Vous êtes sur la page 1sur 45

1

Stiffness and Deflection of Steel-Concrete Composite Beams under Negative Bending


By Jianguo Nie,
1
Jiansheng Fan,
2
C. S. Cai
3

Subject Headings: girder bridges, deformation, stiffness, slip, studs
ABSTRACT: Compared with simply supported beams, continuous steel-concrete composite beams
have many advantages such as higher span/depth ratio, less deflection and higher fundamental frequency
of vibration due to its higher stiffness. However, in negative bending regions near interior supports,
tension in concrete is unfavorable and a complicated issue, which deserves a special study. In this paper, a
mechanics model based on elastic theory was established to investigate the stiffness of composite beams
in negative bending regions by considering slips at the steel beam-concrete slab interface and
concrete-reinforcement interface. In order to validate this approach, a test of three (3) composite beams
with profiled sheeting under negative bending was conducted. Meanwhile, a three-dimensional nonlinear
finite element analysis was conducted to investigate the general behavior of the tested specimens. In
addition, a comparative analysis between results derived from the analytical model, laboratory test, and
FE analysis was performed. The results show that slip always exists for composite beams under negative
bending even with complete shear connection (full composite action) between the steel and concrete
components. The slip effect results in an additional curvature of beam bending and reduces the section
stiffness by 10% to 20% compared with that of a beam without any slip in serviceability condition. This
reduction should be considered in designing process especially for cantilever beams. Formulae under
other loading cases and boundary conditions were also proposed. The results can serve as a basis for
further study on stiffness of continuous steel-concrete composite beams and can directly be used for the

1
Prof., Struct. Engrg. Res. Lab., Dept. of Civil Engineering, Tsinghua University, Beijing, China, 100084
2
Lect., Struct. Engrg. Res. Lab., Dept. of Civil Engineering, Tsinghua University, Beijing, China, 100084.
3
Assist. Prof., Dept. of Civil and Environmental Engrg., Louisiana State Univ., Baton Rouge, LA 70803, USA
2
deflection calculation of cantilever beams.
Introduction
Continuous composite beams are widely used in multi-story buildings and bridges for the advantages
of higher span/depth ratio, less deflection, and higher fundamental frequency of vibration, etc., compared
with simply supported beams. The reduction in beam height and overall dead load results in a decrease in
story height and foundation loads of a structure, which results in a more economical construction. With
the increasing use of high-strength materials of concrete and steel, which results in shallow sections,
deflection may become a critical factor that controls design. For continuous composite beams, the
negative bending moment near interior support regions will generate tension in concrete slab and
compression in steel, which is unfavorable in design.
Even though a beam is designed as a full composite section, a perfect composite action without any slip
cannot be obtained due to the deformation of shear studs, which are often treated as ductile connectors
(Mottram and Johnson 1990). In case of composite beams with profiled sheeting construction, there may
be no enough space in the troughs to provide enough stud connectors in the negative bending regions. A
partial composite design is then required in this condition, and the influence of slip on the beam deflection
may be more pronounced. It is desirable to establish a model for composite beams under negative bending
accounting for slips at the steel beam-concrete slab interface with full or partial shear connection
(Fabbrocino et al. 2000).
A large number of researches have been devoted to develop models for analyzing the behavior of
composite beams, most of which focused on beams under positive bending moments. Previous studies
and tests have shown that the influence of slips between concrete slab and steel beams on the deflection
behavior should be considered in design (Salari et al. 1998, Nie and Cai 2003). Some differential
3
equations were derived with these models (Newmark et al. 1951, Jasim 1999). Alternatively, some more
general and powerful numerical procedures were also used (Wright 1990, Wang 1998). The stiffness of
shear connectors used in these studies was mostly derived from push-out tests, and some of the
conclusions of slip effects on beam deflection have been adopted in design codes.
As for composite section under negative bending, the three main components are reinforcement, steel
beam and shear connectors. The concrete slab that cracks at low tension stress was ignored in most
analyses. Manfredi et al. (1999) proposed a mono-dimensional model to analyze the structural behavior
of steel-concrete composite beams subjected to negative bending. The model takes into account two types
of slips: the slip between steel beam and concrete slab, and the slip between concrete and reinforcement
bars. Although this numerical approach is a very powerful research tool for analyzing complex structures,
it perhaps is too complicated and time consuming for routine design practice. On the contrary, a pure
empirical method may be too approximate and unreliable.
In this paper, differential equations were derived from a model of composite beams under negative
bending, and from which solutions for some loading cases and boundary conditions were obtained. Since
the materials are generally linear elastic under service load conditions, superposition of these loading
cases to predict the deformation of a more complicated case is a valid approach. As cantilevers are often
used in high-rise building to decrease the positive moment in the adjacent span of continuous beams, a
relative simple method for calculating the reduced stiffness of the cantilever composite beams was
developed.
In order to validate the model and the solution of the present study, a test of three (3) composite beams
was completed in laboratory. In addition, since the developed procedure is elastic analysis, a
three-dimensional nonlinear finite element analysis was conducted to study the general behavior of the
4
beams up to failure. The results from the FE analysis were compared with that from experimental and
analytical studies.
Theoretical Analysis
Theoretical Model
In most cases, steel beams and reinforcement bars of composite section will not yield under service
load conditions. To simplify the problem, the material properties of composite beams can thus be
considered as linear elastic, and the following assumptions are made:
1. Shear at the interface of different components of beams is proportional to slips.
2. Shear connection (interface) stiffness is uniform and continuous along the length of composite
beams.
3. Steel beam and concrete slab at the same section have the same curvature and rotation.
4. Reinforcement bars dont provide vertical shear resistance, and concrete tension capacity is
omitted.
5. Material behavior of steel and reinforcement is linear elastic.
The shear slip behavior between the steel and concrete is complicated. It depends on many factors such
as the stiffness of studs, chemical bond at the interface, and cracking of concrete. However, previous
studies (Wang 1998, Nie and Cai 2003) indicated that shear-slip relation can be simplified as linear elastic
under service load.
A segment model under negative bending without distributed load is shown in Fig.1, in which the
concrete interface element is marked with dashed lines. The concrete interface element is a component
that transfers the shear force between reinforcement and steel beam, but will not contribute to the moment
5
resistance. For a case with a distributed load, pressure q will be added on the top surface of the concrete
slab. In this figure, s
1
= slip at the steel beam-concrete slab interface; s
2
= slip at the
concrete-reinforcement interface; y
s
= distance of top steel fiber to the neutral axis of steel section; and y
r

= distance of reinforcement to the top steel fiber.
Shear-slip stiffness of stud
Based on assumptions 1 and 2, the following equation is obtained

1 1
pv K s = (1)
where p = distance (or pitch) between the studs; v = horizontal shear along the steel beam and concrete
slab interface per unit length; and K
1
= shear-slip stiffness of stud connectors.
Determination of the shear connector stiffness is essential for using Eq. (1). According to previous tests
and studies (Johnson et al.1969, Ollgaard et al. 1971, Wang 1998), the relationship between the load of
shear connector and the slip is generally nonlinear with a typical load-slip curve shown in Fig.2. It seems
difficult to find a simple but reliable formula for the shear connector stiffness because of too many
parameters and the scattering test results; the stiffness of studs varies along the beam in terms of the
loading cases and boundary conditions. As a rather simple suggestion (Wang 1998), the shear connector
stiffness may be estimated as the secant stiffness at the shear connector design strength with an equivalent
slip of 0.8mm. Nie and Cai (2003) had proposed a similar formula on the shear connector stiffness as

1 max
0.66
s
K N P = (2)
where N
s
= number of shear studs per row; and P
max
= design shear resistance of stud.
In this paper, the shear resistance of studs given by Eurocode 4 (EC4) is used:

2 2
max
0.8 ( / 4) 0.29
t u s t s ck cm
P k f d k d f E = (3)
where f
u
= ultimate tensile strength of stud; d
s
= diameter of stud; f
ck
= characteristic cylinder strength of
6
concrete; E
cm
= mean secant modulus of concrete; and k
t
= stud reduction factor affected by profiled
sheeting which is given as

0
0.7
1 1.0
d
t
p p s
b h
k
h h N
| |
= |
|
\ .
(4)
where b
0
= width of deck rib; h
p
= height of deck rib; and h
d
= shear stud height after welding.
The stud stiffness discussed above is based on the tests of concrete in compressive (corresponding to
studs in positive moment regions). Tests on continuous composite beams (Daniels and Fisher 1967) and
composite beams in tension (Clinton and Samuel 2000) indicated that the concrete in tension has little
effect on the shear stud strength. Therefore, the load-slip relationship of studs obtained by tests on
positive moment is still applicable in this study for negative moment regions.
Bond-slip stiffness of reinforcement
When concrete stress is low, the longitudinal reinforcement has the same strain as the concrete
surrounding it. As stress increases, slip occurs at the interface of reinforcement and concrete. Similar to
concrete slab and steel beam interface with studs, the following relationship is assumed:

2 2
pv K s = (5)
where K
2
= bond-slip stiffness between reinforcement and concrete.
CEB-FIP (1993) has proposed an idealized slip-bond stress relationship for rebars in concrete. The
curve is composed of 4 segments, and the first part of the curve is

0.4
2
( / )
r u u
s s = (6)
where
u
= the maximum bond stress, which equals
c
f for deformed reinforcement bar in ordinary bond
conditions with no confinement; and s
u
= the maximum slip at first portion of the curve.
The bond-slip relationship between reinforcement and concrete at service condition corresponds to the
7
pre-peak portion of the curve. A linear relationship, using the secant at s
u
as the slope, can be used to model this
bond-slip behavior. Based on test results, s
u
= 0.6mm, thus Eq. (6) is simplified and replaced with

2
0.6
c
r
f
s = (7)
where f
c
= compression strength of concrete (N/mm
2
).
The interface bond stiffness for deformed reinforcement in the slab is derived from Eqs. (5) and (7)
as (by transferring stress into per unit length force)

2
5.2
r r c
K pd N f = (8)
where d
r
= diameter of reinforcement bar (mm); and N
r
= number of reinforcement.
Similarly, in case of plain bars, s
u
= 0.1mm,
u
= 0.15
c
f and the interface bond stiffness is derived as

2
4.7
r r c
K pd N f = (9)
Calculation of deflection
Assume that the relative slip between reinforcement and steel beam is composed of two parts as
s = s
1
+s
2
(10)
then the horizontal shear-slip relationship between steel beam, concrete interface element and
reinforcement bar can be derived from Eqs. (1), (5), and (10) as
pv Ks = (11)
where
1 2
1 2
K K
K
K K
=
+

The following procedure is similar to that of Nie and Cai (2003) where positive bending was
considered, instead of negative bending. Considering the longitudinal equilibrium of the reinforcement
segment gives

dT
v
dx
= (12)
8
Moment equilibrium of the concrete and steel components gives

2
1
2
c r
V dx rdx vy dx + = (13)

2
1
2
s s s
dM V dx vy dx rdx + = + (14)
where V
c
= vertical shear in concrete interface element; dM
s
= moment increment of steel section; V
s
=
vertical shear of steel component; and r = vertical force per unit length at interface. In case of a simply
supported beam with a concentrate load P at mid-span, V
c
+ V
s
= P/2.
Assumption (3) leads to

s
s
M
EI
= (15)
where = curvature of composite section; E = elasticity modulus of steel or reinforcement; M
s
= moment
increment of steel section; and I
s
= moment of inertia of steel section. Eliminating M
s
and v using Eqs. (11)
to (15) and ignoring the small terms, the curvature is derived as

2 ( ) /
2
s r
s
Ks y y p P d
dx EI
+
= (16)
Considering the equilibrium condition T = C, the longitudinal strain at the bottom fiber of concrete
interface element and at the top fiber of steel beam are calculated as

2 cb r s
r
T
y
EA
= + (17)

st s
s
T
y
EA
= (18)
Then the relative slip strain between reinforcement and top fiber of steel beam is derived as

2 s st cb s s r
s r
T T
y y
EA EA
= + = + (19)
From Eqs. (11) (12), (16) and (19), we have

2
2
2
d s
s P
dx
= (20)
where
2
2 0
0 s
1 y K
pE A I

| |
= +
|
\ .
,
0
s
2
y
EI
= ,
0 r s
y y y = + ,
0 r s
1 1 1
A A A
= + .
To be symmetric, the slip at mid-span must be zero. Meanwhile, since the moment at the supports is
9
zero, there is no difference between the longitudinal strains at the interface, meaning that the slip strain at
the beam-ends is zero. Using the boundary conditions ( 0) 0 s x = = ( / 2) 0 s x L = = , the slips can then be
solved as

2
( 1)( )
( 1)
x x L x
L
e e e e P
s
e


=
+
(21)
Differentiating Eq. (21) gives the slip strain as

2
s
( )
(1 )
x L x
L
e e e P
e

=
+
(22)
The slip and slip strain obtained by Eqs. (21) and (22) are plotted in Fig.3 and Fig.4 for the given
information in the figures, and each curve corresponds to different degree of shear connection n/n
f,
, where
n = total studs; and n
f
= studs required for full composite action according to design codes. The figures
indicate that the maximum slip occurs at the beam-ends and a concentration of slip strain occurs near
loading point. As the number of studs (n/n
f
) increases, both the slip and slip strains decrease. The slip
effect generally cannot be neglected even with a full composite design where the n/n
f
is theoretical 1. The
analysis also showed that the slip between reinforcement and concrete is small compared with the total
slip and can thus be neglected, which implies that the slip stiffness factor K depends mainly on the shear
connector stiffness K
1
.
Additional deflection
The longitudinal strains of composite sections under negative bending are shown in Fig.5. Assuming
that the curvature of composite section with full interaction is
1
, and the curvature of composite section
with partial interaction is
2
(it is noted that in the previous equations corresponds to
2
). These two
strain/stress distributions generate the same moment on the section; the equilibrium equation leads to

s r s
( ) M M T y y = + + (23)
The slip strain is derived from Eqs. (15), (19) and (23) as
10

1 2
2 0
0 0
s
s
I I
y
A y



= (24)
where I = moment of inertia for the composite section.
The additional curvature due to the slip strain is then derived from Eq. (24) as

2
0 0
2 1 1 2
0 0
0
0 0
s s
s
s
I I A y
I
A y I
y
A y



= = +
+
+
(25)
Eq. (25) indicates that the additional curvature is determined by the slip strain
s
and curvature
1
.
However, calculation shows that the second term of Eq. (25) is less than 5% of the total additional
curvature for beams in practical range, and its contribution decreases as the slip strain increases. Thus, the
additional curvature can be simplified with sufficient accuracy as

s
a
h

= (26)
where h
a
= y
0
+I
s
/(A
0
y
0
).
Considering the boundary conditions of
1
( / 2) 0 L = and
1
(0) 0 = , the additional deflection of
composite beams with upward concentrate load at mid-span can be calculated by integrating the
additional curvature along the beam as

| |
1 3
2tanh( / 2)
2
a
P L L
h

= (27)
where the minus sign here means the deflection is upwards.
Other loading and boundary conditions
The expression for slip and slip strain along the beam depends on the loading cases, boundary
conditions and arrangement of shear connectors. In the case of continuous composite beams (near the
interior support regions of a continuous composite beam), the loading arrangement and reaction of
negative bending regions near the interior support are similar to that of a point load acting at the bottom of
a simply supported beam. Since the stiffness varies along the beam in serviceability conditions due to the
11
cracking of concrete, moment redistribution should be considered in determining the force and length of
negative bending regions. Therefore, using the reduced stiffness caused by slips in negative bending
region to calculate the deflection of continuous composite beams may not be convenient for hand
calculation in practice. In the case of cantilevers, since the moment distribution is only determined by
equilibrium, calculation of the deflection by considering the slip effect is applicable and reasonable. The
deflection of composite beams under negative bending with different loading cases and boundary
conditions, shown in Fig.6, is discussed below.
The derivation presented earlier corresponds to Case (1). Case (2) is a simply supported composite
beam subjected to two concentrated loads symmetric to the mid-span. The additional deflection is derived
as

2 2 3
2 sinh( ) cosh( ) tanh( / 2)
2
a
P L b b b L
h


(
= +
(

(28)
Case (3) is a cantilever with uniformly distributed load along the full length. The additional deflection
is

2 2
3 4
2 2sech( ) 2 tanh( )
a
q L L L L
h

( +

= (29)
Case (4) is a cantilever with a concentrated load at a distance of a from the fixed end. The additional
deflection is

| |
4 3
2 sech( ) ( cosh( )) sinh( )
a
P a a L L a a
h

+
= (30)
Because the analysis is of linear-elastic, superposition can be used in more general loading conditions.
Simplification of Solutions
The practical value of parameter L varies between 4 and 10 (Nie and Cai 2003), which leads to the
12
assumption that tanh(L/2) = 1 and sech(L) = 0. While simplification of Case (4) depends on the a value,
the formula to calculate the additional deflection due to slip effects of Cases (1) to (3) can be simplified as

1 2 3
1
2
a
P L
h


| |
=
|
\ .
(31)

2 2 3
2
2
b
a
P L b e
h

| |
=
|
\ .
(32)

( )
2 2
3 4
2 2
a
q L L
h

+
= (33)
Thus, the total deflection of composite beam under negative bending by considering the slip effects is

f
= + (34)
where
f
= deflection calculated with the assumption of full interaction between each component of the
composite beam.
Reduced stiffness for cantilevers
The total deflection of Case (3) at the free end is obtained by the classical beam theory and the
additional deflection as

4
3 3
8
ql
EI
= + (35)
where EI = elastic flexural stiffness of the composite section with full interaction under negative moment.
The total deflection of Case (3) can be rewritten in the following simplified form as

4
3
8
ql
B
= (36)
where

1
EI
B

=
+
(37)
and is the stiffness reduction factor defined as

( )
2 2
3 4 4
2
2 2 L L
L

= + (38)
13
where
4
a
EI
h

= .
For Case (4), the total deflection at the free end of cantilever can be calculated as

( ) ( )
2 3 2 3
4 4
2 2
6 6
P a L a P a L a
EI B

+ +
= + = (39)
Correspondingly, the stiffness reduction factor of this case is derived as:

| |
4 2 3
3sech( ) sinh( ) cosh( )
( 3 )
a L a a L a
a a L

+
=

(40)
Eq. (40) can be used to develop a design aid based on the dimensionless factor L and a/L. The
parameter L depends on the properties of composite section and the cantilever length. The computed
factors / are plotted against L in Fig.7 for L from 4 to 10, where different curves are used to
represent the results corresponding to different location of loading. This figure reveals that as the
parameter L increases due to the increase of shear connectors and/or the beam length, the stiffness
reduction factor for load Case (4) decreases. For a given L, the stiffness reduction factor decreases as
the value a/L increases.
A design table is developed for Case (4) based on Eq. (40). The values of which are calculated for the
various values of the factors L and a/L are presented in Table 1. Having calculated the factors L and a/L
for a given cantilever beam, the factor / may be read from the table by interpolation. After the
calculation of that depends on the property of composite section, the maximum deflection of cantilever
beam can be conveniently determined by Eq. (39).
Experimental Study
Description of tests
Three simply supported composite beams, designated as SB6, SB7 and SB8 with profiled sheeting
14
were tested. Since the flexural response is the major consideration in this study, the specimens were
designed to fail in flexure rather than in shear or other types of failures. The tested beams are 4m long
each, and the cross-section (Fig. 8) consists of an I20a steel beam and a slab of 800mm 110mm with
profiled sheeting placed transversely. As shown in Table 2, the details of the three specimens were similar
except for the different longitudinal reinforcement. The force ratio R that is 0.227, 0.362 and 0.498 for the
three specimens, respectively, is defined as

r ry
s sy
A f
R
A f
= (41)
where A
r
= area of longitudinal reinforcement; A
s
= area of steel beam; f
ry
= yield strength of
reinforcement; and f
sy
= yield strength of steel beam.
According to EC4, the compression element of I20a beam is classified as class I, and the moment
capacity is thus not affected by local buckling. The profiled sheeting is of model YXB60-200-600, with
the rib of 60mm in height and 200mm in rib spacing. One shear stud of 16mm in diameter and 90mm in
height was placed on every rib to provide composite actions between the steel beam and the concrete slab.
All studs were welded to the top of steel flange. The degree of shear connection (n/n
f
) for the three tested
beams is 1.85, 1.16, and 0.84 respectively. In order to prevent secondary failure such as longitudinal
splitting of the slab, transverse reinforcement of 6.5mm in diameter was provided at both the top and
bottom of the slab.
The test setup included a load cell over the hydraulic jack and deflection measurements at mid-span.
The deflections were corrected for the small movement of supports to monitor the load-deflection
characteristics. Strain measurements were made at the weak and strong sections respectively in the
constant moment region by strain gages glued on the longitudinal reinforcement, steel beam and concrete
slab. For a composite beam using profiled sheeting, the strong section is located at the rib, and the weak
15
one is just in-between of two ribs. The concrete strength was measured with six 100 mm cubes per beam
the same day as for the experiment. The steel and reinforcement specimens were both tested in tension to
determine the yield strength and ultimate strength.
The beam specimens were each tested to failure in flexure using two symmetric point loads that
produces a constant moment region as shown in Fig.9. In order to facilitate the observation of cracking on
the top surface of the slab, the loading is applied upward. The tests were performed in two stages. In the
first stage, the load was increased at an increment of 5kN until it reached about 80% of the expected
ultimate load P
u
. Load was then kept constant at each step to measure the crack width on top of the slab. In
the second stage, the loading was controlled by displacement increments until a recognizable drop in load
occurred. In the whole procedure, strains and deflections were measured automatically by a data
acquisition system controlled by a computer.
Results of tests
The load-deflection response of the tested composite beams at mid-span is shown in Fig.10. At the
initial stage of loading (i.e., within 15% of P
u
) the response is linear and no cracks is visible on the slab.
As the load increased, the first crack initiated on the top surface of the slab in mid-span and the stiffness
slightly decreased from the first stage. In this stage, the cracks extended slowly with the reinforcement in
elasticity. The cracks were approximately straight and normal to the beam axis with the maximum crack
width no more than 0.3mm. The load-deflection curve kept linear, which confirmed the linear elastic
assumption of the analytical model. When the load exceeded 85% of P
u
, steel and longitudinal
reinforcement became yielding. The cracks expanded rapidly with the maximum crack width exceeding
0.3mm, and then the beam soon reached its ultimate load.
These curves also show a large plateau after the yielding of steel and reinforcement, which means a
16
good ductile failure with large deformation in the inelastic region. The failure loads for each beam are
169.4kN, 189.6kN, and 208.1kN, corresponding to bending moments of 118.6kNm, 132.7kNm, and
145.7kNm in the constant moment region, respectively. The results indicate that the amount of
reinforcement has significant effect on the moment strength of composite beams under negative bending.
Moreover, the three curves in Fig.10 show that, as the amount of reinforcement increases, the ductility of
beams decreases. The reason is that when the compressive height of steel web increases due to the
increased reinforcement, it tends to buckle earlier. For beam SB6 and SB7 with low force ratio R, no
buckling in the steel web was observed. For beam SB8, the failure is due to the web buckling and the
breaking of shear studs.
The load-strain characteristics were measured for each beam at strong section (the section crossing the
rib of the profiled sheet) and weak section (the section between two ribs) respectively. Fig.11 shows the
load-strain characteristics of the bottom fiber of the steel beam. The bottom fiber strain at both strong and
weak sections didnt exceed the yield strain before the load reached 70%P
u
. When the load approached P
u
,
the strains increased rapidly over 0.002, showing that a plastic hinge formed and a full ductile failure
developed. The strains at strong and weak sections did not show noticeable variance during the whole
loading process.
Fig.12 shows the load-strain characteristics for the longitudinal reinforcement. The strains in the
reinforcement are negligible before the cracking of concrete. After cracking, the strains increased faster
and the strains at weak section where most cracks initiated were greater than that at strong sections. For
beam SB8, some studs on one side of the specimen were broken at the maximum load. It led to far less
reinforcement strains than the other two specimens at the failure load. Results from the strains also
confirmed the conclusion that high force ratio results in a low ductility in failure.
17
Discussion of test results
Deflection of beam is generally concerned under working/service load P
s
that is taken as 0.5P
u
in this
study. The comparisons of tests and theoretical results under working load P
s
are given in Table 3. In the
table,
f
is the deflection calculated with assumption of full composite interaction;
a
is the deflection
calculated by the analytical method formulated in this paper, and
FE
is the deflection obtained by FE
analysis that will be discussed later.
Without considering the effect of slip, the method based on the theory of full shear interaction
underestimates the beam deflection by about 10%~20% compared with the experimental results. The
discrepancy increases with the increasing of force ratio R. The reason is that as the number of longitudinal
reinforcement increases with the same number and strength of studs in the tested beam, the shear
connectors become relatively weaker. As a result, more slips will occur as the beam reaches the working
load, which results in more additional deflection.
Numerical Study
In order to study more thoroughly the performance of composite beams under negative bending, a
commercial finite element package ANSYS was used. In the numerical analysis, the brittle property of
concrete is simulated with a solid element that can change its stiffness with the development of cracking
and crushing of concrete. The two types of slips between different materials were modeled with spring
elements. Comparisons of the load-deflection curves between analytical study, FE analysis and laboratory
tests were made. The methodology developed here can also be applied for other analysis on composite
beams.
18
Material properties
Concrete strength is high in compressive and low in tension. When the composite beam is subjected to
negative bending, tension in the slab will result in the formation of cracks perpendicular to the principal
stress direction. The SOLID65 element offered by ANSYS can simulate this nonlinear property of
concrete. The element behaves as a linear elastic material until the stress reaches the tension or
compression strength. Once the principal stress exceeds the strength at an integration point, the
stress-strain relation of the element will be modified by introducing a plane of weakness in the direction
normal to the stress to represent the cracking. The failure surface is defined by a total of five strength
parameters, but it can also be specified by a minimum of two constants f
t
and f
c
with the other three
defaulting to Willam and Warnke criteria (ANSYS 1998). After cracking, the tension stress of the
concrete element is set to zero in the direction normal to the crack plane. The shear transfer coefficient
t

for open cracks and
c
for closed cracks determines the amount of shear transferred across the cracks. The
value of the shear transfer coefficient ranges from 0.0 to 1.0, with 0.0 representing no shear transfer at a
crack section and 1.0 representing full shear transfer. In this study,
t
is assumed to be 0.3 and
c
is
assumed to be 0.5(Kachlakev 2001). The higher values of the shear transfer coefficient were used to avoid
convergence problems during iteration.
The material properties of steel and reinforcement are specified with the typical bilinear idealization in
both tension and compression (Fig.13). The strain hardening modulus E is assumed to be 0.005E. The
deformed reinforcement was simulated using LINK8 truss element, and the steel was simulated using
SHELL43 element.
Bond-slip relationship
In the FE analysis the bond-slip relationship between concrete and reinforcement was simulated using
19
COMBIN14 linear spring element. Every node between two LINK8 elements is jointed with a spring
element parallel to the reinforcement element. Thus, the bond stresses only depend on the longitudinal
slip. The stiffness of the spring was obtained from the secant of bond stress versus slip curve according to
CEB-FIP, and it is given by

E1 E2
r 2
2
l l
k K
+
= (42)
where l
E1
, l
E2
= lengths of two adjacent LINK8 elements.
The shear studs were modeled by nonlinear spring element COMBIN39. Typically, the actual load-slip
curve of the stud connectors is obtained by push-out test. Previous study (Wang 1998) has shown that the
curve is generally nonlinear and it is reasonable to use a nonlinear spring in modeling the mechanic
behavior of the connectors. The constitutive relationship of the spring is given by (Ollgaard 1971)

1
max
(1 )
ns m
F P e

= (43)
where F = load on a shear stud. The parameters m and n define the shape of the curve, and the typical
values used in this study are m = 0.558, n = 1mm
-1
.
Three-dimensional modeling
Three FE models subjected to negative bending were developed to represent the tested specimens. Due
to the symmetric geometry and loading in both transverse and longitudinal directions, only one quarter of
the beam was modeled as shown in Fig.14. The typical length of the element is 50 mm to 65 mm. A
simply supported boundary condition is assumed on the top surface of concrete slab and the gravity of the
materials was included in the element. To consider the effect of slips, two kinds of models with different
level of details have been developed. One model with spring elements was used to simulate the tested
specimens including the slip effect (denoted as FE). Another relative coarse model with full interaction
(without slip) between each component of the beam was used for comparison (denoted as FE-f). The
20
primary parameter varied in the study was the amount of longitudinal reinforcement in the slab. No
buckling in the steel web was considered in this numerical study.
Results and discussions
Fig.15 compares the load deflection response at mid-span between FE analysis with or without slip
effect and experimental measurements. It shows that the results from the FE analysis agree well with that
from experiments in terms of both stiffness and maximum load capacity. Increase of reinforcement,
meaning a lower degree of shear connection and more slip, increases the ultimate strength, and at the
same time reduces more stiffness at working load range compared to beams with full shear interaction. It
is found that the higher the stiffness of the stud connectors and bond-slip stiffness at
reinforcement-concrete interface, the higher flexibility stiffness of the beam in the initial stage of loading.
When the steel beam and the reinforcement begin to yield, the slip has little influence on the load capacity
of the beam. Since the present study focused on the service load behavior, no defections and failure
criteria of steel and reinforcement are considered in the FE model. Therefore, no drop of load was
predicted and significant difference is shown for the descending part of the curves. Further study
including the effects of buckling of the steel and rupture of the reinforcement may refine this FE model
and predict more accurately the descending part of the curves.
Fig. 15 (d) shows the zoomed view of the load deflection curves in the service load range. The load of
first cracking for the three beams is almost the same, which is about 24kNm. After the first cracking, the
curves remain linear and the beam with more reinforcement is stiffer than the beams with less
reinforcement. At the working load of 61.8kNm, 69.2kNm, and 76.0kNm for SB6, SB7 and SB8,
respectively, the deflection of the model considering slip effect is 4.4%, 10.8%, and 16.0% higher than
that without slip, respectively. These results correlate well with those of analytical method as shown in
21
Table 3.
Conclusions
The behaviors of steel-concrete composite beams under negative bending with different degree of
shear interaction have been investigated in this paper. A model for predicting the stiffness behavior of
composite beam under negative bending at serviceability limit state was developed. The model considers
the slips at the steel beam-concrete slab interface and concrete-reinforcement interface, which explains
the decrease of structural stiffness compared with that based on the theory of full shear interaction. A
series of analytical equations based on this model were derived to calculate the maximum deflection of a
composite beam at working load, and different types of loading cases and boundary conditions were
considered. Also, a simplified method for calculating the deflection of cantilever composite beams was
established.
To verify the reliability of the analytical model and the equations, three (3) beams were tested.
Meanwhile, the general purpose FE software ANSYS was used to investigate the behavior of the
composite beam. The nonlinear property of the material, crushing of the concrete, and slips at the
steel-slab and concrete-reinforcement interface were considered. The results from testing and FE analysis
show that the prediction of the deflection by the proposed analytical method was sufficiently accurate in
designing a cantilever, and the procedure can be used in further study on continuous composite beams.
ACKNOWLEDGMENTS
This research is a part of the research program on the continuous steel-concrete composite beams with
profiled sheeting at Tsinghua University, China. The writers gratefully acknowledge the financial
support provided by NSFC program for Outstanding Youth (# 50025822). Assistance for experimental
Comment: This part is changed after
submitting the final.
Deleted: ) and Beijing Natural Science
Fund (# 8992013)
22
studies from the staff at Tsinghua University is also appreciated. The third author appreciates the support
from Louisiana State University.
APPENDIX I. REFERENCES
ANSYS 5.5. Theory reference manual, 8th ed. Editor: Kohnke Peter. 1998
Clinton, O. R., Samuel, E. P. E. (2000). Behavior and modeling of reinforced composite slab in tension.
J. Struct. Engrg., ASCE, 126(7), 764-771.
CEB-FIP Model Code 1990 (1993) Comite Euro-International du Beton. Bulletin Dinformation No.
213/214 (Concrete Structures). Lausanne.
Daniels, J. H., and Fisher, J. W. (1967). Static behaviour of continuous composite beams. Fritz Eng.
Lab. Rep. 324.2, Lehigh University.
European Committee for Standardization (ECS). (1994). Eurocode 4: Design of composite steel and
concrete structures. Brussels.
Fabbrocino G., Manfredi G., Cosenza E. (2000) Analysis of continuous composite beams including
partial interaction and bond. J. Struct. Enrg., ASCE, 126(12): 1288-1294
Jasim, N. A. (1999). Deflections of partially composite beams with linear connector density. J. Constr.
Steel Res., 49(3), 241-254.
Johnson, R. P., Greenwood, R. D., and Van Dalen, K. (1969). Stud shear-connectors in hogging moment
regions of composite beams. The Struct. Engr., London, 47(9), 345-350.
Johnson, R. P., and Hope-Gill, M. C. (1976). Applicability of simple plastic theory to continuous
composite beams. Proc., Instn of Civ. Engrs., 61(3), 127-143.
Johnson, R. P. (1994). Composite structures of steel and concrete., Vol. 1, 2nd Ed., Blackwell Scientific
Publications, Ltd., Oxford, U.K.
Kachlakev, D. I., Miller, T. H., Yim, S., etc. (2001). Finite element modeling of reinforced concrete
structures strengthened with FRP laminates., Final report, FHWA-OR-RD-01-XX, United States
Department of Transportation, Federal Highway Administration and Oregon DOT, Salem, Oregon.
Manfredi, G., Fabbrocino, G., and Cosenza, E. (1999). Modeling of steel-concrete composite beams
23
under negative bending. J. Engrg. Mech. Div., ASCE, 125(6), 654-662.
Mottram, J. T., and Johnson, R. P. (1990). Push tests on studs welded through profiled steel sheeting.
The Struct. Engr., London, 68(5), 187-193.
Newmark, N. M., Siess, C. P., and Viest I. M. (1951). Tests and analysis of composite beams with
incomplete interaction. Proc. Soc. for Experimental Stress Analysis, 9(1), 75-92.
Nie, J. G., and Cai, C. S. (2003). Steel-concrete composite beams considering shear slip effects. J.
Struct. Engrg., ASCE, 129(4), 495-506.
Ollgaard, J.G., Slutter, R. G., and Fisher, J. W. (1971). Shear strength of stud connectors in lightweight
and normal weight concrete. AISC Engrg. J., 8, 55-64.
Salari, M. R., Spacone, E., Shing, P. B., and Frangopol, D. M. (1998). Nonlinear analysis of composite
beams with deformable shear connectors. J. Struct. Engrg., ASCE, 124(10), 1148-1158.
Wang Y. C. (1998). Deflection of steel-concrete composite beams with partial shear interaction. J.
Struct. Engrg., ASCE, 124(10), 1159-1165.
Wright, H. D. (1990). The deformation of composite beams with discrete flexible connection. J. Constr.
Steel Res., 15(1-2), 49-64.
APPENDIX II. NOTATION
The following symbols are used in this paper:
A
r
= aera of reinforcement;
A
s
= aera of steel beam;
b
0
= width of deck rib;
B = reduced stiffness;
C = compression force applied on steel;
d
s
= diameter of stud;
d
r
= diameter of reinforcement bar;
24
E = elasticity modulus of steel or reinforcement;
E
cm
= mean secant modulus of concrete;
f
c
= cylinder compression strength of concrete;
f
ck
= characteristic cylinder strength of concrete;
f
cu
= compressive strength of cubic concrete block;
f
ru
= ultimate strength of reinforcement;
f
ry
= yield strength of reinforcement;
f
su
= ultimate strength of steel;
f
sy
= yield strength of steel;
f
t
= tension strength of concrete;
F = load on shear stud;
h
d
= shear stud height after welding;
h
p
= height of deck rib;
I = moment inertia of composite section;
I
s
= moment inertia of steel section;
k
t
= stud reduction factor;
K = shear-slip stiffness;
K
1
= shear-slip stiffness of stud connectors;
K
2
= bond-slip stiffness between reinforcement and concrete;
l
E1
, l
E2
= lengths of conjoint reinforcement elements;
L = span length of beam;
M = applied moment;
25
M
s
= moment on steel section;
N
r
= number of reinforcement;
N
s
= number of shear studs per row;
p = distance between studs or pitch;
P = applied load;
P
max
= design shear resistance of stud;
P
s
= working (service) load;
P
u
= ultimate load;
q = distributed load;
r = vertical force per unit length at interface;
R = parameter of force ratio;
s = slip;
s
1
= slip at steel-concrete interface;
s
2
= slip at concrete-reinforcement interface;
T = tension force applied on reinforcement;
v = horizontal shear per length unit;
V
c
= vertical shear in concrete;
V
s
= vertical shear in steel;
y
r
= distance of reinforcement to top fiber of steel beam;
y
s
= distance of top steel fiber to neutral axis of steel section;
, , = parameters determined by beam property;

c
= shear transfer coefficient for closed cracks;
26

t
= shear transfer coefficient for open cracks;

a
= deflection calculated by analytical method;

f
= deflection calculated with assumption of full composite interaction;

FE
= deflection obtained with FE analysis;
= additional deflection;
= additional curvature;

s
= slip strain;
= curvature;

1
= curvature of full interaction;

2
= curvature of partial interaction;

u
= maximum bond stress at concrete-reinforcement interface; and
= stiffness reduction factor.
27
List of tables
Table 1. Practical values for /
Table 2. Description of tested beams
Table 3. Comparison of experimental and theoretical results
List of figures
Fig. 1 Segment element model of composite beam
Fig.2 Determination of shear connector stiffness
Fig.3 Slip distribution along beam
Fig.4 Slip strain distribution along beam
Fig.5 Longitudinal strains of composite section across section height
Fig.6 Loading cases and boundary conditions of beam
Fig.7 Stiffness reduction factor for cantilevers
Fig. 8 Beam specimen cross section and slab side view (unit = mm)
Fig. 9 Configuration of beam tests
Fig.10 Load-deflection relationship for tested beams at mid-span
Fig.11 Bottom-fiber steel strain of specimens
Fig.12 Reinforcement strain of beam specimens
Fig. 13 Assumed stress-strain relationship for steel and reinforcement
Fig.14 Finite element model
Fig.15 Moment-deflection response of beam specimens
28



Table 1. Practical values for /
L
a/L
4 5 6 7 8 9 10
0.2 0.395 0.353 0.312 0.274 0.240 0.211 0.185
0.3 0.312 0.257 0.212 0.174 0.145 0.121 0.103
0.4 0.242 0.187 0.146 0.116 0.093 0.076 0.063
0.5 0.188 0.138 0.105 0.081 0.064 0.052 0.043
0.6 0.148 0.106 0.078 0.060 0.047 0.038 0.031
0.7 0.119 0.083 0.061 0.046 0.036 0.029 0.024
0.8 0.098 0.068 0.049 0.037 0.029 0.023 0.019
0.9 0.082 0.056 0.041 0.031 0.024 0.019 0.016
1.0 0.070 0.048 0.035 0.026 0.021 0.016 0.014

29






Table 2. Description of tested beams
Specimen f
cu
(MPa) f
c
(MPa) f
sy
(MPa) f
su
(Mpa) f
ry
(MPa) f
ru
(MPa)
A
s

(mm
2
)
A
r

(mm
2
)
R
SB6 47.0 36.0 290.6 417.6 369.2 560.4 3555 565 0.227
SB7 44.6 34.1 290.6 417.6 369.2 560.4 3555 905 0.362
SB8 48.6 37.1 290.6 417.6 369.2 560.4 3555 1244 0.498
Table 2. Description of tested beams (The previous one is wrong based on Jiangsheng)
Specimen f
cu
(MPa) f
c
(MPa) f
sy
(MPa) f
su
(Mpa) f
ry
(MPa) f
ru
(MPa)
A
s

(mm
2
)
A
r

(mm
2
)
R
SB6 49.7 36.0 290.6 417.6 413.4 607.9 3555 565 0.227
SB7 47.1 34.1 290.6 417.6 413.4 607.9 3555 905 0.362
SB8 51.3 37.1 290.6 417.6 413.4 607.9 3555 1244 0.498



30





Table 3. Comparison of experimental and theoretical results
Specimen P
s
(kN)

test
(mm)

f


(mm)
a
(mm)

FE
(mm)

f
/
test

a
/
test

FE
/
test
SB6 88.3 10.91 9.65 11.13 11.46 0.88 1.02 0.96
SB7 98.8 11.00 9.01 11.34 10.79 0.82 1.03 0.98
SB8 108.5 10.88 8.67 11.74 11.17 0.80 1.08 1.03
31








M
s
C
V
c
T
s
1
s
2
M
s
+dM
s
V
s
T+dT
V
c
C+dC
V
s
s
1
+ds
1 vdx
s
2
+ds
2
y
s
y
r
dx
rdx



Fig.1 Segment element model of composite beam
32




0.0
0.2
0.4
0.6
0.8
1.0
design shear connector resistance
equilvalent slip = 0.8 mm
working load
slip = 0.5mm
s
h
e
a
r

f
o
r
c
e
/
s
h
e
a
r

r
e
s
i
s
t
a
n
c
e
slip


Fig.2 Determination of shear connector stiffness
33
















Fig.3 Slip distribution along beam


-0.50 -0.25 0.00 0.25 0.50
0
5
10
15
I
s
=2.369 mm
7
A
s
=3555 mm
2
A
r
=904.8 mm
2
=1.90
-11
2.0
1.5
1.0
n/n
f
=0.5
s

/

P


(
1
0
-
3
m
m
/
k
N
)
x/L
34















Fig.4 Slip strain distribution along beam
-0.50 -0.25 0.00 0.25 0.50
0
5
10
15
20
2.0
1.5
1.0
n/n
f
=0.5

s

/

P


(
1
0
-
6
/
k
N
)
x/L
35








1

2

s1

s2
M
s
C
T y
r
y
s

(a) Full interaction (b) Partial interaction



Fig.5 Longitudinal strains of composite section across section height
36







L/2 L/2
P
L/2-b
P/2
2b
L
P
q
(1) (2)
(3) (4)
L
L/2-b
P/2
a


Fig.6 Loading cases and boundary conditions of beam
37







4 5 6 7 8 9 10
0.0
0.1
0.2
0.3
0.4
L
a/L = 0.2
0.4
0.6
0.8
1.0



Fig.7 Stiffness reduction factor for cantilevers
38






800
2
0
0
1
1
0
6@100
SB7 812
SB6 512
SB8 1112
I20a
2
5
100
6@65

110 90 110 90 110
6
0
5
0
130 70 130 70 130


Fig.8 Beam specimen cross section and slab side view (unit = mm)
39






Spreader beam
Hydraulic jack
Load cell
1.40m 0.80m 1.40m
Constant moment region
Test beam
40 roller
P/2
Testing frame
P/2



Fig.9 Configuration of beam tests
40






0 50 100 150 200
0
50
100
150
200
SB8
SB7
SB6
P

(
k
N
)
(mm)


Fig.10 Load-deflection relationship for tested beams at mid-span
41

















Fig.11 Bottom-fiber steel strain of specimens
0
50
100
150
200
0 10 20 30
SB8
SB7
{
{
SB6 {
weak
strong
weak
strong
weak
strong
(10
3
)
P

(
k
N
)
42







0
50
100
150
200
0 10 20 30
SB8
SB7
{
{
SB6 {
weak
strong
weak
strong
weak
strong
(10
3
)
P

(
k
N
)


Fig.12 Reinforcement strain of beam specimens
43





f
y
1
-
y
E
1
E'
-f
y

y



Fig.13 Assumed stress-strain relationship for steel and reinforcement

44






Y
X Z




Fig.14 Finite element model
45



0
50
100
150
200
0 50 100 150 200
Exp
FE
FE-f
(mm)
P

(
k
N
)
0
50
100
150
200
0 50 100 150 200
Exp
FE
FE-f
(mm)
P

(
k
N
)

(a) SB6 (b) SB7

(c) SB8 (d) Deflection in working load range

Fig.15 Moment-deflection response of beam specimens

0
50
100
150
200
0 50 100 150 200
Exp
FE
FE-f
(mm)
P

(
k
N
)
0
50
0
50
0
50
100
0 5 10 15
Exp
FE
FE-f
(mm)
P

(
k
N
)
SB8
SB7
SB6

Vous aimerez peut-être aussi