Cyclic Tests of Four Two-Story Narrow Steel Plate Shear Walls - Part 1: Analytical Studies and Specimen Design

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS
Earthquake Engng Struct. Dyn. 2010; 39:775–799

Published online 15 December 2009 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.977
Cyclic tests of four two-story narrow steel plate shear walls—

Part 1: Analytical studies and specimen design
Keh-Chyuan Tsai1,2, ∗, †, ‡, § , Chao-Hsien Li2, ¶ , Chih-Han Lin2, ¶ , Ching-Yi Tsai1, 2,

and Yi-Jer Yu2, ¶
1 Department of Civil Engineering, National Taiwan University, Taipei, Taiwan
2 National Center for Research on Earthquake Engineering, Taipei, Taiwan
SUMMARY
This study consists of two parts. In this two-part research, four two-story narrow steel plate shear walls
(SPSWs) were cyclically tested at the Taiwan National Center for Research on Earthquake Engineering
in 2007. This paper, Part 1, proposes a capacity design method for the first-story boundary column of the
SPSW to ensure that the plastic hinges form at the column bottom ends when the SPSW develops the
plastic mechanism. The design method was developed based on the superposition method considering
the frame sway action and the panel force effects of the SPSW. Restrained steel plate shear wall
(R-SPSW) studied herein adopts pairs of the horizontal restrainers sandwiching over both sides of the infill
panels and connected to the boundary columns. Analytical studies on four SPSW example designs using
nonlinear finite element (FE) models and the simplified strip models confirm that the restrainers could
also effectively reduce the column force demands and allow the infill panel to stretch more uniformly. In
addition, the FE analytical studies verify the effectiveness of the proposed column capacity design method
and the seismic design recommendations for the restrainer. This paper introduces the designs of the four
narrow SPSW specimens, presenting the selections of the boundary beams and columns, the designs of
the beam-to-column connections and the construction details of the restrainers. The experimental results,
key observations and the design implications are reported in the companion paper. Copyright q 2009
John Wiley & Sons, Ltd.
Received 9 August 2008; Revised 7 September 2009; Accepted 11 September 2009
KEY WORDS: steel plate shear wall (SPSW); restrained steel plate shear wall (R-SPSW); capacity design
∗ Correspondence to: Keh-Chyuan Tsai, Department of Civil Engineering, National Taiwan University, No. 1, Sec. 4,
Roosevelt Rd., Taipei 10617, Taiwan.
†
E-mail: kctsai@ncree.org
‡
Professor.
§ Director.
¶ Assistant Research Fellow.
PhD Student.
Copyright q 2009 John Wiley & Sons, Ltd.

776 K.-C. TSAI ET AL.
1. INTRODUCTION
Steel plate shear wall (SPSW) has seen increased usage in North America and Asia in recent
years. As shown in Figure 1(a), an SPSW is composed of a structural frame and infill steel plates.
The beams and columns surrounding the infill panels are named as boundary elements. SPSW can
effectively resist horizontal earthquake forces by allowing the development of diagonal tension
field action after the infill plate buckles in shear. The energy is then dissipated through the cyclic
yielding of the infill plates in tension. The concept of using the post-buckling strength of the steel
plate was first proposed by Thorburn et al. [1]. In their research, a strip model (Figure 1(b)) and an
equivalent brace (EB) model (Figure 1(c)) were developed. The strip model uses at least 10 inclined
truss members pin-connected to the columns and beams to represent the tension field action of a
steel plate. The method of calculating the angle of the tension field action has been developed [2].
The EB model assumes that all tension strips develop equal strain and can be replaced by a
single diagonal truss member that is pin-connected at the two ends to the supporting frame. In
[3], Sabouri-Ghomi et al. proposed a plate–frame-interaction (PFI) model for analysis of SPSW.
Considering the effects of diagonal compressive panel stress, the PFI model could satisfactorily
predict the initial stiffness and shear strength of the SPSW frame with thin or thick plates.
Many experimental studies about the SPSWs have been conducted [4–13]. These prior researches
focused on the thin infill panels with width-to-thickness ratios ranging from 300 to 2000, which
are categorized as ‘slender’ plates [14]. Many of theses studies have confirmed that the elasto-
plastic strip models could satisfactorily simulate the cyclically inelastic responses of the SPSW
specimens with thin infill panels. Two model seismic steel building codes [15, 16] adopt the strip
model as the major analytical tool in the design of SPSWs. The strip model is advantageous
because it can be easily applied to practical structural design. Therefore, this research follows the
simple strip model in exploring the capacity design requirements of the boundary elements and
illustrating the nonlinear responses of typical thin SPSWs. The effectiveness of the strip model in
predicting the large inelastic responses of the thin SPSW has been verified by comparing it with
the test results and the refined finite element (FE) analysis using shell elements. The results will
be presented in this study.
Figure 1. The illustration of (a) an SPSW; (b) a strip model; (c) an equivalent brace model; (d) a restrained
SPSW; (e) and (f) the details of the restrainer connections.
Copyright q 2009 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2010; 39:775–799
DOI: 10.1002/eqe
CYCLIC TESTS OF FOUR TWO-STORY NARROW SPSWS 777
The capacity design requirements for the boundary beams have been investigated by Vian and
Bruneau [10]. Recently, Park et al. [12] proposed a simplified method for estimating the axial and
flexural demands on the columns. However, Park et al.’s method only considers the force demands
induced from the plastic tension field forces without the actions from the swaying of the boundary
element frame. Berman and Bruneau [17] proposed a design method for calculating the ultimate
column axial and flexural demands for SPSWs with pin-supported column bases. This method
was developed based on their previous study of the plastic analysis methods for SPSWs [18]. In
the present study, a capacity design method that ensures the formation of the plastic hinges at the
first-story column bottom ends has been developed considering the combined effects of the frame
sway action and the panel forces.
Two major issues that arise when the tension field action is developed in a typical SPSW include
large out-of-plane deformations and unpleasant sounds from the buckling of the infill panels. These
issues can be alleviated with the application of several types of stiffened shear wall, such as the
addition of welded rib-stiffeners to the infill steel plates, which is a common practice found in
Japan. In other cases, reinforced concrete plates are attached to one or both sides of the infill
plates [19, 20]. The idea of a restrained SPSW (R-SPSW) (Figure 1(d)), bolting pairs of steel
structural member over the two sides of the panel (Figure 1(e)), has also been proposed by Lin
and Tsai [9]. These horizontal steel restrainers can be conveniently connected to the two columns
using slotted or pin-end conditions (Figure 1(f)). In order to prevent the interference of the ‘sliding’
restrainers on the tension stretching of the panel in the tension field direction, enlarged holes for
the bolts through the steel panel have been applied to the steel panels or restrainers. Past studies
[9] have confirmed that the slotted-end restrainers can effectively reduce the out-of-plane buckling
deformations of SPSWs. In this research, the effectiveness of the pin-end horizontal restrainers to
reduce the force demands imposed on the boundary elements has been investigated extensively
using analytical and experimental methods. Seismic design methods for this kind of R-SPSW are
proposed in this paper.
The current model seismic code [16] prescribes that the width-to-height aspect ratio of the panel
be limited within 0.8–2.5. The past researches [21, 22] show that the simplified strip model may
not accurately predict the responses of the narrow SPSWs with relatively light boundary elements
and having an aspect ratio lower than 0.8. The inaccuracy of the strip model could be attributed to
the flexure mode deformation of the narrow SPSW. It is somewhat different from the assumption
for the strip model in which the panel deforms in a shear mode. Lubell et al. [6] also suggested
that in narrow SPSWs, the high axial and flexural column forces caused by the high overturning
moments compared with the story shears would decrease the overall stiffness of the shear wall. The
main problem of the narrow SPSW could be in the design of the boundary column as the flexural
demands imposed on the boundary columns in a narrow SPSW could be more severe than those
in a wide SPSW. The narrow SPSW could be more desirable for architectural reasons. However,
tests conducted on narrow SPSWs with an aspect ratio lower than 0.8 were not available. Based
on the capacity design requirements, the boundary column size in a wide and short SPSW may be
governed by the strong-column weak-girder design. For a narrow SPSW, however, the boundary
column size may be governed by the flexural forces induced from the infill panels. Thus, the
application of the restrainers in a narrow SPSW could be more cost-effective than that in a wide
SPSW. This combined analytical and experimental researches uses narrow SPSWs, as examples,
to investigate its seismic analysis and design methods. First, this research investigates the capacity
design methods for the boundary columns in the typical SPSW and R-SPSW to assure that it can
develop the desirable plastic mechanism under lateral forces. In order to verify the effectiveness
DOI: 10.1002/eqe
of the proposed design and analysis methods for the SPSWs with or without restrainers, four two-
story full-scale narrow SPSW specimens with a panel width-to-height aspect ratio ranging from
0.62 to 0.64 were fabricated and tested at the Taiwan National Center for Research on Earthquake
Engineering (NCREE) in 2007. This Part 1 paper introduces the capacity design for the boundary
column, the analyses of the R-SPSW responses and the designs of the stated four specimens. The
companion paper, Part 2 [23], presents the experimental results and key observations. Additional
analytical studies and design implications are discussed in Part 2.
2. CAPACITY DESIGN FOR FIRST-STORY COLUMN
2.1. Capacity design using method of superposition

The action on the boundary elements when an SPSW is subjected to lateral forces can be considered
as the superposition [10, 11] of two parts as shown in Figure 2. One part is only due to frame sway
whereas the other part is due to panel forces. For properly portioned thin SPSWs, the steel panel
would buckle in shear far before the tension yielding in the tension field direction. As inter-story
drift increases, the tension yielding near the panel central region will grow and propagate into the
corners and the entire panel. Thus, for the capacity design purposes, the fully yield panel forces
acting on the boundary elements can be reasonably considered to be uniformly distributed loads
(Figure 3) with a inclined angle from the vertical consistent with the tension field angle . The
magnitude of the forces acting on the boundary elements can be estimated using the following
equations:
cvJ = bhJ = Ry f yp tp cos sin
chJ = Ry f yp tp sin2 (1)
bvJ = Ry f yp tp cos2
For all the force components given in Equation (1), the first subscript, ‘c’ or ‘b’, denotes that the
force is acting on the column or beam, respectively. The second subscript, ‘v’ or ‘h’, represents
Figure 2. (a) Bending moments and axial forces in the boundary columns of an SPSW
subjected to the lateral forces. It can be considered as the superposition of: (b) the frame
sway action and (c) the panel force effect.
DOI: 10.1002/eqe
Figure 3. Fully yielding panel forces acting on the boundary elements.
the vertical or horizontal force components respectively. The third subscript ‘J ’ represents the
infill panel in the J th story. As an example: ch1 is the horizontal component of the uniformly
distributed panel force acting on the 1F column. In Equation (1), the f yp and tp are the yield stress
and thickness of the infill panel. Ry is the material overstrength factor.
As prescribed in the model seismic steel building code [16], except for the plastic hinges forming
at the beam ends, the boundary elements must be designed to remain essentially elastic under
the panel yield forces. In a typical SPSW building, the columns are often fix-connected to the
base. Thus, plastic hinges will also form at the first-story column bottom ends when an SPSW
develops the desirable uniform yielding mechanism [18]. Figure 2 illustrates the trend of column
bending moment and axial force distribution when an SPSW is subjected to lateral forces in the
elastic level. For simplicity, the panel force effects can be represented as uniformly distributed
loads. For the purpose of discussion, the first-story column is denoted as 1F column through out
this study. At the bottom end of the tensioned 1F column (Figure 2), the bending moments under
the frame sway action and panel force effect are in the same direction. Thus, the peak moment
would be located at the bottom end. Therefore, it is obvious in the tensioned 1F column that the
plastic hinge would form at the bottom end. For the compressed 1F column, however, the bottom
end moments induced from the stated two effects are in the opposite directions. Therefore, the
shape of the quadratic-distributed moment in the compressed 1F column (Figure 2) implies that
the plastic hinge may form within the column height if the panel force effect is strong enough.
On the other hand, the compressed 1F column top end may also develop plastic hinge because the
moments induced from the stated two effects are in the same direction (Figure 2). This issue will
be discussed in greater detail in the end of Section 2.3. Plastic hinges forming at the mid-column
height may adversely affect the seismic performance of an SPSW as described later in this paper.
In addition, the axial force in the compressed column is much greater than that in the tensioned one
because of gravity load effect. Therefore, the capacity design of the 1F column must be carefully
examined for the compressed column.
2.2. Axial force demand
Berman and Bruneau [17] proposed the method to compute the column axial and shear forces
for an SPSW with pin-supported column bases when the SPSW reaches the uniform yielding
DOI: 10.1002/eqe
Figure 4. Force diagrams for calculating the column axial forces.
Figure 5. Plastic hinge locations and moment diagrams for the compressed 1F columns
in the proper and improper designs.
mechanism. In this study, although the 1F column base is fixed, their method is adopted herein
for computing the axial force demand. Figure 4 outlines the details of computing the column axial
forces due to the overturning action. The 1F column axial force demand results from the yielding
of the upper structure, which is caused by the beam-end shear forces and the vertical components
of the panel forces acting along the column. In Figure 4, VcJ or VtJ represents the J th floor
beam-end shear transferred to the compressed or tensioned column, respectively. In addition, the
axial force due to the gravity loads must be considered when computing the axial force on the
compressed 1F column.
2.3. Flexural demand

From the general capacity design principle, the yielding of the boundary column must occur after
the panels fully yield. Superposition method is used to estimate the compressed 1F column flexural
demand developed after the full yielding of the panel but before the yielding of the column.
As shown in Figure 5, the moment from frame sway action can be assumed to be linearly distributed.
Assume that the bottom-end moment is times than that at the top. The location of the inflection
point or the value of is dependent on the stiffness of the boundary elements. The value will
DOI: 10.1002/eqe
increase if the second floor beam stiffness decreases. This location should vary when an SPSW
is deformed from the elastic level to the inelastic level because yielding of the panels and the
beams causes the structure stiffness to change. It is difficult, however, to use a simplified procedure
to compute the actual variation of when an SPSW reaches the plastic mechanism. Therefore,
in this study, the value of is assumed to remain constant from elastic to inelastic ranges. This
paper recommends that the value of in the inelastic range can be estimated by conducting an
elastic static analysis using the EB model [1]. The quadratic-distributed column moment due to
the plastic tension field forces can be estimated by assuming that both ends of the 1F column are
fixed and the column is subjected to a uniformly distributed load ch1 . Before the column plastic
hinge forms, the resulting bending moment distribution M(y) in the compressed column can be
estimated from the superposition:

ch1 1+
M(y) = [−6y +6h 1 y −h 1 ]+Mf 1−
2 2
y (2)
12 h 1
As shown in Figure 5, the Mf is the swaying moment developed at the top of the 1F column. In
addition, Figure 5 displays the differences in the 1F column moment diagrams and locations of the
plastic hinges between the proper and improper designs. Figure 5 can also be used to explain the
effects of the column strength on the locations of plastic hinge formation. If the column strength is
sufficient (refer to Figure 5(a)), it will resist enough swaying moment such that the maximum of
the resulting superposed moment locates at the bottom. This column moment distribution implies
that the plastic hinge will form at the bottom. If the column strength is insufficient, the plastic
hinge will form within the column height (Figure 5(b)). To ensure that the location of the maximum
superposed moment is at the column base, the derivative of M(y) must satisfy the condition of
M (0)0. In this case, the inequality Mf ch1 h 21 /(2(+1)) can be obtained. Then the moment
at the bottom, M(0), can be computed by substituting Mf = ch1 h 21 /(2(+1)) into Equation (2).
The required flexural strength, Mp−req can be computed from the following equation:

ch1 h 21 ch1 h 21 1
M p-req = − = − ch1 h 21 = ch1 h 21 (3)
2(+1) 12 2(+1) 12
Thus, the flexural capacity of the 1F column under the influence of the axial load must be greater
than Mp-req . The index = /(2(+1))− 12 1
can be further simplified to compute the flexural
demand under the combined effects of frame sway and panel forces. Figure 6 shows the relationship
between and . As stated earlier, the value will increase if the second floor beam stiffness
Figure 6. Relationship between and .
DOI: 10.1002/eqe
decreases. From Equation (3) or Figure 6, it can be found that converges to 0.42 when the
connecting beam is much more flexible than the 1F column or when the inflection point is very close
to the column top. Therefore, for preliminary design purposes, Mp-req can be taken as 0.4ch1 h 21
to provide a conservative estimate of the column flexural demand.
In general, for a multi-story SPSW, the swaying bending moment at the compressed 1F column
bottom end is usually much greater than that at the top end. Thus, after superposing the bending
moment due to the panel force effects, the resultant bottom end’s flexural demand is usually greater
than the top end’s. Furthermore, the bottom end’s axial force is indeed greater than the top end’s,
making it highly unlikely that the plastic hinge will only form at the column top whereas its
bottom end remains elastic. However, at the compressed column’s top end, the bending moment
components due to the frame sway action and the panel force effects are in the same direction
(Figure 5). Thus, the top end’s bending moment could be greater than the bottom end’s, suggesting
that a plastic hinge could form at the top end. This situation should be prevented in the design
in order to avoid the possible story mechanism developed in the first story. Thus, the complete
procedures for the design of the 1F column are given as follows: (1) considering the combined
∗ is greater
effects of axial and flexural forces, check that the plastic moment at the bottom end Mpb
than the Mp-req given in Equation (3); (2) based on the superposition method shown in Figure 5,
∗ /+(1+1/)×(
ch1 h 1 /12);
the flexural demand at the top end Mt can be estimated from Mpb 2
(3) check to ensure that the column top end remains elastic under the bending moment Mt and
the axial force, which can be determined from the force diagram shown in Figure 4.
3. RESTRAINED STEEL PLATE SHEAR WALLS (R-SPSWs)
3.1. General introduction

As stated earlier, the interference of the restrainers on the stretching of steel panel can be prevented
by using enlarged bolt holes on restrainers. Thus, in an R-SPSW (Figures 1(d), 1(e) and 1(f)),
the pin-end horizontal restrainers can effectively restrain two kinds of deformations: (1) out-of-
plane deformation of the buckled infill plates and (2) inward flexural deformation of the boundary
columns induced from the panel forces. When an infill panel is constructed with the proposed
restrainers, the out-of-plane buckling deformation can be significantly reduced as the buckling
length is effectively reduced [9]. This kind of restraining effect is similar to the two-way stiffeners
in a stiffened SPSW (S-SPSW). Some differences between the R-SPSW and the S-SPSW include:
(1) In an S-SPSW, the infill panels are stiffened sufficiently such that the panels can develop
shear yield strength before the shear buckling. In the R-SPSW, the panels are not required
to be stiffened to become ‘compact’ plate [14]. The R-SPSW panels are permitted to buckle
in shear at a very low lateral force like those in a typical thin SPSW. For all analytical
examples and the experimental specimens made for this study, the width-to-thickness
ratios
of the panels are much greater than the lower limit for the slender plate 1.37 kv E/Fyp [14].
(2) The main purpose of utilizing restrainers is to reduce the boundary element demands due to
the tension field action. In a typical SPSW, the horizontal component of tension field forces
will pull the boundary columns inward and induce significant boundary element demands
such as beam axial forces and column flexural forces. In an R-SPSW, the restrainers will
help the boundary elements to resist the horizontal tension field forces. Thus, the stated
DOI: 10.1002/eqe
boundary element demands could be reduced. In a typical S-SPSW, the stiffeners are not
designed to reduce the force demands on the boundary elements.
The panel aspect ratio is an important parameter in the design of any SPSW. The inclined folds
developed on the buckled panel must be discontinued where the restrainers are located. However,
the in-plane panel stretching in any direction cannot be stopped by the restrainers as the restrainers
are only sandwiched over the panel using enlarged bolt holes. Unlike the boundary beam elements,
the proposed restrainers do not anchor the inclined tension field forces. Thus, the effects of the
restrainers on the change of the panel aspect ratio should be insignificant.
3.2. Seismic design method for R-SPSW

3.2.1. Design consideration for restrainers. Restrainers must be designed to resist compressive
forces, preventing the columns from being pulled inward. Figure 7 shows that the effects of
restrainers on the columns are akin to lateral supports for the columns. The restrainers are expected
to remain elastic even if the R-SPSW has developed a plastic mechanism and the plastic hinges have
strain-hardened. With the transverse panel forces applied on the fix-ended columns, the compressive
forces on the restrainers can be estimated from the reaction forces at the lateral supports. If a
panel is constructed with NR pairs of uniformly distributed restrainers, each restrainer’s axial
force can be estimated as Hp ch h c /(NR +1). The Hp is the overstrength factor accounting for
the panel strain hardening effect and h c is the clear panel height. In addition, restrainers must
possess sufficient out-of-plane stiffness to prevent the panel from experiencing excessive buckling.
It is suggested that the out-of-plane moment of inertia of the restrainers meets the requirement
for transverse stiffeners, which are used to develop the web shear strength, as described in the
model steel building specifications [24]. The bolts passing through the plate (Figure 1(e)) and the
restrainers must be strong enough to resist the tension force developed during the buckling of
the steel panel. Past research [9] has confirmed that the total bolt tension force can be assumed to
be about 3% of the ultimate in-plane panel shear force.
3.2.2. Effects on beam design. When a typical SPSW is subjected to the lateral forces, the boundary
beams support the boundary columns to resist the horizontal panel forces, causing axial compressive
forces to develop at the beams. When the adjacent panels develop fully yielding tension field forces,
the J th floor beam axial force PbJ due to the horizontal panel forces can be determined from
chJ −1 h J −1 chJ h J
PbJ = + (4)
2 2
Figure 7. Force diagrams for calculating the restrainers’ and the boundary beams’ axial forces
caused by the horizontal panel forces.
DOI: 10.1002/eqe
Figure 8. Schematics of superposition method for computing the bending moments on

the compressed 1F column of R-SPSWs.
As shown in Figure 7, if each panel in an R-SPSW is constructed with NR pairs of uniformly

distributed restrainers, the beam axial force PbJ will be reduced to
chJ −1 h J −1 chJ h J
PbJ = + (5)
2(NR +1) 2(NR +1)
3.2.3. Effects on 1F column design. Restrainers will reduce the 1F column bending moment
induced from panel forces. Figure 8 shows the superposition of two flexural demands for the 1F
column in an R-SPSW. When the 1F panel is constructed with NR pairs of uniform distributed
restrainers, the compressed column bending moment distribution M(y) below the lowest restrainer
(when yh 1 /(NR +1)) can be approximated from the following equation:
2
ch1 h1 y h1 1+
M(y) = −6y +6
2
− −Mf 1− y (6)
12 NR +1 NR +1 h 1
Based on the same logic stated for the analysis of 1F column flexural demand in a typical SPSW, the
required flexural strength, Mp-req , of 1F column with NR pairs of uniformed distributed restrainers
can be estimated from
ch1 h 21 ch1 h 21
Mp-req = − (7)
2(+1)(NR +1) 12(NR +1)2
The effects of the number of restrainers in reducing the 1F column flexural demand are shown in
Figure 9. The reduction ratio is the ratio of the Mp-req (Equation (7)) for an R-SPSW with NR
restrainers to that for a typical SPSW (Equation (3)). If the tension field angle of the panels does
not change significantly after installing restrainers, it can be found that one pair of restrainers can
reduce the 1F column flexural demand Mp-req to about 60%. Attaching two pairs of restrainers
reduces the Mp-req demand to about 40%.
3.2.4. Inclination angle of the tension field action. Past researches [5, 13] have confirmed that
the estimation of the tension filed angle proposed by Timler and Kulak [2] can be satisfactorily
used for strip models in predicting the cyclic responses of SPSWs. In this paper, similar proce-
dures have been adopted to develop the method of computing the tension filed angle for the
DOI: 10.1002/eqe
Figure 9. Relationships between the reduction of the flexural demand versus the number of restrainers.
Figure 10. (a) Schematic of an R-SPSW; (b) the free-body diagram of a portion of the R-SPSW;
(c) the force diagram for estimating the axial forces on the restrainers; and (d) column bending
moments induced from the horizontal panel forces.
R-SPSWs. Figure 10(a) shows a single-story and single-bay R-SPSW with two uniformly
distributed restrainers subjected to the shear force V . Figure 10(b) shows the free-body diagram of
a portion of the R-SPSW. The whole panel is represented using a series of parallel trusses (strips)
with an inclination angle from the vertical. It is assumed that all the inclined strips develop the
same axial stress p . Considering that the restrainers are acting like pin-supports (Figure 10(c)),
the axial force developed in each restrainer HR is approximately equal to ×h/(NR +1). ,
horizontally distributed panel forces acting on the column, can be determined from tp p sin2 .
NR is the number of the uniformly distributed restrainers. From the equilibrium of the horizontal
forces, the beam axial force H is equal to V + F sin − R sin − NR HR . Then, as shown in
Figure 10(b), R = tp p L cos and F = tp p h sin are the resultant panel forces acting across the
beam length L and the column height h, respectively.
Similar to the previous approach [2], the boundary beams’ flexural effects are not included in
computing the inclined angle in this study. However, the following effects have to be modified or
included for the R-SPSW as discussed below: (1) the strain energy due to the reduced beam axial
force H stated above; (2) the change of column flexural strain energy considering the moment
distribution illustrated in Figure 10(d) and (3) the additional strain energy due to restrainers’
compressive force HR . Thus, the equation for the tension field angle in an R-SPSW with NR
DOI: 10.1002/eqe
uniformly distributed restrainers can be computed from the following equation:

tp L
1+
2Ac
tan4 = (8)
tp h NR 1 tp h 4 1
1+ + +
(NR +1)2 AR Ab (NR +1)4 360Ic L
where Ab and Ac is the average cross-sectional area of the boundary beams and that of the boundary
columns, respectively. The Ic is the average moment of inertia of the boundary columns.
4. ANALYTICAL EXAMPLES
4.1. Example designs and FE analysis

The ABAQUS FE analyses of four different designs of a single-story single-bay SPSWs were
compared with strip and EB models to investigate the effectiveness of the restrainers and the
proposed capacity design method. As shown in Figure 11(a), the overall dimension of the example
designs is 2 m wide by 3 m tall. This aspect ratio of 0.67 is somewhat similar to that chosen for
the narrow SPSW specimens adopted in this research. The boundary elements in the four designs
are identical. The H400×200×9×12 beam and two H300×300×12×20 columns are employed.
The differences in the four designs are the infill panel thickness and the use of restrainers. The
four SPSW designs are designated as T2, R2, T3 and R3, where the ‘T’ or ‘R’ represents ‘typical
SPSW’ or ‘R-SPSW’, respectively. The number of ‘2’ or ‘3’ implies that the panel thickness is 2.0
or 3.4 mm, respectively. Designs R2 and R3 are constructed with two pairs of uniformly distributed
restrainers made from steel hollow structural sections (HSSs). Two different HSSs selected for R2
and R3 are listed in Table I. In order to assure that the panel zone remains elastic when the structure
undergoes large lateral displacements, a 10 mm thick doubler plate is used at each of the beam-
to-column joint in the four designs. The steel material for the panels and the boundary elements
are all assumed as elastic–perfect-plastic (EPP) model with the yield stresses f y = 350 MPa.
Figure 11. Schematics of (a) the R-SPSW design example and the corresponding (b) strip
model; (c) EB model; and (d) FE Model.
DOI: 10.1002/eqe
Table I. Key parameters of the four SPSW designs and the design check for the columns.
Column design check
Design Restrainer section Plate thickness (mm) (◦ ) Mp-req (kN mm) Pd (kN) Mp∗ (kN mm)
T2 — 2.0 39.3 452 810 1876 501 938

R2 Tube 75×75×2.5 2.0 41.0 211 126 1867 503 198
T3 — 3.4 37.4 703 960 2804 365 480
R3 Tube 75×75×3.5 3.4 40.8 357 134 2786 368 251
The width-to-thickness ratio of the infill panel ranges from 260 (Design R3) to 850 (Design T2).
It is much greater than the lower-bound limiting width-to-thickness ratio of 1.37 kv E/Fyp = 73.2
for the slender plates. It suggests that the simple strip model is suitable for the analysis of these
thin-panelled SPSWs [14]. Equation (8) and the code prescribed method for the typical SPSW
are used to compute the tension field angles, , of the R-SPSW and typical SPSW, respectively,
which are also listed in Table I. In the column design checks (Table I), the flexural demand Mp-req
is determined from Equations (3) and (7) for typical and R-SPSW, respectively. The Pd is the
axial force demand at the compressed column bottom end. The reduced moment capacity Mp∗ is
determined from the following equation [24]:
Mp∗ = 1.18Mp (1− Pd /Py )Mp (9)
The clear panel width L c f and the clear height h c (Figure 11(a)) are incorporated in computing
the Mp-req and Pd . From Table I, it is clear that only T3 does not comply with the proposed
capacity design requirement (Mp-req >Mp∗ ). For each example design, a nonlinear FE model was
constructed using the ABAQUS program [25]. In addition, a strip model and an EB model have
been made using the PISA3D/GISA3D programs [26]. Pushover analyses for a story drift up to
2.5% radians have been carried out for each model.
4.1.1. Strip model and EB model. Using the tension field angles listed in Table I for the strip
models, 12 or 13 equally spaced inclined truss elements were used to represent each infill panel
in the typical SPSW designs (T2 and T3) or the R-SPSW designs (R2 and R3) (Figure 11(b)),
respectively. The cross-sectional area of each truss element As was calculated from tp ×(L c f cos +
h c sin )/n s , where n s is the number of the strips per panel. The boundary elements were represented
using the plastic hinge model for the beam–column element. The axial–flexural interaction surface
was adopted in each boundary element to model the combined effects of the axial force and
the bending moment. For the R-SPSWs (Figure 11(b)), each pair of the HSS restrainers was
represented using one single truss element with a cross-sectional area equal to that of the sum of
the two tubes.
In the EB models, each infill panel was replaced by using only one single diagonal truss element
(as shown in Figure 11(c)). The cross-sectional area of the EB, AEB , was determined from the
following equation [1]:
tp L sin2 2
AEB = (10)
2 sin sin 2
where = tan−1 (L/ h) is the angle between the brace and the column. Assuming that the total
shear strength of all the strip equals to the EB, the relationship between the yield stress of the EB
DOI: 10.1002/eqe
f yEB and the yield stress of the steel plate f yp is f yEB = (sin 2/ sin 2)×(L c f /L)× f yp . In the strip
and EB models, the 100% rigid-end zone offsets were adopted to represent the beam-to-column
joints where 10 mm thick column web doubler plates have been applied.
4.1.2. Nonlinear FE model. As shown in Figure 11(d), the infill panels, boundary elements, HSS
restrainers and 10 mm thick half-column-depth stiffeners near the restrainer-to-column joints were
all included in the FE model. These models were constructed using the 4-node, quadrilateral,
stress/displacement shell elements with the reduced integration and a large-strain formulation
(ABAQUS S4R Element). The columns and the bottom edge of the infill panels were fully fixed
at their bases. For the R-SPSW models, the out-of-plane displacements of the panel regions sand-
wiched between the two tube restrainers were restrained (Figure 11(d)) considering the presence
of the restrainers. In order to analyze that a thin plate buckles in shear under a very low lateral
load, initial imperfections were specified on the infill panels in the FE models. Before the pushover
analysis, an ABAQUS buckle analysis was performed to obtain the first 20 infill panel buckling
modes. The initial imperfection was determined from the superposition of the first 8 or first 18
mode shapes for typical or R-SPSW models, respectively. The peak amplitude of each mode’s
nodal displacements was set as tp /200, where tp is the panel thickness.
4.2. Analytical results

4.2.1. Lateral force versus displacement responses. Figures 12(a)–(d) present the total shear versus
displacement relationships for all analytical models and provide the panel shear forces obtained
from the FE models. The FE shear force resisted by the panel was obtained from the summation
of the nodal forces at the panel base. In addition, the fully yielding panel strengths, Vyp , calculated
from 0.5 f yp tp L c f sin 2 are also marked in Figure 12. Figures 13(a)–(d) show the Von Mises
stress contours at the 2.5% radians story drift. In theses contours, the white regions represent the
plastic zones. Figures 13(a)–(d) also show the extent of the plastic deformations in all strips and
the boundary elements at the end of the pushover. The size of the plastic hinge rotation and the
strain in each strip can be visually estimated by comparing the size of the circles or length of
the widened lines with respect to the references given at the right bottom of Figure 13(d). From
Figures 12(a)–(d), it can be found that except for Design T3, the total shear versus story drift
responses predicted by three different kinds of analytical models (strip, EB and FE) are rather
similar. In addition, the infill panels in all the design examples, except T3, are cable of developing
the fully yielding panel strength Vyp marked in each of the corresponding figure. The comparison
of the strips’ deformation marks in Figures 13(c) with 13(a) (or 13(d) with 13(b)) shows that the
restrainers can effectively help in promoting a more uniform tension field action in the steel panels.
As shown in Figure 13(a), (c) and (d) for T2, R2 and R3, almost the entire infill panel in each
figure has yielded and most of the plastic zones on the columns concentrate at the bottom ends at
the 2.5% radians story drift.
For Design T3, the total and the panel shears versus displacement responses obtained from the
strip model is very similar to that from the FE model (Figure 12(c)). However, the infill panel
cannot develop its fully yielding strength Vyp . As shown in Figure 13(b)-left for Design T3, the FE
stress contours (darkened zones) and the strips’ axial plastic deformations indicate that the bottom
right corner region of the infill panel near the compressed column bottom end could not be fully
stretched to yield at the end of the pushover. Furthermore, from the stress contours (Figure 13(b)-
left), it can be found that the plastic zones in the compressed column in Design T3 have spread
DOI: 10.1002/eqe
Figure 12. Force versus displacement relationships computed from three different analytical models for
SPSW designs: (a) T2; (b) R2; (c) T3; and (d) R3.
Figure 13. FE stress contours, the EB (for Design T3 only) and the strip model plastic hinge
diagrams for the four SPSW designs at a story drift of 2.5% radians: (a) Model T2; (b) Model
T3; (c) Model R2; and (d) Model R3.
over the bottom half of the column. The range of this kind of plastic zones (white regions) is much
greater than those observed in the other three designs. It can be found that these plastic zones in the
FE stress contours spread from the column bottom over about 40% of the entire column height. The
widest plastic zone (Figure 13(b)-left) appears to be located at 500 mm measured from the column
bottom end. In the strip model shown in Figure 13(b)-middle, a flexural plastic hinge (shown by
a circle) also forms above the compressed column bottom end. The location of this circle is very
close to the widest plastic zone stated above for the FE stress contours for Design T3. Judging from
DOI: 10.1002/eqe
the location of the plastic hinges, the extent and distribution of the strip deformations, and the total
shear versus lateral displacements of the structure shown in Figures 12(a)–12(d) and 13(a)–13(d),
the strip model can satisfactorily predict the FE nonlinear responses of all four designs regardless
of whether restrainers or weak columns exist or not. From Figures 13(b)-left and 13(b)-middle,
it is confirmed that if the column strength is insufficient, the plastic hinge would form near the
mid-height of the compression column. These results agree with the design check listed in Table I.
In addition, it can be found that if a plastic hinge forms near, or plastic zones too far spread out
into, the mid-height of the compressed column, the yielding tension field action cannot be fully
developed by the weak boundary columns as evidenced in the Design T3.
As shown in Figure 12(c), the ultimate strength of Design T3 predicted by the EB model is
somewhat greater than those computed from the strip and FE models. From the same figure, it can
be found that the difference in the ultimate strength between the EB and FE models is almost equal
to the difference between the yielding panel strength Vyp and the peak FE panel strength. This
confirms that the yielding of the EB represents the full yielding of the entire panel, and cannot
realistically reflect the less-stretched, right, bottom corner regions due to the flexural yielding
of the weak column near the mid-height. On the other hand, as shown for Designs T2, R2 and
R3, the nonlinear responses computed from using the EB models agree with the FE responses
when the boundary elements meet the capacity design requirements suggested earlier in this study.
On the contrary, in Design R3, the plastic hinge is formed at the compressed column bottom
end (Figures 13(d)-left and 13(d)-middle). The infill panel could develop its fully yielding strength
Vyp (Figure 12(d)). Comparing Design R3 with T3, it is confirmed that the restrainers could reduce
the panel force effects on the boundary columns, and help the weak boundary columns to anchor
and develop the full yielding of the tension field action. Thus, the lateral load-carrying capacity
of Design R3 (Figure 12(d)) is greater than that of T3 (Figure 12(c)). However, it is worthy to
note that in Figures 12(a) and (b) the ultimate strengths of Designs R2 and T2 are similar. This
is because the same boundary elements themselves are sufficient to anchor the thin steel panel
(2.0 mm) used in Design T2 to develop the panel strength Vyp (Figures 12(a)) and stretch the entire
panel as expected (Figure 13(a)-right).
Past research [21] suggested that the strip model might not satisfactorily predict the narrow
SPSW with a low aspect ratio. However, it appears that the strip model applied in this study could
satisfactorily predict the FE responses of the four narrow SPSW designs. It should be noted that rela-
tively strong and stiff boundary elements have been adopted in these example designs. In these four
designs, the ultimate shear resisted by the boundary columns is about 50% of the total shear force.
This ratio appears higher than most of the experimental specimens observed in the past researches.
4.2.2. Tension field angle. Using the FE model, the accuracy of Equation (8) in estimating the
tension field angle is discussed herein. In fact, the tension field angle changes as lateral deformations
increased during the pushover analysis. In addition, as observed from the analysis stated above,
the severeness of the tension field actions varies from region to region in a deformed SPSW.
In the development of Equation (8), the elastic strain energy of the boundary elements has been
considered. Thus, the FE responses at a story drift of 0.3% radian are utilized. At this stage, the
panels have developed significant shear buckling in the FE models. It can be found that the average
orientations of the FEs’ in-plane principle stresses are 40.7, 38.0, 43.1 and 42.8◦ measured from a
vertical line for Designs T2, T3, R2 and R3, respectively. These are only slightly larger than those
listed in Table I. It appears that the code prescribed method is applicable to the narrow typical thin
SPSW with a frame aspect ratio of 0.67 adopted in this research. In addition, this research also
DOI: 10.1002/eqe
Figure 14. Restrainer axial force versus story drift relationships predicted from the
FE models and the strip models.
confirms the effectiveness of the proposed method (Equation (8)) in computing the tension field
angle for the R-SPSWs. Comparing the FE tension field angles between T2 and R2 (40.7 versus
43.1◦ ) or T3 and R3 (38.0 versus 42.8◦ ), it is evident that the presence of the restrainers increases
the tension field angles to approach 45◦ for narrow SPSWs. This tendency should help the narrow
SPSWs to deform more toward the shear mode.
4.2.3. Axial force demand in the restrainer. In the R-SPSWs (R2 and R3), the lower and upper
restrainers are denoted as Res1 and Res2, respectively (Figure 11(a)). Figure 14 shows the restrainer
axial force versus story drift relationships predicted from the FE and strip models. The restrainer
axial force in the FE model was computed from the total force at the mid-span cross-section of
each tube restrainer. The estimated restrainers’ axial force demands Pres calculated from ch ×
h c /(NR +1) are also plotted in Figure 14. It can be found in Figure 14 that the restrainer axial
forces in the strip models developed earlier than those of the FE models (Figure 14). It appears that,
at the small inter-story drift levels, in the FE models of the R-SPSWs the severe buckling of the
whole panel has not occurred yet. Therefore, the restrainers developed less axial forces in the FE
models than those in the strip models. On the other hand, the strip model considers that the panel
behavior is completely replaced by the tension field action throughout the pushover. The restrainers
in a strip model would develop the axial forces from the beginning of the pushover. Compared with
the restrainers in the FE model, the restrainers in a strip model would overestimate the restrainers’
axial forces when the inter-story drift is relatively small. As shown in Figure 14, the restrainers’
axial forces computed in a strip model could approach those computed from the restrainers in
the FE model as lateral deformations increase. It can be found in Figure 14 that the estimated
demand, Pres , conservatively predicts the restrainers’ ultimate axial force responses in the FE and
strip models. This could be because the Pres is computed by assuming that the restrainers act as
pin-supports of the column. The restrainers in an FE or strip model, however, can deform under
the axial loads. Nevertheless, the Pres can be conveniently applied for the design of the restrainers.
5. DESIGN AND FABRICATION OF SPECIMENS
5.1. Introduction of specimens

Four two-story narrow SPSW specimens 2.14 m wide by 6.5 m tall were constructed in the steel
fabricator’s shop. All four specimens had the same center-to-center line dimensions (Figure 15).
DOI: 10.1002/eqe
Figure 15. Schematics of the typical and restrained SPSW specimens.
Figure 16. Stress–strain relationship of the low-yield strength infill pane.
All the steel panels were 2.6 mm thick low-yield strength (LYS) steel with a measured yield
stress, f yp = 195 MPa. The lateral force was applied through the top of a column in each specimen
(Figure 15). Specimen designs were based on this loading method. No additional gravity load was
applied on the specimens. In order to verify the proposed column design methodologies stated
above, the key differences in the four specimens were included such as the size of the boundary
elements and the use of the restrainers. The detailed design results are shown in Figure 1 in the
companion the paper [23]. For each panel in the four specimens, the frame aspect ratio computed
from the center-to-center line dimensions of the member is 0.66. The panel aspect ratio computed
from the clear width and height ranged from 0.62 to 0.64. These aspect ratios are all less than the
lower limit (0.8) specified by the current seismic provisions [16]. Because the LYS steel plates
were used in the previous tests [9, 10], their strain hardening properties, as shown in Figure 16, are
known and have been carefully incorporated into the capacity design of the boundary elements.
All the boundary elements were made of A572 grade 50 steel. The infill panels were welded at
the edges to the boundary elements using 6 mm thick and 70 mm wide fishplate connection details
(Figure 17).
DOI: 10.1002/eqe
Figure 17. Fishplate connection details.
Figure 18. Predicted plastic hinge locations for the four specimens.
The design of the member sizes of all specimens will be discussed in detail later in this paper.
Specimens N and S are typical SPSWs. On the other hand, Specimens RS and CY are R-SPSWs
using restrainers. The boundary elements are classified and abbreviated into four categories as: the
top beam (TB), the middle beam (MB), the bottom beam (BB) and the column (C). All beam-to-
column connections are welded moment connections. Specimen N was normal, named from the
fact that the columns were complied with the proposed capacity design method. The boundary
elements in Specimen S were smaller than Specimen N. Specimen RS was identical to Specimen
S except that it adopted two pairs of restrainers in each story. The columns and restrainers in
Specimen CY were identical to those in Specimen RS. However, its top beam member size was
intentionally chosen larger than that of Specimen RS, to allow the development of column yielding
(CY) at the column top. Thus, the reduced section details were adopted at the column top for
Specimen CY. Figure 18 shows the predicted plastic hinge diagram for the four specimens. All
beam-to-column panel zone joints were designed to remain elastic under the most critical load
that could possibly develop in the boundary elements. All doubler plates (ranging in thickness
from 6 to 10 mm as shown in Figure 1 in Part 2) were one-sided and welded all around to the
adjacent column flanges and continuity plates. Lastly, all continuity plates were A572 Grade 50
steel having the same thickness as the framing beam flanges.
5.2. Key parameters for specimen design

For Specimens N and S with no restrainer, the design of the boundary elements considers the tension
field action angle, , prescribed in the current U.S. code [16]. The average tension field angles of
the first and second stories are 39 and 37◦ for Specimens N and S, respectively. The tension field
angle of R-SPSW Specimens RS and CY calculated using Equation (8) are both 41◦ , although
DOI: 10.1002/eqe
they were assumed 45◦ for the capacity designs. In fact, the force demands determined from 45◦
tension field angle only result in a slightly more conservative design. For research purposes and to
prevent the fabrication of overly conservative specimens, the material overstrength factor Ry = 1.1
was applied to the nominal yield stresses of all boundary elements as f y = Ry f yn = 385 MPa, in
computing the strengths of the boundary elements. f yn is the nominal yield stress of A572 grade
50 steel.
Based on the results of the pseudo dynamic tests of a full-scale two-story SPSW [11, 13],
the peak story drift could reach 2.5% radians when a properly designed SPSW is subjected to a
collapse prevention-level earthquake. Therefore, the deformation capacities of the specimens were
expected to be about 2.5% radians for severe seismic service. The design of the beam-to-column
connections considers the strain hardening effects of the infill panels and the boundary elements.
For the boundary elements, the strain hardening factor Hbc was taken as 1.15. When an infill panel
develops tension field action, the panel behaves like a series of inclined truss elements (strips).
Assuming that the entire panel deforms like a parallelogram and the tension field angle is 45◦ , all
the strips will develop a strain of 1.25% when the inter-story drift reaches 2.5% radians. Based on
the measured stress–strain relationship of the infill panels shown in Figure 16, the stress in each
strip would reach 1.17 times the yield stress. Therefore, for these four specimen designs, the strain
hardening factor for the infill panels was set as Hp = 1.17.
5.3. Designs for the boundary elements

5.3.1. Design of the top beams and the beam-to-column connections. For Specimen CY, the top
beam was expected to remain elastic throughout the whole test. It was designed based on the
beam axial force resulting from the panel forces. Its flexural demand was determined from the
plastic moment at the top end of 2F column. In addition, the strain hardening effects (1.17 and
1.15 factors) were considered when calculating these force demands. For Specimens N, RS and
S, the top beams are expected to develop plastic hinge at both ends. The reduced beam section
(RBS) was adopted for the beam-to-column connections in the top beam of the three specimens.
The purposes of utilizing the RBS details include: (1) enhancing the ductility of the specimens
and (2) resulting in an economical top beam design [10]. The strength of the top beams with the
RBS in Specimens N, RS and S satisfies the requirements of the design method proposed by Vian
and Bruneau [10]. The lengths, locations and depths of the beam flange reductions comply with
the recommendations described in FEMA 350 [27]. Past experimental results [11, 13] showed that
when the boundary beams were subjected to large vertical and horizontal panel forces, the web at
the RBS would yield in shear before the yielding of the RBS flanges. In our study, however, it was
hoped that the flexural plastic hinge would form at the RBS locations without the web yielding in
shear. Therefore, the webs of the top beams of Specimens N, RS and S were selected according
to their shear demands. The shear strength of the web, VpTB = 0.55 f y tw h ( f y = Ry f yn = 385 MPa),
was found to be greater than the shear force demand at the RBS location when the adjacent panel
fully yields. Figure 19 shows that the shear force demand at the RBS location of the top beam can
be determined from
V(TCR) = [M(TTR) + M(TCR) ]/L ∗T +bvT L ∗T /2 (11)
where M and V are the bending moment and shear force demands, respectively. In Equation (11),
the subscripts in the parentheses denote the location; the first subscript ‘T’ denotes that the member
force is in the top beam. The second subscript, ‘T’ or ‘C’ denotes whether the member forces
DOI: 10.1002/eqe
Figure 19. Force diagram for computing the shear force at the RBS of the top
beam when the top panel fully yields.
Figure 20. (a) Member force distributions; (b) force diagrams for computing the flexural and
shear force demands; (c) top beam-to-column connection details; and (d) illustration of the design
concept of the beam-to-column connection of a top beam with RBS details when the top beam is
subjected to the strain-hardened panel forces.
are at the end that is connected to the tensioned column or compressed column, respectively. The
third subscript ‘R’ denotes the member forces located at the RBS location. The L ∗T (Figure 19)
is the distance between the most reduced sections of the two RBS in the top beam. The bvT
is the vertical component of the uniformly distributed yield panel force acting on the top beam.
The moments at the RBS locations, M(TTR) and M(TCR) , were computed from the reduced plastic
moment capacity Mp∗ at the most reduced cross-section. In these four specimen designs, Mp∗ was
determined using Equation (9) considering the beam axial force at the RBS induced by the fully
yielding panel forces.
The beam-end sections (at the column faces) are expected to remain elastic when the yield panel
and the top beam RBS have strain-hardened. Figure 20(a) shows the member force distributions at
the top beam of an SPSW at this level. The beam end that is connected to the compressed column
has to resist the combined bending moment MU and shear forceVU . The subscript ‘U’ represents
the ultimate member force demands considering the strain hardening effects. The moments at
DOI: 10.1002/eqe
Table II. Dimensions of the shear tabs used in the beams of the four specimens.
Specimen N Specimen RS Specimen S Specimen CY
Shear tab for TB (d ×b ×t mm) 276×120×18 278×120×15 278×120×15 308×120×10
Shear tab for MB (d ×b ×t mm) 276×120×10 278×120×8 278×120×8 278×120×8
t is the thickness of the shear tabs.
d and b are illustrated in Figure 20(c).
the RBS locations, MU(TTR) and MU(TCR) , were computed from Hbc × Mp∗ . The Mp∗ herein was
computed by considering the axial force at the RBS induced by strain hardened panel forces.
Figure 20(b) shows the force diagrams for computing the force demands at the beam ends. Based
on the force and moment equilibriums on the beam segments, the following equations can be used
to design the connections:
VU(TCR) = [MU(TTR) + MU(TCR) ]/L ∗T +(Hp bvT × L ∗T )/2

VU(TCF) = VU(TCR) +Hp bvT × ST (12)
MU(TCF) = MU(TCR) + VU(TCF) × ST −(Hp bvT × ST2 )/2
where ST is the distance from the center of the RBS to the nearby column face. The third subscript
‘F’ in the parentheses denotes the member force MU or VU located at the face of the column.
Compared with the more linearly distributed beam moment in a moment-resisting frame subjected
to lateral forces, the quadratic-distributed beam moment on the top beam of an SPSW suggests
that the typical RBS detail would not prevent the beam-to-column connection from experiencing
fracture. Thus, for each specimen, the shear tab was fillet welded at three edges to the beam web
(Figure 20(c)) and the beam web was welded to the column flange. The beam-end gross cross-
sectional area, consisting of the shear tab and the original beam section, was checked to remain
elastic under the combined axial force, shear force and flexural demands. Figure 20(d) illustrates
the design concept for the connections. Table II lists the sizes of the shear tabs used to detail the
beam-to-column connections at the top beams of Specimens N, S and RS.
5.3.2. Design of the middle beams, the bottom beams and the columns. The RBS details were also
used to detail the beam-to-column connections of the middle beams. Since the plate thicknesses
of the infill panels in both stories were the same, the net effects of the vertical component of the
panel forces acting on the middle beams were negligible. However, since the corner regions of
the top or bottom panel may not yield at the same time, the bending moment distribution in the
middle beam should not be entirely linear when subjected to lateral forces. Therefore, the fillet
welds between the shear tabs and the beam web (Figure 20(c)) were still used for the middle
beam to increase the strength and deformation capacities of the connections. The bottom beams
were fabricated to anchor the tension field action of the 1F infill panels. There is little flexural
demand from the frame sway action for the bottom beam. The whole bottom beams were required
to remain elastic throughout the tests. Thus, no RBS detail was required for the bottom beams.
The bottom beams were designed to resist the flexural and axial force demands due to the strain-
hardened panel forces. For Specimens N, RS and CY, the boundary columns comply with the
capacity design requirements proposed in this research. Only for specimen S, the column strength
was insufficient. Therefore, it was predicted that a plastic hinge would form above the bottom end
DOI: 10.1002/eqe
Figure 21. (a) Locations of bolts for restrainers; (b) restrainers sandwiched over the panel from both
sides of the panel using through bolts; (c) restrainer-to-column connection details; (d) slot holes on the
restrainers; and (e) restrainer ends and connection plates.
of the compressed column as shown in Figure 18. The detailed design check for the columns is
presented in the part 2 of the paper.
5.3.3. Design of restrainers. As stated, Specimens RS and CY were constructed with two pairs
of restrainers. The design of the restrainer was based on the axial force demand of the restrainer
estimated from Hp ch h c /3. As shown in Figures 21(a) and 21(b), pairs of 75×75×3.2 mm
hollow steel tubes were sandwiched over the infill panels from both sides of the panels using
through bolts. Also shown in Figure 21(a), the locations of these out-of-plane bolts were aligned
along the 45◦ tension field direction. This would result in the least loss of the tension field actions.
The holes on the restrainers were slotted (Figure 21(d)), whereas the holes on the infill panels
were circular. The 10 mm thick washer plates were used to prevent the bolts from crushing the thin
flanges of the tubes when the bolts were tightened. Figure 21(c) illustrates a double shear joint
adopted to connect the restrainers to the column flange. A 5 mm thick-end plate was welded to
the ends of each tube. Next, a 20 mm thick connection plate was welded to each end plate to join
the two 10 mm thick plates that were welded to the column flange. The corners of the connection
plates were slightly cut out to avoid the fishplates (Figure 21(e)) and to ensure that the full length
of the restrainers was in continuous contact with the infill panels. A pair of half-column-depth
stiffeners were welded to the column flange and web (Figure 21(c)) at the opposite side of the
restrainer-to-column flange joint.
6. CONCLUSIONS
Based on the stated studies, conclusions can be drawn as follows:

• It is illustrated that the strip model can satisfactorily predict the FE nonlinear local and overall
responses of all four SPSW Designs using thin steel panels regardless of whether of restrainers
or weak columns exist.
DOI: 10.1002/eqe
• When a multi-story SPSW is subjected to the lateral forces, the compressed 1F column is the
critical column. If the column strength is insufficient, the plastic hinge on the compressed 1F
column will form above the bottom end. This kind of inelastic column deformation would
cause the 1F infill panel to fail in developing the full yielding tension field action. This
research proposes a capacity design method for the 1F columns in the SPSW and R-SPSW
to ensure that the plastic hinge forms at the compressed 1F column bottom end.
• The nonlinear responses computed from using the EB models agree well with the FE responses
for Designs T2, R2 and R3, which have the boundary elements meeting the proposed capacity
design requirement.
• From the analysis of Designs T2, R2 and R3 having the boundary elements meeting the
proposed capacity design requirements, the nonlinear responses computed from using the EB
models agree well with the FE responses.
• It is confirmed from the analysis of four example SPSW designs that the proposed sliding
restrainers can effectively help in promoting a more uniform tension field action in the steel
panels. In addition, the restrainers can reduce the force demands on the boundary column
thereby allowing a more cost-effective use of column members.
• The effects of the restrainers on the change of tension field angle have been found to be
insignificant. However, the tension field angle for the R-SPSW can be conveniently computed
by incorporating the strain energies into boundary beams and columns and assuming that the
restrainers act like pin-supports for the column.
• It is confirmed from the FE analyses that the presence of the restrainers has increased the
tension field angles to approach 45◦ for narrow SPSWs. This tendency should have helped
the narrow SPSWs to deform more toward the shear mode.
• This study focuses on the SPSWs and R-SPSWs using thin infill panels. For the design and
analysis of SPSW using thick panels, further study is required.
REFERENCES
1. Thorburn LJ, Kulak GL, Montgomery CJ. Analysis of steel plate shear walls. Structural Engineering Report
No. 107, Department of Civil Engineering, University of Alberta, Edmonton, AB, Canada, 1983.
2. Timler PA, Kulak GL. Experimental study of steel plate shear walls. Structural Engineering Report No. 114,
Department of Civil Engineering, University of Alberta, Edmonton, AB, Canada, 1983.
3. Sabouri-Ghomi S, Ventura CE, Kharrazi MHK. Shear analysis and design of ductile steel plate walls. Journal
of Structural Engineering (ASCE) 2005; 131(6):878–889.
4. Caccese V, Elgaaly M, Chen R. Experimental study of thin steel-plate shear walls under cyclic load. Journal of
Structural Engineering (ASCE) 1993; 119(2):573–587.
5. Driver RG, Kulak GL, Kennedy DJL, Elwi AE. Cyclic tests of four-story steel plate shear wall. Journal of
Structural Engineering (ASCE) 1998; 124(2):112–120.
6. Lubell AS, Prion HGL, Ventura CE, Rezai M. Unstiffened steel plate shear wall performance under cyclic
loading. Journal of Structural Engineering (ASCE) 2000; 126(4):453–460.
7. Astaneh-Asl A, Zhao Q. Cyclic tests of steel plate shear walls. Report No. UCB/CE-Steel-01/01, Department
of Civil and Environmental Engineering, University of California, Berkeley, 2001.
8. Berman JW, Bruneau M. Experimental investigation of light-gauge steel plate shear walls. Journal of Structural
Engineering (ASCE) 2005; 131(2):259–267.
9. Tsai KC, Lin YC. Seismic responses of steel shear wall constructed with restrainers. Technical Report NCREE-
04-023, National Center for Research on Earthquake Engineering, Taipei, Taiwan, 2004 (in Chinese).
10. Vian D, Bruneau M. Steel plate shear walls for seismic design and retrofit of building structures. Technical Report
MCEER-05-0010, Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York, 2005.
DOI: 10.1002/eqe
11. Tsai KC, Lin CH, Lin YC, Hsieh WD, Qu B. Substructural hybrid tests of a full scale 2-story steel plate shear
wall. Technical Report NCREE-06-017, National Center for Research on Earthquake Engineering, Taipei, Taiwan,
2006 (in Chinese).
12. Park HG, Kwack JH, Jeon SW, Kim WK, Choi IR. Frame steel plate wall behavior under cyclic lateral loading.
Journal of Structural Engineering (ASCE) 2007; 133(3):378–388.
13. Qu B, Bruneau M, Lin CH, Tsai KC. Testing of full scale two-story steel plate shear wall with reduced beam
sections connections and composite floors. Journal of Structural Engineering (ASCE) 2008; 134(3):364–373.
14. Astaneh-Asl A. Seismic behavior and design of steel plate shear walls. Steel Technical Information and Product
Services Report, Structural Steel Educational Council, Moraga, CA, 2001.
15. Canadian Standards Association (CSA). Limit States Design of Steel Structures. CSA: Willowdale, Ont., Canada,
2001.
16. American Institute of Steel Construction (AISC). Seismic Provisions for Structural Steel Buildings, Chicago, IL,
2005.
17. Berman JW, Bruneau M. Capacity design of vertical boundary elements in steel plate shear walls. Engineering
Journal (ASCE) 2008; 15(1):55–71.
18. Berman JW, Bruneau M. Plastic analysis and design of steel plate shear walls. Journal of Structural Engineering
(ASCE) 2003; 129(11):1448–1456.
19. Astaneh-Asl A. Seismic behavior and design of composite steel shear walls. Steel Technical Information and
Product Services Report, Structural Steel Educational Council, Moraga, CA, 2002.
20. Zhao Q, Asteaneh-Asl A. Cyclic behavior of traditional and innovative composite shear walls. Journal of Structural
Engineering (ASCE) 2004; 130(2):271–284.
21. Rezai M. Seismic Behaviour of steel plate shear walls by shake table testing. Ph.D., Department of Civil
Engineering, University of British, Columbia, Vancouver, Canada, 1999.
22. Rezai M, Ventura CE, Prion HGL. Numerical investigation of thin unstiffened steel plate shear walls. Proceedings
of the 12th World Conference on Earthquake Engineering, Aukland, New Zealand, 2000.
23. Li CH, Tsai KC, Lin CH, Chen PC. Cyclic tests of four two-story narrow steel plate shear walls. Part 2:
experimental results and design implications. Journal of Earthquake Engineering and Structural Dynamics 2009;
DOI: 10.1002/eqe.964.
24. American Institute of Steel Construction (AISC). Specification for Structural Steel Buildings, Chicago, IL, 2005.
25. Hibbit Karlsson and Sorenson, Inc. (HKS). ABAQUS/Standard 6.7 User’s Manual. Hibbit, Karlsson and Sorenson,
Inc.: Pawtucket, RI, 2007.
26. Lin BZ, Chuang MC, Tsai KC. Object-oriented development and application of a nonlinear structural analysis
framework. Advances in Engineering Software 2009; 40(1):66–82.
27. Federal Emergency Management Agency (FEMA). FEMA 350 NEHRP Recommended Seismic Design Criteria
for New Steel Moment-Frame Buildings, Building Seismic Safety Council for the Federal Emergency Management
Agency, Washington, DC, 2000.
DOI: 10.1002/eqe

Cyclic Tests of Four Two-Story Narrow Steel Plate Shear Walls - Part 1: Analytical Studies and Specimen Design

Uploaded by

Copyright:

Available Formats

Cyclic Tests of Four Two-Story Narrow Steel Plate Shear Walls - Part 1: Analytical Studies and Specimen Design

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cyclic Tests of Four Two-Story Narrow Steel Plate Shear Walls - Part 1: Analytical Studies and Specimen Design

Uploaded by

Copyright:

Available Formats

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2010; 39:775–799

Cyclic tests of four two-story narrow steel plate shear walls—

Keh-Chyuan Tsai1,2, ∗, †, ‡, § , Chao-Hsien Li2, ¶ , Chih-Han Lin2, ¶ , Ching-Yi Tsai1, 2,

Received 9 August 2008; Revised 7 September 2009; Accepted 11 September 2009

Copyright q 2009 John Wiley & Sons, Ltd.

2. CAPACITY DESIGN FOR FIRST-STORY COLUMN

2.1. Capacity design using method of superposition

Figure 3. Fully yielding panel forces acting on the boundary elements.

Figure 4. Force diagrams for calculating the column axial forces.

2.3. Flexural demand

Figure 6. Relationship between  and .

3. RESTRAINED STEEL PLATE SHEAR WALLS (R-SPSWs)

3.1. General introduction

3.2. Seismic design method for R-SPSW

Figure 8. Schematics of superposition method for computing the bending moments on

As shown in Figure 7, if each panel in an R-SPSW is constructed with NR pairs of uniformly

uniformly distributed restrainers can be computed from the following equation:

4.1. Example designs and FE analysis

T2 — 2.0 39.3 452 810 1876 501 938

4.2. Analytical results

5. DESIGN AND FABRICATION OF SPECIMENS

5.1. Introduction of specimens

Figure 15. Schematics of the typical and restrained SPSW specimens.

Figure 16. Stress–strain relationship of the low-yield strength infill pane.

Figure 17. Fishplate connection details.

5.2. Key parameters for specimen design

5.3. Designs for the boundary elements

V(TCR) = [M(TTR) + M(TCR) ]/L ∗T +bvT L ∗T /2 (11)

VU(TCR) = [MU(TTR) + MU(TCR) ]/L ∗T +(Hp bvT × L ∗T )/2

Based on the stated studies, conclusions can be drawn as follows:

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Figure 6. Relationship between and .