You are on page 1of 9

INELASTIC ANALYSIS OF STEEL FRAMES WITH COMPOSITE BEAMS

By J. Y. Richard Liew,1 Hong Chen,2 and N. E. Shanmugam3

ABSTRACT: This paper describes a method of inelastic analysis that provides the necessary degree of accuracy
for studying the limit-state behavior of steel frames with composite floor beams subjected to the combined action
of gravity and lateral loads. An inelastic formulation is proposed to model the composite beams based on the
moment-curvature relationship of a composite beam section under positive and negative moments. Steel columns
are modeled using the plastic hinge approach. To ascertain the accuracy of the composite beam model, two
composite beams and a steel portal frame are analyzed and the results are compared with those obtained from
tests and the more established results. Finally, the robustness of the model is demonstrated by studying 2D and
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

3D building frames using various floor beam models so that their effects on the serviceability deflection and
limit load can be compared.

INTRODUCTION model to investigate the plastic response of composite steel-


concrete girder bridges. Hirst and Yeo (1980) modeled com-
Most steel buildings built in recent years have used concrete posite beams with eight-node isoparametric finite elements,
floor slabs designed to act compositely with steel beams by and stud connectors with quadrilateral elements Razaqpur and
means of shear connectors. On the other hand, there are eco- Nofal (1989) developed a 3D bar element to model shear con-
nomic and structural benefits to utilizing the partially re- nectors. The webs of an I-beam are modeled with plane quad-
strained composite connections with some degree of continuity rilateral elements, and the slab the flanges of the I-beam are
but without the disadvantages associated with the fully rigid modeled with flat shell elements. However, the use of contin-
approach (Nethercot 1995; Viest et al. 1997; ASCE 1998). It uum finite elements for the analysis of composite beams is
is a great advantage to adopt the partially restrained composite usually computationally too intensive with the increase of
frames for low-to-moderate height unbraced frames (ASCE structural size. The cost and effort of such a method are so
1998). great that they often prohibit analysis of a complete frame-
When building frames are subjected to gravity and lateral work.
loads, the distribution of the bending moment in the composite To reduce the computational effect, researchers have pro-
beams varies along the member length, as shown in Fig. 1. In posed the ‘‘line’’ element approaches. The composite beam
the negative moment region, the concrete in tension is cracked element with bond slip was proposed by Arizumi et al. (1981),
and the steel reinforcement in the slab may yield. In the pos- Daniels and Crisinel (1993), and Oven et al. (1997). It consists
itive moment region, a large bending moment may cause of two beam components for the concrete slab and steel girder
yielding of the steel section and crushing of the concrete. Con- and a continuous spring for shear connectors. Hajjar et al.
sequently, the flexural stiffness of beams varies along the (1998) applied a similar method to study the effect of inter-
member length. The nonlinear elastoplastic behavior of the layer slip on the response of composite frames with concrete-
composite beam must be considered in the limit-state analysis. filled tubes. Salari et al. (1998) proposed a composite beam
A number of models have been proposed to model com- element based on the force-based finite-element framework
posite beams. The 1D partial action theory for composite (Spacone et al. 1996), with computational accuracy and effi-
beams was first developed by Newmark et al. (1951). The ciency. However, all these methods include additional degrees
basic equilibrium and compatibility equations for an element of freedom at the element ends, allowing the concrete and steel
of the beam are reduced to a single second-order differential to have independent displacements. When modeling the par-
equation in terms of either the resultant axial force in the con- tially restrained composite frames, it is difficult to enforce the
crete or the interface slip. This approach was further developed compatibility conditions at the semirigid composite connec-
by Yam and Chapman (1972) to incorporate the nonlinear ma- tions.
terial and shear connector behaviors. Al-Amery and Roberts In this research work, a composite beam mode is proposed
(1990) proposed an alternative approach in which the equilib- to provide the necessary degree of accuracy for studying the
rium and compatibility equations are formulated in terms of
the displacements of concrete and steel. The resulting differ-
ential equations are solved by expressing the displacement de-
rivatives in the finite-difference form. However, these methods
are generally applied to the analysis of simple or continuous
beams and are inappropriate for the integrated analysis of the
frame system.
Wegmuller and Amer (1977) proposed a layered beam-plate
1
Assoc. Prof., Dept. of Civ. Engrg., Nat. Univ. of Singapore, BLK E1A
No. 07-03, 1 Engrg. Dr., Singapore 117576.
2
Res. Engr., Dept. of Civ. Engrg., Nat. Univ. of Singapore, Singapore
117576.
3
Prof., Dept. of Civ. Engrg., Nat. Univ. of Singapore, Singapore
117576.
Note. Associate Editor: Sashi Kunnath. Discussion open until July 1,
2001. To extend the closing date one month, a written request must be
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on August 20, 1999.
This paper is part of the Journal of Structural Engineering, Vol. 127,
No. 2, February, 2001. 䉷ASCE, ISSN 0733-9445/01/0002-0194–0202/ FIG. 1. Continuous Composite Beam Subjected to Gravity
$8.00 ⫹ $.50 per page. Paper No. 21689. and Lateral Loads

194 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:194-202.


inelastic behavior of 3D steel frames with composite floor
beams. The composite beam is subdivided along the length
into a finite number of segments. The flexural stiffness of a
segment can be evaluated using the moment-curvature M-⌽
relationship of the composite beam section, which can ap-
proximately account for the partial interaction between the
concrete slab and steel beam. The static condensation method
FIG. 3. Member Analysis
is then used to reduce the segment model to a single beam-
column element. The proposed formulation is not only appli-
cable for the analysis of composite beams, it can also be used discretization, and can capture the inelastic effect in steel
for the inelastic analysis of steel frames if the moment-cur- frames more accurately than the plastic hinge analysis. More-
vature-thrust M-⌽-P relationship of the steel section can be over it can be applied to the inelastic analysis of composite
beams if the M-⌽ relationships of composite beam sections
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

calculated.
The advantage of the proposed approach is that it does not are used.
involve fundamental changes to the existing beam-column for- Simplified M-⌽-P equations have been proposed to relate
mulation for frame analysis. Although the present study fo- the in-plane bending moment acting on and the curvature ex-
cuses mainly on frames with fully rigid beam-to-column con- perienced by a cross section under the axial force (Liapunov
nections, the proposed composite beam model can easily be 1974; Chen and Atsuta 1976; Attalla et al. 1994). Atalla’s M-
implemented into the existing semirigid frame analysis (Hsieh ⌽-P equation is obtained through the calibration with the
and Deierlein 1991) with the proper moment-rotation models force-strain data from the fiber-element analysis (Attalla et al.
for composite connections. Hence the integrated analysis can 1994) in which the cross section is divided into a finite number
be performed to study the inelastic behavior of partially re- of meshes. The strain distribution across the section is linearly
strained composite frames. The proposed formulation is im- varying, and the axial force and bending moment can be de-
plemented and tested using several problems known to be sen- termined based on the stress-strain relationship. Iteration is
sitive to spread-of-plasticity effects. The nonlinear inelastic performed to converge on the values of the strain until the
behavior of building frames is investigated by using various difference between the axial force of the section and the ap-
floor beam models so that their effects on the serviceability plied axial load is less than a predefined tolerance. The pro-
deflection and limit load can be compared. cedure is repeated by increasing the curvature to obtain the
M-⌽ relationship for a given axial force P.
Attalla’s M-⌽-P equation is adopted because it can predict
MODELING OF STEEL BEAM COLUMNS the cross-sectional behavior more accurately than other forms
The proposed 3D beam-column element consists of 12 de- of M-⌽-P equations. The only limitation is that it is a uniaxial
grees of freedom, as shown in Fig. 2. Plasticity is formulated bending model. Further extension of the equation to the axial
based on the cross-sectional stress-resultant constitutive force and biaxial bending is required to capture the 3D beam-
model, which represents the plastic interaction between the column behavior.
axial force and biaxial bending moments (Orbison et al. 1982).
The plastic hinge analysis, which has been proved to be com- MODELING OF STEEL-CONCRETE COMPOSITE
putationally efficient, can yield accurate results for reasonably BEAMS
configured structures in which several hinges form before the In the proposed model, a composite beam is subdivided into
inelastic limit point is reached. However, the method has been a finite number of segments to capture the varying flexural
shown to overestimate the limit load when the structural limit stiffnesses along the member length, as shown in Fig. 3. The
state is dominated by the instability of a few members. In such instantaneous flexural stiffness of a cross section can be de-
cases, the more refined approaches, such as the refined plastic rived using the M-⌽ relationships. Hence, section discretiza-
hinge analysis (Liew et al. 1993) and the quasi-plastic hinge tion is not required. The fiber-element analysis (Attalla et al.
analysis (Attalla et al. 1994), which can simulate the spread- 1994) may be used to generate the M-⌽ relationship of the
of-plasticity effect, can be used for the planar frame analysis composite beam section. However, the computational proce-
and the two-surface plastic hinge analysis can be used for 3D dure involves many curvature steps and the M-⌽ relationship
frame analysis (Liew and Tang 2000). is not a closed-form one. As the analysis assumes plane sec-
An inelastic analysis method based on the M-⌽-P relation- tions remaining plane after deformation, the exact modeling
ship of the cross section is proposed in this section. The beam of the partial shear connection is difficult.
column is subdivided into a finite number of segments to cap- Li et al. (1993) investigated a large number of variables that
ture plasticity spreading along the member length, as shown influence the M-⌽ relationship of a composite beam section
in Fig. 3. The instantaneous flexural stiffness, which can be with full or partial shear connections. For the composite beam
derived from the M-⌽-P relationship, is used in the element section shown in Fig. 4, the M-⌽ relationship in the positive
stiffness formulation to consider the cross-sectional plastifi- moment region is given by
cation. The proposed approach is more efficient than the plas-
M
tic zone analysis (White 1985), which requires cross-sectional ⌽(M) = , 0 ⱕ M ⱕ My (1a)
EI

冉 冊冉 冊
2
M Mu M ⫺ My
⌽(M) = ⫹ ⌽u ⫺ , My < M ⱕ Mu (1b)
EI EI Mu ⫺ My
where EI = elastic flexural stiffness of the composite section
under the positive moment; E = elastic modulus of steel; My
= yield moment; Mu = ultimate moment; ⌽y = curvature at the
yield moment, and ⌽u = curvature at the ultimate moment
given by
My
⌽u = ␤⌽y = ␤ (2)
FIG. 2. Plastic Hinge Beam-Column Element EI

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 195

J. Struct. Eng. 2001.127:194-202.


they refer to the negative moment region. In determining EI ⬘,
M ⬘y , and M ⬘u , the contribution of the effectively anchored re-
bars located within the effective width of the concrete slab can
be included and the contribution of concrete in tension is ig-
nored. If detailed information on rebars is available, (6) can
be used to evaluate the flexural stiffness of he composite beam
under the negative moment. If major reinforcement bars are
not present in the slab, the composite beam under the negative
moment behaves similar to the steel section. Hence Attalla’s
M-⌽-P equation may be used to determine the flexural stiff-
ness.
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

STRATEGIES FOR INELASTIC ANALYSIS


An attempt has been made to approximately account for the
spread-of-plasticity effects, using the simplified expressions
FIG. 4. Composite Beam Section
for the M-⌽ relationships of steel and composite sections. An
important aspect of the proposed approach is that it does not
in which ␤ = 5.7(D/Ds)0.2; D = depth of steel beam; and Ds = require fundamental changes to the existing beam-column for-
overall slab depth (Fig. 4). mulation. The main disadvantage is that the M-⌽ relationships
The ultimate moment Mu with full or partial shear connec- of all sections must be predetermined before a global analysis
tion may be determined by fiber-element analysis (Attalla et can be performed. However, such a procedure can be fully
al. 1994) or design specification procedures [British Standards automated and integrated into the preprocessing software. The
Institution (BSI) 1990; American Institute of Steel Construc- following is a summary of the strategies and computer imple-
tion (AISC) 1993]. The elastic flexural stiffness of the com- mentation of the proposed approach.
posite beam under the positive moment may be calculated To reduce the computational efforts, the lengthwise discret-
(AISC 1993) ization can be minimized by reducing the number of internal


nodes. This can be achieved by performing the yield check at
N the midpoint of segments along the member length, as shown
EI = EIs ⫹ (EIf ⫺ EIs) (3)
Nf in Fig. 3. If several continuous segments are found to be in
the elastic state, they are combined to form one internal ele-
in which EIf = flexural stiffness of the composite beam with ment. The member flexural stiffness is normally large, and the
full shear connection; EIs = flexural stiffness of the steel sec- lateral displacement along the member length is generally
tion; and N/Nf = degree of shear connection. small. Hence the beam member is assumed to be straight in
The instantaneous flexural stiffness of the composite beam assembling the stiffness matrices of internal elements. Thus
section under the positive moment is obtained by the computational effort in storing and updating the internal
nodal coordinates can be reduced.
dM The stiffness equation for the composite beam member with
EIt = = BEI (4)
d⌽ internal nodes is given by
where B = stiffness degradation ratio given by
EIt
[k] 再 冎 冋
du
dui
=
k11
k21 册再 冎 再 冎
k12
k22
du
dui
=
df
dfi
(8)
B= = 1, 0 ⱕ M ⱕ My (5a)
EI in which [k] = member stiffness matrix obtained by assembling
the stiffness matrices of internal elements; {df } and {du} =
EIt 1 incremental force and displacement vectors, respectively, of
B= = , My < M ⱕ Mu (5b)
EI 2(␤␣y ⫺ 1)(m ⫺ ␣y) member ends; and {dfi} and {dui} = incremental force and
1⫹
(1 ⫺ ␣y)2 displacement vectors, respectively, of internal nodes.
As the forces at internal nodes {dfi} = 0, (8) can be written
in which the moment ratio is m = M/Mu ; and the yield moment in terms of the forces and displacements at the member ends
ratio ␣y = My /Mu . through the static condensation
For the negative moment region, it is assumed that, once
cracked, the concrete slab does not contribute to the beam’s ([k11] ⫺ [k12][k22]⫺1[k21]){du} = [kie]{du} = {df } (9)
resistance. Applying the same rule for the positive moment
region, the M-⌽ relationship under the negative moment is (Li in which [kie] = 12 ⫻ 12 inelastic stiffness matrix representing
et al. 1993) the spread-of-plasticity effect across the cross sections and
along the member length. When the moment reaches the sec-
M⬘ tion capacity, the flexural stiffness approaches zero. The matrix
⌽⬘(M) = , 0 ⱕ M⬘ ⱕ M ⬘y (6a)
EI⬘ [k22], which appears in the matrix [kie], will be singular and

冉 冊冉 冊
2 cannot be inverted. To avoid numerical difficulties in inverting
M⬘ M ⬘u M⬘ ⫺ M ⬘y the matrix [k22], a minimum value of EIt /EI = 0.001 is chosen.
⌽⬘(M) = ⫹ ⌽⬘u ⫺ , My⬘ < M ⬘ ⱕ M⬘u
EI⬘ EI⬘ M ⬘u ⫺ M ⬘y Convergence studies indicate that no more than 16 seg-
(6b) ments/member are required to capture the spreading of plas-
ticity along the member length. Because yielding normally oc-
in which curs in the large moment regions, the number of internal
M ⬘y elements is much smaller than the number of segments. Fur-
⌽⬘u = 5.3⌽⬘y ⫹ 2.4 = 5.3 ⫹ 2.4 (7) ther research may be carried out to derive the closed-form
EI ⬘
inelastic stiffness matrix; hence, the computation can be sig-
in meters of radians per meter and the variables with prime nificantly reduced by avoiding the procedure of static conden-
(⬘) have the same meaning as defined in (1), except that here sation.
196 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:194-202.


Eq. (9) is the first-order inelastic formulation at the member In the inelastic analysis, the beam is modeled as eight ele-
level. For a second-order inelastic analysis, it is modified to ments with equal lengths. The uniformly distributed load is
lumped at the internal nodes. The instantaneous flexural stiff-
([kie] ⫹ kg ){du} = {df } (10) ness of the composite section under the positive moment is
in which [kg] = 12 ⫻ 12 geometric stiffness matrix for the used in the inelastic analysis. As shown in Fig. 5(b), the load-
beam-column element (Chen 2000; Liew et al. 2000). deflection curve and limit load predicted by the present anal-
The proposed inelastic formulation has been implemented ysis compare well with those from the finite-difference
in a nonlinear frame analysis program. The program contains method. Fig. 6 shows the moment and EIt /EI diagrams at the
simple incremental and incremental/iterative schemes with limit load. The midspan moment is less than the ultimate mo-
adaptive load increment control (Chen 2000; Liew et al. 2000). ment Mu at the limit load; however, the flexural stiffness has
The element forces and geometry configurations are updated degraded significantly in the part of the beam with a moment
at the end of each iterative step. The rigid-body displacements greater than the yield moment. The failure of the composite
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

serve to rotate the initial forces acting on the element from the beam is due to excessive yielding of the cross section near the
previous configuration to the current configuration, whereas midspan.
the natural deformations constitute the only source for gener-
ating the force increments.
For normal building frameworks, the floor slab may be
modeled as a rigid diaphragm, which is assumed to provide
infinite in-plane stiffness without any out-of-plane stiffness.
The lateral response of the floor slab is characterized by two
translational and one rotational degrees of freedom located at
the floor master node. The analytical methods described in this
paper can be used to analyze 3D frameworks with or without
a rigid floor diaphragm.

VERIFICATION STUDIES
The selected problems for the studies consist of composite
beams and small frames rather than large and more practical
frames because the limit state behavior of such structures is
more sensitive and tends to be dominated by the spreading of
plasticity in individual members.

Simply Supported Composite Beam


The insert in Fig. 5(b) shows a simply supported composite
beam with a 9-m span, subjected to the uniformly distributed
load. It was analyzed by Al-Amery and Roberts (1990) using
the finite-difference method and considering the effect of par-
tial interaction. Properties of the composite beam section are
that the effective width of the slab Be = 1,800 mm, overall
slab depth Ds = 150 mm, steel beam depth D = 412 mm, deck
profile depth Dp = 0, web thickness tw = 9.4 mm, flange width
Bf = 153 mm, and flange thickness tf = 16 mm. The cube
strength of concrete fcu = 30 N/mm2, the yield strength of steel
fy = 280 N/mm2, and the elastic modulus of steel E = 2 ⫻ 105
N/mm2.
The connection between the concrete slab and steel beam is
assumed to be produced by pairs of 10-mm-diameter, 100-
mm-long headed studs, with a spacing of 240 mm. The char-
acteristic resistance of a headed shear stud Qk = 100 kN. The
design capacity of the shear stud (Q = 0.576 and Qk = 57.6 FIG. 5. Inelastic Analysis of Simply Supported Composite
Beam: (a) Flexural Stiffness-Moment Relationship of Composite
kN) is reduced by assuming that lightweight aggregate con- Beam Section; (b) Load-Deflection Curves
crete was used and the steel decking is perpendicular to the
beam. The number of connectors that can be accommodated
in the half span N = 2 ⫻ 4,500/240 = 36. For full composite
action, the number of connectors per half span required Nf =
Rs /Q = 2,371/57.6 = 42, where Rs is the tensile capacity of the
steel section. The degree of shear connection N/Nf = 0.85. For
the composite beam section under the positive moment, the
ultimate moment Mu = 706 kN/m and the yield moment ratio
␣y = 0.74. The flexural stiffness of the steel section EIs = 4.7
⫻ 104 kN/m2. Assuming the effective modular ratio for an
office-type building ␣e = 15 for lightweight concrete, the flex-
ural stiffness of the composite beam with full shear connection
EIf = 1.45 ⫻ 105 kN/m2. Based on (3), the flexural stiffness
with partial shear connection is EI = 1.37 ⫻ 105 kN/m2. The
flexural stiffness-moment relationship of the composite section FIG. 6. Simply Supported Composite Beam at Limit Load: (a)
under the positive moment is shown in Fig. 5(a). Moment Diagram; (b) EIt /EI Diagram

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 197

J. Struct. Eng. 2001.127:194-202.


Two-Span Composite Beam nection is N/Nf = 0.9 and the contribution of rebars is ignored.
The composite beam under the negative moment behaves sim-
Slutter and Driscoll (1965) tested a series of simply sup- ilar to the steel section, and Attalla’s M-⌽-P relationship is
ported and continuous composite beams. Beam B13, which is used to determine the flexural stiffness in the negative mo-
one of the continuous-beam series tested, is analyzed here. The ment. The properties of the composite section under the pos-
insert in Fig. 7(b) illustrates the two-span composite beam itive moment are calculated as Mu = 287 kN/m, EI = 4.30 ⫻
with two concentrated loads applied on each span. Properties 104 kN/m2, and ␣y = 0.72. The properties of the steel section
of the composite beam section are Be = 1,219 mm, Ds = 102 are that the plastic moment Mp = 157 kN/m, EI = 1.70 ⫻ 104
mm, Dp = 0, D = 304 mm, tw = 6.02 mm, Bf = 165 mm, and kN/m2, and yield moment ratio ␣⬘y = 0.63. The flexural stiff-
tf = 10.16 mm. The cylinder strength of concrete f ⬘c = 16 ness-moment relationships of the composite section are shown
N/mm2, the yield strength of steel fy = 252.4 N/mm2, and the in Fig. 7(a). When the moment is greater than the yield mo-
elastic modulus of steel E = 2 ⫻ 105 N/mm2. ment, the flexural stiffness of the composite action under the
Because detailed information of the shear studs and rebars
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

positive moment degrades faster than that of the steel section.


is not available, it is assumed that the degree of shear con- The load-deflection curves shown in Fig. 7(b) indicate that
there is good correlation between the test results and predicted
behavior when using the proposed inelastic analysis. Fig. 8
shows the moment and EIt /EI diagrams at the limit load. The
composite beam reaches the limit load due to the excessive
yielding and stiffness degradation in the area of the two side
midspan sections and middle support section.

El-Zanaty’s Portal Frame


El-Zanaty’s portal frame (Fig. 9) is one of the North Amer-
ican calibration frames for advanced analysis due to its sen-
sitivity to the spreading of plasticity (Chen and Toma 1994).
The frame has little capacity for inelastic redistribution of
forces, and its limit load is governed by the buckling of two
columns. The frame was analyzed by White (1985), using the
plastic zone analysis in which the members are discretized
both along their lengths and through their sections to track the

FIG. 9. El-Zanaty’s Portal Frame

FIG. 7. Inelastic Analysis of Two-Span Composite Beam: (a)


Flexural Stiffness-Moment Relationship of Composite Beam
Section; (b) Load-Deflection Curves

FIG. 8. Two-Span Composite Beam at Limit Load: (a) Moment FIG. 10. Flexural Stiffness-Moment Relationships of W8ⴛ31
Diagram; (b) EIt /EI Diagram Section

198 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:194-202.


gradual spreading of plasticity through the volume of the mem- ANALYSIS OF STEEL FRAMES WITH COMPOSITE
bers. BEAMS
In the present study, the frame is analyzed by using two
approaches: the plastic hinge analysis and the proposed in- Steel Portal Frame with Composite Beams
elastic analysis based on the flexural stiffness of the section
evaluated by using Attalla’s M-⌽-P relationship. Fig. 10 shows The insert in Fig. 12(a) shows a portal frame that consists
the flexural stiffness-moment relationship of the W8⫻31 sec- of a composite beam rigidly connected to two steel columns
tion. For a fixed axial force, the instantaneous flexural stiffness subjected to vertical and lateral loads. The dimensions and
is equal to the elastic flexural stiffness when the bending mo- properties of the composite beam section are the same as those
ment is less than the yield moment. When the bending moment of the two-span composite beam. The cylinder strength of con-
increases from the yield moment to the moment capacity, it crete f c⬘ = 16 N/mm2, the yield strength of steel fy = 252.4
changes from the elastic flexural stiffness to zero. It decreases N/mm2, and the elastic modulus of steel E = 2 ⫻ 105 N/mm2.
To study the spreading of plasticity in the composite beam,
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

significantly with the increase of both axial force and bending


moment. the W12⫻50 section is used for the columns to achieve a
The analysis results shown in Fig. 11(a) indicate that the typical weak-beam, strong-column frame model.
limit loads of the frame are overestimated by the plastic hinge In addition to simulating the composite frame behavior with
analysis. The proposed inelastic analysis can predict the load- the proposed inelastic analysis based on the M-⌽ relationship,
displacement curves and limit loads more closely than those three more cases have been studied by using the following
from the plastic zone analysis. The results from the plastic approaches:
hinge analysis shown in Fig. 11(b) indicate that the classic
‘‘plastic mechanism’’ has not yet formed at the limit load for • Proposed inelastic analysis of pure steel frame based on
the frame with P/Py = 0.4. The limit load is reached due to the M-⌽ relationship
the combined effect of inelastic deformation, axial force, and • Second-order elastic analysis with flexural stiffness
second-order displacements to reduce the structural stiffness. throughout the composite beam approximated by (Viest
The EIt /EI diagram from the proposed inelastic analysis shown et al. 1997)
in Fig. 11(c) indicates that, at the limit load, the flexural stiff-
ness decreases significantly at the top of the two columns due Ic = 0.6I ⫹ 0.4I ⬘ (11)
to the combined actions of axial force and bending moment. in which I and I ⬘ = moments of inertia of the composite
The failure of the frame is due to progressive yielding in col- section under the positive and negative moment regions,
umns, leading to significant stiffness degradation and sidesway respectively
deflection.

FIG. 11. Inelastic Analysis of El-Zanaty’s Portal Frame: (a)


Load-Displacement Curves; (b) Formation of Plastic Hinge; (c) FIG. 12. Inelastic Analysis of Steel Portal Frame with Com-
EIt /EI Diagram posite Beam: (a) Load-Displacement Curves; (b) EIt /EI Diagram

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 199

J. Struct. Eng. 2001.127:194-202.


• Second-order elastic analysis with flexural stiffness beam under the negative moment reaches its nominal strength
throughout the composite beam approximated by at P = 61.2 kN. The predicted limit load is 75% of that of the
composite frame analyzed by the proposed inelastic analysis.
Ic = 0.4I ⫹ 0.6I ⬘ (12) It can be seen from Fig. 12(a) that the proposed weighted
Load-displacement curves obtained by different analysis ap- composite beam stiffness ([(12)] can be used to calculate the
proaches are shown in Fig. 12(a). The pure steel frame col- serviceability lateral stiffness more accurately than (11). This
lapses at P = 62.2 kN. If the composite action between the is due to significant stiffness degradation of the composite
steel beam and concrete slab is considered, the frame collapses beam under the negative moment.
at P = 81.6 kN when a mechanism forms due to the significant
20-Story Steel Frame with Composite Beams
stiffness degradation at the beam ends. This can be seen from
the EIt /EI diagram of the composite frame at the limit load Fig. 13 shows a 20-story space frame with composite beams-
[Fig. 12(b)]. The limit load of the frame, while considering A50 steel is used for all steel sections with fy = 344.8
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

the composite action, is increased by 31%. N/mm2 and E = 2 ⫻ 105 N/mm2, overall slab depth Ds = 127
A practical way to design the frame structures is to obtain mm, and cylinder strength of concrete f c⬘ = 27.6 N/mm2. Full
the member forces by using a direct second-order elastic anal- shear connection is assumed for all composite beams. The con-
ysis. The strength limit state is defined as when the nominal tribution of rebars is assumed to be negligible and therefore
strength of any of the structural elements is reached. Such an ignored in the calculation of the beam capacity. Dimensions
approach is referred to as the ‘‘component’’ limit-state design. and section properties of the composite beams are given in
Second-order elastic analyses with composite beam stiffnesses
approximated by (11) and (12) predict that the right end of the

FIG. 13. 20-Story Steel Frame with Composite Beams: (a)


Plan View; (b) Elevation View FIG. 14. Load-Deflection Curves of 20-Story Frame

TABLE 1. Properties of Composite Beam Sections in 20-Story Frame


Under Positive Moment Under Negative Moment
Be Ds Dp Steel Mu EI Mp EI ⬘
Number (mm) (mm) (mm) section (kN/m) (kN/m2) ␣y (kN/m) (kN/m2) ␣⬘y
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
B1 1,829 127 0 W21⫻57 1,304 2.41 ⫻ 105 0.71 729 9.74 ⫻ 104 0.60
B2 914 127 0 W16⫻36 645 9.03 ⫻ 104 0.73 362 3.73 ⫻ 104 0.62
B3 914 127 0 W12⫻26 413 4.79 ⫻ 104 0.72 185 1.70 ⫻ 104 0.63
B4 1,829 127 0 W12⫻26 446 5.69 ⫻ 104 0.71 185 1.70 ⫻ 104 0.63

200 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:194-202.


Table 1. The frame is analyzed for the combination of gravity Plastic hinge analysis carried out on the pure steel frame
loads = 4.8 kN/m2 and wind loads = 0.96 kN/m2, acting in the model shows that the first plastic hinge occurs at the load ratio
y-direction. Rigid floor diaphragm action is assumed in the of 0.784 and the limit load of the frame is reached at the load
global analysis. Each steel column is modeled using one plas- ratio of 1.031. The load-deflection curves of Nodes A and B
tic hinge beam-column element. Each beam is modeled using (Fig. 14) at the top of the frame are shown in Fig. 15. Because
four elements. The beams are mainly subjected to the bending the frame is asymmetrical in plan, the center of the y-direction
moment about the major axis, and the effect of the axial force wind loads is not aligned with the center of lateral stiffness
and minor-axis bending moment are not significant. and the torsional displacements are induced. This is indicated
by the difference in the lateral deflections at Nodes A and B.
Although the right side of the frame (with respect to the front
elevation) stiffens around the load ratio of 0.9, the center of
lateral stiffness shifts farther to the right due to the further
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

formation of plastic hinges in the left half of the frame. The


lever arm of the lateral load is increased and greater torsional
displacement at Node A is induced to counteract the lateral
displacement at Node B. The formation of plastic hinges is
shown in Fig. 15. At the limit load, most plastic hinges occur
in the beams of the right side of the frame.
In the analysis of the frame while considering the composite
beam effect, the composite beams are analyzed using an in-
elastic approach based on the M-⌽ relationship. The frame
collapses at the load ratio of 1.338, which represents a 30%
increase of limiting strength compared with that of the pure
steel frame. Fig. 16 shows a plot of the EIt /EI ratio of the
third-story floor beams at the limit load. It can be observed
that most beams along the x-axis remain elastic, whereas for
beams in the y-direction, the flexural stiffness degrades sig-
nificantly toward the end region of the beams that are consis-
tent with the plastic hinge positions in the pure steel frame
model.
Second-order elastic analysis with weighted composite
beam stiffness [(12)] predicts that, at the load ratio of 0.773,
the nominal strength is reached at the end of the third-story
beam, which is the same location as in the pure steel frame
model. The strength increase of the composite frame due to
the inelastic redistribution of forces beyond the first member
nominal strength is 73%. It can be seen from Fig. 14 that the
use of the proposed weighted composite beam stiffness pro-
vides a good estimation of the lateral deflection of composite
frames up to the component limit state.
CONCLUSIONS
Inelastic analysis of steel frames with composite floor beams
is presented. The proposed composite beam model is based on
the closed-form M-⌽ relationships, which can approximately
account for the partial interaction between the concrete slab
FIG. 15. Formation of Plastic Hinges in Pure Steel Frame and steel beam. The model can capture the spreading of plas-

FIG. 16. EIt /EI Diagram of Composite Beams in 20-Story Frame

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 201

J. Struct. Eng. 2001.127:194-202.


ticity in composite beams with computational efficiency and strength analysis. Part I: Calculation procedure.’’ J. Struct. Engrg.,
the necessary degree of accuracy. It has been verified by com- ASCE, 119(1), 16–35.
Hajjar, J. F., Schiller, P. H., and Molodan, A. (1998). ‘‘A distributed plas-
paring the predicted results with the established results avail- ticity model for concrete-filled steel tube beam-columns with interlayer
able from the literature. The model does not involve funda- slip.’’ Engrg. Struct., 20(8), 663–676.
mental changes to the existing beam-column formulation. It Hirst, M. J. S., and Yeo, M. F. (1980). ‘‘The analysis of composite beams
can be easily implemented into the existing semirigid frame using standard finite element programs.’’ Comp. and Struct., 11(3),
analysis with the proper moment-rotation models for compos- 233–237.
ite connections. Hence the integrated analysis can be per- Hsieh, S. H., and Deierlein, G. G. (1991). ‘‘Nonlinear analysis of three-
dimensional frames with semi-rigid connections.’’ Comp. and Struct.,
formed to study the inelastic behavior of partially restrained 41(5), 995–1009.
composite frames. Li, T. Q., Choo, B. S., and Nethercot, D. A. (1993). ‘‘Moment curvature
Studies indicate that the limit load of steel frames while relations for steel and composite beams.’’ J. Singapore Struct. Steel
considering the composite beam effect is about 30 % higher Soc., Steel Struct., Singapore, 4(1), 35–51.
Downloaded from ascelibrary.org by University of California, San Diego on 02/01/15. Copyright ASCE. For personal use only; all rights reserved.

than that of the pure steel frames and the lateral stiffness can Liapunov, S. (1974). ‘‘Ultimate strength of multistory steel rigid frames.’’
be significantly enhanced by considering the composite action. J. Struct. Div., ASCE, 100(8), 1643–1655.
Liew, J. Y. R., Chen, H., Shanmugam, N. E., and Chen, W. F. (2000).
Second-order elastic analysis with the composite beam stiff- ‘‘Improved nonlinear plastic hinge analysis of space frame structures.’’
ness adjusted by (12) can be used to evaluate the serviceability Engrg. Struct., 22(10), 1324–1338.
deflection with good accuracy. By using the integrated inelas- Liew, J. Y. R., and Tang, L. K. (2000). ‘‘Advanced plastic hinge analysis
tic analysis, design economy can be achieved if a higher pro- for the design of tubular space frames.’’ Engrg. Struct., 22(7), 769–
portion of the reserved strength built into the composite beams 783.
can be mobilized. Liew, J. Y. R., White, D. W., and Chen, W. F. (1993). ‘‘Second-order
refined plastic-hinge analysis of frame design, Parts I and II.’’ J. Struct.
Engrg., ASCE, 119(11), 3196–3237.
ACKNOWLEDGMENT Nethercot, D. A. (1995). ‘‘Semirigid joint action and the design of non-
sway composite frames.’’ Engrg. Struct., 17(8), 554–567.
This work is funded by research grants (RP3981614) made available
Newmark, N. M., Siess, C. P., and Viest, I. M. (1951). ‘‘Tests and analysis
by the National University of Singapore.
of composite beams with incomplete interaction.’’ Proc. Soc. Exp.
Stress Anal., 9, 75–92.
APPENDIX. REFERENCES Orbison, J. G., McGuire, W., and Abel, J. F. (1982). ‘‘Yield surface ap-
plications in nonlinear steel frame analysis.’’ Comp. Methods Appl.
American Institute of Steel Construction (AISC). (1993). Load and re- Mech. Engrg., 33(1), 557–573.
sistance factor design specifications, 2nd Ed., Chicago. Oven, V. A., Burgess, I. W., Plank, R. J., and Wali, A. A. A. (1997). ‘‘An
Al-Amery, R. I. M., and Roberts, T. M. (1990). ‘‘Nonlinear finite differ- analytical model for the analysis of composite beams with partial in-
ence analysis of composite beams with partial interaction.’’ Comp. and teraction.’’ Comp. and Struct., 62(3), 493–504.
Struct., 35(1), 81–87. Razaqpur, A. G., and Nofal, M. (1989). ‘‘A finite element for modelling
Arizumi, Y., Hamada, S., and Kajita, T. (1981). ‘‘Elastic-plastic analysis the nonlinear behaviour of shear connectors in composite structures.’’
of composite beams with incomplete interaction by finite element Comp. and Struct., 32(1), 169–174.
method.’’ Comp. and Struct., 14(5-6), 453–462. Salari, M. R., Spacone, E., Shing, P. B., and Frangopol, D. M. (1998).
ASCE Task Committee on Design Criteria for Composite Structures in ‘‘Nonlinear analysis of composite beams with deformable shear con-
Steel and Concrete (1998). ‘‘Design guide for partially restrained com- nectors.’’ J. Struct. Engrg., ASCE, 124(10), 1148–1158.
posite connections.’’ J. Struct. Engrg., ASCE, 124(10), 1099–1114. Slutter, R. G., and Driscoll, G. C. (1965). ‘‘Flexural strength of steel-
Attalla, M. R., Deierlein, G. G., and McGuire, W. (1994). ‘‘Spread of concrete composite beams.’’ J. Struct. Div., ASCE, 91, 71–99.
plasticity—Quasi-plastic hinge approach.’’ J. Struct. Engrg., ASCE, Spacone, E., Ciampi, V., and Filippou, F. C. (1996). ‘‘Mixed formulation
120(8), 2451–2473. of nonlinear beam finite element.’’ Comp. and Struct., 58(1), 71–83.
British Standards Institution (BSI). (1990). ‘‘BS5950: Section 3.1: Code Viest, I. M., Colaco, J. P., Furlong, R. W., Griffis, L. G., Leon, R. T., and
of practice for design of simple and continuous composite beams.’’ Wyllie, L. A. (1997). Composite construction design for buildings,
Structural use of steelwork in building, Part 3, London. McGraw-Hill, New York.
Chen, H. (2000). ‘‘Nonlinear inelastic analysis of steel-concrete compos- Wegmuller, A. W., and Amer, H. N. (1977). ‘‘Nonlinear response of com-
ite frames.’’ PhD thesis, Dept. of Civ. Engrg., National Univ. of Sin- posite steel-concrete bridges.’’ Comp. and Struct., 7(2), 161–169.
gapore, Singapore. White, D. W. (1985). ‘‘Material and geometric nonlinear analysis of local
Chen, W. F., and Atsuta, T. (1976). Theory of beam-column, Vol. 1, In- planar behaviour in steel frame using interactive computer graphics.’’
plane behaviour and design, McGraw-Hill, New York. MS thesis, Cornell University, Ithaca, N.Y.
Chen, W. F., and Toma, S., eds. (1994). Advanced analysis of steel Yam, L. C. P., and Chapman, J. C. (1972). ‘‘The inelastic behaviour of
frames: Theory, software, and applications, CRC, Boca Raton, Fla. continuous composite beams of steel and concrete.’’ Proc. Inst. Civ.
Daniels, B. J., and Crisinel, M. (1993). ‘‘Composite slab behavior and Engrs., 53, 487–501.

202 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:194-202.

You might also like