placed on top of the
bridge as an additional dead load.

The actions considered for the design of the model were FIG. 2. Longitudinal and Cross Section of Laboratory Model
self weight; dead load (2.75 kN/m2); live load, consisting of a (1:2 Scale)

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000 / 23

J. Bridge Eng., 2000, 5(1): 22-30

 TABLE 1. Mild Reinforcement Layout in Beams and Slab
Bars
Distance to bottom As Diameter
Segment From To Layer (m) (mm2) Quantity (mm)
(1) (2) (3) (4) (5) (6) (7) (8)
1 (RB) x=0m x=7m 1 0.04 2,614.3 13 16
2 0.10 2,614.3 13 16
3 0.21 452.4 4 12
4 0.43 452.4 4 12
2 (RB) x=7m x=9m 1 0.04 2,614.3 13 16
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

2 0.10 2,614.3 13 16
3 0.21 452.4 4 12
4 0.43 452.4 4 12
5 0.47 2,513.6 8 20
3 (RB) x=9m x = 10.5 m 1 0.04 904.8 8 12
2 0.10 904.8 8 12
3 0.21 452.4 4 12
4 0.43 452.4 4 12
5 0.47 2,513.6 8 20
4 (PB) x = 10.5 m x = 19 m 1 0.03 904.8 8 12
2 0.08 904.8 8 12
3 0.23 452.4 4 12
4 0.53 452.4 4 12
5 (PB) x = 19 m x = 24 m 1 0.03 904.8 8 12
2 0.08 904.8 8 12
3 0.23 452.4 4 12
4 0.38 452.4 4 12
5 0.53 452.4 4 12
6 (S) x=0m x = 3.3 m 1 0.64 3,418.7 17 16
2 0.60 1,922.7 17 12
7 (S) x = 10 m x = 12 m 1 0.64 3,418.7 17 16
2 0.60 1,922.7 17 12
8 (S) x = 3.3 m x = 10 m 1 0.64 8,245.0 41 16
2 0.60 1,922.7 17 12
9 (S) x = 12 m x = 24 m 1 0.64 1,922.7 17 12
2 0.60 1,922.7 17 12
Note: RB: Reinforced beam; PB: prestressed beam; S: slab.

TABLE 2. Construction Steps

Time (Days)
Absolute Relative
Step Operations (t) (t tp)
(1) (2) (3) (4)
1 Pouring of concrete at both beams 0 7
2 Erection of reinforced concrete (RC) beam 5 2
3 Posttensioning of longest beam 7a 0
4 Erection of prestressed concrete (PC) beam 9 2
5 Placement of sand for scale effects 24 17
6 Placement of formwork and reinforcement of 31 24
top slab
7 Pouring of top slab over RC beam (from x = 0 38 31
to 10 m
8 Injection of joint with high-strength mortar 40 33 FIG. 3. Construction Process
9 Posttensioning of joint by means of two 25 mm 42 35
diameter bars
10 Pouring of top slab over joint (from x = 10 to 43 36
12 m)
11 Pouring of top slab over PC beam (from x = 12 46 39 reference measure induced use of only mechanical extensom-
to 24 m) eters for delayed deflections.
12 Placement of first row of hollow core slabs for 79 72
scale effects To measure concrete and steel deformations, strain gauges
13 Placement of second row of hollow core slabs 80 73 were used, most of which were glued to reinforcing steel bars
to simulate dead load of the beam and slab at four critical sections: centers of both
14 Unloading of dead load 573 566
a
spans, intermediate support, and center of the posttensioned
Reference absolute time.
beam. Strain gauges were also glued to the tendons at the
center of the posttensioned beam, to several stirrups near sup-
ports, and to the continuity posttensioning bars. Embedded
applied to the concrete specimens to characterize the concrete gauges were placed in the top and bottom slabs at the center
properties were controlled by means of load cells. Instanta- of the posttensioned beam, as well as inside the concrete spec-
neous deflections were measured with both LVDTs and me- imens, to measure instantaneous and time-dependent concrete
chanical extensometers, including the camber of the postten- strains. LVDTs were also used in the concrete specimens to
sioned beam. The difficulties of the LVDTs to preserve the determine the deformation modulus.
24 / JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000

J. Bridge Eng., 2000, 5(1): 22-30

 Materials Tests Results
Two kinds of material tests, instantaneous and delayed, were
carried out to characterize the properties of each type of con-
crete used in the bridge. For that purpose, specimens were
used subject to the same environmental conditions as the
bridge. The instantaneous tests provided information about the
compressive strength and the deformation modulus at various
concrete ages, as shown in Table 3.
The values obtained for the mechanical concrete properties
at short term were inside the expected range. However, the
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

compressive strength and deformation modulus obtained at

long term were slightly lower than at short term, probably due
to the dry environment, which did not provide enough mois-
ture for cement hydration, or due to shrinkage microcracks in
concrete specimens (Larrard and Atcin 1993).
The delayed tests helped to characterize the shrinkage and
creep strains throughout approximately 700 days. Results of
creep and shrinkage tests of specimens of the beam concrete
subject to permanent loads at the ages of 10 and 100 days are
shown in Table 4. For the first test, both the CEB and ACI
models fit the experimental results well, while for the speci-
mens loaded at 100 days, both codes overestimate creep
strains. Shrinkage strains are between 500 and 600 m/m.
They are slightly underestimated by the CEB and overesti-
mated by the ACI.
FIG. 4. Placement of Sand for Scale Effects Creep strains of the slab concrete were much larger than the
values predicted by any code. Shrinkage of the slab concrete
was also larger than that of the beam concrete, probably due
to differences in the concrete water/cement ratio and the ef-
fective thickness of the beam and the slab.
FINITE-ELEMENT ANALYSIS
Structure Idealization: Geometry, Loads, and
Construction Process
A general step-by-step procedure for the nonlinear and time-
dependent analysis of reinforced concrete, prestressed con-
crete, and segmentally erected, composite steel and concrete
3D frame structures has been used to simulate the structural
behavior and to predict the test results (Mar 1998). The model
combines the features of previous models developed by the
first author and other researchers (Mar 1984; Cruz et al.
FIG. 5. Laboratory Model under Permanent Loads 1998).

TABLE 3. Short-Term Concrete Characterization Tests

Instantaneous Test Results (Mean Values) (MPa)
Concrete
Test location 7 days 10 days 28 days 60 days 90 days 150 days 490 days
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Deformation modulus (Ec,t) Beam 36,900 40,800 40,700 38,800 42,100
Slab 26,800 27,100 25,200
Compression strength ( fcm,t ) Beam 55.2 60.6 66.9 64.1 61.0
Slab 29.4 35.5 35.7

TABLE 4. Long-Term Concrete Characterization Test

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000 / 25

J. Bridge Eng., 2000, 5(1): 22-30

 The nonlinear materials properties, such as cracking and tion allows performing a complete step-by-step analysis with-
yielding, the structural effects of the delayed deformations, and out keeping the whole stress history of each filament at each
the second-order effects are taken into account in the structural section of control. The parameters governing the expansion
analysis under loads and imposed deformations. The structural series are fitted to the corrected CEB-FIP Model Code 1990,
response during the construction process and along the struc- as explained above.
ture service life can be traced throughout the elastic, cracked,
and ultimate load level, providing valuable information about Nonlinear Time-Dependent Analysis Strategy
the influence of the construction sequence on the short- and
long-term service behavior and carrying capacity of the struc- The structural analysis strategy consists of a time-step-by-
ture. Complex phenomena such as delayed cracking or creep step procedure in which the actual construction process and
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

buckling can also be monitored. load history of the structure can be reproduced. The time do-
The analytical tool developed is based on the displacement main is divided into a discrete number of time intervals, and
formulation of the finite-element method (FEM) and incor- a time-step-forward integration is performed in which incre-
porates the Hermitian two-node beam elements, prestressing ments of displacements, strains, and other structural quantities
tendons with arbitrary layout, and prestressed cables to rep- are successively added to the previous totals as we march for-
resent the stays in cable-stayed structures. Several kinds of ward in the time domain.
connections can be defined between ends of beams, such as At each time step, the structure is analyzed under the ex-
generalized hinges and rigid links with finite length. Special ternal applied loads and imposed deformations originated dur-
elements to represent the shores as supports acting only in ing the previous time interval, due to creep, shrinkage, tem-
compression are also defined. perature variations, or construction loads. A direct stiffness
The cross section of a beam element is divided into sub- FEM based on the displacement formulation is used, in which
sections, and each subsection into filaments. A explicit 1D con- the resulting equilibrium equations will be nonlinear in order
stitutive equation is considered at each filament. The concept to be valid for the current state of material properties and
of subsection allows adequate representation of the construc- geometry. A Total Lagrangian formulation is adopted to take
tion of elements in different steps. Several materials with dif- into account the geometric nonlinear effects. Several incre-
ferent rheological behavior, such as structural steel and con- mental and iterative solution strategies to solve the nonlinear
crete or precast and cast-in-place concrete, can be considered. problem have been implemented in a computer program called
Variations with time of the mechanical properties of different CONS, which is an extension of program PCF3D, initially
parts of a cross section can also be simulated in a consistent developed without segmental construction (Mar 1984).
way. Variations of the environmental conditions (temperature,
humidity, exposed surface), which are due to changes in the STRUCTURAL TEST RESULTS AND ANALYTICAL
cross-section geometry that affect the time-dependent behavior PREDICTIONS
of the concrete, are also taken into account in the model.
The model can deal with pretensioned and posttensioned From the large amount of information obtained in this test,
bonded and unbonded tendons. Prestressing losses are ob- only a few representative values will be presented. In Figs. 6
tained throughout the structural analysis. Concentrated and 12, the time scale used is t-tp, tp being the instant of prestress-
distributed loads, equivalent prestressing loads, imposed dis- ing and t the absolute time counted from the instant of fabri-
placements at supports, temperature variations, and time- cation of the beams.
dependent deformations due to creep and shrinkage of con-
crete, as well as the relaxation of prestressing steel, are taken Deformations
into consideration.
The present model has been designed to simulate in a very Strains will be shown at two representative sections: the one
general way all kinds of structural changes that may occur placed at the center of the prestressed beam (CPB), and the
during the construction process and throughout the entire one placed at the central support (CS).
structural service life. Changes in the longitudinal or trans- Fig. 6 shows the strains at the top and bottom reinforcement
verse geometry of the structure can be taken into account by bars at the CPB section during construction and under per-
adding or removing elements or subsections of the cross sec- manent loads. Considerable increments of strains due to creep
tion. Temporary prestressing tendons, as well as stays and sup- and shrinkage of the slab and the beam can be observed in the
ports, can be placed, stressed, unstressed, or removed at any top reinforcement, while the strain at the bottom reinforcement
time, and external support conditions and internal constraints remains practically constant with time. Good agreement be-
and connections between elements can also be modified at any tween experimental and analytically predicted strains in this
time. section is observed, having contributed the absence of cracking
to the accuracy of the predictions. The only significant differ-
Modeling of Materials Behavior ence is the evolution of the strain at the bottom reinforcement
at the CPB section just after the posttensioning, probably due
A multilinear stress-strain relationship with strain hardening to differences between the real and simulated time-dependent
is used both for reinforcing and prestressing steel. The instan- behavior at very early ages.
taneous nonlinear behavior of concrete has been considered by Fig. 7 shows the strains at the top and bottom reinforce-
means of a parabolic model that considers a post-peak de- ments of the slab, at the CPB section. The increment of strain
scending branch and load reversal. A smeared crack approach under permanent loads is remarkable, as it happens at the top
is used with tension stiffening. reinforcement of the beam. It can be seen that the strains at
Segmental construction processes require accurate simula- the top reinforcement are well predicted by the model and are
tion of the time-dependent behavior of the materials, especially considerably bigger than those at the bottom reinforcement of
when prestressing or loading at early ages. For that purpose, the slab, due to the restraint imposed by the previously hard-
the parameters of the CEB model have been corrected to fit ened beam. Such an effect is only partially captured by the
the curves of creep and shrinkage obtained from the delayed model because no gradient of shrinkage and creep has been
tests over concrete specimens. considered along the slab depth, due to environmental condi-
In addition, in the present model, a Dirichlet expansion se- tions.
ries is used to represent the creep function. Such approxima- Fig. 8 shows the strains at the top and bottom reinforce-
26 / JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000

J. Bridge Eng., 2000, 5(1): 22-30

Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

FIG. 7. Strain at Reinforcing Bars of Slab at CPB Section

FIG. 6. Strain at Reinforcing Bars, at CPB Section

ments, respectively, of the beam at the CS section. The top

reinforcement [Fig. 8(a)] is placed at the tensile zone, where
cracking occurs during the construction process. In this case,
after an instantaneous increment of strain due to cracking, the
reinforcement slightly increases its strain with time, due to the
decrease of the tension stiffening effect with time. The bottom
reinforcement [Fig. 8(b)], which is placed at the compression
zone, considerably increases its strain with time.
Fig. 9 shows the development of the strain with time at the
reinforcement placed on the top slab of the same section. The
slab at this section cracks during construction, just after the
concrete slab over the longest span is placed. A remarkable
increment of tensile stresses with time is observed, since ten-
sion-stiffening effects are important.
It can be concluded that considerable complexity exists to
capture the strain distributions along the section depth, due to
the segmental construction. In fact, the top slab is poured over
a beam that is already cracked. In spite of this, the model has
captured the general sectional behavior, including the instant
of cracking and the evolution of strains at the compressive and
cracked areas. The main differences observed are probably due
to the difficulties of accurately reproducing the time-dependent
behavior of the cracked area, such as the evolution of the ten-
sion stiffening and the shear lag effects in the tensile zone. FIG. 8. Strain at Reinforcing Bars of Beam at CS Section

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000 / 27

J. Bridge Eng., 2000, 5(1): 22-30

Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

FIG. 9. Strain at Reinforcing Bars of Slab at CS Section FIG. 12. Evolution of Deflection under Permanent Load and
Unloading

Support Reactions
Fig. 10 shows the experimentally obtained support reactions
from the beginning of the construction process. A very small
decrease of central reactions is observed after providing struc-
tural continuity. Therefore, a small redistribution of forces with
time takes place, with a decrease of the negative bending mo-
ment and an increase of the positive moments at the centers
of both spans, as shown in Fig. 11.
The comparison of the experimentally measured reactions
with those that were obtained numerically shows a generally
good agreement, especially during construction, although the
variation of the reaction under permanent load is slightly big-
ger in the test than in the numerical predictions. Bending mo-
ment redistributions of 12.6% of the maximum negative mo-
ment and 5% of the maximum positive moment were drawn
from the experimentally measured reactions.

Deflections
Fig. 12 shows the analytical and measured deflections at the
FIG. 10. Time Evolution of Total Load and Support Reactions center of the prestressed beam, starting from the placement of
permanent loads in the slab. Creep effects are remarkable, pro-
ducing a delayed deflection of 18.1 mm (which is about 1/825
of the span length).
The predicted maximum delayed deflection was 25%
smaller than the experimental one. The comparison has been
made by taking into account only the delayed deflection be-
cause of the difficulties in accurately measuring this parameter
during the construction process. However, the differences in
the prestressing camber and the instantaneous deflection due
to the application of the permanent loads were only about 5%
with respect to the measured values.
The disagreement in the delayed deflections is probably due
to the existence of inclined cracks produced by the shear-flex-
ure interaction (not taken into account in the numerical model)
and to the difficulties in accurately simulating the time evo-
lution of tension stiffening.
Shear lag combined with creep effects have a big influence
on long-term deflections, especially in box girder bridges
(Kristek and Bazant 1987) such as the one tested, due to the
large ratio between the slab width and web thickness. Since a
1D beam model cannot capture such an effect, an effective
width of the top slab was adopted of constant value along the
FIG. 11. Bending Moment Redistribution under Permanent bridge length. The value of the effective width, obtained ini-
Loads tially from a finite-element linear elastic analysis using shell
28 / JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000

J. Bridge Eng., 2000, 5(1): 22-30

 elements, was 1.40 m on top of each web. Such a value, which the tested bridge, where the maximum negative moment
represents 70% of the total slab width, is very similar to the decreases 12.6% and the maximum positive moment in-
usual values proposed by some international codes (CEB-FIP creases 5% after 500 days. Such small redistributions of
Model Code 90). moments are probably due to a combination of the span
However, the effective width had to be reduced to 1.05 m lengths ratio, position of the joint, permanent load level,
(about 52%) of the total slab width on top of each web in and amount of prestressing force for this particular struc-
order to fit the experimentally measured strains. This reduction ture.
is probably due to transverse cracking of the slab, produced 5. When flexural cracking takes place, the increment of
by restrained shrinkage of the concrete. Such cracks, which strain under permanent loads is smaller at the tensile re-
were observed in the model, reduce the contribution of the top inforcement than at the compressive reinforcement. If
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

slab to the structural stiffness. cracking is very important and tension stiffening is small,
The measured prestressing force at the continuity bars the tensile reinforcement remains practically constant
crossing the joint remained practically constant with time. No with time, as observed in the test at the upper reinforce-
joint opening was observed during construction or after 500 ment of the beam over the central support. However,
days. when tension stiffening is important, as happens in the
The loss of prestressing force at tendon 1 at the central slab, reinforcement incrementally increases its strain with
section of the longest beam, estimated by means of the nu- time, due to progressive cracking and loss of tension
merical model, is 17% of the initial posttensioning force, stiffening.
which is a reasonable amount for this particular kind of bridge. 6. The analytical model has been shown to adequately sim-
This value is consistent with the loss obtained from the mea- ulate the complex time-dependent behavior of the struc-
sured tendon strain multiplied by the tendon modulus of elas- ture. Evolution of strains, stresses, internal forces, reac-
ticity, after adding the expected value of the relaxation loss. tions, deflections, and cracks was predicted with
reasonable accuracy. Therefore, it can be concluded that
CONCLUSIONS the analytical model can be used as a powerful tool at
the design stage or during the erection of the structure
A large-scale model of a continuous precast concrete box in order to optimize the construction process.
girder bridge has been tested in order to characterize its be- 7. The top concrete slab shows transverse shrinkage cracks,
havior during construction and under permanent loads after even at midspan, due to the restraint produced by the
500 days. Evolution of strains in concrete and steel, reactions, beam and by the longitudinal slab reinforcement. Such
and deflections have been measured. Time-dependent redistri- cracks reduce the contribution of the top slab to the sec-
bution of stresses, reactions, and internal forces have been tional stiffness. For the tested structure, the effective
identified and quantified. In addition, analytical predictions of width used in the analytical model necessary to repro-
the nonlinear and time-dependent behavior of the tested bridge duce the test deflections and strains proved to be 52%
have been made and their results compared with those exper- of the total slab width.
imentally measured. The following conclusions can be drawn, 8. Future experimental and analytical research work is
related to the objectives of the present research work: needed to better quantify the contribution of the top slab
to the structural stiffness, taking into account the inter-
1. The laboratory tests have captured the most important action between creep, shrinkage, cracking, and shear lag
phenomena that take place at the material and structural effects.
level in a large-scale model of a real segmentally con-
structed concrete bridge structure. Exhaustive control of ACKNOWLEDGMENTS
the geometry, material properties, environmental condi-
tions, load distribution, and construction sequence has The present investigation has been developed under the Research
Project PB96-0954 funded by the Spanish Ministry of Education. The
been carried out. Most of the variables have been mea- Spanish prefabrication company ALVISA, the construction company
sured at several points of control and the results checked COPCISA, and the quality control company COTCA have also cofi-
among them and over time. Therefore, it can be said that nanced the laboratory test and participated actively in the design and
reliable qualitative and quantitative information has been construction of the structure. The writers want also to express their grat-
obtained about the most important variables influencing itude to their colleagues in the Structural Technology Laboratory of the
the structural behavior. Technical University of Catalonia for their help.
2. The segmental construction process has an enormous in-
APPENDIX I. REFERENCES
fluence on the short- and long-term structural response.
At a given section, the stresses at the end of the con- Abbas, S., and Scordelis, A. C. (1993). Nonlinear geometric, material
struction process can be very different from those that and time-dependent analysis of segmentally erected three-dimensional
cable stayed bridges. Rep. UCB/SEMM-93/09, University of Califor-
would have been produced by a simultaneous construc- nia at Berkeley, Calif.
tion and also very different from those existing at long Cruz, P. J. S., Mar, A. R., and Roca, P. (1998). Nonlinear time-depen-
term. This is due to the evolution of the structural con- dent analysis of segmentally constructed structures. J. Struct. Engrg.,
figuration, the load history, and the effects of creep, ASCE, 124(3), 278287.
shrinkage, and relaxation. Elbadry, M. M. (1988). Serviceability of reinforced concrete struc-
3. Redistribution of stresses between concrete and steel can tures, PhD thesis, University of Calgary, Canada.
Kang, Y. (1989). SPCFRAME: Computer program for nonlinear seg-
be very important. For the tested structure, a 45% incre- mental analysis of planar prestressed concrete frames. Rep. UCB/
ment in the stress at the top reinforcement of the pre- SEMN-89-07, University of California at Berkeley, Calif.
stressed beam occurs at the center of the longest span, Krstek, V., and Bazant, Z. P. (1987). Shear lag effect and uncertainty
reaching a total stress of 400 MPa after 500 days. The in concrete box girder creep. J. Struct. Engrg., ASCE, 113(3), 557
reinforcing bars placed at the top slab showed a stress 574.
increment of 175%, reaching a total value of 300 MPa. Larrard, F., and Atcin, P. C. (1993). Apparent strength retrogression of
silica fume concrete. ACI Mat. J., 90(6), 581585.
The increment of stress is already noticeable at an early Mar, A. R. (1984). Nonlinear geometric, material and time dependent
age of the concrete, probably due to shrinkage strain. analysis of three dimensional reinforced and prestressed concrete
4. Redistributions of reactions and internal forces have also frames. Rep. No. 1, UCB/SESM-84/12, University of California,
been measured, although they are not very important for Berkeley, Calif.

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000 / 29

J. Bridge Eng., 2000, 5(1): 22-30

 Mar, A. R. (1998). Nonlinear and time dependent analysis of segmen- viceability Models, Progress Rep., Comite Euro-International du Beton
tally erected three dimensional concrete frames. Res. Rep. Dept. of (CEB), Lausanne, Switzerland.
Constr. Engrg., Universitat Politecnica de Catalunya, Barcelona, Spain.
Mar, A. R., Montaner, J., and Valero, I. (1996). Evaluacion de estados APPENDIX II. NOTATION
tensodeformacionales y flechas en puentes isostaticos prefabricados de
hormigon pretensado. Hormigon y Acero, No. 202, 2558 (in Span- The following symbols are used in this paper:
ish).
Murcia, J., and Herkenhoff, L. (1994). Analisis en el tiempo de puentes As = area of reinforcing steel;
continuos de hormigon construidos a partir de elementos prefabrica- CPB = central section of prestressed beam;
dos. Hormigon y Acero, No. 192, 5571 (in Spanish). CS = central support section;
Roca, P., and Mar, A. R. (1993). Numerical treatment of prestressing Ect = deformation modulus of concrete;
tendons in the nonlinear analysis of prestressed concrete structures. fcm,t = mean compressive strength of concrete at instant t;
Downloaded from ascelibrary.org by Khulna University of Engineering & Technology on 07/25/17. Copyright ASCE. For personal use only; all rights reserved.

Comp. and Struct., 46(5), 905916. PB = prestressed beam;

Valdes, M. (1997). Comportamiento durante construccion y bajo cargas
RB = reinforced beam;
permanentes de puentes continuos prefabricados de hormigon, PhD
thesis, Escuela Tecnica Superior de Ingenieros de Caminos, Canales y
S = slab;
Puertos, Barcelona, Spain. t = time;
Vitek, J. L. (1997). Long-term deflections of large prestressed concrete t0 = instant of loading;
bridges. Bull. No. 235, Serviceability Models, Progress Rep., Comite t tp = time elapsed after prestressing;
Euro-International du Beton (CEB), Lausanne, Switzerland. cst = shrinkage strain of concrete; and
Vitek, J. L. (1997). Selected monitored bridges. Bull. No. 235, Ser- (t, t0) = concrete creep function.

30 / JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 2000

J. Bridge Eng., 2000, 5(1): 22-30