Structures Congress 2011 © ASCE 2011 2852

Seismic Behavior of Steel Reinforced Concrete Beam-Columns and Frames

Mark D. Denavit1, Jerome F. Hajjar2, and Roberto T. Leon3
1
Graduate Research Assistant, Department of Civil and Environmental Engineering,
University of Illinois at Urbana-Champaign, Urbana, IL 61801-2352.
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

2
Professor and Chair, Department of Civil and Environmental Engineering,
Northeastern University, Boston, MA 02115.
3
Professor, School of Civil and Environmental Engineering, Georgia Institute of
Technology, Atlanta, GA 30332.
ABSTRACT
The ability to perform accurate nonlinear simulations is a key component in the
assessment of the behavior of seismic force resisting systems. A three-dimensional
distributed plasticity formulation for composite beam-columns suitable for nonlinear
static and dynamic analyses of composite seismic force resisting systems has been
developed. New uniaxial constitutive relations are developed for the concrete and
steel elements to simulate the cyclic response of steel reinforced concrete (SRC)
members. The relations account for the salient features of each material, as well as the
interaction between the two, including for concrete: varying levels of confinement
within a section, cracking, crushing, and spalling, and for steel: cyclic plasticity and
residual stresses. The accuracy of the formulation is validated against a
comprehensive set of results from monotonically and cyclically loaded beam-column
specimens. The formulation is suitable for use in parametric studies to quantify the
seismic performance factors of special moment frames using steel reinforced concrete
columns and structural steel beams following recommendations of the recently
released FEMA P695 report.
INTRODUCTION
Steel-concrete composite columns have been shown to have high strength, stiffness,
and ductility. However, there is a lack of quantitatively justified guidance for design
of structures with these members. Notably, there is little data is available to justify the
structural system response factors (e.g., R, Cd, and Ωo) given in the specifications. In
the current work, we strive to fill these gaps through developing system response
factors; assessing beam-column strength; and establishing guidelines for the
computation of equivalent composite beam-column rigidity to be used in seismic
analysis and design of composite frames. Accurate nonlinear static and dynamic
computational formulations are required to achieve these goals. Specifically, for
developing rational system response factors, a model should directly simulate all
predominate inelastic effects from the onset of yielding through strength and stiffness
degradation causing collapse, while being sufficiently robust to track inelastic force
redistribution without convergence problems up to the point of collapse (FEMA
2009).

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2853

Often, SRC members are modeled with constitutive relations taken directly from
structural steel and reinforced concrete theory. This approach neglects the beneficial
effects that the composite section provides, namely the added confinement of the
concrete by the steel shape and the delay of local buckling in steel shape (although
local buckling is often not modeled in structural steel). Several researchers have
developed models specifically for use with SRC members (Mirza 1989, Sanz Picon
1992, Elnashai and Elghazouli 1993, El-Tawil and Deierlein 1999, Chen and Lin
2006) that do account for the beneficial effects provided by the composite section. A
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

defining feature among all of these models is the varying levels of concrete
confinement assumed throughout the cross-section. The differences between the
models arise from the use of different mechanistic assumptions and empirical
relations for the various aspects of section behavior (e.g., differences in determination
of confining pressure and confinement model). The validity the models and thus their
underlying assumptions are shown through comparisons between computation and
experimental results. The model presented in this work utilizes a sophisticated finite
element as well as comprehensive cyclic constitutive relations. Additionally,
extensive validation is performed to ensure accuracy.
THREE-DIMENSIONAL MIXED BEAM FINITE ELEMENT
Frame analyses using distributed-plasticity beam-column elements strike a favorable
balance of computational efficiency and accuracy. Additionally, mixed formulations
(defined here as treating both element displacements and stress resultants as primary
state variables) provide more accurate results with fewer elements as compared to
either displacement- or force-based formulations (Alemdar and White 2005). Tort
and Hajjar (2010) developed a three-dimensional mixed beam-column element for the
analysis of composite frames that include rectangular concrete-filled steel tube
members, validating against a large number of experimental tests of composite
members and frames. This finite element was adapted and further validated against an
additional set of experimental tests on circular concrete-filled steel tube members
(Denavit 2009). The element stiffness and internal force are derived in the
corotational frame using small strain assumptions. When accompanied with an exact
transformation between the corotational and global frame the element is capable of
capturing moderate deformation and rotation behavior. The use of independent force
interpolation functions provides for a more accurate representation of the nonlinear
curvature along the length of a yielded member. Implemented within the OpenSees
framework (OpenSees 2011), the element can be used with the wide variety of other
elements and solution algorithms available in the framework.
UNIAXIAL CYCLIC CONSTITUTIVE RELATIONS
The formulation relies on accurate constitutive relations to achieve accurate results.
The constitutive relations are defined for the finite element at the section level using a
fiber model. A fiber model allows the wide variety of behavior exhibited by SRC
sections to be described by the integration of uniaxial constitutive relations located
throughout the section.
It is desired that the fiber section be defined with minimal input from the analyst. The
required input from the analyst includes: basic dimensions (depth of the composite

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2854

section, H and width of the composite section, B), dimensions of the steel section
(depth, d, flange width, bf, flange thickness, tf, and web thickness, tw), material
properties of steel section (yield strength, Fys, ultimate strength, Fus, and elastic
modulus, Es), concrete material properties (compressive strength, f′c), dimensions and
material properties of the longitudinal and transverse reinforcement, and fiber density.
Using these parameters, the uniaxial constitutive relations and fiber section are
defined. Five distinct regions are identified within the section (Figure 1). Separate
constitutive relations are defined for each of these regions.
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

1) Highly Confined
Concrete
za 2) Medium
Confined Concrete
y
H z 3) Cover Concrete
4) Wide Flange
Steel Shape
5) Reinforcing
Steel Bars

B
Figure 1. SRC Section
The constitutive relation used for the wide flange steel shape and the reinforcing steel
bars is based on the bounding-surface plasticity model of Shen et al. (1995).
Modifications were made to model the effects of residual stress within the steel
section. The residual stress at a fiber is modeled explicitly as an initial elastic stress in
the uniaxial constitutive relation. The Lehigh residual stress pattern (Galambos and
Ketter 1958) is used to define the value of residual stress in the steel section with a
maximum compressive residual stress of 30% of the yield strength occurring at the
flange tips. The confined concrete is assumed to prevent flange and web local
buckling and thus, these effects have not been included. In contrast, the model by
Elnashai and Elghazouli (1993) is for partially encased composite columns and
includes flange local buckling since the concrete only prevents inward buckling of the
flange. For simplicity, the wide flange steel section is modeled with sharp corners
(i.e., neglecting the fillets).
The constitutive relation for the concrete is based on the rule-based model of Chang
and Mander (1994). The backbone stress-strain curve for the concrete is based on the
model by Tsai, which is defined by the initial stiffness Ec, peak coordinate (ε´cc, f´cc),
and r which acts as a shape factor. The confinement model developed by Mander et
al. (1988) for a triaxial state of stress is utilized to determine the peak compressive

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2855

strength from the confining pressure in two orthogonal directions (Equations 1

through 5).
f cc′ ⎛ 0.9 ⎞
K= = 1 + Ax ⎜ 0.1 + ⎟ (1)
f c′ ⎝ 1 + Bx ⎠

A = 6.8886 − ( 0.6069 + 17.275r ) e −4.989 r (2)

4.5
B= −5
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

(3)
5
A ( 0.9849 − 0.6306 e −3.8939 r
) − 0.1

f l1 + f l 2 f l1
x= r= f l1 ≤ f l 2 (4)
2 fc fl 2

ε cc′ = ε c (1 + 5 ( K − 1) ) (5)

where, the strain at peak stress for unconfined concrete, εc, is given by Equation 6

ε c = f c′0.25 1150.0 (6)

In order to accommodate low values of r necessary to model high confinement
pressures, two values of r are used; one prior to the peak stress reached and one after
the peak stress is reached. The value used prior to the peak is based on unconfined
concrete, while the one after the peak accounts for the level of confinement and is
based on a comparison of the models of Tsai and Popovics (Chang and Mander
1994). These and other parameters for the concrete model are as described in Table 1.
The outermost concrete region of concrete, also termed the cover concrete, is
assumed to have zero confining pressure in either direction. Furthermore, it is allowed
to spall. Spalling is modeled by modifying the nonlinear compressive backbone curve
so that the stress and tangent reduce to and remain at zero after a certain strain. The
concrete inside the lateral reinforcing bars is termed the medium confined concrete.
In this region, confining pressure is provided by the lateral reinforcing bars. The
magnitude of the confining pressure is computed in two orthogonal directions with
Equations 7 and 8.
fly ,medium = Ke ρ y Fyr (7)

flz ,medium = Ke ρz Fyr (8)

where, Ke is the ratio of effectively confined cross sectional area to area of the core as
defined by Mander et al. (1988), ρy, ρz are the volumetric ratios of the transverse
reinforce steel in either direction as defined by Mander et al. (1988), and Fyr is the
yield strength of the longitudinal reinforcement.
The concrete between the flanges is termed highly confined concrete. In this region,
confining pressure is provided by the lateral reinforcing bars and the steel shape. El-
Tawil and Deierlein (1999) developed a mechanism model in which the confining
pressure provided by the steel shape acts only in the y direction (Figure 1) and is

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2856

computed considering the plastic moment capacity of the flange (Equation 9). The
distance between the vertex of the parabola defining the boundary between the highly
and medium confined concrete (Figure 1) and the centerline of the steel section is
described by Equation 10. This parabolic boundary is modeled explicitly with
different constitutive relations on either side. In contrast, the parabolic boundary
between the cover and medium confined concrete is implicitly modeled, utilizing the
factor Ke to provide average behavior. The difference in handling these two
boundaries exists to provide greater accuracy in modeling the boundary between the
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

flanges while retaining the flexibility of different possible configurations of

reinforcing steel. A comparison of the monotonic compressive response of the
concrete in each of the three regions is shown in Figure 2.
t 2f Fys
fly ,high = fly ,medium + (9)
0.75 ( b f − tw )
2

za = 0.50b f − 0.25 ( d − 2t f ) ≥ 0.50t w (10)

The discretization of the fiber section (i.e., computation of location and area of the
fibers) is consistent with the geometric description of the cross section presented here.
The sum of the area of the fibers for each material type and the total for the section
are exact for the selected geometry and do not change with the fiber density. The
moment of inertia of the section as computed from the fibers approaches the exact
value with increasing fiber density. The discretization is implemented for general
three-dimensional analyses as well as for two-dimensional analyses where bending is
along either the strong or weak axis. The fiber discretization for two-dimensional
analyses takes the form of strips allowing a significant savings in the number of fibers
used. The fiber density is defined as the nominal number of fibers along the strong
and weak axis with the number of fibers for individual section components (e.g., steel
flanges or concrete cover) determined in proportion to their size relative to the entire
section. A discretization of 20×20 (or simply 20 for two dimensional analyses) is
found to be both accurate and efficient and is used for the analyses presented in this
paper.

1.4
Cover Concrete
1.2 Medium Confined
Normalized Stress (σ/f'c)

1 Highly Confined

0.8

0.6

0.4

0.2

0
0 0.005 0.01 0.015 0.02
Strain (mm/mm)
Figure 2. Concrete Constitutive Relations

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2857

Table 1. Concrete Material Properties

Highly Medium
Cover
Confined Confined
Initial Stiffness Ec Ec [ MPa ] = 12, 400 + 500 f c′[ MPa ]
f ly (Equation 7) (Equation 9) 0
Confinement Pressure
f lz (Equation 8) 0
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

Compression

Peak Stress f' cc (Equation 1) f' c

Strain at Peak Stress ε ' cc (Equation 5) (Equation 6)
r pre rpre = f c′[ MPa ] 5.2 − 1.9
Shape Factor
r post rpost = Ecε cc fcc′
Peak Stress f' t f t [ MPa ] = 0.5 f c′ [ MPa ]
Tension

Strain at Peak Stress ε ' t ε t = 1.23 f t Ec

Shape Factor r 4.0

VALIDATION OF THE FINITE ELEMENT METHOD

To validate the accuracy the model, a large number of comparative analyses were
performed against experimental results. A selection of the validation results for
monotonic proportionally loaded beam-columns and cyclic non-proportionally loaded
beam-columns is presented here.
Proportionally Loaded Beam-Columns
The most common experimental configuration for testing SRC beam-columns is
monotonic proportional loading (Figure 3). Tests of this type have been performed by
Virdi and Dowling (1973), Roderick and Loke (1975), Morino et al. (1984) and Wang
(1999), and others. Details of four specimens from these experimental studies are
presented in Table 2. The results of these experiments are compared to results from
analyses conducted using the model presented in this work (Figure 4). When the
eccentric of the applied load coincided with one of the principal axes of the column
(i.e., strong or weak axis bending), one mid-height lateral deflection is reported.
When the applied load induces biaxial bending, two mid-height lateral deflections are
reported, one for each of the columns principal axes.

Figure 3. Proportionally Loaded Beam-Column Test Schematic

The results (Figure 4) show a good correlation between experimental and
computational results. This is seen in the initial stiffness, peak load, deflection at peak

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2858

load, post-peak degradation, and ratio of strong and weak axis deflections. The two
specimens by Morino et al. (1984) (Figure 4c,d) have a similar cross section and
loading angle, but specimen D8-45 has approximately twice the length and
eccentricity of B4-45. The higher first- and second-order moments resulted in a
significantly lower peak axial load for D8-45. The model predicted the peak axial
load accurately for both specimens indicating that model captures well both material
and geometric nonlinearity.
Table 2. Proportionally Loaded Beam-Column Specimen Data
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

Spec- H B f' c Fy db Fyr L e

Author Year Steel Section Axis
imen (mm) (mm) (MPa) (MPa) (mm) (MPa) (mm) (mm)
Wang 1999 RSJ1 200 200 55.0 RSJ 102x102 310.0 n/a n/a 4,000 Strong 55.0
Virdi and Dowling 1973 G 254 254 44.8 UC 6x6@15.7 314.7 12.7 309 7,315 Biaxial 73.4
Morino et al. 1984 B4-45 160 160 23.4 H-100x100x6x8 287.0 4.0 387 2,309 Biaxial 40.0
Morino et al. 1984 D8-45 160 160 22.9 H-100x100x6x8 302.0 4.0 387 4,619 Biaxial 75.0

1200 700

1000 600
Axial Force (kN)

Axial Force (kN)

500
800
400
600
300
400 Experiment (Dx)
200
Analysis (Dx)
200 Experiment 100 Experiment (Dy)
Analysis Analysis (Dy)
0 0
0 10 20 30 40 0 20 40 60
Mid-Height Deflection (mm) Mid-Height Deflection (mm)
(a) Wang 1999; Specimen: RSJ1 (b) Virdi and Dowling 1973; Specimen G

400 150
Axial Force (kN)

Axial Force (kN)

300
100
200
Experiment (Dx) Experiment (Dx)
50
100 Analysis (Dx) Analysis (Dx)
Experiment (Dy) Experiment (Dy)
Analysis (Dy) Analysis (Dy)
0 0
0 20 40 60 0 50 100 150 200 250
Mid-Height Deflection (mm) Mid-Height Deflection (mm)
(c) Morino et al. 1984; Specimen B4-45 (d) Morino et al. 1984; Specimen D8-45
Figure 4. Proportionally Loaded Beam-Column Validation Results.

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2859

Cyclically Loaded Beam-Columns

A set of carefully controlled and well documented non-proportionally loaded cyclic
SRC beam-columns tests was performed by Ricles and Paboojian (1993, 1994). The
specimens were subjected to a constant axial load and cyclically increasing horizontal
displacements (Figure 5) which induced strong axis bending in the column.

P
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

Figure 5. Cyclically Loaded Beam-Column Test Schematic

Details of two of specimens from this experimental study are presented in Table 3.
The load-deformation results of these experiments are compared to those from
analyses conducted using the model presented in this work (Figure 6a,b).
Additionally, the stress-strain response from the extreme fiber of the steel section
(Figure 6c) and the extreme fiber of the highly confined concrete (Figure 6d) as
predicted from the analysis are shown.

500 500
400 400
300 300
Horizontal Force (kN)

Horizontal Force (kN)

200 200
100 100
0 0
-100 -100
-200 -200
-300 -300
-400 Experiment -400 Experiment
Analysis Analysis
-500 -500
-150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
Tip Deflection (mm) Tip Deflection (mm)
(a) Specimen 6 (b) Specimen 8
500
0

-10
Stress (MPa)

Stress (MPa)

-20
0

-30

-40
-500
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 -0.08 -0.06 -0.04 -0.02 0 0.02
Strain (mm/mm) Strain (mm/mm)
(c) Specimen 6, Steel Section Fiber Stress- (d) Specimen 6, Highly Confined Concrete
Strain Relationship Fiber Stress-Strain Relationship
Figure 6. Cyclically Loaded Beam-Column Validation Results.

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2860

Table 3. Cyclically Loaded Beam-Column Specimen Data

Spec- H B f'c Steel Fy db Fyr L P
imen (mm) (mm) (MPa) Section (MPa) (mm) (MPa) (mm) (kN)
6 406 406 35.8 W8x40 372 28.6 448 1,930 1,490
8 406 406 62.9 W8x40 372 22.2 434 1,930 1,490

The results (Figure 6) show a good correlation between experimental and

computational results. The initial stiffness and peak strength are predicted well by the
model. The unloading stiffness and yield upon unloading are less accurate, leading to
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

the model predicting fuller hysteresis loops than observed in the experiments. Similar
discrepancies between experimental and computational results have been noted in
other SRC models (e.g., Elnashai and Elghazouli 1993).
CONCLUSION
A finite element formulation for analysis of SRC members and frames has been
presented. The formulation included a mixed based distributed-plasticity beam
element and uniaxial constitutive relations. The cyclic uniaxial constitutive relations
account for the salient features of SRC behavior. The concrete constitutive relation
was based on a rule based model from the literature. Three regions of concrete were
defined with separate material properties based on computed levels of confinement
pressure. The steel constitutive relation was based on a bounding surface plasticity
model from the literature with modifications so that residual stresses could be
modeled explicitly within the steel section. A validation was performed showing good
correspondence between experimental and computational results for a selection of
both monotonic and cyclic tests. The finite element formulation provides an accurate
analysis tool for use with composite frames and is suitable for use in parametric
studies such as those necessary to develop system behavior factors.
ACKNOWLEDGMENTS
The authors thank Tiziano Perea for his advice in the completion of this work. The
work described here is part of a NEESR project supported by the National Science
Foundation under Grant No. CMMI-0619047, the American Institute of Steel
Construction, the Georgia Institute of Technology, and the University of Illinois at
Urbana-Champaign. Any opinions, findings, and conclusions expressed in this
material are those of the authors and do not necessarily reflect the views of the
National Science Foundation or other sponsors.
REFERENCES
Alemdar, B. N. and White, D. W. (2005). “Displacement, Flexibility, and Mixed
Beam-Column Finite Element Formulations for Distributed Plasticity
Analysis,” Journal of Structural Engineering, ASCE, 131(12), 1811-1819.
Chang, G. A. and Mander, J. B, (1994). “Seismic Energy Based Fatigue Damage
Analysis of Bridge Columns: Part I - Evaluation of Seismic Capacity,”
National Center for Earthquake Engineering Research, State University of
New York at Buffalo. Department of Civil Engineering, Buffalo, New York.
Chen, C.-C. and Lin, N.-J. (2006). “Analytical Model for Predicting Axial Capacity
and Behavior of Concrete Encased Steel Composite Stub Columns,” Journal
of Constructional Steel Research, 62(5), 424-433.

Structures Congress 2011

 Structures Congress 2011 © ASCE 2011 2861

Denavit, M. (2009). “Nonlinear Seismic Analysis of Circular Concrete-Filled Steel

Tube Members and Frames,” M.S. Thesis, University of Illinois at Urbana-
Champaign, Urbana, Illinois.
Elnashai, A. S. and Elghazouli, A. Y. (1993). “Performance of Composite
Steel/Concrete Members under Earthquake Loading. Part I: Analytical
model,” Earthquake Engineering & Structural Dynamics, 22(4), 315-345.
El-Tawil, S. and Deierlein, G. G. (2001). “Nonlinear Analysis of Mixed Steel-
Concrete Frames. II: Implementation and Verification,” Journal of Structural
Downloaded from [Link] by UNIVERSITE LAVAL on 10/26/15. Copyright ASCE. For personal use only; all rights reserved.

Engineering, ASCE, 127(6), 656-665.

Federal Emergency Management Agency (FEMA). (2009). Quantification of
Building Seismic Performance Factors, FEMA P695, Washington, D.C.
Galambos, T. V., and Ketter, R. L. (1958). “Columns under Combined Bending and
Thrust.” Journal of the Engineering Mechanics Division, ASCE, 85(2), 135-
152.
Mander, J. B., Priestley, M. J. N., and Park, R. (1988). “Theoretical Stress-Strain
Model for Confined Concrete,” Journal of Structural Engineering, ASCE,
114(8), 1804-1826.
Mirza, S.A. (1989). “Parametric Study of Composite Column Strength Variability,”
Journal of Constructional Steel Research, 14(2), 121-137.
Morino, S., Matsui, C., and Watanabe, H. (1984). “Strength of Biaxially Loaded SRC
Columns,” Composite and Mixed Construction, ASCE, New York, 185-194.
OpenSees (2011). “Open System for Earthquake Engineering Simulation,” Open
source software, [Link]
Ricles, R. and Paboojian S. D. (1993). “Seismic Performance of Composite Beam-
Columns,” Report No. 93-01, ATLSS Engineering Research Center, Lehigh
University, Bethlehem, Pennsylvania.
Ricles, R. and Paboojian S. D. (1994). “Seismic Performance of Steel-Encased
Composite Columns,” Journal of Structural Engineering, ASCE, 120(8),
2474-2494.
Roderick, J. W. and Loke, Y. O. (1975). “Pin-Ended Composite Columns Bent About
the Minor Axis,” Civil Engineering Transactions, 17(2), 51-58.
Sanz Picon, C. F. (1992). “Behavior of Composite Column Cross Sections under
Biaxial Bending” M.S. Thesis, Cornell University, Ithaca, New York.
Shen, C., Mamaghani, I. H. P., Mizuno, E., and Usami, T. (1995). “Cyclic Behavior
of Structural Steels. II: Theory,” Journal of Engineering Mechanics, ASCE
121(11), 1165-1172.
Tort, C. and Hajjar, J. F. (2010). “Mixed Finite-Element Modeling of Rectangular
Concrete-Filled Steel Tube Members and Frames under Static and Dynamic
Loads,” Journal of Structural Engineering, ASCE, 136(6), 654-664.
Virdi, K. S. and Dowling, P. J. (1973). “The Ultimate Strength of Composite
Columns in Biaxial Bending,” Proceedings of the Institute of Civil
Engineering (London) 55(2), 251–72.
Wang, Y. C. (1999). “Tests on Slender Composite Columns,” Journal of
Constructional Steel Research, 49(1), 25-41.

Structures Congress 2011