Collapse Fragility Evaluation of Ductile Reinforced

Concrete Block Wall Systems for Seismic

Risk Assessment
Mustafa A. Siyam, Aff.M.ASCE 1; Dimitrios Konstantinidis, M.ASCE 2;
and Wael El-Dakhakhni, F.ASCE 3
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Seismic risk assessment is a critical first step toward mitigating the social and economic losses resulting from earthquakes. The
FEMA P-58 document, prepared by the Applied Technology Council (ATC), provides a methodology for seismic performance assessment of
buildings. The methodology consists of five main steps: assembling the building performance model, defining earthquake hazards, analyzing
building response, developing collapse fragility, and quantifying performance. Owing to the probabilistic nature of the methodology frame-
work, each step incorporates uncertainty. After the 2011 earthquake in Christchurch, New Zealand, a significant number of reinforced ma-
sonry low-rise buildings were deemed unusable, although the damage was repairable. This created a need to investigate the seismic collapse
performance of a reinforced masonry shear wall (RMSW) system. In this study, the FEMA P-58 methodology is implemented to assess the
performance of a typical RMSW office building designed according to the National Building Code of Canada (NBCC) 2010 and the Canadian
Standards Association (CSA) S304-14 masonry design code provisions. In order to develop the building collapse fragility curves, an
analytical model to predict the behavior of RMSW was developed. The model was calibrated using experimental results from an earlier
study by the authors on scaled RMSW. Subsequently, an incremental dynamic analysis (IDA) procedure was performed to study the RMSW
performance under a suite of ground motions consistent with the NBCC 2010 design spectrum for a high seismicity site in Victoria, British
Columbia. The results of the IDA were used to develop collapse fragility curves for RMSW in order to facilitate the quantification of the
collapse risk under different levels of seismic demand. This study is expected to catalyze the development of design provisions and
subsequent adoption of RMSW in the next generations of seismic codes in North America. DOI: 10.1061/(ASCE)CF.1943-5509
.0000895. © 2016 American Society of Civil Engineers.
Author keywords: Collapse; Fragility curves; Incremental dynamic analysis; Reinforced masonry; Structural walls; Seismic performance;
FEMA P-58; FEMA P-695.

Introduction owners, government officials, users, and so forth) (Hamburger

Seismic risk assessment is a critical first step toward mitigating
the social and economic losses resulting from earthquakes. In
the ongoing effort toward the development of next-generation per-
formance-based seismic design criteria and guidelines, the Applied
Technology Council (ATC) developed the FEMA P-58 (ATC 2012)
methodology for seismic performance assessment of buildings. The
methodology includes assembling the building performance model,
defining earthquake hazards, analyzing building response, develop-
ing collapse fragility, and, quantitatively assessing the building
performance. This performance quantification in terms of meaning-
ful metrics, such as probable casualties, repair costs, and downtime,
facilitates the decision-making process for stakeholders (building
1
Ph.D. Candidate, Dept. of Civil Engineering, McMaster Univ.,
Hamilton, ON, Canada L8S 4L7. E-mail: siyamm@mcmaster.ca
2
Assistant Professor, Dept. of Civil Engineering, McMaster Univ.,
Hamilton, ON, Canada L8S 4L7. E-mail: konstant@mcmaster.ca
3
Martini Mascarin and George Chair in Masonry Design, and Director
of McMaster University Institute for Multi-Hazard Systemic Risk Studies,
McMaster Univ., Hamilton, ON, Canada L8S 4L7 (corresponding author).
E-mail: eldak@mcmaster.ca
Note. This manuscript was submitted on June 5, 2015; approved on
January 29, 2016; published online on May 11, 2016. Discussion period
open until October 11, 2016; separate discussions must be submitted for
individual papers. This paper is part of the Journal of Performance of
Constructed Facilities, © ASCE, ISSN 0887-3828.
2014).
In this respect, because of the large estimated repair costs, a
significant number of reinforced masonry shear wall (RMSW)
buildings were deemed unusable following the 2011 Christchurch
earthquake, even after realizing that the damage was repairable.
Haach et al. (2010) studied the seismic behavior of RMSW tested
using truss-type vertical and horizontal reinforcement. The study
showed that the presence of horizontal reinforcement improved
the distribution of cracks, even if the increase in lateral strength
was marginal. Shedid et al. (2010a, b) performed cyclic tests on
scaled RMSW with various configurations and aspect ratios and
analyzed the walls’ seismic behavior. Shedid et al. (2010a) con-
cluded that higher ductility values than those currently adopted
in the existent North American codes can be realized by walls with
rectangular cross sections. Shedid et al. (2010b) also concluded that
flanged and end-confined RMSW can be cost-effective alternatives
to enhance the seismic performance of midrise RMSW construc-
tion in North America. Banting and El-Dakhakhni (2012) reported
on the performance of RMSW detailed with confined boundary el-
ements and subjected to fully reversed cyclic loading. The study
showed experimental evidence of the superior seismic performance
of this type of RMSW seismic force resisting systems (SFRSs) com-
pared to RMSW with typical rectangular cross sections.
Heerema et al. (2014) studied the seismic response of an asym-
metrical RMSW SFRS and investigated how it compares to the cor-
responding individual wall behaviors. The study provided useful

© ASCE 04016047-1 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 benchmarking data that contributes to the understanding of the
relation between component- and system-level performance of
RMSW SFRS, respectively. Ahmadi et al. (2014) performed an ex-
perimental study to analyze the seismic behavior of cantilever
RMSW. The study concluded that similar behavior exists between
walls constructed with recycled units and ordinary units. Moreover,
it showed that lap splices in the longitudinal reinforcement can neg-
atively affect the RMSW performance and that walls with concen-
trated longitudinal reinforcement behaved similarly to walls with
distributed reinforcement. The two previously reported studies by
Siyam et al. (2015a, b) evaluated force-, displacement-, and perfor-
mance-based seismic design parameters of RMSW classified as
ductile shear walls/special reinforced walls according to the CSA
S304-14 (CSA 2014) and the American MSJC-13 (MSJC 2013)
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.
masonry design codes. The studies highlighted the fact that RMSW
which are detailed following the same prescriptive code require-
ments for ductile shear walls SFRSs can experience significantly
different force-, displacement- and performance-based seismic
design parameters at the component level depending on the wall
configuration.
Recent studies focused on collapse fragility assessment for dif-
ferent types of SFRSs subjected to different earthquake effects.
Azarbakht et al. (2012) investigated the effects of ground motion
spectral shape parameters on collapse fragility curves. The study
proposed a closed-form formulation to predict the collapse capacity
of structures as a function of their structural behavior parameters
(i.e., spectral shape parameters). Li and Van De Lindt (2014) de-
veloped collapse fragility curves for a special moment-resisting
steel frame with fully restrained reduced beam sections after being
subjected to damage from a main shock. This was done to inves-
tigate the effect of aftershocks on these types of structures. The
study concluded that the structural collapse capacity might reduce
significantly when the building is subjected to high intensity main
shocks. Nazari et al. (2015) conducted a similar study on a two-
story wood-frame house to investigate the effect of aftershocks on
these types of structures. The study showed that the building model
developed to represent the wooden structure was not significantly
affected from the aftershocks if it survived the main shock. This
indicated that the effect of aftershocks on the collapse of low-rise
to midrise wood-frame buildings is not as significant as was per-
ceived. Purba and Bruneau (2015) conducted collapse assessment
of steel plate shear walls having infill plates using two different
design philosophies. They concluded that infill plates for the walls
should be designed to resist the total specified story shears and that
walls designed by sharing story shears between the boundary frame
and infill plates will undergo significantly larger drifts (Purba and
Bruneau 2015).
To the best of the authors’ knowledge there is no published
study that assesses the seismic performance of RMSW SFRSs us-
ing the FEMA P-58 and P-695. Considering a complete SFRS
(i.e., system-level) is thought to align better with the objectives
of the next generation performance-based seismic design guide-
lines since the seismic performance of the whole building system
is investigated. Currently, there is no established performance-
based evaluation methodology to quantify the seismic performance
of RMSW SFRS in Canada. As such, the FEMA P-58 and P-695
procedures were followed using Canadian design spectrum-
consistent ground motions that have been used to generate seismic
hazard curves (Assatourians and Atkinson 2010) for the NBCC,
(NRCC 2010).
As building collapse is the main cause of earthquake casualties,
the FEMA P-58 (ATC 2012) presents a procedure for collapse fra-
gility assessment and the development of collapse fragility curves,
which describes the probability of incurring structural collapse as a
© ASCE 04016047-2
J. Perform. Constr. Facil., 04016047
function of ground motion intensity (ATC 2012). The procedure
involves the selection and scaling of a suite of ground motions.
In this study, a total of 15 simulated ground motion pairs were
used and matched to the NBCC (NRCC 2010) target design re-
sponse spectrum, corresponding to a highly seismic site in Victoria
(Gonzales Heights), British Columbia. A considerable portion of
the study focused on the development of an analytical model that
was used to perform an incremental dynamic analysis (IDA) using
the selected ground motion suite. The use of IDA as a method for
collapse assessment of structures is explained in detail by Vamvat-
sikos and Cornell (2002). Results from the IDA were then used to
generate collapse fragility curves for the RMSW SFRS. The fol-
lowing sections discuss the RMSW building design and configu-
ration and give a brief summary of the experimental program and
test results presented in Siyam et al. (2015a, b). The paper also
presents the details of the developed inelastic model and the
relevant analysis methodologies utilized to evaluate the collapse
fragility and the adjusted collapse margin ratio of a typical RMSW
building.
Building Design Configuration
The RMSW SFRS shown in Fig. 1 was used in the study. The build-
ing plan shows that the SFRS consists of four different RMSW con-
figurations: W1, W2, W5, and W6. For ease of cross-referencing,
the same wall designation, presented in Siyam et al. (2015a) was
used herein. Walls W1, W5, and W6 have rectangular cross-
sections, while W2 is flanged. The walls shown at the corners of
the building are not connected, to maintain consistency with the
walls tested in the experimental program. Fig. 1 shows that the
SFRS in the N-S direction is composed of Walls W1, W5, and
W6a. Table 1 lists details pertaining to each wall where all walls
are classified as ductile and special according to CSA S304-14
(CSA 2014) and MSJC-13 (MSJC 2013), respectively. As shown
in Table 1, the horizontal reinforcement ratio for all walls is
0.26% at the first story, and 0.14% at the second story, conforming
to capacity design principles. The seismic forces were transferred to
the shear walls through a rigid diaphragm action in the 20-cm-thick
reinforced concrete (RC) floor slabs. The building was designed
according to the NBCC (NRCC 2010) and CSA S304-14 (CSA
2014) code requirements.
Summary of Previous Work
The experimental program in Siyam et al. (2015b) focused on
evaluating the seismic response of six fully-grouted scaled RMSW
subjected to reversed cycles of quasi-static loading. The aspect ra-
tios and steel reinforcement ratios of the one-third-scale model
walls are the same as those of the building walls, presented in
Table 1, in which the vertical and horizontal steel ratios, and the
vertical and horizontal bar diameters, are denoted by ρv, ρh , dv ,
and dh , respectively. In general, the experimental results showed
that the ductile/special RMSW failed in a flexural manner reaching
a displacement ductility between 5.4 and 7.6 at 20% strength deg-
radation. The results also showed that RMSW that were detailed
following the same ductile/special RMSW classification, and hav-
ing the same overall aspect ratio and reinforcement ratio, could ex-
perience significantly different seismic design parameters and may
have different ductility capacities depending on different wall con-
figurations. Further details of the test program, the experimental
results, and the evaluated seismic design parameters can be found
in Siyam et al. (2015a, b).
J. Perform. Constr. Facil.
 25 m 30 m

6.6 m

(a)
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

N-S
W2

W6 W6a W1

W5

(b)

Fig. 1. (a) Archetype full-scale masonry building; (b) 3D top view

Table 1. Building Wall Details and Specifications

Vertical Horizontal
reinforcement reinforcement
Height Length Aspect Masswall = CSA shear wall
Wall Type (mm) (mm) ratio mass1 st floor ρv (%) ρh1 (%) ρh2 (%) classification
W1 Rectangular 6,400 4,590 1.4 0.059 0.6 0.26 0.14 Ductile
W2 Flanged 6,400 4,590 1.4 0.062 0.6 0.26 0.14 Ductile
W5 Rectangular 6,400 1,790 3.6 0.023 0.6 0.26 0.14 Ductile
W6 Rectangular 6,400 1,390 4.6 0.018 0.6 0.26 0.14 Ductile
W6a Rectangular 6,400 1,590 4.6 0.021 1.2 0.26 0.14 Ductile

Analytical Model
Model Development and Modeling Process
A modeling process, shown in Fig. 2, was established to create the
RMSW SFRS model representing the N-S SFRS of the building in
Fig. 1. The process starts by development of the scaled wall stick
models using the OpenSees platform to simulate the inelastic flexu-
ral behavior of the walls. In step 1, numerical models of the scaled
walls are created in OpenSees. In steps 2 and 3, calibration of the
model is carried out by comparing the results of the numerical
model for Wall W5 to the experimental hysteresis results for that
wall. The model parameters that are adjusted include some of the
parameters in the material models, as well as the number of ele-
ments used in the numerical model. If the model compares well
with the experimental results (step 4), then the next step is to model
the other scaled walls. Otherwise, modeling returns to step 1. In
step 5, the calibrated scaled wall models are scaled up to represent
full-scale walls. Step 6 involves checking the flexural capacity of
the wall by comparing full-scale model pushover analysis with
flexural analysis using mechanics. Step 7 involves assembling the
RMSW SFRS model using the full-scale wall stick models to
represent the RMSW building in Fig. 1. Step 7 also involves the
selection and scaling of ground motions in parallel to the assembly
of the RMSW SRFS model. Step 8 performs the IDA to generate
IDA curves. Step 9 involves postprocessing of results obtained
from the IDA.
The numerical model adopted in the current study is a 1D fiber-
based macromodel originally developed by Ezzeledin et al. (2014).
Macromodeling was adopted because the study focuses on the
global response of the SFRS, and because the model was to be
used in a large number of nonlinear response history analyses
(NLRHA). Moreover, macromodeling of shear walls using fiber
elements have shown considerable accuracy when modeling shear
wall systems (Waugh and Sritharan 2010). Each fiber element is
a displacement-based beam-column element. The element is based

© ASCE 04016047-3 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Modeling process used in this study

on the displacement formulation and therefore considers the spread
of plasticity along the element (OpenSees). Additionally, this for-
mulation assumes a constant axial strain and a linear curvature dis-
tribution. Most fiber elements do not account for the effect of shear
deformations that occur due to lateral load, although experimental
research shows that the flexure and shear displacements are
coupled for most of the walls, even for walls with relatively high
aspect ratio (Massone and Wallace 2004). Subsequently, it was
necessary to account for shear deformations in the adopted model.
Therefore, the shear deformations in the walls were aggregated
using a uniaxial material model available in the OpenSees platform
(Pinching4 material) to facilitate accurate predictions of wall
displacements. Pinching4 is a one-dimensional hysteretic load
deformation response model that involves a response envelope,
an unload-reload path, and three damage rules that control the evo-
lution along these paths (Lowes et al. 2003). Global material
response parameters describing the walls’ load-displacement enve-
lopes are estimated and used as input for the Pinching4 material.
The global force parameters indicated in the Pinching4 mate-
rial model as ePf1, ePf2, ePf3, and ePf4 were estimated using
mechanics-based flexural analysis of RMSW including the
self-weight of the walls. The model displacements parameters des-
ignated as ePd1, ePd2, ePd3, and ePd4 were estimated using de-
flection calculation of cantilever walls with top load application
taking into account flexural and shear displacements when calcu-
lating the gross stiffness of the walls.
As shown in Fig. 3, the RMSW building’s SFRS is represented
by six walls aligned along the building’s N-S direction. Since the
building is symmetrical, only half of the building was modeled.
Each wall is discretized into seven elements in total: five ele-
ments distributed along the first floor, one element representing
the top story, and a zero-length section element added at the
wall-foundation interface (Fig. 3). The rationale for this discreti-
zation is accurate and efficient modeling of the plastic hinging
mechanism that develops in the bottom story and the tensile strain
penetration that occurs below the wall base level. The model had a
total of 48 nodes, 42 displacement beam-column elements, 10 elas-
tic truss elements and 12 lumped masses. The masonry was mod-
eled using the Concrete 7 model, which represents the simplified
Chang and Manders (1994) material. The vertical reinforcement

Rigid links
7 14 21 28 35 42

lumped mass element 6

6 13 20 27 34 41
5 12 19 26 33 element 5
40
18 25 32 39 element 4 beam-column
4 11
element 3 elements
3 10 17 24 31 38
23 37 element 2
2 9 16 30
8 22 29 36 element 1
1 15
43 44 45 46 47 48
zero section element
W1 W5 W5 W5 W5 W6a
(a) (b)

Fig. 3. Model discretization in OpenSees: (a) N-S SFRS model; (b) wall model

© ASCE 04016047-4 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 Table 2. Mass and Axial Load Assigned to Walls at Each Floor Level
Floor level Mass number Areamass ðm2 Þ Areaaxial ðm2 Þ Weight (kN) Mass (t) Axial load (kN) Nodes in OpenSees
First floor m1 19.64 12.63 80.01 8.16 109.38 6
m2 20.66 18.82 84.14 8.58 145.97 13
m3 20.66 18.82 84.14 8.58 145.97 20
m4 19.64 12.63 80.01 8.16 109.38 27
m5 266.75 53.67 1,086.40 110.74 341.27 34
m6 13.83 0.00 56.35 5.74 45.00 41
Roof level m7 19.64 12.63 75.79 7.73 82.65 7
m8 20.66 18.82 79.71 8.13 115.25 14
m9 20.66 18.82 79.71 8.13 115.25 21
m10 19.64 12.63 75.79 7.73 82.65 28
m11 266.75 53.67 1,029.19 104.91 294.32 35
m12 13.83 0.00 53.38 5.44 45.00 42
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

was modeled using the Steel 02 material model, which represents fy , and the strain hardening coefficient, h. The parameters R0,
uniaxial Giuffre-Menegotto-Pinto steel material (Menegotto and cR1, and cR2 were calibrated with values shown in Table 3 to
Pinto 1973). Bond_SP01 material was added to model the effect of best simulate the cyclic behavior of the rebars. The parameters in
tensile strain penetration of the vertical reinforcement. The param- Bond_SP01 material include the yield and ultimate strengths, fy
eters used in each material model are explained in the subsequent and f u , respectively, the rebars slip at member interface under yield
section. stress, Sy , the rebars slip at the loaded end at the bar fracture
The following assumptions were made in the development of strength, Su , the initial hardening ratio in the monotonic slip versus
the fiber-based element model: bar stress response, g, and the pinching factor for the cyclic slip
• The walls were supported on a rigid foundation system; versus bar response, R. Parameters g and R were calibrated using
• Floor masses were assumed to be lumped at each floor level at the experimental results.
the center of the walls. Assuming a total area of the building of The Pinching4 material was also used to describe the load deg-
750 m2 , the mass and axial load assigned to each wall at each radation behavior past the peak load and the pinching behavior
node is listed in Table 2. The axial loads were calculated accord- within the hysteresis loops. Discrete load and displacement points
ing to the tributary area method using the NBCC 2010 (NRCC were estimated in the Pinching4 material to represent the limits of
2010) load factors and combination; whereas the seismic masses the backbone curve. As mentioned previously, the load estimations
were distributed as a function of the wall stiffness values. Values of the Pinching4 material were calculated using flexural analysis
of the ratio of the mass of each wall to the mass of the first floor of the walls while the displacement points were estimated using
are listed in Table 1; elastic deflections of the walls using the gross stiffness and ac-
• The floor slab systems were modeled using rigid elastic truss counting for flexural and shear deformations. Moreover, the num-
elements with high axial stiffness. Table 2 shows that the elastic ber of elements representing the first floor portion of the wall
stiffness (Eslab ) of the slabs is almost 20 times as stiff as the
was calibrated to simulate the hinging mechanism at the bottom
walls (Em ). Such floor-slab systems will cause all the nodes
section of the walls. A total number of six elements (excluding
within the same floor level to displace equaly; and
• Only the horizontal components of the ground motions were
considered in the NLRHA. Table 3. Material Model Parameter Values Used in This Study
Materials Properties Values
Model Parameters Evaluation and Calibration
Concrete masonry (concrete07) f m0 ðMPaÞ 13.5
It is important to make the distinction between the strength param- εm ðmm=mmÞ 0.002
eters that were used in the scaled- and full-scale models (steps 1 and ft ðMPaÞ 0.5
5 in Fig. 2). In the scaled model, material parameters obtained from εt ðmm=mmÞ 0.00011
experimental testing of the masonry constituents were used. On the Em ðMPaÞ 11,475
other hand, for the full-scale model, design values from Canadian xp 2
code provisions were used to define the material parameters. xn 2.3
The Concrete 7 material model in OpenSees was used to define r 2.3
Eslab ðMPaÞ 200,000
the masonry material with values of f m0 and strain at peak load, εm .
Currently there is no specific material model developed to simulate Vertical reinforcement (steel02) f y ðMPaÞ 500
the behavior of masonry, and therefore the Concrete 7 model in Es 200,000
OpenSees was used for this purpose. In this concrete model h 0.0025
(step 2 in Fig. 2), the parameters xp and xn shown in Table 3 were R0 10
calibrated to accurately model the masonry material. These two cR1 0.925
cR2 0.15
parameters define the strain at which the straight line descent
begins in tension and compression, respectively. The r parameter Bond steel (bond_SP01) f u (MPa) 600
accounts for the inelastic descending branch in the curve. It is im- Sy 0.35
portant to note that values of fm0 and εm in Table 3 correspond to the Su 13.2
full-scale wall model (step 5 in Fig. 2). These values were used in g 0.4
the design calculations of the RMSW building as discussed earlier. R 0.6
The vertical steel material parameters included the yield strength, Note: Material models are defined in OpenSees documentation.

© ASCE 04016047-5 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 100
the zero-length element) were thought to be sufficient after per-
Experimental Wall W1
forming sensitivity analyses. Fig. 3 shows the layout of the ideal- 80
ized building model. Numerical Model
60
The calibrated numerical model was compared to the experi-
40

Lateral Load (kN)

mental results of the two-story RMSW (hysteresis loops) presented
in Siyam et al. (2015b). This fiber-based model was used to gen- 20
erate hysteretic relationships for three rectangular scaled walls 0
(W1, W5, and W6) of the test specimens presented in Siyam et al. -20
(2015b). The same loading protocol used for the experimental pro-
-40
gram was used to compare the numerical and experimental models.
Fig. 4 shows comparisons between the experimental and the -60
numerical model hysteresis loops for the walls. In addition, Table 4 -80
compares the model strength predictions with the corresponding -100
experimental values at specified displacement cycles given as the
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

-50 -40 -30 -20 -10 0 10 20 30 40 50

input to the model. As shown in Fig. 4 and listed in Table 4, (a) Top Displacement (mm)
the model captures the experimental response with good accuracy
(10–30% for Walls W5 and W6, which have high aspect ratios. The 20
numerical result for Wall W1 matches the experimental ones for the Experimental Wall W5
15
first five cycles, and then the accuracy decreases for the last cycle
(i.e., sixth). In general, the model is capable of capturing the inelas- Numerical Model
10
tic behavior of the RMSW. As described earlier in the modeling

Lateral Load (kN)

process, after calibration, step 5 commenced where the walls were 5
scaled up to represent full-scale walls. Pushover analysis was per-
formed on the full-scale wall models and compared with theoretical 0

calculation of flexural strength. The results are presented in Table 5,

-5
showing the agreement between the yield and ultimate lateral load
capacity predictions of the S304-14 code and the numerical model. -10
Step 7 was then performed in order to assemble the building model
shown in Fig. 3. At this stage, selection and scaling of ground mo- -15
tions were performed, as discussed in the next section.
-20
-80 -60 -40 -20 0 20 40 60 80
(b) Top Displacement (mm)
Analysis Methodology and Discussion
12
Experimental Wall W6
Selection of Ground Motions
8
Numerical Model
A set of representative ground motion pairs were used to estimate
the intensity measure (IM) adopted within the seismic fragility
4
Lateral Load (kN)

assessment, following the FEMA P-58 (ATC 2012) guidelines.

Three sets of 45 simulated ground motions from Assatourians and
Atkinson (2010) were used as a basis to obtain several Canadian 0
design spectrum-consistent motions for the region of interest. From
a total of 135 simulated motions, 15 pairs (Table 6) were selected -4
for the IDA. The characteristics of the selected motions reflected
the seismicity in the region of Western Canada (shallow crust or -8
within the underlying subducting slab earthquakes) as described
in Adams and Halchuk (2003) and Adams and Atkinson (2003).
-12
In the current study, the intensity-based performance assessment -100 -80 -60 -40 -20 0 20 40 60 80 100
procedure in FEMA P-58 (ATC 2012) was used for ground motion (c) Top Displacement (mm)
selection and scaling. This procedure involves evaluating the per-
formance of the RMSW SFRS aligned along the N-S direction. The Fig. 4. Numerical model validation of force-displacement relation-
procedure first requires the selection of a target design spectrum to ships: (a) W1; (b) W5; (c) W6
which the response spectra of the simulated ground motions are to
be matched (Table 6). The Western Canada suite contains elastic
response spectra with 5% damping and their corresponding time qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
histories at magnitudes ranging between 6.5 and 7.5, for site class Sgm ðTÞ ¼ Sx ðTÞ Sy ðTÞ ð1Þ
C and focal distance ranging between 10 and 100 km. Then the
geometric mean spectrum, Sgm , given by Eq. (1) is computed for The target response spectrum is shown in Fig. 5. The selected
each ground motion pair over the period range between T min ¼ ground motion pairs had geometric mean response spectra with
0.04 s and T max ¼ 0.54 s (Fig. 5), where Sx and Sy are orthogonal shapes similar to the target spectrum over the range of interest
components of spectral acceleration at period T. The notation Sgm is (T min ≤ T ≤ T max ). The average mean response spectrum, along
used since most of ground motion prediction equations provide with that for the 16th and 84th percentiles, are also shown in Fig. 5.
geometric mean spectral response accelerations (ATC 2012) The final step involves scaling each ground motion pair by the ratio

© ASCE 04016047-6 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 Table 4. Experimental and Numerical Model Comparison (Cyclic Analysis)
Cycle
Wall Displacement/load Model 1 2 3 4 5 6
Wall W1 Displacement (mm) (input) Experimental 6.4 12.8 19.4 25.6 32 38.5
Numerical 6.4 12.8 19.4 25.6 32 38.5
Lateral load (kN) (output) Experimental 68.4 90.7 89.8 85.7 78.1 53.1
Numerical 78.8 95.5 85.5 79.9 77.0 75.0
Experimental/numerical 0.9 0.9 1.1 1.1 1.0 0.7
Wall W5 Displacement (mm) (input) Experimental 9.5 18.9 28.3 37.6 47.1 56.4
Numerical 9.5 18.9 28.3 37.6 47.1 56.4
Lateral load (kN) (output) Experimental 9.9 15.6 14 13.95 13.7 12
Numerical 10.8 14.4 13.2 11.32 10.8 11.2
Experimental/numerical 0.9 1.1 1.1 1.2 1.3 1.1
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

Wall W6 Displacement (mm) (input) Experimental 14.9 29.5 44.3 59.2 73.8 —
Numerical 14.9 29.5 44.3 59.2 73.8 —
Lateral load (kN) (output) Experimental 6.8 9.9 8.4 7.56 7.21 —
Numerical 7.7 9.34 7.97 7.65 7.65 —
Experimental/numerical 0.9 1.1 1.1 1.0 0.9 —

Table 5. Wall Capacities Using CSA S304-14 Code Provisions and Full of Sa ðT̄Þ to the Sgm ðT̄Þ, where T̄ ¼ 0.22 s is the average funda-
Scale Numerical Model mental period of the orthogonal directions (x and y) of the building.
Lateral load Lateral load The scaled ground motions are then used in the IDA of the N-S
capacity CSA capacity RMSW SFRS.
S304-14 numerical model
Yield Ultimate Yield Ultimate Estimation of System Capacity: Pushover and
Wall Length Thickness Area (kN) (kN) (kN) (kN) Hysteretic Relationship
W1 4,590 190 872,100 772 966 798.1 967.1 The system pushover curve for N-S RMSW is plotted in Fig. 6(a).
W5 1,790 190 340,100 118 148 119.3 144.5
The curve was computed by subjecting the N-S SFRS to constant
W6 1,390 190 264,100 77 93 76.7 92.2
displacement increments of 6 mm imposed at the roof level (Fig. 3).

Table 6. Ground Motion Records Used in IDA

Record Record Earthquake # as per Spectral Soil Fault Scale
number designation Assatourians and Atkinson (2010) ordinates Magnitude class distance (km) Sgm ðT 1 ÞðgÞ factor
1 west6c1 22 x 6.5 C 11.2 0.73 0.89
2 west6c1 23 y 6.5 C 11.2 0.73 0.89
3 west6c1 25 x 6.5 C 11.2 0.67 0.98
4 west6c1 26 y 6.5 C 11.2 0.67 0.98
5 west6c1 37 x 6.5 C 13.0 0.53 1.22
6 west6c1 38 y 6.5 C 13.0 0.53 1.22
7 west6c1 43 x 6.5 C 13.0 0.52 1.25
8 west6c1 44 y 6.5 C 13.0 0.52 1.25
9 west6c2 1 x 6.5 C 19.7 0.52 1.24
10 west6c2 2 y 6.5 C 19.7 0.52 1.24
11 west6c2 10 x 6.5 C 21.6 0.34 1.90
12 west6c2 11 y 6.5 C 21.6 0.34 1.90
13 west6c2 13 x 6.5 C 21.6 0.39 1.66
14 west6c2 14 y 6.5 C 21.6 0.39 1.66
15 west6c2 19 x 6.5 C 14.6 0.62 1.05
16 west6c2 20 y 6.5 C 14.6 0.62 1.05
17 west6c2 22 x 6.5 C 25.8 0.47 1.39
18 west6c2 23 y 6.5 C 25.8 0.47 1.39
19 west6c2 25 x 6.5 C 26.3 0.49 1.32
20 west6c2 26 y 6.5 C 26.3 0.49 1.32
21 west6c2 31 x 6.5 C 30.0 0.38 1.73
22 west6c2 32 y 6.5 C 30.0 0.38 1.73
23 west6c2 34 x 6.5 C 31.1 0.29 2.25
24 west6c2 35 y 6.5 C 31.1 0.29 2.25
25 west7c2 1 x 7.5 C 47.4 0.45 1.43
26 west7c2 2 y 7.5 C 47.4 0.45 1.43
27 west7c2 7 x 7.5 C 48.8 0.43 1.52
28 west7c2 8 y 7.5 C 48.8 0.43 1.52
29 west7c2 10 x 7.5 C 50.7 0.29 2.25
30 west7c2 11 y 7.5 C 50.7 0.29 2.25

© ASCE 04016047-7 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 1.4
of NLRHA in which the ground motion IM selected for collapse
Victoria (Gonzalez Heights) RS investigation is incrementally increased until the collapse of the
1.2 RS of Simulated Ground Motions
SFRS is realized. Although multiplication an actual ground motion
16th, 50th, 84th percentile
record by a scale factor does not necessarily result in a physically
Spectral Acceleration, Sa (g)

1 Tmin
Tmax realizable earthquake motion, it facilitates a better understanding
of the behavior of inelastic systems over a range of intensities
0.8
(Konstantinidis and Nikfar 2015). The analysis results in what is
known as IDA curves, also known as dynamic pushover curves,
0.6
which are relationships between IM and the engineering demand
parameter (EDP). Such curves are produced by subjecting the struc-
0.4
ture to multiple ground motions, with each ground motion scaled
to multiple levels of intensity (Vamvatsikos and Cornell 2002).
0.2
The IM chosen in this study is the spectral acceleration (5%
damping) at the fundamental period of the structure (T 1 ¼ 0.2s) de-
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

0
0 0.5 1 1.5 2 2.5
Period of Vibration, T (s)
Fig. 5. Scaled response spectra pairs from simulated Western Canada
earthquakes (data from Assatourians and Atkinson 2010)
The graph also shows the pushover curve obtained from superim-
posing the pushover curves of the individual walls as shown in
Fig. 6(b). The curve corresponding to the RMSW SFRS shows
a higher ultimate load capacity (normalized base shear of 0.52)
compared to the summation of the lateral load capacities at the cor-
responding IDRmax of 0.47%. The hysteresis loops from cyclic
analysis of the model are also shown in Fig. 6(c), using the loading
protocol illustrated in the lower-right corner of the figure. The lat-
eral load capacity from cyclic analysis was lower than the pushover
curve by 15.4% at IDRmax ¼ 0.47% indicating more damage in-
duced in the structure due to repeated cycles.
Incremental Dynamic Analysis for Building Collapse
Capacity and Response Histories
Inelastic demands for the RMSW building used in this study
were determined using IDA. The method was first developed by
Luco and Cornell (1998) and explained in detail by Vamvatsikos
and Cornell (2002). The method adopts a parametric analysis ap-
proach to estimate structural performance under seismic loads
(Vamvatsikos and Cornell 2002). It involves performing a series
0.6
= 4.0
T
0.4
Normalized Base Shear
3 3.5 4 4.5 5
noted as Sa (T 1 , 5%). The code-defined empirical period of vibration
for shear wall structures (T n ¼ 0.05h0.75 n ) provided in the NBCC
(NRCC 2010) was used to calculate T 1 . Ten intensity scale factor
values, ranging from 1.0 to 5.5, with 0.5 increments, were chosen
to generate the IDA curves starting with Sa ðT 1 ; 5%Þ ¼ 0.7 g. The
maximum inter-story drift ratio (IDRmax ), was used as the EDP in
this study. The IDA curves for the 30 ground motions along with
16th, 50th, and 84th percentile IDA curves are illustrated in Fig. 7.
The interpretations from the data are summarized as follows:
• The curves show that the N-S RMSW SFRS remains linear
elastic at IM ¼ 0.7 g, reaching a maximum IDR of approxi-
mately 0.2%;
• An increase in level of intensity does not necessarily result in a
monotonic increase in IDR demand (Fig. 7). In some cases, the
SFRS exhibits a severe softening behavior with increasing IM,
while in other cases the SFRS exhibits hardening (Fig. 7);
• The collapse median intensity, θ, can be quickly estimated by
locating the intersection of the 50th percentile IDA curve with
the vertical line of the x-axis (IDRmax ) that defines collapse.
This is depicted in Fig. 7, where θ was estimated roughly to be
2.7 g; and
• Results from the IDA curves were used to generate collapse frag-
ility curves after defining the specific collapse state of the structure.
Fig. 8 shows the computed weight-normalized base shear
against the IDR response for two ground motions (records 1 and 6
in Table 6) with increasing levels of intensity: 0.7, 1.6, and 2.6 g.
The figure shows the transition of the structure into the inelastic
range with increasing IM. At IM ¼ 0.7 g, the structure remains
elastic, reaching maximum IDR of 0.2%. As the IM increases to
1.6 g, the structure deforms inelastically past its maximum load
(a)
(b)
(c) 3.5

3.0
Sa (T1=0.2s, 5% Damping) (g)

0.2 Intermediate
collapse region
2.5 severe
0
hardening case
200

150 2.0
100
Displacement (mm)

severe
-0.2 50
softening case
0
1.5
-50

-0.4
-100
1.0
-150 Single IDA Curve
-200
0 10 20 30 40 50 NBCC interstory drift limit
Number of Cycles 0.5
-0.6 16th, 50th and 84th percentile
-3 -2 -1 0 1 2 3
IDRmax (%) 0.0
0 2 4 6 8 10
Fig. 6. Force-displacement relationships: (a) pushover curve of system; IDRmax (%)
(b) pushover curve of individual walls superimposed; (c) cyclic hyster-
esis loops Fig. 7. Incremental dynamic analysis (IDA) curves for N-S SFRS

© ASCE 04016047-8 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 0.6  
lnðx=θÞ
PðCjIM ¼ xÞ ¼ Φ ð2Þ
β
0.4
where PðCjIM ¼ xÞ = probability of collapse given an intensity
measure, i.e., IM ¼ Sa ðT 1 ; 5%Þ = x; Φ = standard normal cumula-
Normalized Base Shear

0.2 tive distribution function, θ = collapse median intensity (i.e., the IM

level with 50% probability of collapse), and β = standard deviation
0 of the IM. Eq. (2) implies the assumption of lognormality of the
IM values of the ground motions causing collapse of a specific
structure (Baker 2015). This assumption was considered a reason-
-0.2 able assumption according to a number of researchers (Ibarra and
Krawinkler 2005; Comerio 2005; Porter et al. 2007; Bradley and
IM = 0.7 Dhakal 2008; Ghafory-Ashtiany et al. 2010; Eads et al. 2013).
-0.4 IM = 1.6 Collapse fragility curves are used for collapse safety assessment
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

IM = 2.6 of building structures. Using such curves, the probability of collapse

-0.6 at a specific level of earthquake hazard can be predicted. The next
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 section outlines the method used to identify collapse using the IDA.
(a) IDR (%)

0.6 Identifying Collapse State from IDA

Defining collapse of an SFRS is an essential step for creating col-
0.4 lapse fragility curves. According to FEMA P-58 (ATC 2012), an
SFRS collapse is defined as: sidesway failure (lateral dynamic in-
Normalized Base Shear

stability), loss of vertical-load-carrying capacity, or exceedance of

0.2
0
-0.2
-0.4
-0.6
-1.5 -1 -0.5 0 0.5 1 1.5
(b) IDR (%)
Fig. 8. N-S RMSW load-displacement relationships for sample ground
motions: (a) record 1; (b) record 6
capacity at IDRmax of 0.5%, resulting in significant residual drifts
from this point on.
Collapse Fragility Assessment
Fragility curves can be generated using several approaches, includ-
ing field observations, experimental testing, numerical simulations,
expert judgment, or a combination of these (Kennedy and Ravindra
1984; Kim and Shinozuka 2004; Calvi et al. 2006; Villaverde 2007;
Porter et al. 2007; Shafei et al. 2011). This study developed analytical
collapse fragility curves based on NLRHA results. In general
terms, fragility functions are mathematical functions that define the
conditional probability that a system or a component will experience
damage equal to or exceeding a specific damage state (DS) as a func-
tion of demand, characterized by the EDP. Recently, collapse fragility
curves obtained from IDA results have been becoming increasingly
popular in seismic assessment procedures (ATC 2012) since protec-
tion against collapse is a crucial objective in performance-based
seismic design. A collapse fragility function gives the probability
that a structure will collapse as a function of the intensity measure.
It can be described mathematically (Baker 2015) as
nonsimulated failure criteria. In this study, two methods were used
to define the collapse of the SFRS. The first method assumes that
collapse occurs when a certain limit along the x-axis of the IDA
curve, i.e., the EDP (which in this study is the IDRmax ), is exceeded
and is referred to as the EDP-based rule. The other method defines
collapse to occur when a certain limit along the y-axis of the IDA
curve, i.e., the IM½Sa ðT 1 ; 5%Þ, has exceeded and is referred to as
IM = 0.7
IM-based rule. The methods are described in detail in Vamvatsikos
IM = 1.6 and Cornell (2002) and are reviewed herein briefly. It should be
IM = 2.6 noted that what is nowadays referred to as the EDP-based rule
was introduced as the EDP-based rule by Vamvatsikos and Cornell
2 2.5 3 3.5 4 (2002). EDP was later changed to EDP to align with the terminol-
ogy used by the Pacific Earthquake Engineering Research (PEER)
Center (Vamvatsikos and Cornel 2005).
Researchers have proposed several types of EDPs for wall com-
ponents. Some involve displacement-based measures such as
maximum drift ratio and maximum top displacement, while others
use energy-based measures that quantify the hysteretic energy at a
given level of demand. Currently, FEMA 58-1 adopts the maxi-
mum interstory drift ratio to assess the level of damage attained
by structural elements and SFRS. To be consistent with FEMA
58-1 guidelines, the collapse point is set to CEDP ¼ 2.5% IDRmax
for EDP-based rule. This value is chosen conforming to the
NBCC-10 (NRCC 2010) IDR interstory drift limit set for collapse
prevention. To be on the conservative side, the lowest IM value for
which CEDP ¼ 2.5% IDRmax is used to identify collapse. Fig. 9
shows that there is approximately a 70% probability that RMSW
will exceed the DS3 level. The three damage states, DS1, DS2,
DS3, are defined in the FEMA P-58/BD-3.8.10 document (ATC
2009a) showing the different failure stages of flexural-critical
masonry shear walls. DS1, DS2, and DS3 characterize slight, mod-
erate, and severe flexure damage, respectively. The damage state
descriptions for RMSW in Table 7 are based on FEMA P-58.
Table 8 shows the identification criteria of each damage state from
load-displacement relationships respectively.
Another collapse assessment approach uses the IM-based rule
to define collapse. In this method, the IDA curve is divided into a
collapse and a noncollapse region, with CIM being the IM value at
which the tangent slope of the IDA curve reaches 20% of the elastic
slope is regarded as collapse (Vamvatsikos and Cornell 2002). The

© ASCE 04016047-9 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 1 shows the collapse fragility curves for the three methods where
0.9 DS collapse is defined using the EDP-based rule, while Fig. 10(b) plots
collapse fragility curves for the IM-based rule.
0.8 DS2
Probability of Exceedance

0.7 Method of Moments

The method of moments involves estimating θ̂ and β̂ by taking the
0.6 DS3 logarithm of each ground motion’s IM value associated with the
0.5 onset of collapse, which can be either EDP- or IM-based. Eqs. (3)
and (4) can be utilized to compute these values (Baker 2015)
0.4
1X n
0.3 ln θ̂ ¼ ln IMi ð3Þ
n i¼1
0.2 DS1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0.1 DS2 1 X n
β̂ ¼ ½lnðIMi =θ̂Þ2 ð4Þ
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

DS3
n − 1 i¼1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 where n = number of ground motions considered, and IMi defines
IDRmax (%) the IM value associated with the onset of collapse for the ith
ground motion. The quantities ln θ̂ and β̂ are estimates of the mean
Fig. 9. Actual data fragility curves
and standard deviation (i.e, the moments) of the normal distribution
of the ln IM values (Baker 2015). In this study, the θ̂ values com-
puted using the EDP- and the IM-based rules are 2.71 g and 2.31 g,
flattening of the IDA curve is used as an indication that collapse has respectively. The corresponding β̂ values are 0.20 and 0.26. The
occurred. As suggested by FEMA 350 (FEMA 2000), an inelastic collapse fragility curves generated using this method were plotted
stiffness level of 20% of the corresponding elastic value marks the with dashed lines in Fig. 10.
capacity point. These two collapse definitions were utilized to gen-
erate collapse fragility curves as explained in the next section. Maximum Likelihood Method
The maximum likelihood fitting method uses the theoretical fragil-
ity function [Eq. (2)] and the binominal distribution function
Collapse Fragility Curve Fitting  
nj z
Using the aforementioned definitions of collapse, the collapse fra- Pðzj collapses in nj ground motionsÞ ¼ pjj ð1 − pj Þnj −zj
zj
gility curves from IDA were computed for the N-S RMSW. The
first step involved the use of the IDA data to estimate the collapse ð5Þ
median intensity, θ̂, and the dispersion of the IM, β̂ (also known as to formulate the likelihood expression (Baker 2015)
record-to-record uncertainty), which are required to fit the fragility
Ym       
function. Three methods were used to estimate θ̂ and β̂; namely, the nj lnðxj =θÞ zj lnðxj =θÞ nj −zj
likelihood ¼ Φ 1−Φ
method of moments, the maximum likelihood method, and the least
j¼1
zj β β
squares regression (Cornell et al. 2002; Baker 2015). The details of
each method are summarized in the following sections. Fig. 10(a) ð6Þ

Table 7. Damage State Description for RMSW (Data from ATC 2012)
Damage state Description Repair measure fully grouted
DS1 slight damage flexure 1. Few flexural and shear cracks with hardly noticeable residual crack widths 1. Cosmetic repair
2. Slight yielding of extreme vertical reinforcement 2. Patch cracks and painteach side
3. No spalling
4. No fracture or buckling of vertical reinforcement
5. No structural significant damage
DS2 moderate damage flexure 1. Numerous flexural and diagonal cracks 1. Epoxy injection to repair cracks
2. Mild toe crushing with vertical cracks or light spalling at wall toes 2. Remove loose masonry
3. No fracture or buckling of reinforcement 3. Patch spalls with non-shrink grout
4. Small residual deformation 4. Paint each side
DS3 severe damage flexure 1. Severe flexural cracks 1. Shore
2. Severe toe crushing and spalling 2. Demolish existing wall
3. Fracture or buckling of vertical reinforcement 3. Construct new wall
4. Significant residual deformation

Table 8. FEMA 58-1 Damage State Identification Criteria from Load-Displacement Curves (Data from ATC 2012)
Damage state Identification criteria
DS1 slight flexural damage When a flexural-critical wall has been loaded to 80% of its peak in plane lateral resistance
DS2 moderate flexural damage When a flexural-critical wall has been loaded to its peak in plane lateral resistance
DS3 severe flexural damage When a flexural-critical wall has been loaded beyond its peak resistance and exhibited a load drop of 20% with
respect to its peak

© ASCE 04016047-10 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 1.0 2

1.5
IM = 1.80(IDRmax) - 0.96
0.8 1
Probability of Collapse

0.5

ln IM
0.6
0

-0.5
0.4
-1
Empirical
-1.5
0.2 Maximum Likelihood
Least square regression -2
Method of moments estimator -1 -0.5 0 0.5 1 1.5
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

0.0 ln IDRmax
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Sa (T1) (g) Fig. 11. Log-log relationship between seismic demand, IM, and
(a)
IDRmax
1.0

dispersion signifying more uncertainty in the θ̂ value. Nevertheless,

0.8 the collapse median intensity values are in excellent agreement
Probability of Collapse

using both collapse rules. The fragility curves generated using the
maximum likelihood method were plotted with solid lines in
0.6 Fig. 10.

Least Squares Regression Method

0.4 The IDRmax is assumed to depend on the IM following a power law
Empirical
IDRmax ¼ aIM b ð8Þ
0.2 Maximum Likelihood
Method of moments estimator From the IDA results, the IM versus the EDP log-log data is
0.0 shown in Fig. 11. Least squares regression (LSR) provides estimates
0 2 4 6 8 of the regression parameters a and b in Eq. (8) as a ¼ expð−0.959Þ
and b ¼ 1.805. Subsequently, Eq. (8) can be expressed as
(b) Sa (T1) (g)
 
1 IDRmax
Fig. 10. Collapse fragility curves for N-S RMSW SFRS using different ln θ̂ ¼ ln ð9Þ
fragility fitting methods and collapse defined by (a) EDP-based rule; b a
(b) IM-based rule
which, assuming IDRmax ¼ CEDP ð2.5%Þ, yields θ̂ ¼ 2.83 g. The
dispersion of demand conditioned on the IM can be estimated
using
assuming interdependence of collapse data between ground mo-
tions. In Eqs. (5) and (6), nj denotes the total number of ground rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn b 2
i¼1 fln ½Δi − lnðaIM Þ g
motions (30 in the current study), zj the number of collapse at a β̂ ≅ ð10Þ
specific intensity IM, and pj the probability that a ground motion n−2
with IM ¼ xj will cause collapse. The goal is to identify the fragil- which gives β̂ ≅ 0.18. With these values for θ̂ and β̂, the fragility
ity function that estimates pj , which can be done using the maxi- curve based on the LSR method is plotted with a dotted line in
mum likelihood method. The method identifies the fragility function Fig. 10(a). Only the EDP-based rule can be used when performing
that gives the highest probability of collapse as obtained from the the LSR method since a fixed collapse value of IDRmax is needed to
IDA (Baker 2015). Estimates of the fragility function parameters, θ solve for θ̂. The curve gives very close approximation to the maxi-
and β are denoted by θ̂ and β̂. Such values are obtained by maxi- mum likelihood method where the fragility parameter values differ
mizing the logarithmic likelihood function described by by 0.7% and 11%, respectively.
Xm      Using the EDP-based rule collapse definition, θ̂ and β̂ lie
nj lnðxj =θÞ
fθ̂; β̂g ¼ arg max ln þ zj ln Φ between 2.71 and 2.81 g, and 0.20 and 0.21, respectively. On
θ;β j¼1
zj β
the other hand, following the IM-based rule collapse definition,
  
lnðxj =θÞ θ̂ and β̂ lie between 2.31 and 2.76 g, and 0.26 and 0.48, respec-
þ ðnj − zj Þ ln 1 − Φ ð7Þ tively. The EDP-based rule shows more robustness as compared
β
to the IM-based rule for collapse definition, where there are minor
where the arg max function represents the point of θ̂ and β̂ for variations between the fragility statistical parameters. Being on the
which the preceding whole expression attains a maximum value. conservative side, the lowest collapse median intensity of 2.31 g
Eq. (7) was evaluated using a nonlinear solver tool generating and the highest dispersion value of 0.48 were used to evaluate
values of 2.81 g and 0.21 for θ̂ and β̂, respectively, following the the performance and base the decision of the seismic vulnerability
EDP-based rule (i.e., CEDP defined at 2.5%); and 2.76 g and 0.48 of the RMSW SFRS using the adjusted collapse margin ratio, as
following the IM-based rule. The IM-based rule shows higher will be discussed in the next section.

© ASCE 04016047-11 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 Table 9. SSF Factors for Different Ductility Levels and Fundamental Table 10. Uncertainty Dispersion Values from Different Sources
Period, T
Uncertainty parameter Value
Period-based ductility (μT )
β DR 0.3
T (s) 1 1.1 1.5 2 3 4 6 ≥8 β TD 0.3
β MDL 0.3
≤ 0.5 1.00 1.05 1.1 1.13 1.18 1.22 1.28 1.33
β RTR 0.4
0.6 1.00 1.05 1.11 1.14 1.2 1.24 1.3 1.36
β TOT 0.66
0.7 1.00 1.06 1.11 1.15 1.21 1.25 1.32 1.38
0.8 1.00 1.06 1.12 1.16 1.22 1.27 1.35 1.41
0.9 1.00 1.06 1.13 1.17 1.24 1.29 1.37 1.44 1.0
1 1.00 1.07 1.13 1.18 1.25 1.31 1.39 1.46
1.1 1.00 1.07 1.14 1.19 1.27 1.32 1.41 1.49 0.9
1.2 1.00 1.07 1.15 1.2 1.28 1.34 1.44 1.52
0.8
1.3 1.00 1.08 1.16 1.21 1.29 1.36 1.46 1.55
1.4 1.00 1.08 1.16 1.22 1.31 1.38 1.49 1.58

Probability of Collapse
0.7
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.

IM-based rule Fragility Curve

≥ 1.5 1.00 1.08 1.17 1.23 1.32 1.4 1.51 1.61 AMCR10%
0.6

Performance Evaluation of RMSW SFRS
Having obtained the collapse median intensity, θ̂, from the fragility
fitting curves, the performance of the RMSW SFRS can be evalu-
ated using the performance criteria outlined in FEMA P-695 (ATC
2009b). Performance evaluation was defined by the collapse
median ratio, CMR. The ratio is formulated by dividing the collapse
median intensity by the ground motion spectral demand, SMT ,
which is equal to 1.2 g at the fundamental period of the structure
θ̂
CMR ¼ ð11Þ
SMT
To be more conservative, the collapse median intensity using
IM-based rule (θ̂¼ 2.31 g) was used to calculate the CMR
(CMR ¼ 1.93). Moreover, in order to account for the effects of
the spectral shape on the collapse capacity of the structure, the col-
lapse margin ratio was modified. This was done by multiplying the
CMR by a factor known as spectral shape factor (SSF) in order to
obtain the adjusted collapse margin ratio, ACMR [Eq. (12)] (ATC
2009b). SSFs depend on the ductility of the system, μT , and the
applicable seismic design category assumed to be Dmax in this case.
A value of μT ¼ 4.0 was obtained from the pushover curve in Fig. 6,
and the value of SSF¼1.22 was obtained from Table 9. Using
ACMR ¼ SSF × CMR ð12Þ
a value of ACMR ¼ 2.35 was obtained. This value was compared
with an acceptable value of ACMR symbolized as ACMR10% . The
ACMR10% value was obtained by considering total system collapse
uncertainty denoted as β TOT [Eq. (13)], which accounts for other
sources of uncertainties in the performance assessment process.
These sources of uncertainties include design requirements, test
data, nonlinear models, and record-to-record uncertainty
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
β TOT ¼ β DR þ β TD þ β MDL þ β̂ ð13Þ
The value of β TOT was calculated to be 0.66 after assuming a
value for uncertainties corresponding to different parameters, as
shown in Table 10. Uncertainty parameters were derived from
the range specified in FEMA P-695 (ATC 2009b). The adjusted
collapse fragility curve is plotted in Fig. 12 after considering β TOT .
As shown in Fig. 12, the same collapse median intensity is pivoted
at 2.31 g but the curve has more dispersion due to increase in the
uncertainty. The calculated ACMR10% value was 2.31. As defined
in FEMA P-695 (ATC 2009b), the system is deemed acceptable
if ACMR is greater than ACMR10% . Based on this criteria, the
system passes (2.35 > 2.31) and therefore one can conclude that
the selected RMSW SFRS, which was designed to meet the
0.5
0.4
0.3 AMCR10% = 2.31/1.0 = 2.31
0.2
10%
0.1
0.0
0 1 2 3 4 5 6 7 8 9
Sa (T1) (g)
Fig. 12. Adjusted collapse fragility curve
prescriptive requirements of the ductile masonry walls classifica-
tion of the CSA S304 (CSA 2014), shows potential capacity against
collapse under high intensity earthquakes in one of the highest seis-
mic zones in Canada.
Conclusions
This study presented a collapse fragility assessment of a two-
story RMSW building located in the region of Victoria (Gonzales
Heights), British Columbia. The seismic performance assessment
process presented in the FEMA P-58 document (ATC 2012) was
followed. The process involves selection and scaling of ground
motions, the development of an analytical model that is used to
perform incremental dynamic analysis, and the development of fra-
gility curves to describe the probability of collapse of the structural
system given a seismic intensity measure, IM.
The analytical model developed is based on a 1D fiber-based
macromodel simulating the inelastic flexural behavior of the walls
originally developed by Ezzeledin et al. (2014). Shear deformations
in the walls were aggregated using a uniaxial material model avail-
able in the OpenSees platform to facilitate accurate prediction of
the wall top displacements. The calibrated numerical model com-
pared well to the experimental results presented in Siyam et al.
(2015b).
The intensity-based assessment was the selected procedure for
ground motion selection and scaling, following the FEMA P-58
(ATC 2012) procedure. The procedure requires a target response
spectrum to which the response spectra of the selected ground mo-
tions records are matched. In this study, the NBCC-10 design spec-
trum for Victoria, British Columbia, was used. From a total of 135
simulated ground motions (Assatourians and Atkinson 2010), a
suite of 30 (15 pairs) was selected. Inelastic demands for a model
masonry building were determined using IDA procedure. Collapse
fragility assessment of the RMSW SFRS by different methods

© ASCE 04016047-12 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 04016047

 (method of moments, maximum likelihood method, and least
squares regression method) was then conducted using IDA data
for EDP-based and IM-based collapse rules.
The EDP-based rule shows more robustness in the fragility
parameters as compared to the IM-based rule for collapse definition
criteria. Conservatively, the lowest collapse median intensity of ap-
proximately 2.31 g with highest dispersion value of 0.48 was used
to evaluate the RMSW SFRS performance using the ACMR. The
system was deemed acceptable since the ACMR was greater than
ACMR10% (2.35 > 2.31). Therefore, it can be concluded that the
selected RMSW SFRS, designed to meet the prescriptive require-
ments of the ductile masonry walls classification of the CSA S304
(CSA 2014), shows potential capacity against collapse under high
intensity earthquakes in one of the highest seismic zones in Canada.
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.
The study thus indicates that properly detailed RMSW buildings
constitute viable SFRSs that can withstand a significant seismic
demand prior to collapse. Looking back on the post-Christchurch
earthquake RMSW building inventory, it might be argued that the
CMR and the ACMR quantification approaches might have pro-
vided additional means (in conjunction with visual inspection
and perceived consequence of damage) to assist in the decision-
making process (in terms of building demolition or repair, and
the extent of the needed repair).
Notation
The following symbols are used in this paper:
ACMR = adjusted collapse margin ratio;
ACMR10% = adjusted collapse margin ratio at 10% probability
of collapse using β TOT ;
a, b = least squares regression parameters;
CEDP = value of the engineering demand parameter
(EDP) at which collapse occurs;
CIM = value of intensity measure at which collapse
occurs;
CMR = collapse margin ratio;
DS1 = slight flexural damage state;
DS2 = moderate flexural damage state;
DS3 = severe flexural damage state;
E = Young’s modulus of the wall’s
material (MPa);
EDP = engineering demand parameter;
Em = Young’s modulus of masonry (MPa);
Es = Young’s modulus of steel (MPa);
f m0 = compressive strength of masonry prism (MPa);
fu = ultimate stress of vertical reinforcement (MPa);
fy = yield stress of vertical reinforcement (MPa);
ft = tensile stress of masonry (MPa);
g = Initial hardening ratio in the monotonic slip
versus bar stress response;
h = strain hardening ratio;
IDRmax = maximum inter-story drift;
IM = intensity measure;
PðCjIM ¼ xÞ = probability of collapse given that the intensity
measure is equal to x;
ni = ground motion i;
nj = total number of ground motion j;
pj = probability that ground motion IM ¼ xj will
cause collapse to the structure;
R, R0, cR1, = parameters to control transition from elastic to
cR2 plastic branch in Steel 02;
r = parameter that controls the nonlinear descending
branch;
© ASCE 04016047-13
J. Perform. Constr. Facil., 04016047
Sa ðT 1 ; 5%Þ = spectral acceleration at fundamental period of
structure for 5% damping;
Sgm ðTÞ = geometric mean spectral acceleration at period
TðgÞ;
Sx ðTÞ = x-direction component spectral acceleration (g);
Sy ðTÞ = y-direction component spectral acceleration (g);
SSF = spectral shape factor;
T = period of vibration (s);
T 1 = fundamental period of vibration (s);
xn = nondimensional term that defines the strain at
which the straight line descent begins in
compression;
xp = nondimensional term that defines the strain at
which the straight line descent begins in tension;
zj = number of collapses at a specific intensity;
θ = collapse median intensity (g);
θ̂ = estimated collapse median intensity (g);
β = dispersion of seismic demand conditioned on
the IM;
β̂ = estimated dispersion of seismic demand
conditioned on the IM;
β DR = design requirements-related collapse uncertainty;
β TD = test data-related collapse uncertainty;
β MDL = modeling related collapse uncertainty;
β TOT = total system collapse uncertainty;
εm = ultimate compressive strain of masonry;
εt = ultimate tensile strain of masonry;
μT = period-based ductility of index archetype model;
and
Φ = standard normal cumulative distribution function,
CDF.
References
Adams, J., and Atkinson, G. (2003). “Development of seismic hazard maps
for the 2003 National Building Code of Canada.” Can. J. Civ. Eng.,
30(2), 255–271.
Adams, J., and Halchuk, S. (2003). “Fourth generation seismic hazard maps
of Canada: Values for over 650 Canadian localities intended for
the 2005 National Building Code of Canada.” Geological Survey of
Canada, Ottawa.
Ahmadi, F., Hernandez, J., Sherman, J., Kapoi, C., Klingner, R., and
McLean, D. (2014). “Seismic performance of cantilever-reinforced
concrete masonry shear walls.” J. Struct. Eng., 10.1061/(ASCE)ST
.1943-541X.0000941, 04014051.
Assatourians, K., and Atkinson, G. (2010). “Database of processed time
series and response spectra for Canada: An example application to study
of the 2005 MN 5.4 Riviere du Loup, Quebec. Earthquake.” Seism. Res.
Lett., 81, 1013–1031.
ATC (Applied Technology Council). (2009a). “Background document:
Damage states and fragility curves for reinforced masonry shear walls.”
FEMA 58-1/BD-3.8.10, FEMA, Washington, DC.
ATC (Applied Technology Council). (2009b). “Quantification of build-
ing seismic performance factors.” FEMA P-695, FEMA, Washington,
DC.
ATC (Applied Technology Council). (2012). “Seismic performance assess-
ment of buildings: Volume 1—Methodology.” FEMA P-58-1, FEMA,
Washington, DC.
Azarbakht, A., Mousavi, G., and Ghafory-Ashtiany, M. (2012). “Adjust-
ment of seismic collapse fragility curves of structures by considering
the ground motion spectral shape effects.” J. Earthquake Eng., 16(8),
1095–1112.
Baker, J. W. (2015). “Efficient analytical fragility function fitting using
dynamic structural analysis.” Earthquake Spectra, 31(1), 579–599.
Banting, B. R., and El-Dakhakhni, W. W. (2012). “Force- and displacement-
based seismic performance parameters for reinforced masonry structural
J. Perform. Constr. Facil.
 walls with boundary elements.” J. Struct. Eng., 10.1061/(ASCE)ST
.1943-541X.0000572, 1477–1491.
Bradley, B. A., and Dhakal, R. P. (2008). “Error estimation of closed-form
solution for annual rate of structural collapse.” Earthquake Eng. Struct.
Dyn., 37(15), 1721–1737.
Calvi, G. M., Pinho, R., Magenes, G., Bommer, J. J., Restrepo-Vélez, L. F.,
and Crowley, H. (2006). “Development of seismic vulnerability assess-
ment methodologies over the past 30 years.” ISET J. Earthquake
Technol., 43(3), 75–104.
Chang, G. A., and Mander, J. B. (1994). “Seismic energy based fatigue
damage analysis of bridge columns: Part 1—Evaluation of seismic
capacity.” NCEER Technical Rep. No. NCEER-94-0006, State Univ.
of New York, Buffalo, NY.
Comerio, M. C., ed. (2005). “PEER testbed study on a laboratory building:
Exercising seismic performance assessment.” Rep. PEER 2005/12,
Downloaded from ascelibrary.org by University Of Florida on 06/02/16. Copyright ASCE. For personal use only; all rights reserved.
Pacific Earthquake Engineering Research Center, Univ. of California,
Berkeley, CA.
Cornell, A. C., Jalayer, F., Hamburger, R. O., and Foutch, D. A. (2002).
“Probabilistic basis for 2000 SAC Federal Emergency Management
Agency steel moment frame guidelines.” J. Struct. Eng., 10.1061/
(ASCE)0733-9445(2002)128:4(526), 526–533.
CSA (Canadian Standards Association). (2014). “Design of masonry struc-
tures.” CSA S304-14, Mississauga, Canada.
Eads, L., Miranda, E., Krawinkler, H., and Lignos, D. G. (2013). “An
efficient method for estimating the collapse risk of structures in seismic
regions.” Earthquake Eng. Struct. Dyn., 42(1), 25–41.
Ezzeledin, M., Lydell, W., Shedid, M., and El-Dakhakhni, W. (2014).
“Numerical modelling of reinforced concrete block structural walls
under seismic loading.” 9th Int. Masonry Conf., International Masonry
Society, Guimarães, Portugal.
FEMA. (2000). “Recommended seismic design criteria for new steel
moment frame buildings.” Rep. No. FEMA-350, Washington, DC.
Ghafory-Ashtiany, M., Mousavi, M., and Azarbakht, A. (2010). “Strong
ground motion record selection for the reliable prediction of the mean
seismic collapse capacity of a structure group.” Earthquake Eng. Struct.
Dyn., 40(6), 691–708.
Haach, V., Vasconcelos, G., and Lourenço, P. (2010). “Experimental
analysis of reinforced concrete block masonry walls subjected to in-
plane cyclic loading.” J. Struct. Eng., 10.1061/(ASCE)ST.1943-541X
.0000125, 452–462.
Hamburger, R. O. (2014). “FEMA P-58 seismic performance assessment of
buildings.” Proc., 10th National Conf. in Earthquake Engineering,
Earthquake Engineering Research Institute, Anchorage, AK.
Heerema, P., Shedid, M., and El-Dakhakhni, W. (2014). “Seismic re-
sponse analysis of a reinforced concrete block shear wall asymmetric
building.” J. Struct. Eng., 10.1061/(ASCE)ST.1943-541X.0001140,
04014178.
Ibarra, L. F., and Krawinkler, H. (2005). “Global collapse of frame struc-
tures under seismic excitations.” John A. Blume Earthquake Engineering
Center, Stanford, CA.
Kennedy, R. P., and Ravindra, M. K. (1984). “Seismic fragilities for nuclear
power plant risk studies.” Nucl. Eng. Des., 79(1), 47–68.
Kim, S.-H., and Shinozuka, M. (2004). “Development of fragility curves of
bridges retrofitted by column jacketing.” Probab. Eng. Mech., 19(1–2),
105–112.
Konstantinidis, D., and Nikfar, F. (2015). “Seismic response of sliding
equipment and contents in base-isolated buildings subjected to broad-
band ground motions.” Earthquake Eng. Struct. Dyn., 44(6), 865–887.
Li, Y., Song, R., and Van De Lindt, J. (2014). “Collapse fragility of steel
structures subjected to earthquake mainshock-aftershock sequences.”
J. Struct. Eng., 10.1061/(ASCE)ST.1943-541X.0001019, 04014095.
Lowes, L. L., Mitra, N., and Altoontash, A. (2003). “A beam-column joint
model for simulating the earthquake response of reinforced concrete
© ASCE 04016047-14
J. Perform. Constr. Facil., 04016047
frames.” Pacific Earthquake Engineering Research Rep., Univ. of
California, Berkeley, CA.
Luco, N., and Cornell, C. A. (1998). “Effects of random connection frac-
tures on demands and reliability for a 3-storey pre-Northridge SMRF
structure.” Proc., 6th U.S. National Conf. on Earthquake Engineering,
Earthquake Engineering Research Institute, Seattle.
Massone, L., and Wallace, J. (2004). “Load-deformation responses of
slender reinforced concrete walls.” ACI Struct. J., 101(1), 103–113.
Menegotto, M., and Pinto, P. E. (1973). “Method of analysis of cyclically
loaded RC plane frames including changes in geometry and non-elastic
behavior of elements under normal force and bending.” Proc., IABSE
Symp. on Resistance and Ultimate Deformability of Structures Acted on
Well Defined Repeated Loads, Lisbon, Portugal, 15–22.
MSJC (Masonry Standards Joint Committee). (2013). “Building code
requirements for masonry structures (TMS 402-13/ASCE 5-13/ACI
530-13).” Detroit.
Nazari, N., van de Lindt, J., and Li, Y. (2015). “Effect of mainshock-
aftershock sequences on wood frame building damage fragilities.”
J. Perform. Constr. Facil., 10.1061/(ASCE)CF.1943-5509.0000512,
04014036.
NRCC (National Research Council of Canada). (2010). “National Building
Code of Canada.” Ottawa.
OpenSees [Computer software]. Pacific Earthquake Engineering Research
Center, Univ. of California, Berkeley, CA.
Porter, K., Kennedy, R., and Bachman, R. (2007). “Creating fragility
functions for performance-based earthquake engineering.” Earthquake
Spectra, 23(2), 471–489.
Purba, R., and Bruneau, M. (2015). “Seismic performance of steel plate
shear walls considering two different design philosophies of infill
plates. II: Assessment of collapse potential.” J. Struct. Eng., 10.1061/
(ASCE)ST.1943-541X.0001097, 04014161.
Shafei, B., Zareian, F., and Lignos, D. G. (2011). “A simplified method for
collapse capacity assessment of moment-resisting frame and shear wall
structural systems.” Eng. Struct., 33(4), 1107–1116.
Shedid, M., El-Dakhakhni, W., and Drysdale, R. (2010a). “Alternative
strategies to enhance the seismic performance of reinforced concrete-
block shear wall systems.” J. Struct. Eng., 10.1061/(ASCE)ST.1943
-541X.0000164, 676–689.
Shedid, M., El-Dakhakhni, W., and Drysdale, R. (2010b). “Characteristics
of rectangular, flanged and end- confined reinforced concrete masonry
shear walls for seismic design.” J. Struct. Eng., 10.1061/(ASCE)ST
.1943-541X.0000253, 1471–1482.
Siyam, M., El-Dakhakhni, W., Banting, B., and Drysdale, R. (2015a).
“Seismic response evaluation of ductile reinforced concrete block struc-
tural walls. II: Displacement and performance-based design parameters.”
J. Perform. Constr. Facil., 10.1061/(ASCE)CF.1943-5509.0000804,
04015067.
Siyam, M., El-Dakhakhni, W., Shedid, M., and Drysdale, R. (2015b).
“Seismic response evaluation of ductile reinforced concrete block struc-
tural walls. I: Experimental results and force-based design parameters.”
J. Perform. Constr. Facil., 10.1061/(ASCE)CF.1943-5509.0000794,
04015066.
Vamvatsikos, D., and Cornell, C. A. (2002). “Incremental dynamic
analysis.” Earthquake Eng. Struct. Dyn., 31(3), 491–514.
Vamvatsikos, D., and Cornell, C. A. (2005). “Developing efficient scalar
and vector intensity measures for IDA capacity estimation by incorpo-
rating elastic spectral shape information.” Earthquake Eng. Struct. Dyn.,
34(13), 1573–1600.
Villaverde, R. (2007). “Methods to assess the seismic collapse capacity of
building structures: State of the art.” J. Struct. Eng., 10.1061/(ASCE)
0733-9445(2007)133:1(57), 57–66.
Waugh, J. D., and Sritharan, S. (2010). “Lessons learned from seismic
analysis of a seven-story concrete test building.” J. Earthquake Eng.,
14(3), 448–469.
J. Perform. Constr. Facil.