Beruflich Dokumente
Kultur Dokumente
Research Paper
Composite steel-concrete beams are common structural members used in floor systems and
are evaluated by well known procedures. Evaluation of composite member flexural strength
requires an expectation of strain transfer between the two materials, however, perfect transfer
does not occur, therefore, predictions of flexural strength may not be consistent with measured.
Composite beam test demonstrations conducted at the Pennsylvania State University on
identically constructed members have revealed that composite beams fail before reaching
predicted strength. Observations of failed composite beams include longitudinal cracking of the
concrete slab and interlayer slip between the concrete slab and the steel beam. This paper
presents laboratory test results and analyzes the ultimate flexural strength of the tested composite
beams using currently available methods and compares to observed behavior. Additionally,
simplified finite element models of the composite beam were developed with results also
compared to the data obtained from the tests. Based on the results of these analyses,
recommendations for design are made that allow accurate determination of composite beam
flexural strength.
1
Bridge Designer, HDR, Inc. 1016 W. 9th Avenue, Suite 110, King of Prussia, PA 19406-1331.
2
Professor of Civil Engineering, The Pennsylvania State University, Department of Civil and Environmental Engineering, 231J Sackett
Building, University Park, PA 16802.
90
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
91
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
analysis methods, and methods that use finite at the interface, no uplift, and linear strain
element analysis. distributions. Finite element models were
found to be very accurate, however, are most
TESTS often used for academic purposes and special
A limited number of composite beam cases. Leon and Viest (1997) conclude that
laboratory studies have been conducted to two concerns require development: (1)
confirm prediction models currently in use. develop a refined shear connection model;
Naithani and Gupta (1988) tested three and (2) simplify finite element models.
composite beams to compare measured to Sapountzakis and Katsikadelis (2003)
predicted strength. The three beam presented an analog equation method as a
configurations tested the effect of shear solution to the case where deformable shear
connection type and transverse reinforcement studs are used. The method neglects uplift, but
placement. One configuration failed by considers in-plane shear forces and
concrete crushing with a longitudinal crack. deformation of the slab and axial forces and
The ultimate strength of the beams exceeded deflection of the steel beam. The results show
the predicted strength by 8%, 4%, and 45%. that as the stiffness of the connection
The configuration that exceeded predicted decreases, the interface slip increases, the
strength by 45% utilized double, but smaller, shear forces at the interface decrease, and
shear studs. These tests demonstrate that the the lateral deflections of the beam increase.
shear connection significantly affects the Liu et al. (2005) investigated strain differences
ultimate strength of a composite beam. Ramm between the steel and concrete in a composite
and Jenisch (1997) investigated interface beam using partial interaction and the theory
longitudinal shear forces and recognized that of elasticity. Taking slip and curvature into
additional transverse bending in the concrete account, it was demonstrated that the strain
slab affects composite beam strength. Full difference affects the displacement and load
scale beam tests were performed to study the capacity of a beam. In comparing full
effect of transverse bending. It was observed interaction and partial interaction, the
that negative transverse bending moments calculated deflection difference was 11% in
reduced the ductility of the composite beams one case, therefore, it was suggested that the
and longitudinal cracks and transverse bending full interaction assumption should not be used
moments significantly influence capacity. for design. Girhammar et al. (1993) proposed
a complex composite beam-column analysis
ANALYTICAL MODELS of partial interaction to determine internal
A number of theoretical/analytical studies have actions and displacements. The analysis
been conducted. Leon and Viest (1997) incorporates a first- and second-order analysis
conducted a review of elastic and inelastic that accounts for interlayer slip. Girhammar et
composite beam theories based on al. (2007) proposed a refined composite beam
incomplete interaction. Assumptions for or beam-column static analysis with partial
inelastic analyses include no friction or bond interaction that accounts for interlayer slip
92
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
where more accurate boundary conditions are differential equation for a partially composite
included. The methods of exact analysis beam loaded transversely as discussed by
presented by Girhammar et al. (2007) are, Girhammar et al.
however, extremely long, intensive, and
detailed resulting in a need to simplify. Kr 2
L L
EI o 1
EI o ...(2)
APPROXIMATE ANALYTICAL EI
PREDICTION METHODS
Fabbrocino et al. (1999) also proposed an
Girhammar et al. (2009) presents an
analysis method to predict composite beam
approximate strength prediction method that
flexural strength. Their results demonstrate that
is simpler and more readily applied. The
actual behavior of a composite beam differs
approximate, simplified method incorporates
from predictions based on a full interaction
an effective bending stiffness that was used
assumption. The Fabbrocino analysis
for buckling in the more exact analysis. This
recognizes that composite beam behavior
analysis uses a differential element to define
depends on slip distribution and the resulting
equations for moment, section shear, interface
forces at the interface. In this model, slip is
shear, deflection, and axial force for a fully
related to the rotation of the beam and
composite member. For approximation of
displacement of the centroids of the two
these terms for a partially composite beam,
sections. By defining this relationship, the
the bending stiffness for the full composite
derivative of the slip is shown to be a function
member is replaced with the effective bending
of the curvature and strains at the centroids of
stiffness for the partially composite member.
the two sections. The dependency of the
While this method is much simpler than the
relationships causes the solution to be involved
previously proposed methods by Girhammar
and non-linear, however, the method predicted
et al. it is suitable for approximation only and
ultimate load to within 1% of experimental
not for codified design. Effective stiffness:
results.
1
EI / EI 0 1 Qiongxi Lui (2011) proposed a composite
EI eff 1 2
EI ...(1)
EI 0 ( / ) ( L)
2 beam ultimate load prediction method that
accounts for reduced flexural rigidity due to
where EI = bending stiffness of the fully cracking of the concrete and slip strain. Total
composite section, EI0 = bending stiffness of slip is determined by integrating the slip strain
the non-composite section, K = slip modulus along the beam length. In addition, as the slip
of shear connection, r = distance between the is greatest at beam ends, end slip is
two centroids, L = length of composite beam, compared to shear connector strength. The
= buckling length coefficient (for a simply method is lengthy, but accuracy was not
supported beam, = 1), L = the partial evaluated against experimental results.
composite action parameter (defined below
Other researchers have proposed
in equation (2)) obtained from the general
composite beam flexural strength prediction
solution, in terms of displacement, to the
93
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
methods involving longitudinal shear stresses and two Timoshenko beams. All three models
in the composite beam. Segura (1990) predicted the ultimate load to within 2%,
proposed a method to evaluate the shear however, where beams were controlled by
stresses at the steel and concrete interface. shear, none of the three models provided
Gara et al. (2010) proposed a strength accurate results. In this case, the models using
prediction method based on the effects of a Timoshenko beam provided much closer
shear lag. Along with others, these methods failure load predictions than the model using
are long and complex, and are not practical two Euler-Bernoulli beams. Earlier work by
for codified design. Ranzi and Zona (2007) presented a detailed
analysis using an Euler-Bernoulli beam for the
FINITE ELEMENT MODELING concrete and a Timoshenko beam for the steel
A number of finite element studies have been to include shear deformation. He et al. (2011)
conducted in the recent several years that have used an element between the steel beam and
investigated a number of issues relating to concrete slab to represent the shear
composite flexural members with shear studs. connection in a composite beam. The element
Mirza and Uy (2008) describe the effects of was assigned a stiffness to model the average
strain profiles on the headed shear stud effect of the shear connectors along the beam.
connection in composite beams. Because The results of this finite element model were
shear studs are subjected to both flexure and compared to two different experimental tests
shear, Mirza and Uy emphasize the importance through the ultimate load. The model prediction
of evaluating the combined effect. The overestimated the ultimate load by 5.5% and
presence of both moment and shear causes 3.5%. da Silva and Sousa (2009) presented
a nonlinear response, therefore ABAQUS was a family of interface elements to account for
used to create a 3D solid element shear stud interlayer slip in a composite beam. Vertical,
model. A push test analysis under different horizontal, and rotational displacement fields
strain regimes was performed to determine were considered. Different degrees of
the shear stud load limit. The finite element freedom were used in the different elements
model predicted the strength of the shear studs with some elements considering shear strain.
to within 0.7% of the experimental data and Da Zilva and Sousa determined that a
concluded that strain regimes in a solid Timoshenko interface element with quadratic
concrete slab do not significantly affect the displacement accounting for shear strain is the
shear stud performance. Ranzi and Zona most reliable.
(2011) compared three different numerical
models incorporating a uniformly deformable AISC STRENGTH PREDICTION
shear connection to link the concrete slab and It is presumed that the reader is familiar with
steel beam. Ranzi and Zona evaluated the AISC strength prediction of composite beams
model using two Euler-Bernoulli beams, a on the basis of both full and partial composite
combination of an Euler-Bernoulli beam action and, therefore, the method is only
(concrete) and a Timoshenko beam (steel), described conceptually here. In general the full
94
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
concrete strength and plastic yield strength of area remained the same. The model was
the materials is expected to develop and the subjected to an incrementally increasing point
internal moment is computed as Mn. In the case load at mid-span to study the load
of partial composite action, the shear concentrations at the shear studs, the shear
connector strength is less than the controlling lag development in the concrete slab, the
internal, longitudinal shear force developed by longitudinal distribution of force in the shear
either the steel or concrete. studs, and composite beam load versus
deflection relationship.
LABORATORY TEST BEAM
AND NUMERICAL ANALYSIS MEASURED AND SIMULATED
Physical Test RESULTS
The tested beam geometry and loading Laboratory Test Results
configuration under the present study is A number of behavior observations from both
presented in Figure 3. Each beam was the laboratory tests and numerical results were
constructed with 24 identical and equally collected. Longitudinal cracks often developed
spaced, ¾ inch diameter headed shear studs. in the concrete slab of a composite beam
The test data sets are presented in Figure 6 directly above the headed shear studs as
and Figure 7. The AISC, Part 16, Chapter I presented in Figure 2. In addition, the slab was
composite beam strength prediction for a autopsied to examine the headed stud
f'c = 4,000 psi, solid concrete slab and W10×17 condition after loading. The four studs at the
A992 steel beam is Mn = 158 ft-kips with a end of the beam were examined (see Figure
corresponding center point load of 31.7 kips. 5) and deflection of each of the four end studs
was measured to be 1/8" toward the end of
Finite Element Model
To better study the composite beam behavior Figure 2: Observed Longitudinal
of the tests, the tested beam was modeled Cracking of Test Beam Slab
numerically using a very simple, 3D, SAP2000
with shell elements for the slab and frame
elements for the beam and studs (see Figure
4). To incorporate material nonlinear behavior,
hinges that allow yield were assigned at
discrete locations along the steel beam.
Modeled concrete material is f'c = 4000 psi
on a 1" × 2" grid. To model the shear studs,
frame elements were used with the cross
section of the shear stud and the properties of
ASTM A108 steel. Shear stud length was
equal to half the thickness of the concrete slab
is used, 17/16", which is less than the actual
length of the shear stud, however, the shear
95
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
Figure 4: SAP2000 Finite Element Model Figure 5: Shear Studs with Concrete
Views of Composite Beam Numerical Removed after Failure
Model of Composite Beam
b) SAP2000 Extruded View of Basic Elements b) Stud Failure after Low Impact
96
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
In comparing the numerical model Uniform shear stud shear force along the
predicted deflections in the elastic range, AISC beam is normally assumed, however,
elastic deflection predictions, and measured numerical modeling demonstrates that shear
deflections, the numerical model predictions forces are concentrated at the end studs.
97
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
98
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
...(3)
2
Vs As F y (4.99 in )(50 ksi) 250 kips
99
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
The predicted end stud shear force of 26 thickness of 2¾", Vmax/Vassumed decreases,
kips loads the stud beyond capacity. Stud therefore, expecting a shear of 1.25 times
shear forces of this magnitude results in stud greater than predicted by AISC remains
deflection, slab to beam slip, and ultimately conservative.
premature failure. Additionally, for this To compare this method to other typical
f'c = 4,000 psi example, the stud shear force composite beams, two additional models were
is only slightly below the value at which the created. The fourth and fifth numerical models
surrounding concrete is expected to fail. were assigned the same material properties
The composite beam laboratory as all previous models. The models included
observations of the studs after load are a single row of shear studs and loaded with a
consistent with this analysis. However, this concentrated load at mid-span. The fourth
evaluation so far is limited to a W10×17 beam model consisted of a W18×35 beam and a
with a 37/8" concrete slab and 24 headed shear 64" wide by 37/8" thick concrete slab with
studs. To evaluate the effect of slab thickness headed studs spaced at 8" and a span length
of 32 ft. The fifth model consisted of a W14×26
and width on the distribution, two more models
beam and an 80" wide by 3" thick concrete
were constructed and evaluated.
slab with headed studs spaced at 10" and a
The initial numerical model was modified span length of 30 ft. The simulation results
to create two additional models: (1) W10×17 demonstrate that these beams also
beam with a 60" wide by 37/8" concrete slab experience end shear stud forces that were
and 24 headed shear studs; and (2) W10×17 20.2% and 17.7% higher than a uniform
beam with a 28" wide by 2¾" concrete slab distribution would predict, respectively. It
and 24 headed shear studs. The beff used for appears, based on this limited study, that
these two models is the AISC maximum larger steel beams will experience larger than
allowable, taking span/8 on each side of the predicted end shear stud forces.
beam. Because Vmax/Vassumed was observed to
DISCUSSION AND
decrease after first steel yield, the mid-span
EVALUATION
load corresponding to the elastic limit was
used. Based on this study, it was determined Composite beam tests demonstrated that
neither the steel beam nor the concrete slab
that Vmax/Vassumed increases as the effective slab
attained full plastic strains despite loading to
width increases. Also, a decrease in Vmax/
Vassumed is observed with a decrease in slab failure, indicating that full composite action, as
normally assumed, did not develop. The error
thickness. In the case of a large effective slab
in predicting ultimate flexural capacity ranged
width, increasing the expected shear by 1.25
may not provide a conservative estimate. from –0.30% (under prediction) to +19.2%
(over prediction) with an average over
However, Vmax/Vassumed for the numerical models
prediction of 6.6%.
with effective widths equal to 60" and 28" are
within 2%. Also, Vmax/Vassumed is expected to To evaluate the degree of composite action
decrease after the steel yields. For a slab that did develop in the test beams, a partial
100
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
composite analysis was conducted. From Figures 7a through 7j present the measure
Equation 8, the flexural capacity of a partially composite beam response as compared to
composite section with the PNA in the top the analysis methods suggested by
flange of the steel is determined. Girhammar et al. (2009) and AISC (2010). A
lower bound moment of inertia (ILB) is used to
d a x
M n Ts C c t s 2( bf F y x ) ...(8) develop the AISC load vs. deflection
2
2 2
relationship. I LB is an AISC, theoretical
Because the calculated total shear stud minimum moment of inertia including only that
strength has been shown to be sufficient, Cc portion of the concrete slab within Whitney’s
becomes an unknown variable of interest. stress block and is therefore considered
Equation 8 can be rearranged by making conservative. For many of the data sets,
several substitutions and Cc can be calculated however, it can be observed that this
from the quadratic Equation 9: seemingly conservative approach is actually
very accurate other than for data sets 1, 2, and
2 3 where the measured elastic deflections were
Cc
1
1 C c t s As
2(0.85)fc ' beff 4 F y bf 2bf
notably greater than deflections predicted
2 ...(9) using ILB.
d As F y PL
As F y 0
2 4 bf 4 The deflections predicted utilizing
Girhammar et al. (2009) are presented in
which is valid when the PNA is in the steel Figures 7a though 7j. Girhammar et al. uses
flange. For the f ' c = 4,000 psi example an effective bending stiffness that accounts for
examined here, the actual C c that was partial composite interaction involving an
experienced is calculated to be 189 kips. estimated slip stiffness, K. For a non-
Knowing the value of x, the tensile force composite section, the slip stiffness is zero
component of the flexural moment at failure can and increases with higher degrees of
be calculated. This evaluation demonstrates composite interaction. The slip stiffness, K, is
that 86% of the steel plastic tensile strength difficult to estimate because the slip is small
was coupled against the concrete at the time and concrete properties are nonlinear. It was
of flexural failure. Among the ten sets of determined that using a slip stiffness of 9 ksi
measure beam response data, the developed for every test data set provided an accurate or
tension ranged from 64% in one test to two somewhat conservative estimate. While the
tests of 100% with an average of 87%. Girhammar et al. method is attractive and can
provide accurate predictions, a reliable
Load versus deflection plots are also
method to quantify the slip stiffness is not
presented for measured composite beam
available.
responses. Figure 6 presents a compilation
of all data sets normalized to predicted SUMMARY AND
ultimate flexural strength due to different CONCLUSIONS
measured concrete strengths with each test.
Composite steel-concrete beams tested to
101
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
102
Int. J. Struct. & Civil Engg. Res. 2013 Jeffrey A Laman and David Leaf, 2013
103