Engineering Structures 30 (2008) 546–560

www.elsevier.com/locate/engstruct

Effect of lap reinforcement in link slabs of highway bridges

T. Charuchaimontri a , T. Senjuntichai a,∗ , J. Ožbolt b , E. Limsuwan a
a Department of Civil Engineering, Chulalongkorn University, Bangkok 10330, Thailand
b Institute of Construction Materials, University of Stuttgart, 70569 Stuttgart, Germany

Received 14 December 2006; received in revised form 21 March 2007; accepted 17 April 2007
Available online 31 May 2007

Abstract

This paper investigates the influence of lap reinforcement in link slabs of highway bridges under four independent boundary conditions by
using a three-dimensional nonlinear finite element code based on the microplane model. Numerical solutions for load–deflection relationships,
internal force distribution and failure cracking planes are presented for link slabs with different details of lap reinforcement. A full-scale test was
performed on three reinforced concrete long span link slabs with various lap reinforcement details subjected to mid-span loading. The comparison
indicates a good agreement between the results from finite element analysis and the experiment. The present model can be used to predict the
effective moment of inertia of the link slab under mid-span loading, end rotation and end translation for the development of design criteria for a
link slab.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Microplane model; Link slab; Reinforcement detailing; Effective moment of inertia; Finite element analysis; Highway bridge

1. Introduction slab was subjected to rotation due to thermal effects greater

Jointless construction of a multiple span bridge deck
of a simply supported girder may be considered as a
continuous pavement supported by the girders. It is generally
more effective than integral bridge construction practices [1].
Continuity of the deck can eliminate gaps and joints between
adjacent spans to assure smooth riding, noise reduction and
serviceability improvement. In addition, it reduces initial
and maintenance costs of joints, malfunction due to debris
accumulation and deterioration of bridge bearings due to
water leakage. A deck connecting the two adjacent spans is
called a link slab (see Fig. 1), and it can accommodate axial
deformation, end rotation and end translation from adjacent
spans. The ends of the link slab are subjected to vertical and
longitudinal movements due to the flexibility of elastomeric
bearing pads as well as longitudinal and rotational movements
due to girder deformation. The first link slab jointless bridge
in the state of North Carolina was studied with a remote data
acquisition system and monitored for over a year. The link
∗ Corresponding author. Tel.: +66 2 218 6460 3; fax: +66 2 251 7304.
E-mail address: fcetsj@eng.chula.ac.th (T. Senjuntichai).
than that due to traffic loads [2]. A total of 32 overpasses with
jointless bridge decks in the earthquake region during the 1999
Izmit earthquake in Turkey typically performed well because
dislocation of the girders was controlled by link slabs [3].
In addition, the seismic performance of multi-simple span
bridges retrofitted with link slabs was also investigated by
Caner et al. [4] who suggested that the overall response of
the highway bridge should be checked to evaluate the force
distribution due to the addition of link slabs.
Effects of contraction and expansion of a jointless deck
on the interaction forces between the deck and girders as
investigated by Zuk [5] showed that the stresses built up
in both the deck and girders in the longitudinal direction
were not extremely high and can be accommodated with
suitable materials. The finite element method was employed by
Gastal [6] to study the load–deflection response of a jointless
bridge deck. In his model, each bridge span was divided into
a number of isoparametric beam elements, except for the link
slab where a spring element with only axial stiffness was used.
Richardson [7] proposed a simplified design procedure based
on an elastic analysis for partially debonded and continuous
decks by assuming that the debonded portion of the deck was a
tension member. His study indicated that the support conditions

c 2007 Elsevier Ltd. All rights reserved.

0141-0296/$ - see front matter

doi:10.1016/j.engstruct.2007.04.015
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
(a) Long span link slab.
(b) Short span link slab.
Fig. 1. Link slabs in highway bridges.
had a great influence on stresses and potential cracking in the
deck. El-Safty [8] later modified the finite element program by
Gastal [6] and concluded that the optimum debonded length to
get the highest ultimate load is between 2% and 6% of the girder
span depending on the conditions of support and loading.
Subsequently, a design procedure for the link slab based
on analytical studies and a test program was presented [9,10],
in which each span of a bridge was considered as simply-
supported since the flexural stiffness of link slab is much
smaller than that of the girders. In addition, the debonding
of the link slab over the girders for a length equal to 5% of
each girder span would not alter the load–deflection behavior
of jointless bridge decks with simple-span girders. It was also
assumed that the link slab was subjected to flexure rather
than axial deformation and cracks occurred at the mid-span
section. If the moment in the link slab exceeded the cracking
moment, the moment of inertia of the cracked section would
be used at the middle part of the slab and the effective
moment of inertia for the debonded portion was computed
from Branson’s formula [11]. Recently, partial continuity due
to the axial stiffness of link slab and support configurations
was investigated by Okeil and El-Safty [12] using a modified
three-moment equation. The influence of span length ratio, slab
axial stiffness, and girder stiffness on the behavior of a two-
span jointless bridge was studied for roller and hinged supports.
It was found that the bottom of the girders at a hinged support
was prevented from longitudinal movement, the support then
developed higher continuity moments and tension forces in the
link slab.
Generally, two types of link slabs are constructed in highway
bridges and can be considered as long and short span link
slabs with respect to the gap between the two main girders (see
Fig. 1). A long span link slab is constructed over inverted T cap
beams and the span is long enough to form a low stiffness link
slab. For a short span link slab, the cap beam is rectangular in
shape and the space between the two main girders is narrow.
547
The link slab is then constructed with a certain amount of
debonding between the deck and girders to reduce its stiffness.
In practice, link slabs in highway bridges are often constructed
with various lengths of lap reinforcement. A review of the
literature indicates that the influence of lap reinforcement on the
behavior of a link slab has never been considered in the past. In
addition, the majority of existing studies have considered a link
slab as an axial member or a flexural member by using an elastic
theory with the assumption that a major crack occurs at mid-
span. However, the behavior of the link slab is predominantly
nonlinear due to cracks that are distributed over the entire
slab. Therefore, the effective moment of inertia of the link
slab to restrain the adjacent girder movements, the crack width
and crack pattern to control the serviceability, and the failure
mode for the development of a strut-and-tie model based design
can not be correctly predicted without an appropriate model.
If the end movements of adjacent girders are appropriately
controlled, the internal forces in a reinforced concrete link slab
can then be determined provided that the actual stiffness is
known. In addition, the relative stiffness can also be adjusted
to accommodate the internal force in the design of a link slab
according to the procedure shown in Fig. 2. In practice, a link
slab design is performed in two stages, one in a service level and
the other in a strength level by following existing design codes
(e.g. see [13,14]). Details on the link slab design are beyond the
scope of this paper.
In this paper, the behavior of long span link slabs,
which are normally constructed in Thailand, with various lap
reinforcement details are investigated under four independent
boundary conditions, namely axial deformation, end rotation,
end translation and mid-span loading with fixed-end supports.
In reality, a link slab is subjected to these boundary conditions
almost simultaneously due to interaction between adjacent
girders, the link slab and the support stiffness. To develop a
complete interaction model one needs to well understand the
link slab behavior under each condition. Work is currently
548 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
Fig. 2. Design procedure of a link slab.
underway to incorporate all boundary conditions for the
development of such a model. The three-dimensional finite
element code based on the microplane model, namely MASA,
is employed by taking into account nonlinearities in materials
and the fracture criteria. It is noted that although this model
is appropriate for the present problem, it is computationally
quite demanding for practical design of a link slab. The
numerical results on the load–deflection relationships, internal
force distribution, and failure cracking planes are presented
for three types of link slabs with different lap reinforcement
detailing. A full-scale test of three reinforced concrete link
slabs with various reinforcement details subjected to mid-
span loading was performed to verify the numerical model. A
long span link slab is chosen in this test as the behavior of
the short span link slab with debonded portion for different
lap reinforcement cannot be observed clearly under mid-span
loading. Note that the behavior of a short span link slab under
the other three conditions (axial deformation, end rotation and
end translation) is similar to that of the long span one. The
effective flexural stiffness of the link slab under mid-span
loading, end rotation and end translation is determined by using
Branson’s formula [11] for all types of reinforcement detailing.
The proposed effective stiffness of a link slab considering
cracks leads to more accurate prediction of restraint end
movement of highway girders and internal forces in the slab,
which is necessary in the design of a link slab.
2. Finite element analysis of a link slab
2.1. Finite element code
A finite element code MASA (MAcroscopic Space
Analysis) was developed at the Institute of Construction
Materials, University of Stuttgart. The code is based on the
microplane model and it can be used for analysis of two- and
three-dimensional structures made of quasi-brittle materials.
The smeared crack approach is employed and the constant or
secant stiffness method (CSM or SSM) is applied as a system
solver. The crack band approach is used in order to obtain mesh
independent results. In addition, the time-dependent behavior
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
Fig. 3. Microplane model.
(a) Concrete.
Fig. 4. Constitutive models employed in finite element analysis.
of concrete and other materials, such as creep and shrinkage,
can be simulated as well. Concrete can be discretized by
using 8-node or 4-node elements, whereas reinforcement is
modeled by using 2-node bar elements. A uniaxial elasto-
plastic stress–strain relationship can be employed with or
without strain hardening. Pre and post processes are performed
by using a commercial program FEMAP [15]. A number of
numerical studies performed so far have confirmed that the code
is capable of realistically predicting the behavior of reinforced
concrete structures [16].
The basic concept of the microplane model was first
described by Taylor [17] and further developed for quasi-brittle
materials by Bažant and co-workers [18,19]. The model is
aimed at analyzing damage and fracture phenomena in concrete
structures under a three-dimensional state of stresses. In the
microplane model, the material is characterized by a uniaxial
relationship between the stress and strain components on planes
of various orientations as shown in Fig. 3. These orientations
or planes may represent damage planes or weak planes in the
microstructure, such as contact layers between aggregates and
mortar in concrete.
To obtain mesh objective results for the nonlinear smeared
fracture finite element analysis of concrete structures one
needs a so-called localization limiter. Principally there are two
approaches, i.e. a crack band method and so-called higher order
methods (nonlocal continuum or Cosserat continuum). Both
approaches are implemented in the finite element code MASA
resulting in finite element analysis with mesh independence.
549
(b) Reinforcing steel.
2.2. Modeling of concrete material and reinforcing steel
The concrete material is modeled by using 8-node solid
elements. A comparison of the concrete behavior using a
microplane constitutive law and that from a stress–strain
relationship given by Hognestad [20] for uniaxial compression
is shown in Fig. 4(a). The microplane model implies
that Young’s modulus, Poisson’s ratio and 12 microplane
parameters have to be obtained such that the model fits the
macroscopic properties of concrete. In the finite element code,
the microplane model parameters are automatically generated
for given macroscopic concrete properties. It can be seen from
Fig. 4(a) that the microplane model realistically simulates both
pre-peak (hardening) and post-peak (softening) responses of a
concrete material.
The constitutive law for reinforcing steel is given by a
uniaxial stress–strain law as shown in Fig. 4(b). The yielding
strength f y , Young’s modulus E, hardening modulus E h and
strength f s have to be given as the input material parameters for
a bar element. It is also assumed that the constitutive laws for
reinforcing steel under tension and compression are the same in
the model.
The spatial finite element model (FEM) for a link slab is
shown in Fig. 5. Four boundary conditions are considered in
the numerical study, namely axial deformation, end rotation,
end translation and mid-span loading. In each case, loading
is applied by displacement control with boundary conditions
being prescribed as shown in Fig. 6 and a link slab is modeled
according to the symmetry of the problem. The reinforcing
550 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
(a) Link slab coordinate system.
Fig. 6. Boundary conditions for each loading case.
steel and the surrounding concrete are assumed to be perfectly
bonded in the analysis, i.e. no slip in reinforcement occurs. In
addition, both vertical and horizontal loads are applied over
an elastic plate, which is discretized by using 8-node solid
elements with linear elastic behavior, as shown in Fig. 5(b).
Three link slabs with different lengths of lap reinforcement,
namely fully continuous (LS 183), semi continuous (LS 025)
and mid-span hinge (LS 000), are considered in the numerical
study as shown in Fig. 7. For a semi continuous slab, the length
of lap reinforcement or loop joint is chosen to be 250 mm
following Ryu et al. [21] who found that the mechanical
behavior of the proposed loop joint is similar to that of an
ordinary slab without overlapping reinforcement if the loop
joints are enough in terms of the development length concept
for the anchorage of hook reinforcement [13,14]. The size of
the link slab in numerical examples was selected according to
practical constructions [22]. The support of the link slab for
the mid-span loading case is assumed to be fixed because the
stiffness of the girders is much higher than that of the link
slabs [9]. In addition, the supports are designed to produce both
positive and negative bending moments as well as inflection
points in the link slab to study the effect of lap reinforcement
detailing on the internal force distributions along the span
length of the slab.
2.3. Finite element solutions
Finite element solutions for load–deflection relationships
and crack patterns of all lengths of lap reinforcement for link
slabs under axial deformation, end rotation, end translation and
mid-span loading are presented in Figs. 8–11 respectively.
(b) Finite element discretization of a link slab (quarter of a specimen).
Fig. 5. Model of link slabs.
The numerical results shown in Fig. 8 indicate that the length
of lap reinforcement has a significant influence on the crack
pattern (crack distribution) and the cracking load of a link slab
under axial deformation. In a mid-span hinge slab (LS 000) the
first crack, which is also the widest one, is observed at the saw
cut. On the other hand, cracks in the other two slabs (LS 025
and LS 183) are distributed over the entire slab. It is also found
that the LS 000 slab yields the minimum cracking load whereas
the cracking loads for the LS 025 and LS 183 slabs are almost
the same. After the slab is cracked, the slab stiffness shows
a notable dependence on the stiffness of reinforcing bars that
have enough development length along the span, as shown in
the load–deformation curves in Fig. 8(a).
The crack patterns under end rotation in Fig. 9 show a similar
trend to those observed in the slabs under axial deformation.
Cracks in a LS 025 slab start from the support end and mid-
span sections to the quarter-span section, whereas for a LS 183
slab, cracks start from the support end to the mid-span sections.
In addition, the widest cracks in both slabs are found in the
earliest cracks. Similar to what is observed in axial deformation
case, the cracking and ultimate moments of LS 000 are lower
than those of the LS 025 and LS 183 slabs because of a
weak section at the saw cut. After cracking, the slab stiffness
once again depends significantly on the lap reinforcement
details.
Numerical results shown in Fig. 10 indicate that the cracking
moment, crack distribution and stiffness of link slabs under end
translation show negligible dependence on the length of lap
reinforcement since the maximum moment in this case occurs
at the support where all types of slabs have the same section
properties.
For link slabs with fixed-end supports under mid-span
loading in Fig. 11, the crack distribution near the support
(negative moment cracks) is the same for all reinforcement
detailing. However, the cracking pattern near the mid-
span section (positive moment cracks) shows a significant
dependence on the length of lap reinforcement. It is found that
a LS 183 slab has the smallest crack width followed by the
LS 025 and LS 000 slabs respectively. Additional results for
the mid-span loading case are presented in the next section.
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560 551

(a) Mid-span hinge (LS 000).

(b) Semi continuous (LS 025).

(c) Fully continuous (LS 183).

Fig. 7. Three types of reinforcement details.

(a) Load–deformation curves. (b) Crack distribution.

Fig. 8. Numerical results for the axial deformation case.

3. Model verification finite element analysis. Only the mid-span loading case
was simulated and the length of the lap reinforcement was
3.1. Test set-up the key parameter in the test. The quantities that were
measured from the test included crack distribution, crack width,
An experimental program of full-scale testing was carried load–deflection relationship and ultimate load capacity. Three
out at Chulalongkorn University to verify the results from link slab specimens, with thickness of 200 mm, span length of
552 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560

(a) Load–deformation curves. (b) Crack distribution.

Fig. 9. Numerical results for the end rotation case.

(a) Load–deformation curves. (b) Crack distribution.

Fig. 10. Numerical results for the end translation case.

2000 mm and width of 1000 mm as shown in Fig. 7, were tested
to failure under a central point load from an actuator of 1000 kN
capacity. To represent the AASHTO HS20-44 wheel load, the
load was transmitted over a rectangular steel plate of 500 mm ×
200 mm × 20 mm [23]. The characteristic compressive strength
of concrete was 30 MPa and the yield strength of reinforcing
steel was 400 MPa. Table 1 shows actual dimensions of test
specimens with their corresponding concrete and reinforcement
steel strength as well as the reinforcement details.
The test set-up is shown in Fig. 12. Electrical resistance
strain gauges with a length of 5 mm were fixed across the
mid and edge sections of the link slab at both top and bottom
reinforcements in order to measure the tensile and compressive
strains. Out-of-plane deflections at four points along the mid-
section, at one-sixth span and at both supports were measured
by using linear variable deflection transducers (LVDT) as
shown in Fig. 12(b). The test was performed right after the
concrete reached the specified strength, and the LVDT, load cell
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560 553

Table 1
Properties of test specimens

Specimens Average thickness (mm) Level of reinforcement Average Reinforcement Lap length Top Bottom
(mm) upper/lower concrete yield stress (mm) reinforcement reinforcement
strength (MPa) (MPa)
Left Middle Right Left Middle Right
LS 000 202 203 202 162/32 168/35 164/32 23.3 553 0 DB16@100 DB16@100
LS 025 200 200 205 165/31 165/38 160/31 26.1 575 250 DB16@100 DB16@100
LS 183 203 205 203 168/30 169/33 169/32 33.3 576 1830 DB16@100 DB16@100

(a) Load–deformation curves. (b) Crack distribution.

Fig. 11. Numerical results for the mid-span loading case.

and strain gauge signals were the input for the computerized
data acquisition system.
The specimens were gradually loaded up to 10 kN and the
load was subsequently released to ensure that the loading edges
remained in proper contact with the specimens. Thereafter,
the specimens were subjected to two cyclic loads. First, they
were cyclically loaded to 200 kN with a 10 kN increment to
observe cracking load and tension stiffening behavior before
and after cracking. A cyclic of 93 kN loading, which represents
the factored load for the AASHTO HS20-44 wheel load (plus
30% impact), was then applied for 5 times to observe the link
slab behavior under the service level. Thereafter, they were
loaded with a 50 kN increment until a substantial increase in
displacement was observed to establish a base line curve and
to detect specified crack widths. Finally, the specimens were
loaded until they failed.
3.2. Test results
The first cracks of both LS 025 and LS 183 slabs were
observed simultaneously at the mid-span and near the supports,
whereas for the LS 000 slab, the first crack was found only at
the mid-span. Cracking loads for specified crack width obtained
from the experiment are summarized in Table 2. It is also
observed that the base line curves of each loading cycle (from
0 to 93 kN) are similar for all lengths of lap reinforcement. The
reinforcing steel did not yield before shear failure took place.
The LS 000 slab acts as a cantilever slab with horizontal
compression at the upper part of the mid-span section. Before
failure, a horizontal crack occurs at the uppermost side of the
mid-span section due to the compressive force, and an inclined
crack is found at the support ends. These two cracks are the
main causes of the failure. For the LS 025 slab, it acts as a
continuous slab until the flexural shear crack occurs because
the positive moment cannot be fully developed due to the
insufficient double lap reinforcement. The weak section is at
the location of lap reinforcement, the failure cracking plane is
then close to the loading pad and lap reinforcement. For the
LS 183 slab, the positive moment can be fully developed due to
double reinforcement at the mid-span section. In addition, the
failure crack has an axis of symmetry at the half length between
the mid-span and end-span, and horizontal cracks are clearly
observed at both the end-span and mid-span sections. Failure
occurs due to either crushing of the reduced concrete section
at the bottom of the end-span and the top of the mid-span in
all slabs, called shear compression failure, or crack splitting of
the concrete cover along the reinforcing steel in the LS 025 and
LS 183 slabs.
554 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560

(a) Loading arrangement.

(b) LVDT position.

Fig. 12. Test set-up.

Table 2
Cracking and failure loads from the experiment

Specimens Load at specified crack width (kN) Failure load (kN) Deflections at failure (mm)
First crack 0.2 mm 0.3 mm 0.4 mm Experiment FEM Experiment FEM
LS 000 30 60 90 150 635.9 655.2 10.89 6.86
LS 025 110 300 400 700 809.6 908.2 7.28 6.69
LS 183 120 400 700 a 777.8 824.7 3.73 4.52
a The specimen failed before this crack width.

The results also indicate that the inclined tie angle and
the position of failure plane for each reinforcement detailing
are different. In the LS 025 and LS 183 slabs, the angle is
approximately 30◦ , which is an angle that usually occurs in a
prismatic beam. However, the failure plane of the LS 183 slab
is located at the quarter-span of the specimen but that of the
LS 025 slab is close to the mid-span. In addition, the LS 000
slab shows inclined angles at the cracking planes of 10◦ , 20◦ ,
and 30◦ at the mid, quarter and end of the span respectively.
Because the applied loading is not a line load, the crack across
the section is not uniform. It is very difficult to detect the first
inclined crack since it occurs inside the slab and the inclined
crack angle observed at the side surface is just lower than that
which occurs inside.
The shear strength of a reinforced concrete slab or beam
without shear reinforcement provided by an uncracked concrete
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560 555

(a) LS 000. (b) LS 025.

(c) LS 183.

Fig. 13. Comparison of tensile stress in reinforcing steel at end-span and mid-span between FEM and experiment.

section, aggregate interlock and dowel bar was developed in LS 025 and LS 000 slabs. At the same load, curvatures at
ACI318-05 [13] and is given by the end-span and mid-span of the LS 000 slab are higher
  than those of the LS 183 and LS 025 slabs. These results
p
0
Vu d bw d p confirm that the locations of inflection points and curvatures
Vc = f c + 120ρw ≤ 0.3 f c0 bw d (1)
Mu 7 depend significantly on the length of lap reinforcement. Since
the reinforcement for the positive bending moment in the
where ρw = As /bw d, As is the area of nonprestressed tension
LS 183 slab is heavier than the negative moment reinforcement,
reinforcement, d is the effective depth of the section, bw is
the inflection point then moves toward the supports. On the
the width of the beam web. From the test results, failure
contrary, the reinforcement at the mid-span section of the
loads of the LS 000 and LS 183 slabs are about 25% higher
LS 000 slab cannot carry the bending moment since the slab
than the shear strength due to the beam behavior, whereas the
behaves after cracking as a cantilever slab, and the inflection
failure load of the LS 025 slab is about 40% higher because of
point then moves to the mid-span. Therefore, the LS 000
the arch behavior. The lateral restrained link slab exhibits an
slab yields the largest deflection followed by the LS 025 and
internal arching action that strengthens the link slab through
LS 183 slabs respectively. The behavior of the LS 025 slab
the confinement provided by a complex three-dimensional
is somewhere between the LS 000 and LS 183 slabs, i.e. it
membrane stress around the loaded area. As a result, the slab
behaves like a semi continuous slab.
fails due to shear rather than flexure.
The deflections of the test specimen were measured at four 3.3. Comparison between finite element analysis and test
equally-spaced points from the support to the mid-span. These results
values can be used to predict the deflected curves, the inflection
points and curvatures of the slab provided that the deflected Failure loads and corresponding mid-span deflections from
curve is approximated by a third-order parabola. It is found the full-scale test are presented in Table 2 and they are
that the inflection points for the LS 183, LS 025 and LS 000 compared with the finite element solutions. Comparisons
specimens occur at 0.18L, 0.22L and 0.27L from the support of tensile stresses in reinforcing steel at the end-span (top
respectively where L is the span length of the link slab. The reinforcement) and the mid-span (bottom reinforcement) for all
inflection points determined from finite element analysis are reinforcement details between finite element and experimental
found at 0.22L, 0.23L and 0.25L respectively for the LS 183, results are also shown in Fig. 13. Both results agree well for the
556 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
top reinforcement stresses at the end-span for all reinforcement
details because the locations of the strain gauges were close
to the estimated crack positions. However, the stresses in
the bottom reinforcement at mid-span from FEM and the
experiment are quite different due to the discrepancy in the
locations of the strain gauges and random cracks as well as
the deformed curve of the experiments under the loading pad.
The numerical results presented in Fig. 13(a) also indicate
that the stresses in reinforcement at the mid-depth across the
saw cut obtained from the test of the LS 000 specimen are
much higher than the finite element results. The reason for this
inconsistency is that the depth of the compression zone at the
mid-span section from the finite element analysis is about twice
as large as that from the experiment. Therefore, the mid-span
section of the LS 000 slab from the present FEM is stiffer, and
a different FEM that can simulate a very large concentrated
crack is recommended for the analysis of a link slab with no
lap reinforcement.
Fig. 14 presents a comparison between the maximum crack
widths estimated from reinforcement stresses due to finite
element analysis and those measured in the experiment. The
maximum surface crack width at each loading step from finite
element analysis is estimated by using the AASHTO [23] and
ACI Building Code [13], which is given by
p
wcr = 0.000076β f s 3 dc A (2)
where wcr is the crack width in inches; β is the ratio between
the distance from the extreme tension fiber to the neutral axis,
and the distance from the centroid of the main reinforcement
to the neutral axis, and it can be taken as 2 [24]; f s is the
rebar stress in ksi; and dc is the concrete cover from the
extreme tension fiber to the centroid of the nearest rebar level.
In addition, A is the effective area per reinforcing bar in in2 .
It can be seen from Fig. 14 that a good agreement between
both results is obtained for the LS 025 and LS 183 slabs.
The maximum crack width measured from the LS 000 slab is
different from that estimated from finite element analysis due to
the discrepancy in the level of the neutral axis at the mid-span
section discussed in the foregoing paragraph. Nevertheless, the
difference between the maximum crack widths in the LS 000
slab estimated from the FEM and the experiment is acceptable
for practical applications. Therefore, the FEM presented in this
study can be used to develop the serviceability limit state for a
link slab. The advantage of making a saw cut in the LS 000 slab
is to generate a concentrated crack in one location, which can
be easily repaired, instead of several cracks. The drawback is
that the variation of stress in reinforcement across the saw cut
is so high that it has to be carefully monitored to ensure that the
stress variation is within the acceptable limit to prevent fatigue
failure.
The crack distribution and the direction of failure cracking
planes from the experiment and finite element analysis are
presented in Fig. 15(a) and (b) respectively. Those numerical
results can be used to verify the inclined angle of a tie in the
strut-and-tie model, in which the angle of 60◦ for a flexural
member without transverse reinforcement is proposed [25]. The
angle of inclined tie is perpendicular to the direction of the
cracking plane. A good agreement between the two results is
observed for all specimens as shown in Fig. 15(a) and (b).
A comparison of the load–deflection curves in Fig. 16
indicates a close agreement between the experiment and finite
element results for the LS 025 and LS 183 slabs. However
for the LS 000 slab, in addition to the different depth of the
compression zone at the mid-span section between the two
results as previously discussed, the tension stiffening is partially
lost resulting in lower stiffness in the test specimen than the
FEM when the load is sustained or applied repeatedly. This is
due to the fact that the bending moment cannot be redistributed
in a determinate structure after a crack is formed at the mid-
span of the LS 000 slab.
3.4. Effective moment of inertia of link slabs
Under service loading, the section properties of a link slab
change from an uncracked section to a fully or severely cracked
section since cracks will normally occur in the link slab.
Determination of the effective moment of inertia for a link
slab is necessary to accurately predict the deflection of the
slab and the restraint force from adjacent girders. Branson [11]
presented a well-known expression to compute the average
effective moment of inertia for a simply-supported rectangular
or T-beam under uniformly distributed loading as follows:

Mcr m
  
Mcr m
 
Ie = Ig + 1 − Icr . (3)
Ma Ma
For a fixed-end beam, the above equation becomes
Ie = 0.70Im + 0.15 (Ie1 + Ie2 ) (4)
where Icr and Ig are the moments of inertia of the completely
cracked and uncracked sections respectively; Im , Ie1 and Ie2
are the effective moments of inertia at the mid-span and at both
ends respectively; and Iexp is the experimental value of Ie . In
addition, Mcr and Ma are the cracking moment and the applied
service moment respectively. The value of the exponent m in
Eq. (3) is given by
h i
I −Icr
log Iexp −I
m= h
g cr
i . (5)
Mcr
log Ma
The suggested value of m in Eq. (3) due to Branson [11]
is 3. For a load level greater than 1.5Mcr , the reduction in the
value of m is very small. Therefore, an accurate prediction of
Ie can be computed from Eq. (3) with m = 3 provided that
the loading level is greater than 1.5Mcr . If the loading level
is less than 1.5Mcr , the value of m = 3 can still be used but
the estimated value of Ie should be reduced by 10% [26]. In
this study, the range of loading level (Ma /Mcr ) to determine
the value of m is varied from 1.5 to 4.0. Note that the value
of Ie also depends on the reinforcement ratio, especially for
heavily reinforced concrete members, and the cracked length
of the member. Those effects are not considered in this study.
In addition, the models mentioned above are valid when the
section property is uniform along the span length, which is not
the case under consideration in this paper.
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
(a) LS 000.
(c) LS 183.
Fig. 14. Comparison for the maximum crack width between the FEM and experiment.
In the case of a beam with fixed-end supports subjected to a
central point load, the effective moment of inertia is calculated
from the following equation
P L3
E Iexp = (6)
192∆
where P and ∆ denote the magnitude of applied loading and the
mid-span deflection respectively. For a given load–deflection
diagram, the effective moment of inertia from either the FEM
or experimental results can be calculated by using Eq. (6). The
effective moment of inertia can also be obtained from Eqs.
(3) and (4) according to Branson [11] for given values of the
weight factors and the exponent m. The appropriate values of
the weight factors and the exponent m from both FEM and the
experimental results are determined by using a least squares
method with the initial values of (0.30, 0.70) and 3 respectively
for the weight factors and m. It is found that the difference
between the effective moments of inertia computed from Eqs.
(3) and (4), and that from Eq. (6) is within 10%. According
to Eq. (4), the weight factors are 0.3 and 0.7 for negative and
positive bending moments respectively. However, the weight
factors obtained by applying Eq. (4) to the experimental results
are (0.23, 0.77), (0.34, 0.66) and (0.39, 0) for the LS 183,
LS 025 and LS 000 slabs respectively. These values are in
good agreement with the weight factors computed from finite
557
(b) LS 025.
element analysis, which are (0.22, 0.78), (0.31, 0.69) and (0.34,
0) for the LS 183, LS 025 and LS 000 slabs respectively. The
weight factor for a negative bending moment of the LS 183
slab is less than that of the LS 025 slab since the length of
lap reinforcement at the mid-span of the LS 183 slab is much
longer. As a result, the mid-span section of the LS 183 slab
is stiffer and the negative bending moment zone is then shorter.
Note that the weight factor for a positive moment of the LS 000
slab is zero since this slab behaves like a cantilever slab.
The value of the exponent m in Eq. (3) was originally
proposed to be 3 by Branson [11]. Al-Zaid et al. [26] later
suggested that this value should be 2.8, 2.3 and 1.8 for a beam
under uniformly distributed loading, third-point loading and
mid-span concentrated loading respectively. Since the mid-span
loading in this study is applied through a loading plate, which
is not an actual point load, the proposed value for m is then
varied from 1.8 to 2.3. It is found that the optimal values of m
from the experimental results are found to be 2.5, 2.7 and 0.9
for the LS 183, LS 025 and LS 000 slabs respectively, whereas
from the finite element analysis, they are 2.3, 2.5 and 0.7 for the
LS 183, LS 025 and LS 000 slabs respectively. This implies
that the suggested value of m between 1.8 and 3 [11,26] can not
be used for the LS 000 slab, which has no lap reinforcement
and a large crack is concentrated at the mid-span section.
Three different depth-to-span ratios, i.e. d/L = 200/2000,
300/2000 and 300/1000, which are widely employed in
558 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560

(a) Experiment.

(b) Finite element analysis.

Fig. 15. Crack pattern and failure planes.

Fig. 16. Comparison of load–displacement curves between the FEM and experiment.

practical constructions, are considered for the LS 183, LS 025 moment of inertia (Ie ) since it is difficult to set up and control
and LS 000 slabs under end rotation and end translation. Only these two types of boundary conditions in the full-scale testing.
the finite element method is employed to determine the effective In the case of end rotation, the exponent m for the effective
 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
(a) End rotation.
Fig. 17. Estimated values of the exponent in Eq. (5) from the FEM.
moment of inertia is calculated from the following equation
Ma L
E Iexp = (7)
2θ
where θ is the end rotation angle and Ma in Eq. (7) is the
moment due to reduced stiffness and end rotation, which is
different from the mid-span loading case where Ma is computed
from the applied load and it does not depend on stiffness. It is
found that the estimated values of m are equal to 0.5, 0.8 and
0.3–0.5 (depending on the depth-to-span ratio of the link slab)
for the LS 183, LS 025 and LS 000 slabs respectively as shown
in Fig. 17(a) and the difference between the effective moment
of inertia computed from Eq. (7) and that from Eq. (3) is less
than 10%. Caner [24] studied a link slab under end rotation by
using the same assumptions as given in the references [9,10].
He suggested that the effective stiffness can be computed from
Branson’s formula [11], Eq. (3), with m = 1. However, the
influence of lap reinforcement was not considered in his study.
Finally, the exponent m for the translational loading case is
given by
Ma L 2
E Iexp = (8)
6∆
where ∆ is the support settlement and Ma is the moment due
to reduced stiffness and end translation. The estimated value
of m in this case is varied from 0.70 to 0.90 for all lengths of
lap reinforcement (see Fig. 17(b)) and the difference between
the effective moment of inertia computed from Eq. (8) and that
from Eq. (3) is less than 10%. This is due to the fact that the
maximum bending moment for translational loading is close to
the ends of the slabs, where the section properties are identical
for all types of reinforcement detailing. In addition, the bending
moment at the mid-span, in which the section properties for
each link slab are different, is zero.
4. Conclusions
The influence of reinforcement detailing in the link slab for
highway bridges under end movements and applied loading
is presented in this paper. The behavior of link slabs with
different lengths of lap reinforcement is investigated by using a
three-dimensional finite element code based on the microplane
559
(b) End translation.
model. Nonlinearities in materials and fracture criteria are
also considered in the model. It is found that the stiffness,
crack distribution and crack width in a link slab depend
significantly on reinforcement detailing. Good agreement is
obtained between the finite element analysis and the full-scale
test of long span link slabs under mid-span loading, except for
the LS 000 slab where no lap reinforcement is used and a large
concentrated crack occurs at the mid-span section. The internal
force distribution can also be obtained from the inflection point,
crack distribution and stress distribution in reinforcement. The
effective moment of inertia of the link slab under mid-span
loading, end rotation and end translation is determined based on
the formula given by Branson [11]. It is found that the values of
the weight factors and exponent m recommended for predicting
the effective moment of inertia can be used for a link slab with
adequate length of lap reinforcement under mid-span loading.
However, for a link slab that behaves like a cantilever slab,
the appropriate values should be carefully chosen. In addition,
the exponents m for the effective moment of inertia for end
rotation of the LS 000, LS 025 and LS 183 slabs are different,
whereas for the end translation case, the value of m does not
depend on the lap reinforcement details. The results from the
present study are thus very useful in the development of a
design procedure of a link slab. Generally, the restrained forces
from link slabs and highway girders as well as the serviceability
and strength limits of the link slab must be known in the design
of a link slab. These restrained forces can be estimated from
end movement of girders associated with the link slab stiffness
and other stiffnesses. For a certain amount of end movement of
the girders, the link slab can be designed according to the crack
width control for serviceability. However, if the movements are
too large to control the crack width, a saw cut in the link slab
(LS 000) may be the best solution.
Acknowledgements
We would like to thank TRF (Thailand Research Fund)
for the financial support of the Ph.D. study in Chulalongkorn
University and DAAD (Deutscher Akademischer Austausch
Dienst) for financial support of the 6-month experience at the
University of Stuttgart, Germany.
560 T. Charuchaimontri et al. / Engineering Structures 30 (2008) 546–560
References
[1] Alampalli S, Yannotti AP. In-service performance of integral bridges and
jointless decks. Transportation research record. 1998. Paper no. 98-0540.
[2] Wing KM, Kowalsky MJ. Behavior, analysis, and design of an
instrumented link slab bridge. J Bridge Eng ASCE 2005;10(3):331–44.
[3] Youd LT, Bardet JP, Bray JD. Kocaeli, Turkey, earthquake of August 17,
1999 reconnaissance report. Earthquake Engineering Research Institute
2000. Earthquake Spectra. 2000–03.
[4] Caner A, Dogan E, Zia P. Seismic performance of multisimple-span
bridges retrofitted with link slabs. J Bridge Eng ASCE 2002;7(2):85–93.
[5] Zuk W. Jointless bridges. Research report no. FHWA/VA.81/84. Virginia
Highway and Transportation Research Council; 1981.
[6] Gastal F. Instantaneous and time-dependent response and strength
of jointless bridge beams. Ph.D. dissertation. Department of Civil
Engineering, North Carolina State University; 1986.
[7] Richardson DR. Simplified design procedures for the removal of
expansion joints from bridges using partial debonded continous decks.
Master’s thesis. Department of Civil Engineering, North Carolina State
University; 1989.
[8] El-Safty AK. Analysis of jointless bridge decks with partially debonded
simple span beams. Ph.D. dissertation. Department of Civil Engineering,
North Carolina State University; 1994.
[9] Zia P, Caner A, El-Safty AK. Jointless bridge decks. Research report no.
FHWA/NC/96-006. Raleigh (NC): Center for Transportation Engineering
Studies, North Carolina State University; 1995.
[10] Caner A, Zia P. Behavior and design of link slab for jointless bridge decks.
PCI J 1998;43(3):68–80.
[11] Branson DE. Instantaneous and time-dependent deflections of simple and
continuous reinforced concrete beams. Alabama highway research report.
Alabama Highway Department. Bureau of public roads report no.7 Part 1;
August 1963 (1965).
[12] Okeil A, El-Safty A. Partial continuity in bridge girders with jointless
decks. Pract Period Struct Des Constr J ASCE 2005;10(4):229–38.
[13] ACI 318-05. Building code requirements for structural concrete and
commentary and notes. 1999.
[14] DIN 1045. Plain, reinforced and prestressed concrete structures. 2001.
[15] FEMAP. Software package for finite element modeling and postprocess-
ing version 8.1. Exton: Structural Dynamics Research Corp.; 2001.
[16] Ožbolt J. MASA3, Finite element program for 3D nonlinear analysis
structures, MASA manual. Institute for Construction Materials, Univer-
sity of Stuttgart; 2002.
[17] Taylor GI. Plastic strain in metals. J Inst Metals 1938;62:307–24.
[18] Ožbolt J, Li J, Kožar I. Microplane model for concrete with relaxed
kinematic constraint. Internat J Solids Structures 2001;38(16):2683–711.
[19] Bažant ZP, Ožbolt J. Nonlocal microplane model for fracture, damage and
size effect in structures. J Eng Mech ASCE 1990;116(11):2485–504.
[20] Hognestad E. A study of combined bending and axial load in reinforced
concrete members. Bulletin series no. 399. University of Illinois
Engineering Experimental Station; 1951.
[21] Ryu HK, Kim YJ, Chang SP. Experimental study on static and fatigue
strength of loop joints. Eng Struct 2007;29(2):145–62.
[22] Goris A. Beitrag zum Tragverhalten von Federplatten bei
Strassenbrücken. Doktor-Ingenieurs dissertation: Der Rheinisch-
Westfälischen Technischen Hochschule Aachen: Aachen; 1985 [in
German].
[23] AASHTO Highway Subcommittee on Bridges and Structures. Standard
specifications for highway bridges. 16th ed. Washington (DC): AASHTO;
1996.
[24] Caner A. Analysis and design of jointless bridge decks supported by
simple-span girders. Ph.D. dissertation. Department of Civil Engineering,
North Carolina State University; 1996.
[25] Reineck KH. Ultimate shear force of structural concrete members without
transverse reinforcement derived from a mechanical model. ACI Struct J
1991;88(5):592–602.
[26] Al-Zaid RZ, Al-Shaikh AH, Abu-Hussein MM. Effect of loading type on
the effective moment of inertia of reinforced concrete beams. ACI Struct
J 1991;88(2):184–90.