10 1061@ (Asce) ST 1943-541X 0001492

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Practical Nonlinear Modeling of Reinforced

Concrete Structural Walls


Kristijan Kolozvari, Ph.D., P.E., M.ASCE 1; and John W. Wallace, Ph.D., P.E., M.ASCE 2

Abstract: Current engineering practice typically relies on the use of fiber-based modeling approaches with uncoupled axial-bending (P-M)
and shear (V) responses to simulate nonlinear behavior of reinforced concrete (RC) structural walls. However, more sophisticated numerical
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

models are available that incorporate coupled P-M-V behavior. The effect of using uncoupled and coupled modeling approaches and the
influence of various modeling assumptions, particularly modeling parameters related to wall shear behavior, on computed global and local
building responses are reported. A five-story archetype RC wall-frame building designed according to current U.S. code provisions is used for
the assessment. The results indicate that modeling parameters associated with wall shear behavior have a significant effect on computed
responses for uncoupled models; use of commonly recommended effective shear stiffness of 0.2Ec Aw to account for effects of concrete
cracking provides a reasonable estimate of roof displacement response. However, wall shear demands and interstory drift at stories where
wall yielding occurs tend to be overestimated and underestimated, respectively, in comparison with results obtained using coupled wall
models. DOI: 10.1061/(ASCE)ST.1943-541X.0001492. © 2016 American Society of Civil Engineers.
Author keywords: Analysis and computation.

Introduction and Background available computer programs (e.g., PERFORM-3D) and have been
widely used to model RC walls. Studies that compare model and
Reinforced concrete (RC) structural walls are effective for resisting experimental results (e.g., Orakcal and Wallace 2006; PEER and
lateral loads imposed by wind or earthquakes on building struc- ATC 2010) show that uncoupled fiber models provide reasonably
tures because they provide substantial lateral strength and stiff- accurate predictions of flexure responses. However, the inability of
ness, in addition to deformation capacity, to meet the demands of fiber models to account for interaction (coupling) between axial-
strong earthquake shaking. In general, it is not feasible to de- flexural and shear behavior is a significant drawback, as studies
sign a structural wall to remain elastic during a severe earthquake have shown that uncoupled models tend to underestimate axial
[i.e., R-factors used in ASCE 7-10 (ASCE 2013) are typically 5 compressive strains in wall boundaries even in relatively slender
or 6]; therefore, inelastic wall deformations are expected, with RC walls controlled by flexure (Orakcal and Wallace 2006), and
the distributions and magnitudes of the inelastic deformations de- overestimate the lateral load capacity in moderately slender walls
pending on the attributes of the structural system. Because struc- (Kolozvari 2013). In addition, when using an uncoupled model, a
tural walls are primary, and in some cases, the only lateral-force shear force-deformation relation must be defined. Common ap-
resisting elements, it is essential that analytical tools capable of proaches are to use a (1) linear relation with an effective shear
capturing the hysteretic behavior of structural walls, in addition to stiffness (e.g., GAeff ¼ 0.2Ec Aw ), (2) bilinear relation with an un-
the interaction of walls with other structural members, are available cracked (i.e., GAuncracked ¼ 0.4Ec Aw ) and cracked region (e.g.,
to researchers and engineers to investigate the design of new build- GAeff ¼ 0.2Ec Aw ), and (3) trilinear relation that includes a bilinear
ings and evaluation of existing buildings. relation like (2) along with a postyield region. The approach used
Current practice for design and evaluation of buildings where will affect the obtained results; however, a systematic assessment of
nonlinear analysis procedures are used involves application of the effects and implications has not been presented.
fiber-based models with flexural response simulated using a series A number of analytical models have been proposed to cap-
of uniaxial elements (or macro-fibers) based on stress-strain/force- ture the observed shear-flexure interaction (SFI) in RC walls
deformation relations for concrete and steel, along with the plane (e.g., Kolozvari et al. 2015a; Fischinger et al. 2012; Panagiotou
sections assumption, whereas shear behavior is typically accounted et al. 2011; Massone et al. 2006; Henry and Lu 2014; Dashti et
for by using a horizontal spring with a specified force–deformation al. 2014; Belletti et al. 2013). An effective approach to capture
(backbone) relation that is usually uncoupled from flexural behav- the SFI was proposed by Petrangeli et al. (1999) and Massone et al.
ior. Fiber models have been implemented into various research- 2006)) and implemented this approach for monotonic analysis;
oriented (e.g., OpenSees, McKenna et al. 2000) and commercially Kolozvari et al. (2015a) extended it to address reversed cyclic load-
ing. The model proposed by Kolozvari et al. addresses the issues
1
Assistant Professor, Dept. of Civil and Environmental Engineering, identified earlier for the uncoupled models, as coupling between
California State Univ., Fullerton, CA 92831 (corresponding author). shear and flexural responses is captured at the model element
E-mail: kkolozvari@fullerton.edu level. Therefore, the effect of shear behavior on concrete compres-
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of
sive strain is directly incorporated, and shear stiffness evolves ac-
California, Los Angeles, CA 90095-1593. E-mail: wallacej@ucla.edu
Note. This manuscript was submitted on April 6, 2015; approved on
cording to computed responses and assumed material behavior
December 10, 2015; published online on February 25, 2016. Discussion (versus use of a backbone relation); for example, Kolozvari (2013)
period open until July 25, 2016; separate discussions must be submitted showed that the wall shear stiffness depends on wall shear demand,
for individual papers. This paper is part of the Journal of Structural En- wall axial load, the ratio of M u =V u lw , and the extent of nonlinear
gineering, © ASCE, ISSN 0733-9445. flexural deformations. In addition, the model has been shown to be

© ASCE G4016001-1 J. Struct. Eng.

J. Struct. Eng., G4016001


an effective tool for the analytical modeling of nonlinear behavior conducted for shaking in the transverse direction only, in which
of RC walls, capable of successfully describing global and local the lateral-force-resisting elements include two identical one-bay
wall responses under cyclic loading (Kolozvari et al. 2015b). frames located at the building perimeter (axis 1 and 8) and two
identical walls located near the center of the building (axis 4 and
5), as shown in Fig. 1. An iterative design procedure, outlined in the
Objectives and Scope following section, resulted in cross section dimensions of walls,
beams, and columns of 0.30 × 6.10 m (12 × 240 in:), 0.46 × 0.81 m
Fiber-based modeling approaches, with uncoupled axial-flexural (18 × 32 in:, width × depth), and 0.71 × 0.71 m (28 × 28 in:),
(P-M) and shear (V) behavior, are used widely for the seismic respectively.
design of new buildings and evaluation and retrofit of existing
buildings. The recent development, validation, and implementation Design Approach
of a coupled P-M-V model that captures the interaction between Structural design is performed for a residential building character-
shear and flexural behavior under cyclic loading into the nonlinear ized with an importance factor of I ¼ 1.0, risk category I, and
range for RC walls (e.g., Kolozvari et al. 2015a, b) provides an design category D, according to ASCE 7-10 (ASCE 2013; S11.5
opportunity to conduct novel systematic studies to assess the effects and S11.6). The frame was designed to resist 25% of the earthquake
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

of the various modeling approaches and assumptions on computed lateral load, which classifies the building structural system as a
global and local responses of RC structural walls. In this study, dual system according to ASCE 7-10 (ASCE 2013). Concrete com-
comparative analytical studies are conducted using a fiber-based pressive strength of fc0 ¼ 34.47 MPa (5,000 psi) and reinforcing
wall model with uncoupled shear and flexural responses (Taucer steel (longitudinal and transversal) with yield strength of fy ¼
et al. 1991) and a model with coupled flexural and shear behavior 413.69 MPa (60,000 psi) were used. A uniformly distributed dead
(Kolozvari et al. 2015a), both implemented in the widely used com- load of 7.18 kN=m2 (150 psf) and live load of 1.91 kN=m2 (40 psf)
putational platform OpenSees (McKenna et al. 2000). Particular per ASCE 7-10 (ASCE 2013; Table 4-1, Residential Building) were
emphasis is placed on the effect of modeling assumptions related used for the design, whereas load combinations were adopted ac-
to wall shear behavior (e.g., effective shear stiffness, linear versus cording to ASCE 7-10 (ASCE 2013; S2.3). Because only one bay
bilinear relationship) on computed responses by identifying ap- of seismic-resisting perimeter frame on each side of the structure is
proaches and/or assumptions that produce unreliable or inconsis- used, which resists less than 35% of the seismic force, the redun-
tent results, in addition to providing recommendations for practical dancy factor [ASCE 7-10 (ASCE 2013; S12.3.4)] was taken as
applications for the nonlinear modeling of RC walls for design ρ ¼ 1.3.
and assessment. To provide context to the comparative studies pre- Seismic lateral loads on the building were calculated using the
sented, the modeling approaches selected are used to analyze the equivalent lateral force procedure (ELFP) according to ASCE 7-10
behavior of an archetype five-story RC wall-frame building de- (ASCE 2013; S12.8), based on the code prescribed spectrum char-
signed according to current U.S. practice [i.e., ACI 318-11 (ACI acterized with mapped short period and 1-s period accelerations
2011) and ASCE 7 standards] under a single ground motion and of SS ¼ 1.5 and S1 ¼ 0.6 g, respectively, assuming Site Class B
a suite of seven ground motions. (Fa ¼ 1.0, FV ¼ 1.5) and design spectral acceleration parameters
of SDS ¼ 1.0 g and SD1 ¼ 0.6 g, which yielded T 0 ¼ 0.12 s and
T S ¼ 0.60 s. Based on the fundamental period of the building
Nonlinear Modeling Studies computed according to ASCE 7-10 (ASCE 2013; S12.8.2)
(T ¼ 0.60 s), the base shear of V ¼ CS W ¼ 0.19W ¼ 5,338 kN
The following sections provide information on archetype building
(1,200 kips) was obtained, in which the total effective seismic
geometry, design approach, and member proportioning and detail-
weight of the building was W ¼ 28,024 kN (6,300 kips). Axial
ing, as well as descriptions of the analytical models.
demands resulting from gravity loads on walls and columns are
computed based on prescribed dead and live loads [reduced in line
Archetype Building with ASCE 7-10 (ASCE 2013; S4.7)] according to tributary areas
for the considered load combinations, whereas axial load resulting
Building Geometry from seismic actions (horizontal and vertical) were computed ac-
Plan and elevation/section views of the archetype building cording to ASCE 7-10 (ASCE 2013; S12.14.3).
are shown in Fig. 1. The building footprint is 42.7 × 18.3 m Instead of using linear analysis, a simple (and approximate)
(140 × 60 ft), with 6.1 m (20 ft) long spans. An analysis is collapse mechanism approach was used to determine the design

(a) (b) (c)

Fig. 1. Archetype building: (a) plan view; (b) frame elevation; (c) wall elevation

© ASCE G4016001-2 J. Struct. Eng.

J. Struct. Eng., G4016001


strength requirements for walls, beams, and columns using the A one-bay, five-story frame [Fig. 1(b)] was designed to resist
vertical distribution of seismic lateral forces [ASCE 7-10 (ASCE 25% of the lateral seismic demand obtained using ELFP. Based
2013; S12.8.3)]. The assumed collapse mechanism of the structural on axial and flexural demands, a 0.71 × 0.71 m (28 × 28 in:) col-
system included wall and column yielding at the base, and the beam umn with 12 #11 bars [db ¼ 35.81 mm (1.410 in.)] was adopted
yielding at each level [negative beam nominal moment capacity along the height of the building [Fig. 2(b)]; the column P-M
was assumed to be one-half of the positive beam nominal capacity strengths satisfy ACI 318-11 (ACI 2011; S10.3) requirements. The
in accordance with ACI 318-11 (ACI 2011; S21.5.2.2)], and the beam design was characterized by a 0.46 × 0.81 m (18 × 32 in:)
strong-column weak-beam condition at each joint, determined the cross section, with seven #9 bars [db ¼ 28.65 mm (1.128 in.)] at
flexural demands of MuW ¼ 37,638 kN · mð27;760 kips · ftÞ, M uB ¼ the top and five #8 bars [db ¼ 25.4 mm (1.0 in.)] at the bottom of
1,082 kN · mð798 kips · ftÞ, M uW ¼ 649 kN · m ð479 kips · ftÞ for a the beam [Fig. 2(c)], which was sufficient to resist the beam flexu-
wall, beams and columns, respectively. ral demand and satisfy the requirements of ACI 318-11 (ACI 2011;
A detailed assessment of the building lateral load resisting sys- S10.5 and S21.5.2). The strong column–weak beam provision of
tem was conducted according to ASCE 7-10 (ASCE 2013) and ACI ACI 318-11 (ACI 2011; S21.6.2) was checked at all floor levels.
318-11 (ACI 2011) provisions. A 6.10 m (20 ft) long and 0.30 m Beams and columns were assumed to satisfy the shear strength
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

(12 in.) thick structural wall, with 16 #11 bars [db ¼ 35.81 mm requirements of ACI 318-11 (ACI 2011; S21.5.4 and S21.6.5), re-
(1.410 in.)] located at each boundary [Fig. 2(a)], satisfied P-M spectively, and the detailing requirements of S21.5.3 and S21.6.4,
strength requirements (ϕMnW ¼ 41,081 kN · m > M uW ) and the respectively. The design of beam-column joints was performed ac-
“stress-based” approach [ACI 318-11 (ACI 2011; S21.9.6.3)] cording to ACI 318-11 (ACI 2011; S21.7) for an exterior connec-
was used to determine that special boundary elements satisfying tion, assuming that beams that frame into beam-column joints yield
S21.9.6.4 [ACI 318-11 (ACI 2011)] were required [Fig. 2(a)]. before the columns. Finally, building lateral displacements com-
Axial load at the base of the wall of 3,839 kN (863 kips) resulted puted according to ASCE 7-10 (ASCE 2013; S12.8.6) yielded
inter-story drifts of 0.0035, 0.0091, 0.0126, 0.0144, and 0.0151,
in axial load ratio of Pu =ðAg fc0 Þ ¼ 3,839,000 N=ð6,100 mm ×
which were less than the allowable inter-story drift of 0.02hsx =ðρ ¼
300 mm × 34.47 MPaÞ ¼ 0.06. Horizontal and vertical web rein-
1.3Þ ¼ 0.0154hsx [ASCE 7-10 (ASCE 2013; Table 12.12-1)].
forcement consisting of two layers of #5 bars [db ¼ 15.875 mm
(0.625 in.)] spaced at 457 mm (18 in.) satisfied the minimum hori-
zontal web reinforcing ratio according to ACI 318-11 (ACI 2011) Analytical Modeling
(ρl ¼ ρt ¼ 0.00284 > ρt;min ¼ 0.0025, S21.9.2.1) and provided
wall shear strength of ϕV n ¼ 3,011 kN (677 kips) according to Model Description
ACI 318-11 [ACI 2015; S21.9.4; Eqs. (21)–(7)], which was suffi- Analytical models of the archetype building lateral-load-resisting
cient to resist wall shear demand corresponding to 100% of seismic system are generated in OpenSees (McKenna et al. 2000) according
force (V u ¼ 5,338=2 ¼ 2,669 kN). Alternative design procedures to adopted geometry, cross sections, and material properties of
(e.g., capacity design approach) could lead to larger wall shear the structural elements. For the purposes of this study, a two-
capacity than the one obtained using the ACI code; however, the dimensional model is used; therefore, torsion is neglected and
primary objective and conclusions of the study depend on the me- symmetry is used such that the model consists of one frame and one
chanics of the modeling approaches used to assess nonlinear RC wall [Fig. 3(a)]. Given that the primary objective of this study is to
wall behavior (i.e., uncoupled versus coupled models) and are compare two conceptually different modeling approaches for RC
not sensitive to the adopted design methodology. walls, these simplifications in the building’s nonlinear model are

(a)

(b) (c)

Fig. 2. Structural element cross sections: (a) wall; (b) column; (c) beam

© ASCE G4016001-3 J. Struct. Eng.

J. Struct. Eng., G4016001


Element EIeff/EIg Mn (kN-m) a (%) b (%)
Elastic
Rigid Beams 0.20 106top/62bot 0.23 0.20
Beam/Column
Rigid diaphragms Col. L1 0.55 106 0.27 0.23
Beam Col. L2 0.42 106 0.27 0.23
Column
Col. L3 0.34 106 0.27 0.23
Hinges
Col. L4,5 0.30 106 0.27 0.23

Beam Column
Detail A 1.0 Hinge
Hinge

Rigid

M/Mn
Beam Hinges
Detail A
Node
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

SFI-MVLEM or
a b 0.2
Fiber Element EIeff
RC Wall RC Frame Chord Rotation, θ
(a) (b)

Fig. 3. Analytical model of archetype building: (a) wall-frame system; (b) plastic hinges

appropriate and will not affect the overall conclusions. The gravity element [B-C or fiber model (Taucer et al. 1991)] with uncoupled
system is not included in the model based on ASCE 7-10 (ASCE axial/flexural and shear behavior, both briefly described in fur-
2013) requirements, and the assumption of a rigid diaphragm is ther text.
implemented within each story level. Tributary mass is assigned
at the element nodes at each story level along axes of the wall Wall Models
and columns, whereas gravity load (dead and live) was assigned The analytical model proposed by Kolozvari et al. (2015a)
according to the corresponding tributary areas as either nodal load [Fig. 4(a)], called the SFI-MVLEM, incorporates biaxial con-
at wall-element nodes of each story or uniformly distributed load stitutive RC panel behavior (Ulugtekin 2010), described with the
along the beams of the frame. fixed-strut angle approach, into a two-dimensional macroscopic
RC frame elements (i.e., beams and columns) are modeled using fiber-based model formulation of the multiple-vertical-line-element
elastic beam-column elements by assuming the location of plastic model (MVLEM, Orakcal et al. 2004); axial-shear coupling is
hinges at the faces of beam-column joints [Fig. 3(b)], the behavior achieved at the panel (macro-fiber) level, which further allows cou-
of which was simulated using the moment-rotation hysteretic pling of axial/flexural and shear responses at the SFI-MVLEM
model (Lignos and Krawinkler 2011), with modeling parameters element level. Biaxial behavior of concrete within each RC panel
adopted according to beam and column flexural capacities and the element is described using a uniaxial stress-strain relationship for
ASCE 41 backbone relationships [Fig. 3(b)]. Stiffness modifiers for concrete (Chang and Mander 1994), applied along fixed compres-
elastic portions of beam and column elements were adopted accord- sion struts and parameters representing compression softening
ing to ASCE 41 [Table 6.5, ASCE 41-13 (ASCE 2014)] to account (Vecchio and Collins 1993), and also incorporates hysteretic biax-
for cracking. Given the objectives of the study, the behavior of RC ial damage (Mansour et al. 2002) and tension stiffening effects
walls was simulated using two conceptually different modeling (Belarbi and Hsu 1994). The implemented uniaxial constitutive
approaches: (1) SFI wall model (Kolozvari et al. 2015a) that in- relationship for reinforcing steel is the nonlinear hysteretic model
corporates interaction between axial-flexural and shear behavior of Menegotto and Pinto (1973). In addition, the RC panel model
under cyclic loading condition; and (2) nonlinear beam-column incorporates shear aggregate interlock effects along concrete cracks

(a) (b)

Fig. 4. RC wall modeling approaches: (a) SFI-MVLEM; (b) beam-column element

© ASCE G4016001-4 J. Struct. Eng.

J. Struct. Eng., G4016001


and reinforcement dowel action (Kolozvari 2013; Kolozvari et al. Table 1. Concrete Material Modeling Parameters
2015a). Parameter Boundary (confined) Web (unconfined)
In the nonlinear beam-column (B-C) element (Taucer et al.
Compression
1991), the flexural response is simulated by a series of uniaxial
fc0 (MPa) 53.12 34.47
elements (or macro-fibers) representing the wall cross section and εc0 0.005 0.002
vertical reinforcement along with the assumption that plane sec- Ec (MPa) 34,494 27,789
tions remain plane after loading [Fig. 4(b)]. The stiffness properties xcr− a 1.030 1.015
and force-displacement relationships of the fibers are defined ac- Rb 13 15
cording to uniaxial stress-strain relations for concrete and steel,
Tension
evaluated at integration points along the each uniaxial fiber and
ft (MPa) 1.82 1.82
corresponding areas of the materials in the RC wall cross section. εt 0.00008 0.00008
In this study, the uniaxial material model proposed by Chang and Ec (MPa) 27,789 27,789
Mander (1994) is used to simulate the behavior of concrete, xcrþ a ∞ ∞
whereas the uniaxial stress-strain law developed by Menegotto rb 1.2 1.2
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

and Pinto (1973) is used to describe the behavior of reinforcing a


Parameter defining the normalized strain (relative to εc0 =εt ) where the
steel. Shear behavior is uncoupled from the axial/flexural behavior, compression/tension envelope curve starts to follow a straight line until
and is simulated using a horizontal spring with a specified shear zero stress is reached (Chang and Mander 1994).
force–deformation (backbone) relation. The modeling parameters b
Parameter defining the shape of the compression or tension envelope curve
governing the shear behavior, including hysteretic shape (e.g., lin- (Tsai 1988).
ear, bilinear), stiffness and capacity, are typically adopted based on
available experimental data, and can vary significantly based on the
wall characteristics such as wall aspect ratio, amount of shear and of the monotonic tension envelope of the Chang and Mander
flexural reinforcement, and level of axial load. Therefore, the shear (1994) model was calibrated using the parameter r to reasonably
force-deformation relationships that are commonly used in practice represent the average postcrack stress-strain relation proposed by
are typically ad hoc. Belarbi and Hsu (1994), which represents the tension stiffening
As shown in Fig. 3(a), the RC wall is modeled using 10 equal- effects on concrete. The parameters used for calibrating the constit-
length elements along the building height (i.e., two elements per utive models for concrete are presented in Table 1.
story height) for both considered wall models. Wall discretization Shear Force–Deformation Relationship. Two different force–
in the horizontal direction was performed using six macro-fibers to deformation relationships are used in this study to represent the
represent the wall cross section, where two outer macro-fibers were shear behavior of the beam-column model: (1) linear-elastic and
used to represent the confined wall boundaries and the remaining (2) bilinear backbone curves; conceptual relationships and the
four represented the unconfined wall web. In addition, parameters modeling parameters are presented in Fig. 5. A relatively wide
of the material models for concrete and reinforcement are the same range of effective shear stiffness can be found in current provisions
for both modeling approaches to allow direct comparison of the and recent research. For example, ASCE 41-13 (ASCE 2014) sug-
results, as described in the following section. gests the use of bilinear or trilinear relationships with 1.0 G for
uncracked shear modulus; LATBSDC (2015) proposes the use
Material Calibration of 0.5 G as the effective shear modulus; and PEER/ATC 72p(PEER ffiffiffiffiffi
Reinforcing Steel. The reinforcing steel stress-strain relationship and ATC 2010) provides 0.05pGffiffiffiffiffi(wall shear stress of 5 fc0 ) or
described by the Menegotto and Pinto (1973) model was calibrated 0.1 G (wall shear stress of 10 fc0 ) for the secant shear modulus
to represent the typical properties of Grade 60 reinforcing bars corresponding to flexural yielding. In addition, based on recent
using a yield strength of 413.69 MPa (60,000 psi) and strain hard- tests results on moderately slender RC walls (Tran 2012), a shear
ening ratio of 0.02. The parameters describing the cyclic stiffness modulus of 0.25 G was found to be the most appropriate to re-
degradation characteristics of the reinforcing bars were calibrated present the wall effective shear stiffness, whereas validation against
as R0 ¼ 20, a1 ¼ 18.5, and a2 ¼ 0.15, as proposed by Menegotto tests on slender RC walls (Gogus 2010) revealed that the effective
and Pinto (1973). shear stiffness after concrete cracking of 0.025 G captures the tests
Concrete. The monotonic envelope of the stress-strain model results the most appropriately. Based on a previous discussion,
for unconfined concrete in compression proposed by Chang and three different values of shear stiffness are used (0.5, 0.2, and
Mander was calibrated to agree with the monotonic envelope pro- 0.1GAw ) for two considered backbone relationships (linear and
posed by Saatcioglu and Razvi (1992) by matching the compres-
sive strength fc0 , the strain at compressive strength εc0 , initial tangent
modulus Ec , and the parameter r defining the shape of the mon-
otonic stress-strain curve (Tsai 1988). The stress-strain envelopes
for confined concrete in compression were obtained by computing
0
the peak stress of confined concrete (fcc ) and the strain at peak
0 ) based on the area, configuration, spacing, and yield
stress (εcc
stress of the transverse reinforcement, using the confinement model
by Mander et al. (1988), whereas the initial tangent modulus for
confined concrete (Ecc ) and corresponding shape parameter (rc )
are obtained based on the peak stress of confined concrete (fcc 0 )

using empirical relations proposed by Chang and Mander (1994)


and Orakcal and Wallace (2006). The tensile strengthpffiffiffiffiffi of concrete
was determined from the relationship f t ¼ 0.31 f c0 ðMPaÞ, and a
value of 0.00008 was selected for the strain at the peak monotonic
Fig. 5. Considered force–deformation relations for shear behavior
tensile stress εt , as suggested by Belarbi and Hsu (1994). The shape

© ASCE G4016001-5 J. Struct. Eng.

J. Struct. Eng., G4016001


bilinear), shown in Fig. 5; the shear yielding in the bilinear relation- effective stiffness of 0.5GAw and shear capacity equal to the wall
ship was adopted based on wall nominal shear capacity according nominal shear capacity according to ACI 318-11 (ACI 2011)
to ACI 318-11 (ACI 2011) (V n;ACI ). Although the actual wall shear (V n;ACI ). Comparisons of analytical results are conducted at both
capacity could be larger than the capacity obtained from the ACI global and local response levels. The sensitivity studies presented
code (Wallace 1996; Orakcal et al. 2009), a commonly used value are obtained for a single selected ground motion with a response
of V n;ACI was adopted in this study to be consistent with the current spectrum that matches reasonably the design spectrum (Fig. 6);
design and evaluation practice; using a larger value (e.g., 1.5V n;ACI ) however, the same trends were observed for the remaining six
would not change the overall conclusions reported in this study. ground motions.

Building Fundamental Period


Ground Motion Records
Variations of the initial fundamental building period (T 1;start ) that
Nonlinear analysis of the archetype building was performed using were predicted by the B-C wall model for the range of effective
seven horizontal ground motions from sites located greater than shear stiffness of 1.0, 0.5, 0.2, 0.1, and 0.05GAw are presented
or equal to 10 km from fault rupture (soft rock/stiff soil, strike- in Table 2. The initial fundamental building period obtained using
slip, and thrust mechanism); vertical earthquake components are the uncracked shear stiffness (i.e., GAeff ¼ 1.0GAw ) is equal to
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

not considered important for evaluation. Ground motions in- 0.54 s, which is in agreement with the initial fundamental period
cluded in the set are selected from the PEER National Geospatial- predicted by the SFI model that corresponds to uncracked RC panel
Intelligence Agency (NGA) database, and have peak ground macro-fibers within the model elements. However, reduction of the
accelerations (PGAs), peak ground velocities (PGVs), and magni- effective shear stiffness from 0.5 to 0.05GAw increases the initial
tudes greater than 0.20 g, 15 cm=s, and 6.5, respectively; records fundamental period from 0.57 to 0.98 s (by approximately 70%),
were used on previous ATC 63 (ATC 2009) studies. Individual and revealing considerable sensitivity of the fiber model to the adopted
mean response spectra of the seven ground motion records used are value of effective shear stiffness, which could significantly affect
provided in Fig. 6 and compared with the adopted code design the reliability of analytical predictions obtained using fiber models,
spectrum; as observed from the figure, the mean of seven response given the relatively wide range of effective shear stiffness available
spectra match reasonably well the code design spectrum. in current modeling recommendations. Table 2 also presents the
average of seven fundamental periods corresponding to end of the
time-history analyses (T 1;end ), which indicates that the fundamental
Analysis Results period shifts for approximately the same amount, of approximately
0.20 s, for all considered models.
Detailed comparisons of analytical results obtained for the arche-
type building are provided in the following sections, including Response Histories of Roof Displacement and Base Shear
sensitivity studies of the B-C model results to the adopted shear Force
force–deformation relationship and their comparisons against the The comparison of wall roof displacement and base shear force
results obtained using the SFI-MVLEM for a single ground motion, histories are presented in Fig. 7 for the wall models considered.
in addition to a comparison of the average responses for seven It can be observed from Figs. 7(a and b) that the roof displacement
ground motions. and base shear force histories predicted using the fiber model with
linear-elastic shear relationship and effective stiffness of 0.5GAw
Sensitivity of Beam-Column Model to Shear match reasonably well the responses predicted by the SFI model,
Force–Deformation Relationship although approximately 30% larger maximum shear force at 5.5 s
Sensitivity of analytical predictions obtained using the nonlinear was obtained using the fiber modeling approach. Figs. 7(a and b)
B-C model to shear force–deformation relationship is presented, further reveal that the roof displacement and base shear force
and the results are compared with the predictions obtained using histories predicted by the fiber model with elasto-plastic shear re-
the SFI-MVLEM. Two force–deformation relationships are con- lationship (Fig. 5) agree well with the responses predicted by the
sidered to represent the wall shear behavior in the B-C model: SFI model up to approximately 5.0 s, when the base shear force in
(1) linear-elastic relation with effective shear stiffness (GAeff ) equal the wall reached the adopted wall shear capacity V n;ACI [4,017 kN
to 0.5, 0.2, and 0.1GAw ; and (2) elasto-plastic relation with initial (903 kips)]. After the wall shear capacity is reached, the predicted
wall response becomes numerically unstable, suggesting shear fail-
ure within the first story, as the shear stiffness within the model
element reduced abruptly from the initial value to a post-yield value
(Fig. 5); the node located at the mid-height of the first story expe-
rienced unreasonably large displacements as a result of pure-shear
deformations. To avoid the described numerical instabilities, and to

Table 2. Sensitivity of Beam-Column Model Fundamental Periods to


Shear Stiffness
Model GAeff =GAw T 1;start (s) T 1;end (s)
Beam-column 0.05 0.98 —
0.10 0.78 1.02
0.20 0.66 0.84
0.50 0.57 0.80
1.00 0.54 —
SFI —a 0.53 0.75
Fig. 6. Ground motion record set response spectra and design spectrum a
Shear stiffness degrades throughout the analysis.

© ASCE G4016001-6 J. Struct. Eng.

J. Struct. Eng., G4016001


Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Time histories of wall responses: (a) top displacement and (b) base shear force for SFI model and beam-column model with linear elastic
(GAeff ¼ 0.5GAw ) and elasto-plastic (GAeff ¼ 0.5GAw ) shear behavior; and (c) top displacement and (d) base shear force for beam-column model
with linear elastic (GAeff ¼ 0.1 and 0.2GAw ) and elasto-plastic (GAeff ¼ 0.5GAw , one element per level) shear behavior

allow further investigation of the fiber model with elasto-plastic the model with shear stiffness of 0.5w or 0.2GAw to 4,000 kN
shear relationship, an alternative analytical model of the building (900 kips) for the model with 0.1GAw .
with one wall model element per level was created. It can be
observed from Figs. 7(c and d) that this modification resulted in Vertical Profiles of Wall Deformation Characteristics
stable roof displacement and base shear force responses that match Vertical profiles of wall maximum drifts, lateral displacements,
closely the responses obtained using the SFI model and fiber and contributions of shear deformations to lateral displacements
model with a linear-elastic shear relationship of GAeff ¼ 0.5GAw are presented in Fig. 8. It can be observed from the figure that
[Figs. 7(a and b)]. Furthermore, Figs. 7(b and d) reveal that the profiles of interstory drifts [Fig. 8(a)] and lateral deformations
trend of roof displacement time history obtained using the reduced [Fig. 8(b)] predicted by the SFI wall model and fiber model with
values of effective shear stiffness of 0.2 and 0.1GAw diverge from effective shear stiffness of 0.5GAw (linear-elastic and elasto-plastic
the two previously considered cases; the reduction of effective shear relationship) are in good agreement over the building height,
shear stiffness from 0.5 to 0.1GAw increases the maximum roof except within the first level where the SFI model predicts approx-
displacement by approximately 40%, from 100 mm (3.9 in.) to imately 50% larger lateral displacements and interstory drifts.
140 mm (5.5 in.). Finally, the comparisons of wall base shear force Figs. 8(a and b) also show that a reduction in the effective shear
histories show similar trends for the three cases of fiber models with stiffness in the fiber (B-C) model from 0.5 to 0.1GAw resulted
linear-elastic shear stiffness, although considerable sensitivity of in improved predictions of interstory drifts and lateral deformations
peak shear force can be observed relative to the adopted effective within the first level relative to the SFI model results. However,
shear stiffness; maximum base shear occurs at 5.5 s for all fiber deformations at levels 2–5 are now overestimated by approximately
models and varies from approximately 6,200 kN (1,400 kips) for 50%. Furthermore, Fig. 8(c) illustrates that contributions of shear

Fig. 8. Vertical profiles of maximum wall responses for selected ground motion: (a) interstory drifts; (b) lateral displacements; (c) contributions of
shear deformation

© ASCE G4016001-7 J. Struct. Eng.

J. Struct. Eng., G4016001


Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. First-level wall responses for selected ground motion: (a) shear force versus total lateral displacement; (b) shear force versus shear displace-
ment; (c) moment versus curvature

deformations to total lateral displacement over the wall height are the effective shear stiffness of 0.5GAw . The occurrence of nonlinear
significant within the first level for all considered models. The SFI shear deformations in the SFI-MVLEM results does not necessarily
model predicts the largest contribution of shear deformations of suggest shear failure, as the mechanics of the analytical model cap-
73% within the bottom wall element, whereas the corresponding ture coupled nonlinear shear and flexural behavior. Furthermore,
contributions of shear deformations obtained from fiber model the B-C models with effective shear stiffnesses of 0.2 and 0.1GAw
with linear-elastic shear relationships are 23, 41, and 58% for ef- predict approximately 30 and 60% of the maximal shear deforma-
fective shear stiffness values of 0.5, 0.2, and 0.1GAw , respectively. tions obtained using the SFI model, respectively, suggesting that
Finally, the fiber model with elasto-plastic shear backbone curve an even lower value of effective shear stiffness (e.g., 0.05GAw )
and initial effective shear stiffness of 0.5GAw predicts approxi- would be more appropriate to account for the nonlinear shear de-
mately two times larger contributions of shear deformations over formations obtained using the SFI model. Finally, the B-C model
the first story, compared with the corresponding linear elastic with elasto-plastic shear backbone relation (with one element per
case (i.e., 43 versus 23%), but 40% smaller shear contributions level) predicts the maximal nonlinear shear deformations that are
compared with the SFI model. approximately 30% smaller than the shear deformations obtained
using the SFI-MVLEM, but captures cyclic shear stiffness degra-
Wall Responses within the First Story dation, although the shape of the hysteretic loops could be im-
To investigate the predicted wall behavior in the region where proved by calibration of material parameters used to represent the
nonlinear deformations are expected, the force–deformation and cyclic shear behavior.
the moment–curvature responses over the first story height are pre- Moment versus curvature relations plotted in Fig. 9(c) reveal
sented in Fig. 9. It can be observed from the shear force versus first that the moment yield capacity predicted by all considered
story lateral displacement responses [Fig. 9(a)] that the B-C wall modeling approaches is the same and equal to approximately
models with linear-elastic shear behavior and effective shear stiff- 50; 000 kN · m ð36; 900 kips · ftÞ, which is slightly larger than
ness of 0.5 and 0.2GAw predict wall demands that exceed V n;ACI by the nominal moment capacity obtained using the section analysis
approximately 50%, whereas the first story maximum shear force of 45; 284 kN · m ð33; 400 kips · ftÞ, whereas curvatures over the
of the B-C model with effective shear stiffness of 0.1GAw and the first story are significantly different. Fiber models with linear-elastic
SFI-MVLEM are in good agreement and approximately 10% larger shear behavior predict the largest maximum curvature, whereas com-
than V n;ACI . It can be also observed from Fig. 9(a) that using puted curvature is approximately two times smaller when the fiber
elasto-plastic shear backbone relation in the B-C model resulted model with elasto-plastic shear behavior is used as a result of pure-
in hysteretic loops that are characterized with an abrupt change of shear deformations over the first level after the shear capacity is
stiffness after the adopted shear capacity (V n;ACI ) is reached, reached [Fig. 9(b)]. The maximum curvature predicted by the SFI
whereas the overall stiffness and lateral deformations are in reason- model is in between the two extreme cases of fiber models (linear-
ably good agreement with the SFI-MVLEM results. As expected, elastic and elasto-plastic shear behavior) as a result of coupled non-
the B-C model with an effective shear stiffness of 0.1GAw pro- linear shear and flexural deformations at the model element level.
duced the largest first-story lateral displacements from the three
cases of linear-elastic shear stiffness [22 mm (0.87 in.)], which Vertical Profiles of Shear Force and Bending Moment
are slightly smaller than the lateral deformations predicted by the Fig. 10 plots the distribution of maximum shear force and bending
B-C model with elasto-plastic shear backbone relation [25 mm moment over the height of the wall for selected ground motion
(0.99 in.)] and the SFI model [28 mm (1.10 in.)]. record obtained using the SFI and B-C wall models. It can be ob-
Shear force versus shear deformation response [Fig. 9(b)] served from Fig. 10(a) that the B-C model with a linear elastic shear
further reveals that the total shear deformation predicted by the SFI- stiffness of 0.5GAw predicts generally higher shear force demands
MVLEM over the first story is 14 mm (0.55 in.) and 12 mm over the entire height of the wall compared with the SFI model, in
(0.48 in.) in the positive and negative directions, respectively. Shear which a significant difference can be observed within the first level
deformation at shear yielding is approximately 2.0 mm (0.08 in., where the predicted shear force is approximately 30% larger. It can
shear strain of 0.0006), which agrees well with the shear deforma- be also observed from Fig. 10(a) that the fiber model with a reduced
tion corresponding to V n;ACI predicted by the B-C wall model with effective shear stiffness of 0.1GAw provides predictions of the shear

© ASCE G4016001-8 J. Struct. Eng.

J. Struct. Eng., G4016001


Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Vertical profiles of maximum wall responses for selected ground motion: (a) shear force; (b) overturning moment

force over the wall height, including the first level, which is in very Wall Vertical Strains and Rotations
good agreement with the shear force predicted by the SFI model. Fig. 11 illustrates the profiles of maximal tensile and compressive
Furthermore, Fig. 10(a) reveals that the shear force over the first vertical axial strains and maximum rotations computed over the
and the second levels predicted by the B-C model with elasto- wall height. Fig. 11(a) shows that the tensile strains over the first
plastic shear relationship reached the prescribed wall shear capacity model element (i.e., bottom 1.82 m or 6.0 ft of the wall) are sig-
of 4,017 kN (903 kips, V n;ACI ), whereas the B-C models with a nificantly larger than strains in the second model element for all
linear-elastic shear relationship and effective shear stiffness of 0.5 considered models, indicating that the plastic hinge length of the
and 0.2GAw predicted shear force that exceeds 1.5V n;ACI , a value wall predicted by the adopted models is approximately lw =3. Using
that is commonly considered to represent the mean shear strength more elements over the first story might slightly change the distri-
of shear-controlled walls (Wallace 1996; Orakcal et al. 2009). bution of strains over the height; however, the plastic hinge length
The distributions of bending moment over the height of the wall would still be within lw =2, which is typically considered as a plas-
presented in Fig. 10(b) illustrate that predicted moment demand is tic hinge length for a wall with well-confined boundaries. The SFI
significantly less sensitive to the choice of modeling approach than model predicts a maximum tensile strain of 1.0%, whereas the B-C
the shear force demand. Absolute maximum moment at the base models with a linear-elastic shear relationship predict maximum
of the wall of approximately 54; 000 kN · m ð39; 850 kips · ftÞ is tensile strains of approximately 1.25% (25% larger) on average of
predicted by all considered modeling approaches, which is approx- the three considered values of effective shear stiffness. Fig. 11(a)
imately 15% larger than the nominal flexural capacity of the wall also reveals that the maximum vertical tensile strain predicted by
obtained from section analysis, suggesting that flexural yielding the B-C model with elasto-plastic shear behavior and one element
occurred for all models as shown in Fig. 9(c). over the story height is only 0.60%, which corresponds to the

Fig. 11. Vertical profiles of maximum wall responses for selected ground motion: (a) boundary compressive/tensile vertical strains; (b) rotations

© ASCE G4016001-9 J. Struct. Eng.

J. Struct. Eng., G4016001


average strain over the first level obtained from the corresponding Comparison of Results for a Suite of Seven Ground
model with linear-elastic shear relationship, indicating that sparse Motions
discretization might not capture appropriately the plastic hinge
Comparisons of predicted wall behavior are further conducted for
length of the wall and could significantly underestimate the com-
the average responses obtained from the suite of seven ground
puted strain demands. Maximum compressive strains are approx-
motions using the SFI model, B-C model with linear-elastic
imately 0.15% for all models, and no significant variations are
(GAeff ¼ 0.5GAw ) shear behavior, and B-C model with elasto-
observed in the presented results, likely because of the relatively
plastic shear behavior (GAeff;initial ¼ 0.5GAw , V y ¼ V n;ACI ) and
slender wall used; the effect of the modeling approach on concrete
discretization of only one element per level.
compressive strain is expected to be more significant for lower
Vertical profiles of maximum interstory drifts, lateral deforma-
aspect ratio walls (Orakcal and Wallace 2006).
tion profiles, and contributions of shear deformations are presented
Similar trends can be observed for distributions of rotations over
in Fig. 12. Fig. 12(a) reveals that magnitudes of interstory drifts
the wall height [Fig. 11(b)]. Maximum and minimum rotations of
predicted using the SFI-MVLEM and the B-C model are consid-
0.3 and −0.4% are predicted by the SFI model, whereas the B-C
models with linear-elastic shear behavior predict approximately erably different within the first story, where the flexural yielding is
reported. Maximum average interstory drifts within the first level as
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

25% larger wall rotations in both positive and negative directions.


Although the B-C model with elasto-plastic shear relationship predicted by the SFI model are approximately 0.56%, whereas the
predicts maximum wall rotations that match reasonably well the B-C model with linear-elastic shear behavior predicts the maximum
rotations obtained using linear-elastic B-C models, the distribution average interstory drifts of 0.34%, which is approximately 40% less
of rotations (i.e., plastic hinge length) is not accurately captured than the drifts obtained using the SFI model. In addition, interstory
because of sparse model discretization of one element per building drifts at the top story level predicted by the B-C model are approx-
level. imately 10% larger than the drifts obtained using the SFI model.
Overall, the sensitivity studies presented in this section reveal The maximum average interstory drifts within the first level ob-
that global responses (e.g., roof displacement, interstory drifts, tained using the B-C model with elasto-plastic shear behavior are
shear force) predicted using the B-C wall model are considerably approximately 0.45%, which is larger than the drifts computed
sensitive to the adopted relationship used to represent wall shear using the linear-elastic B-C model, but still approximately 20%
behavior, whereas the sensitivity of local responses (e.g., strains smaller than the drifts predicted by the SFI model. The SFI wall
and rotations) is not that significant. Reduction of effective shear model generally predicts larger drifts (and smaller shear force de-
stiffness from 0.5 to 0.1GAw increased the computed interstory mands) within the plastic hinge region (i.e., at the location of wall
drifts for approximately 50%, whereas the computed base shear yielding) as a result of the interaction between nonlinear flexural
force varied by approximately 30%, depending on the adopted ef- and shear deformations and cyclic degradation of shear stiffness, as
fective shear stiffness. In addition, the use of elasto-plastic shear captured at the model element level (Fig. 9), as opposed to the fiber
relation resulted in either unstable response at mid-level nodes modeling approach in which this interaction/degradation is not
(using two wall elements per story) or underestimated shear force captured. Fig. 12(b) plots the lateral deformation profiles for the
and vertical strains (using one wall element per story). Finally, the three modeling approaches, illustrating that larger lateral deforma-
predicted wall responses within the plastic hinge region (i.e., first tions over the first story are predicted using the SFI wall model,
building level) using the B-C model with an effective shear stiffness which is consistent with the maximum lateral drift profiles
of 0.1GAw are in good agreement with the predictions of SFI- presented in Fig. 12(a), although the three profiles are generally
MVLEM in terms of lateral deformation profiles, interstory drifts, similar in shape and the amount of maximum lateral displacement
base shear force, and bending moment, whereas the best match predicted at the roof level. Fig. 12(c) further reveals that the
among these responses at the remaining building levels are ob- nonlinear shear deformations predicted by the SFI-MVLEM
tained using the effective shear stiffness of 0.5GAw . are significant within the first level, where nonlinear flexural

Fig. 12. Average vertical profiles of maximum wall responses: (a) interstory drifts; (b) lateral displacements; (c) contributions of shear deformation

© ASCE G4016001-10 J. Struct. Eng.

J. Struct. Eng., G4016001


Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Average vertical profiles of maximum wall responses: (a) shear force; (b) overturning moment

Fig. 14. Average vertical profiles of maximum wall responses: (a) boundary compressive/tensile vertical strains; (b) rotations

deformations are reported, by contributing approximately 70% to within the top three levels of the building, whereas significant
the total lateral deformation of the bottom wall element and approx- differences in predicted shear force demand exist within the first
imately 45% (on the average of two wall elements) over the first level. The average shear force demand within the bottom level pre-
level. This is in close agreement with the contributions of shear dicted using the B-C model with linear-elastic shear behavior
deformations predicted by the B-C model with an elasto-plastic [5,811 kN (1,306 kips)] is 31% larger than the shear force obtained
shear backbone relation over the first story height, although this using the SFI model [4,423 kN (994 kips)] and approximately 45%
model is not capable of producing refined responses because of larger than the maximum shear force developed in the B-C model
its sparse discretization. The B-C model with a linear-elastic shear with a prescribed shear capacity of V n;ACI [4,017 kN (903 kips)].
force-deformation relation predicts the contribution of shear defor- All three considered models predicted almost identical average mo-
mations of only 41% over the bottom wall element and approxi- ment distributions over the wall height [Fig. 13(b)], with the wall
mately 20% over the first story level (average of two elements), maximum moment at the base of 49;258 kN · mð36;330 kips · ftÞ,
given the inability of the modeling approach to capture the nonlin- which is approximately 10% larger than the wall nominal flexural
ear shear deformations and shear stiffness degradation. capacity of 45; 284 kN · m ð33; 400 kips · ftÞ as a result of strain
Distributions of average maximum shear force [Fig. 13(a)] and hardening.
bending moment [Fig. 13(b)] over the height of the wall obtained Comparisons of the average maximum vertical strains and the
using the three considered cases are further explained. It can be wall rotations over the building height are presented in Fig. 14. The
observed from Fig. 13(a) that shear force distributions that are com- distribution of vertical strains [Fig. 14(a)] predicted by the SFI
puted using the three considered models are in good agreement and fiber modeling approach are similar in terms of distribution

© ASCE G4016001-11 J. Struct. Eng.

J. Struct. Eng., G4016001


Table 3. Comparisons of Absolute Maximum Wall Responses
SFI-MVLEM Beam-column model
Standard deviation COV ¼ μ=σ Standard deviation COV ¼ μ=σ
Response Mean (μ) (σ) (%) Mean (μ) (σ) (%) μBC =μSFI
Max drift (%) 0.56 0.30 53 0.34 0.30 49 0.61
Shear force (kN) 4,423 541 12 5,811 944 16 1.31
Moment (kN · m) 48,408 3,579 7 50,109 4,238 9 1.04
Min strain (%) −0.07 0.01 9 −0.08 0.01 8 1.11
Max strain (%) 1.00 0.48 48 1.10 0.60 55 1.10
Rotation (%) 0.29 0.16 55 0.33 0.19 59 1.11
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

of strains over the height, whereas the fiber model predicts approx- beam-column model (fiber model) with uncoupled axial/flexural
imately 10% larger tensile and compressive vertical strains. The and shear behavior. Shear behavior in the fiber-based modeling
corresponding wall rotations [Fig. 14(b)] obtained from the vertical approach was represented with either a linear-elastic relationship
strains at the wall boundaries follow similar trends as the vertical with effective shear stiffness of 0.5, 0.2, and 0.1GAw or elasto-
strains, with rotations predicted by the fiber model that are approx- plastic (bilinear) shear backbone relationship with effective shear
imately 10% larger than the rotations predicted by the SFI model at stiffness of 0.5GAw and shear capacity calculated according to
the base of the wall. Based on results presented in Fig. 14, the an- ACI 318-11 (ACI 2011) (V n;ACI ). A five-story archetype RC build-
alyzed wall experienced moderate (average) strain/rotation ductility ing with a lateral-force-resisting system consisted of RC walls and
demands of approximately 5.5 under the suite of seven selected special moment frames that resisted 25% of the seismic load
ground motions. Although using stronger ground motions could (i.e., dual system) was designed to satisfy the current U.S. code
lead to generally larger ductility demands in the wall, the overall provisions [i.e., ACI 318-11 (ACI 2011) and ASCE 7-10 (ASCE
conclusions of this study would not change as they are primarily 2013)]; the wall shear strength corresponding to the minimum
based on the mechanics of the modeling approaches used, and not reinforcing ratio according to ACI 318-11 (ACI 2011) was suffi-
by the severity of the ground motions. cient to resist the shear force demand corresponding to 100% of
Analytically predicted wall responses over the first level ob- seismic code-level force. Nonlinear response-history analyses were
tained using the B-C model with a linear-elastic shear stiffness performed using a set of seven far-field ground motion records
of 0.5GAw and SFI-MVLEM are summarized in Table 3. The most from soft rock/stiff soil sites with peak ground accelerations larger
significant discrepancies among the results computed by the un- than 0.25 g. The main conclusions of the paper are direct conse-
coupled and coupled wall models can be observed for interstory quences of conceptually different formulations of coupled and un-
drifts and shear force. The average maximum interstory drift pre- coupled modeling approaches, and are not affected by the design
dicted by the B-C model is approximately 60% of the interstory approach used (e.g., displacement-based, capacity-based) and/or
drift predicted by the SFI model, suggesting that the use of un- severity of ground motions.
coupled approaches might significantly underestimate the level of The sensitivity of analytical responses to variations of effective
damage in structural elements of a lateral-load-resisting (or gravity) shear stiffness used for the nonlinear beam-column model revealed
system within the plastic hinge region in comparison with the SFI that both global and local responses are considerably sensitive to
model. In contrast, the uncoupled wall model imposes approxi- the adopted relationship used to represent wall shear behavior.
mately 30% higher shear force demand on the wall than the SFI It has been observed that, with a reduction of effective shear stiff-
model, which could result in differences in distribution of total ness from 0.5 to 0.1GAw , the initial building period increased
shear force within the elements of the lateral-load-resisting system. approximately 37% (from 0.57 to 0.78 s), the predicted lateral dis-
Furthermore, tensile or compressive strains and rotations obtained placements increased approximately 40%, and the interstory drifts
using the B-C model are generally 10% larger than the responses increased approximately 50%. In addition, the nonlinear beam-
predicted by SFI-MVLEM, whereas predictions of the moment at column model with elasto-plastic shear force-deformation rela-
the wall base using the two modeling approaches are very similar. tionship was shown to be numerically unstable during OpenSees
Table 3 further reveals that the largest variation in predicted wall analyses at nodes that are not connected to beams of the special
responses are obtained for interstory drifts, tensile strains and ro- moment frame. The instability resulted after the prescribed wall
tation, where the coefficient of variation (COV) is approximately shear capacity defined by the backbone relation was reached, re-
50%, moderate variation is observed for the shear force (COV ¼ sulting from the pure-shear deformation of the wall elements,
12%), whereas the least variation of responses for both models which led to unreasonable analytical predictions of building re-
can be observed for the compressive strains and bending moment sponses. Therefore, only one element over the story height for the
(COV < 10%). bilinear shear relation was considered. The results further revealed
that the predicted wall responses using the beam-column model
with effective shear stiffness of 0.5GAw (linear and bilinear shear
Summary and Conclusions behavior with one element per story) are in good agreement with
the predictions obtained with SFI-MVLEM in terms of vertical pro-
This paper presents the results of a detailed assessment of analyti- files of maximum interstory drift, lateral deformation, and shear
cally predicted RC wall behavior of an archetype building obtained force, except within the plastic hinge region (i.e., first level), where
using two conceptually different RC wall modeling approaches the best match among first level responses are obtained using the
available in OpenSees: (1) SFI-MVLEM, which captures the effective shear stiffness of 0.1GAw .
interaction among axial/flexural and shear responses and nonlin- A comparison of maximum average responses obtained for a
ear shear deformations (coupled model); and (2) a nonlinear set of seven ground motions revealed that the beam-column wall

© ASCE G4016001-12 J. Struct. Eng.

J. Struct. Eng., G4016001


model with an effective shear stiffness of 0.5GAw (typically used in Henry, R. S., and Lu, Y. (2014). “Research to address the performance of
practice) generally predicts the maximum shear force within the lightly reinforced concrete walls during the 2010/2011 Canterbury
plastic hinge region of the wall (i.e., within the first story), which earthquakes.” 2nd European Conf. on Earthquake Engineering and
is approximately 30% larger than the maximum shear force pre- Seismology, Istanbul, Turkey.
dicted by the SFI modeling approach. In contrast, the interstory Kolozvari, K. (2013). “Analytical modeling of cyclic shear-flexure interac-
tion in reinforced concrete structural walls.” Ph.D. dissertation, Univ. of
drift and contributions of the shear deformation within the first level
California, Los Angeles.
predicted by the uncoupled model are approximately 40% lower in
Kolozvari, K., Orakcal, K., and Wallace, J. W. (2015a). “Modeling of cyclic
comparison with the coupled modeling approach. The analyzed shear-flexure interaction in reinforced concrete structural walls. Part I:
wall experienced a moderate (average) strain ductility demand of Theory.” J. Struct. Eng., 10.1061/(ASCE)ST.1943-541X.0001059,
5.5 according to all considered modeling approaches, whereas the 04014135.
beam-column model predicted tensile vertical strains and rota- Kolozvari, K., Tran, T., Orakcal, K., and Wallace, J. W. (2015b). “Modeling
tions that were generally 10% larger than the responses predicted of cyclic shear-flexure interaction in reinforced concrete structural
using the SFI-MVLEM. The results for both the single ground walls. Part II: Experimental validation.” J. Struct. Eng., 10.1061/
motion and the suite of seven ground motions suggest strongly (ASCE)ST.1943-541X.0001083, 04014136.
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

that the use of a single model to determine the responses to evalu- LATBSDC (Los Angeles Tall buildings Structural Design Council).
ate local and global force and deformation demands is biased, (2015). “An alternative procedure for seismic analysis and design
based on the assumed wall shear stiffness in the uncoupled modeling of tall buildings located in the Los Angeles region.” Los Angeles.
approaches. Lignos, D. G., and Krawinkler, H. (2011). “Deterioration modeling of steel
Future studies could focus on extensive investigation of wall components in support of collapse prediction of steel moment frames
under earthquake loading.” J. Struct. Eng., 10.1061/(ASCE)ST.1943
responses for a large set of ground motion records to verify the
-541X.0000376, 1291–1302.
trends observed in this study, and conduct further reliability studies
Mander, J. B., Priestley, M. J. N., and Park, R. (1988). “Theoretical stress-
using the considered modeling approaches. Future work could also strain model for confined concrete.” J. Struct. Eng., 10.1061/(ASCE)
focus on the investigation of wall responses for a range of building 0733-9445(1988)114:8(1804), 1804–1826.
heights (e.g., 3, 8, 12, 15 stories) and different relative strengths of Mansour, M. Y., Hsu, T. C., and Lee, J. Y. (2002). “Pinching effect in
walls and frames (e.g., frames that resists 10 and 50% of seismic hysteretic loops of R/C shear elements.” ACI, 205, 293–321.
load), to investigate the sensitivity of the analytical predictions over Massone, L. M., Orakcal, K., and Wallace, J. W. (2006). “Shear–flexure
a wider range of building configurations for uncoupled and coupled interaction for structural walls.” ACI SP-236-07, ACI (American
wall modeling approaches. Comparison studies could also be per- Concrete Institute), Farmington Hills, MI.
formed for coupled wall systems with various coupling beam McKenna, F., Fenves, G. L., Scott, M. H., and Jeremic, B. (2000). “Open
strengths and configurations, and for tall buildings (e.g., 40 stories) system for earthquake engineering simulation (OpenSees).” Pacific
in which the discrepancies among the model predictions could be Earthquake Engineering Research Center, Univ. of California, Berkeley,
more significant as a result of a larger contribution of higher modes CA.
to building seismic behavior. Menegotto, M., and Pinto, E. (1973). “Method of analysis for cyclically
loaded reinforced concrete plane frames including changes in geometry
and non-elastic behavior of elements under combined normal force and
bending.” Proc., IABSE Symp. on Resistance and Ultimate Deformabil-
References ity of Structures Acted on by Well-Defined Repeated Loads, Lisbon,
Portugal.
ACI (American Concrete Institute). (2011). “Building code requirements Orakcal, K., Conte, J. P., and Wallace, J. W. (2004). “Flexural modeling of
for structural concrete.” ACI 318-11. Farmington Hills, MI. reinforced concrete structural walls—Model attributes.” ACI Struct. J.,
ASCE. (2013). “Minimum design loads for buildings and other structures.” 101(5), 688–698.
ASCE 7-10, Reston, VA. Orakcal, K., Massone, L. M., and Wallace, J. W. (2009). “Shear strength
ASCE. (2014). “Seismic evaluation and retrofit of existing buildings.” of lightly reinforced wall piers and spandrels.” ACI Struct. J., 106(5),
ASCE 41-13, Reston, VA. 455–465.
ATC (Applied Technology Council). (2009). “Quantification of build- Orakcal, K., and Wallace, J. W. (2006). “Flexural modeling of reinforced
ing seismic performance factors.” Rep. No. ATC-63, Redwood concrete walls—Experimental verification.” ACI Struct. J., 103(2),
City, CA. 196–206.
Belarbi, A., and Hsu, T. C. (1994). “Constitutive laws of concrete in tension Panagiotou, M., Restrepo, J., Schoettler, M., and Geonwoo, K. (2011).
and reinforcing bars stiffened by concrete.” ACI Struct. J., 91(4), “Nonlinear cyclic model for reinforced concrete walls.” ACI Struct.
465–474. Mater. J., 109(2), 205–214.
Belletti, B., Damoni, C., and Gasperi, A. (2013). “Modeling approaches PEER (Pacific Earthquake Engineering Research Center) and ATC
suitable for pushover analyses of RC structural wall buildings.” (Applied Technology Council). (2010). “Modeling and acceptance
Eng. Struct., 57, 327–338. criteria for seismic design and analysis of tall buildings.” PEER/ATC
Chang, G. A., and Mander, J. B. (1994). “Seismic energy based fatigue 72-1), Richmond, CA.
damage analysis of bridge columns. Part I: Evaluation of seismic PERFORM 3D [Computer software]. Computers and Structures, Walnut
capacity.” NCEER Technical Rep. No. NCEER-94-0006, State Univ. Creek, CA.
of New York, Buffalo, NY. Petrangeli, M., Pinto, P. E., and Ciampi, V. (1999). “Fiber element for
Dashti, F., Dhakal, R. P., and Pampanini, S. (2014). “Simulation of out- cyclic bending and shear of RC structures. I: Theory.” J. Eng. Mech.,
of-plane instability in rectangular RC structural walls.” 2nd Euro- 10.1061/(ASCE)0733-9399(1999)125:9(994), 994–1001.
pean Conf. on Earthquake Engineering and Seismology, Istanbul, Saatcioglu, M., and Razvi, S. R. (1992). “Strength and ductility of con-
Turkey. fined concrete.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(1992)118:
Fischinger, M., Rejec, K., and Isaković, T. (2012). “Modeling inelastic 6(1590), 1590–1607.
shear response of RC walls.” Proc., 15th World Conf. on Earthquake Taucer, F. F., Spacone, E., and Filippou, F. C. (1991). “A fiber beam-
Engineering, Lisbon, Portugal. column element for seismic response analysis of reinforced concrete
Gogus, A. (2010). “Structural wall systems—Nonlinear modeling and structures.” Rep. No. UCB/EERC-91/17, Earthquake Engineering
collapse assessment of shear walls and slab-column frames.” Ph.D. Research Center, College of Engineering, Univ. of California,
dissertation, Univ. of California, Los Angeles. Berkeley, CA.

© ASCE G4016001-13 J. Struct. Eng.

J. Struct. Eng., G4016001


Tran, T. A. (2012). “Experimental and analytical studies of moderate Ulugtekin, D. (2010). “Analytical modeling of reinforced concrete panel
aspect ratio reinforced concrete structural walls.” Ph.D. dissertation, elements under reversed cyclic loadings.” M.S. thesis, Dept. of Civil
Dept. of Civil and Environmental Engineering, Univ. of California, Engineering, Bogazici Univ., Istanbul, Turkey.
Los Angeles. Vecchio, F. J., and Collins, M. P. (1993). “Compression response of cracked
Tsai, W. T. (1988). “Uniaxial compressional stress-strain relation of reinforced concrete.” J. Struct. Eng., 83(2), 219–231.
concrete.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(1988)114:9(2133), Wallace, J. W. (1996). “Evaluation of UBC-94 provisions for seismic
2133–2136. design of RC structural walls.” Earthquake Spectra, 12(2), 327–348.
Downloaded from ascelibrary.org by Auburn University on 02/28/16. Copyright ASCE. For personal use only; all rights reserved.

© ASCE G4016001-14 J. Struct. Eng.

J. Struct. Eng., G4016001

You might also like