Seismic Collapse Assessment of Steel Plate Shear Walls Considering The
Seismic Collapse Assessment of Steel Plate Shear Walls Considering The
Seismic Collapse Assessment of Steel Plate Shear Walls Considering The
a r t i c l e i n f o a b s t r a c t
Article history: This study evaluates the effect of mainshock–aftershock sequences on the collapse of steel plate shear wall
Received 13 January 2021 (SPSW) systems. For this purpose, three 9-, 14- and 20-story SPSWs are selected considering the practical
Received in revised form 1 April 2021 range of heights used in this system. After numerical modeling of the structures, they are initially subjected to
Accepted 3 April 2021
static-pushover analyses to evaluate their plastic behavior in large inelasticity regions. The incremental dynamic
Available online 10 April 2021
analysis is subsequently performed to capture the collapse capacity of the structures under aftershock excitations
Keywords:
preceded by mainshocks that pose various levels of damage expressed in terms of maximum inter-story drift
Steel shear wall (MID). Evaluation of the median collapse capacities under aftershock excitations exhibit the same reduction ra-
Mainshock–aftershock sequence tios at similar mainshock MIDs for the different structures. This reduction is predicted via a quadratic polynomial
Seismic-collapse capacity for given mainshock MID values. Investigation of the mainshock-aftershock effects on distribution of the MIDs
Collapse mechanism and maximum ductility demands at the collapse state shows that the prior mainshock damage mainly affects
Inelastic deformation demand the collapse capacity by consuming the energy-absorption capacity of the infill plates, thus causing strength de-
OpenSees terioration. Irrespective of the mainshock damage level, the structures were found to collapse when their first-
story columns reach a threshold softening level.
© 2021 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.jcsr.2021.106688
0143-974X/© 2021 Elsevier Ltd. All rights reserved.
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
modal response history analysis method to determine the consequent mainshock sequences used in seismic assessment of RC structures was
stochastic structural responses and damage patterns. carried out by [32]. Reference [33] investigated the seismic-energy dis-
In the study in [14], the effect of aftershocks on the collapse and in- sipation of RC shear wall-frame structures subjected to single and se-
tegrity of wood-frame structures was studied through some experimen- quential natural records. Furthermore, [34] studied the probabilistic
tal tests. In [15], the inelastic responses of low- to mid-rise reinforced aftershock–mainshock collapse risk of buckling-restrained braced
concrete (RC) frames with and without irregularities to five as- frames.
recorded aftershock–mainshock seismic sequences and 40 synthesized Among the previously mentioned studies, a method for imple-
sequences were analyzed. In [16], the maximum and residual-drift de- menting IDA under aftershock excitation was proposed by [18]. To per-
mands of three steel-frame buildings were studied under 64 as- form IDA, each aftershock was introduced on the structure with
recorded aftershock–mainshock seismic sequences. Furthermore, in increasing scales, which ultimately led to the collapse of the structure.
[17], a probabilistic framework was proposed to generate aftershock– To investigate the mainshock effect, before the application of each
mainshock seismic sequences. The evaluation of the seismic losses of level of aftershock, the structure was subjected to a mainshock at a scal-
light-frame wood buildings subjected to aftershock-mainshock se- ing factor that introduced a certain MID level. The aftershock intensity
quences generated by this framework demonstrated the substantial that led to the structure collapse was extracted for the recorded set
amplifying effects of aftershock on the seismic losses. In [18], the incre- and was statistically presented using lognormal fragility.
mental dynamic analysis (IDA) method proposed in [19] was adopted The steel plate shear wall (SPSW) system has been known as one of
for the analysis of aftershock excitations. In this method, the structure the most effective methods of resisting lateral loads, particularly forces
was first subjected to a mainshock with a scale factor that would result applied to a structure during seismic events [35–39]. Among the numer-
in a certain level of maximum inter-story drift (MID) and then to an in- ous structural systems in which the effect of the aftershock–mainshock
creasing level of aftershocks in the IDA [20–22]. sequence is still not evaluated is the SPSW system. Many experimental
In 2012, Jeon et al. [23] simulated a mainshock-induced structural [40–45] and numerical [46–51] studies have been carried out that dem-
damage by conducting cyclic-pushover analysis for different levels of onstrate the superiority of the SPSW system in terms of ductility and
maximum roof drift. For each mainshock-damage level, the structure energy-absorption capacities. Recent studies have also addressed the
was then subjected to 80 different ground-motion records that acted as design methods used in the SPSW system ([52–54]). Developments in
aftershocks. In [24], the effect of different mainshock-induced damage the numerical modeling methods used for the system have also been
levels on the collapse capacity of a four-story steel moment frame was proposed by studies such as [55,56]. The efficiency of this structural sys-
investigated using repeated, randomized, and as-recorded aftershock– tem in mitigating seismic lateral drifts has made it an attractive alterna-
mainshock seismic sequences. The results of the IDA performed in [24] tive for medium- to high-rise structures. Owing to these advantages and
following the instruction in [18] demonstrated that a high mainshock- with regard to the aforementioned significance of aftershock–
damage level significantly reduced the capacity of a structure to resist mainshock effects, the seismic collapse of SPSW structures from 9 to
collapse under an aftershock. Reference [24] study also reported the 20 stories is evaluated in the current work by considering the aftershock
low dispersion of the estimated aftershock collapse capacities under re- effects. The non-inclusion of short-rise SPSWs in this study is considered
peated aftershock–mainshock sequences, the moderate dispersion of based on the following factors:
these capacities under randomized sequences, and their large dispersion
under as-recorded aftershock–mainshock sequences. In [25], NTHA was i) The use of the SPSW system for the design of low-rise structures
performed on three-story and six-story non-ductile RC-frame buildings is not economically justified. In other words, the capacity-based
using 32 as-recorded aftershock–mainshock sequences from the western design of short-rise SPSWs leads to overly large vertical bound-
United States as well as as-recorded mainshocks followed by different ary elements (VBEs) that deteriorate the economic aspects of
randomly synthesized aftershocks. The probability of the defined dam- the system compared with the other alternatives such as the mo-
age states was then approximated based on the estimated peak drift ment frames.
and acceleration at each story. In [26], bare and masonry-infilled RC ii) The high lateral stiffness provided by the SPSW system is re-
frames were subjected to aftershock–mainshock sequences that were quired in structures in which control of lateral drifts is hardly
scaled to specific intensity levels to examine the effect of these sequences achieved. However, low-rise structures are not common among
on such a building. The result demonstrated that introduction of the this group of buildings.
mainshock on the bare frames significantly reduced the vibration period iii) Minor P-delta and plasticity effects are suffered by excessively
and increased the inter-story drifts, but the addition of masonry infills re- sized short SPSW structures in their shear-dominated response
duced these effects. Again, in 2015, [27] employed IDA to analyze four under mainshocks. Thus, the multi-shock effects posed on these
ductile RC-frame buildings, which found that 4% mainshock–induced structures are expected to have minor importance.
inter-story drift reduced the median collapse capacity (MCC) of the
frames (subjected to an aftershock) by approximately 40%. In [28], an in- Accordingly, three SPSWs with 9, 14 and 20 stories are selected in
vestigation was carried out to determine how the ground-motion polar- the presented study to assess the collapse performance of the system
ity and the vertical component of the ground motion affect irregular when multi-shock effects are accounted for. To sufficiently address the
structures with asymmetric plans. According to [28], irregular structures performance of the studied SPSW structures, the probabilistic aspects
were more strongly affected by the mainshock and their final residual of the mainshock and aftershock records are taken into account,
drifts were almost twice as much affected by an aftershock as those of which are achieved by incorporating the multi-shock IDA method first
regular structures. proposed by [18]. In addition to the IDA and collapse fragility curves,
Other investigations on the effect of aftershock–mainshock se- the multi-shock effect on the distribution of the structure response at
quences have also been carried out since 2017 which involved various the collapse state is assessed. For this purpose, the MID and maximum
types of buildings. These studies include [29] which assessed the effect DD (MDD) parameters are considered.
of modeling assumptions and aftershock hazard level on the seismic re-
sponse of steel-framed buildings. Another study [30] investigated the 2. Design and modeling of the structures
effect of an aftershock on steel moment frames designed using a
performance-based approach. One more study was performed by [31] As mentioned in the previous section, three SPSW structures with 9,
which investigated the effect of hysteretic behavior of inelastic SDOF 14, and 20 stories are considered in this study to be evaluated under the
systems on their collapse performance under multi-shock excitations. effect of aftershock–mainshock excitations. These structures are taken
Employment of artificial neural network to predict the aftershock– from a previous study on seismic-collapse performance of the code-
2
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
3
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
4
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
Fig. 5. Story drifts and DD contours of the different structures at the FEMA P695 performance point.
Fig. 6. Maximum story DDs of various members at the ultimate state suggested by FEMA P695.
Despite the significant plasticity undergone by the web plates and excess of unity and reach about 10 or higher. To obtain a quantitative
HBEs, the maximum DD experienced by the VBEs remains in a small criterion for determination of the threshold DD corresponding to effec-
range. This is due to the fact that the computed DDs reflect the maxi- tive yielding in the VBEs, an obvious softening in the story that leads to
mum strain values of the sections and are only experienced at very global instability in the structure should be considered. According to the
few points across them. That is while the effective yielding of the sec- drift profiles computed at this state, story softening and the resulting
tion, and thereby the element, occurs at DD values that are well in drift concentration are still not reached. As a result, the overall
5
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
Fig. 7. Story softening and subsequent plasticity distribution in structures with a) low, and b) high number of redundancies at the stories.
assessment is that the FEMA P695's suggestion is not a good criterion for will also suffer from sudden softening at the yield point. In this case,
predicting the ultimate state of SPSW structures. after the yielding of that story, its softness will be significantly reduced
Because of the non-occurrence of severe plasticity in the VBEs in the compared with that of the other stories. As a result, the increase in the
ultimate state proposed by FEMA P695, the structural behavior should roof displacement will lead to a considerable concentration of deforma-
be investigated in a state closer to the collapse state. Therefore, a state tions in that story. This phenomenon is known in the literature and in
corresponding to a 10% global drift is considered. It may be said that guidelines as a soft-story formation. In contrast, Fig. 7(b) shows that
the numerical results obtained at a global drift as much as 10% might re- the story strength can be provided by a significant number of elements
quire experimental approval since it may seem as an overly large value. that act in parallel to one another using a rigid-enough interface. In this
While this fact is not neglected here, it should be added that the thor- case, the resisting force of the story will be the sum of the forces of these
ough modeling of the plasticity and damage mechanisms in current elements, and the successive yielding of these elements will lead to a
study are believed to support the obtained results. In other words, all gradual softening in the force-deformation curve of the story. Conse-
the damage sources that may prevent reaching a 10% global drift in quently, the softening process of this story is associated with a relatively
the real structures are also simulated here except the local buckling of large range of deformations in which other stories will also have the op-
the columns. As mentioned earlier in section 2, the compactness of the portunity to enter the inelastic range. As a result, no large difference ap-
utilized column sections heavily designed to remain elastic after com- pears between the flexibility of the different stories, and a relatively
plete yielding of all the web plates and the beams is able to prevent to uniform deformation distribution will be achieved over the height of
prevent their local failure at very large lateral drifts. To further validate the structure.
this, the DD values reported at the coming graphs can be regarded. Ac- These explanations show that the uniform distribution of deforma-
cording to those results, the maximum DD undergone by the columns at tions over the height of SPSW structures and the more economical use
a 10% global drift does not exceed 15 value. As mentioned before, such of the load-bearing capacity of the stories than the other systems such
value is corresponding to an effective section ductility demand of as moment frames and bracing systems are attributed to the following
about 5. Regarding the experimental findings and the ductility capacity factors:
estimates provided by previous researchers [65], local failure of the sec-
tions used in this study occur at ductility demands that are larger than 5. 1) existence of multiple resisting elements and a high strength re-
This indicates that occurrence of the non-simulated failure modes is in- dundancy in the story, which is provided by the continuous
directly prohibited in this study and the estimated collapse capacities area of the plate;
can be considered reliable. 2) anchorage provided by the high-stiffness columns that allow op-
Before examining the nonlinear behavior of structures at this condi- timal contribution of different areas of the plate in a parallel
tion, we need to describe how such large lateral drift capacities can be mechanism; and
provided by SPSW structures. For this purpose, we should first note 3) differences in the length of the diagonal strips representing the
that the load-bearing mechanisms provided by the different floors in a plate after its buckling; these differences provide the strips
building form a series system together. In other words, the loading with various yield strengths so that an optimal yield sequence
mechanism imposed on each floor can be considered as a ring of a and gradual softening is achieved.
chain, as shown in Fig. 7. With this concept, application of a lateral de- The drift distribution diagram and the DD distribution contour over
formation to the roof of a structure will result in distribution of defor- the height of the different structures are demonstrated in Fig. 8 for the
mations in different stories according to their flexibilities. limit state corresponding to a 10% global drift. By considering these
In the elastic-behavior range, the distribution of deformations is a re- plots, we can see that an almost 12% MID has been reached in this
sult of the elastic design procedure and can be almost uniform. How- state indicating a 20% increase with respect to the applied global drift
ever, with the increase in the global deformations, one of the stories of 10% which denotes the average of the story drifts. The relatively uni-
will eventually reach its yield point earlier than the others. At this form distribution of the drifts and the prevention of plasticity localiza-
stage, the shape of the nonlinear force deformation of the yielding tion in one or more specific stories indicates that the soft-story
story will significantly affect the building-wide distribution of the defor- mechanism does not occur even in this large lateral drift. Regarding
mations at the following steps of the loading process. Fig. 7(a) shows the maximum values of DDs shown in Fig. 9 for various member types
that if the strength of a story is supplied by a few elements with sudden and in different stories, rupture of web plates is implied by the observed
yielding behavior, the overall force-deformation diagram of that story large DDs in all stories. The large DDs observed for the beams should be
6
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
Fig. 8. Story drifts and DD contours at the suggested failure point of the different structures.
still cautiously interpreted since these values do not reflect response of 4. Multi-shock IDA results
the entire fiber sections, but the fibers with the maximum strains across
the section. Regarding the column DDs, values as large as 20 are ob- The methodology used for investigating the effects of mainshock-
served for the 9-story SPSW that are decreased for the taller buildings induced damage on the collapse capacity of the studied structures are
and reach as low as 10 for the 20-story structure. Considering the afore- presented in more details in this section. An important aspect of this
mentioned fact and the lack of story mechanism observation in the methodology is how the uncertainties associated with the spectral
structures, it can be said that columns are still in the safe side of their shape and intensity of the mainshock-aftershock sequence is addressed.
behavior. Another source of variability is the number of aftershock excitations to
7
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
be applied on a structure. However, most of the studies in this field have capacity of the structure under the aftershock excitation triggered on a
limited the number of sequential excitations to two to avoid too much presumed mainshock-damage level. Whereas scaling the mainshock
complication. to a specific intensity level is easy, scaling it to produce a given damage
Previous studies have utilized four different methods for consider- level is not straightforward. This process requires another IDA using the
ing the uncertain spectral shape of the record pairs. One method was mainshock record to establish the response-intensity relationship. By
to randomly and independently choose the mainshock and aftershock considering the MID to represent the mainshock damage, the intensities
records from a set of pre-selected natural records. An alternative corresponding to various damage levels are then interpolated using the
method was to generate artificial records in the emulation of naturally IDA data. The use of this relationship facilitates scaling of the mainshock
observed relationships between the mainshock and aftershock. A third for structural analysis. After imposing the mainshock and before apply-
approach assumed that the spectral shapes of the mainshock and after- ing the aftershock, a free-vibration analysis is performed to have the
shock records were fully correlated or the mainshock records were re- structure at its residual stage when it is excited by the aftershock. The
peated in the aftershock. However, the most popular approach was to free-vibration phase should be sufficiently long until an acceptably
use as-recorded mainshock-aftershock pairs. The popularity of this low MID vibration amplitude is reached. Previous studies have made
method originated from the fact that it encompasses the geological use of either a constant value or a multiple of the mainshock duration
variables that inherently affect the spectral shape of the mainshock- as the free-vibration time. However, the required free-vibration time
aftershock pair. is a function of the damping possessed by the structure and the peak re-
The intensity of the aftershock–mainshock pair is also an important sponse recorded during the mainshock phase. Thus, ensuring the ade-
source of uncertainty. As described in Section 1, a universally accepted quacy of the selected duration for all structures and for every record
method is to scale the mainshock to various damage levels while pair is difficult. On the other hand, lengthy free vibrations can also be
conducting IDA using the aftershock. IDA is used to assess the variations detrimental to the efficiency of the analyses and should be avoided.
in the aftershock intensity, and the results are used to determine the To address these requirements, a programmatic technique suggested
Fig. 10. IDA curves obtained in three of the considered aftershock–mainshock scenarios.
8
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
Fig. 11. Comparison of median IDA curves and fragilities of the different aftershock–mainshock scenarios.
Fig. 12. Correlation between MCC and the mainshock MID level. a) MCC values. b) MCC reduction factors.
by [66] is used for accurate determination of the free-vibration time command that directly invoked response values and eliminated time
after each mainshock excitation. This method involved customized consumption in opening and closing the response-data files.
response-recording capabilities in the OpenSees software. These abili- A remarkable parameter that should also be regarded when apply-
ties allowed the definition of drift recorders that reported the maximum ing an aftershock after a mainshock has been posed, is the polarity of af-
and minimum values of the response during the last desired seconds of tershock. Regarding polarity effect, turning the aftershock direction by
the analysis. The software modification included the addition of a new 180 degrees will cause different results in the already inclined structure.
9
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
Fig. 13. Distribution of the PISDs at the aftershock-induced collapse state under different mainshock MID levels.
To address this variability, a conservative approach is regarded, as in the residual response actually appears as an initial state for the structure
previous studies, and the direction causing the larger response is used in at the start of IDA under an aftershock. Because of this initial drift, the
the IDAs. drifts from the aftershock record are larger than or equal to those of
The current study employs 32 aftershock–mainshock pairs ex- the lag.
plained by Han et al. in [25] to demonstrate the spectral shape uncer- After extracting the Sa values that correspond to the aftershock col-
tainties. The 32 sequences were recorded from earthquakes in the lapse, they are used to draw a lognormal fragility curve. These diagrams
West United States. The aftershock with the largest magnitude that oc- are shown in Fig. 11 for the different levels of collapse caused by the
curred within one week after the mainshock was selected. Furthermore, mainshock. The collapse capacity corresponding to a 50% probability is
the sequences were selected in such a manner that the aftershock mag- extracted from these curves and is referred to as the median collapse ca-
nitudes were larger than or equal to five. The site-to-source distance of pacity (MCC). We can see that the initial damage caused by the
all the records was larger than 10 km; thus, the records were termed as mainshock leads to a reduction in the aftershock MCCs. To quantify
far-field events. The moment magnitudes of the mainshocks and after- this reduction under varying conditions, the MCC values corresponding
shocks varied in 5.80–7.20 and 5.0–6.70 ranges, respectively. More de- to the different levels of mainshock damage are compared for the differ-
tails on characteristics of the records are provided in [25]. ent structures.
The adopted response levels to represent the mainshock damage in Fig. 12(a) shows the comparison of the MCC values for structures
this study correspond to 0.007, 0.025, and 0.05 MID values. Additionally, with different number of stories as well as for different damage levels
the damage level of the structure at the collapse threshold (CT) is con- caused by the mainshock. Despite the higher collapse capacity of the
sidered. CT is defined as the stage in which a stable response is experi- 9-story structure than that of the two structures with heights of 14-
enced by the structures, but minor increases in the ground motion and 20- stories, the reduction in the MCC value due to the response of
scale factor will lead to collapse occurrence. The Hunt–Fill algorithm is the mainshock follows the same trend in all structures.
used to perform IDAs [19]. The collapse of the SPSW structures under af- To more closely evaluate this trend, Fig. 11(b) shows the MCCs nor-
tershock excitation is identified by considering two criteria. The first cri- malized with respect to the capacity computed in absence of mainshock
terion addresses the global instability of the structure in the presence of damage. The normalized values indicate the MCC reduction factor and
P-Delta effects and damage of the structural members. To quantify this also allow for a comparison between mainshock effects in the different
criterion, a limit value is set for the slope of the IDA curve following structures.
the FEMA P695 [63] and other relevant guidelines. This limit value is By plotting the values of these factors against the MID values gener-
equal to 20% of the initial slope of the IDA curve. The second criterion ated by the mainshock in Fig. 11(b) and fitting a quadratic polynomial to
considers the maximum allowable MID value to be experienced by the the results, we can accurately measure the amount of MCC drop using
SPSW system at the collapse stage. A limit MID value of 10% was pro- Eq. (1). With this equation, predicting the collapse capacity of similar
posed by [63] and other references as a rationale-based upper bound structures is possible under different levels of mainshock damage and
for moment-resisting frames. Although this value was shown in without the need for multi-shock analyses.
Section 3 to underestimate the collapse capacity of SPSW structures, it
is also regarded in this study.
y ¼ −32:022x2 –1:762x þ 0:9194 ð1Þ
The results of IDAs under the aftershock records following the
mainshocks are shown in Fig. 10 for the different structures. In this fig-
ure, because of space limit, the results are shown for only three of the
mainshock and aftershock combination scenarios. To compare the IDA 5. Collapse-mechanism assessment
curves of the structures with different number of stories under different
scenarios of the mainshock-aftershock application, the median values of In this section, our attempt to study the distribution of nonlinear de-
the IDA curves shown in Fig. 11 are compared with one another in formations in different parts of the structures at their collapse state
Fig. 11. We can generally deduce from this comparison that as the under the aftershock is presented. For this purpose, we examine distri-
level of damage caused by the mainshock increases, the capacity of bution of two response variables at the collapse and in presence of var-
the structures to withstand the collapse under an aftershock decreases. ious mainshock damage levels. These parameters include peak inter-
This increase in the mainshock damage also leads to a horizontal lag in story drift (PISD) (equal to the maximum drift experienced by a story
the beginning of the IDA curves, resulting from the MID response gener- over time) and maximum ductility demand (MDD, which is the maxi-
ated in the structure at the end of the mainshock excitation. This mum of DD over time).
10
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
Fig. 14. Median MDD contours at the collapse state of the different mainshock-aftershock
scenarios (9-story SPSW).
Fig. 15. MDD contours at the collapse state of the different mainshock-aftershock
scenarios (14-story SPSW).
5.1. PISD distribution at collapse
The PISD distribution plots shown in Fig. 13 are the median of the re-
sults obtained using the 32 records described in the previous section.
Regardless of the number of stories, we can state that the effect of
0.007 and 0.025 damage levels imposed by the mainshock on the PISD
distribution graph is insignificant. By increasing the mainshock MID
level to 0.05 or more, a relatively significant change is observed in this
graph. Thus, to derive better conclusions, the studies presented in this
section focus on the three damage levels of zero initial failure (after-
shock only), 0.05, and CT.
As a general observation, the initial damage caused by the
mainshock leads to insignificant changes in the lower stories' PISDs at
the collapse state. For the 14- and 20-story structures, these changes
are also minor in the topmost stories.
The changes in the distribution of PISDs reflect the attempt of the
structures to dissipate aftershock energy in stories that suffer less
from the previous damage caused by mainshock. From a mathematical
perspective, strength deterioration in stories which are damaged by
the mainshock reduces their energy absorption capability; thus, large
deformations accumulate in more intact stories when the structures
are hit by the aftershock. To further describe this, we need to state
that the excitation of the structures by the mainshock reduces the col-
lapse capacity of structures under an aftershock excitation in two man-
ners. The first is the strength deterioration due to the nonlinear
deformations and hysteretic energy dissipation under the mainshock.
The second is the residual plastic deformations at the end of mainshock
excitation, which tend to shift the structure from its intact configuration
to an inclined condition. This initial inclination may align with the polar-
ity of the aftershock record, leading to an increase in the maximum de-
formation of the structure under the aftershock. However, as
demonstrated by the study in [66], the residual deformations in SPSW
structures are easily revertible due to the inherent self-centering char-
acteristic provided by the system. This characteristic is a result of the
pinched hysteresis curves of infill plates that lead to very slight
unloading stiffness. This slight stiffness, accompanied with the signifi- Fig. 16. MDD contours at the collapse state of the different mainshock-aftershock
cant returning forces provided by the surrounding columns, can return scenarios (20-story SPSW).
11
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
the system back near to its original state in the following free-vibration diagrams, the maximum MDD values are calculated for the members
phase of at the beginning of the aftershock excitation. As a result, we can of each story, and the graph of the MDD changes through the height is
state that the main factor responsible for the influence of mainshock on plotted for different members.
the behavior of the SPSW structures is the behavioral deterioration due By considering the diagrams shown in Fig. 17, the first significant
to the dissipation of hysteretic energy. point is the similarity of the web plates' MDD distribution to the PISD
distribution diagrams shown in Fig. 13. This similarity testifies that the
5.2. Plasticity distribution at collapse deformations experienced by the plates exactly obeys the story drift,
which indicates the perfect anchorage provided by the boundary ele-
To study the collapse mechanism of the structures under an after- ments for the web plates. On the other hand, the distribution of MDDs
shock, distribution of the MDD values in different parts of the structures in the HEB and VBE members, as will be described, is affected by the dis-
is investigated. MDD is the ratio of the maximum element deformation tribution of the plate plasticity over the height. Therefore, in general, the
over time to its yielding deformation. For fiber sections, this value is effect of mainshock damage on the collapse mechanism under the after-
maximized at the cross section and along the element. Whereas this shock follows the explanations mentioned in the previous section re-
variable indicates the extent of nonlinear response undergone by the garding the PISD variable distribution at the collapse state.
members, it is also a good indicator of the amount of hysteretic energy To clarify the relation between the plates MDD and the MDDs under-
damped by the elements. Fig. 14 shows the contours of MDD values ex- gone by the HBEs and VBEs, the statics principles also deployed by the
perienced by the 9-story structure at the collapse state caused by after- capacity-based design method should be recalled. In this regard, the
shocks when various mainshock damage levels are applied. The forces undergone by the HBEs are equal to the unbalanced forces applied
contours represent the median of results obtained using the 32 by the lower and upper plates connected to it. The plate forces on the
mainshock-aftershock record pairs. Figs. 15 and 16 also show this con- other hand are a function of plate thickness and the strain-related
tour for the 14- and 20-story structures respectively. We can observe stresses. Therefore, in addition to the magnitude of the plate MDD, the
that the distribution of the MDD values in the collapse state is very sim- difference in the thickness of the upper and lower plates affects the
ilar to the distribution of DD values obtained from the pushover analysis force demand of each HBE. This force demand in turn leads to yielding
at the 10% global drift state (Fig. 7). In other words, the largest values of and plastic behavior at the end of the HBE. According to the irregular
the MDDs undergone by the plates and beams are observed in the last trend dominating the plate thicknesses resulted from the design process,
story and decrease by moving downward. A more detailed study of the level of the plasticity experienced by the HBEs also follows an irreg-
the distribution pattern of these values in different stories and members ular trend. Despite this irregularity, the overall trend of these changes
is possible by referring to the diagrams shown in Fig. 16. In these shows increase of the plates' plasticity with the increase in height.
Fig. 17. Distribution of the maximum story MDDs in the HBEs, VBEs, and infill plates at the aftershock-induced collapse state.
12
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
On the other hand, the axial and flexural forces experienced by each deformations in the first story eventually led to the collapse of the struc-
VBE are result of the forces applied by the connected plates and HBEs ture. The threshold column MDD values at the collapse state were about
and also those located at the upper floors. According to these consider- 10 for the 9-story structure and about 15 for the taller 14- and 20-story
ations, the increase in the force demand and thereby the MDD of the col- structures. However, the collapse MDDs of the beams and plates varied
umns by approaching the first floor can be justified. However, the in the 20–30 and 45–60 ranges, respectively, depending on the level of
columns form the last layer of the parallel mechanism resisting the damage caused by the mainshock.
forces acting on each floor. Therefore, their yielding is accompanied by
a significant softening of the story. As a result, according to the explana-
tions presented in Section 3 and Fig. 7, this softening leads to a signifi- Declaration of Competing Interest
cant plasticity concentration in the first story. As a result, a remarkable
difference between the MDDs of the first-story columns and those of We declare that we have no financial and personal relationships
the other stories occurs that magnifies as the number of floors increases. with other people or organizations that can inappropriately influence
The p-delta effects due to the softening in the columns of the first floor our work, there is no professional or other personal interest of any na-
and the large deformations eventually lead to the failure of this story ture or kind in any product, service and/or company that could be con-
and thereby collapse of the whole structure. strued as influencing the position presented in, or the review of, the
We can observe that irrespective of the severity of the damage level manuscript entitled, “Seismic collapse assessment of steel plate shear
caused by the mainshock, the collapse of the structures is always accom- walls considering the mainshock–aftershock effects” by S.A. Jalali, A.
panied with column MDDs about 10 in the 9-story structure and around Amini, I. Mansouri, and J. W. Hu.
15 in the 14- and 20-story structures. The reason for the larger column
MDD at the collapse state of the taller structures can be the higher mode Acknowledgement
effects. Regarding the HBEs and the web plates, the collapse MDDs vary
in 20–30 and 45–60 ranges, respectively, depending on the level of This research was supported by a grant (19CTAP-C152266-01) from
damage caused by the mainshock. Technology Advancement Research Program (TARP) funded by the
Ministry of Land, Infrastructure, and Transport of the Korean
6. Conclusions government.
13
S.A. Jalali, A. Amini, I. Mansouri et al. Journal of Constructional Steel Research 182 (2021) 106688
[19] D. Vamvatsikos, C.A. Cornell, Incremental dynamic analysis, Earthq. Eng. Struct. Dyn. [42] P. Timler, G.L. Kulak, Experimental study of steel plate shear walls, Structural Engi-
31 (3) (2002) 491–514. neering Report No. 114, Department of Civil Engineering, University of Alberta, Ed-
[20] I.P. Mitseas, M. Beer, First-Excursion Stochastic Incremental Dynamics Methodology monton, AB, 1983.
for Hysteretic Structural Systems Subject to Seismic Excitation, Computers and [43] Z. Ahmadi, A.A. Aghakouchak, S.R. Mirghaderi, Cyclic behavior of moment frames
Structures, 2021 242. with conventional or slit Infills, J. Constr. Steel Res. 178 (2021).
[21] D. Vamvatsikos, Performing incremental dynamic analysis in parallel, Comput. [44] J.C. Cui, J.D. Xu, Z.R. Xu, T. Huo, Cyclic behavior study of high load-bearing capacity
Struct. 89 (1–2) (2011) 170–180. steel plate shear wall, J. Constr. Steel Res. 172 (2020).
[22] A. Hardyniec, F. Charney, A new efficient method for determining the collapse mar- [45] M. Wang, X. Zhang, L. Yang, W. Yang, Cyclic performance for low-yield point steel
gin ratio using parallel computing, Comput. Struct. 148 (2015) 14–25. plate shear walls with diagonal T-shaped-stiffener, J. Constr. Steel Res. 171 (2020).
[23] J. Jeon, R. DesRoches, I. Brilakis, L. Lowes, Aftershock Fragility Curves for Damaged [46] L.J. Thorburn, G.L. Kulak, C. Montgomery, Analysis of steel plate shear walls, Struc-
Non-ductile Reinforced Concrete Buildings, 15th World Conference on Earthquake tural Engineering Report No. 107, Department of Civil Engineering, University of Al-
Engineering, 2012. berta, Edmonton, AB, 1983.
[47] E. Tromposch, G.L. Kulak, Cyclic and Static Behaviour of Thin Panel Steel Plate Shear
[24] Y. Li, R. Song, J.W. Van De Lindt, Collapse fragility of steel structures subjected to
Walls, Department of Civil Engineering, University of Alberta, 1987.
earthquake mainshock-aftershock sequences, J. Struct. Eng. 140 (12) (2014),
[48] M. Elgaaly, Y. Liu, Analysis of thin-steel-plate shear walls, J. Struct. Eng. 123 (11)
04014095, .
(1997) 1487–1496.
[25] R. Han, Y. Li, J. van de Lindt, Assessment of seismic performance of buildings with
[49] M. Rezai, C.E. Ventura, H.G. Prion, Numerical investigation of thin unstiffened steel
incorporation of aftershocks, J. Perform. Constr. Facil. 29 (3) (2014), 04014088, .
plate shear walls, 12th World Conference on Earthquake Engineering 2000, pp. 1–4.
[26] S. Tesfamariam, K. Goda, G. Mondal, Seismic vulnerability of reinforced concrete [50] M.R. Behbahanifard, G.Y. Grondin, A.E. Elwi, Experimental and numerical investiga-
frame with unreinforced masonry infill due to main shock–aftershock earthquake tion of steel plate shear walls, Struct. Eng. Report 254 (2003).
sequences, Earthquake Spectra 31 (3) (2015) 1427–1449. [51] J. Bai, J. Zhang, S. Jin, K. Du, Y.H. Wang, A multi-modal-analysis-based simplified seis-
[27] M. Raghunandan, A.B. Liel, N. Luco, Aftershock collapse vulnerability assessment of mic design method for high-rise frame-steel plate shear wall dual structures, J.
reinforced concrete frame structures, Earthq. Eng. Struct. Dyn. 44 (3) (2015) Constr. Steel Res. 177 (2021).
419–439. [52] R. Purba, M. Bruneau, Seismic performance of steel plate shear walls considering
[28] F. Hosseinpour, A. Abdelnaby, Effect of different aspects of multiple earthquakes on two different design philosophies of infill plates. I: deterioration model develop-
the nonlinear behavior of RC structures, Soil Dyn. Earthq. Eng. 92 (2017) 706–725. ment, J. Struct. Eng. 141 (6) (2015), 04014160, https://doi.org/10.1061/(ASCE)ST.
[29] J. Ruiz-García, J.D. Aguilar, Influence of modeling assumptions and aftershock hazard 1943-541X.0001098.
level in the seismic response of post-mainshock steel framed buildings, Eng. Struct. [53] R. Purba, M. Bruneau, Seismic performance of steel plate shear walls considering
140 (2017) 437–446. two different design philosophies of infill plates. II: assessment of collapse potential,
[30] G. Abdollahzadeh, A. Mohammadgholipour, E. Omranian, Seismic evaluation of steel J. Struct. Eng. 141 (6) (2015), 04014161, https://doi.org/10.1061/(ASCE)ST.1943-
moment frames under Mainshock–aftershock sequence designed by elastic design 541X.0001097.
and PBPD methods, J. Earthq. Eng. (2017) 1–24. [54] S.A. Jalali, M. Banazadeh, Computer-based evaluation of design methods used for a
[31] X.-H. Yu, S. Li, D.-G. Lu, J. Tao, Collapse capacity of inelastic single-degree-of-freedom steel plate shear wall system, Struct. Design Tall Spec. Build. 25 (17) (2016)
systems subjected to mainshock-aftershock earthquake sequences, J. Earthq. Eng. 904–925.
(2018) 1–24. [55] I.-R. Choi, H.-G. Park, Steel plate shear walls with various infill plate designs, J. Struct.
[32] G. Abdollahzadeh, E. Omranian, V. Vahedian, Application of the artificial neural net- Eng. 135 (7) (2009) 785–796.
work for predicting Mainshock-aftershock sequences in seismic assessment of rein- [56] S.A. Jalali, M. Banazadeh, Development of a new deteriorating hysteresis model for
forced concrete structures, J. Earthq. Eng. (2018) 1–27. seismic collapse assessment of thin steel plate shear walls, Thin-Walled Struct.
[33] H.H. Jamnani, J.V. Amiri, H. Rajabnejad, Energy distribution in RC shear wall-frame 106 (2016) 244–257.
structures subject to repeated earthquakes, Soil Dyn. Earthq. Eng. 107 (2018) [57] S.A. Jalali, Developing a modified hysteresis model for seismic performance evalua-
116–128. tion of steel plate shear wall system, Ph.D. thesis. , Department of Civil and Environ-
[34] S. Veismoradi, A. Cheraghi, E. Darvishan, Probabilistic mainshock-aftershock col- mental Engineering, Amirkabir University of Technology, Iran, 2016.
lapse risk assessment of buckling restrained braced frames, Soil Dyn. Earthq. Eng. [58] AISC, ANSI, AISC 341–16: Seismic Provisions for Structural Steel Buildings, American
115 (2018) 205–216. Institute for Steel Construction, Chicago, IL, 2016.
[59] ASCE, ASCE 7–16: Minimum Design Loads and Associated Criteria for Buildings and
[35] A. Farzampour, I. Mansouri, C.-H. Lee, H.-B. Sim, J.W. Hu, Analysis and design recom-
Other Structures, American Society of Civil Engineers, Reston, Virginia, 2016.
mendations for corrugated steel plate shear walls with a reduced beam section,
[60] S. Mazzoni, F. McKenna, M.H. Scott, G.L. Fenves, OpenSees Users Manual, PEER Cen-
Thin-Walled Struct. 132 (2018) 658–666.
ter, University of California, Berkeley, 2004.
[36] S. Sabouri-Ghomi, T.M. Roberts, Nonlinear dynamic analysis of thin steel plate shear
[61] V. Macillo, S. Shakeel, L. Fiorino, R. Landolfo, Development and calibration of a hys-
walls, Comput. Struct. 39 (1–2) (1991) 121–127.
teretic model for CFS strap braced stud walls, Adv. Steel Constr. 14 (3) (2018)
[37] N. Paslar, A. Farzampour, F. Hatami, Investigation of the infill plate boundary condi- 337–360.
tion effects on the overall performance of the steel plate shear walls with circular [62] S. Sabouri-Ghomi, S.R. Asad Sajjadi, Experimental and theoretical studies of steel
openings, Structures 27 (2020) 824–836. shear walls with and without stiffeners, J. Constr. Steel Res. 75 (2012) 152–159.
[38] A. Farzampour, I. Mansouri, J.W. Hu, Seismic behavior investigation of the corru- [63] FEMA P695, in: F.E.M. Agency (Ed.), Quantification of Building Seismic Performance
gated steel shear walls considering variations of corrugation geometrical character- Factors, 2009 , Washington, DC.
istics, Intern. J. Steel Struct. 18 (4) (2018) 1297–1305. [64] J.W. Berman, Seismic behavior of code designed steel plate shear walls, Eng. Struct.
[39] I. Mansouri, A. Arabzadeh, A. Farzampour, J.W. Hu, Seismic behavior investigation of 33 (1) (2011) 230–244.
the steel multi-story moment frames with steel plate shear walls, Steel Compos. [65] D.G. Lignos, H. Krawinkler, Deterioration modeling of steel components in support
Struct. 37 (1) (2020) 91–98. of collapse prediction of steel moment frames under earthquake loading, J. Struct.
[40] M. Elgaaly, V. Caccese, C. Du, Postbuckling behavior of steel-plate shear walls under Eng. 137 (11) (2010) 1291–1302.
cyclic loads, J. Struct. Eng. 119 (2) (1993) 588–605. [66] S.A. Jalali, E. Darvishan, Seismic demand assessment of self-centering steel plate
[41] R.G. Driver, G.L. Kulak, D.L. Kennedy, A.E. Elwi, Cyclic test of four-story steel plate shear walls, J. Constr. Steel Res. 162 (2019) 105738.
shear wall, J. Struct. Eng. 124 (2) (1998) 112–120.
14