Mathematical and Computational Applications, Vol. 16, No. 4, pp. 947-958, 2011.

© Association for Scientific Research

O
LI
EAR FI
ITE ELEME
T A
ALYSIS OF A
R/C FRAME U
DER

LATERAL LOADI
G

Yunus Dere
Department of Civil Engineering
Selcuk University
42075 Konya, Turkey
[email protected]

Fatma Tuba Dede

Sari Engineering
Kemerli Is Merkezi A Blk K:4/408
Selcuklu, Konya, Turkey
[email protected]

Abstract- Recent developments in computer technology have made possible the use of
finite element methods for 3D modeling and analysis of reinforced concrete structures.
In this study, the failure behavior and crack formation of an R/C frame under monotonic
and reversed-cyclic lateral loading are studied by 3D nonlinear finite element analysis
using ANSYS software. Modeling the nonlinear behavior of concrete material as well as
the reinforcing steel embedded within concrete is a difficult task. Different
methodologies and modeling options are considered in the computer model. The
application of reversed cyclic displacement loading and the execution of nonlinear
analyses are explained in detail. Load-displacement relationships and concrete crack
profiles are obtained in order to compare with the experimental data and observed crack
profile. The analysis results compared well with the experimental data based on the
comparison of load-displacement graphs. The failure mode of the frame is identified by
the crack profiles displayed on the structure.
Keywords- Failure analysis, Cracks, Finite element analysis, Reinforced concrete

1. I
TRODUCTIO

In order to properly design Reinforced Concrete (R/C) structures to withstand

lateral loads resulting from earthquakes as well as high winds, the failure behavior is
usually investigated through laboratory tests. Due to the difficulties faced during the
fabrication and testing of real size models of R/C buildings and frames within a
laboratory, Finite Element (FE) analysis of the computer models of R/C structures is
quite popular among researchers. Moreover, computer simulations of planned
experiments prior to laboratory fabrication and testing processes provide significant
advantages such as reduced amount of construction materials, labor and time. However,
the nonlinear stress-strain relationship of concrete, time dependent deformations such as
creep and shrinkage, aggregate interlock, tension cracks, compression failure and the
adhesion between concrete and reinforcing steel cause difficulties in the modeling of
R/C members. There are many commercially available FE software packages capable of
948 Y. Dere and F.T. Dede

modeling nonlinear behavior of materials such as ABAQUS, ANSYS, DIANA, MSC

NASTRAN and ADINA. This work is focused on FE modeling and analysis of R/C
frames with ANSYS [1] since it is a sophisticated software with a 3D eight-node
isoparametric FE element which can be used to model concrete material and its cracking
behavior. ANSYS modeling of reinforced concrete structures has been studied by many
researchers [2-12]
In this study, the FE model of an R/C frame fabricated and experimentally tested
by Akin [13] at Selcuk University, Civil Engineering Structures Laboratory is
constructed and analyzed under monotonic and reversed-cyclic lateral loading. The
process of constructing the model, modeling and analysis parameters and options
required to be specified and the concerns related to the convergence of the analyses are
described. The failure behavior is investigated through load and displacement
relationships obtained from analyses and which are compared later with the
experimental data. Moreover, the crack profiles obtained at various stages of the
reversed-cyclic loading are significant in the identification of the failure modes of the
R/C frames.

2. MODELI
G REI
FORCED CO
CRETE MEMBERS

The Solid65 element type (Figure 1) available in ANSYS element library is a

solid element with 8 nodes and has three translation degrees of freedom at each node. It
can be used to model concrete with or without rebars and is capable of modeling
cracking in tension, crushing in compression, plastic deformation and creep behavior.
Rebars are smeared within the element and can be defined in three different
axes. Full adherence between concrete and rebar and is assumed. Solid65 can crack in
three orthogonal axes at each Gauss integration point. As a result, the maximum number
of cracks in an element is 8x3=24. These cracks are labeled as “first”, “second” and
“third” according to their order of occurrence. Analysis crack profiles can be visualized
based on their labels. Crack widths are not supplied by the element and therefore
existence of a crack does not mean that it is visible to the eye. Crack profiles with
“third” cracks are usually more consistent with the experimental ones than the ones with
“first” and “second” cracks. An integration point with a “third” crack is cracked at all
three axes; therefore the crack width is probably larger and more visible.

Figure 1. Solid65 element type as defined in ANSYS

 Nonlinear Finite Element Analysis of an R/C Frame under Lateral Loading 949

2.1. Concrete material model

The concrete material model in ANSYS uses a failure model developed by
Willam and Warnke [14] for multi-axial stress state. Solid65 element decides the
cracking and crushing of concrete through this material model. A material model may
be composed of two or more material definitions. Concrete material should have at least
Elastic and Concrete material definitions. In Elastic definition, the modulus of
elasticity and Poisson’s ratio are necessary. The modulus of elasticity of concrete can be
determined by either experiments or existing formulations. For Concrete definition,
axial tension strength of concrete and shear transfer coefficients between crack surfaces
for open and closed cracks are required. If the shear transfer from one crack surface to
the other does not exist then the shear transfer coefficient is 0.0, if it fully exists then the
coefficient is 1.0. In the literature, there are different suggestions for this coefficient by
researchers.
ANSYS does not allow for the definition of an additional material model for the
tension behavior of concrete. However, if requested/required, an additional stress-strain
relationship for compressive behavior can be defined through a hardening model such as
Multilinear Isotropic Hardening. If this is the case, the modulus of elasticity must be
the same as the slope of the initial tangent of the defined stress-strain curve.

2.2. Modeling the reinforcement

The rebars can be either smeared within the Solid65 elements or defined
discretely by spar elements (such as Link8) attached to the nodes of Solid65 elements. If
the smeared rebar model is used, then the reinforcement should be defined using the real
constants of Solid65 element.
Confinement reinforcement, in addition to the stated methods, can also be
included in the concrete material model for the confined concrete region. There is no
need to model rebars using smeared or discrete approaches, if confined concrete stress-
strain relationship given by one of the existing mathematical models is used in the
concrete model. During the modeling of confinement reinforcement, this method
requires usually less effort than discrete or smeared modeling.

3. FE MODELI
G A
D FAILURE A
ALYSIS OF A
R/C FRAME

The 3D FE model of an R/C frame designed and tested by Akin [13] at Selcuk
University, Civil Engineering Structures Laboratory is created using ANSYS. The R/C
frame specimen consisting of three stories and three bays is subjected to reversed-cyclic
loading applied at the top story level. The FE model is initially analyzed under
monotonic lateral loading. It is later subjected to reversed cyclic loading analysis.

3.1. The properties of the frame specimen

The geometry, section and reinforcement details of the R/C frame are shown in
Figure 2. Column and beam dimensions are 120x120mm and 120x150mm, respectively.
The frame is arranged symmetrically and the middle bay is constructed narrower than
the side bays. At the beam-column connections of the frame, confinement reinforcement
was not placed on purpose. The columns are reinforced with 4ϕ7 straight bars, and
beams are reinforced with 2ϕ6 straight bars, 2ϕ6 top bars to hold the stirrups and 1ϕ6
950 Y. Dere and F.T. Dede

bent bar. ϕ4/6 stirrups are used as the confinement reinforcement in both columns and
beams. The characteristic compression strength of concrete is 180kg/cm2 and the yield
strength of steel is 5200 kg/cm2.
20kN 40kN 40kN 20kN

150 P(t)

450

150

450
2250
150

650

250

200 120 880 120 680 120 880 120 200

3320
(All dimensions are in mm)
Column Section Beam Section (mid-span)
Longitudinal bars: 4φ7 Longitudinal bars: 3φ6
120 Stirrups: φ4/6 Top bars: 2φ6
150
Stirrups: φ4/6
120
120
Figure 2. The geometry, cross-sections and rebar details of the R/C frame

3.2. FE Model of the Frame

In ANSYS, the frame model created with Volume objects in consistence with the
frame’s geometry is meshed with Solid65 type finite elements. Mesh discretization is
performed in a way that the longest edge of an element should not exceed 30 mm. As a
result, the frame model is made up of 8080 elements and 12500 nodes. The FE mesh,
support restraints at the bottom of the columns and the load application points at the top
story level are shown in Figure 3. Since Solid65 element type does not have rotational
degrees of freedom, bending deformations along the columns under horizontal loading
may be computed with some errors depending on the element edge dimensions.
However, for the existing model, the chosen element edge length was verified to be
sufficient by comparing the experimental and analytical results.
In the FE model of the frame, the straight rebars in the beams and columns are
represented using smeared rebar option for the Solid65 elements located closest to the
physical rebar positions. Therefore discrete elements are not required to model the
reinforcement. The properties of the smeared rebars are given using element real
constants (Table 1). It is necessary to define a separate real constant for every
reinforcement ratio and orientation. In Figure 4, the real constant numbers at a typical
beam column connection is shown.
 Nonlinear Finite Element Analysis of an R/C Frame under Lateral Loading 951

The stress-strain relationship of the steel reinforcement is defined using Bilinear

Isotropic Hardening option (Figure 5). The yield strength and modulus of elasticity of
steel are assumed as 5,200 kg/cm2 and 2,000,000 kg/cm2, respectively. Therefore the
yielding strain is computed as 0.0026. In order to help overall convergence of the FE
analyses, the slope of the second line segment is assumed to be 1,000 kg/cm2.

Table 1. The element real constants defined for Solid65 element type.
Real Material
Constant
umber Ratio θo φo Explanation
1 - - - - No reinforcement
2 2 (Steel) 0.0428 90 0 Column reinforcement
3 2 (Steel) 0.0248 0 0 Beam reinforcement
0.0428 90
4 2 (Steel) 0 Beam-column reinforcement
0.0248 0

Figure 3. The FE mesh of the R/C frame model

Figure 4. The real constant numbers at the beam-column connection

952 Y. Dere and F.T. Dede

The stress-strain relationship of concrete is modeled through the Multilinear

Isotropic Hardening option (Figure 6). In order to obtain the stress-strain curve for both
unconfined and confined concrete regions, Modified Kent and Park concrete material
model [15] is adopted. The maximum stress of the confined concrete is computed as
216.14 kg/cm2. Unit strains corresponding to the maximum stresses for unconfined and
confined concrete are assumed to be 0.002 and 0.00235, respectively. When creating the
Multilinear Isotropic Hardening models, the stress-strain curves are assumed to be
composed of straight line segments and the modulus of elasticity is calculated as the
slope of the initial line segment. The modulus of elasticity for unconfined and confined
concrete is computed as 160,540 kg/cm2 and 164,200 kg/cm2, respectively.

Figure 5. The ε-σ relationship of Bilinear Isotropic Hardening Model for the steel
(SIG: kg/cm2)

Figure 6. The ε-σ relationship of Multilinear Isotropic Hardening Model for: (a)
Unconfined concrete (b) Confined concrete (SIG: kg/cm2)

Since the beam-column connections did not have any stirrups, unconfined
concrete material model is defined in these regions (Figure 6a). Confinement
reinforcement is modeled through the confined concrete material model (Figure 6b) for
the Solid65 elements located at the confined region rather than creating separate spar
elements for stirrups.
 Nonlinear Finite Element Analysis of an R/C Frame under Lateral Loading 953

The tensile strength of concrete is assumed to be 10% of its compressive

strength. The Poisson’s ratios for concrete and steel are taken as 0.2 and 0.3,
respectively. The control for concrete crushing in compression is prevented by inputting
-1 for the concrete compressive strength in Solid65 options. Activating concrete
crushing control and using hardening models (Figure 6) concurrently is known to cause
analysis and convergence errors and therefore avoided.

3.3. Analysis under lateral monotonic displacement loading

In order to investigate the monotonic behavior and crack profiles of the frame,
an FE analysis is performed and the results are partially compared with the experimental
data obtained under reversed cyclic loading of the frame in the laboratory. The loading
used for the FE analysis is applied at the top corner of the frame as in the case of
experimental loading. A displacement controlled loading is preferred and its magnitude
is incrementally increased. The loading is continued until the termination of the analysis
due to convergence errors. Convergence using nonlinear element types such as Solid65
is so sensitive to the analysis and element options. The size of the load increment is
initially set to an optimum value and then, let ANSYS change automatically according
to the number of nonlinear iterations performed at each load step. The shear transfer
coefficients for the cracks should be assumed considering both the actual behavior of
the crack interface and the convergence of the analysis. In order to see the effect of
different assumptions for these coefficients on the load-displacement relationship, two
analyses are performed: Analysis (a) and Analysis (b). The coefficients for open and
closed cracks are assumed to be 0.2 and 0.8 in Analysis (a); 0.5 and 1.0 in Analysis (b),
respectively. Analysis (a) uses a more realistic assumption whereas Analysis (b) uses a
more convergent assumption.
In Figure 7, the relationship between the applied forces and the measured
displacements obtained from the experiment and FE analyses is shown. As expected, in
Analysis (b), the frame has demonstrated a more rigid behavior. Analysis (a) with more
realistic assumption has stopped earlier producing results much closer to the
experimental ones.
As it is seen, both analyses have stopped at around the ultimate load capacity of
the frame. Due to excessive cracking, significant convergence problems are experienced
in the solution of nonlinear problems using ANSYS Solid65 element type. Several
attempts have been made to overcome these convergence problems through adjustment
of load step sizes and loosening the convergence tolerances. However it was impossible
to get the expected softening regime after the peak strength.
In Figure 8, the crack profile with “third” cracks obtained from the Analysis (a)
at the maximum displacement (40 mm) is shown. It is observed that the cracks are
concentrated at beam-column connections and distributed in accordance with the
direction of the applied load.
954 Y. Dere and F.T. Dede

Figure 7. Comparison of analyses results under monotonic loading with the

experimental data

Figure 8. The crack profile with “third” cracks obtained from ‘Analysis (a)’ at
maximum displacement
 Nonlinear Finite Element Analysis of an R/C Frame under Lateral Loading 955

3.4. Analysis under lateral reversed cyclic loading

The reversed cyclic displacement loading shown in Figure 9 is applied at the
nodes of the elements located at the top corners of the frame model in 9 load steps
(Figure 10). The load step size is chosen to be 10 mm. For Solid65 concrete elements, to
ease the convergence of the analysis, crack shear transfer coefficients are assumed to be
0.5 and 1.0, respectively. The loading process is terminated due to non-convergence at
the 9th load step with a displacement value of 29.3 mm. A total of 3889 nonlinear
analysis steps are executed.

Figure 9. Reversed-cyclic displacement loading (∆u=10 mm)

In Figure 11, the experimentally obtained reversed cyclic loading vs.

displacement data are compared with the analysis results. In Figure 11(a), to allow a
better comparison by displaying more details, the graph is bounded by the ultimate
displacement and force values obtained by the analysis. As it is seen, the analysis results
are in good agreement with the experimental data. However, it is not possible to make a
complete comparison for the loading-unloading curves due to the convergence problem.
Similar convergency problems were experienced by Miller et al. [6].

10 mm 0 mm 10 mm 0 mm 20 mm

LS1 LS2 LS3 LS4 LS5

0 mm 20 mm 0 mm 30 mm

LS6 LS7 LS8 LS9

Figure 10. The displacement load steps applied at the RC frame

In Figure 12, the crack profiles with “third” cracks for different load steps are
shown. In Figure 13, the experimental cracks at the maximum displacement are shown
schematically. As it is seen, the number of cracks on the model in Figure 12(d) is much
larger than that on the test specimen. It should be noted that cracks are displayed on the
model regardless of their widths; even they are not visible to the eye. In the initial load
steps, the cracks are concentrated only at the beam-column connections. As the
956 Y. Dere and F.T. Dede

magnitude of the load increases, the number of cracks also increases and the cracks
propagate away from the connections.

Figure 11. Comparison of analysis results with the experimental data (a) Focused view
(b) General view

Figure 12. Calculated crack profiles at various load steps

 Nonlinear Finite Element Analysis of an R/C Frame under Lateral Loading 957

Figure 13. Observed crack profile at the maximum displacement by Akin [13]

4. CO
CLUSIO
S

In this work, the failure analyses of R/C structural frames under monotonic and
reversed cyclic loading are carried out by using ANSYS software. The 3D FE model of
an R/C frame specimen tested by Akin [13] at Selcuk University, Civil Engineering
Structures laboratory is constructed by the use of Solid65 element type which can
simulate three axial cracking of concrete. The reinforcing steel can be modeled either by
using discrete spar elements or by the smeared rebar option within Solid65. Smeared
reinforcement option is preferred since it is found to be easy and reliable. The effect of
various crack shear transfer coefficient assumptions for open and closed cracks is
discussed. The steps of applying reversed cyclic loading onto the frame are presented.
Stress-strain relationship of concrete material obtained from the confined and
unconfined Kent & Park mathematical models produced quite reliable load-
displacement relationship and crack profiles.
The load-displacement figures obtained from ANSYS analyses and the
experimental data are compared and found to be in good agreement with each other;
however, complete load-displacement relations from the analyses could not be obtained
due to convergence problems near ultimate loading capacity of the frame.
The crack profiles obtained from the ANSYS model are plotted for different
loading steps and compared with the experimental crack schematic figure. Although the
number of cracks on the finite element model is quite larger, the crack locations
compare well with the experimental ones. Unfortunately, it is not possible to define and
hide the cracks invisible to the eye in ANSYS. For researchers doing experimental
studies, it is proposed to carry out computer simulations of their planned experiments
prior to laboratory fabrication and testing processes. This might provide significant
advantages such as lesser amount of construction materials, labor and time.

5. REFERE
CES

1. SAS, ANSYS 7.1 Finite element analysis system, Theoretical manual, Swanson
Analysis Systems IP, Inc., 2003.
958 Y. Dere and F.T. Dede

2. H.O. Köksal and B. Doran, Finite element applications using non-linear octahedral
elastic and plastic relations in concrete and reinforced concrete members, Technical J.
of Turkish Chamber of Civil Eng. 8(Digest), 437-441, 1997.
3. A.F. Barbosa and G.O. Ribeiro, Analysis of reinforced concrete structures using
ANSYS nonlinear concrete model, In: Idelsohn S, Onate E, Dvorkin E (Eds)
Computational Mechanics: New Trends and Applications, Barcelona, Spain, 1998.
4. D. Kachlakev and T. Miller, Finite element modeling of reinforced concrete
structures strengthened with FRP laminates, Technical Report, No. SPR 316, Corvallis,
Oregon, USA, 2001.
5. P. Fanning, Nonlinear models of reinforced and post-tensioned concrete beams,
Electronic Journal of Structural Engineering 2, 111-119, 2001.
6. C.A. Miller, C.H. Hofmayer, Y. Wang, N. Chokshi, A.J. Murphy and Y. Kitada,
Prediction of NUPEC's multi-axis loading tests of concrete shear walls, 16th
International Conference on Structural Mechanics in Reactor Technology, SMiRT 16,
12-17 August, Washington, DC, USA, 2001.
7. Y. Calayir and M. Karaton, Non-linear dynamic analysis of arch dams using Drucker-
Prager approach, Technical J. of Turkish Chamber of Civil Eng. 15, 3085-3103, 2004.
8. E. Erduran and A. Yakut, Drift based damage functions for reinforced concrete
columns, Computers and Structures 82, 121-130, 2004.
9. A. Rahman, R. Hawileh and H. Tabatabai, FE model of precast hybrid beam column
connection subjected to cyclic loads, 16th Engineering Mechanics Conference, ASCE,
16-18 July, University of Washington, Seattle, USA, 2003.
10. F.E. Farvashany, S.J. Foster and B.V. Rangan, Computer simulation of tests on
reinforced concrete shear walls, 8th International Conference on Damage and Fracture
Mechanics: Computer Aided Assessment and Control, Crete, Greece, 2004.
11. F.T. Dede, Nonlinear finite element analysis of reinforced concrete frames subjected
to reversed-cyclic loading using ANSYS software, M.Sc. Thesis (in Turkish), Selcuk
University Graduate School of Nat. and Appl. Sci., Konya, Turkey, 2006.
12. L. Dahmani, A. Khennane and S. Kaci, Crack identification in reinforced concrete
beams using ANSYS software, Strength of Materials 42(2), 232-240, 2010.
13. S.K. Akin, Rehabilitation of RC frames by tension only bracing, Ph.D. Thesis (in
Turkish), Selcuk University Graduate School of Nat. and Appl. Sci., Konya, Turkey,
2006.
14. K.J. Willam and E.D. Warnke, Constitutive model for the triaxial behavior of
concrete, In: Proceedings of International Association for Bridge and Structural
Engineering: Seminar on Concrete Structures Subjected to Triaxial Stress, ISMES,
Bergamo, Italy, p. 174-186, 1975.
15. R. Park, M.J.N. Priestley and W.D. Gill, Ductility of square-confined concrete
columns, ASCE J Struct Div. 108(4), 929-950, 1982.