Engineering Structures 43 (2012) 105119

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Performance limits for structural walls: An analytical perspective

_
Ilker
Kazaz a, Polat Glkan b,, Ahmet Yakut c
a

Department of Civil Engineering, Atatrk University, 25240 Erzurum, Turkey

Department of Civil Engineering, ankaya University, 06810 Ankara, Turkey
c
Department of Civil Engineering, Middle East Technical University, 06800 Ankara, Turkey
b

a r t i c l e

i n f o

Article history:
Received 4 July 2011
Revised 13 May 2012
Accepted 14 May 2012
Available online 18 June 2012
Keywords:
Damage limits
Seismic codes
Plastic rotation
Shear walls
ASCE/SEI 41
Eurocode 8
Turkish Seismic Code

a b s t r a c t
Recently proposed changes to modeling and acceptance criteria in seismic regulations for both exure
and shear dominated reinforced concrete structural walls suggest that a comprehensive examination is
required for improved limit state denitions and their corresponding values. This study utilizes nonlinear
nite element analysis to investigate the deformation measures dened in terms of plastic rotations and
local concrete and steel strains at the extreme ber of rectangular structural walls. Response of nite
elements models were calculated by pushover analysis. We compare requirements in ASCE/SEI 41, Eurocode 8 (EC8-3) and the Turkish Seismic Code (TSC-07). It is concluded that the performance limits must
be rened by introducing additional parameters. ASCE/SEI 41 limits are observed to be the most accurate
yielding conservative results at all levels except low axial load levels. It is shown that neither EC8-3 nor
TSC-07 species consistent deformation limits. TSC-07 suggests unconservative limits at all performance
levels, and it appears to fall short of capturing the variation reected in the calculated values. Likewise
EC8-3 seems to fail to represent the variation in plastic rotation in contrast to several parameters
employed in the calculation. More accurate plastic rotation limits are proposed.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
One of the most important steps of performance based assessment of RC buildings relies on comparison of deformations
obtained from nonlinear structural analyses (static or dynamic)
with the performance based limits. These deformation limits
signicantly affect the assessment result so their accuracy plays
a critical role. Provisions for performance assessment of reinforced
concrete structures, such as FEMA356 [1], Eurocode 8 [2] and
ASCE/SEI 41 [3] include deformation limits for both exure and
shear controlled wall members at specic limit states to estimate
the performance of components and structures. The criteria are dened in terms of plastic hinge rotations and total drift ratios for the
governing behavior modes of exure (ductile members) and shear
(brittle members), respectively. Recently, strain limits are dened
for concrete in compression and steel in tension at serviceability
and damage-control limit states as a vital component of direct
displacement-based design procedures [4]. The recently revised
Turkish Seismic Code (TSC-07) [5] species limiting strain values
associated with different performance levels of reinforced concrete
members. While deformations are specied in relation to global
parameters, local damage indicators in terms of strain limits are
used inconsistently to determine the expected performance. For
Corresponding author. Tel.: +90 312 233 1403.
E-mail address: polatgulkan@cankaya.edu.tr (P. Glkan).
0141-0296/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2012.05.011

results of nonlinear pushover analyses to be evaluated according

to either of the acceptance criteria, i.e. whether local or global
response will imply similar performance states is a matter that
must be established because full calibration of the requirements
is lacking.
Another criticism raised against the rotations associated with
different limit states is that they may turn out to be lower than
the actual rotations expected to develop in reinforced concrete sections [6]. So, it is postulated that the given limits may be unduly
conservative. In a way this is a direct consequence of adaptations
performed for the plastic hinge analysis method employed in the
guidelines. In most applications the momentcurvature relation
of a section is calculated using the plane section assumption, the
limiting plastic rotations in codes were adjusted to conform to
the resulting plastic rotations calculated by multiplying the assumed plastic hinge length and plastic curvature rather than the
actual rotations. These issues will be examined in this study to test
adequacy of the limits specied by codes and guidelines. In previous studies [7,8] analytical modeling techniques for shear wall elements both in micro- and macro-levels were investigated and
closed-form equations for estimation of building collapse capacity
of moment resisting frame and shear wall structural systems
subjected to seismic excitations were developed. We employ nonlinear nite element analysis for reinforced concrete structural
components that has been thoroughly veried by benchmark problems as can be found in Kazaz [9].

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

106

Structural performance assessment, done either by the family of

methods grouped under nonlinear static procedures (NSPs) or by
using the more reliable nonlinear dynamic analysis needs to provide insight into how severe individual members are deformed.
The bridge that gaps the global deformations and member end
deformations must be reliable and stable. In this article we develop
a tool that has been derived for walls that will provide improved
estimates for member level deformations. Current modeling and
acceptance criteria available in the provisions and codes will be
summarized in the next section.

above, but the differences between the deformation measures

(drift, rotation, curvature and strain) and limiting values used as
modeling and acceptance criteria for structural members can lead
to signicant differences in the estimations of global structural
performance. For instance, while ASCE/SEI 41 and EC8-3 uses
plastic rotation as the primary deformation parameter, TSC-07 uses
section strains to assess the performance level of a member.
Considering that these measures are interchangeable, the need
for consistent limit values in different deformation measures is
apparent.

2. Code performance limits

2.1. ASCE/SEI 41 performance limits

A performance level describes a limiting damage condition that

may be considered likely to be brought into existence for a given
building under seismically induced deformations. Building seismic
performance level is determined on the basis of structural member
damage states. Basically all deformation-based provisions employ
similar damage state denitions for reinforced concrete members.
Structural members are classied as ductile and brittle with respect to their mode of failure in determining the damage limits.
Fig. 1 shows the conceptualized force versus deformation curve
used in ASCE/SEI 41 [3], TSC-07 [5] and EC8-3 [2] to specify member modeling and acceptance criteria for deformation-controlled
actions. Three discrete Component Performance Levels and two
intermediate Component Performance Ranges are dened in
Fig. 1 to identify the performance level of a member. The terminology used in reference to damage states of a member differs among
the documents. In ASCE/SEI 41 [3], the discrete Performance Levels
are Immediate Occupancy (IO), Life Safety (LS), and Collapse Prevention (CP). The intermediate Structural Performance Ranges are
designated as Damage Control Range and the Limited Safety Range.
In EC8-3 [2] the discrete Limit States are named as Damage Limitation (DL), Signicant Damage (SD), and Near Collapse (NC). In TSC07 [5] Damage Limits are Minimum Damage Limit (MD), Safety
Limit (SL) and Collapse Limit (CL) (Fig. 1).
As indicated in Fig. 1, at the Collapse Prevention level (CP)
member deformation capacities are taken at ultimate strength or
at lateral displacement demand at which capacity begins to rapidly
degrade for primary components. At the Life Safety level (LS),
member deformation capacities are reduced by a (safety) factor
of 4/3 over those applying at Collapse Prevention. For the Immediate Occupancy (IO) two denitions arise in reference to Fig. 1.
While ASCE/SEI 41 and TSC-07 anticipate some degree of nonlinear
deformation beyond the global yield for the immediate occupancy
level and minimum damage, respectively, EC8-3 adopts the global
yield point as the limit state for the damage limitation on the
member.
Seismic assessment provisions investigated in this study establish the damage states on semantically similar denitions as stated

ASCE/SEI 41 [3] basically adopts the same performance limits

proposed in the wall provisions of FEMA 356 [1] for the seismic
assessment and rehabilitation of existing buildings. According to
ASCE/SEI 41 shear walls shall be considered slender if their aspect
ratio is Hw/Lw > 3.0, and shall be considered short or squat if their
aspect ratio is Hw/Lw < 1.5. Slender shear walls are normally controlled by exural behavior; short walls are normally controlled
by shear. The response of walls with intermediate aspect ratios is
inuenced by both exure and shear. For walls deforming inelastically under lateral loading governed by exure, the rotation (h)
over the plastic hinging region at the base of member will be used.
For shear walls whose inelastic response is controlled by shear, the
deformation limits are expressed in terms of the lateral drift ratios.
For multi-story shear walls the drift shall be the story drift.
Table 1 gives the ASCE/SEI 41 plastic rotation limits for members controlled by exure where P/Po is the axial load ratio and v
is the maximum average shear stress in the member
normalized
p
with respect to concrete compressive strength fc calculated as

V max
p
fc

t w Lw

Here Vmax is the maximum shear force carried by the member. The
knowledge inherited in normalized shear stress expression given in
Eq. (1) covers the parameters that affect the wall response signicantly, so normalized shear is a useful parameter that discriminates
the distinct behavior modes of wall response. ASCE/SEI 41 adopts
the ACI 318-02 [10] requirements for the denition of a conned
boundary.
Elwood et al. [11] proposes further changes to acceptance and
modeling criteria for walls controlled by both exure and shear,
in order to make them more consistent with experimental results.
For exural walls the
limiting
average
shear stress in Table 1 was

p
p
increased from 0:25 fc to 0:33 fc (MPa) to obtain a better match
with experimental results. Linear interpolation between tabulated
values is to be used if the member under analysis has conditions
that are between the limits given in the tables.
2.2. Eurocode 8
The deformation capacity of beam-columns and walls is dened
as the chord rotation h, i.e., the angle between the tangent to the
axis at the yielding end and the chord connecting that end with
the end of the shear span (Lv = M/V = moment/shear), i.e., the point
of contra-exure. The chord rotation is also equal to the element
drift ratio, i.e., the deection at the end of the shear span divided
by the length. The state of damage in a member is dened in
EC8-3 [2] by three Limit States:

Fig. 1. Component performance levels.

2.2.1. Limit state of Near Collapse (NC)

The value of the total chord rotation capacity (elastic plus
inelastic part) at ultimate hum of concrete members under cyclic
loading may be calculated from the following expression:

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

107

Table 1
Plastic rotation limits for shear wall members controlled by exure in ASCE/SEI 41.
Shear walls and wall segments
As  A0s fy
t w lw fc0
60.10
60.10
P0.25
P0.25
60.10
60.10
P0.25
P0.25
a

Acceptable plastic hinge rotation (rad)

Shear force
p
t w lw fc0

Conned boundary

60.25 (3)a
P0.50 (6)a
60.25 (3)a
P0.50 (6)a
60.25 (3)a
P0.50 (6)a
60.25 (3)a
P0.50 (6)a

Yes
Yes
Yes
Yes
No
No
No
No

Performance level
IO

LS

CP

0.005
0.004
0.003
0.0015
0.002
0.002
0.001
0.001

0.010
0.008
0.006
0.003
0.004
0.004
0.002
0.001

0.015
0.010
0.009
0.005
0.008
0.006
0.003
0.002

The values in parentheses are in psi.


0:225
max0:01; x0
0:0160:3v
fc
max0:01; x
cel
 0:35  fyw 
aqsx f
Lv
c
25
1:25100qd
h

hum

2.3. Turkish Seismic Code (TSC-07) limit states

where cel = 1.5 for primary elements and 1.0 for secondary elements, h = depth of cross-section, m = N/bhfc (b width of compression zone, N axial force positive for compression), x and
x0 = reinforcement ratio of the longitudinal tension (including the
web reinforcement) and compression reinforcement, respectively,
fc is the concrete compressive strength (MPa), qsx = Asx/shbw = ratio
of transverse steel parallel to the direction x of loading (sh = stirrup
spacing), qd = steel ratio of diagonal reinforcement (if any), in each
diagonal direction, a = connement effectiveness factor. In walls the
value given by Eq. (2) is multiplied by 0.625.
The value of the plastic part of the chord rotation capacity of
concrete members under cyclic loading may be calculated from
the following expression


0:3
max0:01; x0
0:01450:25v
fc0:2
max0:01; x
cel
 0:35  fyw 
aqsx fc
Lv
25
1:275100qd
h

hpl
um

cel = equal to 1.8 for primary elements and 1.0 for secondary elements in Eq. (3).
2.2.2. Limit state of Signicant Damage (SD)
The chord rotation capacity corresponding to signicant damage hSD may be assumed to be 75% of the ultimate chord rotation
hum given by Eq. (2).
2.2.3. Limit state of Damage Limitation (DL)
The capacity for this limit state used in the verications is the
yielding bending moment under the design value of the axial load.
In case the verication is carried out in terms of deformations the
corresponding capacity is given by the chord rotation at yielding hy,
evaluated for walls using the following equation



Lv av z
h
db fy
0:13/y p
0:002 1 0:125
hy /y
3
Lv
fc

where /y is the yield curvature and aVz is the tension shift of the
bending moment diagram, db is the (mean) diameter of the tension
reinforcement. z is the internal lever arm, taken equal to 0.8Lw in
walls with rectangular section. av should be set equal to 1 if shear
cracking is expected to precede exural yielding, otherwise
av = 0.0. The rst term in the above expressions accounts for exure,
the second term for shear deformation and the third for bond slip of
bars.

In a perplexing divergence from either of these two approaches

the Turkish Seismic Code [5] species strain limits to evaluate the
performance of reinforced concrete members. Depending on the
analysis tool, concrete compression and steel tension strain
demands at the member section extreme bers can be obtained
directly (if the ber section modeling technique is used) or must
be transformed from section rotations, which are obtained from
pushover analyses or time history analyses (if generalized load
deformation models were used in the member modeling). When
the latter case holds the plastic rotations obtained at the member
plastic hinge locations are used for calculating the plastic curvature
demands at these critical sections followed by the calculation of
total curvature, /t, by adding the yield curvature, /y using the following equation:

/p

hp
;
lp

/t /p /y

Concrete compressive strains and steel tensile strain demands

corresponding to the calculated total curvature demand at the
plastic regions are calculated from the momentcurvature diagrams obtained by conventional sectional analyses of the critical
section. Momentcurvature diagrams of the critical sections are
obtained by using appropriate stressstrain rules for concrete
and steel. Finally, the calculated strain demands are compared
with the damage limits given below to determine the member
damage states.
Concrete and steel strain limits at the bers of a cross section
for Minimum Damage Limit (MD) are given as

ecu MD 0:0035;

es MD 0:010

Concrete and steel strain limits at the bers of a cross section for
Safety Limit (SL) are

ecg SL 0:0035 0:01qs =qsm  0:0135;

es SL 0:040

and for Collapse Limit (CL) they are specied as

ecg CL 0:004 0:014qs =qs m  0:018;

es CL 0:060

In Eqs. (6)(8), ecu is the concrete strain at the outer ber, ecg is the
concrete strain at the outer ber of the conned core, es is the steel
strain and (qs/qsm) is the ratio of existing connement reinforcement at the section to the connement required by the Code. The
ASCE/SEI 41 [3] standard recommends that the maximum compressive strain in longitudinal reinforcement shall not exceed 0.02, and
maximum tensile strain in longitudinal reinforcement shall not
exceed 0.05 in conned concrete. The Turkish requirements seem
to transcend both limits, but no justication is provided as to the
provenance of these generous limits.

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

108

The limits given in TSC-07 [5] are mostly based on the studies
and proposals of Priestley et al. [4], where strain limits for tension
and compression in relation to serviceability and damage-control
limit states to be used in momentcurvature analysis were listed.
Damage-control limit state corresponds to the collapse limit in
TSC-07. The substitution of strain limits for performance acceptance criteria is not appropriate for walls, because Priestley
et al.s work is based on momentcurvature analysis that relies
on the plane section hypothesis.
3. Analytical framework for the investigation of performance
limits
3.1. Parameters
A parametric study calculating the static response of idealized
cantilever structural wall models under monotonically increasing
lateral point load applied at the top was conducted. Instead of an
inverted triangular load distribution mimicking the rst mode
response, a point load applied at the effective height (2Hw/3)
was used as shown in Fig. 2. The effective height in such case is also
referred as shear span (Lv). Elastic solutions obtained from both
cases result in only minor difference in lateral deection calculated
at the effective height. Effect of different design and wall parameters on the deformation measures of structural walls was investigated. The variables of the parametric study are summarized
below. The schematic description of these variables is given in
Fig. 2. The parameters are
 Wall length (Lw): 3 m, 5 m and 8 m.
 Effective shear span (Lv): 5 m, 6 m, 9 m, 15 m, 24 m.
 Wall boundary element longitudinal reinforcement ratio (qb):
0.5, 1, 2, 4%.
 Wall axial load ratio at the base (P/fc/Aw): 0.02, 0.05, 0.1, 0.15,
0.5.
3.2. Design requirements
The walls were designed according to TSC-07 specications.
Concrete strength was taken as 25 MPa for all cases. Wall boundary
elements were assumed to extend over a region of 0.2Lw at the
edges according to TSC-07. A constant value of tw = 250 mm was
assigned for the wall thickness. A sketch of a typical wall section

Fig. 2. Illustration of variables of the parametric study.

is displayed in Fig 3. For any given combination of above parameters, such as wall length (Lw), ratio of boundary element longitudinal reinforcement area to the boundary region cross section area
(qb) and axial load ratio (P/Po), the wall yield moment (My) was
calculated. For a given qb the steel area (As = 0.2Lwtwqb) in the
boundary element was calculated and converted to discrete longitudinal bars of specic diameter. In the following step, using the
specied shear span length (Lv) the design shear force was calculated (Vd = My/Lv). The ratio of the horizontal and vertical web reinforcement was assumed to be nominally 0.0025. If the factored
shear force (Ve = kVd) exceeded the shear safety limit calculated
with V n Aw 0:65f ctd qt fyd according to TSC-07, the required
amount of web horizontal reinforcement was recalculated employing the same equation. Since codes specify that the amount of vertical reinforcement should not be less than the horizontal
reinforcement in the web, the same steel ratio of web reinforcement was used in the vertical direction. The design shear force
was factored only for exural over-strength. The amplication in
the base shear due to higher mode effects was disregarded.
The deformation capacity of structural walls is controlled by the
level of connement in the boundary elements. TSC-07 and ACI
318-02 calculate the amount of transverse reinforcement that is
required at the wall boundaries with similar expressions. The
expression in TSC is given as Ash 0:05sbc fck =fytk . This is 2/3 of
the amount of transverse reinforcement used to conne the column elements. The same equation with a multiplier of 0.09 is given
in ACI 318. Since the thickness of walls was taken as constant,
8 mm bars at 100 mm spacing (/8/100 mm) was used as transverse reinforcement at the boundaries. The yield strength of transverse bars was assumed to be 420 MPa. If the ACI 318-02 [10] had
governed the design, /8 hoops at 85 mm spacing would have been
required as connement steel at the boundary elements. In conclusion wall boundaries can be considered as well conned for TSC-07
and adequately conned for ACI 318-02. Obviously connement
should be considered among the variables of the parametric study,
but since this would have increased the analysis permutations signicantly, the study was limited to conned members. Since the
conned concrete model employed here requires only area and
spacing of the transverse reinforcement to achieve the intended
conned section, no specic detailing of the section has been
shown herein.
3.3. Finite element model
Response of the designed walls was calculated using nonlinear
nite element analysis program ANSYS [12]. A new wall model
was developed for analysis purposes. Trial analysis of cantilever
walls under monotonically increasing uniform and inverted triangular load patterns demonstrated that even when cracking may extend up to mid-height of the wall, signicant steel yielding extends
over only lower one or two stories. The upper stories can be effectively treated as a cracked beam. Using this analogy the nite
element model displayed in Fig. 4 was developed to reduce the
computation time. The regions with different colors at the edges
designate the boundary elements. Conned concrete material
model was used in this region. Web concrete was taken as unconned concrete. The rst two stories of the cantilever wall were
discreticized with solid continuum elements whereas the upper
stories were modeled with Timoshenko beam elements. The nonconformance between the nodal degrees of freedom of beam and
solid elements was overcome by providing the transition with
constraint elements. To dene the behavior of beam elements
generalized nonlinear section properties were used. The load
deformation behavior of beam elements was assigned in the form
of bilinear forcedistortion angle (Fc) and momentcurvature
(M/) relation. The initial exural rigidity was taken as 0.5EIw

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

109

Fig. 3. Sketch of a typical wall section and reinforcement detailing of analyzed models.

Fig. 4. Finite element model.

[1]. This model proved to be adequate because all the response

parameters under investigation were focused at the lower stories.
Nonlinear loaddeformation response of the wall models under
monotonically increasing force was determined using static analysis with a point load applied at the top as displayed in Fig. 4.
3.4. Material models
Several physical effects causing nonlinearity in the system were
included in the material models adopted for the concrete and steel
materials. For concrete material the ve-parameter WillamWarnke [13] criterion was used with the solid element. The model assumes linear elastic stressstrain relationship until crushing.
When used without a plasticity law it underestimates the deformation capacity of concrete because it neglects the nonlinearity in the
ascending branch and the post-crushing strength of concrete in
compression. ANSYS offers a number of rate independent kinematic and isotropic hardening plasticity options that can be used
with the concrete element to model the compression behavior.
Multi-linear isotropic work hardening plasticity (MISO) was combined with the tensile failure criteria of WillamWarnke material
model (CONC). When plasticity based models are combined with
the WillamWarnke concrete material option (CONC), the plasticity check is done before the cracking and crushing checks. MISO is

similar to von Mises yield criterion except that a multilinear curve

is used instead of a bilinear curve. Yielding or cracking of any material point within the element is evaluated on the basis of principal
stresses. If the wall response is simplied to a plane stress condition, it is obvious from Fig. 5a that in the quadrants for tension
tension and tensioncompression the WillamWarnke model will
prevail until the cracking of concrete. Upon cracking a plane of
weakness will form orthogonal to the crack direction which reduces the principal stress in this direction to zero as the solution
converges. Following the stress relaxation due to cracking in the
quadrant tensioncompression both models will interact. In the
quadrant for compression-compression pure plastic behavior will
apply. The combined multi-surface failure model is displayed in
Fig. 5a. The failure (yield) stress value in each surface is input
through the multilinear stressstrain curve displayed in Fig. 5b.
This curve was obtained using available concrete models.
The conned concrete compressive stressstrain curve at the
boundary elements was calculated according to the Saatcioglu
and Razvi [14] model. Then the smooth curve was brought to a
multilinear form in ve segments. Typical conned concrete curves
are plotted in Fig. 5b. For the unconned concrete classical Hognestad model was used.
Uniaxial behavior of longitudinal and transverse steel bars was
modeled with a bilinear isotropic hardening using the von Mises
yield criterion. Modulus of elasticity of the steel material was taken
as 200,000 MPa. The yield stress and tangent modulus at the strain
hardening was taken as 420 MPa and 1500 MPa, respectively. A
strain hardening stiffness helps in achieving a quicker
convergence.
Buckling of longitudinal reinforcement in the compression
boundary element was modeled according to Dhakal and Maekawa
[15]. In this model the average compressive stressstrain relationship including the softening in the post-buckling
p range can be
completely described in terms of the product of fy and the slenderness ratio, s/db, of the reinforcing bar, where s is the unconned
length of the longitudinal reinforcement between the two transverse reinforcement and db is the diameter of the longitudinal
bar. Stressstrain curves modied to take into account bar buckling
were implemented for the longitudinal bars at the compression
boundaries. Typical curves are plotted in Fig. 5c.
A sample simulation study is described next. The nite element
model and analysis results of the wall specimen with rectangular
cross section (RW2) tested by Thomsen and Wallace [20] is
displayed in Fig. 6. The wall was 3.66 m tall and 102 mm thick. Wall
length was 1.22 m. Reinforcement conguration being similar to
Fig. 3, well-detailed boundary elements of 153 mm length
(0.125Lw) were provided at the edges of the wall over the bottom
1.22 m of the wall. The volumetric ratio of the longitudinal reinforcement in the boundary element was approximately
qb = 0.033. Vertical web reinforcement amounts to qsh = 0.00325.
The ratio transverse reinforcement for the connement of boundaries was qs = 0.01. The average compressive strength of concrete
at the time of testing was measured to be 42.8 MPa. The longitudinal bars were Grade 60 (fy = 414 MPa) and the ultimate steel strain

110

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

Fig. 5. Description of the material models for concrete and steel: (a) Biaxial stress state representation of the combined material model for concrete, (b) Multilinear isotropic
hardening plasticity used to model the behavior of conned concrete under compression, (c) Bilinear steel uniaxial base stressstrain curve and curves modied for buckling.

Fig. 6. (a) Dimensions of the wall tested by Thomsen and Wallace [20], (b and c) Finite element models, (d) Vertical strain plot, (e) Variation of vertical strain prole over
229 mm gage length along the web in the plastic hinge region, (f) Force displacement response.

at rupture from material tests was obtained as 0.08 for the #3 rebar
that was used as longitudinal reinforcement at the boundary elements. The specimen was loaded cyclically by hydraulic actuators
at the top. An axial stress of approximately 0.075Agfc (P/
Po = 0.075) was maintained throughout the duration of the test.

Analysis of the test specimen was performed under both cyclic

and monotonic static loads. The comparison of calculated and
experimental loaddeformation curves are given in Fig. 6f. The
FEM model successfully captured the experimental global
response. The wall shear stress normalized with respect to square

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

root of concrete strength was m = 0.19. Fig. 6d displays the vertical

distribution of strains along the edges at the ultimate limit state.
The maximum concrete compression and steel tensile strains were
calculated as 0.0113 and 0.0335, respectively. The distribution of
vertical strains over the 221 mm high gage length along the web
of the wall at the base is plotted in Fig. 6e. These strain distributions give the curvature proles for intermediate and severe damage levels at the base section of the wall. Although the response of
the wall was governed by exure (m = 0.19 < 0.33), the curvature
prole is not linear. Using the edge strains, the base curvature at
2% drift was calculated as (0.0335 + 0.0113)/1.22 = 0.0367 rad/m.
Using the given base section properties, the momentcurvature
relation (M/) at the base of the wall was calculated by section
analysis. At the calculated curvature the concrete and steel strains
were obtained as 0.007 and 0.0344, respectively, from section
analysis. Wall edges would be considered as well conned in both
TSC-07 and ACI 318-02 (qs = 0.01). Under such conditions Eq. (8)
calculates the ultimate strains for concrete and steel as 0.018 and
0.06, respectively. The same strain limit states were proposed for
the damage-control by Priestley et al. [4]. These limits are greater
than the actual strains. This situation jeopardizes the reliability of
the performance evaluation procedures based on strain limits for
reinforced concrete members.
Further details of the modeling approach and verication
studies on 23 shear wall specimens that were tested under
monotonic, cyclic and dynamic conditions can be found in Kazaz
[9] and Kazaz et al. [16]. It is believed that further discussion on
the nite element modeling aspects, which can be found in the
given references, will contribute much to the scope of this
article.
4. Relation between drift, curvature, rotation and section
strains
In the interest of consistency with the conventional analyses
procedures and deformation measures of modeling and acceptance
criteria specied in codes and standards, curvatures and rotations
on the model should be evaluated in a way that will not contravene
the plastic hinge analysis method [17]. The method is especially
appealing for structural wall buildings because it is simple and it
is possible to idealize a wall member inside the building as isolated
cantilevers as displayed in Fig. 7. In the plastic hinge analyses the
tip displacement of a cantilever is obtained as the sum of its exural yield displacement, Dy, and plastic displacement component,
Dp. While the yield displacement is calculated by double integrating the curvature distribution along the cantilever, plastic displacement component is calculated by multiplying the height of the
cantilever (measured from the center of plastic hinge region) by
the plastic rotation, hp, at the base as expressed in the following
equation:

D Dy Dp

/y H2
/  /y Lp H  0:5Lp
3

111

The term, (//y)Lp, in Eq. (9) is the plastic rotation hp and is

based on the governing assumption that the plastic curvature is
lumped at the center of the equivalent plastic hinge length, Lp.
The actual physical length over which the plasticity spreads may
be larger and referred to as plastic hinge region, Lpz. The plastic
zone length that yields accurate plastic rotation can best be determined from experiments. Eq. (9) may then be used to calculate the
equivalent plastic hinge length.
Relating the local deformation demands (strains) of the wall to
the curvatures so that exural deformations can be interpreted
correctly requires correct determination of plastic hinge length or
spread of plasticity along the member. The method that was
employed to evaluate the length of the plastic zone, Lpz, can be described as follows. Curvature prole computed from element
strains calculated at the same height at the two wall ends were
used to determine the spread of plasticity along the wall. The limiting yield curvature to determine the spread of plasticity along the
wall can be calculated with the expression proposed by Priestley
et al. [4]

/y 2

ey
Lw

10

where ey is the yield strain of the reinforcement and Lw is the wall

length.
As a second step, the base curvatures and rotations were calculated over the identied plastic zone lengths. The rotations (hb),
which were assumed to represent the rotation of the base section,
were calculated just above the plastic zone length by using the vertical displacements calculated at tensile and compressive edges in
the same row. The sketch in Fig. 8 illustrates the calculation of the
base section curvature and rotation in a way that is consistent with
ASCE/SEI 41. The base curvature was calculated in two different
ways to ensure the accuracy in the calculation of this parameter,
because all the performance criteria and assessment procedure depends on it. /b1 was obtained by using the moment-area theorem.
Since the rotation above the plastic zone is known (hb) now by integrating the curvature prole, which was assumed to be linear along
the plastic zone length Lpz, /b1 was obtained from 2hb/Lpz. Then /b2
was obtained by tting a best line to the curvature prole along the
plastic zone length. The intercept of the best t line equation at the
base level was adopted as /b2. The two methods of curvature calculation yielded very similar results. The base curvatures used in
this study are those calculated by best line t method, i.e. /b = /b2.
Violation of the plane section remains plane after deformation
hypothesis due to concentration of compression strains at the
section of maximum moments at the base of the wall precludes a
direct comparison of the curvatures calculated from experimentally

Fig. 7. Denition of plastic hinge length and analysis [17].

112

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

Fig. 8. Schematic descriptions of base curvature and rotation calculation.

measured strains and those obtained from nite element analyses

such as in this article with the curvatures obtained from momentcurvature based section analysis. Base curvature calculated
in the plastic hinge length may provide a more appropriate means
of linking local deformations measured experimentally (or results
of FE analysis) at the base of the wall to the results from momentcurvature analysis. The plastic rotations calculated above
the plastic hinge region should be compatible with the modeling
and acceptance criteria values tabulated in ASCE/SEI 41 [3].
In the current study the strains and curvatures were evaluated
according to section analyses, because the current state-of-the-art
analysis methods for shear walls depend primarily on plane section deformations (plastic rotation-plastic hinge, ber elements,
multiple vertical spring line elements). So, the presented results
and recommendations must be consistent with section based analysis methods. Matching the momentcurvature relations of nite
element analysis and section analysis on the basis of curvature,
the strain limits were determined.
The section based momentcurvature analyses were conducted
with a modied version of CUMBIA [18]. Comparison of typical
momentcurvature relations obtained from section analyses and
nite element analyses is shown in Fig. 9. The curves agree in
the initial segment and in terms of moment capacities, but the ultimate curvature capacities obtained from the two different analyses
differ signicantly. The ratio of ultimate curvatures obtained from
section and nite element analyses are plotted as a function of normalized shear stress in Fig. 10. It is seen from the gure that the
section analysis results deviate from the nite element analysis
results in terms of curvature limits as the wall length and unit
shear on the member increase. In most cases the sectional analysis
results overestimated the deformation capacity of walls, which
may lead to un-conservative assessment of the structural walls.
Even when section analyses indicate very large deformation
capacities, the limiting value for the curvatures was adopted as

the capacities obtained from the nite element analysis in this

study.
The second oor drift ratio was used to display the deformation
capacity of the walls. Roof drift of cantilever walls is not a meaningful measure to investigate the deformation capacity because if
the drift value is used as modeling criterion for structural walls it
would be unconservative. Additionally, the base stories are the
most critical regions when the deformability of the walls is considered. In this way the deformation components can be directly compared with code specied values. The second oor displacements
were preferred over rst oor values because the total drift composed of shear and exural deformation components can be calculated more representatively over two story height due to excessive
diagonal cracking and localized damage at the base story.
5. Calculated damage states
The response quantities of wall models are presented at the
three damage levels given above namely, global yield, ultimate
and an intermediate damage level, i.e. life safety, dened as the
percentage of ultimate. In the analysis the ultimate point was
determined on the basis of one of the criteria dened as the point
on the loaddeformation curve where strength drops abruptly or
degrades to 85% of the ultimate strength (Vmax), or the steel strain
at the tension side exceeds es = 0.1, or the reinforcing bars at the
compression side buckles (accompanied by signicant crushing
of concrete). The degrading effect of cyclic loading regimes on
the stiffness and strength of reinforced concrete was not considered in the analyses carried out in this study. Vallenas et al. [19]
proposed that as a general rule the overall deformation capacity
under a realistic ground motion could be expected to be over
75% of the deformation capacity under monotonic loading conditions. This is due to deterioration of concrete, and the development
of cyclic failure mechanisms associated with the load history and
characteristics of the specimens. Typical loaddeformation curves
validating this assumption was obtained from analyses of the
RW2 wall specimen tested by Thomsen and Wallace [20] as displayed in Fig. 6f. 75% of the ultimate displacement capacity of
the analytical model analyzed under monotonically increasing
loading agrees well with the ultimate displacement capacity of
the specimen tested under cyclic loading regime. Further validation can be found in Kazaz [9]. In conclusion, it is assumed that collapse prevention performance level is taken at the 75% of the
ultimate point of the nite element analyses in this study.
For damage limitation or minimum damage for immediate
occupancy two denitions arise in reference to Fig. 1, global yield

Fig. 9. Comparison of typical momentcurvature relationships obtained from section and nite element analysis.

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

Fig. 10. Ratio of ultimate curvature capacities obtained from section (/SEC) and
nite element analysis (/SIM).

and a point slightly greater than the yield point including slight
nonlinear action. Global yielding of wall models was determined
on the basis of momentcurvature relation at the wall base section
employing Eq. (10). As discussed previously ASCE/SEI 41 and TSC07 anticipate that slight nonlinear action can be accommodated by
the member. Explicitly the damage limitation point was adopted as
the point where concrete compressive strain of 0.0035 and steel
tensile strain of 0.01 is rst reached in the section level.

6. Results of analyses
6.1. Evaluation of global and local deformations
Fig. 11 presents the various deformation measures at particular
damage states, corresponding to global yield, ultimate deformation
and 75% of the ultimate deformation. These general terms are used
to dene damage states of the wall models, because although seismic design documents indicate similar damage condition, they use
different terminology for these damage states. Another objective is
to present the limiting values for global yielding of the member as
a benchmark point, because ASCE/SEI 41 and TSC-07 anticipate the
point of immediate occupancy and minimum damage beyond the
global yield. Drift limits at different damage states are plotted as
a function of normalized shear stress in the wall. The shear stress
limits used to determine the plastic rotations in ASCE/SEI 41 are
superposed on the artwork. ACI 318-02 states that the normalized
p
shear carried by a wall member should not exceed 0:83Aw fc in
MPa. The deformation limits are categorized with respect to axial
load ratio as well. As seen in Fig. 11, increased shear stress and axial load ratio signicantly reduce the deformation capacity of
structural walls. The data presented here agrees with the normalized shear stress limits used to differentiate between the exural
and shear type of behavior in ASCE/SEI 41. For walls with
m < 0.33 exural behavior governs.
According to Fig. 11a, the drift at yield varies between 0.1% and
0.4%. The drift capacities of walls that respond in the exural mode
range from 1.5% to 4.5% depending on the level
p of axial load as
shown in Fig. 11c. For exural walls (m < 0:33
pfc ) and walls
punder

combined exure and shear action (0:33 fc < m < 0:50 fc ) the
lower bound of drift can be taken as 11.5% for life safety and collapse prevention performance levels, respectively, even under very
high axial
conditions. Under moderate loading conditions
pload

(m < 0:50 fc and P/Po < 0.10) the limiting values can be extended
to 1.52.5% for life safety and collapse prevention performance
levels, respectively. These limits should be interpreted as the interstory drift ratio. The roof drift at these performance levels may take
a slightly higher value.

113

The second row plots in Fig. 11 display the dimensionless base

curvature obtained by multiplying the calculated base curvature
with the wall length. It is seen in Fig. 11d that yield base curvature
can be effectively calculated using Eq. (10). Dimensionless curvature yields the total strain excursion (/Lw = et  ec) on the section.
Restrepo-Posada [21] states that the total strain excursion (et  ec)
is better suited for characterizing the onset of buckling and fracture of reinforcing bars than the tensile strain et alone. Priestley
et al. [4], investigating the effect of several parameters on the
ultimate curvature set the damage control curvature limit as /
Lw = 0.072, and assert that the variation in limit curvatures is
small, and average values provide an adequate estimate, within
10% of the data for all except high axial load combined with high
reinforcement. From measurements on six exural wall specimens
Dazio et al. [22] proposed that fracture of boundary reinforcing
bars occurs at an average value of /Lw = 0.7esu, where esu is the
strain at rupture of the reinforcing bar, yet the scatter was relatively high. Dazio et al. [22] proposed that the ultimate damage
state limit should be set to 0.50.6esu for plastic hinge analysis
(which yields /Lw = 0.050.06 for esu = 0.1). When Fig. 11f is evaluated in the light of this discussion, it is seen that the proposed
limits can only be assumed to form the upper bound
p of the data
at the low-to-medium shear stress range (m < 0:50 fc ).
The total rotations calculated above the plastic zone as illustrated in Fig. 8 are plotted in Fig. 11gi. The smooth trend observed
in Fig. 11i conrms the correctness of including shear stress and
axial load in specifying structural wall rotation limits at ultimate
limit state. However, no pronounced effect of shear stress and axial
load is observed on the yield rotation of walls. When three distinct
data clouds are investigated in Fig. 11g, it is noticed that wall yield
rotation correlates well with wall length. This is the direct consequence of the procedure used to calculate the rotations described
above. Yield rotation is calculated over the region where the yield
curvature specied in Eq. (10) is exceeded for the rst time along
the wall. Since yielding starts at the base, where the moment is
maximum and then moves up along the wall as the top drift is
increased, for all models the rst yield occurs on the rst row of
elements on the nite element model. This length over which the
yield rotation is calculated varies between 0.25 and 0.30 m. Then
rotation becomes the product of the yield curvature and that
length. It must be noticed that if a greater length, such as the plastic hinge length that will fully develop in the inelastic response
phase, is used in deriving the yield rotation the order of rotations
will be much higher. The order of yield rotations calculated at
the very base of the walls agrees with reported experimental values [23]. Although guidelines enforce the use of plastic rotations
as assessment criteria, the section curvature may be a more appropriate means of drawing performance limits since curvature is the
direct product of simple section analyses. This way, the uncertainty
due to plastic hinge length assumption is avoided.
The last two rows in Fig. 11 present the compressive (ec)SEC and
tensile strains (et)SEC obtained by matching section curvatures from
moment curvature analysis with the base curvature (/b2) at limit
states determined from nite element analyses. There is a signicant difference between the nite element analyses strains and
section analyses strains calculated at the same curvature that arise
from plane sections assumption. Fig. 12 displays the comparison of
compressive strains from two types of analyses at the global yielding and ultimate limit states. The compressive strains calculated
from nite element analysis are especially larger than the compressive strains from section analysis at the ultimate limit state
by an order of 2.1 in average due to localized compression damage
at the boundary element. The tensile strains from the two separate
analyses methods vary but agree in the mean. The strains
presented in this study are from section analysis because nite element analysis strains are meaningless unless the tools employed in

114

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

Fig. 11. Variation of deformation parameters with normalized wall shear stress and axial load ratio.

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

115

Fig. 12. Comparison of extreme ber compression strains at global yield and ultimate limit states for section and nite element analyses.

seismic assessment of reinforced concrete members and structures

incorporate such effects in the analysis.
The limiting concrete compressive strain at wall yield can be taken as 0.002 as shown
pin
Fig. 11j. For walls responding in the exural mode (m < 0:33 fc ) under low axial load ratios (P/Po < 0.1)
compressive strains are much lower than 0.002. Conversely, the
steel strains are larger at the exure controlled region and decrease
as the shear carried by wall increases. The total strain excursion
et  ec, which is given indirectly in Fig. 11d, is nearly constant summing to approximately 0.004. However, et  ec = 2ey, which stems
from the position of the neutral axis located at the center of the
section assumption, is valid only for medium range shear stresses
when Fig. 11j and m are examined together.
Fig. 11l and o display the compressive and tensile strains calculated at the ultimate limit state, respectively. The strain limits
proposed by Priestley et al. [4] and TSC-07 [5] are superposed on
the same gure for comparison. The available limits lead to unconservative estimates of the deformation capacity at the ultimate
limit state. The limiting strains are sensitive to the axial load ratio
on the section.

6.2. Comparisons with codied limits

In Fig. 11 ve different damage parameters were addressed at
the described limit states. Among these parameters rotations are
of primary importance since structural elements are modeled as
nonlinear frame elements with lumped plasticity by dening plastic hinges at both ends of the members in structural analysis programs. While ASCE/SEI 41 and EC8-3 directly specify the plastic
rotation limits for the modeling and acceptance criteria, TSC-07
dictates strain limits for momentcurvature analysis that are required to be converted to plastic rotations. Figs. 1315 display
the comparison of the calculated rotation limits with the limits
specied in these documents. Data in each plot is also classied

with respect to axial load ratio. The description of the damage limits used in the analysis was given previously.
Fig. 13ac compare ASCE/SEI 41 limits with the calculated plastic rotation limits. At the immediate occupancy performance level
the ASCE/SEI 41 limits yield conservative estimates for medium
and high axial load ratios, but for low axial load ratios the limits
are on the unsafe side. This situation contradicts expectations,
yet it is the consequence of the procedure used in the calculation
of plastic rotations. Analysis results indicate that while yielding
initiates at a limited region near the base of the wall under low
axial load ratios, it has a distributed pattern in walls subjected to
high axial load ratios. Consequently the plastic region length is
larger in high axial load cases yielding larger plastic rotations. At
the collapse prevention performance level, ASCE/SEI 41 limits are
below 0.02 rad, yielding conservative estimates. However, it
appears that a cap has been applied for greater values. The capped
data falls into the region characterized by low shear stress and
exural response as displayed in Fig. 11i. Since Eq. (10) seems to
yield good estimation of yield curvature in reference to Fig. 11d,
the yield rotation according to ASCE/SEI 41 can be obtained using
Eq. (10) as hy = (2ey/Lw)0.5Lw = 0.0021 rad.
The rotation limits presented in EC8-3 that are dened in terms
of chord rotation differ from the rotations that will be advocated in
this study where it is intended to obtain rotation limits that account for the member deformation characteristics within a story.
The chord rotation cannot be representative of base rotation of
shear walls under higher mode effects and walls interacting with
frames, especially with the strong ones. In order to be consistent
with EC8-3 denitions, chord rotations were calculated and compared with EC8-3 limits in Fig. 14. The chord rotation is calculated
as the tip drift ratio using the exural displacement component.
EC8-3 adopts the yield point as the damage limitation point.
Fig. 14a displays the correlation of the yield rotation calculated
using Eq. (4) with the element drift ratio calculated at the tip (roof)
of cantilever nite element model. The inference of Fig. 14a is that

116

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

Fig. 13. Comparison of calculated plastic rotation limits at specic performance levels with the limits available in ASCE/SEI 41.

Fig. 14. Comparison of calculated plastic rotation limits at specic performance levels with the plastic chord rotation limits in EC8-3.

Fig. 15. Comparison of calculated plastic rotation limits at specic performance levels with the limits available in TSC-07.

the yield rotation given by Eq. (4) signicantly overestimates the

analysis results, especially as the shear-span-to-wall-length ratio
increases. No pronounced effect of axial load ratio is observed on
the yield rotation. The order of yield rotations calculated according
to EC8-3 ranging from 0.003 to 0.009 rad is high for stiff shear wall
elements. The yield rotation limits proposed by EC8-3 are unconservative for shear walls especially when the shear-span-to-walllength ratio is high.
For shear wall elements, Eq. (3) proposed in EC8-3 used to calculate the ultimate limit state plastic rotation yields conservative,
yet unrealistic limits as shown in Fig. 14c. The equations seem to
be insensitive to the most of the design parameters, except the
shear-span-to-wall-length ratio and axial load ratio. The resulting
plastic rotation limits vary between 0.006 rad and 0.019 rad on

average. The plastic rotation limits calculated according to EC8-3

are observed to be smaller than the ones given in ASCE/SEI 41,
which is contrary to the expectations. EC8-3 denes chord rotation
with respect to shear-span (M/V) as opposed to the plastic rotation
in ASCE41 dened over the plastic hinge region. The life safety limits are calculated as 3/4 of collapse prevention limits.
Fig. 15 displays the correlation of plastic rotation limits calculated according to strain limits given in TSC-07 using section based
momentcurvature analysis with the plastic rotations calculated
from the analytical model. The plastic hinge length was taken as
0.5Lw as proposed by TSC-07 in the calculation of plastic rotations
from the curvatures from section analyses. For all damage states
(or performance levels) TSC-07 overestimates the rotation
signicantly without showing much variation. The horizontally

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

extending trend in Fig. 15ac reects that the given TSC-07 limits
fall short of predicting the variation in the rotation due to varying
design parameters. As discussed previously the same is also valid
for EC8-3. The average limit curvatures in dimensionless form for
serviceability (/sLw) and damagecontrol (/dcLw) states were proposed as 0.0175 and 0.072, respectively, by Priestly et al. [4] as discussed previously. The procedure implemented in TSC-07 for the
seismic assessment results in rotations as hMD = 0.5/sLw =
0.00875 rad for minimum damage and hCL = 0.5/dcLw = 0.036 rad
for collapse limit. These limits are consistent with what is presented in Fig. 15, yet they are insufcient to calculate the actual
rotations for a range of walls. Another reason that causes differences between the analytical results and TSC-07 predictions is
the basic crudeness of plastic hinge length.
7. Discussion
The analysis results suggest that the deformation capacity of
structural walls with conned boundary elements is larger than
the limits given in ASCE/SEI 41 provisions. It is seen that ASCE/
SEI 41 yields conservative estimations of the structural performance. On the other hand, if the strain based performance criteria
dened in TSC-07 or as suggested by Priestley et al. [4] is used in
the determination of structural performance, unconservative
estimates of performance are obtained for reinforced concrete

117

rectangular walls. With reference to trends in Fig. 11i and l it can

be concluded that plastic rotation is a much more stable parameter
than the strains to establish the limit states of reinforced concrete
members. The dispersion in the strain data in Fig. 11k and l indicates that this measure of deformation is excessively sensitive to
member dimensions, material properties, reinforcement and the
level of axial load than the plastic rotations. So, the results of this
study favor the use of plastic rotations as performance limits in
the assessment of reinforced concrete wall members. Equations given in EC8-3 for calculating performance based rotation limits lead
to unconservative values at yield. The plastic rotation limits show
insignicant variation and appear to be inadequate in estimating
the nite element results. It is worth noting that similar equations
are used for beams, columns and walls despite signicant differences in their behavior within a structure. We address one part
of this oversimplication by the renement for walls.
Fig. 16 displays the lower bound plastic rotation limits of the
data obtained from analyses at different performance levels. In this
gure, ASCE/SEI 41 limits are also displayed for comparison. As
seen here, the major differences are for limits in the immediate
occupancy level and at low and high shear stress ranges for the
otherptwo
levels. ASCE/SEI 41 limits are specied at
performance
p
0:33 fc and 0:50 fc , where intermediate values can be obtained
by linear interpolation. However, outside these limits we have little basis about what the extrapolated trend may be. The proposed

Fig. 16. Lower bound plastic rotation limits at different performance levels.

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

118

Table 2
The proposed plastic rotation limits for shear wall members controlled by exure
(underlined values corresponds to ASCE/SEI 41 limits).

P/Po 6 0.10
m < 0.33
m = 0.33
0.33 < m < 0.50
m > 0.50
m = 0.50
P/Po = 0.15
m < 0.33
m = 0.33
0.33 < m < 0.50
m > 0.50
m = 0.50
P/Po P 0.25
m < 0.33
m = 0.33
0.33 < m < 0.50
m > 0.50
m = 0.50
a

IO

LS

CP

0.0050.0106m

0.0180.0242m

0.0250.0303m

0.0015/0.005a
0.0015
0.0015

0.01/0.01
0.01390.0118m
0.0120.008m

0.015/0.015
0.02470.0294m
0.0150.01m

0.0015/0.004

0.008/0.008

0.01/0.01

0.0050.0045m

0.0180.0273m

0.0250.0394m

0.0025/0.0043
0.00350.0029m
0.00250.001m

0.009/0.0087
0.01480.0176m
0.00850.005m

0.012/0.013
0.0160.012m
0.0160.012m

0.002/0.0032

(0.006/0.0063

0.01/0.0083

0.0040.0045m

0.0150.0242m

0.020.0333m

0.0025/0.003
0.00350.0029m
0.00250.001m

0.007/0.006
0.01180.0109m
0.0070.004m

0.009/0.009
0.0120.009m
0.0120.009m

0.002/0.0015

0.005/0.003

0.0075/0.005

Fig. 17. Correlation of predicted plastic rotations with analysis results.

This study/ASCE/SEI 41.

Table 3
Coefcients of Eq. (12) to calculate the ultimate plastic rotation limit of structural
walls with conforming boundary elements.
P/Po

60.10
=0.15
=0.25

0.138
0.087
0.034

0.220
0.148
0.037

1.814
1.779
1.485

0.071
0.066
0.037

limits for the plastic rotations of walls controlled by exure are

tabulated in Table 2. These limits are given as alternative values
to Table 6.18 in ASCE/SEI 41 for conforming members. The limits
are derived as a function of normalized shear stress (m) and axial
load level (P/Po) for different ranges of these variables in order to
obtain more accurate representation of plastic rotation limits at
the specied performance levels. Limits in relation to mid-range
axial load levels (P/Po = 0.15) are also introduced to increase the
accuracy of the assessment procedure. In case the ASCE/SEI 41
format, i.e. numerical limit values at specic m and P/Po, the underlined number corresponds to the existing ASCE/SEI 41 limit and the
left side number is the value proposed by this study.
Alternative to limits proposed in Table 2, the lower bound plastic rotation limits can also be expressed with a more general exponential expression as given in the following equation:

(
hp

0:031  0:053P=Po  e1:75m

if P=Po  0:10

0:031  0:053P=Po  e1:901:47P=Po m

if P=Po > 0:10

11

As seen in Fig. 16 the proposed values correspond to the lower

bound limits of the collective results. A general expression is
derived through regression analysis to calculate the actual rotation
limits at collapse prevention performance level. Using the parameters that govern the response of shear walls an equation is derived
to estimate the rotation capacity of shear walls with moderately
conned boundary elements. The equation from a regression analysis reads as

hp Aqb B  eCmDLw  rhp

12

where A, B, C and D are coefcients dened in Table 3 as a function

of axial load ratio, the boundary element reinforcement ratio (qb),
the normalized shear stress (m and the wall length (Lw). While
normalized shear and axial load have already been used to express

the plastic rotations in the seismic documents, the existence of Lw in

the expression can be legitimized by the fact that the rotations in
our model were calculated above the plastic zone where the spread
of plasticity depends primarily on the wall length. Strain hardening
and tension shift affecting the spread of plasticity are related to the
amount of boundary element longitudinal reinforcement (qb). r(hp)
is the standard deviation of the calculated plastic rotation limits dened as a function of plastic rotation. The following set of equations
can be employed in the calculation of r(hp).

8
0:1429hp 0:0005
>
>
>
< 0:0025
rhp
> 0:8125hp  0:0219
>
>
:
0

if hp  0:014 rad
if 0:014 rad < hp  0:03 rad
if 0:03 rad < hp  0:038 rad
if 0:038 rad < hp
13

The limit obtained through Eq. (12) is greater than the limits given in Table 2. The predictions omitting the standard deviation are
compared with the analytical results in Fig. 17. The predicted
values agree quite well with the computational results. If the predicted values are reduced by 0.75 the limits for life safety performance level is obtained.
8. Conclusions
Nonlinear static or response history analysis procedures are the
tools with which the deformation response of structural components are estimated in displacement based procedures. Regardless
of the method of analysis employed, local and global quantities in
terms of internal forces and deformations form the basis of judgment. These are then used to assess performance of structural
assemblies. The most challenging part of the displacement based
assessment procedures is the determination of the deformation
limits that strongly inuence the results. Therefore, the primary
objective of the study here was to evaluate the limits recommended by the codes and guidelines we have chosen to examine
in detail. The results of this study that were obtained from comprehensive parametric analyses of the walls provide data to adequately test performance based deformation limits specied in
different documents. Among the documents evaluated, ASCE/SEI
41 limits were observed to be the most accurate ones yielding conservative results at all levels except the low axial load levels. It has
been shown that neither EC8-3 nor TSC-07 species adequately
consistent deformation limits. TSC-07 suggests unconservative
limits at all performance levels, and it appears to fall short of
capturing the variation reected in the calculated values. Likewise
EC8-3 seems to fall short of representing the variation in plastic
rotation in contrast to several parameters employed in the

_ Kazaz et al. / Engineering Structures 43 (2012) 105119

I.

calculation. The estimations are unconservative at damage limitation. Although conservative estimations are obtained at life safety
and collapse prevention levels, the values are not logical.
For displacement-based design and assessment structural performance limits are best dened in terms of slowly varying
structural deformation measures. Member end rotations and drifts
are such measures because they represent integration along deformed sections where the effect of local singularities is evened.
Strain is a poor indicator not only because it is a local index, but
also because its recovery from other measures is vulnerable to signicant modeling errors. The ASCE/SEI 41 and EC8-3 requirements
are reasonable because they refer to plastic deformations in terms
of chord rotations. The TSC-07 requirements are inconsistent, and
seem not to be based on any careful prior evaluation, victimized
by the fundamental false premise of arriving at global indices of response from very local quantities of deformation such as strain.
The computational modeling approach for a range of parameters applied to a structural wall governed by exural deformation
described in this article permits derivation of expressions for plastic rotations that expand and improve ASCE/SEI 41 limit states.
These are summarized in Fig. 16 and Table 2. These expressions
are adjusted to err on the safe side, and represent a signicant
improvement over existing provisions. The provisions of the Turkish Code seem to lack basis in fact, and must be revised.
Nonlinear static procedures or response history analyses must
ultimately be interpreted in terms of member deformations formulated as acceptability criteria. The superiority of one set of procedures against another becomes moot when acceptability criteria
have fallen prey to internal inconsistencies. By incorporating
explicitly the effects of axial load and unit shear the expressions
that have been derived in this study would permit an improved
judgment basis for performance.
References
[1] NEHRP Guidelines for the seismic rehabilitation of buildings. FEMA 356.
Federal Emergency Management Agency, Washington DC; 2000.
[2] Eurocode 8: Design of Structures for Earthquake Resistance-Part 3: Assessment
and Retrotting of Buildings. BS EN 19983. Comit Europen de
Normalisation, Brussels, Belgium; 2005.

119

[3] Seismic rehabilitation of existing buildings. ASCE/SEI 41. American Society of

Civil Engineers, Reston, Virginia; 2006.
[4] Priestley MJN, Calvi GM, Kowalsky MJ. Displacement-based seismic design of
structures. Pavia, Italy: IUSS Press; 2007.
[5] Turkish Seismic Design Code for Buildings, Specication for Structures to be
Built in Disaster Areas. TSC-07. Ministry of Public Works and Resettlement,
Ankara, Turkey; 2007.
[6] Acun B, Sucuoglu H. Performance of reinforced concrete columns designed
for exure under severe displacement cycles. ACI Struct J 2010;107(3):
36471.
[7] Shafei B, Zareian F, Lignos DG. A simplied method for collapse capacity
assessment of moment-resisting frame and shear wall structural systems. Eng
Struct 2011;33(4):110716.
[8] Jalali A, Dashti F. Nonlinear behavior of reinforced concrete shear walls using
macroscopic and microscopic models. Eng Struct 2010;32(9):295968.
[9] Kazaz I. Dynamic characteristics and performance assessment of reinforced
concrete structural walls. Ph.D. Thesis, Civil Engineering Dept., Middle East
Technical University, Ankara, Turkey; 2010.
[10] Building Code Requirements for Structural Concrete and Commentary. ACI
318-02. American Concrete Institute, Detroit, Michigan; 2002.
[11] Elwood KJ, Matamoros AB, Wallace JW, Lehman DE, Heintz JA, Mitchell AD,
Moore MA, Valley MT, Lowes LN, Comartin CD, Moehle JP. Update to ASCE/SEI
41 concrete provisions. Earthquake Spectra 2007;23(3):493523.
[12] ANSYS R11.0. Swanson analyses system; 2007.
[13] Willam KJ, Warnke ED. Constitutive model for the triaxial behavior of concrete.
Int Assoc Bridge Struct Eng Proc 1975;19:174203.
[14] Saatcioglu M, Razvi SR. Strength and ductility of conned concrete. J Struct Eng
(ASCE) 1992;118(6):1590607.
[15] Dhakal RP, Maekawa K. Modeling for postyield buckling of reinforcement. J
Struct Eng (ASCE) 2002;128(9):113947.
[16] Kazaz I, Yakut A, Glkan P. Numerical simulation of dynamic shear wall tests: a
benchmark study. Comput Struct 2006;84(8):54962.
[17] Park R, Paulay T. Reinforced concrete structures. New York: Wiley; 1975.
[18] Montejo L. Generating set of codes for the analysis of reinforced concrete
members, CUMBIA. Technical, report no. IS-07-01; 2007.
[19] Vallenas MV, Bertero VV, Popov EP. Hysteretic behavior of reinforced concrete
structural walls. EERC Report 79/20, Earthquake Engineering Research Center,
University of California, Berkeley; 1979.
[20] Thomsen JH, Wallace JW. Displacement-based design of reinforced concrete
structural walls: an experimental investigation of walls with rectangular and
T-shaped cross-sections. CU/CEE-95/06, Dept of Civil and Environmental
Engineering, Clarkson University, Potsdam, New York; 1995.
[21] Restrepo-Posada JI. Seismic behavior of connections between precast concrete
elements. Research report 93-3. Department of Civil Engineering, University of
Canterbury, Christchurch; 1993.
[22] Dazio A, Beyer K, Bachmann H. Quasi-static cyclic tests and plastic hinge
analysis of RC structural walls. Eng Struct 2009;31(7):155671.
[23] Oesterle RG, Fiorato AE, Johal LS, Carpenter JE, Russell HG, Corley WG.
Earthquake resistant structural walls tests of isolated walls, PCA R/D series
1571, Portland Cement Association, Skokie, Illinois; 1976.