Finite Element Analysis of Damage and Failure of Reinforced Concrete Members Under Earthquake Loading

Received: 12 December 2016 Revised: 14 April 2017 Accepted: 22 May 2017
DOI: 10.1002/eqe.2932
RESEARCH ARTICLE
Finite element analysis of damage and failure of reinforced

concrete members under earthquake loading
Mohammadreza Moharrami1 | Ioannis Koutromanos2
1
Staff II‐Structural Engineer, Simpson
Summary
Gumpertz & Heger, Inc., Waltham, MA
2
Department of Civil and Environmental
This paper presents a three‐dimensional analysis framework, based on the explicit
Engineering, Virginia Polytechnic Institute finite element method, for the simulation of reinforced concrete components under
and State University, Blacksburg, VA cyclic static and dynamic loading. A recently developed triaxial constitutive model
Correspondence for concrete is combined with a material model for reinforcing steel which can
Ioannis Koutromanos, Department of Civil account for rupture due to low‐cycle fatigue. The reinforcing bars are represented
and Environmental Engineering, Virginia
with geometrically nonlinear beam elements to account for buckling of the rein-
Polytechnic Institute and State University,
103 Patton Hall, 750 Drillfield Drive, forcement. The strain penetration effect is also accounted for in the models. The
Blacksburg, VA 24061. modeling scheme is used in a commercial finite element program and validated with
Email: ikoutrom@vt.edu
the results of experimental static and dynamic tests on reinforced concrete columns
and walls. The analyses are supplemented with a parametric study to investigate the
impact of several modeling assumptions on the obtained results.
K EY WO R D S
bar buckling, bar rupture, explicit method, finite elements, reinforced concrete, three‐dimensional models
1 | INTRODUCTION
Reinforced concrete (RC) structures are widely used in earthquake‐prone areas. The lateral force‐resisting system in such struc-
tures consists of moment frames, structural walls, or a combination thereof. Modern design codes such as the current ACI 318
document1 include detailing requirements for RC members in seismically active regions, to ensure that satisfactory performance
will be attained through the inelastic deformability of the structure. Strong ground motions due to rare, extreme events can lead
to opening of large cracks, cover spalling, rebar yielding, and crushing of the concrete. Inelastic deformations can lead to buck-
ling and subsequent rupture of the longitudinal reinforcement due to low‐cycle fatigue. The rupture of the longitudinal rein-
forcement is associated with significant degradation in the resistance of a member and may lead to full or partial collapse of
a structural system.
Experimental research is the most reliable means to investigate the damage patterns and evaluate the performance of RC
structures. However, experimental testing is limited to either small‐scale tests or tests of single structural components. Thus,
the reliable determination of the seismic response of RC structures critically hinges on the availability of computational simu-
lation tools, which can capture the inelastic behavior of concrete and reinforcing steel and its effect on the structural response.
An analytical model should be capable of accounting for all the salient mechanisms associated with inelastic response, such as
concrete cracking and crushing and rebar buckling and subsequent rupture.
Analytical models of RC structures under extreme loading events typically rely on uniaxial stress‐strain laws, which may
include limit strain definitions to account for rebar buckling and rupture and for concrete crushing.2-5 Such uniaxial laws can
be combined with beam formulations having a fiber sectional model. Although beam‐based models are numerically efficient
and allow systematic parametric investigations of structural systems, they may lack the capability to accurately capture several
aspects of material behavior, such as the inelastic buckling and subsequent rupture of longitudinal bars. Additionally, recent studies
Earthquake Engng Struct Dyn. 2017;1–19. wileyonlinelibrary.com/journal/eqe Copyright © 2017 John Wiley & Sons, Ltd. 1
2 MOHARRAMI AND KOUTROMANOS
(eg, Lu and Panagiotou6) have concluded that beam models may not be capable of accurately describing the response of RC walls
with a nonrectangular cross section. Given the above issues, continuum‐based finite element (FE) models are necessary to provide
the response and damage patterns of RC structures, supplementing large‐scale experimental investigations and allowing the
calibration and refinement of simpler, beam‐based models. Such FE models require constitutive laws and kinematic
formulations which can capture the interaction of the various stress and strain components and the effect of geometric nonlinearity.
Previous studies have used two‐dimensional FE models for the simulation of members subjected to monotonic7-9 and cyclic
loading.10-15 Three‐dimensional FE models have also been used to investigate the behavior of RC structures under impact16,17
and monotonic loading.18,19 Different structural components have been considered, including reinforced or fiber‐reinforced
concrete columns,18,20 frames,21,22 precast beam‐to‐column connections,23 slabs,24 post‐tensioned bridge girders,25 composite
bridge decks,26,27 and hybrid moment‐resisting frame systems.28 These studies have validated the capability of FE models to
capture the monotonic force‐displacement response of RC components.
A limited number of studies have used three‐dimensional FE models for the analysis of shear walls29,30 and beam‐to‐column
connections31-33 under cyclic loading. Faria et al29 have used a simplified version of the material model described by Faria
et al34 to investigate the seismic behavior of an arch dam and to reproduce the hysteretic response of a small‐scale, six‐story,
RC shear‐wall subjected to base excitation. The analytical results satisfactorily matched the experimental observations.
Eligehausen et al31 used the microplane model by Ožbolt et al35 to simulate an RC beam‐to‐column joint subjected to
quasi‐static monotonic and cyclic loading. The reinforcement was modeled with truss elements having a simplified, trilinear,
stress‐strain curve. The bond‐slip response of the reinforcement was explicitly accounted for using appropriate elements. The anal-
yses showed that the continuum model can potentially capture shear failure associated with the formation of a large diagonal crack;
however, the analytically obtained strength and stiffness degradation were more severe than those observed in the experimental
tests. A similar modeling approach for the analysis of beam‐to‐column connections was also used by Deaton.33 The concrete
was modeled using the triaxial material law by Selby and Vecchio,36 and the behavior of the steel was described with the uniaxial
law by Filippou et al.37 The model was validated using the results of an experimental test on a beam‐to‐column connection.
Although the aforementioned studies have used three‐dimensional FE analysis for simulation of structural components
under cyclic loading, there are important aspects of behavior, such as rebar buckling and rupture, which have not been
accounted for. Additionally, many studies have identified and described limitations of existing triaxial models for concrete.3-
41
For example, the continuum models used by Faria et al29 and by Deaton33 cannot account for the development of irreversible
(plastic) compressive strains in the concrete. There have also been cases (eg, Hawileh et al32) where important features of mate-
rial response, such as compressive strength degradation due to concrete crushing, have been neglected in the models, to elim-
inate the possibility for convergence difficulties in the analysis.
The above considerations lead to the realization that there is a need for reliable, three‐dimensional, FE models to capture the
damage and failure in RC structures under earthquake loading. The present study is aimed to address this need through the
formulation and validation of an FE modeling scheme. The scheme relies on two recently proposed constitutive models, namely,
a triaxial material model for concrete under cyclic loading41 and a uniaxial material model for steel.42 The proposed method-
ology can capture all the aspects of concrete behavior, such as compressive crushing, strength and stiffness degradation due
to cracking, and the effect of confinement on the material strength and ductility. Furthermore, the modeling technique explicitly
accounts for rebar buckling—by means of geometrically nonlinear beam elements—and rebar rupture due to low‐cycle fatigue.
The material models have been implemented in a commercial FE program and used for the simulation of RC walls under cyclic
loading and a bridge pier subjected to a sequence of ground motions. The solution of the global equations relies on an explicit
dynamic integration scheme. Parametric studies are also conducted to investigate the effect of viscous damping coefficient,
global integration scheme, and accounting for strain penetration.
2 | DE SCRIPT I ON OF M ODE L I NG S C H E M E
This section describes the element types and constitutive laws used in the analysis of RC structures subjected to cyclic loading.
A detailed description of the concrete and steel material models is provided by Moharrami and Koutromanos41 and Kim and
Koutromanos,42 respectively. The salient features of these two models are presented here for completeness.
2.1 | Element formulation and constitutive model for concrete

In the proposed analysis scheme, the concrete is modeled using three‐dimensional, hexahedral, eight‐node solid elements with
uniform reduced integration (URI), i.e. with a single quadrature point. Although URI enhances the efficiency of FE
MOHARRAMI AND KOUTROMANOS 3
formulations by substantially reducing the computational cost of the stress update at the quadrature points of the mesh, it also
entails the presence of spurious zero‐energy modes (hourglass modes), for which URI‐based elements cannot develop resis-
tance.43 To prevent such modes from polluting the solution, hourglass control methods must be used to allow the development
of hourglass resistance.
The complete procedure for hourglass control, including the determination of the hourglass vectors and of the correspond-
ing resistance forces, is described by Flanagan and Belytschko.44 This procedure typically provides the hourglass forces, which
would correctly reproduce the elastic resistance of an element for bending modes of deformation. For analyses involving inelas-
tic material behavior, these forces are typically multiplied by a reduction factor, which is much lower than 1, to eliminate the
possibility for excessive resistance.
Two different approaches can be established for hourglass control, depending on the nodal response quantity used to
determine the hourglass component of the motion. Using the nodal displacements to obtain the hourglass part of the motion
leads to stiffness‐based hourglass control, and using the nodal velocities leads to viscous hourglass control. The analyses pre-
sented herein primarily rely on stiffness‐based hourglass control. The hourglass coefficient is set equal to 1 for elements with
elastic material (in regions which do not incur damage), and 0.05 for elements with nonlinear material behavior. For cases
involving quasi‐static cyclic loading, the combination of a stiffness hourglass control and an explicit solution scheme was
found to occasionally result to slightly oscillatory results due to undesired excitation of dynamic response. This issue was
found to be resolved by adopting viscous hourglass control for the portion of the cross‐section enclosed by transverse
reinforcement. The viscous hourglass control uses the formulation described by Belytschko and Bindeman45 and an hourglass
coefficient of 0.15.
The constitutive model used for the hexahedral elements to describe the behavior of concrete is the one described by
Moharrami and Koutromanos,41 which combines an elastoplastic law and a rotating smeared‐crack formulation. The specific
material model accounts for the stiffness and strength degradation due to cracking and crushing, and for the increase in strength
and ductility of the material under confined compression. The strain vector is decomposed into three parts, namely, an elastic
part, a plastic part, and a cracking part. The elastoplastic law relies on a yield surface in the principal stress space, fb
σ g, which is
described by the following equation:
1 h pffiffiffiffiffiffiffi
f ðfb
σ g; κÞ ¼ α⋅Ι1 þ rðθ; eÞ 3J 2 − cc ðκÞ ¼ 0 (1)
1−α
where I1 and J2 are the first invariant and the second deviatoric invariant of the stress tensor, cc is a strength parameter that can
be shown to be equal to the uniaxial compressive strength, κ is a hardening variable expressing the cumulative effect of inelastic
deformation, α is a dimensionless material parameter, and r(θ,e) is a quantity describing the effect of the third deviatoric invari-
ant on the yield criterion, as explained by Kang et al.46 The shape of the yield surface in the deviatoric plane, as compared with
that of the well‐known Drucker‐Prager criterion, is shown in Figure 1a. The proposed yield surface is inscribed to the Drucker‐
Prager one. For increasing values of confining pressure, the projection of the yield surface on the deviatoric plane gets closer to
that of the Drucker‐Prager surface.
The evolution of the plastic strain is described by a plastic potential function, g, given by the following equation:
pffiffiffiffiffiffiffi
g ¼ αp ⋅Ι 1 þ 2J 2 (2)
Parameter αp in Equation 2 controls the material dilatation for compressive inelastic states.
The strength of the material changes with the accumulation of inelastic strains in accordance with the following hardening‐
softening law47:
fo h pffiffiffiffiffiffiffiffiffi i
cc ðκ Þ ¼ ð1 þ aÞ φðκ Þ−φðκ Þ ≥ f res (3)
a
where φ(κ) = 1 + a(2 + a)κ, fres is the residual compressive strength in the material, and fo and a are material constants. The
variation of the strength parameter cc with κ is schematically presented in Figure 1b. The maximum value of cc = fc is
obtained for κ = κο. The values of κο and parameter a in Equation 3 can be found if the strain εο at the peak compressive
strength (for uniaxial stress states) and the area gc under the plot of the hardening‐softening law are given. To eliminate the
possibility for spurious mesh size effects due to localization,41 the parameter gc is set equal to the ratio Gc/h, where Gc is a
compressive fracture energy quantity and h is the element size. The evolution of κ is governed by the following rate
equation:

cc ∂g p
κ_ ¼ ð1−r Þ: ⋅ ⋅ed:ð1þX Þ f c (4)
gc ∂b
σ bσ min

pffiffiffiffiffiffiffi ∂g
where p = − I1/3 is the pressure in the material, X ¼ I 1 = 3J 2 , r is a weight factor, and is the component of the
σ bσ min
∂b
plastic strain rate in the direction of the minimum principal stress. Parameter d in Equation 4 controls the effect of the con-
fining pressure on the evolution of the hardening variable. For confined compressive states, the value of d will lead to a
more ductile behavior as explained by Moharrami and Koutromanos.41
A rotating smeared crack model is used to capture the damage associated with tensile cracking. The model allows up to three
cracks to be simultaneously open. The principal stress of the cracked material is a function of the corresponding principal strain
in accordance with the following equation:
" #
−λt
bεi −bεini
σ i ¼ ct ⋅ ð1−M Þe
b ft þ M ; i ¼ 1; 2; 3 (5)
where M is the ratio of the residual tensile strength over the tensile strength, bεini is the strain at the onset of softening, and λt is a
parameter controlling the rate of tensile softening. The value of the material parameter ct is obtained by the following expression:
ct ¼ f t ; if κ ≤ κο
cc (6)
ct ¼ f t ; if κ > κο
fc
where ft is the uniaxial tensile strength. The value of parameter λt in Equation 5 is obtained by stipulating that the area under the
softening portion of the cracking stress‐strain law is equal to the ratio Gt/h, where Gt is the mode‐I (tensile) fracture energy of the
material and h is the element size. Specifically, the following equation applies for λt:
ð1−M Þf t 2
λt ¼ h (7)
Gt
The softening law of the cracked material and the unloading‐reloading behavior in the cracked regime are schematically pre-
sented in Figure 1c.
2.2 | Element formulation and constitutive model for reinforcing steel

The reinforcing steel bars are modeled with beam elements that use the formulation by Hughes and Liu.48 The specific formu-
lation, which is based on the concept of continuum‐based structural elements (eg, Belytschko et al43), was originally created for
shell elements 49 and accounts for large rotations and large deformations. The sectional stress resultants (axial forces, shear
forces, bending moments, and torsional moments) are obtained through the integration of the stresses over the cross section
of the beam. The nodal forces and moments are then obtained from the integration of the sectional stress resultants along the
length. Although such beam elements are computationally more demanding compared to truss elements, they can naturally
FIGURE 1 Yield surface and hardening‐softening law of concrete. A, Yield surface in deviatoric plane. B, Hardening‐softening law. C, crack
model [Colour figure can be viewed at wileyonlinelibrary.com]
capture the effect of rebar buckling.50 The beam elements used in the present study include a single quadrature location along
the length, and a total of nine quadrature points are used for the integration of stresses over the cross section. The beam elements
have common nodes with the corresponding continuum elements which represent the concrete material.
The reinforcing steel material is described by the uniaxial constitutive law of Kim and Koutromanos,42 which is essentially
an enhanced version of the material model by Dodd and Restrepo.51 The enhancements include the elimination of the need for
iteration in the stress update algorithm and the capability to account for the material failure (rupture) due to low‐cycle fatigue.
The material model can be fully calibrated if the stress‐strain curve in monotonic tension, schematically shown in Figure 2a, is
known. The constitutive model includes a hysteretic law to capture the cyclic response of reinforcing steel, as qualitatively
shown in Figure 2b.
To account for the rebar rupture, a criterion based on the accumulation of a continuous quantity D is adopted.42 52 Specifically,
given the true material stress, f, and the plastic strain rate, ε_ p , the evolution of D is governed by the following rate equation:
8 !2t
>
< f
ε_ p ; if f >0
D_ ¼ fy (8)
>
:
0 otherwise
where t is a material constant. Rupture, i.e. material failure, occurs when the value of D becomes equal to Dcr, which is also a
parameter of the material model.
2.3 | Accounting for the strain penetration effect

The analysis methodology also accounts for the strain penetration effect, i.e. the deformations associated with rebar bond‐slip at
the base of the members considered. Specifically, the beam elements and continuum elements in this region use separate sets of
nodes. One‐dimensional contact elements are used to connect the nodes of the beam elements representing the rebars with the
nodes of the surrounding solid elements representing the concrete. As shown in Figure 3a, the contact elements, which are used
over the entire embedment length of the bars in the base, are essentially springs aligned with the axis of the beam elements. The
coupling forces of these springs are a function of the slip of the beam elements, i.e. the relative displacement of these elements
with respect to the surrounding concrete. The displacements of the beam elements in directions other than the axial one are
constrained to be equal to those of the surrounding solid elements. The relation between bond forces and slip is assumed to
be elastoplastic, with a softening post‐yield branch to account for local bond strength degradation, as depicted in Figure 3b.
The curve describing the bond‐slip law uses the peak bond strength, τmax, and the corresponding slip value, Speak, obtained from
the bond‐slip model of Murcia‐Delso and Shing,53 to provide an approximation of the bond‐slip curve obtained from the spe-
cific model. This approximation can be deemed adequately accurate, provided that the inelastic response in an analysis is not
dominated by bond failure and pullout of reinforcing bars.
3 | VA L I DAT I O N O F A NA LYSI S M E T H O D O L O GY
The proposed modeling methodology is implemented in the commercial FE program LS‐DYNA 54 to allow the simulation of
structural components and systems under cyclic loading. An explicit transient integration algorithm is used for the solution
of the global equations in the models. Specifically, the global response is formulated as a dynamic problem, and a central‐
FIGURE 2 Behavior of reinforcing steel

model (figures from42). A, Monotonic tensile
response. B, Cyclic hysteretic response
[Colour figure can be viewed at
wileyonlinelibrary.com]
FIGURE 3 Methodology to account for strain penetration effect. A, Element types and connectivity. B, Schematic comparison between
approximate bond‐slip law adopted in the present study and Murcia‐Delso and Shing53 [Colour figure can be viewed at wileyonlinelibrary.com]
difference scheme is used for the integration of the equations of motion in time. The analyses account for geometric
nonlinearities by means of a large deformation, updated Lagrangian formulation.43 The equations of motion at each step
in an analysis are solved on the updated geometry of the mesh. This allows to capture the P‐Delta effect at the structural
component level and the local effect of geometric nonlinearity, such as rebar buckling.
The modeling approach is validated through the simulation of two post‐tensioned walls subjected to cyclic loading,55 a
U‐shaped wall under bidirectional cyclic loading,56 and a bridge pier subjected to a sequence of ten ground motions.57 Ray-
leigh damping is used in the analyses, with a prescribed damping ratio of 1% for the walls and 0.5% for the bridge pier, for the
first and fourth elastic natural frequency of the models. Since all analyses use an explicit dynamic time‐marching scheme,
special care is taken to simulate the cases involving quasi‐static loading. Specifically, the cyclic loads in these cases are applied
at a sufficiently slow rate, to ensure that the analytical solution is not affected by dynamic effects.
The material parameters of the constitutive model for the concrete are calibrated in a consistent fashion. The uniaxial com-
pressive strength, fc, is obtained from material tests that accompanied the experiments of the structural components. In all
analytical models, the Poisson0 s ratio, v, is set equal to 0.2. The values of parameters fo and fres in Equation 3 are taken equal
to 2/3fc and 0.05fc, respectively, whereas the value of the tensile strength, ft, is set equal to 10% of fc. The values of parameters
εo, ap, and a are set equal to 0.0025, 0.15, and 0.375, respectively. The modulus of elasticity of the concrete is assigned a value
of 2fc/εο. The parameter M in Equation 5 is calibrated in accordance with Moharrami et al,58 and the value of λt is found using
the equation by Bažant59 for the mode‐I fracture energy, Gt. The compressive fracture energy, Gc, is set equal to 100Gt, based on
recommendations by Feenstra.60 Finally, the parameter d, controlling the effect of confinement on material ductility in accor-
dance with Equation 4, is assumed to be equal to the ratio of fc over 6.89 MPa (1 ksi). This value of d has been found by
Moharrami and Koutromanos41 to give satisfactory results at the material and component level. The values of several material
parameters for the concrete model are presented in Table 1. The material failure of concrete associated with cover spalling is
accounted for through element removal. Specifically, solid elements are removed from the analysis whenever the hardening
variable, κ, becomes equal to 1.
The material parameters of the uniaxial stress‐strain model for reinforcing steel have been assigned the values presented in
Table 2, based on the results of monotonic rebar tensile tests that accompanied each of the component tests. The value of the
Poisson ratio, v, which is also required in the continuum‐based beam elements, is set equal to 0.3 for all the analyses.
TABLE 1 Concrete material model parameters
Specimen fc (MPa) d Gt (N/m) Gc (N/m) M
Pakiding et al,55 Wall 1 43.4 6.3 88.9 8890 0.053

55
Pakiding et al, Wall 2 47.6 6.9 92.7 9270 0.044
56
Beyer et al 77.9 11.0 108.9 10890 0.025
Schoettler et al57 41.3 6.0 83.6 8360 0.036
Note: fc is the concrete compressive strength, εo is the strain at peak compressive strength, ap is the dilatancy parameter, a is a parameter affecting the increase in strength
due to confinement, d is a parameter affecting the increase in material ductility due to confinement, Gt is mode−1 fracture energy, Gc is compressive fracture energy, M is
the ratio of residual tensile strength over tensile strength.
TABLE 2 Steel material parameters
Specimen Bar type Es (MPa) fy (MPa) εsh εsh1 fsh1 (MPa) εu fu (MPa)
Pakiding et al,55 Wall 1 #3 200000 473 0.010 0.04 621 0.100 742
#4 200000 441 0.010 0.04 531 0.130 683
#7 200000 519 0.010 0.04 621 0.160 744
Pakiding et al,55 Wall 2 #3 200000 474 0.010 0.04 621 0.100 746
#4 200000 441 0.010 0.04 531 0.130 683
#5 200000 436 0.010 0.04 517 0.140 605
Beyer et al56 D6 200000 518 0.004 0.04 650 0.084 681
D12 200000 488 0.025 0.06 550 0.126 595
Schoettler et al57 #11 195990 519 0.011 0.04 607 0.110 706
#5 200000 378 0.002 0.04 517 0.125 592
Note: Es is Young0 s modulus, fy is the rebar yield stress, εsh is the strain at onset of strain hardening, εsh1 and fsh1 are the strain and stress of an intermediate point on
hardening regime of the monotonic curve, fu is the ultimate strength, and εu is the strain corresponding to the ultimate strength.
Calibration is also necessary for two parameters, t and Dcr, controlling the low‐cycle fatigue criterion. The value of param-
eter t in Equation 8 is taken equal to 1. The value of parameter Dcr (i.e. the threshold value of the continuous quantity D that
corresponds to material failure due to low‐cycle fatigue) is set equal to 2:4DcrðmonÞ , where DcrðmonÞ is the value of Dcr required to
capture the instant of rupture for monotonic tension. The value of Dcr used for the various types of rebars in the analytical
models is presented in Table 3. Longitudinal rebar rupture due to low‐cycle fatigue is accounted for through element removal.
A beam element is removed when the material failure criterion, D = Dcr, is satisfied for all the sectional integration points of
that element.
3.1 | Analysis of post‐tensioned RC walls under quasi‐static loading

The first set of analyses is conducted for two post‐tensioned RC wall specimens tested by Pakiding et al.55 In the specific exper-
imental study, the two specimens were termed Wall 1 and Wall 2, respectively. The configuration and reinforcing details of the
wall specimens are depicted in Figures 4a and 4b. Wall 1 represented a shear wall detailed in accordance with modern design
standards, whereas Wall 2 included a higher amount of prestressing tendons to ensure a self‐centering (rocking) capability. Two
prestressing tendon bundles, each consisting of five strands, were used to provide a prestressing force of 1561 kN (351 kips) for
Wall 1, and three bundles, one with five strands and two with seven strands each, were used to provide a prestressing force of
2907 kN (653.5 kips) for Wall 2. The cross‐sectional dimensions were identical for the two specimens. The number and size of
the reinforcing steel bars was the same in both specimens, except for the vertical reinforcement in the boundary regions of the
sections, where Wall 1 included eight # 7 bars with a diameter of 22 mm and Wall 2 included eight # 5 bars having a diameter of
16 mm. The walls included additional vertical # 3 bars, with a diameter of 10 mm, in the web and boundary regions of the sec-
tion, as shown in Figure 4a. The same figure shows the size and spacing of the transverse (horizontal) reinforcement for the two
walls. Each wall specimen was subjected to a prescribed cyclic horizontal displacement history applied at a height of 3.81 m
(150 in) from the foundation block until the occurrence of severe damage. The FE model of the walls is presented in
Figure 4c. An elastoplastic constitutive law with linear kinematic hardening is used to represent the material of the prestressing
TABLE 3 Calibration of parameter controlling rupture in steel material model
Specimen Bar type DcrðmonÞ Dcr

55
Pakiding et al, Wall 1 #3 0.23 0.55
#7 0.35 0.83
Pakiding et al,55 Wall 2 #3 0.20 0.48
#5 0.28 0.65
Beyer et al56 D12 0.18 0.44
D6 0.13 0.31
Schoettler et al57 #11 0.19 0.45
FIGURE 4 Reinforcement details, geometric configuration and computational model for post‐tensioned walls tested by Pakiding et al.55 A, Cross
section. B, Elevation view. C, FE model [Colour figure can be viewed at wileyonlinelibrary.com]
tendons in the model. The yield strength and the hardening slope of the material are set equal to 1675 MPa (243 ksi) and
2038 MPa (295 ksi), respectively, based on provided experimental data for the tendon stress‐strain curve. The prestressing
of the tendons was introduced in the models through the definition of an autogenous strain in the tendon material.
The analytically obtained hysteretic response for the two specimens is compared to the corresponding experimental obser-
vations in Figure 5. The points of the hysteretic curves corresponding to some key events in the analysis are also marked in the
figure. The capacity of both specimens is well captured in the negative direction, and it is slightly underestimated in the positive
direction. The analysis predicts the same peak strength for both positive and negative directions, while the experimentally
obtained strength in the positive direction was higher than that in the negative direction. Pakiding et al55 have not provided
any discussion or explanation for this aspect of the experimentally recorded response.
The sequence of damage accumulation obtained in the analysis of the two specimens is in good agreement with the exper-
imental observations. For the analysis of Wall 1, the first flexural cracks are formed for the cycles with a drift ratio of 0.1%, as
shown in Figure 6a. The longitudinal steel reinforcement first yields for the cycles with a drift ratio of 0.9%, as shown in
Figure 6b. Spalling of the cover concrete, described in the models through element removal, first occurs for the cycles with
a drift ratio of 1.35%, as shown in Figure 6c.
The occurrence of tensile yielding of the vertical reinforcement is followed by the accumulation of inelastic tensile strains.
During reversal of the applied lateral loads, the vertical rebars that have previously yielded in tension begin carrying compres-
sive stresses before the open cracks in the vicinity of these bars fully close. The vertical rebars also incur transverse deforma-
tions due to the multiaxial strain states in the surrounding concrete material. The combined effect of transverse deformations
and of compressive stresses following the development of inelastic tensile strains leads to buckling of the vertical reinforcement.
In the analytical models, buckling of reinforcement in the boundary regions of the specimens is visually detected, as depicted in
Figure 7.
FIGURE 5 Hysteretic response of post‐tensioned walls tested by Pakiding et al55 The instants in the analysis corresponding to the occurrence of
various damage states are provided in the plots. A, Wall 1. B, Wall 2 [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 6 Pre‐peak damage sequence in Wall 1 tested by Pakiding et al.55 A, Flexural cracking at drift ratio of 0.1%. B, Rebar yielding at drift
ratio of 0.9%. C, Cover spalling at drift ratio of 1.35% [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 7 Reinforcement buckling in wall specimens tested by Pakiding et al.55 A, Wall 1. B, Wall 2 [Colour figure can be viewed at
Both wall specimens eventually incurred significant strength degradation because of rebar buckling and subsequent
rupture.55 In the analytical models, the first rebar rupture for Wall 1 and Wall 2 occurs during the third and second cycle,
respectively, with a drift ratio of 4%. The bars that rupture first for the analyses of the two Wall specimens are presented in
Figure 8. Three vertical bars simultaneously rupture for Wall 1, whereas rupture of a single bar initially occurs in the
analysis of Wall 2. In the experimental test of Wall 1, rupture was obtained at the same exact cycle as in the analysis.
For Wall 2, rupture was experimentally obtained during the last loading cycle with a drift ratio of 3%. Given the inherent
variability in the yield strength, ultimate strength and low‐cycle fatigue resistance of rebars of the same material, the
agreement between the analysis and the experimental observations is deemed satisfactory. Finally, it is worth mentioning
that the damage pattern of Wall 1 was described by Pakiding et al.55 as a shear failure due to extensive damage in the
web of the wall as shown in Figure 9a. The specific damage pattern is satisfactorily captured by the analytical model, as
depicted in Figure 9b.
Table 4 compares the drift values corresponding to several damage states, for the experimental tests and the analytical
models. Overall, a satisfactory agreement between analysis and experiment is observed.
FIGURE 8 Fracture initiation for post‐tensioned walls tested by Pakiding et al.55 A, Wall 1, third cycle of 4% drift ratio in both the analytical model
and experiment. B, Wall 2, second cycle of 4% drift ratio in the analytical model, and third cycle of 3% drift ratio during the experiment [Colour figure
can be viewed at wileyonlinelibrary.com]
FIGURE 9 Experimental and analytical damage patterns of Wall 1 tested by Pakiding et al,55 at the end of the analysis. A, Experimental damage
pattern. B, Analytical crack opening strain contours [Colour figure can be viewed at wileyonlinelibrary.com]
TABLE 4 Drift ratio values corresponding to various damage states in the experiments and analytical models for the post‐tensioned walls tested by
Pakiding et al55
Wall 1 Wall 2
Damage state Experiment Analysis Experiment Analysis
Cracking 0.016% 0.012% 0.20% 0.20%

Rebar yielding 0.57% 0.59% 0.21% 0.22%
Concrete spalling 1.35% 1.12% 0.90% 0.92%
First rebar rupture 3.56% 3.62% 3.05% 2.98%
3.2 | Analysis of U‐shaped wall under bidirectional, quasi‐static loading

The next validation analysis is conducted for a U‐shaped RC wall specimen tested by Beyer et al56 and referred to therein as
specimen TUA. As shown in Figure 10a, the height of the specimen was 2.65 m (104 in), and the thickness of the wall was
150 mm (5.9 in). The same figure defines a coordinate system to facilitate the description of the bidirectional loading, which
corresponded to forces applied along the X‐ and Y‐directions. Figure 10b presents the cross‐sectional dimensions and reinforc-
ing details of the specimen. The boundary regions of the wall section included twenty‐two longitudinal # 4 bars with a diameter
of 12 mm, whereas the web portion of the section included a total of twenty‐eight # 2 rebars with a diameter of 6 mm. The
transverse reinforcement at the wall boundaries consisted of ties with a diameter of 6 mm and a spacing of 50 mm (2 in), while
the diameter and spacing of the transverse reinforcement at the web regions of the wall section were equal to 6 mm and 125 mm
(4.9 in), respectively. The specimen was subjected to a prescribed cyclic bidirectional displacement history until the occurrence
of severe strength and stiffness degradation due to rebar rupture and concrete crushing. The displacement histories were applied
at a height (above the base) of 3.35 m (132 in) and 2.95 m (116 in), for loading in the X‐ and Y‐direction, respectively. The FE
model of the wall specimen is depicted in Figure 10c.
The analytically obtained hysteretic response of the specimen is compared to the corresponding experimental observations
in Figure 11. The peak strength and the initial stiffness of the specimen are well captured by the analysis for the Y‐direction,
whereas they are slightly underestimated for the X‐direction. The analysis also provides a satisfactory representation of the hys-
teretic response of the specimen.
In both the analytical model and the experimental test, severe damage occurred due to rebar rupture, as shown in Figure 12.
Similar to the experiment, the first rebar rupture in the analysis occurs during the first loading cycle with a drift ratio of 2.5%
along the X‐direction. The analytical model successfully captures the damage associated with cover spalling and flexural and
shear cracking in the specimen, as depicted in Figure 13.
3.3 | Analysis of bridge column subjected to seismic loading

Analysis is also conducted for a full‐scale bridge column tested by Schoettler et al.57 The specimen was designed based on
modern seismic provisions and subjected to a sequence of ten ground motions. The column had a height of 7.31 m (288 in)
and a circular cross section with a diameter of 1.22 m (48 in), which was reinforced with eighteen # 11 vertical bars, each having
a diameter of 36 mm. The transverse reinforcement consisted of double # 5 ties at a spacing of 152 mm (6 in). The column
FIGURE 10 Configuration and computational model of the U‐shaped wall tested by Beyer et al.56 A, Isometric view. B, Cross section. C, FE
model [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 11 Hysteretic response of U‐shaped wall tested by Beyer et al.56 A, X‐direction. B, Y‐direction [Colour figure can be viewed at
FIGURE 12 Rebar rupture in the U‐

shaped wall tested by Beyer et al56 A,
Experiment. B, Analysis [Colour figure can
be viewed at wileyonlinelibrary.com]
configuration and reinforcing details are depicted in Figure 14a. The specimen included a concrete block at the top, with a total
weight of 2.32 MN (521 kips), to represent the seismic mass of the bridge column. The FE model of the specimen is presented
in Figure 14b and has been analyzed for the entire motion sequence of the experimental test.
The analytically obtained drift time histories for nine of the ten motions of the sequence are compared to the corresponding
experimental records in Figure 15. A good agreement is obtained in terms of the peak displacements and drift histories, except
FIGURE 13 Damage at the base of the U‐shaped wall tested by Beyer et al56 A, Experiment. B, Analysis [Colour figure can be viewed at
FIGURE 14 Configuration and analytical model for the column tested by Schoettler et al57 A, Dimensions and reinforcement details. B, FE model
[Colour figure can be viewed at wileyonlinelibrary.com]
for the last motion in the sequence. During the experimental test for that motion, the column came to contact with a safety sup-
port tower that was provided to prevent damage of the shake table from total collapse of the specimen.57 Since the effect of this
support tower is not accounted for in the analytical model, the observed discrepancy between the analytically predicted and the
experimentally recorded drift time histories for this motion is expected.
Similar to the experimental observations, the analytical model predicts that yielding of the longitudinal reinforcement first
occurs during the third ground motion. During the same motion, concrete cover spalling is observed for the first time in the
analytical model, as depicted in Figure 16a. During the fifth and sixth motions, more extensive loss of cover material occurred
in the test, and more cover elements are removed in the analysis, as shown in Figure 16b. At the end of the sixth motion, the core
concrete in the vicinity of the base of the column was significantly damaged, as shown in Figure 17. In the specific figure, the
concrete damage in the analytical model is quantified through a contour plot of the hardening‐softening variable κ, which is a
measure of the accumulated inelastic strains in the concrete constitutive model.
The experimental test specimen eventually incurred significant damage in the form of rebar buckling and subsequent rup-
ture, as shown in Figure 18a. In the analytical model, the buckling of rebars is visually observed during the seventh motion,
similar to the corresponding experimental observation, and it is significant during the eighth motion, as shown in Figure 18b.
Additionally, as depicted in Figure 18c, the first rebar rupture is obtained during the eighth ground motion, which was also
the case in the experimental test.
4 | PARAMETRIC ANALYS ES
Despite the capability to fully calibrate the values of all parameters of the employed constitutive laws in accordance with the
procedure presented in Section 3, there are still aspects of the analytical models—such as the use of viscous damping for anal-
yses involving quasi‐static loading—which can affect the analytical results. Furthermore, it is deemed necessary to assess the
FIGURE 15 Comparison of analytically obtained and experimentally recorded drift histories for the column tested by Schoettler et al57 A, EQ2:
100% Loma Prieta‐900. B, EQ3: 100% Loma Prieta. C, EQ4: 100% Loma Prieta‐900. D, EQ5: 100% Kobe. E, EQ6: 100% Loma Prieta. F, EQ7:
100% Kobe. G, EQ8: −120% Kobe. H, EQ9: 120% Kobe. i, EQ10: 120% Kobe [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 16 Loss of cover during the test sequence, for the specimen by Schoettler et al.57 A, After the third motion. B, After the fifth motion
significance of several assumptions regarding material behavior, such as accounting for the increased concrete material ductility
due to confinement and for the strain penetration effect. This section presents a set of parametric analyses to investigate the
effect of the aforementioned aspects on the analytical results.
FIGURE 17 Damage at base of column specimen tested by Schoettler et al57 after the sixth motion [Colour figure can be viewed at
FIGURE 18 Rebar buckling and first rebar rupture observed during the eighth motion for the specimen tested by Schoettler et al57 A, Damage in
experimental test. B, Rebar buckling in analysis. C, Rebar rupture in analysis [Colour figure can be viewed at wileyonlinelibrary.com]
4.1 | Effect of viscous damping for analysis of components under quasi‐static loading
A first set of parametric analyses is conducted to determine the significance of viscous damping for analyses involving quasi‐
static loading. To this end, the analyses of the two specimens tested by Pakiding et al55 are repeated, this time without any vis-
cous damping in the models. The hysteretic response obtained for the two specimens is depicted in Figure 19. The overall
response obtained without damping is similar to that of the analysis with damping, except for the initial cycles of loading, where
high‐frequency oscillations are observed in the hysteretic curves. Damping primarily affects the structural response in the elastic
stage of behavior, and its effect becomes negligible after the occurrence of inelastic deformations.
FIGURE 19 Analytically obtained hysteretic response of specimens by Pakiding et al55 with and without viscous damping. A, Wall 1, B, Wall 2
4.2 | Effect of increased concrete ductility due to confinement

To demonstrate the effect of the increased concrete ductility due to confinement on the nonlinear performance, the analysis of
Wall 2 by Pakiding et al55 is repeated for a value of the parameter d of the concrete constitutive law equal to zero. This modified
calibration essentially neglects the effect of local confining pressure on the compressive ductility of the concrete. The analytically
obtained hysteretic response is compared with the experimental data and with the analytical results of Section 3.1 in Figure 20a.
The figure shows that neglecting the effect of confinement on the material ductility leads to significant underestimation of the
wall strength and ductility. Thus, the use of multiaxial material models for concrete which do not account for increased material
ductility due to confinement would entail predictions of premature lateral strength degradation in the walls.
4.3 | Effect of strain penetration

The effect of strain penetration on the response of the wall specimens is elucidated by repeating the analysis of Wall 2 by Pakiding
et al,55 this time assuming perfect bond between the rebars and the concrete in the foundation block. The analytically obtained
hysteretic response obtained with and without the strain penetration effect is compared to the experimental observations in
Figure 20b. The figure indicates that ignoring the strain penetration effect has a minor impact, leading to a slight increase in
the stiffness and peak strength. This conclusion only applies for the walls considered in the parametric analysis, as previous
studies have shown that strain penetration is important for analysis of columns58 or beam‐to‐column connections.61
5 | DISCUSSI ON
The proposed methodology has been shown to satisfactorily capture the global response and damage patterns of RC compo-
nents under cyclic loads. The validation analyses have all been conducted with a consistent calibration scheme, without any
effort to fine‐tune material parameters to improve the agreement with the experimental tests. This fact emphasizes the very good
predictive capabilities of the proposed method.
The modeling scheme proposed herein may require several enhancements before it is applied to simulation of components
such as beam‐to‐column joints. Specifically, it is well‐established that the hysteretic response of such components is signifi-
cantly affected by the bond‐slip behavior of the reinforcing steel.62 For this reason, the implementation of a more refined
bond‐slip model that can account for cyclic deterioration of bond resistance, such as the one by Murcia‐Delso and Shing,53
is deemed necessary for the simulation of such components.
An important requirement for refined FE models pertains to capturing full collapse of structural components. The present
study has considered tests that were conducted up to the occurrence of severe damage, but full specimen collapse was not
pursued, to avoid damaging the laboratory equipment. A preliminary verification of the capability of the proposed scheme to
capture component collapse is deemed necessary. To this end, an additional analysis of the specimen tested by Schoettler
et al57 is conducted, wherein the parameter Dcr is set to 50% of the corresponding value in Table 3. The reason for assigning
a fictitiously low value to Dcr is to obtain rupture in more rebars than in the actual experimental test, leading to more severe
strength degradation and, ultimately, to collapse in the analysis. The analysis for the fictitiously low value of Dcr does predict
collapse during the eighth motion of the sequence, as shown in the drift time history and the deformed configuration presented
FIGURE 20 Parametric analysis for Wall 2 tested by Pakiding et al.55 A, Effect of confinement on ductility. B, Effect of strain penetration [Colour
figure can be viewed at wileyonlinelibrary.com]
in Figure 21. The analysis has given full collapse of the column without any problem in the stress update process of the con-
stitutive laws or in the global solution scheme. It is important to emphasize that the collapse of Figure 21 did not occur in
the actual experimental test or in the corresponding validation analysis of Section 3.3.
The analyses presented herein rely on explicit transient integration. To verify that the inertial effects due to the use of the
explicit solution scheme for quasi‐static loading do not significantly affect the results obtained in the present study, the analysis
of the wall specimen tested by Beyer et al56 is repeated using an implicit static solution scheme. This analysis also allows the
verification of the capability of the proposed modeling approach to be combined with an implicit static solver. The analytically
obtained hysteretic response is compared with the corresponding experimental data in Figure 22. As can be seen from the figure,
the lateral strength of the wall is satisfactorily captured in both directions. Additionally, the analytical model can capture the
strength degradation due to rebar rupture. The only significant disagreement with the hysteretic curve obtained for explicit anal-
ysis is in the initial stiffness, which is higher for implicit analysis. This discrepancy in initial stiffness is attributed to the fact that
the analysis program uses stiffness‐based hourglass control for all elements in the implicit analysis, while viscous‐based hourglass
control has been used for the solid elements representing the core concrete in the explicit analysis presented in Section 3.2.
For simulations focused on quasi‐static loading, implicit static solution algorithms may be deemed preferable because they
eliminate the possibility for spurious inertial effects on the solution. Another issue which may negatively affect the efficiency of
explicit schemes is that they are conditionally stable, i.e. they require a sufficiently small time‐step to preclude the occurrence of
numerical instability.43 On the other hand, implicit algorithms may not be well suited for analyses of large models because they
typically entail large memory requirements, associated with the need to store a global tangent stiffness matrix. Furthermore,
implicit schemes may not be appropriate for simulations of failure or collapse due to the probability of failure to satisfy the con-
vergence criteria of the global solution algorithms.
FIGURE 21 Demonstration of capability to analytically simulate collapse. A, Drift history. B Deformed mesh and crack strain contours [Colour
figure can be viewed at wileyonlinelibrary.com]
FIGURE 22 Hysteretic response of U‐shaped wall by Beyer et al,56 obtained using implicit static global solution scheme. A, X‐direction. B, Y‐
direction [Colour figure can be viewed at wileyonlinelibrary.com]
A comparative assessment of the efficiency of explicit and implicit schemes in the context of parallel computation is pursued
here. To this end, the analysis of the wall by Beyer et al56 has been repeated for a varying number of processors. The dependence
of the analysis time on the number of processors (CPUs), NCPU, employed in the simulation, constitutes an indicator of the effi-
ciency of parallel processing. A plot providing this information is presented in Figure 23a for the explicit analysis, and in
Figure 23b for the implicit analysis. For the range of NCPU values examined here, an increase in NCPU for the explicit analysis
always leads to a decrease in the required computation time. On the other hand, the time required for a single global iteration
in the implicit scheme is a decreasing function of the number of CPUs for relatively small values of NCPU. An increase of NCPU
beyond 16 is found to lead to an increase in the computation time. This increase is attributed to the increasing contribution of the
operations involving communication between different processors to the computation time. This communication among
processors is introduced by the need to solve linear systems of equations. Although the exact dependence of the computation time
on the number of CPUs depends on the specific iterative scheme used and on the algorithm for the solution of the system of
equations, the trend in efficiency obtained herein is expected to characterize all implicit solution schemes.
6 | CONCLUSIONS
A three‐dimensional, continuum‐based, FE analysis framework has been developed for the simulation of RC structures sub-
jected to lateral cyclic loads. The analysis method uses recently developed material laws, which can capture the behavior and
failure of concrete and reinforcing steel. The rebars are simulated with geometrically nonlinear beam elements to naturally cap-
ture the effect of inelastic buckling on the structural response. Loss of concrete cover and rebar rupture are accounted for with
element removal techniques.
The modeling scheme has been validated with experimental tests of RC components under quasi‐static and dynamic loads.
The proposed method has been found capable of satisfactorily reproducing the global response and damage accumulation for
the experimentally tested components. Additional parametric studies have shown that viscous damping for the analysis of com-
ponents under quasi‐static loading has a significant effect on the stage of response before the development of inelastic deforma-
tions and that the strain penetration effect is not significant for RC walls. Although collapse did not occur in the experimental
tests considered, the capability of the modeling scheme to capture full collapse, without failure in the stress update process of
the material laws, has been demonstrated.
AC KN OWL ED GM E NT S
The authors thank the personnel of the Livermore Software and Technology Corporation (LSTC) for support in the implementa-
tion of the constitutive models in LS‐DYNA. Furthermore, the support of the Advanced Research Computing (ARC) division of
Virginia Tech for conducting the simulations in the cluster Blueridge is greatly appreciated. Finally, the authors thank Professor R.
Sause, Professor K. Beyer, and Dr. M. Schoettler for providing the recorded data from the experimental tests on the post‐tensioned
walls, U‐shaped wall, and bridge column, respectively. Any opinions expressed in this paper are those of the authors alone.
FIGURE 23 Effect of number of processors (CPUs) on the time required for the analysis of the wall specimen by Beyer et al.56 A, Explicit
integration scheme. B, Implicit integration scheme [Colour figure can be viewed at wileyonlinelibrary.com]
R E F E RENC E S
1. ACI 318‐14. Building Code Requirements for Structural Concrete. Farmington Hills, MI: American Concrete Institute; 2014.
2. Talaat MM, Mosalam MK. Computational modeling of progressive collapse in reinforced concrete frame structures. In: PEER Report 2007/10,
Pacific Earthquake Engineering Center. Berkeley, CA: University of California; 2008.
3. Iribarren BS, Berke P, Bouillard P, Vantomme J, Massart T. Investigation of the influence of design and material parameters in the progressive
collapse analysis of RC structures. Eng Struct. 2011;33(1):2805‐2820.
4. Sagiroglu S, Sasani M. Progressive collapse‐resisting mechanisms of reinforced concrete structures and effects of initial damage locations. J Struct
Eng. 2013;140(3), 04013073):1‐12.
5. Lu X, Lu X, Guan H, Ye L. Collapse simulation of reinforced concrete high‐rise building induced by extreme earthquakes. Earthq Eng Struct Dyn.
2013;42(5):705‐723.
6. Lu Y, Panagiotou M. Three‐dimensional cyclic beam‐truss model for nonplanar reinforced concrete walls. J Struct Eng. 2013; 04013071;1‐11.
7. Seible F, LaRovere HL, Kingsley GR. Nonlinear analysis of reinforced concrete masonry shear wall structures—monotonic loading. TMS J.
1990;9(1):70‐77.
8. Lotfi HR, Shing PB. Interface model applied to fracture of masonry structures. J Struct Eng. 1994;120(1):63‐80.
9. Mehrabi AB, Shing PB. Finite element modeling of masonry‐infilled RC frames. J Struct Eng. 1997;123(5):604‐613.
10. Seible F, LaRovere HL, Kingsley GR. Nonlinear analysis of reinforced concrete masonry shear wall structures—cyclic loading. TMS J.
1990;9(1):60‐69.
11. Kwan WP, Billington SL. Simulation of structural concrete under cyclic load. J Struct Eng. 2001;127(12):1391‐1401.
12. Palermo D, Vecchio FJ. Compression field modeling of reinforced concrete subjected to reversed loading: verification. ACI Struct J. 2004;101(2):155‐164.
13. Kim TH, Lee KM, Chung YS, Shin HM. Seismic damage assessment of reinforced concrete bridge columns. Eng Struct. 2005;27(4):576‐592.
14. Koutromanos I, Stavridis A, Shing PB, Willam K. Numerical modeling of masonry‐infilled RC frames subjected to seismic loads. Comput Struct.
2011;89(11):1026‐1037.
15. Koutromanos I, Shing PB. Cohesive crack model to simulate cyclic response of concrete and masonry structures. ACI Struct J. 2012;109(3):349‐358.
16. Murray YD, Abu‐Odeh AY, Bligh RP. Evaluation of LS‐DYNA concrete material model 159. Report No. FHWA‐HRT‐05‐063. McLean, VA:
Federal Highway Administration; 2007.
17. Lin X, Zhang YX, Hazell PJ. Modelling the response of reinforced concrete panels under blast loading. Mater Des. 2014;56(1):620‐628.
18. Grassl P, Jirásek M. Damage‐plastic model for concrete failure. Int J Solids Struct. 2006;43(22):7166‐7196.
19. Zheng Y, Robinson D, Taylor S, Cleland D. Finite element investigation of the structural behavior of deck slabs in composite bridges. Eng Struct.
2009;31(8):1762‐1776.
20. Teng JG, Xiao QG, Yu T, Lam L. Three‐dimensional finite element analysis of reinforced concrete columns with FRP and/or steel confinement.
Eng Struct. 2015;97(1):15‐28.
21. Bao Y, Lew HS, Kunnath SK. Modeling of reinforced concrete assemblies under column‐removal scenario. J Struct Eng. 2012;140(1),
04013026):1‐13.
22. Sasani M, Werner A, Kazemi A. Bar fracture modeling in progressive collapse analysis of reinforced concrete structures. Eng Struct.
2011;33(2):401‐409.
23. Kaya M, Arslan AS. Analytical modeling of post‐tensioned precast beam‐to‐column connections. Mater Des. 2009;30(9):3802‐3811.
24. Thiagarajan G, Kadambi AV, Robert S, Johnson CF. Experimental and finite element analysis of doubly reinforced concrete slabs subjected to
blast loads. Int J Impact Eng. 2015;75(1):162‐173.
25. Gangi MJ. Analytical modeling of the repair impact‐damaged prestressed concrete bridge girders. Blacksburg, VA: Master0 s thesis, Virginia
Tech; 2015.
26. Barth KE, Wu H. Efficient nonlinear finite element modeling of slab on steel stringer bridges. Finite Elem Anal Des. 2006;42(14):1304‐1313.
27. Chung W, Sotelino ED. Three‐dimensional finite element modeling of composite girder bridges. Eng Struct. 2006;28(1):63‐71.
28. Noguchi H, Uchida K. Finite element method analysis of hybrid structural frames with reinforced concrete columns and steel beams. J Struct Eng.
2004;130(2):328‐335.
29. Faria R, Oliver J, Cervera M. Modeling material failure in concrete structures under cyclic actions. J Struct Eng. 2004;130(12):1997‐2005.
30. Spiliopoulos KV, Lykidis GC. An efficient three‐dimensional solid finite element dynamic analysis of reinforced concrete structures. Earthq Eng
Struct Dyn. 2006;35(2):137‐157.
31. Eligehausen R, Ožbolt J, Genesio G, Hoehler MS, Pampanin S. Three‐Dimensional Modelling of Poorly Detailed RC Frame Joints. Napier, New
Zealand: NZSEE Conference; 2006.
32. Hawileh RA, Rahman A, Tabatabai H. Nonlinear finite element analysis and modeling of a precast hybrid beam–column connection subjected to
cyclic loads. App Math Model. 2010;34(9):2562‐2583.
33. Deaton J. Nonlinear finite element analysis of reinforced concrete exterior beam‐column joints with non‐seismic detailing. PhD. Atlanta, GA:
dissertation, Georgia Institute of Technology; 2013.
34. Faria R, Oliver J, Cervera M. A strain‐based plastic viscous‐damage model for massive concrete structures. Int J Solids Struct. 1998;35(14):1533‐1558.
35. Ožbolt J, Li Y, Kožar I. Microplane model for concrete with relaxed kinematic constraint. Int J Solids Struct. 2001;38(16):2683‐2711.
36. Selby RG, Vecchio FJ. A constitutive model for analysis of reinforced concrete solids. Can J Civ Eng. 1997;24(3):460‐470.
37. Filippou F, Popov E, Bertero V. Effects of bond deterioration on hysteretic behavior of reinforced concrete joints. Report No. UCB/EERC‐83/19.
Berkeley, CA, 1983.
38. Sasani M, Kropelnicki J. Progressive collapse analysis of an RC structure. Struct Design Tall Spec Build. 2008;17(4):757‐771.
39. Yu T, Teng JG, Wong YL, Dong SL. Finite element modeling of confined concrete: I. Drucker‐Prager type plasticity model. Eng Struct.
2010;32(3):665‐679.
40. Murcia‐Delso J. Bond‐slip behavior and development of bridge column longitudinal reinforcing bars in enlarged pile shafts. PhD dissertation,
University of California: San Diego, La Jolla, CA, 2013.
41. Moharrami M, Koutromanos I. Triaxial constitutive model for concrete under cyclic loading. J Struct Eng. 2016;142(7), 04016039):1‐15.
42. Kim SH, Koutromanos I. Constitutive model for reinforcing steel under cyclic loading. J Struct Eng. 2016;142(12): 04016133
43. Belytschko T, Liu WK, Moran B, Elkhodary K. Nonlinear Finite Elements for Continua and Structures. 2nd ed. New York, NY: Wiley; 2014.
44. Flanagan DP, Belytschko T. A uniform strain hexahedron and quadrilateral with orthogonal hourglass control. Int J Numer Methods Eng.
1981;17(5):679‐706.
45. Belytschko T, Bindeman LP. Assumed strain stabilization of the eight node hexahedral element. Comput Methods Appl Mech Eng.
1993;105(2):225‐260.
46. Kang HD, Willam K, Shing PB, Spacone E. Failure analysis of R/C columns using a triaxial concrete model. Comput Struct. 2000;77(5):423‐440.
47. Lee J, Fenves GL. Plastic‐damage model for cyclic loading of concrete structures. J Eng Mech. 1998;124(8):892‐900.
48. LSTC. Hughes‐Liu. Beam element. In: LS‐DYNA Theory Manual. Livermore Software and Technology Corporation, Livermore, CA, 2006.
Available online at: http://www.lstc.com/pdf/ls‐dyna_theory_manual_2006.pdf
49. Hughes TJ, Liu WK. Nonlinear finite element analysis of shells: part I. Three‐dimensional shells. Comput Methods Appl Mech Eng.
1981;26(3):331‐362.
50. Maekawa K, Pinanmas A, Okamura H. Nonlinear Mechanics of Reinforced Concrete. New York, NY: Spon Press; 2003.
51. Dodd LL, Restrepo‐Posada JI. Model for predicting cyclic behavior of reinforcing steel. J Struct Eng. 1995;121(3):433‐445.
52. Huang Y, Mahin SA. Simulating the inelastic seismic behavior of steel braced frames including the effects of low‐cycle fatigue. PEER Report
2010/104. Berkeley, CA: Pacific Earthquake Engineering Center, University of California; 2010.
53. Murcia‐Delso J, Shing PB. Bond‐slip model for detailed finite‐element analysis of reinforced concrete structures. J Struct Eng. 2014;141(4),
04014125):1‐10.
54. LSTC. LS‐DYNA User0 s Manual. Livermore, CA: Livermore Software and Technology Corporation; 2007.
55. Pakiding L, Pessiki S, Sause R, Rivera M. Experimental testing of cast‐in‐place seismic resistant unbonded post‐tensioned special reinforced con-
crete walls. ATLSS Report No. 14–07, Lehigh University, Bethlehem, PA, 2014.
56. Beyer K, Dazio A, Priestley MJN. Quasi‐static cyclic tests of two U‐shaped reinforced concrete walls. J Earthq Eng. 2008;12(7):1023‐1053.
57. Schoettler MJ, Restrepo JI, Guerrini G, Duck DE, Carrea F. A full‐scale, single‐column bridge bent tested by shake‐table excitation. In: Center for
Civil Engineering Earthquake Research. Reno, NV: University of Nevada; 2012.
58. Moharrami M, Koutromanos I, Panagiotou M, Girgin SC. Analysis of shear‐dominated RC columns using the nonlinear truss analogy. Earthq Eng
Struct Dyn. 2015;44(5):677‐694.
59. Bažant ZP. Concrete fracture models: testing and practice. Eng Fract Mech. 2002;69(2):165‐205.
60. Feenstra, P. Computational aspects of biaxial stress in plain and reinforced concrete. Ph.D. dissertation, Delft University of Technology, Delft,
The Netherlands, 1993.
61. Zhao J, Sritharan S. Modeling of strain penetration effects in fiber‐based analysis of reinforced concrete structures. ACI Struct J.
2007;104(S14):133‐141.
62. Beckingsale CW. Post elastic behavior of reinforced concrete beam‐column joints. PhD dissertation, University of Canterbury: Christchurch, New
Zealand, 1980.
How to cite this article: Moharrami M, Koutromanos I. Finite element analysis of damage and failure of reinforced
concrete members under earthquake loading. Earthquake Engng Struct Dyn. 2017;1–20. https://doi.org/10.1002/
eqe.2932

Finite Element Analysis of Damage and Failure of Reinforced Concrete Members Under Earthquake Loading

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Finite Element Analysis of Damage and Failure of Reinforced Concrete Members Under Earthquake Loading

Caricato da

Copyright:

Formati disponibili

Received: 12 December 2016 Revised: 14 April 2017 Accepted: 22 May 2017

Finite element analysis of damage and failure of reinforced

Mohammadreza Moharrami1 | Ioannis Koutromanos2

2.1 | Element formulation and constitutive model for concrete

2.2 | Element formulation and constitutive model for reinforcing steel

2.3 | Accounting for the strain penetration effect

FIGURE 2 Behavior of reinforcing steel

TABLE 1 Concrete material model parameters

Specimen fc (MPa) d Gt (N/m) Gc (N/m) M

Pakiding et al,55 Wall 1 43.4 6.3 88.9 8890 0.053

TABLE 2 Steel material parameters

3.1 | Analysis of post‐tensioned RC walls under quasi‐static loading

TABLE 3 Calibration of parameter controlling rupture in steel material model

Specimen Bar type DcrðmonÞ Dcr

Cracking 0.016% 0.012% 0.20% 0.20%

3.2 | Analysis of U‐shaped wall under bidirectional, quasi‐static loading

3.3 | Analysis of bridge column subjected to seismic loading

FIGURE 12 Rebar rupture in the U‐

4.2 | Effect of increased concrete ductility due to confinement

4.3 | Effect of strain penetration

Potrebbero piacerti anche