Engineering Structures 172 (2018) 346–357

Contents lists available at ScienceDirect

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

Quasi-static test of assembled steel shear panel dampers with optimized T

shapes
⁎
Baijie Zhu, Tao Wang , Lingxin Zhang
Key Laboratory of Earthquake Engineering and Engineering Vibration, Institute of Engineering Mechanics, CEA, Harbin 150080, China

A R T I C LE I N FO A B S T R A C T

Keywords: A metallic shear panel damper with the shape optimized by stress contour lines is proposed in this study to
Metallic shear panel damper mitigate stress concentration, reduce the effect of hot welds, and improve energy consumption efficiency. The
Stress contour line stress contour line is defined according to the J2 plasticity theory, and the optimized shape is obtained by
Shape optimization assuming that the points on the same contour line yield simultaneously. Different optimized shapes are devel-
Assembled damper
oped considering various loading conditions. The design formulas for the stiffness and the strength are then
Mechanical properties
derived, and further examined by nine dampers tested quasi-statically. Four are tested laterally under the ver-
tical axial load to simulate real boundary conditions. All dampers can be easily installed or replaced because of
the all-bolt connections. The test results demonstrate that the proposed metallic shear damper has a stable
energy-dissipation capacity and a better low-cycle fatigue capability than traditional shear dampers without
shape optimization. The stiffness and strength design values match the test values very well. The axial de-
formation in the specimen has been observed and identified due to the interaction among the cyclic axial-shear
coupled plasticity, the geometric nonlinearity, and the higher lateral buckling modes. Compared with the non-
optimized damper, the distribution of plastic deformation in the proposed dampers is more uniform, and the
stress concentration is reduced significantly.

1. Introduction
Traditional building structures consume seismic energy through the
plasticity of structural components, resulting in extensive damage
within these components. Although this can protect human life, the
difficulty of repairing these components can lead to enormous eco-
nomic losses [1]. Enhanced building performance and resilience are
required in modern societies, so human activity is not interrupted or
can be quickly restored after strong earthquakes. Installing passive
dampers to dissipate the seismic energy is an effective method, in which
the main gravity-bearing structural components remain almost elastic.
Various types of dampers have been developed and applied in the past
decades, such as displacement dependent buckling restrained braces
(BRB), metallic shear panel dampers, friction-based Pall dampers, and
velocity dependent viscous and viscoelastic dampers [2]. Of these,
metallic dampers are found to be most economic and have a stable and
large capacity to dissipate seismic energy.
Kelly et al. [3] put forward the steel shear panel damper in 1972,
which has since been widely used because of its simple configuration,
clear mechanical mechanism, and excellent low cycle fatigue features.
The damper consists of flanges, stiffeners, steel shear panels to consume
energy, and end plates connecting to the main building structure. All
components are welded together. In subsequent research [4–7] it was
observed that the steel shear panels buckled and the welds fractured
due to the significant stress concentration and the plastic strain accu-
mulation, particularly in the regions most affected by the weld heat. To
increase the energy dissipation, some studies have developed some
metallic panel dampers with special shapes. Chan et al. [8] conducted
experimental research for the Steel Slit Damper (SSD), which was fab-
ricated using a wide-flange steel with a number of slits on the web, and
a larger energy dissipation capacity can be achieved. Yong et al. [9]
studied the cyclic behavior of non-uniform steel strip dampers, such as
dumb bell-shaped strip and a tapered strip. Such configurations can
avoid both brittle failure and excessive force to the primary structure.
Fairs et al. [10] investigated a metallic energy dissipater called Perfo-
rated Yielding Shear Panel Device (PYSPD), which comprised of a thin
perforated diaphragm plate welded inside a short length square hollow
section. The perforations reduced the elastic stiffness and yield
strength, while increased the plastic areas to consume energy. Yong
et al. [11] improved the conventional slit damper by use of an hour-
glass-shaped strip. It has been found the plastic bending moment
reached at all cross sections simultaneously. Valizadeh et al. [12]

⁎
Corresponding author.
E-mail addresses: baijie_zhu@126.com (B. Zhu), wangtao@iem.ac.cn (T. Wang), lingxin_zh@126.com (L. Zhang).

https://doi.org/10.1016/j.engstruct.2018.06.004
Received 8 September 2017; Received in revised form 2 June 2018; Accepted 3 June 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
B. Zhu et al.
opened some holes with different dimensions on a steel plate shear wall
to improve the ductile and energy absorption capacity. Other studies
[13–18] have focused on the deployment and design of stiffeners to
solve the premature fracture problem, but the effectiveness is in general
limited.
To solve this problem, the panel shape is often optimized to enlarge
the plastification area and the stress concentration can then be miti-
gated. Zhang et al. [19] improved the low cycle fatigue performance of
steel shear panel dampers by weakening the thickness in the center area
of the energy dissipater. Liu et al. [20] used an arc shape at four corners
and a rectangular middle panel. Quasi-static tests demonstrated that the
ultimate bearing capacity was constant while the ultimate shear angle
increased by 44%. In subsequent studies [21], a parabolic shape was
used for the shear panel damper as a local optimization, and a nu-
merical study indicated that the maximum accumulative plastic strain
was reduced by 82.2% while the total energy consumption was only
reduced by 2.3%. A globally optimal solution for the panel shape was
obtained by Deng et al. [22,23] with a simulated annealing algorithm.
Both numerical and experimental results revealed that the maximum
accumulative plastic strain was reduced by 70%.
A new type of steel shear panel damper is proposed in this study,
with the shape optimized by stress contour lines considering different
loading patterns. All components of the damper are assembled using
high-strength bolts for convenient installation or replacement. Design
formulas for the stiffness and strength are then developed for the en-
gineering application. Finally, the effectiveness of the shape-optimized
damper and the correctness of the design formulas are demonstrated
through the testing of nine specimens. The influence of the vertical
axial load on the lateral bearing force is also examined.
2. Configuration of assembled steel shear panel dampers
This new type of assembled steel shear panel damper (ASSD) con-
sists of one energy dissipater, two L-shaped connectors, two buckling
restrainers, and two pieces of partition plates. These are assembled
using friction type high-strength bolts, as shown in Fig. 1(a), which
both reduce the adverse effect of welding heat and can be conveniently
installed or replaced after severe earthquakes. The energy dissipater
with effectiveness height h and width b, as shown in Fig. 1(b), deforms
laterally in the plane. The surface of the panel has two treatments: first,
the two shadow regions are treated by sand-blasting to increase the
friction force when clamped by the L-shaped connectors and the
buckling restrainers; and second the rest area is deformed to dissipate
the seismic energy and galvanized to protect it from rust. The de-
formation area can be designed with a special shape to maximize the
energy dissipation capacity, which is defined by the profile function
f (x ) . The clamped areas are designed elastically in any case. Note that
all corners are chamfered and rounded to avoid stress concentration.
The L-shaped connectors clamp the energy dissipater tightly through
the high-strength friction-type bolts, and are connected to the main
structural components securely also by another set of high-strength
bolts. The buckling restrainers serve as the confinement to avoid lateral
buckling of the energy dissipater. The well confinement to the dis-
sipater is realized by the enhanced restraining area that has sufficient
vertical and horizontal stiffeners, as shown in Fig. 1(c). The thickness of
the confinement plate is reduced by 2 mm to provide a gap to accom-
modate the higher mode out-of-plane buckling deformation of the dis-
sipater panel, so that a larger deformability in the plane can be
achieved. The confinement surface is attached by a layer of stainless
steel shim to reduce the friction force once the dissipater is in contact
with the confinement surface. The pair of buckling restrainers are
connected through high-strength bolts but separated by the two parti-
tion plates, as shown in Fig. 1(a). Note that the buckling restrainers also
serve as the connector to the main structural components using a si-
milar mechanism of the L-shaped connectors. This mechanism prevents
the buckling restrainer from sliding during the lateral loading and
347
Engineering Structures 172 (2018) 346–357
deformation.
3. Shape optimization design method based on J2 theory
A steel shear panel damper is often installed at positions with large
deformations during a severe earthquake, such as two adjacent stories
and the coupling beams. The damper commonly deforms in the plane
and the entire area is supposed to yield simultaneously under the pure
shear force. However, the moment associated with the shear force often
results in an early yielding at the four corners where the largest
equivalent stress exists. The J2 theory to define the yield criterion can
thus be used to develop the yield line of the panel under the shear force.
A more complex load condition of the damper bearing both axial and
shear forces can also be considered within the J2 theory framework.
3.1. Derivation of stress contour line under shear force
3.1.1. Type 1: Optimized by von Mises yield criterion (J2 theory)
The thickness of the panel is uniform and defined as t . The given
height is denoted as h . The panel is symmetrical on the vertical and
horizontal axes. The xoy coordinate system is defined with the origin at
the center of the panel, as shown in Fig. 2. The height is along the x
axis. The left side of the damper is assumed to be fixed as the constraint
boundary, while the right side moves vertically as the loading direction.
The rotation about the z axis (out of plane) at the right side is pro-
hibited, while the deformation along the x direction is free. Considering
the yielding mechanism, a small deformation is assumed. At any section
perpendicular to the x axis, the shear stress is assumed to be uniformly
distributed along the cross-section, while the normal stress has a linear
distribution with the maximum at the both ends and zero at the middle
point.
Given the designed yield force as V , the stress state of a micro unit
in the plane at any point (x , y ) is as shown in Fig. 2, where σ and τ
represent the normal and shear stress respectively, and θp is the direc-
tion angle of the principal plane where the shear stress is zero. Ac-
cording to the equilibrium of the micro unit, the direction angle can be
calculated as Eq. (1), and the two principal stresses in the plane are
expressed as Eqs. (2) and (3), respectively:
2τxy
tan2θp = −
σx −σy (1)
σx + σy 1 2
σ1 = + (σx −σy )2 + 4τxy
2 2 (2)
σx + σy 1 2
σ2 = − (σx −σy )2 + 4τxy
2 2 (3)
By applying the J2 yield criterion [24] and introducing the stress
condition that σy = 0 , the equivalent von Mises stress, σe , is expressed as
Eq. (4):
1
σe = (σ1−σ2)2 + σ12 + σ22 = σx2 + 3τxy
2
2 (4)
When the damper sustains the shear force V , the bending moment of
a section M at x can be given as M = Vx . Suppose the shape function of
the panel is f1 (x ) , the maximum normal stress on the section at x is
calculated by Eq. (5):
M|f1 (x )| 3Vx
σx = =
Iz 2tf12 (x ) (5)
2 3
where the moment of inertia about the z axis is Iz = 3
tf (x ) .
The shear stress τxy introduced by the shear force V is actually dis-
tributed parabolically along the cross-section, but simplified as a uni-
form distribution:
V
τxy =
2t|f1 (x )| (6)
B. Zhu et al. Engineering Structures 172 (2018) 346–357

(a) Assembly (b) Energy dissipater

(c) Buckling restrainer

Fig. 1. Configuration of ASSD.

Substituting Eqs. (5) and (6) into Eq. (4), the von Mises stress can be 3.1.2. Type 2: Optimized by dominative stress
calculated by Eq. (7): At the mid-span of the damper, the bending moment is zero and
there are no normal stresses where the shear stress dominates the sec-
V 9x 2 3 ⎞ tional yield performance. At both ends of the damper the bending
σe = σx2 + 3τxy
2
= + ⎛⎜ ⎟
2t f14 (x ) ⎝ f12 (x ) ⎠ (7)
moment is at the maximum. The section may start yielding from the
edges where the shear stress is zero. Therefore, the yielding is con-
Suppose that all the points on the shape function f1 (x ) yield si- trolled by the normal stress. If only shear stress occurs in each section,
multaneously, that is, let σe be the material yield strength f y , then the the shape of the panel can be determined by a constant line P (x ) .
shape function can be derived as Eq. (8), which defines the yield con- Similarly, if there is only normal stress in each section, the shape can be
tour line: defined by a function Q (x ) , as shown in Fig. 3. The panel shape can then
be determined by the envelop curve of P (x ) and Q (x ) .
V f y tx If the shear stress dominates the yielding behavior, the profile shape
f1 (x ) = ± 3+ 9 + 144( )2
2 2 fy t V (8) function of the panel P (x ) can be governed by Eq. (9):

Fig. 2. Stress state of energy dissipater in plane.

348
B. Zhu et al. Engineering Structures 172 (2018) 346–357

contour line:
N
σ0 =
2|R (x )|t (14)
The total normal stress σ ′ of a micro unit in the cross-section is given
as Eq. (15):
N M|R (x )|
σ ′ = σ0 ± σx = ±
2|R (x )|t Iz (15)
Replace σx by σ ′ in Eq. (7) and let σe be the material yield strength f y ,
then the yield contour line R (x ) can be governed by Eq. (16) with the
participation factor of axial force β introduced as Eq. (17):
4t 2f y2 R 4 (x )−(3 + β 2) V 2R2 (x )−6|β|V 2xR (x )−9V 2x 2 = 0 (16)
Fig. 3. Stress state of the energy dissipater in plane.
N = βV (17)
V Given the design shear force V , the height h and the participation
τxy =
2t|P (x )| (9) factor β , the implicit equation can be solved by Matlab®. The envelop
where τxy is the shear stress uniformly distributed along the cross-sec- curve of the four solutions, R1 (x ) , R2 (x ) , R3 (x ) , and R 4 (x ) , defines the
tion. If all cross-sections yield simultaneously, the shape function P (x ) shape function f3 (x ) of the panel as Eq. (18):
can be obtained as Eq. (10) by letting τxy equal the shear yield strength f3 (x ) = ± max {|R1 (x )|, |R2 (x )|, |R3 (x )|, |R 4 (x )|} (18)
stress fv , which is 1/ 3 of the tensile yield strength f y .

3V 3.3. Stiffness of dampers with optimized shapes

P (x ) = ±
2tfy (10)
Once the profile shape function of the energy dissipater is de-
If the normal stress dominates the yielding behavior the profile termined by the stress contour line, its elastic stiffness will be further
shape function of the panel Q (x ) can be obtained as derived for the design purpose. The deformation of a panel, denoted as
M|Q (x )| 3Vx Δ , is the sum of the shear deformation Δs and the bending deformation
σx = ± =
Iz 2tQ 2 (x ) (11) Δb , as Eq. (19). Correspondingly, the elastic stiffness k of the damper
can be calculated as the series combination of the shear stiffness ks and
where σx is the normal stress at the panel edge. If the edge points yield the bending stiffness kb , as Eq. (20).
simultaneously for all cross sections, which means that σx equals f y , the
Δ = Δs + Δb (19)
shape function Q (x ) is calculated as:
1 1 1
3Vx = + = Δs + Δb
Q (x ) = ± k ks kb (20)
2tfy (12)
Given a unit shear force, the shear deformation Δs and the bending
Finally, the envelope curve of P (x ) and Q (x ) is taken as the shape deformation Δb of the panel are given as Eqs. (21) and (22), respec-
function f2 (x ) : tively.
f2 (x ) = ± max {|P (x )|, |Q (x )|} (13) 1 h
dx
Δs =
G
∫− h2 A (x ) (21)
2

3.2. Derivation of stress contour line under shear and axial forces h h
MM 3 x2
Δb = ∫− h2 EI (x )
dx =
2E
∫− h2 tf 3 (x )
dx
(22)
2 2
During earthquake shaking, axial forces may be introduced into the
damper. Suppose that a constant axial force is uniformly distributed where A (x ) is the area at the cross-section perpendicular to the axis x ,
along the right edge as shown in Fig. 4, then the normal stress caused by which equals 2f (x ) t , M is the bending moment, E is the young’s
the axial force is given by Eq. (14), where R (x ) defines the stress modulus, and G is the shear modulus. Substituting Eqs. (21) and (22)

Fig. 4. Stress state of energy dissipater under shear and axial loads.

349
B. Zhu et al. Engineering Structures 172 (2018) 346–357

Table 1
Material properties (Unit: N, mm).
Specimen Thickness Yield Ultimate Max. Ultimate strength
strength strength strain (%) Yield strength

BLY-160-1 10.91 196.1 298.8 64.9 1.52

BLY-160-2 10.90 203.2 303.9 69.7 1.50
BLY-160-3 10.92 209.7 309.4 73.5 1.48
Average 10.91 203.0 304.1 69.4 1.49
Q235-B-1 25.20 241.3 403.5 53.7 1.67
Q235-B-2 25.10 240.4 401.9 53.2 1.67
Q235-B-3 25.20 249.3 404.8 49.2 1.62
Average 25.16 243.7 403.4 52.0 1.66

Table 2
Configuration of specimens.
Fig. 6. Test setup.
Specimen number b (mm) h (mm) V (kN) β (N/V) Optimization type

ASSD-RS-1 303.5 400 180 0 Rectangle shape

ASSD-OS-2 318.3 400 180 0 Type-1
ASSD-OS-3 328.2 400 180 0.25 Type-1
ASSD-OS-4 338.2 400 180 0.5 Type-1
ASSD-RS-5 371.7 400 270 0 Rectangle shape
ASSD-OS-6 399.3 400 270 0 Type-1
ASSD-OS-7 414.0 400 270 0.25 Type-1
ASSD-OS-8 429.1 400 270 0.5 Type-1
ASSD-OS-9 371.7 400 270 0 Type-2

into Eq. (20) gives the elastic stiffness as Eq. (23):

1
k= h h
3 2 x2 1 dx
2E
∫
− h2 tf 3 (x )
dx + tG
∫−2h 2f (x ) (23)
2

4. Specimens and test scheme

Fig. 7. Loading protocol.

To demonstrate the effectiveness of the optimized shape on the

mitigation of plastic strain concentration, quasi-static tests are con-
ducted on nine specimens. The accuracy of the design formula for the
yield strength and the initial stiffness is also examined.
4.1. Material properties
Low yield steel with a nominal yield strength of 160 MPa, denoted
as BLY160, is used to fabricate the energy dissipater, because of its
excellent low cycle fatigue properties [25]. The standard coupon test
was conducted using three specimens. The thickness, the yield strength,
and the ultimate tensile strength are listed in Table 1. The yield strength
and the ultimate strength of BLY 160 are 203 MPa and 304 MPa, re-
spectively. The yield over-strength ratio is 1.27, while the ultimate to
yield strength ratio is 1.49. The maximum strain is about 69%. These
property parameters satisfy China’s seismic design code. The steel
Q235-B is used for the connectors and the buckling restrainers. Both
should remain elastic.
4.2. Configuration and dimensions of specimens
Nine dampers were designed for the quasi-static test, as listed in
Table 2, and classified into two groups according to their design shear
forces, i.e., 180 kN and 270 kN. In each group, there is one non-opti-
mized specimen with a rectangular shape, denoted as RS for compar-
ison, and three specimens with shapes optimized by the Type 1 method,
which are to be loaded vertically with axial force participation factors
of 0, 0.25, and 0.5 respectively. In the 270 kN group, one specimen’s
shape is optimized by the Type 2 method. The specimens with opti-
mized shapes are denoted as OS. All dampers are designed with a given
height of 400 mm. The width of each optimized damper is determined
by the corresponding shape function, while the width of that without
optimization is governed by the maximum nominal stress at each corner
introduced by the bending moment as the following equation:

(a) RS (b) OS-Type 1 (c) OS-Type 2

Fig. 5. Chamfering for corners and intersections.

350
B. Zhu et al.
Fig. 8. Hysteresis curves.
3vh
b=
tfy
(24)
Note that when designing the shape of the energy dissipater, the
measured thickness of the plate, 10.91 mm, and the averaged yielding
strength of BLY-160, 203 MPa, are used for the calculation. The curved
edge of the optimized dampers is processed by the wire cutting. The
diameter of the wire is a maximum of 0.3 mm. The whole cutting
process requires cooling and must be kept at a speed slower than
0.1 mm/s, to reduce the residual stress caused by the accumulated heat.
As shown in Fig. 5, the four corners are chamfered with a radius of
20 mm, and the intersection points close to the mid-height are cham-
fered with a radius of 50 mm. This treatment is used to avoid any po-
tential strain concentration.
351
Engineering Structures 172 (2018) 346–357
Friction-type 10.9-class 30 mm diameter high-strength bolts of
Chinese standard are used to assemble the L-shaped connectors, the
buckling restrainers, and the energy dissipater. All contact surfaces are
treated by sandblasting, and the friction coefficient is 0.45.
4.3. Test scheme
A test frame, as shown in Fig. 6, is used to test the specimens, which
consists of four columns and three beams to form a steady loading
frame. Each specimen is installed between the bottom beam and the
loading beam. The bottom beam is securely fixed on the strong reaction
floor, while the loading beam is connected to a parallelogram system
through which the rotation of the loading beam is restrained while the
translational movements remain free. One end of the loading beam is
B. Zhu et al. Engineering Structures 172 (2018) 346–357

Fig. 8. (continued)

Table 3
Primary mechanical properties (Units: kN, mm).
Specimen K Fy Δy Fu Δu ∑Δ ∑iE
(Fy Δy / 2)
Test Calculated Error (%) Test Calculated Error (%)

ASSD-RS-1 345.8 350.4 −1.3 184.2 180.0 +2.3 0.53 341.5 32.35 1802 8342.3
ASSD-OS-2 265.8 273.9 −3.0 173.5 180.0 −3.6 0.65 367.2 49.97 2163 9737.6
ASSD-OS-3 288.9 293.4 −1.5 175.8 180.0 −2.3 0.61 490.4 48.85 2016 11714.4
ASSD-OS-4 307.2 314.5 −2.3 178.5 180.0 −0.8 0.58 511.3 48.15 2109 13033.6
ASSD-RS-5 468.2 487.1 −3.9 260.5 270.0 −3.5 0.56 399.6 12.25 1128 5071.1
ASSD-OS-6 424.8 427.2 −0.6 261.8 270.0 −3.0 0.62 569.2 47.83 2355 7244.3
ASSD-OS-7 442.6 457.2 −3.2 260.2 270.0 −3.6 0.59 667.3 47.25 2127 10819.4
ASSD-OS-8 481.5 489.9 −1.7 267.3 270.0 −1.0 0.56 713.5 48.25 2071 11659.0
ASSD-OS-9 378.2 388.9 −2.8 215.5 270.0 −20.2 0.57 424.5 48.35 1992 8647.4

Note: K is the initial stiffness, Fy is the yield force, Δy is yield displacement, Fu is the average of positive and negative peak forces, Δu is the ultimate displacement, and
∑ Δ is cumulative ductility.

connected to the horizontal actuator, which is displacement-controlled
and used for the shear loading on the damper specimen. The vertical
actuator is connected at the mid points of the loading beam, which is
force-controlled and balances the gravity of the loading beam. It is also
used to exert the axial load on the specimens when considering a
constant design axial force.
To investigate low-cycle fatigue behavior, deformability, and the
energy dissipation capacity, the loading protocol, as shown in Fig. 7, is
used. This contains 30 cycles loading with an amplitude of 12 mm,
corresponding to the deformation under a design-basis earthquake,
after which the amplitude increases steadily until the horizontal force
decreases to 85% of the maximum [26,27]. The loading speed is no
more than 5 mm/min.
The horizontal and vertical loading forces are measured through the
actuator load cells. Two displacement transducers are set in the hor-
izontal, one at the bottom and the other on the top, but both are at-
tached directly to the energy dissipater of each damper. The relative
displacement of the two transducers is used as the displacement control
target of the horizontal actuator. Another two displacement transducers
are used to measure the vertical displacement and the rotation of the
loading beam.
5. Test results and discussion
5.1. Hysteretic performance
The hysteretic curves of all specimens are plotted in Fig. 8. A good
energy dissipation capacity is observed for all dampers, demonstrating

352
B. Zhu et al.
Fig. 9. Rotation of the loading beam.
that the shape-optimized dampers deformed with a shear dominated
pattern, and an apparent yielding point can be observed from the
curves. The dampers without optimization, however, behaved in a
bending flexural pattern, and the global yielding point is difficult to
identify. The drift angles of the optimized dampers all achieved 12%,
while the non-optimized dampers, ASSD-RS-1 and ASSD-RS-5, lost 15%
of their maximum bearing force at amplitudes of 30 mm and 12 mm,
respectively. All the optimized dampers experienced 30 cycles loading
of 12 mm amplitude, and failed at the amplitude of 48 mm, indicating a
better low cycle fatigue performance than the two without optimiza-
tion.
5.2. Primary mechanical properties
The initial stiffness of each specimen is calculated by Eq. (23) and
listed in Table 3. The measured values match the calculated values very
well, but are slightly lower with the maximum error being less than
-4%, possibly due to the rotation of the loading beam. From the data
obtained from the two vertical displacement transducers, a rotational
angle of a maximum of 0.003 rad occurred during the cyclic loading, as
shown in Fig. 9, which released the boundary constraint of the energy
dissipater. From the experimental data, the yield force can be easily
identified for the optimized damper. However, there are no obvious
yield points in the two non-optimized dampers. Instead, their yield
forces were obtained by the graphical method introduced by [28]. The
design equations for yield force all predicted the measured values very
well except for specimen ASSD-OS-9, for which a large discrepancy of
−20.2% was observed because of the assumption of uniformly dis-
tributed shear stress in the design equations. The shear stress is in fact
(a) ASSD-OS-6
Fig. 10. Development of over-strength ratio with loading cycles.
353
Engineering Structures 172 (2018) 346–357
parabolically distributed along the cross-section. The shear-dominated
middle part of ASSD-OS-9 yielded earlier than expected and for the
same reason, its ultimate force capacity was also much smaller than
expected. The ultimate force of ASSD-RS-1 is similar to that of ASSD-
OS-2. However, those considered as having the axial force effect have a
greater ultimate force capacity. The ultimate forces of ASSD-OS-3 and
ASSD-OS-4 are 1.34 and 1.39 times that of ASSD-OS-2, respectively,
and those of ASSD-OS-7 and ASSD-OS-8 are 1.17 and 1.25 times that of
ASSD-OS-6, respectively. Considering the axial force in the design
process leads to a larger cross-section. As for the ultimate deformation,
the optimized panel is on average 60% larger than that without opti-
mization for the 180kN group, while the for 270kN dampers, this value
is 300%. The superior ductility could help the shape-optimized damper
survive very severe earthquakes. Accumulated energy is one of the key
parameters for the energy-based design. To make it more general, the
accumulated energy ∑ Ei is scaled by the elastic strain energy of Fy Δy /2 ,
and listed in Table 3. It is found that the scaled energy ratio highly
dependent on the aspect ratio b/h. A recursive equation of the Type 1
damper without axial force is obtained as Eq. (25). However, it shall be
noted that more data is needed to improve the precision of the equa-
tion.
∑ Ei b
= −12282 + 19501.8 (0.8 ⩽ b/ h ⩽ 1)
(Fy Δy /2) h (25)
5.3. Over-strength of bearing force
It should be noted that all energy dissipaters have similar strain-
hardening characteristics. In the 5 mm amplitude cycles, the over-
strength ratio, defined as the ultimate force Fu′ at a cycle divided by the
yield strength Fy , increased slightly with the accumulation of plastic
deformation. During the low fatigue cycles of 12 mm amplitude, the
hardening developed quickly in the first cycles and then remained al-
most constant, as shown in Fig. 10(a), where ASSD-OS-6 without axial
load is taken as an example. In the following cycles with larger am-
plitudes, the over-strength ratio increased again, which means that the
strain-hardening is amplitude sensitive. The axial compression often
resulted in contact between the L-shaped connectors and the buckling
restrainers, as in the ASSD-OS-7 specimen shown in Fig. 10(b). Once in
contact, an increase of the over-strength ratio from 1.85 to 2.02 was
observed. This was a point to surface contact because of the slight ro-
tation of the loading beam, as shown in Fig. 11(a). A significant contact
force was observed in the strain history of the buckling restrainer, as
shown in Fig. 11(b), and the associated friction force led to the increase
of the over-strength ratio. However, because the rotation in the positive
direction is a little different from that in the negative direction, as
shown in Fig. 9, the associated friction force introduced by the contact
are different correspondingly, as shown in Fig. 8- (3), (4), (7), and (8).
The over-strength ratios at an amplitude of 12 mm and the
(b) ASSD-OS-7
B. Zhu et al. Engineering Structures 172 (2018) 346–357

(a) Strain gauges at contact points (b) Strain history

Fig. 11. Strain of buckling restrainer for ASSD-OS-7.

Table 4 strength, the overturning resistance from the axial load, and the contact
Over-strength ratios of all specimens. friction led to this result, which requires further research to explicitly
Specimen type Constant ratio at 12 mm specify the influence of each factor.
F
Maximum ratio ⎛ u ⎞
amplitude ⎝ Fy ⎠
5.4. Failure mode
V= 180 kN V= 270 kN V= 180 kN V= 270 kN
The damage patterns differed with the optimized shapes. ASSD-RS-
ASSD-RS (N = 0) 1.82 1.62 1.94 1.65
ASSD-OS type1 1.76 1.41 2.23 2.22 1, ASSD-OS-2, and ASSD-OS-9 are representatives of dampers with a
(N = 0) non-optimized shape, Type 1 optimized shape, and Type 2 optimized
ASSD-OS type1 1.95 1.85 2.91 2.70 shape, respectively. To investigate the deformation patterns of these
(N = 0.25 V) dampers, the surface of the energy dissipation panel was sculpted with
ASSD-OS type1 2.03 1.86 3.09 2.73
(N = 0.5 V)
4 cm by 4 cm grids. The depth of the sculpted lines was less than
ASSD-OS type2 – 1.64 – 2.13 0.1 mm, which avoided causing any adverse influence on the mechan-
(N = 0) ical properties of the energy dissipater.
The final damage patterns of the three dampers are shown in
Fig. 12. ASSD-RS-1 suffered a premature failure at the corners of the
maximum values for all specimens are listed in Table 4. The non-opti- dissipater where the crack occurred, due to the significant plasticity
mized specimen ASSD-RS-5 experienced pre-matured failure in the concentration, and finally penetrated through the closest bolt hole so
amplitude of 12 mm. Its maximum value is therefore similar to that at that the energy dissipater could not be tightly clamped, thus resulting in
the amplitude of 12 mm. The over-strength ratios of ASSD-OS-2 and complete failure of the damper. Fig. 12(a) shows the plastification area
ASSD-OS-6 are 1.76 and 1.41 at the amplitude of 12 mm, respectively. of the surface. The areas along both edges close to the mid-height re-
The height-to-width aspect ratio of ASSD-OS-2 is larger than ASSD-OS- mained elastic. The grids were found to be not particularly distorted,
6. The deformation of ASSD-OS-2 was dominated by a mixed shear- implying that the shear deformation was limited. In contrast, all cross-
bending behavior, where both material over-strength and sectional sections of the two types of optimized energy dissipater yielded en-
plastification determined the overall over-strength. As for ASSD-OS-6 tirely. The plasticity concentration at the four corners was mitigated
dominated by shear deformation, the material over-strength primarily and the fracture was not observed. ASSD-OS-2 yielded along the opti-
took the largest portion. The over-strength ratio is similar to the ma- mized edges, as shown in Fig. 12(b). Strain concentration was not ob-
terial average over-strength given in Table 1. Both specimens finally served in the area close to the mid-height, though it was dominated by
approached a value of 2.25, implying that both sectional plastification shear deformation because of the continually reduced section width.
and material over-strength were fully developed, each providing an However, this was not observed for ASSD-OS-9 where the strain con-
amplifier of 1.5. However, for the specimen ASSD-OS-9, the maximum centration resulted in a fracture in the four corners of the shear link at
ratio was 2.13, implying that the plasticity was not as fully developed as mid-span. Fig. 12(c) shows that the shear link at the midspan was
the Type-1 specimens. The over-strength ratios are much larger for the dominated by the shear deformation, and both ends were dominated by
specimens that sustained axial loads than those without, but the ratios the flexural behavior. The design target has also been achieved. An-
changed slightly when the axial loads increased from 0.25 V to 0.5 V. A other photo for ASSD-OS-7 under compression was given as Fig. 12(d).
complex combination of sectional plastification, the material over- There’s no fracture observed in the energy dissipation area. Instead, the

Fig. 12. Typical failure mode.

354
B. Zhu et al.
Fig. 13. Bilinear mechanical model of ASSD-OS Type1 without axial force.
plastification has been fully developed in the entire area.
5.5. Mechanical model
For the convenience of design practice, a mechanical model has
been developed for this metallic damper with such specific configura-
tion as Type 1 which has been found more favorable from experimental
results. The mechanical behavior can be simplified as a bilinear model
with kinematic hardening. An recursive procedure as shown in Fig. 13
suggested a second stiffness ratio of 0.018. Experimental data sum-
marized in Table 3 demonstrated a good prediction of stiffness of Eq.
(23) so that it will be used to calculate the initial stiffness once the
shape is determined by Eq. (8) given a yielding force V. Similar model
Fig. 14. Axial deformation with loading step.
355
Engineering Structures 172 (2018) 346–357
can be developed for those dampers sustaining a given axial load. It will
not be presented because of the length limit of the paper.
5.6. Axial deformation
Under shear and axial forces, the damper may experience complex
behavior and even premature failure. Therefore, the axial deformation
of the proposed dampers is closely examined in this section. Axial de-
formation is caused by four actions: geometric nonlinearity introduced
compression, compression caused by lateral buckling, extension caused
by the shear-plasticity, and direct axial compression from the axial load.
The global axial deformation given in the following was obtained by
averaging the measured values from the two vertical displacement
transducers installed at the two ends of the loading beam.
5.6.1. Without axial load
A vertical actuator was used to balance the gravity of the loading
beam so no axial force was exerted on the specimen. The first axial
deformation is due to the geometric nonlinearity. When the specimen
was loaded horizontally, there was a vertical displacement due to this
nonlinearity, as shown in Fig. 14(a), which would recover completely
with a horizontal displacement of zero. No obvious lateral buckling was
observed on the two dampers without optimization. Therefore, the
steady increase in vertical displacement was completely due to the
extension caused by the shear plasticity, which is along the diagonal of
the panel and it cannot be recovered. It was more pronounced at the
large horizontal deformation when fracture occurred. Fig. 14(b) shows
the axial deformation of the optimized damper with the design shear
force of 180 kN. Taking ASSD-OS-2 as an example, the high mode of
lateral buckling occurred at the horizontal loading with small ampli-
tudes. Therefore the compression introduced by the lateral buckling
B. Zhu et al.
Fig. 15. Finite element model.
dominated the axial deformation at the beginning until the panel was
damaged, because of the large shear plasticity when the axial de-
formation became extension. As shown in Fig. 14(c), ASSD-OS-6 be-
haved similarly in the axial, but the shear plasticity is more pronounced
than in ASSD-OS-2 because of the smaller height-width ratio. The axial
displacement was first dominated by the shear-plasticity introduced
extension, and then controlled by the high mode of lateral buckling.
Finally, it was again extended by a larger shear plasticity.
5.6.2. Under axial load
All specimens under axial load behaved similarly, as shown in
Fig. 14(d) where the damper ASSD-OS-3 is used as an example. Al-
though the other three actions did exist, the axial deformation was
primarily dominated by the axial load. The axial deformation
Fig. 16. Nephogram of von Mises stress for typical models.
356
Engineering Structures 172 (2018) 346–357
accumulated with the horizontal deformation until it reached the 16th-
cycle of low cycle fatigue loading with 12 mm amplitude when the L-
shaped connectors came into contact with the buckling restrainers, thus
becoming almost constant. Finally, at the larger shear angle, the axial
deformation increased because of the large shear plasticity effect.
6. Numerical simulation
6.1. Finite element model (FEM)
A general-purposed finite element software MSC. MARC® is em-
ployed to develop the numerical models for the four dampers, i.e.,
ASSD-RS-5, ASSD-OS-6, ASSD-OS-8 and ASSD-OS-9 for further com-
parison. As shown in Fig. 15, only the energy dissipater is modeled by
use of the four-node element 3D-SHELL No. 75 with reduced integration
for simplicity. And the out-of-plane deformation is restrained for all
nodes because the buckling restrainer is deemed strong enough. The
nodes on the left connection region are all fixed through the rigid link
confinement, while the nodes on the right connection region are all
confined and only movements along x and y directions are allowed. The
target displacement is applied in the y direction, while the axial force
applied in the x direction. The material property is simulated by the
bilinear model with kinematic hardening rule based on Table 1. And the
loading protocol is the same as that used in the quasi-static test, as
shown in Fig. 7.
6.2. Simulation results
At the moment of yield, the nephogram of von Mises stresses for
four dampers are shown in Fig. 16. The maximum stresses are
B. Zhu et al.
concentrated in the four corners of ASSD-RS-5 with rectangle shape as
observed from test. All the while, ASSD-OS-6 and ASSD-OS-8 start yield
along the optimized shape edges because of bending and the center of
the section due to the shear. The design target has been well achieved.
Stress concentration exists at the four corners of shear stress-dominated
region of ASSD-OS-9, where the pre-matured facture was also observed
during the test. All the yielding patterns agree well with those observed
from the test.
7. Conclusions
A new type of assembled steel shear panel damper with an opti-
mized shape is presented. The stress contour line is developed ac-
cording to the J2 plasticity theory, and the optimized shape is obtained
by assuming that the points on the same contour line yield simulta-
neously. The design formulas for the shear yield strength and the elastic
stiffness are further derived. Nine specimens are then designed con-
sidering different loading conditions, and tested quasi-statically to
verify the effectiveness of the proposed optimization method, and the
accuracy of the design formula. The main findings are summarized as
follows:
(1) The design formulas for the elastic stiffness agree well with the test
results, with an error range from −3.9% to −0.6%. The yield
strength formula for type-1 is from −3.6% to +2.3%. Both can be
used in engineering applications.
(2) The shape optimized dampers have better mechanical behavior
than that without optimization. The consumed energy increased by
40.65% to 63.69%, and the ultimate shear deformation is much
larger than for the one without optimization because of the low
concentration of plastic strains.
(3) The axial deformation is the result of four actions: geometric non-
linearity, compression caused by lateral buckling, extension caused
by the shear plasticity, and axial compression directly from the
axial load. Each accounts for different proportions during various
loading amplitudes.
(4) The over-strength ratio of shear type dampers is preliminarily
analyzed. The four factors affecting the over-strength behavior are
the sectional plastification, the material over-strength, the over-
turning resistance from axial force, and the contact friction. It is
difficult to quantify the influence of each factor because of the
limited number of specimens. Further studies can examine the over-
strength behavior experimentally and numerically.
Acknowledgements
This research was funded by the Scientific Research Fund of the
Institute of Engineering Mechanics, CEA (2016A06), the National Key
Research and Development Program of China (2016YFC0701101), the
International Science and Technology Cooperation Program of China
(2014DFA70950), the National Natural Science Foundation of China
(51378478, 51678542), and the Program for Innovative Research Team
in China Earthquake Administration. The authors are also grateful to
the research team of Huixian Key laboratory, Institute of Engineering
357
Engineering Structures 172 (2018) 346–357
Mechanics, China Earthquake Administration.
References
[1] Xiong L, Lan R, Wang Y, Tian X, Feng B. Earthquake damage investigation of
structures in 7.0 Lushan strong earthquake. J Earthquake Eng Eng Dyn
2013;04:35–43.
[2] Zhou Y. Design theory and application of metal energy dissipation structure. Wuhan
University of Technology Press; 2013. p. 2–10.
[3] Kelly JM, Skinner RI, Heine AJ. Mechanisms of energy absorption in special devices
for use in earthquake resistant structures. Bull N. Z. Soc Earthquake Eng
1972;5(3):63–88.
[4] Ji X, Ma Q, Wang Y, et al. Cyclic tests of replaceable shear links in steel coupling
beams. J Build Struct 2014;35(06):1–11.
[5] Okazaki T, Arce G, Ryu HC, Engelhardt MD. Experimental study of local buckling,
overstrength and fracture of links in eccentrically braced frames. J Struct Eng, ASCE
2005;131(10):1526–35.
[6] Okazaki T, Engelhardt MD. Cyclic loading behavior of EBF links constructed of
ASTM A992 steel. J Constr Steel Res 2007;63(6):751–65.
[7] Mc Daniel C, Uang CM, Seible F. Cyclic testing of built-up steel shear links for the
New Bay Bridge. J Struct Eng, ASCE 2003;129(6):801–9.
[8] Chan Ricky WK, Albermani Faris. Experimental study of steel slit damper for passive
energy dissipation. Eng Struct 2008;30(4):1058–66.
[9] Lee Chang-Hwan, Jub Young K, Minc Jeong-Ki, Lhod Seung-Hee, Kimb Sang-Dae.
Non-uniform steel strip dampers subjected to cyclic loadings. Eng Struct
2015;192–204.
[10] Chan Ricky WK, Albermani Faris, Kitipornchai Sritawat. Experimental study of
perforated yielding shear panel device for passive energy dissipation. J Constr Steel
Res 2013:14–25.
[11] Lee Chang-Hwan, Lho Seung-Hee, Kim Do-Hyun, Jintak Oh, Young KJu. Hourglass-
shaped strip damper subjected to monotonic and cyclic loadings. Eng Struct
2016;122–134.
[12] Valizadeh H, Sheidaii M, Showkati H. Experimental investigation on cyclic behavior
of perforated steel plate shear walls. J Constr Steel Res 2012;70(2):308–16.
[13] Ge H, Chen Z, Usami T. Hysteretic model of stiffened shear panel dampers. J Struct
Eng 2006;132(3):478–83.
[14] Koike Y, Yanaka, Tsutomu U, Akihisa, et al. An experimental study on developing
high-performance stiffened shear panel dampers. J Struct Eng 2008;54:372–81.
[15] Koike Y, Yanaka T, Kasugai. A performance test of stiffened shear panel dampers.
Yokogawa bridge holdings group technical report; 2008. p. 30–37.
[16] Chusilp P, Usami T, Ge H, et al. Cyclic shear behavior of steel box girders: ex-
periment and analysis. Earthquake Eng Struct Dyn 2002;31(11):1993–2014.
[17] Tanaka K, Miyama T, Meng L, et al. Study on the passive-vibration controlled
building with low-yield-point steel damper: Part 6. Test of dampers with shear
panel stiffened by rib plates: outline and results of the tests. Architectural Institute
of Japan; 1995. p. 651–2.
[18] Ohsaki M, Nakajima T. Optimization of link member of eccentrically braced frames
for maximum energy dissipation. J Constr Steel Res 2012;75:38–44.
[19] Zhang C, Zhang Z, Shi J. Development of high deformation capacity low yield
strength steel shear panel damper. J Constr Steel Res 2012;75(7):116–30.
[20] Liu Y, Aoki T, Shimoda M. Strain distribution measurement of a shear panel damper
developed for bridge structure. J Struct 2013:1–11.
[21] Liu Y, Shimoda M. Shape optimization of shear panel damper for improving the
deformation ability under cyclic loading. Struct Multidiscip Optim 2013:1–9.
[22] Deng K, Pan P, Sun J, et al. Shape optimization design of steel shear panel dampers.
J Constr Steel Res 2014;99(8):187–93.
[23] Deng K, Pan P. Experimental study of steel shear panel dampers with varying cross-
sections. Eng Mech 2016;99(8):187–93.
[24] Chen M. Elasticity and plasticity. Beijing China: Science Press Ltd, vol. 10. 01; 2015.
p. P230.
[25] GB50011-2010. Code for seismic design of buildings. Beijing: Ministry of
Construction of the People’s Republic of China; 2010.
[26] CMC (Ministry of Construction). Specification of testing methods for earthquake
resistant building. JGJ 101-96. Beijing; 1997 [in Chinese].
[27] CMC (Ministry of Construction). Technical specification for seismic energy dis-
sipation of buildings. JGJ 297-2013. Beijing; 2013 [in Chinese].
[28] Xu L, Nie X, Fan J. Cyclic behavior of low-yield-point steel shear panel dampers.
Eng Struct 2016:391–404.