Journal of Constructional Steel Research 65 (2009) 1751–1762

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

journal homepage: www.elsevier.com/locate/jcsr

Steel–concrete composite bridge deck slab with profiled sheeting

Hyeong-Yeol Kim ∗ , Youn-Ju Jeong
Structural Engineering Department, Korea Institute of Construction Technology, Gyeonggi-Do 411-712, Republic of Korea

article info abstract

Article history: This paper presents an experimental study of a steel–concrete composite bridge deck slab with profiled
Received 22 September 2008 sheeting and perfobond shear connectors. Two full-scale deck slab specimens cast onto three concrete
Accepted 20 April 2009 blocks were fabricated and tested under static loading to examine the ultimate load-carrying capacity of
the proposed deck slab system under sagging and hogging bending actions. The ultimate behaviour of
Keywords: the full-scale deck slab specimens is also compared with that of simply supported deck specimens under
Steel–concrete composite
hogging bending only. In addition, the load–deflection behaviour of the proposed deck system is compared
Bridge deck
Perfobond rib shear connector
with that of a reinforced concrete (RC) deck slab. The test results indicate that the ultimate load-carrying
Profiled sheeting capacity of the proposed deck system is at least 220% greater than that of the RC deck system and that
Experiment the deck weighs about 23% less than the RC deck system. The paper summarizes the test results, findings,
and recommendations for future study.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction
Various types of deck slab systems have been developed and
used for girder bridges, though a full-depth cast-in-place (CIP)
reinforced concrete (RC) deck remains the most common system
due to its economical advantages. Stay-in-place steel forms are
often used for CIP RC decks because a full-depth CIP RC deck takes
a relatively long time to erect. The use of steel forms saves a
significant amount of time during the construction of an RC deck
slab for girder bridges.
Steel forms are generally profiled and made of light-gauge steel
sheets with indentations or embossments. If the composite action
between the form and the concrete can be obtained, the forms can
partially act as a tensile reinforcement and the cross-sectional area
of the deck can be somewhat reduced. However, the composite
action is generally neglected in the design of RC bridge deck slabs
because the indentations or embossments in the forms fail to
provide the desired level of horizontal shear resistance between
the form and the concrete under a live load.
Jeong et al. [1] proposed a steel–concrete composite deck
system with profiled steel sheeting for girder bridges. In such a
system, the interaction between the sheeting and the concrete
generally governs the strength and behaviour of the deck. Headed
studs were used as shear connectors in the study of Jeong et al.,
and the studs were welded onto the profiled sheeting. However,
the results of push-out and flexural tests indicate that a horizontal
∗ Corresponding author. Tel.: +82 31 910 0582; fax: +82 31 910 0578.
E-mail addresses: hykim1@kict.re.kr (H.-Y. Kim), yjjeong@kict.re.kr
(Y.-J. Jeong).
0143-974X/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2009.04.016
slip occurs between the sheeting and concrete under a service live
load.
Although headed studs are the most common type of shear
connector, they may be unsuitable for profiled sheeting because of
need to weld the studs to a thin steel plate. If the studs have a large
diameter, they may cause a welding problem during fabrication.
On the other hand, if the studs have a small diameter, a greater
number of studs must be used. Furthermore, the studs welded onto
thin steel plate may cause a fatigue problem during the service life.
The perfobond rib shear connector may be an alternative shear
connector that can be used with profiled sheeting. This type of
shear connector consists of a steel plate with a number of uni-
formly spaced holes. If the holes are filled with concrete, con-
crete dowels are formed; and the dowels provide horizontal shear
resistance between the sheeting and the hardened concrete. The
perfobond rib shear connector has several potential advantages:
they are easy to customize and fabricate; they help overcome the
fatigue problem; and a single perfobond rib can replace a number
of headed studs.
In 2006 Kim and Jeong [2] used perfobond rib shear connectors
for a composite deck slab with profiled sheeting and experimen-
tally investigated the behaviour of the deck slab. The type of com-
posite deck slab that they proposed for girder bridges had a longer
span but weighed less than a typical CIP RC deck slab.
More recently, 16 deck specimens were tested with different
shear span lengths for the purpose of evaluating the horizontal
shear capacity of the proposed deck system. The horizontal shear
capacity of the proposed deck system was evaluated by means of
the empirical design method described in [3], and the test results
are briefly summarized in [4]. The estimated horizontal shear
capacity of the proposed deck was identified as being at least two
1752 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762
Fig. 1. Schematics for the proposed deck profile.
Fig. 2. Cross-section of a typical precast PSC I-girder bridge.
times greater than the required horizontal shear strength. The test
results [2,4] indicate that the perfobond rib shear connection can
be effectively used for the proposed deck system.
This study elaborates the work of [2] with a series of test
programs. This paper presents the ultimate behaviour of a
steel–concrete composite deck slab designed for a prototype of a
precast prestressed concrete (PSC) I-girder bridge. The design is
based on the deck profile proposed in [2].
The clear slab span length of the proposed deck slab was
assumed to be 2.5 m, whereas that of a typical CIP RC deck slab for a
PSC I-girder bridge is normally less than 2.2 m. Two full-scale deck
slab specimens cast onto three concrete blocks were fabricated and
tested under static loading to examine the ultimate load-carrying
capacity of the proposed deck slab specimens under sagging and
hogging bending actions. The test results are compared with the
results of a finite element (FE) analysis of a full-length bridge. The
ultimate behaviour of the full-scale deck slab specimens is also
compared with that of the simply supported deck specimens tested
in [4]. In addition, the load–deflection behaviour of the proposed
deck system is compared with the behaviour of the RC deck slab
specimens tested in this study. The test results and findings are
summarized in this paper.
The use of the proposed deck system in bridge construction
may be costly unless consideration is given to the savings in
construction time and maintenance costs. The proposed deck
system is roughly estimated to cost at least 1.6 times more than CIP
RC decks but it may be competitive with CIP RC decks under certain
circumstances. The user cost might be reduced, for example, when
the proposed deck system is used in a deck replacement project
for a high-traffic bridge. If the user cost is considered, the use of
the proposed deck system in construction may be a less expensive
choice because the expected erection time of the proposed system
is about half that of RC deck system.
As discussed in [2], the proposed deck system has several
advantages over conventional CIP RC decks but also a few
disadvantages. During service, the condition of the deck, especially
the concrete, cannot be inspected visually. In addition, the profiled
sheeting must be protected from steel corrosion. For practical
applications, there is also a need for a simple but effective sheet-to-
sheet connection method to be developed because the size of the
profiled sheeting is limited due to fabrication and transportation
requirements.
2. Proposed deck system
2.1. Deck profile
Fig. 1 shows a schematic of the steel–concrete composite bridge
deck system proposed in [2]. The deck consists of concrete, steel
reinforcements, perfobond rib shear connectors, and profiled steel
sheeting set perpendicular to the girders. Because the profiled
sheeting serves as a tensile reinforcement in the design of the deck
slab, no reinforcing bars are required at the bottom.
Fig. 2 shows a cross section of a prototype of a precast PSC I-
girder bridge selected for the design of the proposed composite
deck slab system. The clear slab span length (L) was assumed to
be 2.5 m, which is approximately 1.2 times longer than a typical
CIP RC deck slab for a precast PSC I-girder bridge.
The standard design truck load specified in the Korean Bridge
Design Code [5] is used as the design live load. According to
the code, the gross weight of the design truck is 43 200 kg.
The self-weight of the deck slab together with the 50 mm-thick
asphalt wearing surface is used for the dead load. The ultimate
strength method was used for the deck design. Note also that the
overall design process is driven by the strength limits because
the horizontal shear capacity of the deck is obtained from test
information.
Fig. 3 shows cross-sectional dimensions of the proposed
composite deck profile. The dimensions of the profiled steel
sheeting were determined on the basis of structural optimization.
The design procedure for the deck profile is briefly summarized
 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762 1753

Fig. 5. Test set-up for a composite deck specimen.

Fig. 3. Cross-sectional dimensions of a deck profile (units: mm).

Fig. 4. Dimensions of a perfobond rib.

in [2]. In this study, the dimensions of the profiled sheeting are

assumed to be the same as those of the ‘Deck-A’ type in [2], though
the depth of the deck was increased from 200 to 220 mm. Fig. 6. Load versus mid-span displacement curves for the specimens tested in [4].

2.2. Perfobond rib

In the mid-80s, Zellner [6] endeavoured to overcome the fatigue

problem of studs under a live load by introducing a new type
of shear connector, known as a perfobond rib. Several studies
have been successfully conducted on composite deck slabs of
steel bridges with a perfobond rib shear connection [7–11]. More
recently, a T-shaped perfobond shear connector was developed for
building structures [12–15]. However, the existing studies focus on
a composite deck slab that rests on steel beams or frames.
To date, no design information is available for composite decks
with perfobond ribs welded onto profiled sheeting. Existing studies
by other groups [6–15] on the perfobond rib shear connection
indicate that the horizontal shear capacity of perfobond ribs is
mainly influenced by the diameter and spacing of the holes. The
hole diameter (D) is generally assumed to be about half the height
of the perfobond rib. The hole spacing that produces the maximum
shear resistance varies from 2D to 2.5D. Fig. 7. Regression model of the test data.
A series of push-out tests was conducted to evaluate how the
hole spacing within a perfobond rib influences the horizontal shear The horizontal shear capacity of a composite deck can be
resistance [4]. In light of the test results and findings, the hole estimated by an empirical design method known as the m–k
spacing of 2D was selected for the perfobond ribs for the proposed method [3,16]. The m–k method involves a series of flexural tests
deck system. Fig. 4 shows the dimensions of a perfobond rib shear with different shear span lengths until the deck specimen fails as
connector used for the proposed deck system. a result of horizontal shear failure. In this method, an empirical
design relation for the horizontal shear capacity of the composite
2.3. Horizontal shear strength deck is obtained from test data.
Eurocode 4 [3] specifies that the design shear resistance of a
In [4], a total of 16 full-scale deck specimens were tested composite deck with profiled steel sheeting, VRd , can be computed
with different shear span lengths to evaluate the horizontal shear as
capacity of the proposed deck system. Eight of the specimens had
VRd = bdp [(mAp /bLs ) + k]/1.25 (1)
a clear span length (L) of 2.5 m, and the other eight had a length of
3.5 m. Highlights of the test results for the specimens with a clear where b is the width of the deck, dp is the distance from the top
span length of 2.5 m are briefly discussed below. of the deck to the centroid of the effective area of the profiled
The shear span length (Ls ) of the specimens was selected as sheeting, and Ap is the effective area of the profiled sheeting.
L/7.1, L/4.0, L/3.0, and L/2.0, where L = 2.5 m. As illustrated in The characteristic values of m and k in Eq. (1), which can be
Fig. 5, Ls is the distance from a loading point to a support. Fig. 6 obtained from the graphical relation illustrated in Fig. 7, are 0.2716
presents the load versus mid-span vertical displacement curves and 0.00052, respectively. In Fig. 7, V is the vertical shear force.
for the deck specimens with different shear span lengths and also From Eq. (1), the minimum value of VRd for the proposed deck
shows the number of specimens tested for each shear span length. system is computed as 0.92 N/mm2 , which is approximately two
1754 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762

Fig. 8. Test set-up for the RC deck specimens: (a) RC-1 and (b) RC-2.

times greater than the required horizontal shear strength of the Table 1
deck slab. Summary of the test results.
Specimen ID Ls Pu (kN) Pcr (kN) δu (mm) δs (mm) Failure mode
2.4. Comparison of composite and RC decks PF-3 [4] L/4.0 1,554 360 20.2 1.6 Shear
RC-1 L/4.0 696 390 14.8 1.3 Shear
For a comparison of the load–deflection behaviour of the PF-6 [4] L/2.0 863 150 21.8 1.7 Shear–flexural
proposed deck system with that of a conventional CIP RC deck slab, RC-2 L/2.0 375 50 51.6 3.7 Flexural

two 1000 mm × 2700 mm RC deck specimens were fabricated and

tested. The clear span length (L) of the specimens was 2.5 m. The
depth of the RC deck specimens was 250 mm, and the 16 mm-
diameter reinforcing bars were uniformly spaced at 220 mm and
125 mm in the longitudinal direction for the top and bottom
reinforcements, respectively. Used as a transverse reinforcement
to control cracking, the 16 mm bars were uniformly spaced at
250 mm. The mean yield strength of the reinforcing bars was
431 MPa for the 16 mm bars and 412 MPa for 19 mm bars. The
design strength of the normal density concrete was 27 MPa, and
the compressive strength of the air-cured concrete at the time of
the test was 26.1 MPa.
Fig. 8 shows the test set-up of the RC deck specimens, for
which Ls was assumed to be L/4 for specimen RC-1 and L/2 for
specimen RC-2. For specimen RC-1, two equal concentrated loads
were applied through 230 mm × 580 mm rigid blocks. The size of
the rigid blocks was selected to simulate the tire-contacting area
of a wheel load of the design truck [5]
Two linear variable displacement transducers (LVDTs) were Fig. 9. Comparison of test data from the specimens of the PF and RC series.
placed at the mid-span and quarter-span sections of the specimen
so that the vertical displacement of the specimen could be
measured at different loading stages. A number of strain gauges lists the test results of the RC deck specimens; it also lists the
were mounted on different locations of the specimen to measure results of the composite deck specimens tested in [4]. In the table,
the strains. A vertical monotonic loading with a displacement Pu is the ultimate load, Pcr is the initial concrete cracking load, and
control of 0.02 mm/s was applied to the specimens by means of δu and δs are the mid-span vertical displacement at the ultimate
a servo-controlled actuator with a 3500 kN capacity. load level and service load level, respectively. The service live load
The location of the neutral axis at the mid-span section of the RC is a rear wheel load of the design truck [5]; the wheel load, which
deck specimens was estimated on the basis of the strains measured includes a dynamic load allowance factor of 0.3, is 122.3 kN.
at the top and bottom reinforcing bars. The neutral axis of the RC Fig. 10 shows the failure mechanisms of the RC deck specimens.
deck specimens is about 200 mm from the bottom. In contrast, Specimen RC-1 shows a linear load–displacement behaviour up to
the neutral axis of the composite deck specimens tested in [4] is the point of failure. As the load is increased, a number of flexural
located in the concrete slab and has a mean value of about 90 mm cracks occur at the bottom of the concrete. Furthermore, as the load
from the bottom. approaches the failure load level, a diagonal shear crack spreads
For a load–deflection behaviour comparison of the proposed progressively from the support towards the top of the concrete at
deck system and the RC deck system, Fig. 9 shows a plot of the the loading point near the rigid block. The ultimate load is obtained
load versus mid-span vertical displacement curves of the deck just before the failure, which is a brittle shear mode of failure. The
specimens with the same shear span length. In addition, Table 1 design vertical shear strength of the RC deck specimens is 224 kN.
 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762
Fig. 10. Failure mechanisms of the RC deck specimens: (a) RC-1 and (b) RC-2.
Fig. 11. Load versus strain curves of the specimens of the PF and RC series.
Specimen RC-2, on the other hand, shows a typical flexural
mode of failure. As illustrated in Fig. 11, the failure of specimen RC-
2 is associated with the yielding of tensile reinforcement. The yield
load level of the tensile reinforcement of specimen RC-2 is 236 kN.
Under a service load level of 122.3 kN, the response of specimen
PF-6 [4] is approximately five times stiffer than that of specimen
RC-2.
The composite deck specimen weighs 23% less than the RC
deck specimen. However, regardless of the shear span length, the
RC deck specimens have at most only 45% of the ultimate load-
carrying capacity and 33% of the initial concrete cracking load of
the composite deck specimens tested in [4].
Concrete topping slab
a 220
Block
600 1 Concrete
block
50
700 2,500
b
Screw jack
Fig. 12. A full-scale deck slab specimen: (a) dimensions in millimeters and (b) the testing fixture.
1755
3. Full-scale deck slab test
3.1. Fabrication of the specimens
Two full-scale deck slab specimens were fabricated and tested
to evaluate the ultimate strength of the proposed deck system
under sagging and hogging bending actions and to experimentally
validate the deck-to-girder connection.
Fig. 12(a) shows a 1000 mm × 7600 mm full-scale deck slab
specimen that was continually cast onto three concrete blocks. The
concrete blocks were used to simulate a precast concrete I-girder,
and each block was reinforced with 16 mm bars. The size of the
concrete block covers the full width of the top flange and about
one-fourths of the full depth of the web section of the concrete
I-girder. Fig. 13 shows the reinforcement details of the concrete
block.
Due to the symmetry of the superstructure, approximately two-
thirds of the superstructure of the bridge shown in Fig. 2 was
modeled as a three-girder system. As illustrated in Fig. 12(b), the
three concrete blocks are horizontally locked up with screw jacks
and floor-anchored rigid blocks to prevent them from moving
horizontally during the casting and testing of the specimens. Note
however, as explained in Section 3.4, this test set-up cannot reflect
the actual behaviour of a deck-on-girder system.
Fig. 14(a) shows the simple but effective deck-to-girder
connection that was designed for the proposed deck system. The
profiled sheeting with the perfobond rib shear connectors, which
is shown in Fig. 14(b), sits on the concrete block-outs that formed at
both corners of the top flange of the concrete block with a bearing
length of 75 mm.
The dimensions of the profiled sheeting and the perfobond rib
shear connectors are the same as those presented in Figs. 3 and
200
Block Block
2 Profiled 3
sheeting
700 2,500 700 450
7,600
Rigid block Anchor
1756 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762

75 550 a
a
50
180

100
650

320 b

280

6@125
b
Fig. 15. Reinforcement details of a full-scale deck slab specimen.

140
Fig. 15(b), 16 mm bars (H16-2) are uniformly spaced at 250 mm as
100 a transverse reinforcement to control the cracking. The use of the
perfobond rib as a chair for the reinforcement obviates the need to
140 use conventional chairs during the erection stage.
As shown in Fig. 15(b), the 16 mm bars (H16-1) pass through
140
every second hole of the perfobond ribs to reinforce the concrete
dowels. Fig. 14(b) shows that two 22 mm bars are placed at both
1000 ends of the perfobond ribs to provide end-anchorage. The end
anchorage was used because the test results in [4] indicate that the
Fig. 13. Reinforcement details for a concrete block (units: mm): (a) front view and use of the end-anchorage can effectively increase the horizontal
(b) side view. shear resistance of the composite deck.
In the longitudinal direction of the specimen, the top reinforce-
ment over the composite section is extended to the cantilevered
a overhang. As a means of resisting the compressive stresses induced
by the sagging moment of the deck, the 19 mm bars that are uni-
formly spaced at 125 mm are also placed for the bottom reinforce-
ment in the cantilevered concrete overhang section. Because the
profiled sheeting serves as a stay-in-place form, no form work is
needed for the composite sections between the concrete blocks,
though a conventional plywood form is used for the cantilevered
overhang.
The pouring of normal density concrete with a design strength
of 30 MPa, a slump of 120 mm, and entrained air of 5.1% began
b Center bars after the placement of the reinforcements. The mean compressive
Perfobond rib strength of the air-cured concrete at the time of the test, as
evaluated in three cylinder tests, is 33.1 MPa.

Anchor bars
3.2. Test procedure

Profiled sheeting
Fig. 14. Fabrication of a full-scale deck slab specimen: (a) deck-to-girder
connection and (b) profiled sheeting with perfobond ribs.
4, respectively. An SS400-grade mild steel plate with a thickness
of 6 mm was profiled through a press braking process to form the
profiled sheeting. The perfobond rib was fabricated by first cutting
a 6 mm thick steel plate and then cutting holes in the plate with
a plasma jet machine. The perfobond rib was welded onto the
profiled sheeting with an automatic welding machine.
Fig. 15(a) shows that 19 mm bars (H19) are uniformly spaced at
125 mm as the primary reinforcement. At the top layer, as shown in
Fig. 16 shows the test set-ups for the full-scale deck slab speci-
mens, and Fig. 17 shows the two loading schemes used in the test.
The specimens have identical dimensions but are named PGF-1
and PGF-2 to distinguish the loading schemes. Both specimens are
subjected to a three-point load; however, the shear span length (Ls )
of the composite deck sections is assumed to be L/4 for specimen
PGF-1 and L/2 for specimen PGF-2. Thus, the behaviour of the deck
slab under a hogging and sagging bending action can be compared
with that of the simply supported deck slab specimens tested in [4].
Fig. 17(a) also shows the locations of the strain gauges and
LVDTs. Three LVDTs (D1, D2, and D3 in Fig. 17(a)) are placed
underneath the specimens to measure the vertical displacements
of the deck slab at different loading stages. As shown in Fig. 17(a),
a number of strain gauges are mounted on the bottom flange of the
profiled sheeting (S1 and S2).
The load is incrementally applied to the specimens in a two-
stage of loading step. In the first stage, loads P1 and P2 are
 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762 1757

Fig. 16. Test set-up of the specimens: (a) PGF-1 and (b) PGF-2.

Fig. 17. Instrumentation and loading scheme of the specimens: (a) PGF-1 and (b) PGF-2.

simultaneously applied to the specimen until concrete cracking
occurs, either at the top of the cantilevered overhang or the
concrete deck section above the concrete block. In the second stage
of loading, load P1 maintains the final load level of the first stage
of loading but load P2 is continually increased until the specimen
fails.
Fig. 16 shows that load P2 was applied through a rigid spreader
beam. Note also that the load was applied through a line over the
full width of the deck to prevent the possibility of the specimens
overturning during the test, especially at the failure stage. In
addition, vertical monotonic loadings with a displacement control
of 0.02 mm/s were independently applied to the specimens by
means of two servo-controlled actuators with a 3500 kN capacity.
3.3. Test results and discussion
Fig. 18 shows the failure mechanism of specimen PGF-1, and
Fig. 19 shows the failure mechanism of specimen PF-3, which was
1758 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762
(a) Cantilevered overhang.
(b) Section above Block 2.
Fig. 18. Failure mechanisms of specimen PGF-1 (Ls = L/4.0).
Fig. 19. Failure mechanisms of specimen PF-3 (Ls = L/4.0) tested in [4].
tested in [4] with the same shear span length as specimen PGF-1.
Fig. 18(a) shows that the first concrete crack occurs on top of the
concrete slab above the concrete block (Block 3 in Fig. 12(a)) at
P1 = 250 kN. The tip displacement at the cantilevered overhang is
6.12 mm at the initial cracking load.
When load P2 reaches about 450 kN, hairline cracks occur at the
bottom of the concrete within the shear span (Fig. 18(b)). These
cracks spread progressively towards the top of the concrete at the
loading point as load P2 is further increased.
The cracking pattern within the composite section of specimen
PGF-2 (Fig. 18(a)) is quite different from that of the simply
supported specimen PF-3 (Fig. 19) tested in [4]. Because specimen
PF-3 is simply supported its ultimate behaviour is mainly affected
by flexural and vertical shear strength of the composite deck. On
the other hand, the failure of the composite section in specimen
PGF-2 is characterized by the development of a number of diagonal
shearing cracks which appear in the concrete at an angle of
approximately 45 degrees. This outcome may be due to the fact
that the hogging bending of the composite section in specimen
PGF-2 is restrained by the concrete sections above the concrete
block whereas specimen PF-3 is a rotation-free condition at the
supports. Thus, the ultimate behaviour of the composite section in
specimen PGF-1 is mainly affected by the vertical shear strength of
the deck slab.
Fig. 20 shows the failure mechanism of specimen PGF-2, and
Fig. 21 shows the failure mechanism of specimen PF-6, which was
tested in [4] with the same shear span length as specimen PGF-2.
The first concrete crack occurs on top of the concrete slab above
the concrete block (Block 3) at P1 = 290 kN. The tip displacement
at the cantilevered overhang is 2.62 mm at the initial cracking load.
When load P2 reaches about 640 kN, flexural cracks occur at
the bottom of the concrete underneath the loading point. When
load P2 approaches the ultimate load level, diagonal shearing
cracks develop within the shear span of the composite section
and spread progressively towards the top of the concrete at the
loading point. The cracking pattern within the composite section
of specimen PGF-2 (Fig. 20(a)) is similar to that of the simply
supported specimen PF-6 (Fig. 21). The ultimate behaviour of the
composite section in specimen PGF-2 is affected by the flexural
and shear strength of the composite deck. The tensile stresses
that develop in a diagonal direction within the shear span of the
composite section cause diagonal shear cracking of the concrete.
During and after the test, a concrete section at the deck-
to-girder connection was visually inspected and no visible
deformation or crack was observed. The test results of the full-scale
deck slab specimens indicate that the bending and shear stresses
that developed in the composite section are effectively transferred
to the concrete blocks through the deck-to-girder connection.
The load–deflection behaviour of the composite deck section
under hogging and sagging bending actions was quantitatively
analyzed on the basis of the load versus vertical displacement
curves of the slab specimens. The load–vertical displacement
curves are presented in Fig. 22 for specimens PGF-1 and PGF-2 and
for the simply supported composite deck specimens with the same
shear span lengths. The vertical displacements presented in Fig. 22
are the displacements measured at the mid-span of the composite
section between Block 2 and Block 3.
 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762
(a) Section between Block 2 and Block 3.
(b) Cantilevered overhang.
Fig. 20. Failure mechanisms of specimen PGF-2 (Ls = L/2.0).
Fig. 21. Failure mechanisms of specimen PF-6 (Ls = L/2.0) tested in [4].
Fig. 22. Load versus mid-span displacement curves of the specimens of the PGF
and PF series.
Specimen PGF-1 showed linear load–displacement behaviour
up to 400 kN. After a number of cracks occurred at the bottom of the
concrete within the shear span, the stiffness of the composite deck
slab gradually decreased and the deck showed nonlinear behaviour
until failure. Under the service load level (122.3 kN), the composite
section in specimen PGF-1 gives a response that is 4.9 times stiffer
than the response of specimen PF-3 with the same shear span
length. On the other hand, the ultimate strength of the composite
deck slab of specimen PGF-1 is 8% less than that of the simply
supported deck (PF-3). This lower level of ultimate strength is due
to the fact that the ultimate strength of the deck slab with a shear
1759
span of Ls = L/4 is mainly affected by the vertical shear strength
of the deck slab.
Specimen PGF-2 also shows linear load–displacement beha-
viour up to 360 kN. Under the service load level, the composite
section in specimen PGF-2 gives a response that is 3.1 times
stiffer than the response of specimen PF-6 with the same shear
span length. The ultimate strength of the composite deck slab of
specimen PGF-1 is 160% greater than that of the simply supported
deck (PF-6). This greater level of ultimate strength is due to the fact
that the ultimate strength of the deck with a shear span of Ls = L/2
is mainly affected by the flexural strength of the deck slab.
Fig. 23 shows the load versus strains curves of specimens
PGF-1 and PGF-2 together with those of the simply supported
deck specimens tested in [4] with the same shear span lengths.
The strains were measured with a strain gauge (S2 in Fig. 17(a))
mounted at the bottom flange of the profiled steel sheeting. The
tendency of the measured strains correlates fairly well with the
load and vertical displacement curves of the specimens. Note that
the yield strain of the profiled sheeting is 1795 micro strains. The
curves of the measured strains indicate that the failure of the
composite section in specimen PGF-1 occurs without any yielding
of the profiled sheeting. On the other hand, the yielding of the
profiled sheeting accelerates the failure of the composite section
in specimen PGF-2.
3.4. Finite element analysis
The ultimate behaviour of the proposed deck system was
experimentally evaluated by means of a failure test of the full-
scale deck slab specimens. However, the test set-up described in
1760 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762

Table 2
Comparison of the test data and the results of the FE analysis.
Description Gauge no. PGF-1 PFG-2
Test FE Test FE

Vertical displacement (mm) D1 0.59 0.15 1.25 0.28

D2 1.27 0.10 1.50 0.22
D3 6.12 0.37 2.62 0.42
Stress, σT (MPa) S1 15.34 4.79 47.12 16.77
S2 10.22 4.88 42.74 16.87
Stress, σL (MPa) S1 – 1.47 – 2.75
S2 – 1.34 – 2.61

Figs. 26 and 27 show the contours of the normal stresses induced

Fig. 23. Load versus strain curves of the specimens of the PGF and PF series.
Section 3.1 cannot reflect the actual behaviour of a deck-on-girder
system because the test set-up prevented any vertical deflections
or rotations of the girders. Furthermore, although an actual deck
slab was subjected to a biaxial bending moment, the deck slab of
the specimen was stressed only in the longitudinal direction of
the specimen. Thus, an FE analysis was conducted on a full-length
hypothetical bridge to simulate the actual behaviour of the deck-
on-girder system, and the results of the full-length FE analysis are
compared with the test data.
Fig. 24 shows an FE model of a 30 m-long hypothetical bridge.
The cross-sectional dimensions of the deck slab are assumed to
be the same as those of specimens PGF-1 and PGF-2. A 2 m-deep
precast PSC I-girder in typical drawings [17] is used to model the
concrete girders.
The commercial package LUSAS [18] was used for the FE
analysis. In the FE model, the profiled sheeting without any
perfobond rib shear connectors was discretized by a four-node
shell element (QSI4) whereas the concrete girder and slab were
discretized by an eight-node continuum element (HX8M). The
loading conditions used in the FE analysis are the same as those
used in the test. For the purpose of comparison with the test data,
the value of load P1 is assumed to be 250 kN for specimen PGF-
1 and 290 kN for specimen PGF-2 whereas a design live load of
268.3 kN is assumed for load P2.
Fig. 25 shows the deflected shape of the bridge at the mid-
span section with the same loading conditions as specimen PGF-1.
in the profiled sheeting and the concrete topping slab near the
loading points, respectively. Table 2 lists the vertical displacement
and the induced stresses of the deck slab at the mid-span section
of the bridge. The locations of the LVDTs and strain gauges are
the same as those shown in Fig. 17(a). In Table 2, σL and σT
are the normal stresses induced in the bottom flange of the
profiled sheeting in the longitudinal and transverse directions
of the bridge, respectively. The vertical net displacement of the
deck slab is approximately calculated by subtracting the vertical
displacements of the girders from the total displacements.
As expected, the deflection and rotation of the girders
significantly influence the behaviour of the deck slab. The vertical
displacements of the deck slab obtained from the full-length FE
analysis are at most 24.7% of the test results, regardless of the
loading conditions. The full-length FE analysis shows that the
normal stress values (σT ) in the transverse direction are about
31.2%–47.8% of the normal stresses measured in the tests. The test
results are conservative compared to the results of the full-length
FE analysis. This outcome is due to the fact that, in the full-length
FE model, the load applied to the deck slab can be transferred to
the girders. On the other hand, the deck slab in the test specimens
must support all the applied loads because rigid girders are used in
the test.
The results of the full-length FE analysis confirm that the normal
stress values (σL ) developed in the longitudinal direction are at
least 27.4% of σT with the loading conditions of specimen PGF-1
and 15.5% of σT with the loading conditions of specimen PGF-2.
This outcome is due to the fact that the test set-up does not capture
the biaxial bending behaviour, which was predicted by the full-
length FE analysis. Thus, a full-length bridge must be analyzed or
tested to evaluate the behaviour of an actual bridge deck of a deck-
on-girder system.

Fig. 24. An FE model of a full-length bridge.

 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762 1761

Fig. 25. Deflected shape of the superstructure at mid-span (PGF-1).

(a) Stress in transverse direction. (b) Stress in longitudinal direction.

Fig. 26. Stress contour of the profiled sheeting (PGF-1).

(a) Stress in transverse direction. (b) Stress in longitudinal direction.

Fig. 27. Stress contour of the concrete topping slab (PGF-2).

4. Conclusion
The behaviour of a steel–concrete deck slab system with
profiled steel sheeting and perfobond rib shear connectors was
experimentally and numerically investigated in this study. The
conclusions from the test results are summarized below.
The load–deflection behaviour of the proposed deck system
is compared with that of a CIP RC deck system. The test results
of the simply supported deck slabs indicate that the ultimate
load-carrying capacity and initial concrete cracking load of the
proposed deck system is at least 220% and 230% greater than the
corresponding values of an RC deck slab, respectively, while the
proposed deck system weighs about 23% less than the RC deck
system.
The consequent reduction in the dead load of the deck slab may
lead to lighter superstructures and reduced foundation loads. If the
design of a long span deck is considered, the lighter deck weight
and higher initial cracking load can contribute to the potential
advantages of the proposed deck system over a conventional RC
deck.
Two full-scale deck slab specimens that were continually cast
onto three-concrete blocks were tested under sagging and hogging
bending actions. Under the service load level, the response of the
full-scale deck slab specimens is at least three times stiffer than
that of the simply supported deck slab specimens, regardless of
the shear span length. The test results confirm that, if the failure
is associated with the flexural strength of the deck, the ultimate
strength of the proposed deck under sagging and bending actions
is at least 1.6 times greater than that of the simply supported
deck slab under a hogging bending action alone. However, if the
failure is associated with the vertical shear strength of the deck,
the difference is insignificant.
The vertical displacements and normal stresses measured in
the full-scale deck specimens are many times smaller than those
1762 H.-Y. Kim, Y.-J. Jeong / Journal of Constructional Steel Research 65 (2009) 1751–1762
obtained by a full-length bridge analysis, regardless of the loading
conditions. This outcome is due to the fact that the test set-up used
in the present study prevents any vertical deflections or rotations
of the girders.
References
[1] Jeong YJ, Kim HY, Kim SH. Partial-interaction analysis with push-out tests.
J Construct Steel Res 2005;61:1318–31.
[2] Kim HY, Jeong YJ. Experimental investigation on behaviour of steel–concrete
composite bridge decks with perfobond ribs. J Construct Steel Res 2006;62:
463–71.
[3] Eurocode 4. prEN 1994. Design of composite steel and concrete structures.
European Committee for Standardization (CEN). 2005; 1994.
[4] Jeong YJ, Kim HY, Koo HB. Longitudinal shear resistance of steel–concrete
composite slabs with perfobond shear connectors. J Construct Steel Res 2009;
65:81–8.
[5] Ministry of Land, Transportation, and Maritime Affairs. Highway bridge design
code. Seoul, Korea; 2008 [in Korean].
[6] Zellner W. Recent designs of composite bridges and a new type of
shear connectors. Composite construction in steel and concrete. In: Proc.
engineering foundation conference. 1987. p. 240–52.
[7] Veldanda MR, Hosain MU. Behaviour of perfobond rib shear connectors: Push-
out tests. Can J Civil Eng 1992;19:1–10.
[8] Oguejiofor EC, Hosain MU. Behaviour of perfobond rib shear connectors in
composite beams: Full-size tests. Can J Civil Eng 1992;19:224–35.
[9] Oguejiofor EC, Hosain MU. A parametric study of perfobond rib shear
connectors. Can J Civil Eng 1994;21:614–25.
[10] Klaiber FW, Wipe TJ. An alternate shear connector for composite action. In:
Proc. mid-continent transportation symposium 2000. 2000.
[11] Valente I, Cruz PJS. Experimental analysis of perfobond shear connection
between steel and lightweight concrete. J Construct Steel Res 2004;60:465–79.
[12] Vellasco PCG, Andrade SAL, Ferreira LTS, Lima LRO. Semi-rigid composite
frames with perfobond and T-rib connectors Part 1: Full scale tests. J Construct
Steel Res 2007;63:263–79.
[13] Andrade SAL, Vellasco PCG, Ferreira LTS, Lima LRO. Semi-rigid composite
frames with perfobond and T-rib connectors Part 2: Design models
assessment. J Construct Steel Res 2007;63:280–92.
[14] Vianna JC, Costa-Neves LF, Vellasco PCG, Andrade SAL. Experimental
assessment of perfobond and T-perfobond shear connectors. J Construct Steel
Res 2009;65(2):408–21.
[15] Vianna JC, Costa-Neves LF, Vellasco PCG, Andrade SAL. Structural behaviour
of T-perfobond shear connectors in composite girders: An experimental
approach. Eng Struct 2008;30(9):2381–91.
[16] Johnson RP. Composite structures of steel and concrete. 2nd edition Oxford:
Blackwell Scientific Publications; 1994.
[17] Ministry of Land, Transportation, and Maritime Affairs. Highway bridge design
manual: Typical drawings. Kim HY, editor. Seoul (Korea). 2008 [in Korean].
[18] Finite Element Analysis Ltd. LUSAS reference manual. Ver. 14.2–3. Surrey, (UK).
2008.