Seismic Performance of Steel Plate Shear Walls Considering Two Different Design Philosophies of Infill Plates. I Deterioration Model Development
Seismic Performance of Steel Plate Shear Walls Considering Two Different Design Philosophies of Infill Plates. I Deterioration Model Development
Seismic Performance of Steel Plate Shear Walls Considering Two Different Design Philosophies of Infill Plates. I Deterioration Model Development
Abstract: Research was conducted to investigate the seismic performance of steel plate shear walls (SPSWs) having infill plates designed to
resist different percentages of the applied lateral loads. The FEMA P695 methodology, which defines the performance in terms of collapse
potential under maximum considered earthquake (MCE) ground motions, was used to compare the performance of the SPSWs under
consideration. This paper describes the development of component strength deterioration models that are needed to perform the collapse
assessment of SPSWs, focusing on stress-strain or force-deformation relationships for infill plates and boundary elements. The approach
began with identifying the deterioration and failures modes of SPSW from 36 tested specimens. Cyclic deformation capacities of these
SPSWs when reaching their ultimate strength, failure points, and rates of degradation were statistically quantified. Based on these statistical
results, initial deterioration models for SPSW components were developed in a format compatible for use with the FEMA P695 procedures.
The chosen deteriorated material models for infill plates and boundary elements were calibrated to four selected SPSW specimens varying
from one to four stories. A companion paper presents the steps and results of the seismic performance assessments. DOI: 10.1061/(ASCE)ST
.1943-541X.0001098. © 2014 American Society of Civil Engineers.
Author keywords: Seismic performance; Steel plate shear walls; Deterioration and failure modes; Deterioration model; Collapse potential;
Infill plates design; FEMA P695 methodology; Seismic effects.
Introduction (AISC 2010), but one possible interpretation of the AISC design
specifications may lead to the same design approach. In this
In seismic design applications, the primary energy dissipating approach, HBEs and VBEs are designed to resist the tension field
elements of steel plate shear walls (SPSWs) resisting lateral loads forces generated by the fully yielded infill plates, and the boun-
are their unstiffened infill plates (webs), which buckle in shear and dary frame moment resisting action contributing to the global plas-
form a series of diagonal tension field actions (TFAs). From a tic lateral strength of SPSW effectively provides overstrength for
capacity design perspective, the tension force from the infill plates resisting the lateral loads. As reported in past experiments, this
must be resisted by the surrounding horizontal and vertical boun- overstrength in conventional SPSWs can be quite significant.
dary elements (HBEs and VBEs). When rigid connections are For example, Driver et al. (1997) reported that boundary frame
specified between HBEs and VBEs, and between VBEs and the moment resisting action contributed approximately 25% of the
ground (as specified in many applications of SPSWs), SPSWs also global plastic strength of their four-story SPSW specimen. The
benefit from the moment resisting action of the boundary frame to same observation was made by Berman and Bruneau (2005),
resist the applied lateral loads. Recognizing the contribution of who indicated that the boundary frame of their single story SPSW
the boundary frame to the overall strength of the system, Qu and specimen contributed 38% to the total strength of the wall. Qu and
Bruneau (2009) accounted for the attraction that may exist as a Bruneau (2009) demonstrated that the boundary frame moment
means to optimize SPSW designs, rather than relying on the pres- resisting action can contribute up to 50% of the total strength
ence of the system overstrength that it may provide to resist the of an SPSW with an aspect ratio of 2.0 when its boundary ele-
specified lateral loads. ments are designed per capacity design principles. In other words,
In the current Canadian standard [Canadian Standards Associ- in such a case, the total lateral strength of the SPSW is twice that
ation (CSA) 2009] for the design of steel structures, it is specified needed to resist the total specified lateral loads. This provides a
that the infill plates of SPSWs must be designed to resist the entire significant incentive to reduce overstrength by explicitly consid-
lateral loads, without considering the possible contribution from ering boundary frame moment resisting action as contributing
the surrounding boundary moment resisting frame. Such a state- to the overall lateral strength of the SPSW.
ment is not explicitly included in the American seismic provisions Qu and Bruneau (2009) investigated this concept of sharing
lateral loads between the boundary frame and infill plates, and re-
1
Instructor, Dept. of Civil Engineering, Univ. of Minnesota, 121 ported that SPSWs designed per this approach, although having
Swenson Civil Engineering, 1405 University Dr., Duluth, MN 55811 smaller steel quantities, exhibited larger drifts, which suggested
(corresponding author). E-mail: [email protected] that they may need to be designed according to different response
2
Professor, Dept. of CSEE, Univ. at Buffalo, 130 Ketter Hall, Buffalo,
modification coefficients (i.e., R-factors). They left it to future
NY 14260. E-mail: [email protected]
Note. This manuscript was submitted on September 19, 2013; approved research to investigate.
on April 11, 2014; published online on August 5, 2014. Discussion period This paper, along with a companion paper (Purba and Bruneau
open until January 5, 2015; separate discussions must be submitted for in- 2014b), investigates this matter by quantifying the seismic perfor-
dividual papers. This paper is part of the Journal of Structural Engineer- mance factors (SPFs) for SPSWs having infill plates designed per
ing, © ASCE, ISSN 0733-9445/04014160(12)/$25.00. two different philosophies, to sustain different percentages of the
J. Struct. Eng.
applied lateral loads. The FEMA P695 methodology (2009b) was components share similarities with the FEMA 356 (2000) back-
selected to accomplish this objective. This methodology provides a bone model. Two examples can be found in Song and Pincheira
rational basis for establishing SPFs for seismic-force resisting (2000) and Ibarra and Krawinkler (2005), which describe models
systems by assessing the system risks against collapse under implemented to investigate the cyclic response of nonductile
maximum considered earthquake (MCE) ground motions. and ductile structures (Song and Pincheira 2000; Haselton and
The validity of the results obtained using this methodology Deierlein 2007; Lignos and Krawinkler 2009).
critically depends on the accuracy of the structural numerical One of the key differences between those two models and the
models used in these analyses to simulate the component strength FEMA 356 (2000) model is that they have less of a sudden
deterioration that will eventually lead to global collapse of the degradation slope (i.e., postpeak or postcapping stiffness), which
system, because these will affect the results of the incremental alleviates computational difficulties and the inability to converge
dynamic analyses used to quantify the SPF [as described in the in nonlinear dynamic analysis. In addition, they have hysteretic
companion paper (Purba and Bruneau 2014b)]. Although many rules to account for cyclic deterioration modes, but this is not the
SPSW specimens in past research have been tested until they case for the FEMA 356 (2000) model. In other words, structural
exhibited significant strength degradations, and although fragility components modeled according to the FEMA 356 (2000) model
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
curves have been developed to relate SPSW damage states to drift only experience strength degradation when cycled beyond Point
values (Baldvins et al. 2012), no attempt to simulate complete C, while the component stiffness from one cycle to another cycle
strength degradation through numerical investigation was found in of loading remains unchanged. By contrast, structural components
the existing literature. Hence, the first key step toward this objective modeled according to the last two models may experience stiffness
is to develop degradation models for SPSW components (i.e., boun- and strength degradation (even before reaching Point C) in addition
dary elements and infill plates). to strength degradation beyond the capping point. In fact, the Ibarra
This paper describes the development of these component and Krawinkler (2005) model can simulate strength deterioration in
strength deterioration models to be used in the collapse assessment structural components owing to increasing inelastic displacement
of SPSWs, focusing on stress-strain or force-deformation relation- and repeated cycles of the same displacement, unloading and
ships for infill plates and boundary elements. The approach starts reloading stiffness degradations, and pinching cyclic strength deg-
with identifying deterioration and failures modes of SPSW radation. It can also simulate structural components that have more
observed from 36 tested specimens. Cyclic deformation capacities severe strength degradation under cyclic loading than under mon-
of these SPSWs when reaching their ultimate strength, failure otonic loading (i.e., the cyclic envelope is smaller than the force-
points, and rates of degradation are statistically quantified. Based displacement backbone boundary).
on these statistical results, initial deterioration models for SPSW Beyond backbone-type models, there exist many sophisticated
components are developed. The chosen deteriorated material hysteretic models that are able to incorporate stiffness degradation,
models for infill plates and boundary elements are calibrated to four strength degradation, or a combination of both deteriorations. An
selected SPSW specimens varying from one to four stories. The extensive review of these hysteretic models can be found elsewhere
development of global structural numerical models and the results (FEMA 2009a). The question remains as to which models are
of the collapse assessment to determine the corresponding SPFs are appropriate to model SPSW components up to failure, and whether
presented in the companion paper (Purba and Bruneau 2014b). cyclic degradation rules should be considered. Observations of
Beyond the current purpose, the resulting calibrated models can force-displacement hysteretic curves for SPSW specimens in past
also be valuable to engineers for progressive collapse assessments experimental research provides guidance about how to select
or performance-based design of individual buildings. appropriate deterioration models for SPSW components (several
examples are presented in later sections). Relating those to the
hysteretic rules for cyclic deterioration, several observations can
Selecting Degradation Models for SPSW be made, as follows:
Components 1. Up to the capping point (i.e., defined as the point of maximum
The FEMA 356 document (2000), which was adapted to become strength before strength degradation starts to occur), stiffness
the ASCE 41 document (2013), provides provisions for the both during loading and unloading are slightly similar between
evaluation and rehabilitation of buildings to improve their seismic hysteretic loops. Moderate changes in stiffness are only
performance. In this document, force-displacement capacity of observed after strength degradation occurs. This behavior
structural components with reliable ductility (prior to exhibiting indicates that incorporating stiffness degradation in the dete-
strength deterioration) are modeled as shown in Fig. 1, commonly rioration models for SPSW components is not crucial for the
known as the backbone curve. Many other backbone models de- type of collapse analysis of SPSWs intended in this research.
veloped in past research on modeling the deterioration of structural 2. Strength degradation owing to repeated cycles at the same dis-
placement is relatively small; as such, it can be neglected in
developing deterioration models for the SPSW components.
3. Strength degradation primarily occurs when the increasing
inelastic displacements pass the capping point in the backbone
curve. Therefore, the cyclic envelope can be designated as
similar to the force-displacement backbone boundary.
4. The significant pinching behavior exhibited in the hysteretic
curve of these SPSWs is a consequence of the fact that the
unstiffened infill plates behave analogously to slender, ten-
sion-only bracing. That behavior can be captured directly
when modeling the wall with tension-only nonlinear braces.
Detailed information on the behavior of unstiffened infill
Fig. 1. Force-deformation relations for structural components (courtesy
plates can be found elsewhere (Roberts and Sabouri-Ghomi
of FEMA 2000)
1991; Berman and Bruneau 2005).
J. Struct. Eng.
Hence, the simple force-displacement backbone curve without Failure of Boundary Elements
cyclic degradation rules is sufficient to represent degradation
Failure of boundary elements (FBE) can be classified as either
models for boundary elements and infill plates in SPSWs. Here, flexural failure, shear failure, or member instability. Flexural failure
degradation parameters, such as plastic deformation when reach- observed in past experiments can be classified into the following
ing the capping point and postcapping stiffness (i.e., rate of types of damage: (1) plastic hinge development at boundary
degradation) will be estimated in a probabilistic framework (Ibarra element ends with ductile strength degradation owing to localized
and Krawinkler 2005), determined based on the results of pre- flange or web buckling; (2) weld fracture at the connections be-
vious SPSW experiments. Quantification of these degradation tween HBEs and VBEs or between VBEs and their base; (3) frac-
parameters for both infill plates and boundary elements, and ture of the boundary elements away from the connections [i.e., at
whether or not residual strength should be considered, is addressed the center line of plastic hinge, particularly in HBEs with reduced
in this paper. beam section (RBS) connections where fractures have occurred
near the midlength of the reduced flange segment]; and (4) shear
tab failure that leads to HBE web and flange fractures.
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
Identification of Deterioration and Failure Modes Shear failure is characterized by shear yielding of a significant
length across the web of VBEs, which causes lower expected VBE
To quantify the degradation parameters for SPSW components, a
plastic moments and can result in significant VBE inward defor-
detailed literature review of SPSW experimental results from
mations (i.e., hourglass shape deformations) owing to the pulling
research in the past 30 years was conducted. This review was
forces from yielded infill plates. Several examples of flexural
intended to investigate three specific objectives: (1) to identify the
and shear failures can be found in past studies (Driver et al. 1997;
deterioration modes associated with the loss of strength and failure
Lubell et al. 2000; Vian and Bruneau 2005; Park et al. 2007; Qu and
modes that occurred at the end of each test; (2) to statistically quan-
Bruneau 2008).
tify at which story drift the capping and failure points occurred; and Deterioration mode associated with instability of boundary
(3) to statistically quantify the rate of degradation, which is defined elements (IBE) has been occasionally reported in past experiments.
as the ratio between strength drop and drift range from the failure to Global instability was reported to associate with out-of-plane
the capping points. A total of 36 conventional unstiffened slender- (weak-axis) buckling of VBEs or lateral-torsional buckling of
web SPSW specimens were examined for which testing data were HBEs; this occurred in the early stages of loading (i.e., as early as
adequately reported and accessible at the time of this writing. The 1% drift). Examples of IBE deterioration mode can be found in past
specimens varied from single-story to four-story SPSWs with studies (Caccese et al. 1993; Elgaaly 1998; Lubell et al. 2000).
aspect ratios ranging from 0.7 to 2.2. Both welded and bolted con- Shear failure and IBE are unlikely to occur in a well-designed
nections were used in these walls, either connecting infill plates by SPSW because they can be prevented by designing boundary ele-
means of fish plates to boundary frames or connecting horizontal to ments according to capacity design principles and by selecting
vertical boundary elements. seismically compact sections to prevent local and global buckling.
Inferred from the experimental data considered here, the Therefore, these failure modes are ruled out as limit states to be
possible causes of deterioration of structural components that lead considered in applying the FEMA P695 methodology (2009b);
to failures of SPSWs consist of deteriorations associated with only the two ultimate failure modes of well-designed SPSWs,
web tearing, flexural or shear failure of boundary elements, and WT and FBE, are considered in this research for quantifying the
instability of boundary elements. Each deterioration mode is deterioration parameters in SPSW. The results are presented in
described in detail in the following. the following section.
Web Tearing
Statistical Estimation of Cyclic Deformation
Past experiments have shown that relatively small fractures of infill Capacity
plates have insignificant degrading impact on the structure’s ulti-
mate strength, (Lubell et al. 2000; Astaneh-Asl and Zhao 2002; Among the 36 SPSW specimens under examination, a large vari-
Vian and Bruneau 2005; Li et al. 2010). Because infill plates pro- ability of experimental outcomes was observed. To avoid a biased
vide superior redundancy to transfer tension forces to surrounding statistical interpretation of cyclic deformation capacity at the ulti-
boundary elements, cracks in one part of a plate strip may only mate (capping) and failure points, only the specimens that were
locally disturb the regularity of the tension stress flows. As such, pushed beyond the ultimate point and exhibited stable deterioration
stresses can flow around and redistribute tension stresses to adja- with gradual strength drop were considered, as listed in Table 1
cent uncracked parts of the infill, maintaining TFA throughout the for 17 selected specimens. A complete list of SPSW specimens
loading history. can be found in (Purba and Bruneau 2014a). On average, SPSW
Web tearing (WT) contributes to the deterioration of SPSW specimens reached their ultimate strength at 3.1% drift.
strength only if fractures of infill plates propagate to significant Regarding the condition at the failure point, cyclic deformation
lengths, which can render parts of the infill plates unable to capacity and percentage of strength degradation need to be
develop TFA. As a result, infill plates progressively lose their estimated. The average SPSW cyclic capacity when reaching the
capacity to sustain loads. Examples of this deterioration mode failure point (when substantial WT, FBE, or a combination of
can be found in specimens tested by Qu and Bruneau (2008) both deterioration modes occurred and the tests ended) was 4.5%
and Choi and Park (2009). In the first example, the cracks initially drift. When reaching the failure points, SPSW specimens lost
occurred at the corner of the panel, then propagated to the connec- approximately 25% of their ultimate strengths. Many tests were
tions of infill plates with vertical and horizontal fish plates; in stopped after a substantial drop in strength was observed (25%,
the second example, the cracks also initially occurred at the same on average), and the actual rate of progression of further damage
location, but this time propagated to the middle part of the infill that would have occurred beyond that point is not known for those
plates. specimens.
J. Struct. Eng.
Table 1. Steel Plate Shear Walls of Tested Specimens
Type of
Geometric properties connectiona Condition at ultimate Condition at end
Number of Lp Hs Aspect V max Drift V end Drift
Researcher Specimen stories (mm) (mm) ratio Frame Infill Mode (kN) (%) Mode (kN) (%) μg
Driver et al. (1997) — b
4 3,050 1,776 1.7 W W WT 3,080 2.2c
FBE 2,618 4.0c
9.0
Lubell et al. (2000) SPSW2 1 900 900 1.0 W W FBE 250 4.00 FBE 175 5.0 7.5
Astaneh-Asl and Zhao (2002) UCB-1 2 —b 3,100 —b W W FBE 4,005 3.3c FBE 2,403 4.0c 5.7
UCB-2 3 —b 2,067 —b W W FBE 5,451 2.2c FBE 4,066 3.0c 4.3
Behbahanifard et al. (2003) —b 3 3,050 1,678 1.8 W W FBE 3,500 2.6c WT 2,850 3.7c 7.9
Berman and Bruneau (2005) F2 1 3,658 1,829 2.0 P W WT 620 3.0 WT 420 3.7 12
Vian and Bruneau (2005) P 1 4,000 2,000 2.0 W+RBS W FBE 1,790 2.0 FBE 1,650 3.0 10
CR 1 4,000 2,000 2.0 W+RBS W FBE 2,050 2.5 FBE 1,340 4.0 13.3
Park et al. (2007) SC2T 3 1,750 1,100 1.6 W W FBE 1,663 2.6d FBE 1,338 3.8e 7.0
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
Qu et al. (2008) —b 2 4,000 4,000 1.0 W+RBS W FBE 4,245 3.3c,f WT 2,387 5.2c,f 10.4
Choi and Park (2008) FSPW1 3 1,650 1,075 1.5 W W FBE 1,392 3.6d FBE 1,364 5.2e 8.1
FSPW2 3 2,350 1,075 2.2 W W FBE 1,817 4.5d WT 1,776 5.6e 11.8
FSPW3 3 2,350 1,075 2.2 W W FBE 1,565 2.7d FBE 1,100 5.4e 10.6
Choi and Park (2009) BSPW1 3 2,350 1,075 2.2 W P WT 1,882 3.6d WT 1,200 5.3d 11.8
BSPW2 3 2,350 1,075 2.2 W P WT 1,961 3.3d FBE 1,05d5 5.3d 11.0
Li et al. (2010) N 2 2,140 3,250 0.7 W+RBS W FBE 1,300 4.0e FBE 1,105 5.0e 12.5
S 2 2,140 3,250 0.7 W+RBS W FBE 1,070 3.0e WT 910 5.3e 12.5
a
P = pin (simple) or partial welded connection; W = welded (rigid) connection; RBS = reduced beam section.
b
Not available.
c
First story drift.
d
Top story drift.
e
Maximum interstory drift.
f
Information from Phase II (i.e., cyclic test).
g
μ ¼ Δend =Δyield .
In the preceding approach, both cyclic deformation capacity Initial Deterioration Models for SPSW Components
and percentage of strength degradation at the failure point were
treated as independent statistic variables. Another approach in The preceding cyclic deformation capacities for the capping and
which both parameters were treated as two related statistical var- failure points were capacities at the structural level. To model
iables was investigated. The process started by preparing a back- deterioration for SPSWs using structural analysis software, similar
bone curve of the cyclic hysteresis for each specimen from the information needs to be developed at the component level. In other
ultimate to the failure points. For a certain percentage of strength words, assuming that the degradation pattern at the structural level
degradation beyond the capping point, the corresponding story is an expression of behavior at the component level, moment-
drift at this degradation level was recorded for each specimen; rotation and axial force-deformation degradation models are re-
then, for all specimens, average and SD values were calculated. quired for boundary elements and infill plates, respectively. These
Interestingly, this slightly more elaborate approach produced models are usually developed from past component tests or past
results only marginally different from those obtained by using the assembly of structural system tests that reported these local behav-
first approach (i.e., 25% strength degradation at approximately iors. However, because neither of these sources is available for
4% drift). SPSWs at the time of this writing (in a way that would uncouple
Several other factors may affect cyclic deformation capacities infill and boundary element contributions), initial deterioration
of specimens when reaching the capping and failure points, such models were developed for boundary elements and infill plates
as panel aspect ratio, specimen scales that dictate boundary indirectly from the available structural-level responses.
element sizes, boundary element compactness, and types of con- This can be achieved by first developing numerical models of
nections between SPSW components and between specimens and several selected specimens, performing monotonic pushover analy-
their bases. In addition to these factors related to specimen geo- sis, and recording local behaviors when the structure experienced
metric properties, the design approaches used to size the SPSWs 3 and 4% interstory drift (i.e., corresponding to the capping and
(e.g., elastic design, plastic design, or capacity design principles global failure points). In the analysis of strip models, both steels
designing specimens as shear-type structures, flexural-type struc- used for boundary elements and infill plates were represented
tures, or a combination of both) and loading protocol imposed by an idealized elasto-perfectly plastic stress-strain material. To
during experiments (e.g., level of axial loads imposed, distribu- consider strength degradation, yield strengths of both steels used
tion of lateral loads along the specimen height for multistory for boundary elements and infill plates were afterward assigned
specimens) may affect specimen capacities to resist earthquake a 25% strength drop at the failure point based on the statistical re-
loads. However, in light of the relatively small number of SPSW sults presented in the previous section. For this purpose, the spec-
specimens tested at the time of this writing, the effect of these imens selected were those of Vian and Bruneau (2005), Qu and
factors on specimen cyclic deformation capacities cannot be stat- Bruneau (2008), Choi and Park (2009), and Driver et al. (1997).
istically investigated and is considered to be beyond the context Detailed information and the rationale for selecting these single-
of this research. As new data become available, the effect of story to four-story specimens are addressed in the next section.
these factors and relationships between these factors may be The resulting initial degradation models are shown in Fig. 2 for
quantified. boundary elements and infill plates. The degradation parameter
J. Struct. Eng.
σ σ
σy B C
σy B C E
0.75σy D E 0.75σy
Es D
Es
A 15εy -εd A
ε ε
εy 11εy -εc -εy εy εc εd
No Compression D -0.75σy
Strength E -σy
C B εc = 0.5h/L p ×0.03
εd = 0.5h/L p ×0.04
(a) (b)
values shown in the figure represented the average values of four Calibration of Proposed Deterioration Model
specimen local behaviors at 3 and 4% drifts.
Although both deterioration modes (i.e., WT and FBE) gener- Calibration of the proposed deterioration models was conducted by
ally take place once SPSW specimens are cycled to relatively large using four selected specimens that represent single-story to four-
lateral displacements (or drifts), many past experiments reported story SPSWs with panel aspect ratios ranging from 1.0 to 2.2. Each
that infill plates can continue to dissipate energy even after failure specimen has a unique characteristic for which observation of
of boundary elements. In such cases, infill plates exhibited non- different scenarios of strength degradation can be made. The
deteriorating WT with significant inelastic deformations. Hence, single-story specimen examined by Vian and Bruneau (2005) ex-
it may be tempting to consider modeling deteriorating SPSWs hibited fractures of boundary elements but no fractures of its infill
by implementing strength deterioration models only for the plates that contributed to the specimen strength degradation (i.e., the
boundary elements while providing elasto-perfectly plastic model infill plates exhibited significant plastic deformations instead). The
without deterioration for infill plates. The problem with this ap- reverse scenario was observed in the three-story specimen tested by
proach is that when significant WT occurs, as observed in several Choi and Park (2009), in which strength deterioration was attrib-
other tests, it may provide incorrect results. uted to WT in the absence of significant damages to boundary
Another legitimate question is whether the residual strength elements. A case in which both fracture of boundary elements
defined for infill plates in Fig. 2(a) is appropriate, considering and deterioration owing to WT was reported by Qu and Bruneau
(2008) for a two-story specimen. Although both deterioration
that once significant tears start to propagate across the infill,
modes were also observed in the four-story specimen of Driver et al.
the corresponding strip plate used in the model at this location
(1997), the strength degradation rate and magnitude of this degra-
should lose its entire strength to sustain loads (i.e., the stress it
dation were not as severe as in the two-story specimen. Considering
carries should completely drop to zero when tears start to propa-
that the four calibrated specimens already covered the ranges of
gate). Furthermore, given that WT that is correlated to strength
aspect ratio, number of stories, drift capacities at the ultimate
degradation generally starts from one of the panel corners, an
and end conditions, and amount of strength degradation for the
accurate model should presumably account for the fact that strips
specimens reported in Table 1, additional calibration was not con-
attached closer to the panel corner lose strength faster than the
ducted and the four calibrated specimens were deemed adequate to
others. The impact is unclear of using different deterioration
represent the intended calibration results.
models for strips, depending on their location from the panel Based on documented experimental information reported for
corner. each specimen (e.g., geometric and section properties, material
To investigate these concerns, a series of monotonic pushover properties, presence of gravity loads, lateral load distributions,
analyses was conducted with various deterioration models as- and cyclic pushover displacement loading histories), a strip model
signed to boundary elements and infill plates. More specifically, was developed in OpenSees for each specimen. The strip model
these analytical investigations were intended to compare the consists of series of tension strips, typically of equal width, pin-
global SPSW deterioration behavior when deterioration models connected to the surrounding boundary elements, and inclined
were assigned to both infill plates and boundary elements for in the direction of the tension field, α, estimated per Eq. (F5-2)
the case in which a deterioration model was only assigned to of AISC (2010) that considers geometric properties of the infill
either boundary elements or the infill plates. In addition, impact plates and boundary elements at the story level of interest. A mini-
of severe and moderate deterioration models assigned to corner mum of 10 strips per panel is required to accurately represent the
and middle strips, respectively, was investigated and compared infill plate behavior (Thorburn et al. 1983) and the width of each
with the case of uniform deterioration models. Finally, different strip (Sdiag ) can be calculated as (L cos α þ H sin α) divided by the
deterioration rates for infill plates were investigated, considering number of specified strips, where L and H are the width and height
abrupt drop to zero stress at a certain specified strain level and of the panel. In the case of multistory SPSWs, equally spaced strips
gradual drop to zero stress at various rates. Based on the findings with slightly different tension field angles from one story level to
of this parametric study, reported comprehensively in Purba and another most likely result in a strip model having staggered node
Bruneau (2014a), the initial degradation models were modified by points at the HBE in adjacent stories. However, for practical pur-
calibrating proposed degradation models to available experimental poses, it is preferable to use a strip model in which the strips have
data. By matching the results of analyses using these numerical the same inclination angle at all stories (i.e., using the average of
models to their corresponding experimental results, key deteriora- the tension field angle of all panels) and have common nodes at the
tion model parameters (i.e., the capping Point C and the residual HBE in adjacent stories. Such slight variations in angle have insig-
strength Point D in Fig. 2) for boundary elements and infill plates nificant consequences on the results (Dastfan and Driver 2008).
could be estimated. Furthermore, an eccentricity exists between boundary element
J. Struct. Eng.
centerlines and the edge of the infill plate; this rigid offset was not Single Story SPSW: Vian and Bruneau Specimen
included, but instead, strips were directly connected the boundary
The first calibration was conducted on the single-story SPSW
elements. This simplification is reasonable in the current context
specimen tested by Vian and Bruneau (2005). Bottom flange frac-
and is of negligible impact on the results of this particular study.
tures at both RBS locations of the lower HBE occurred at 3.0%
The validity of the strip model to accurately represent the nonlinear
interstory drift and caused a strength degradation of 17.5% from
behavior of SPSWs is well established (Driver et al. 1997; Elgaaly
the specimen ultimate strength of 2,060 kN, which was reached
1998; Berman and Bruneau 2005; Qu and Bruneau 2008).
in the previous displacement cycle of 2.5% interstory drift.
Moreover, unstiffened infill plates were modeled as series of
Moreover, fractures at the connection of the upper HBE to the VBE
truss elements in OpenSees oriented in the direction of the tension
occurred at 4.0% interstory drift and caused a strength degradation
field. The hysteretic uniaxial material model was selected to
of 24% from the ultimate strength of the specimen. Regarding the
define the inelastic behavior of these truss elements. Boundary
infill plate, noticeable plastic deformations and minor cracks were
elements were modeled by using the beam with concentrated
observed but had insignificant impacts on the overall deterioration
hinge (BCH) element with fiber sections and hysteresis uniaxial
behavior of the specimen. In the strip model developed for this
material model. The concentrated hinge with fiber sections auto-
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
J. Struct. Eng.
(a) 354 mm (b) M
Mcap
42.7° SH
My = 2%
40%Mcap
44.5° EI
θ
4000 mm
-θy θy
2000 mm
Symmetric 0.031 0.045 P = 16%Py
0.039 0.064 P=0
Left-leaning
strips are not -My 3.0% 3.7%
shown for first story drift
clarity For non-deteriorating
boundary elements
Left-leaning strips are P
not shown for clarity 11
(c) Pcap
12 10 SH = 2%
9 Py
8
7
EA
4000 mm
4
Axial hinge with Fiber hinge with 2
850 mm
1 δy δ
m non-deteriorating material deteriorating material 3 7δy
m 6
6 model (c) model (b) No Compression
26 5
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
Strength
Legend:
| = Fiber Hinge with non-deteriorating
| deteriorating material model
4000 mm | = Axial Hinge with non-deteriorating
4000 mm | deteriorating material model
(a) (d) P
Pcap Seq. δC δD ∆C ∆D
M P SH = 2%
Mcap Pcap Py 1 7.6δy 8.1δy
3.0% 3.2%
SH SH = 2% 2 8.0δy 8.6δy
My = 2% 40%Mcap Py EA 3 7.6δy 9.0δy
3.2% 3.7%
EI EA δ 4 9.0δy 10.6δy
δy δC δD 5 7.1δy 9.1δy
-θy θy θ δy δ No Compression
0.038 11.6δy 6 8.2δy 10.2δy
Symmetric
0.023 Strength ∆C ∆D 3.7% 4.3%
No Compression 7 10.4δy 12.8δy
2.5% 4.0% Strength 2.5%
8 10.2δy 12.6δy
-My interstory drift interstory drift δC | δD = Strip elongation at the capping
9 12.0δy 13.9δy
| failure point
∆C | ∆D = Corresponding first story drift 10 10.8δy 13.0δy
4.3% 4.8%
(b) (c) when reaching the capping 11 8.4δy 10.3δy
| failure point 12 1.6δy 2.4y
Interstory Drift
-4% -3% -2% -1% 0% 1% 2% 3% 4% 1st Story Drift
2500 (e)
-4% -2% 0% 2% 4% 6%
5000
1500
3000
Base Shear (kN)
500
Base Shear (kN)
1000
-500
-1000
-1500 Experiment
Analytical -3000
Experiment
-2500 Analytical
-80 -60 -40 -20 0 20 40 60 80
-5000
Interstory Displacement, ∆ (mm) -161 -80 0 80 161 241
(d) 1st Story Disp. (mm)
Fig. 3. Calibration of the single story SPSW specimen tested by Vian Fig. 4. Calibration of the two-story SPSW specimen tested by Qu and
and Bruneau (2005): (a) strip model; (b) material model for boundary Bruneau (2008): (a) strip model; (b) material model for boundary
elements; (c) material model for strips; (d) resulting hysteresis curve elements; (c) material model for nondeteriorating strips; (d) material
model for deteriorating strips; (e) resulting hysteresis curve
J. Struct. Eng.
infill plate, causing the base shear strength to drop by approxi- This started from middle strips at the onset of 3.3% top story drift
mately 46%, to 1,055 kN at 5.3% top story drift. Based on this and gradually propagated toward strips adjacent to the upper left
failure mechanism, in the strip model developed for this specimen and lower right corners at 5.3% top story drift. The resulting
(Fig. 5), deteriorating material models were assigned to most of the force-displacement hysteresis of the specimen model is shown
strips at the second story. Minor plate tearing and plate kinking in in Fig. 5(d), plotted with that from the cyclic pushover test. Overall,
the first and third story infill plates were considered to be nonde- the two hysteresis values agree well with negligible discrepancies at
teriorating web tearing (NWT); therefore, an elasto-plastic material the onset of effective yielding, the capping point, and the degrada-
model with 2% strain hardening up to the capping point was used at tion backbone up to completion of the test. However, the two
these stories instead of a deteriorating material model. Moreover, hysteresis values exhibit slightly different pinching behavior.
flange fracture at the upper end of the second story VBE occurred Similar to the calibration results for the Vian and Bruneau (2005)
in the last cycle of displacement loading. However, because no in- case, the strip model exhibited severe pinching behavior, as
dication reported that this fracture initiated at an earlier cycle of opposed to the moderate pinching behavior observed during the
displacement loading, all boundary elements were considered to experiment.
have nondeteriorated material model similar to the strips at the first
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
SH
My = 2% corner of the first story infill plate and flange local buckling at both
Left-leaning strips are
not shown for clarity EI ends of the east VBE1 (i.e., below HBE1 and near the base) and at
-θy θy θ the upper end of the west VBE1 were reported in earlier cycles
9 87 0.032
1150 mm
Symmetric
6 4 2 1 3 5 before reaching the capping point. As the WT propagated to a
3.3%
-My top story drift
larger size and the severity of VBE flange local buckling increased,
150 mm
the lateral strength of the specimen started to deteriorate. However,
the degradation rate was relatively slow compared to that of the
1075 mm
J. Struct. Eng.
(a) (b) (as opposed to cycling testing). Hence, the compressive strength
M
Mcap
of the strip was excluded from the final strip deterioration model.
SH
My = 2%
40%Mcap
42.3°
EI
For non-deteriorating BEs Considering the preceding calibration results, particularly the fact
that different models were used to replicate each of the four exper-
(c) P
Pcap imental results, one might ask which of these degradation param-
Py SH For non- eters should be selected to generally capture the degradation of
= 2% deteriorating
strips (SH up boundary elements and infill plates in any specific SPSW. Because
EA to 8δy)
only a limited number of specimens was calibrated, the approach
δ y 3.7δ y δ
No 7.6δ y selected here was to examine the worst degradation parameters
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
Compression
265 mm Strength from the preceding cases, as opposed to the average values of
all four specimens. Hence, for the collapse assessment of steel plate
See Fig. 4 for shear walls, degradation models for boundary elements and infill
legend definitions
plates were selected from the calibration results of the two-story
SPSW specimen of Qu and Bruneau (2008). The model for boun-
3050 mm
dary elements in Fig. 4(b) was calibrated for the condition when
(d) 1st Story Drift HBE1 experienced axial loads, P, equal to 2,193 kN (i.e., P =
-2.1% -1.0% 0.0% 1.0% 2.1% 3.1% 4.2% 16% Py). To apply the model to other specimens, the model
3200
was first modified for the condition at zero axial loads. The updated
2400 rotation capacities for the capping point and the point correspond-
ing to 40% degradation of moment capacity for boundary element
1st Story Shear, V (kN)
1600
model are 0.039 and 0.064 rad, respectively [Fig. 4(b)]. Regarding
800 the infill plates, the strip model was based on results of deteriora-
0
tion of the first strip. As shown in Fig. 4(d), the strip deforma-
tion capacities at the capping and failure points were 7.6δ y and
-800 8.1δy , respectively (corresponding to 1.3 and 1.4% axial strain,
-1600 respectively).
Experiment To examine the impact of selecting the most conservative deterio-
-2400 Analytical ration models (i.e., the models with the most severe degradation),
-3200 another set of analyses was conducted on all calibrated specimens.
-40 -20 0 20 40 60 80
1st Story Lateral Displacement, ∆ (mm)
The resulting hysteresis values for all calibrated specimens using the
conservative deterioration models are shown in Fig. 7. As expected,
Fig. 6. Calibration of the four-story SPSW specimen tested by Driver because the conservative deterioration models were selected from the
et al. (1997): (a) strip model; (b) material model for boundary elements; Qu and Bruneau (2008) specimen, the analytical result was close to
(c) material model strips; (d) resulting hysteresis curve the calibration result of the experimental hysteresis in that case. In
the specimen of Vian and Bruneau (2005), the conservative model
actually delayed degradation, which had moved from 2% interstory
drift in the experiment to 3.3% interstory drift in the analysis. How-
Although, in all cases, the calibration process successfully ever, beyond this point, strength degradation was significant and
matched the capping point and the backbone of the degrading hys- abrupt. In other words, deterioration of infill plates and boundary
teresis, discrepancies in pinching behavior were observed between elements occurred at similar times. Significant differences were
the numerical and experimental results. One possible explanation observed in the Choi and Park (2009) specimen. Strength degrada-
for this underestimation of the experimental pinching behavior is tion started to occur at 2.7% drift, which is sooner than in the actual
that the strip material models exclude the compressive strength that experiment, where it happened at 3.5% drift. In addition to boundary
can develop during loading reversal. This assumption is reasonably element deterioration, all strips lost strength by 3.5% drift and only
correct to simulate infill plate behavior in earlier cycles. However, it the second story infill plate lost strength at 5.3% drift in the actual
has been observed in some cyclic tests that after SPSWs undergo experiment. Finally, for the Driver et al. (1997) specimen, a similar
substantial inelastic elongation, the infill plates can exhibit signifi- response to that reported in the actual experiment was observed in
cant compressive strength during load reversal, as reported by this specimen, up to 2.1% top story drift. Beyond this point, however,
Clayton et al. (2012). When including compressive strength during faster strength degradation was observed in the conservative case
reversal of loading in the material model for the infill plates, Choi within the positive excursions, but not for the negative excursions.
and Park (2010) successfully simulated the pinching behavior At the conclusion of the analysis, the specimen had lost 67.5% of its
observed in the Driver et al. (1997) experiment. However, their strength, down to only 1,000 kN. This outcome was a consequence
numerical model failed to simulate the deteriorating hysteresis of the fact that, in the model, all strips and HBEs at each floor con-
behavior after the capping point. During the calibration process, tributed to the degradation, compared to only the first floor elements
no effort was made to resolve the underestimated pinching behavior in the actual experiment.
because it was considered to have a marginal impact on the overall As mentioned previously, the conservative strip model selected
collapse performance of steel plate shear walls. Furthermore, recent from the calibration of the Qu and Bruneau (2008) specimen was
tests by Dowden and Bruneau (2014) showed that the effect of based on result of its first strip to deteriorate. In the initial calibra-
this compressive strength was insignificant in shake table tests tion, 15 strips were used to model the first story infill plate
J. Struct. Eng.
Story Drift Top Story Drift
-4% -3% -2% -1% 0% 1% 2% 3% 4% -7.0% -5.3% -3.5% -1.8% 0.0% 1.8% 3.5% 5.3% 7.0%
2500 3000
2000
2000
1500
1000
Total Force (kN)
1000
0 0
-500
-1000
-1000
-2500 -3000
-80 -60 -40 -20 0 20 40 60 80 -240 -180 -120 -60 0 60 120 180 240
(a) Displacement, ∆ (mm) (c) Top Story Displacement (mm)
4000 2400
3000
1600
800
1000
0 0
-1000 -800
-2000
-1600
-3000 Matching Case
Conservative Case
-2400
-4000 Matching Test Conservative Case
-5000 -3200
-161 -80 0 80 161 241 -40 -20 0 20 40 60 80
(b) 1st Story Disp. (mm) (d) 1st Story Lateral Displacement, ∆ (mm)
Fig. 7. SPSW specimen hysteresis with the most conservative degradation models applied to calibrated specimens: (a) Vian and Bruneau (2005);
(b) Qu and Bruneau (2008); (c) Choi and Park (2009); (d) Driver et al. (1997)
degradation behavior, each with different degradation parameters linearly decrease to zero strength when cross-sectional rotation
(i.e., deformation capacities at the capping and failure points) to reached 0.103 rad. At the fiber level, this corresponds to 0.057
match the experimental results [Fig. 4(a)]. On average, the defor- strains in the farthest fiber from the neutral axis.
mation capacities of these strips at the capping and failure points During the first analysis of the Qu and Bruneau (2008) specimen
were 9δy and 10.7δy , respectively (corresponding to 1.5 and 1.8% subjected to such an extreme drift, a numerical convergence prob-
axial strain, respectively). When using these average results for the lem developed after the specimen experienced 6% drift. Further
infill plate deterioration models instead of using the result of the investigation revealed that the primary source of the problem
first strip to deteriorate, marginal differences were observed in was the inability of the boundary elements modeled with fiber
the two-story and three-story SPSW specimens (detailed results elements to sustain axial loads after most of the fibers had lost their
are not reported here). Assuming that the marginal differences flexural strength at large drift. To solve this problem, it an elasto-
would also be the case for the other two specimens (i.e., comparable plastic material was assigned for fibers on the web of boundary
to the results shown in Fig. 7), the average deformation capacities elements and a deteriorating material model was assigned for
were selected for the final deterioration model for strips, as shown the fibers located on the bottom and top flange of the cross section,
in Fig. 8(a). This was necessary to avoid an overly conservative as shown in Fig. 8(b). This technique made it possible for the analy-
deterioration model for strips in the general case of SPSWs. ses to execute fully when the structures experienced drifts of up to
At the time of this writing, no SPSW specimen has been tested 10%. In other words, this approach was equivalent to having the
up to extreme drifts (i.e., 10% drift). Among the currently available fibers on the web of a cross section reserved to sustain axial loads
experimental data (Table 1), the maximum drift to which SPSW when all other fibers had lost their capacity owing to significantly
specimens have been tested is 5.6% drift. For collapse assessment large cross-sectional rotations. As a consequence of this approach,
of SPSW using incremental dynamic analysis [as addressed in the boundary elements actually exhibit residual flexural strength when
companion paper (Purba and Bruneau 2014b)], it is important to reaching 0.103 rad, as opposed to the zero strength originally
investigate specimen behaviors up to such extreme drifts because intended. However, for W-sections commonly used in North
this may condition the results. For this purpose, another set of America, the web of W-shapes contributes, on average, approxi-
analyses was conducted on all calibrated specimens up to 10% mately up to 20% of the total plastic moment of a section. There-
drift. The deterioration of boundary elements was modeled to fore, the preceding approach was deemed acceptable, particularly
J. Struct. Eng.
quickly lost their capacity to sustain loads owing to plate tearing,
at 1.8% axial strain (i.e., 10.7δ y ). All strips were established with
the same deterioration model, irrespective of the location of strips
(i.e., corner or middle strips or floor level). This model exhibited
stable behavior and was found to provide an acceptable match with
experimental results for the perspective of investigating seismic
performance of SPSW having infill plates designed considering
two different design philosophies. Using these deterioration mod-
els, the companion paper (Purba and Bruneau 2014b) presents the
results of these seismic performance assessments, including the
development of SPSW archetypes, the formulation of a nonlinear
analytical model, the results of nonlinear static and dynamic analy-
ses (i.e., pushover and incremental dynamic analyses), and the
evaluation of collapse performance.
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
Acknowledgments
J. Struct. Eng.
Dowden, D., and Bruneau, M. (2014). “Analytical and experimental Lubell, A. S., Prion, H. G. L., Ventura, C. E., and Rezai, M. (2000).
investigation of self-centering steel plate shear walls.” Tech. Rep. “Unstiffened steel plate shear wall performance under cyclic loading.”
MCEER-14-0010, Multidisciplinary Center for Earthquake Engineering J. Struct. Eng., 10.1061/(ASCE)0733-9445(2000)126:4(453), 453–460.
Research, State Univ. of New York at Buffalo, Buffalo, New York. OpenSees version 2.0 [Computer software]. Berkeley, CA, Pacific Earth-
Driver, R. G., Kulak, G. L., Kennedy, D. J. L., and Elwi, A. E. (1997). quake Engineering Research Center, Univ. of California.
“Seismic behavior of steel plate shear walls.” Structural Engineering Park, H. G., Kwack, J. H., Jeon, S. W., Kim, W. K., and Choi, I. R. (2007).
Rep. 215, Dept. of Civil Engineering, Univ. of Alberta, Edmonton, “Framed steel plate wall behavior under cyclic lateral loading.”
AB, Canada. J. Struct. Eng., 10.1061/(ASCE)0733-9445(2007)133:3(378), 378–388.
Elgaaly, M. (1998). “Thin steel plate shear walls behavior and analysis.” Purba, R., and Bruneau, M. (2014a). “Seismic performance of steel plate
Thin Walled Struct., 32(1–3), 151–180. shear walls considering various design approaches.” Tech. Rep.
FEMA. (2000). “Prestandard and commentary for seismic rehabilitation of MCEER-14-0005, Multidisciplinary Center for Earthquake Engineering
buildings.” FEMA Rep. No. 356, ASCE for FEMA, Washington, DC. Research, State Univ. of New York at Buffalo, Buffalo, New York.
FEMA. (2009a). “Effects of strength and stiffness degradation on seismic Purba, R., and Bruneau, M. (b). “Seismic performance of steel plate shear
response.” FEMA Rep. No. P440A, Applied Technology Council for walls considering two different design philosophies of infill plates. II:
FEMA, Washington, DC. Assessment of collapse potential.” J. Struct. Eng., 10.1061/(ASCE)ST
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.
J. Struct. Eng.