Computers and Structures 82 (2004) 13–23

www.elsevier.com/locate/compstruc

Effects of superstructure flexibility on strength

of reinforced concrete bridge decks
S. Zhou, D.C. Rizos *, M.F. Petrou
Department of Civil and Environmental Engineering, University of South Carolina, 300 Main Street, Columbia, SC 29208, USA
Received 31 December 2002; accepted 26 August 2003

Abstract
This research studies the impact of the relative rigidity between the concrete bridge deck and the remaining structural
components of the bridge superstructure on the behavior of the concrete deck. The study uses non-linear 3-D FEM
models, which are developed using ANSYS 5.7 software package. Experimental data from one-span non-composite
bridge superstructure are used to validate and calibrate the proposed FEM models. A series of parametric studies is
conducted with respect to three parameters: (a) composite action, (b) slenderness ratio, and (c) presence of diaphragms.
The analysis results are discussed in detail and conclusions on the behavior of the bridge deck are presented.

2003 Elsevier Ltd. All rights reserved.

Keywords: Reinforced concrete bridge deck; Superstructure flexibility; FEM; ANSYS

1. Introduction the effects of superstructure flexibility on slab perfor-

mance. Burke [1] stated that bridge superstructure flex-
Durability of the concrete bridge deck is directly re- ibility had adverse influence on the integrity and
lated to cracking, since the latter increases the suscep- durability of the reinforced concrete deck slabs. In-
tibility of the embedded reinforcing bars to corrosion. creasing the stiffness of the bridge superstructure could
Cracking patterns and propagation are affected by the be a remedy to reduce cracking. James and his co-
load distribution on the slab and the stiffness charac- workers [2] reported that the most important cause of
teristics of the superstructure. A typical deck-type high- the deck damage was the increased flexibility of steel
way bridge consists of reinforced concrete deck, steel girders. Cao and Shing [3] stated that the maximum
girders, and diaphragms, which are the primary load bending moment in a deck depends on the stiffness ratio
bearing elements that determine the flexibility of the of girders and the ratio of the girder spacing to bridge
superstructure. The relative rigidity between the con- span length.
crete bridge deck and the remaining structural compo- Although in-depth study of the relationship between
nents of the bridge superstructure is considered as one superstructure flexibility and slab durability has not
of the important factors that affect the durability and been the major concern for many researchers, much
lifespan of the concrete deck. Altering the configura- research work has been focused on the configuration of
tion of the structural components of the superstruc- superstructure components and the effects on the wheel
ture changes the support conditions of the concrete load distribution. As an example, Azizinamini et al. [4]
deck. In recent years, a number of researchers reported reported that cross frames (diaphragms) have little effect
on the load capacity. Ebeido and Kennedy [5] argued
that the presence of cross frames would not affect the
*
Corresponding author. Tel.: +1-803-777-6166/8953; fax: distribution factor. Stallings and Cousins [6] concluded
+1-803-777-0670/8953. that diaphragms have a small influence on the distribu-
E-mail address: [email protected] (D.C. Rizos). tion factor. It is stated that the presence of diaphragms

0045-7949/$ - see front matter
2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruc.2003.08.009
14 S. Zhou et al. / Computers and Structures 82 (2004) 13–23

00 00
after construction may cause fatigue cracks at the dia- (M6 · 4.4 ) spaced at 18.18 in. Diagonal bracing
00 00 00
phragm–girder connection. However, Burke and Seif [7] (L1 · 1/2 · 1/8 ) between adjacent steel girders is placed
stated that the elimination of mid-span transverse braces at the ends and at the third points of the span. The
may have profound effects on the durability of rein- modulus of elasticity of the steel is assumed as
forced concrete deck slabs. Although it is generally ac- Es ¼ 29; 000 ksi. The thickness of the concrete slab is 1.4
knowledged that the flexibility of the superstructure has in. and the compressive strength of concrete has been
strong effects on the durability of the concrete deck, the determined experimentally as fc0 ¼ 6:43 ksi. Steel rein-
relative importance of the influence of the superstructure forcement is placed at the top and bottom of the con-
components on concrete deck durability is not clear. crete slab in the transverse and longitudinal directions.
Published research works are based on parametric The steel ratio is 0.3% for each layer in both directions.
studies, using computer simulations. However, there are Annealed D-1 wire, with yielding strength fy ¼ 60 ksi
significant limitations in such studies. As an example, and Young modulus Ews ¼ 28; 000 ksi, is used. The lo-
yielding of the reinforcement of the concrete deck and cation of diaphragms and the girders demarcate nine
the non-linear behavior of bridge structure has been regions (‘‘panels’’) as shown in Fig. 1(a). A concentrated
routinely ignored in finite element models (FEM mod- load is applied gradually on the center of each panel
els). In addition, the limitation of computing equipment until each panel failed.
and supporting software packages has also been
restricting factors. Therefore, refined and completed
studies are needed to evaluate the impact of super-
structure components to durability of concrete slab.
Sophisticated models are needed to simulate the actual
bridge superstructure behavior.
The objective of this research is to study how the
relative rigidity between the concrete bridge deck and
the remaining structural components of the bridge su-
perstructure impact the response of the concrete deck.
To this end, a number of commercial software, such as
ADINA and ANSYS, which accommodate non-linear
3-D FEM models, can be employed. The latter software
was readily available at the time this research was con-
ducted, and has been used. However, it has been re-
ported that ADINA offers a more reliable reinforced
concrete element with improved convergence charac-
teristics [8]. The bridge models focus on non-linear be-
havior of materials at early stages of loading. The
development of these models is based on a one-span
non-composite bridge superstructure, which was exam-
ined experimentally by Petrou et al. [9,10]. The experi-
mental data are used to validate and calibrate the
proposed FEM models. Subsequently, a series of para-
metric studies is conducted. The main variables investi-
gated are: (a) composite action, (b) girder spacing to slab
thickness ratio (slenderness ratio), (c) presence of dia-
phragms.

2. Model development, calibration and validation

2.1. Physical model

A 1/6.6-scale non-composite slab on girder bridge

superstructure, which was considered in experimental
studies by Petrou et al. [9,10] is used as the control
model in this work. This model is 90.9 in. long, simply
supported, non-composite structure with an isotropic
reinforced concrete deck supported on four steel girders Fig. 1. (a) Top view of physical model and (b) FE model.
 S. Zhou et al. / Computers and Structures 82 (2004) 13–23 15

2.2. FEM bridge model action. In the case of composite action, full coupling of
the degrees of freedom of the nodes on the girder–slab
Two FEM models are developed to simulate bridge interface is enforced. Because of the non-linear nature of
non-composite and composite action between the con- the models, the full Newton–Raphson solution method
crete slab and the girders. The geometry of a typical fi- is activated along with line search, and the adaptive
nite element model is shown in Fig. 1(b). ANSYS descent technique.
elements Shell 63, Link 8 and Solid 65 are selected to
model the steel girders, steel diaphragms and the rein- 2.3. Reinforced concrete material model and calibration
forced concrete deck respectively. In ANSYS, the Solid
65 element is a ‘‘reinforced concrete element’’ that ac- Concrete shows a complex structural response due to
commodates cracking and crushing of concrete and non-linearities associated with cracking, crushing, non-
yielding of the reinforcement, among other non-linear linear stress–strain relations, reinforcement bond, ag-
characteristics. However, this element is not free of gregate interlock, as well as time-dependent effects such
shortcomings, as discussed in the following section. The as creep, shrinkage, temperature, and load history.
reinforcement is assumed to be ‘‘smeared’’ throughout Three main aspects are of importance in the concrete
the element in this work with typical elastic–perfectly- models associated with the present scope of work, i.e.,
plastic stress–strain behavior. Reinforcement is placed in stress–strain relations, material failure criteria, and post-
the transverse and longitudinal directions (0.3% in each cracking and post-crushing behavior. Drucker-Prager
direction––top and bottom). Full bond is assumed be- (DP) models included in ANSYS are applicable for the
tween reinforcement and concrete. As shown in Fig. 2, simulation of granular materials, such as concrete. The
in the non-composite FE model the degrees of freedom multilinear isotropic hardening (MISO) behavior is
of the coincident nodes between the top flange and the used, because experimental measurements for concrete
bottom surface of the slab are coupled in the vertical and properties are available for the uniaxial stress–strain
transverse directions, but are uncoupled in the longitu- behavior only. A maximum tensile stress, as defined
dinal direction to simulate the particular non-composite from the cracking strength of concrete, is provided to
cut off the tensile part of the MISO curve. The definition
of the failure surface depends on such parameters as the
Solid elements uniaxial crushing and tensile stresses, biaxial crushing
stress, and ambient hydrostatic stress state. In ANSYS
concrete models, the formation of a crack at an inte-
Interface nodes
UX, UY coupled
gration point represents a plane of weakness in a di-
rection normal to the crack face. While crushing occurs,
concrete strength which contributes to the element
stiffness at the integration point could be ignored. In the
present model, the compressive strength of concrete is
adopted from the experimental measurements reported
in [10] as fc0 ¼ 6:5 ksi. A pilot study and preliminary runs
Shell 63 elements revealed that accurate prediction of the experimental
measurements by the numerical models depends on the
(a) Noncomposite Structure tensile strength of the concrete and the value of the shear
transfer coefficients, especially at early stages of loading.
Solid 65 elements It is determined that the uniaxial tensile strength should
pffiffiffiffi
Areas shared by be taken as ft ¼ 379 psi that corresponds to ft ¼ 4:7 fc0 .
concrete deck and The shear transfer coefficients are taken as 0.3 and 0.5
steel girder
for the open and closed crack, respectively, and are
adopted in all models. The default values provided in
ANSYS are used for the remaining parameters.
It is observed that at high load levels, beyond the
Y range of interest in this study, the overall solution does
Z not converge. It has been reported in the literature that if
X both cracking and crushing capabilities are activated in
Shell 63 elements
ANSYS simultaneously, fictitious crushing of the con-
crete may be caused due to the coupling of excessive
(b) Composite Structure
cracking strains to the orthogonal uncracked directions
Fig. 2. (a) Non-composite and (b) composite superstructure through Poisson’s effect [11,12]. This may be one of the
FEM model. reasons that cause divergence of the solution at later
16 S. Zhou et al. / Computers and Structures 82 (2004) 13–23

stages. Alternatively, Bathe and his co-workers [8] have panel. The ultimate load in this case is approximately
reported on a hypoelastic model for the concrete mate- 7500 lb. After failure of the central panel, the stiffness of
rial that is based on generalization of the uniaxial stress– the slab and the support conditions of the adjacent
strain behavior to represent the biaxial and triaxial panels are significantly altered. However, the ANSYS
states. Their model has shown better performance and solution for the same load case encountered convergence
stability characteristics than the ANSYS model and has problems when the load exceeded 5500 lb. It is well
been adopted in software ADINA. documented [11,12], that the ANSYS concrete model
does not produce acceptable results at high loads when
the cracking and crushing options are activated simul-
2.4. Validation taneously, even when refined mesh is considered.
Therefore, for future studies on the subject different
The experimental data presented by Petrou [9,10] software will be considered. Because of the aforemen-
pertain to the loading of the slab at the center of each tioned convergence issue, comparison with experimental
‘‘panel’’ shown in Fig. 1(a) starting from the central data is appropriate for the loading of the central panel

8000
7000
7000
6000
6000
5000
Load (lb)

5000
Load (lb) 4000
4000
Experiment 3000
3000 Composite
ANSYS
2000 2000 Non-Composite
1000 1000

0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.02 0.04 0.06 0.08 0.1 0.12
Net Deflection/Slab Thickness Net Deflection/Thickness

Fig. 3. Comparison of FEM results with experimental data. Fig. 5. Composite vs. non-composite action.

Fig. 4. Comparison of cracking pattern of bottom surface between (a) FEM model and (b) experimental model.
 S. Zhou et al. / Computers and Structures 82 (2004) 13–23 17

only. In this case, the net deflection of the bottom of the recorded. Subsequently, the normalized net deflection,
concrete slab with respect to the deflection of the girders u
s of the slab is calculated as,
is computed from the FE analysis and compared to the
reported experimental measurements. To this end, the us  0:5ðu1g þ u2g Þ
u
s ¼ ð1Þ
deflection, us of the point on the bottom surface directly ts
under the point of application of the load is recorded at
every load step. In addition, the deflections, u1g and u2g , of where ts ¼ 1:4 in. represents the thickness of the slab.
the bottom flanges of the two adjacent girders are also Fig. 3 shows the load–net-deflection curve of both the

Fig. 6. Deformed configuration for (a) non-composite bridge and (b) composite bridge slab under 5500 lb load.

Fig. 7. Cracking pattern of bottom surface of: non-composite (a–c) and composite deck (d–f).
18 S. Zhou et al. / Computers and Structures 82 (2004) 13–23

experimental and numerical investigation. The FEM 4. Effects of slenderness ratio

solution is in excellent agreement with experimental data
and it reached levels well above the linear elastic be- The objective of this section is to investigate the effects
havior of the system. Fig. 4(a) shows the cracking pat- of the slenderness ratio on the stiffness and cracking of
tern of the bottom surface of the FEM model at load the concrete slab. The slenderness ratio, k, is defined as:
level P ¼ 4000 lb. Fig. 4(b) shows the correspond- sg
ing cracking pattern of the experimental model at fail- k¼ ð2Þ
ts
ure. Comparison of Fig. 4(a) and (b) reveals the
similarities of the characteristic diagonal and longitudi- where ts is the slab thickness and sg is the center-to-
nal cracks between the two models. It is concluded that center girder spacing. Non-composite and composite
the FEM models and solutions are suitable for the ob- decks with diaphragms are considered.
jectives of the parametric studies of this work, the For the non-composite decks, four different slender-
findings of which are presented and discussed in the ness ratios corresponding to two groups of models are
following sections. considered. The first group adopts the girder spacing
sg ¼ 18:18 in. of the experimental model while the slab
thickness is assumed ts ¼ 1:2, 1.3, 1.4, and 1.6 in.
yielding slenderness ratios k ¼ 15:15, 14, 13, and 11.4.
3. Effects of composite and non-composite action The second group adopts the slab thickness ts ¼ 1:4 in.
of the experimental model while the girder spacing is
A concentrated load is applied on the central node of adjusted appropriately to obtain the same slenderness
the top surface of panel CC on both composite and non- ratios of the first group. The control model in this study
composite decks, the normalized net deflections are corresponds to the experimental model reported by
computed as in Eq. (1) and are shown in Fig. 5. Fig. 6(a)
and (b) show the deformed configurations of the non-
composite and composite structures, respectively, at the 7500

last recorded load step. In view of the load–deflection

curves, Fig. 5, and the deformed configurations, Fig. 6000
λ
6(a) and (b), is evident that the net deflection of the slab ts
11.4 1.4
of the non-composite deck is much larger than the de-
Load (lb)

4500 11.4 1.6

flection of the composite one. In the composite deck,
when a load is applied at the center of the concrete slab, 14 1.3

only the central panels develop significant deformations 3000

14 1.4

since the girders and slab are fully bonded. On the other 15.15 1.2
hand, in the non-composite slab, when a load is applied, 15.15 1.4
the deformations are distributed across the entire mid- 1500 13 experimental
span of the concrete slab. These findings are in agree- 13 FEM control
ment with the general conclusion that composite bridge
0
decks carry higher loads than non-composite decks 0 0.1 0.2 0.3 0.4 0.5
with identical girders (e.g., [13]). More profound differ- Net Deflection
ences that affect the lifespan of the concrete deck are (a) Non-composite bridge
revealed in the cracking pattern diagrams shown at
different load levels in Fig. 7(a)–(f) for the non- 7500
composite and composite case. It is observed that: (1)
In the non-composite deck, cracking starts at the cen- 6000
tral panel of the slab and propagates outwards in both
Load (lb)

directions. In contrast, cracking of the composite 4500

λ ts
deck takes place only along the three middle panels in 15.2 1.4
the longitudinal direction. (2) In the non-composite 3000 15.2 1.6
case, as the load increases, diagonal cracking develops 11.4 1.4

first. In the composite case cracking appears first in 1500 11.4 1.2

the central panel in the longitudinal direction and sub- 13

sequently propagates within the central panel and 0

0 0.05 0.1 0.15 0.2 0.25
along both sides of the interior girders. There is no
Net Deflection
significant diagonal cracking, and slightly minor trans- (b) Composite bridge
verse cracking is observed in the composite superstruc-
ture. Fig. 8. Load–deflection curves for different slenderness ratio.
 S. Zhou et al. / Computers and Structures 82 (2004) 13–23 19

Petrou et al. [9,10] which has a slenderness ratio of 13. for each of the slenderness ratios of k ¼ 15:15 and
The load–net-deflection curves for all cases are shown in k ¼ 11:4, two groups of models are considered. These
Fig. 8(a), and the corresponding cracking patterns are models have different slab thickness while the girder
shown in Fig. 9. It is evident in Fig. 8(a) that the slen- spacing is adjusted appropriately. The load–net-deflec-
derness ratio has an effect on the flexibility of the slab. tion curves for all cases are shown in Fig. 8(b), and the
Higher slenderness ratios yield larger net deflection of corresponding cracking patterns are shown in Fig. 10. It
the slab. However, at low load levels below 500 lb in the is also evident in Fig. 8(b) that the slenderness ratio has
scaled model, which correspond to typical values of an effect on the flexibility of the slab. A higher slender-
wheel load, the girder deflections were very similar for ness ratio yields a larger net deflection. Even at low load
all slenderness ratios. For slenderness ratios k ¼ 11:4 or levels below 500 lb in the scaled model, which corre-
k ¼ 14, the net-deflection curves of the slabs of the two sponds to typical values of wheel load, the girder de-
models for the same slenderness ratio appear to be flections were different for all slenderness ratios. For
similar. The model with thicker slab is slightly stiffer slenderness ratio k ¼ 11:4, the net deflection of the slabs
than the model with thinner slab. The load–net-deflec- of the two models for the same slenderness ratio is very
tion curve of the control model is located between the close. The model with thicker slab is slightly stiffer than
curves of slenderness ratio k ¼ 11:4, and k ¼ 14. For the model with thinner slab. The load–net-deflection
slenderness ratio k ¼ 15:15, the thinner slab model curve of the control composite model is located between
shows high non-linear behavior due to appearance of the curves of k ¼ 11:4, and k ¼ 15:15. For the slender-
cracking at early stages. Both models started showing ness ratio k ¼ 15:15, the net-deflection curves of the
non-linear behavior at comparatively low load levels. slabs of the two models for the same slenderness ratio
Composite decks with diaphragms are considered appear to be similar. The thinner slab model showed
next. The three different slenderness ratios k ¼ 15:15, 13, high non-linear behavior due to appearance of cracking
and 11.4 are used, while the slenderness of the control at early stages. Both models started showing non-linear
model is k ¼ 13. Similarly to the non-composite model, behavior at comparatively low load levels. It is evident

Fig. 9. Non-composite deck––cracking pattern of bottom surface under 3000 lb for various slenderness: (a) due to variable thickness,
(b) due to variable girder spacing.
20 S. Zhou et al. / Computers and Structures 82 (2004) 13–23

Fig. 10. Composite deck––cracking pattern of bottom surface under 3000 lb for various slenderness: (a) due to variable thickness, (b)
due to variable girder spacing.

that the models with higher slenderness ratios enter the 5.1. Effects on stiffness
non-linear range at lower load levels, as opposed to,
models with lower slenderness ratio. In view of Figs. 9 The load–net-deflection curves for the four cases are
and 10, different slenderness ratios would show similar plotted in Fig. 11. Fig. 12(a) and (b) show the deformed
cracking patterns. However, the cracking densities of configurations of the non-composite structures with and
these patterns showed a significant difference altering the without diaphragms under the 3000 lb load, respectively.
slab stiffness. Comparison of net slab deflection between Similarly, Fig. 13(a) and (b) show the deformed con-
composite and non-composite decks shows that com- figurations of the composite structures with and without
posite decks are more sensitive to the change of the
slenderness ratio, especially at lower load levels, as evi-
denced in Fig. 8(a) and (b). 3500

3000 Non-composite (NC)

Composite
2500
Load (lb)

5. Effects of diaphragms 2000 NC with Diaphragms

NC w/o Diaphragms
1500
The objective of this section is to determine the in- C with Diaphragms

fluence of diaphragms on the stiffness and cracking of 1000 C w/o Diaphragms

the concrete slab. To this end, four models are consid- 500
ered, i.e.: (1) non-composite superstructure with dia-
0
phragms, (2) non-composite superstructure without 0 0.01 0.02 0.03 0.04 0.05 0.06
diaphragms, (3) composite superstructure with dia- Net Deflection/Thickness
phragms, and (4) composite superstructure without di-
aphragms. Fig. 11. Effects of diaphragms on slab stiffness.
 S. Zhou et al. / Computers and Structures 82 (2004) 13–23 21

Fig. 12. Non-composite deck––deformed configuration under 3000 lb load: (a) with diaphragms, (b) without diaphragms.

Fig. 13. Composite deck––deformed configuration under 3000 lb load: (a) with diaphragms, (b) without diaphragms.
22 S. Zhou et al. / Computers and Structures 82 (2004) 13–23

Fig. 14. Cracking patterns of bottom surface of non-composite Fig. 15. Cracking patterns of bottom surface of composite
deck: (a) with diaphragms, (b) without diaphragms. deck: (a) with diaphragms, (b) without diaphragms.

reinforced concrete, such as cracking, local crushing,

diaphragms under the 3000 lb load, respectively. The
and plastic deformations and yielding of reinforcement.
girder rotation in the absence of diaphragms is evident.
In the load–deflection curve is evident that the presence
of diaphragms in the composite superstructure has little 5.2. Effects on cracking
effect on its net deflection while the load remains small.
The difference is 12% when the load reaches 2000 lb. For Fig. 14(a) and (b) show the cracking pattern on the
the non-composite superstructure, the effect of the dia- bottom of the slab for the non-composite bridge with
phragm is also small while load is less than 1000 lb. and without diaphragms, respectively. Fig. 15(a) and (b)
However, when the load increases above the 1000 lb show the cracking pattern on the bottom of the slab for
level, this difference increases sharply. In both cases, the the composite case with and without diaphragms, re-
effects of diaphragms on the stiffness of the deck become spectively. In the non-composite case, the difference in
significant under higher loads. This is due to (a) the non- the cracking patterns between the model with dia-
linear behavior of concrete and (b) the tendency of the phragms and the model without diaphragms are: (1)
girders to rotate under the higher loads. The diaphragms Four small groups of cracks at the intersection of the
prevent the rotation from happening. These results are girders and one-third point diaphragms at the third
different from studies reported in the literature due to points appeared on the model with diaphragms. How-
two reasons. The load used in the reported studies per- ever, these cracks do not appear on the model without
tains to the wheel load, which corresponds to loads diaphragms. This may be due to the stress concentration
approximately 500 lb in the scaled model considered in at the diaphragm–girder connection. (2) The diagonal
this study. However, the load used in this work is much cracks are the most obvious cracks in the superstructure
higher than the wheel load. Furthermore, other studies without diaphragms. The diagonal cracks propagate
have either used oversimplified finite element models or significantly as the load increases. Several cracks appear
not considered the non-linear material behavior of the parallel to the diagonal crack. Therefore, the zone of
 S. Zhou et al. / Computers and Structures 82 (2004) 13–23 23

diagonal cracks becomes much wider than in the model net deflection of the slab. Cases of same slenderness
with diaphragms. (3) There are no transverse cracks in ratio involving thinner slabs show more extensive dam-
the non-diaphragm model, in contrast to the diaphragm age of the slab. These effects are more profound at
model, which has cracks that appear alongside the dia- higher loads. Composite decks are more sensitive to
phragms. (4) The cracks along the inside girders ap- changes of slenderness ratio, especially at lower load
peared at the outside panel in the longitudinal direction levels.
in the model without diaphragms. This kind of cracks
does not appear in the model with diaphragms.
References

6. Conclusions [1] Burke MP. The adverse effect of AASHTO’S LRFD

criteria for deflection on the integrity and durability of
This work considers single span slab-on-girders reinforcement concrete deck slabs. Transportation Re-
bridges and studies the effects of the flexibility of the search Board 79th Annual Meeting, Washington DC,
2000.
bridge superstructure on the strength of the concrete
[2] James RW, Zimmerman RA, McCreary CRJR. Effects of
slab. To this end, software ANSYS 5.7 is effectively used overloads on deterioration of concrete bridges. Transpor-
to conduct a series of parametric studies with respect to tation research record 1118, Transportation Research
composite action, existence of diaphragms and slender- Board, National Research Council, Washington DC,
ness ratio. The accuracy of the FEM models is validated 1987. p. 65–72.
by experimental results reported in the literature. Con- [3] Cao LC, Shing PB. Simplified analysis method for slab-on-
vergence issues due to smeared reinforcement assump- girder highway bridge decks. J Struct Eng 1999;125(1):49–
tion along with simultaneous checking of cracking and 59.
crushing of concrete appear at relatively high load levels [4] Azizinamini A, Kathol S, Beacham MW. Influence of cross
and do not affect the presented parametric studies. The frames on load resisting capacity of steel girder bridge. Eng
J 1995;33(3):107–17.
parametric studies have identified and addressed the
[5] Ebeido T, Kennedy JB. Punching strength of deck slab in
following issues: skew composite bridges. J Bridge Eng 1996;1(2):59–66.
1. At low load levels, the effects of the composite [6] Stallings JM, Cousins TE. Evaluation of diaphragm
action and the presence of the diaphragms are very requirements in existing bridges. In: Proceedings of Struc-
small. The stiffness of the concrete slab is proportional tures Congress XV sponsored by The Structural Engineer-
to the slenderness ratio. ing Institute of ASCE held in Portland, Oregon, 1997.
2. At intermediate load levels, and while the slab has p. 1494–8.
entered the inelastic range, the effects of the composite [7] Burke MP, Seif JS. The adverse effect of AASHTO’S
action are more profound both in the stiffness of the slab LRFD specifications for transverse braces and deflections
and crack distribution. on the integrity and durability of reinforcement concrete
deck slabs. The Transportation Research Board’s 79th
3. The presence of diaphragms has minimal effects on
Annual Meeting, Washington, DC, 2000.
the composite bridge deck at lower load levels. However, [8] Bathe K, Walczak J, Welch A, Mistry N. Nonlinear
at intermediate loads the diaphragms stiffen the non- analysis of concrete structure. Comput Struct 1989;32(3/4):
composite deck to the levels of the composite deck. In 563–90.
both the composite and non-composite cases and when [9] Petrou MF, Perdikaris PC, Duan M. Static behavior of
diaphragms are present, local failures of the slab are non-composite concrete bridge decks under concentrated
detected in the vicinity of the connection of the dia- loads. J Bridge Eng 1996;1(4):143–54.
phragms to the girders. In the non-composite case, the [10] Petrou MF. A model study on highway concrete bridge
load distribution is significantly different between the decks subjected to concentrated load (ONTARIO vs.
case of a structure with diaphragms and without dia- AASHTO Design). MS Thesis, Department of Civil
Engineering, Case Western Reserve University, Cleveland,
phragms. However, the distribution of cracks between
OH, 1991.
composite structures with and without diaphragms [11] ANSYS theory manual 5.7, 2001. ANSYS Inc., 12th ed.,
shows little difference. 2001.
4. The slenderness ratio has significant effect on the [12] ANSYS user’s manual for revision 5.7. ANSYS Inc., 2001.
stiffness of the non-composite superstructure and the [13] Narendra T. Design of modern highway bridges. USA:
load distribution. Higher slenderness ratio yields greater McGraw-Hill; 1998.