Journal of Constructional Steel Research 138 (2017) 701–713

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Static and seismic behaviours of innovative hybrid steel reinforced

concrete bridge
Mohammad Hamid Elmy a, Shunichi Nakamura b,⁎
a
School of Engineering, Nangarhar University, Jalalabad, Nangarhar, Afghanistan
b
Department of Civil Engineering, Tokai University, 4-1-1 Kitakaname, Hiratsuka 259-1292, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Bridges using rolled steel H-section seem to be economical compared with plate girder bridges due to lower ma-
Received 24 February 2017 terial and fabrication cost in the short-span length. The rolled steel H-section is a compact section and no stiffener
Received in revised form 16 August 2017 is necessary. However, as the maximum available web height of the rolled H-sections is 900 mm, the applicable
Accepted 22 August 2017
span-length is 20–25 m. A new composite girder bridge was proposed using rolled steel H-section to extend the
Available online xxxx
span-length. The superstructure consists of continuous rolled steel H-girders composite with the RC slab, and the
Keywords:
substructure is RC piers. The girder and the pier are rigidly connected by concrete and reinforcements to resist
Composite girder bridge large negative bending moment at the joints. Experiments were conducted with the partial SRC bridge model,
Rolled steel H-beam which showed that the proposed structure with a rolled H-girder had high ultimate strength with good ductile
FE analysis property, and attained the full plastic moment. A FE model was established considering material non-linearity
Limit states design and shear transfer between steel and concrete and compared with the experiments, which showed good agree-
Dynamic analysis ment and clarified the load transfer mechanism. A trial design with the SRC highway bridge was carried out, con-
Seismic performance firmed that the applicable span length can be extended to 50 m. Seismic performance of the SRC system was
verified by dynamic analysis, showed sufficient strength and ductility for strong earthquake with peak accelera-
tion of 810 m/s2. Cross beams were eliminated in the new SRC bridge because the girder height was low, which
was verified by FEM analysis with the full bridge model.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction superstructure is a continuous girder with the rolled steel H-section

The girder bridge using steel rolled H-sections is competitive and
economical in short spans due to low material and fabrication costs.
The rolled steel H-beam is a compact section and no stiffener is neces-
sary for girders. However, the maximum available web height of the
rolled H-beam is about 900 mm, the applicable span length for a road
bridge is about 20 m for simple spans and 25 m for continuous spans.
In a continuous structural form, the intermediate support area is sub-
jected to negative bending moment and is usually more critical than
the span centre, which limits the span length.
For a simply supported girder where the bending moment is positive
along the span length, the composite girder consisting of the
steel H-girder and the reinforced concrete (RC) slab is rational. The
steel girder is in tension and the RC slab is in compression. However,
for a continuous composite girder, the RC slab is tension and does not
contribute much at support joints, where large negative bending mo-
ment exists.
A new form of steel and reinforced concrete (SRC) composite
girder bridge was proposed using a rolled steel H-section [1]. The
⁎ Corresponding author.
E-mail address: snakamu@tsc.u-tokai.ac.jp (S. Nakamura).
which is composite with the RC slab. Steel/concrete composite bridges
are commonly used all over the world because of the attractive appear-
ance and efficient structural rationality. The steel girders and the RC
piers are rigidly connected by reinforced concrete at the pier top. The
rolled H-beams are strengthened around the joints by being covered
with reinforced concrete, which forms the SRC section and increases
the resisting capacity of the section at support joints. The proposed
SRC bridge using rolled steel H-section is basically a multi-span rigid
frame bridge structure and is expected to be competitive and economi-
cal compared with the welded plate girder bridges.
Experiments were conducted with the partial SRC bridge model [2],
and the bending strength and the load transfer mechanism at the steel-
concrete rigid joint was investigated. Finite element model was then
developed, considering material non-linearity and shear connections
between steel and concrete. The FE model was applied to the experi-
ments and compared with the test results to clarify the behaviours of
the rigid joints and to verify the FE model.
A trail design was conducted with a road bridge with the maximum
span length of 50 m, and safety verification of the sections was per-
formed by the limit states design method. Furthermore, the structural
performance of the proposed bridge form against ultra-strong earth-
quakes was studied by dynamic analysis. These studies were intended

http://dx.doi.org/10.1016/j.jcsr.2017.08.025
0143-974X/© 2017 Elsevier Ltd. All rights reserved.
702 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713
to confirm that the new SRC bridge using rolled H-girders is feasible for
a much longer span bridge than the existing one.
In this SRC bridge with rolled H-girders, the web height is only about
900 mm and the steel girder is composite with relatively thick RC slab,
which can provide sufficient stiffness against transverse forces without
cross beams. Therefore, the cross beams could be unnecessary in the
new SRC bridge, which was also verified by FEM analysis with the full
bridge model.
2. Structural form
The structural form of the proposed steel and reinforced concrete
composite girder bridge with rolled H-beam section is illustrated in
Fig. 1 and Fig. 2. The super-structure is a continuous girder with the
rolled steel H-section and the RC slab, and the sub-structure is the RC
piers. The girder and the pier are rigidly connected by concrete and
reinforcements. The concept of this new structural form is that the
steel/concrete composite girder resists the positive bending moment
at the span centre and the steel girder covered by reinforced concrete
section (SRC) resists the negative bending moment at the rigid joints.
As shown in Fig. 3, this new SRC bridge is basically a multi-span rigid
frame structure and the applicable span length is expected to reach
almost double the existing ones.
In order to achieve a reliable composite rigid frame bridge, shear
forces must be transferred between the steel girder and the RC part.
Perfo-bond rib (PBL) shear connectors were welded to the upper and
lower flanges of the steel girders in the rigid joint area. The PBL consists
of the steel plate with holes, into which reinforcements go through. The
steel girders were covered by reinforced concrete at around 15% of the
span length near the support joints. Sufficient amount of reinforcements
in the cross-section were installed longitudinally and stirrups were also
arranged at proper intervals.
3. Experiments
Bending tests were conducted with the partial SRC specimen [2],
which has a reverse L-shape shown in Fig. 4. The objective of the test
was to investigate the bending strength and ductility of the steel-
concrete rigid joint and to clarify the load transfer mechanism from
the girder to the pier at the rigid joint. The test specimen was scaled
down to half of the actual bridge model, using rolled H-beam section.
The embedded length of the steel girder at the rigid joint was approxi-
mately twice the girder height to provide adequate embedded length.
A half of the length of the steel girder was covered by RC to form SRC.
3.1. Experimental set-up
The structural steel used in the test was mild steel SS400 with a yield
stress of 338 MPa. The specimens had a length of 3500 mm and a height
of 1720 mm. The RC slab had a width of 1000 mm and a depth of
Fig. 1. The SRC girder bridge with rolled steel H-section.
Span centre Support
Fig. 2. Cross-section of the SRC bridge.
100 mm, with an additional 50 mm depth of the haunch. The concrete
slab had reinforcement steel bars of SD295 with a yield stress of
388 MPa. The reinforcement bars were 19 mm in diameter and spaced
at a pitch of 100 mm. It was arranged in one layer at the slab centre.
The rolled H-beam consisted of a web with 440 mm in height and
11 mm in thickness, and upper and lower flanges with 300 mm in
width and 18 mm in thickness. About half length of the steel girder
was covered by RC with a section of 500 × 590 mm, forming the SRC sec-
tion. The measured cylinder strength of concrete was 30.0 MPa. The RC
pier had a cross section of 1000 mm × 900 mm. The reinforcement was
SD345 with a yield stress of 374 MPa and with a diameter of 25 mm at a
pitch of 125 mm (D25@125). The general layout and dimension of the
designed specimen is shown in Fig. 5.
The Perfo-bonded rib (PBL) shear connectors were used at the
beam-column rigid corner joint to prevent shear slippage between the
steel and concrete interface. The PBL consisted of steel plates with a
height of 100 mm, a thickness of 12 mm and a length of 800 mm.
There were 60 mm diameter holes at the centre. The PBLs were welded
to the upper and lower flanges of the rolled H-girder. Reinforcement
(D22) were installed in the holes. Stirrups were also used to bond the
steel girder and the RC pier. The PBL arrangement and details are
shown in Fig. 6.
3.2. Test procedure
The specimen was incrementally loaded at the end of a cantilever
beam and the vertical and horizontal displacements were monitored.
The capacity of the loading machine was 625 kN and the test was termi-
nated when the jack reached the maximum capacity. Concrete cracks
were carefully checked at each loading step at the slab and around the
beam-column rigid joint.
Vertical and horizontal displacements were measured at the loading
position. Strains of the steel girder, the RC slab, the RC rigid joint, the RC
column and the reinforcements were measured by strain gauges. The
results obtained by experiments are presented with the FEM analysis
in Section 5.
4. Finite element model
4.1. FE model
FEM has been commonly used to simulate structural behaviours of
composite bridges and experimental results were often verified by
FEM analysis [3–11]. Considering these past studies, a finite element
model was developed for the current structure using the FEM program
ABAQUS [12] and applied to the experimental specimen to clarify the
behaviours of the rigid joints. The steel girder and the concrete slab, col-
umn and beam were modelled as a solid element (C3D8R), and the steel
reinforcements as a truss element (T3D2) with two nodes and three
translational degree of freedom were adopted [3,4,6,7,9,10]. The shear
 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713 703

Fig. 3. Side view of the proposed SRC bridge.

Fig. 4. Test set-up.

connectors (PBLs and headed studs) were modelled as spring elements
[3,7,8]. Fig. 7 shows the developed finite element model.
A parametric study was conducted about the finite element mesh
size to obtain accurate results with the minimum computational time.
It was found that a mesh size of approximately 50 × 50 mm (length
by width of the element) provides adequate accuracy in modelling
the concrete slab and steel beam, while a mesh size of approximately
100 × 75 mm for the concrete pier. The smallest mesh size was equal
to the steel plate thickness (50 × 18 mm). The total number of elements
used in the model shown in Fig. 7 was 28,168.
The vertical load was applied incrementally as a concentrated load at
the top surface of the concrete slab along the full width of the slab,

B C

Fig. 5. Dimension and set up of test model (unit: mm).

704 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713

T-L T-M T-R

B-L B-M B-R

Fig. 6. Details and arrangement of PBLs.

which is identical to the experimental procedure. The total load was di-
Fig. 8. Stress-strain relationship of concrete in compression.
vided into several steps in the same way as the experiment and analysis
was conducted at each step. Geometrical non-linearity was considered
in the FE analysis to account for large deformation. of concrete in compression is expressed by the following Eqs. (1), (2),
(3) and (4) and the curve is shown in Fig. 8.
4.2. Contact and interaction models

σ 0 ¼ E0 K ε0c −ε0p ≥0 ð1Þ
A mechanical interaction, the surface to surface contact in ABAQUS,
was used between the steel girder and the concrete. The friction coeffi- 0
cient between steel plates and concrete in the tangential direction has 2 f cd
E0 ¼ ð2Þ
been assumed as 0.4–0.55 in most of the FEM modelling [4,9,10]. The ε0peak
different values in this range were applied to the current FEM models, ( !!)
which showed that there was not significant difference in interface ε0 ε0
slips or global behaviours. Therefore, the average value of 0.45 in this K ¼ exp −0:73 0max 1− exp −1:25 0max ð3Þ
εpeak εpeak
range was adopted in the model. In the normal direction, the hard con-
tact property available in ABAQUS element library adopted between the ( !)
ε0max
steel girder and the covered concrete interface [3,4,6,10]. The rigid con- ε0p ¼ ε0max −2:86ε0peak 1− exp −0:35 ð4Þ
straint was used between the steel girder and the vertical stiffeners at ε0peak
the welded parts. The reinforcements were embedded into the con-
crete, by which the reinforcing bars were perfectly connected with the where, E0: initial modulus of elasticity (26,000 N/mm2), ε′peak : strain
surrounding concrete [3,7,8,10]. corresponding to compressive strength (0.002), σ′: c compressive con-
crete stress, K: residual rate of elastic stiffness, εp′: plastic strain, ε′max :
4.3. Material modelling maximum strain in the past and f′cd: design compressive strength of
concrete (30.0 N/mm2). Fig. 8 illustrates the compressive behaviour of
As for the concrete in compression, the non-linear elastic-plastic be- concrete [6]. The density and Poisson's ratio for the concrete was
haviour was expressed by an equivalent uniaxial stress-strain curve in- assumed to be 2400 kg/m3 and 0.2, respectively.
cluding the softening effect after the post-peak point based on the JSCE The stress-strain relation of concrete in tension is assumed that the
specification [6]. The modulus of elasticity and compressive behaviour tensile stress linearly increases with the tensile strain up to the concrete

Reinforcements (T3D2) Concrete (C3D8R)

Loading plate

Steel beam (C3D8R)

Spring elements

Fig. 7. Finite element model of SRC girder.

 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713
Fig. 9. Stress-strain relationship of concrete in tension.
cracking stress and decreases non-linearly because of the tension stiff-
ening effect, as expressed by Eqs. (5), (6), (7) and (8). Fig. 9 illustrates
the behaviour of concrete in tension [13].
σ t ¼ E0 εcr 0 ≤εt ≤εcr
σt ¼ ft εcr bεt ≤εtu
 0:4
εtu
σt ¼ ft εtu bεt
εt
 0 2
f t ¼ 0:23 f cd 3
where, fcd′ : design compressive strength of concrete (30.0 N/mm2), εt:
average of concrete tensile strain, εcr: initial cracking strain, εtu: crack
developing strain (0.0002), ft.: design tensile strength of concrete
(2.2 N/mm2). The complex non-linear material behaviour of the con-
crete material both in compression and tension was modelled using
damage plasticity model available in the ABAQUS material library [15].
A tri-linear relation was used as constitutive law for the steel beam,
shown in Fig. 10 [8]. The yield strength fsy (290 N/mm2), yield strain εsy
(0.00145) and the ultimate strength fsu (450 N/mm2) need to be input
to define the stress-strain curve, while the strain at the onset of strain
hardening εsh was assumed to be 0.015 and the slope of non-linear
hardening state was assumed to be Es/100 (Es: modulus of elasticity of
steel). The density of the steel in all the components was assumed to
be 7800 kg/m3. The modulus of elasticity and Poison's ratio of steel
was taken as 200 GPa and 0.3, respectively. For the steel reinforcement
materials, a simple elastic-perfectly plastic model without strain hard-
ening behaviour is employed as the constitutive law [14] with the
yield strength fsy (388 N/mm2) and the yield strain εsy (0.00194).
Fig. 10. Stress-strain relationship for steel.
705
5. Comparison of FEM and experimental results
To validate the developed finite element model and clarify the load
transfer mechanism, the results by FEM analysis were compared with
the experimental results. The load-deflection curve, load-strain relation
and strain distribution patterns obtained experiments and FEM using
the finite element model were compared.
5.1. Load vs. displacement relationship
Vertical displacements at the loading position are shown in Fig. 11. It
was found by the experiment that the concrete crack started on the slab
surface at A (Fig. 5) at the applied load of about 90 kN, the concrete
cracks started on the pier surface at B (Fig. 5) at the applied load of
about 130 kN, the pier rebar yielded at thee applied load of 400 kN
and the slab rebar yielded at the applied load of 625 kN.
The load versus displacement curve changed when cracks occurred
at the RC pier and the steel rebar yielded at the pier/girder joint, and
the loading was terminated when the loading jack reached the limit.
FEM and experimental load-deflection curve agreed in the initial elastic
region up to about 130 kN, which indicates that no degradation occurs
ð5Þ in stiffness of the SRC girder. After then, although there observed
a small difference in the curve between the test results and FEM calcu-
lations, reasonably good agreement was obtained. The maximum
ð6Þ
deflection obtained by FEM (98.46 mm) was nearly the same as the ex-
perimental value (97.5 mm) at the final load, at which the experiment
was terminated.
ð7Þ
5.2. Load vs. strain relationship
ð8Þ
The longitudinal strains of reinforcements in the pier (at point B in
Fig. 5) were measured in the experiment and compared with FEM re-
sults (Fig. 12). There was a change of curve at the load of about
130 kN by both the experiment and the FEM analysis. This was caused
by the concrete cracks occurred at the pier top. Therefore, the assumed
crack initiation strength of Eq. (6) was validated. This analysis intended
to calculate the strength of concrete crack initiation but not to predict
crack distribution. Therefore, the applied load at the crack initiation cal-
culated by the FEM analysis was compared with that observed by the
experiment. Comparison of crack distribution was not carried out,
which showed the limit of the current study and was taken as a future
task.
The pier rebar yielded at a load of 400 kN in the FE analysis, which
almost agreed with the experimental result (440 kN). Both experiment
and FEM curves followed the same tendency until yielding point, after
that the experiment showed larger tensile strain compared to FEM re-
sults. As mentioned before, the experiment was terminated when the
Fig. 11. Vertical displacements at the loading position.
706 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713
Fig. 12. Strains of reinforcement at point B in Fig. 5.
applied load reached 625 kN due to the limit of the loading machine.
The model did not collapse at this load and had a little extra capacity.
The FEM results of Fig. 12 indicted that the reinforcement became
yield and the strain continued to increase, and the plastic hinge would
formulate at the top corner of the pier (point B in Fig. 5). This would
make the vertical displacement diverge and the failure would occur.
Fig. 13 shows the tensile strain of the slab concrete, the slab rebar
and the upper flange at the rigid joint (section-A in Fig. 5) calculated
by the FE analysis. There was a change of curve at the load of about
100 kN which was nearly the same load when the slab concrete cracked
at the experiment. This validated the assumed concrete crack initiation
strength of Eq. (6).
It was observed that all three strains linearly increased in the early
stage of loading. However, the strain of the concrete exhibited a differ-
ent curvature compared with the rebar and the upper flange as the
load increased. This could be caused by the interface slip and the crack
of concrete slab.
Fig. 14 shows the compressive strain distribution on the concrete
surface of the pier face at 50 mm below the lower flange (point C in
Fig. 5). It was understood that the strain distribution was almost uni-
form in the transverse direction at each load stage and the FE results
agreed well with those obtained by the experiments.
5.3. Load vs. slip behaviour of the interface
The upper and lower flanges were connected to the concrete by
PBLs, which were modelled by springs with the spring constant of
Fig. 13. Tensile strains at point A in Fig. 5.
Fig. 14. Strain distribution along the pier surface at point C in Fig. 5.
120 kN/mm in FEM model. The interface slip was computed by relative
displacements in the longitudinal direction between the adjacent nodes
of the concrete and the steel flanges. The slip between the upper flange
and concrete is shown in Fig. 15 and that between the lower flange and
the concrete is shown in Fig. 16. There was a change of curve at about
100 kN which corresponded with the load when the slab concrete
started to crack. Then, the slip increases with the load and the interface
slip is only 1.4 mm even at the ultimate load. Also, the slip along the
three points at PBL behaved similarly, which indicated that PBLs worked
uniformly. These slip results proved that PBL shear connectors had a sig-
nificant effect on reducing the interface slip and provided adequate
bonding properties for load transfer at the rigid joint.
6. Design example
6.1. Bridge model
To confirm the feasibility and the applicability of the proposed struc-
tural form, a trial design was carried out with a full SRC bridge model.
The bridge was a four-span continuous girder road bridge with the
span length of 40 + 50 + 50 + 40 m, accommodating two lanes, and
consisting of five steel rolled H-girders. The concrete piers were 10 m
high, 3.0 m wide, and rigidly connected to the girders. The layout and
dimensions of the model bridge is shown in Fig. 17 and Fig. 18.
The steel/concrete composite girder was assumed for the positive
bending moment part at the span centre and the steel girder covered
by the reinforced concrete section was assumed for the negative
Fig. 15. Load vs interface slip at position of upper flange PBLs.
 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713
Fig. 16. Load vs interface slip at position of lower flange PBLs.
bending moment at the support joints. The concrete slab was assumed
220 mm thick. The steel girders were covered by reinforced concrete
at about 15% of the span length near the support joints. The SRC section
had reinforcements with about 2% of the total cross-sectional area. The
steel with tensile strength of 490 MPa was assumed for H-girders, and
the reinforcement with yield strength of 390 and 490 MPa were as-
sumed for the slab and the RC section, respectively. For the RC slab
and piers, concrete with a compressive strength of 40 MPa was
assumed.
6.2. Design loads
The two kinds of design dead loads were considered: pre-composite
dead load (D1) due to the self-weight of girders, fresh concrete of the
slab and form work, and the post-composite dead load (D2) due to
the pavements, railings and traffic barriers. The design live load
(L) and the impact effect (I) was adopted from the Japanese road bridge
specification [16]. The design live load consists of the equivalent con-
centrated load (p1 = 10 kN/m2) due to a train of large vehicle acting
over a length (D = 10 m), and the uniformly distributed load (p2 =
3.5 kN/m2) which represents all kinds of small vehicles and imposed
over the entire length. The p1 and p2 are loaded with a width of 5.5 m
Fig. 17. Side view of the bridge model (unit: m).
Fig. 18. Cross section of the bridge model (unit: m).
707
and half of them loaded on other parts. The live load impact factor
was assumed as 20% based on the span length.
6.3. Sectional forces
Linear static analysis with a 3D bridge model with beam elements
was performed to obtain sectional forces and deformations. The contin-
uous beam model was used for pre-composite case, and the 3D rigid
frame model was used for the post-composite load case. Influence line
analysis was performed for the live loads, using SAP 200 v.16 [17]. The
distributed p1 and p2 line loads was applied as moving loads so that
they produced the maximum and minimum sectional forces and
deformations.
The sectional forces of the edge girder (G1) were largest among all
the girders. The bending moment diagram of G1 is illustrated in
Fig. 19, which showed the bending moment diagram due to pre-
composite dead load (D1), post-composite dead load (D2) and the max-
imum and minimum bending moments obtained by the influence line
analysis due to live load (L). The maximum deflections due to the design
live load obtained was 68.6 mm in the P2-P3 span of G1, which was
within the allowable value of 100 mm (1/500 of the span length).
6.4. Safety verification
Safety verification was carried out by the limit states design method
to ensure load carrying capacity of the composite girders throughout
the service life of the structure against possible actions. The verification
method and safety factors conform to the Standard specifications for hy-
brid structures [14].
Fig. 20 shows the stress distribution of composite section at the span
centre subjected to positive bending moment to find the resistant de-
sign bending moment. As the neutral axis usually lies within the con-
crete slab due to relatively thick concrete slab, the lower part below
the neutral axis is in tension which can be resisted by the steel
H-girder. Fig. 21 shows the stress distribution of the SRC section at the
intermediate support subjected to negative bending moment to find
the resistant design bending moment. The neutral axis lies within the
web because the slab concrete is in tension and does not to contribute.
The compressive forces at the lower part is resisted by the steel girder,
reinforcing bars and covered concrete. The tensile forces at the upper
708 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713
Fig. 19. Bending moment diagram (G1).
part are resisted by the reinforcing bars and the upper part of the steel
H-beam.
The fibre model was used to obtain the bending moment-curvature
(M-ϕ) relation and the ultimate bending capacity of SRC sections. The
cross section was divided into small fibres and each fibre conformed
to the constitutive law of steel and concrete explained in Section 4.
Fig. 22 shows the M-ϕ relation of the composite and the SRC sections,
by which the maximum values were obtained.
The safety of member at each section was verified before and after
formation of composite action in accordance to the Standard specifica-
tions for hybrid structures [14]. The basic verification equation is
expressed by:
γi ðMd =Mud Þ≤1:0
where γi: structure factor which is usually set at 1.1, Md: design bending
moment and Mud: design resistant bending moment. In this equation,
Fig. 20. Stress distribution at composite section.
Fig. 21. Stress distribution at SRC section.
the load factor is 1.1 for pre-composite dead load, 1.2 for post-
composite dead, and 1.98 for live load. The material factor is 1.05 for
steel, 1.3 for concrete and 1.0 for steel bars, in addition to the member
factor of 1.1 for section overstretching.
The design bending moment due to the pre-composite load (Md1) is
taken only by the plastic bending capacity of the steel H-girder (Msud),
while the design bending moment due to the post-composite load
(Md) is resisted by the ultimate bending resisting capacities (Mud) of
the composite and the SRC sections at the span centre and the support
joints, respectively. Table 1 and Table 2 are the verification results of
G1, showing that all the sections satisfy Eq. (9).
6.5. Effects of temperature change
ð9Þ
Temperature change may affect structural behaviours in the rigid
frame structure. Then, structural analysis was also carried out with the
3D frame model to evaluate the effects of stresses and displacements
caused by temperature change. A uniform temperature of 15°C was ap-
plied to the girders of the bridge model, resulting in the expansion
movement of 6.7 mm at the end pier top. As this movement is small,
the expansion joints can be simple and compact.
This temperature change caused bending moments and axial forces
at the girder sections. The bending moment was not large, as illustrated
in Fig. 19, whereas the axial forces reached 2,600 kN. It also caused
bending moments and axial forces at the pier sections. The bending mo-
ment at the pier base was significantly large with the maximum values
of 82.0 MN·m and 35.6 MN·m for P1 and P2, respectively.
Fig. 22. M-ф relation of girder sections.
 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713 709

Table 1 Table 3
Verification of girder due to pre-composite dead loads. Verification of girder and pier due to dead and live loads and temperature change.

Position Md1 Msud Safety check γi Remarks Position Md Mud Nd Safety check γi Remarks
(kN·m) (kN·m) (Md1/Msud) ≤ 1 (MN·m) (MN·m) (kN) (Md/Mud) ≤ 1

N1 0 3200 0.00 OK N1 6.12 13.5 1220 0.50 OK

C1 1950 3200 0.67 OK N3, N4 12.11 14.0 2600 0.95 OK
N2 −2860 −3200 0.98 OK P1 108.13 141.0 −8900 0.84 OK
N3 −2950 −4100 0.79 OK P2 50.32 141.0 −12,100 0.39 OK
C2 1780 4100 0.48 OK
N4 −3080 −4100 0.83 OK

earthquakes with high probability of occurrence during the bridge ser-

The girder and the pier were verified against the combination of
dead load, live loads and thermal load by the Eq. (9), where the load fac-
tor was 1.2 for the temperature load and 1.65 for live load. The results
are summarized in Table 3, showing that the thermal load affects the
girder only a little but can be critical for the pier in this design example.
All of the member sections satisfied Eq. (9). It is reminded that the pier
also suffers large bending moment caused by earthquakes, which is ver-
ified in Section 7.
6.6. Estimate of construction costs
The construction cost of the designed SRC bridge with the span
length of 40 + 50 + 50 + 40 m was estimated and compared with
that of the conventional plate girder bridge which had the same span
length and width. The plate girder bridge consisted of two steel
I-girders with a web height of 2,900 mm and the RC slab with 320 in
thickness. RC piers were assumed for both the SRC and the plate girder
bridges. As the sub-structure was not very different for both types of
bridge, the construction cost of the super-structures was compared.
The required amount of steel for the SRC girders was 9.3% higher
than the plate girders, whereas the unit material and fabrication cost
of the SRC girder was 47.5% lower than the plate girder mainly due to
low fabrication cost. The required amount of concrete of the SRC bridge
and the plate girder bridge was nearly equal because the SRC bridge
needed extra concrete for the SRC parts but less concrete for the slab
which was supported by multi-girders consequently reduced its
thickness.
The above amounts gave that the construction cost of the SRC bridge
was 22% smaller than the plate girder bridge. In addition, the SRC bridge
requires no bearing, which provides extra benefits in the initial and
maintenance costs.
7. Seismic analysis
According to Japanese Seismic Code for Road Bridges, two levels of
design earthquake ground motions shall be considered in the seismic
design of bridges [16]. The strong earthquake level corresponds to
Table 2
Verification of girder and pier due to dead and live loads.
vice life (Level-1 earthquake, L1-EQ), and the ultra-strong earthquake
level corresponds to earthquakes with less probability of occurrence
during the bridge service life but very strong enough to cause critical
damage (Level-2 earthquake, L2-EQ).
The structure should satisfy the required performance against the
design earthquake forces. Important bridges in highways and national
roads must satisfy the seismic performance-1 (SP-1), which specifies
that members must be kept elastic and no damage is allowed, for
L1-EQ. For L2-EQ, they must satisfy the seismic performance-2 (SP-2),
which specifies that members can be elastic and plastic and minor
damage is allowed but no major damage, or the seismic performance-
3 (SP-3), which specifies that the bridge does not collapse. The push-
over analysis and the time history analysis were conducted for L1-EQ
and L2-EQ.
7.1. Structural model for dynamic analysis
The 3D frame model with beam elements was used for the seismic
analysis. The girder and the piers were divided into fibre elements of
steel and concrete. The non-linear material constitutive relation of
steel and concrete of each fibre was the same as those used for the
FEM analysis in Section 4. The RC slab was replaced by the discrete
beam elements, as illustrated in Fig. 23 [18]. Good and hard grounds
such as the diluvium and rock ground was assumed in this design exam-
ple. The interaction between the sub-structure and the ground was con-
sidered by linear springs in the three directions and three rotations. The
equivalent stiffness of these springs was obtained by the Design Specifi-
cation for Railway Structures [19] and the values are shown in Table 4.
7.2. Push-over analysis
Eigen value analysis was conducted, finding the dominant natural
period, which was 0.435 sec. in the transverse direction and 0.384 sec.
in the longitudinal direction. For L1-EQ, the corresponding design seis-
mic coefficient was 0.2 in two directions. For L2-EQ, the corresponding
design seismic coefficient was 2.0 in two directions.
The push-over analysis was conducted in the longitudinal and trans-
versal directions. The horizontal force H was incrementally applied to
the structure until the structure collapsed. The horizontal force (H) is
the dead load (D) multiplied by the seismic coefficient (Kh).

Position Md Mud Nd Safety check γi Remarks H ¼ Kh D ð10Þ

(kN·m) (kN·m) (kN) (Md/Mud) ≤ 1

N1 −6435 −14,300 0 0.50 OK

N1′ −2775 −5900 0 0.52 OK
Material non-linearity was considered in this process. Fig. 24 shows
C1 4700 6400 0 0.81 OK the obtained relation of the seismic coefficient with the displacements
N2 −9775 −14,300 0 0.75 OK in two directions. The seismic coefficients at the yield points are 2.34
N2′ −4230 −5900 0 0.79 OK and 3.05 in the longitudinal and transverse directions, respectively.
N3 −13,100 −15,300 0 0.94 OK
They are much higher than the design seismic coefficient of L1-EQ.
N3′ −4940 −6900 0 0.79 OK
C2 6160 7600 0 0.89 OK The seismic coefficients at the ultimate points are 3.37 and 4.04 in the
N4 −13,270 −15,300 0 0.95 OK longitudinal and transverse directions, respectively. They are also
N4′ −5070 −6900 0 0.81 OK much higher than the design seismic coefficient of L2-EQ. These results
P1 9910 142,000 −8800 0.07 OK show that the proposed structure has sufficient seismic strength and
P2 7860 142,000 −12,000 0.06 OK
ductility capacity in the longitudinal and transverse directions.
710 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713

Fig. 23. Proposed bridge 3D model with beam element.

7.3. Time history analysis where γi: structural factor (1.0 for seismic analysis), Vd: design shear
force, Vud: design shear resistance capacity. The results are presented
The proposed bridge is a multi-degrees of freedom structure and in Table 5, which show that the RC piers satisfy Eq. (11).
needs to be verified by dynamic analysis. The time history analysis
was then carried out for the proposed bridge model by the design seis-
mic wave for L1-EQ and L2-EQ [9]. Fig. 25 shows a typical design accel-
eration wave for L2-EQ.
The response at the pier top calculated by the time history analysis is
shown in Fig. 26 and Fig. 27. The maximum response values were
93.1 mm and 167.7 mm in the longitudinal and transverse directions,
respectively. Fig. 28 shows the hysteretic response of bending moment
vs. curvature of the SRC girder section near the support joint in the lon-
gitudinal direction with the maximum bending moment of 6730 kN·m
and the curvature of 0.00184 m−1. The hysteretic response exhibits al-
most linear. The hysteretic responses of bending moment vs. curvature
at the RC pier base (P3) in the longitudinal and transverse directions are
shown in Fig. 29 and Fig. 30, respectively. They show that elastic and
plastic behaviours and the area inside the oval shapes absorbed the seis-
mic energy.

7.4. Seismic performance verifications

The response of the RC pier obtained by time history analysis due to Fig. 24. Pushover results of proposed bridge model.
L2-EQ should satisfy the shear force verification and the restorability
criteria, according to the JSCE Specification for Concrete Structures
[20]. The shear force should satisfy the following equation.
 
γi Vd =Vyd ≤1:0 ð11Þ

Table 4
Foundation spring constant.

Degree of freedom Spring constant

Translation along Y-axis 4500 MN/m

Translation along X-axis 1500 MN/m
Translation along Z-axis 1500 MN/m
Rotation about X-axis 13,600 MN·m/rad
Rotation about Z-axis 30,000 MN·m/rad
Rotation about Y-axis 30,000 MN·m/rad
Fig. 25. The earthquake input for level-2 wave acceleration.
 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713 711

Fig. 26. Displacement-history curve in longitudinal direction.

Fig. 29. Hysteretic response at the bottom of P3 in the longitudinal direction.

Fig. 27. Displacement-history curve in transverse direction.

It is necessary to control the damage level according to the seismic

performance grade. Table 6 shows the restorability verification of the
bridge piers. Bending moment vs. rotation angle (M-θ) relationship of
the bridge pier is used to verify their restorability. The verification Fig. 30. Hysteretic response at the bottom of P3 in the transverse direction.
equation proposed by limit states design method can be written as in
Eq. (12).
of the proposed bridge for L1-EQ and L2-EQ, showing good seismic per-
formance and satisfy the design requirements.
γi θsd =θrd ≤1:0 ð12Þ

7.5. A new bridge system without cross beams

where, θsd: design response angle of rotation at pier base, and θrd: de-
sign limit angle of rotation at pier base. The design limit rotation angles
correspond to the yield point and the maximum point for SP-1 and SP-2,
respectively. Table 6 shows the restorability verification of the RC piers
Cross beams or floor beams are essential for conventional plate gird-
er bridges. They distribute the vertical loads such as the dead and live
loads among the main girders. Without cross beams the vertical loads
tend to concentrate on particular girders, which are connected with rel-
atively wide-span RC slabs. In addition, when transverse forces such as
earthquake and wind forces act on the plate girders, lateral torsional
buckling is likely to occur without cross beams because the steel web
is high and slender [21].
The proposed SRC girder bridge is different from the conventional
plate girder bride. The RC slab is supported by multiple H-girders and
the slab span is shorter than the plate girder bridge. The rolled
H-beam is a compact section and has higher torsional stiffness. In addi-
tion, the web height is much lower and composite with RC slab. These
structural characteristics are expected to enable the new SRC bridge

Table 5
Verification of shear strength by nonlinear dynamic history method.

Direction Designation Vyd (MN) Vd (MN) Vd Safety check

γi Vyd ≤1:0

Long. Pier (P2) 30.40 20.8 0.68 OK

Trans. Pier (P3) 36.75 33.5 0.91 OK
Fig. 28. Hysteretic response of SRC section in the longitudinal direction.
712 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713

Table 6 Mechanical interaction between the steel/concrete interfaces was

Restorability verification of the pier. modelled by surface-to-surface contact elements same as in Section 4.
Level Direction SP-1 SP-2 Mechanical properties and non-linear constitutive relations of steel
θsd θrd Verification θsd θrd Verification
and concrete are the same as in the design example. Different mesh
sizes were tried to find mesh sizes that provide reliable results and
L1-EQ Longitudinal 0.0018 0.0094 0.19 b 1
less computational time. It was found that a mesh size of 300
Transverse 0.0036 0.0167 0.22 b 1
L2-EQ Longitudinal 0.014 0.0579 0.24 b 1 × 300 mm and 300 × 141 mm (length by width of the element) provid-
Transverse 0.031 0.044 0.71 b 1 ed adequate accuracy in modelling the concrete slab and steel beam, re-
spectively. The total number of elements used in the developed FEM
model shown in Fig. 31 was 40,534.
stiffer against the vertical and transverse loads. Therefore, cross beams The horizontal static force corresponding to L1-EQ, 0.2 of the dead
may not be necessary and a new bridge system without cross beams loads, was applied at the concrete slab centre, the upper and lower
was proposed. flanges of the steel H-girder, and at the web centre along the entire gird-
The FEM model for the full SRC bridge with the RC slab and rolled H- er length. Geometrical non-linearity was also considered.
beam was developed using ABAQUS. The steel girder and the RC slab Fig. 32 shows the deformed shape with transverse displacement cal-
were modelled as 3D solid element (C3D8R). One interior span of the culated by the developed FEM model. The maximum displacement took
proposed SRC bridge was studied considering the symmetry, as shown place at the bottom of steel beam flange at the span centre. No signifi-
in Fig. 31. Displacements and rotations were all fixed at all the nodes cant deformation was observed. The transverse displacements at the
at the girder ends. The shear forces between the steel beam and the span centre obtained by the FE analysis and by the frame model
RC slab were transferred in the transverse direction by non-linear spring (Fig. 23) are shown in Fig. 33. The displacement by the frame model
element at the positions of shear connectors. In the vertical direction, agreed very well with that of the FEM model at the steel lower flange.
the RC slab and steel beam was connected by vertical rigid springs. The displacements at the slab centre and the upper and lower flange

10,500 mm

Concrete slab
(C3D8R)

Fixed boundary

Concrete web (C3D8R)

Steel beam (C3D8R)

Fig. 31. FEM model of full SRC bridge.

Fig. 32. Deformed shape with transverse displacement (mm).

 M.H. Elmy, S. Nakamura / Journal of Constructional Steel Research 138 (2017) 701–713
Fig. 33. Load-displacement relationship.
and the web centre all linearly increase with horizontal forces. The
maximum displacement obtained by the FEM and the frame model
was 1.45 mm, which was very small. These results show that the system
without cross beams are stable and feasible against transverse seismic
forces.
8. Conclusion
A new form of steel reinforced concrete (SRC) girder bridge using
the steel rolled H-section was proposed. The super-structure consisted
of continuous rolled steel H-girders which was composite with the RC
slab, and the sub-structure was RC piers. The girder and the pier were
rigidly connected by concrete and reinforcements to resist large nega-
tive bending moment at the intermediate support joints. This new
bridge was studied by experiments and FEM analysis. A design example
and seismic analysis were also conducted. The main conclusions obtain-
ed by this study are as follows.
1) Experiments were conducted with the partial SRC bridge model,
showing that the new SRC structural form had high bending
strength and ductile property. The applied load was smoothly trans-
ferred from the girder to the pier through the rigid joint, and the
cracks observed on the concrete surfaces were within the acceptable
limit.
2) Finite element model was developed and the analysed results were
compared with the experimental results. The displacement and
strains obtained by FEM agreed well with the test results, which ver-
ified the established FE model. The local behaviour of steel/concrete
interface at the rigid joint was also simulated with non-linear
springs, showing that the shear forces were smoothly and sufficiently
transferred.
3) A design example with a four-span continuous road bridge with the
span-length of 40 + 50 + 50 + 40 m was conducted, finding that the
proposed bridge satisfied the safety and serviceability requirements.
4) Seismic analysis was carried out by two methods, the push-over
analysis and the time history analysis, considering material non-
linearity. The proposed SRC bridge was proved to have sufficient
ultimate strength and ductility against ultra-strong earthquakes.
713
5) A new bridge system without cross beams was proposed for the new
SRC bridge because the girder height was low and composite with
the RC slab. The structural behaviours of this new bridge system
due to horizontal forces were studied by FEM analysis, showing
that the new system without cross beams were stable and feasible
against transverse seismic forces.
6) This study shows that the proposed steel reinforced concrete girder
bridge using rolled H-beam was structurally rational, feasible and
economical. The applicable span length could be extended to almost
double the existing H-girder bridge.
References
[1] S. Nakamura, T. Momiyama, T. Hosaka, K. Homma, New technologies of steel/
concrete composite bridges, J. Constr. Steel Res. 58 (2002) 99–130.
[2] M. Takagi, S. Nakamura, S. Muroi, An experimental investigation on rigid frame
steel-concrete composite girder bridge, JSCE, Journal of Structural Engineering 49
(32) (2003).
[3] E. Ellobody, Finite element analysis and design of steel and steel-concrete composite
bridges, first ed. Butterworth-Heinemann, USA, 2014.
[4] Xinpei Liu, A. Mark Bradford, Qing-Jun Chen, Huiyong Ban, Finite element modelling
of steel-concrete composite beams with high-strength friction-grip bolt shear
connectors, Elsevier, Journal of Construction Steel Research 108 (2016) 54–65.
[5] Francesco Bencardino, Antonio Condello, Luciano Ombers, Numerical and analytical
modelling of concrete beams with steel, FRP and hybrid RFP-steel reinforcements,
Elsevier, Journal of Construction Steel Research 140 (2016) 53–65.
[6] Huiyong Ban, Ee Loon Tan, Brian Uy, Strength of multi-span composite beams
subjected to combined flexure and torsion, Elsevier, Journal of Construction Steel
Research 113 (1–12) (2015).
[7] G. Vasdravellis, B. Uy, E.L. Tan, B. Kirkland, Behavior and design of composite
beams subjected to sagging bending and axial compression, Elsevier, Journal of
Construction Steel Research 110 (2015) 29–39.
[8] Weiwei Lin, Teruhiko Yoda, Experimental and numerical study on mechanical be-
haviour of composite girders under hogging moment, Adv. Steel Constr. 9 (4)
(2013) 309–333.
[9] Mahesan Bavan, Sharizan Bahrom, Azrul A. Mutalib, Siti Aminah Osman, Numerical
prediction of composite beam subjected to combined negative bending and axial
tension, J. Eng. Sci. Technol. 8 (4) (2013) 428–447.
[10] Prakash Amar, N. Anandavalli, C.K. Madheswaran, J. Rajasankar, N. Lakshmanan,
Three dimensional FE model of stud connected steel-concrete composite girders
subjected to monotonic loading, International Journal of Mechanics and Applications
1 (1) (2011) 1–11.
[11] Jun He, Yuqing Liu, Airong Chen, Dalei Wang, Teruhiko Yoda, Bending behaviour of
concrete-encased composite I-girder with corrugated steel web, Elsevier, Journal of
Construction Steel Research 74 (2014) 70–84.
[12] ABAQUS Business User's Manual, Version 6.142014.
[13] Japanese Society of Civil Engineers, Standard specification for concrete structures,
Design (2012).
[14] Japanese Society of Civil Engineers, Standard Specification for Hybrid Structures,
2014.
[15] J. Lubliner, J. Oliver, S. Oller, E. Onate, A plastic damage model for concrete, Int. J.
Solid Structure (1989).
[16] Japan Road Association, Specifications for Road Bridges, 2012.
[17] SAP2000 Structural Analysis Program. Computer and Structures, Inc. Advanced,
2016.
[18] C. Chung Fu, Shuqing Wang, Computational Analysis and Design of Bridge
Structures, CRC press, USA, 2015.
[19] Japanese Railway Research Institute, Design Specification for Railway Structures,
Maruzen, 2009.
[20] Japanese Society of Civil Engineers, Standard Specification for Concrete Structures,
Seismic Performance Verification, 2002.
[21] L.P. Carden, I.G. Buckle, A.M. Itani, Transverse displacement capacity and stiffness of
steel plate girder bridge superstructures for seismic loads, Elsevier, Journal of
Constructional Steel Research 63 (2007) 1546–1559.