Engineering Structures 138 (2017) 105–119

Contents lists available at ScienceDirect

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

Development of a novel high-performance all-steel fish-bone shaped

buckling-restrained brace
Liang-Jiu Jia a, Hanbin Ge b,⇑, Rikuya Maruyama b, Kazuki Shinohara b
a
Research Institute of Structural Engineering and Disaster Reduction, Tongji Univ., Shanghai 200092, China
b
Dept. of Civil Eng., Meijo Univ., Nagoya 468-8502, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Based on investigation of recent strong earthquakes, there is a potential that BRBs may rupture during a
Received 2 February 2015 strong earthquake or subsequent repeated aftershocks. This study aims to propose a novel type of light-
Revised 24 November 2016 weighted all-steel dismountable BRB with fish-bone shaped core plate, which is termed FB-BRB in this
Accepted 2 February 2017
paper. The FB-BRB consists of a core plate, two filling plates, two restraining plates and unbonding mate-
rial. Deformation capacity of the proposed FB-BRB is to be maximized by generating several necking loca-
tions at the core plate, and details to avoid strain concentration at stoppers are also proposed.
Keywords:
Experimental study is carried out using four scaled specimens with different configurations. Favorable
Bucking-restrained brace
All-steel
seismic performance is obtained through comparison with that of a conventional BRB. The failure mech-
Fish-bone anisms of the newly proposed FB-BRBs are also further verified through numerical study using a combi-
Seismic nation of a ductile fracture model and a cyclic plasticity model, where further improvement is required to
Cyclic fully achieve the expected deformation mechanism.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction where Dmax = BRB maximum deformation; Dy = BRB yield deforma-

Buckling-restrained braces (BRBs) have been increasingly
employed in building and bridge structures since the 1970s when
BRBs, e.g., [1–4], were first developed in Japan. Applications of
BRBs in structural engineering have been approved by more and
more countries, especially after several strong earthquakes in
recent decades, such as the 1994 Northridge earthquake [5,6],
the 1995 Kobe earthquake [7], the 1999 Chi-Chi earthquake [8],
the 2008 Wenchuan earthquake [9] and the 2011 Tohoku earth-
quake [10].
Two indices are generally employed to evaluate ductility capac-
ities of BRBs, i.e., a maximum ductility index, lmax , and a cumula-
tive ductility index, lc . The two can be obtained using the
following formulae
Dmax
lmax ¼
Dy
P responding cumulative ductility demand is around 193 [21]. A
Dp
lc ¼
Dy
⇑ Corresponding author.
E-mail addresses: lj_jia@tongji.edu.cn (L.-J. Jia), gehanbin@meijo-u.ac.jp (H. Ge).
P
tion; Dp = accumulated BRB plastic deformation. These variables
are all calculated for the yielding portion of the core plate.
According to a number of experimental and analytical studies,
the maximum ductility demand, lmax , for buckling-restrained
braced frames (BRBFs) under a seismic input with a 2% exceedance
probability in 50 years in the US, ranges from 20 to 25 [11]. Based
on available experimental results to date, e.g., [12–19], there are a
number of BRBs which cannot meet the maximum ductility
demand of a BRBF. The cumulative ductility index, lc , of a BRBF
is required to be larger than 200 in the US design code [20]. It
has been found that the required cumulative ductility of 200 can
be readily achieved for BRBFs [11]. The maximum ductility
demands for BRBs employed in steel arch bridges are close to those
a BRBF building. When a material with a yield stress of 235 MPa is
employed for the core plate, the seismic demanded maximum
ð1Þ average elongation of a BRB in an arch bridge is around 3%, corre-
sponding to a maximum ductility index of 26. Meanwhile, the cor-
ð2Þ ductility capacity lower than the demand, will lead to failure of
the BRBs due to either buckling or rupture of the core plate during
a strong earthquake. Besides, strong aftershocks following a strong
earthquake were observed in recent years [21,22], and higher duc-
tility demands are required for this type of seismic waves with long
durations and strong aftershocks.

http://dx.doi.org/10.1016/j.engstruct.2017.02.006
0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.
106 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
Nomenclature
a initial imperfection at the mid-length of a BRB
bc width of the core plate
bs width at root of the stopper
CID cumulative inelastic deformation
D half of the reduced size at mid-width of the core plate
d gap between the core plate and the restraining plate
d0 gap between the core plate and the filling plate
E Young’s modulus
Es allowable maximum elongation of each segment
Fc, i cross sectional total force of the i-th segment of core
plate
Fc, i+1 cross sectional total force of the (i + 1)-th segment of
core plate
Fs shear force sustained by a single branch of stoppers
e eccentricity of the compressive force of the core plate
L length of the yielding portion
M Ry yield moment of the restraining plate
PRE buckling load of the restraining plate
Py yield load of yielding portion of the core plate
Pu peak load of a coupon
Typical requirements of a BRB include [23]: (1) stable hysteretic
characteristic and high energy absorption capacities; (2) large
deformation capacities under both compression and tension; (3)
high ductile fracture-resistant properties of the core plate under
a small number of loading cycles with large plastic strain ampli-
tudes; (4) simple and low-cost fabrication and construction (e.g.,
connection details); (5) high low-cycle fatigue properties; (6) good
weathering properties; (7) ease of replacement or no need to be
replaced during service life. A number of all-steel BRBs with large
maximum ductility capacity and cumulative ductility capacity, e.g.,
[24–35], have been developed. Compared with BRBs with mortar
filled tubes as restraining components, all-steel BRBs have more
stable hysteretic properties owing to high manufacturing accuracy
and reliable mechanical properties of steel. Besides, compared all-
steel BRBs with mortar filled steel tube type BRBs, the mass of the
restraining components can be reduced by around 40%. Mean-
while, it should be noted that welding within the core can lead
to poor low-cycle fatigue property compared with that of all-
steel BRBs with non-welded core, e.g., [35].
Stoppers [24,25] are also employed to ensure that the restrain-
ing plates move simultaneously with the core plate, and to prevent
premature local buckling at the transition parts close to the con-
nections. A number of experimental studies on all-steel BRBs with
stoppers are experimentally investigated. Typical failure modes
reported in previous studies (see Fig. 1 [36]) include: (1) cracking
Fig. 1. Several typical failure modes of conventional all-steel buckling-restrained braces.
S number of stoppers at one side of the core plate
t thickness of the core plate
Dmax maximum deformation of a BRB
Dy yield deformation of a BRB
d elongation of a material
du deformation at instant of peak load of a coupon
d0:95Pu deformation at instant when Pu decreases by 5%
dmax maximum deformation of a coupon
ep average plastic strain of the yielding portion
gs stopper strength index
lc cumulative ductility index
lmax maximum ductility index
mF index to control global buckling of a BRB
ry yield strength of a material
rt tensile strength of a material
sy yield shear stress of a material
REp
P cumulative plastic energy dissipation
Dp cumulative plastic deformation
at the weld toe of the stiffener-to-core-plate joint due to poor
fracture-resistant capacity under cyclic loading; (2) cracking at
the weld toe of the stopper-to-core-plate joint; (3) cracking at base
metal of the core plate; (4) local buckling at the transition portion
of the core plate due to large unrestrained length of the core plate;
(5) local buckling of the restraining plate; (6) global buckling; (7)
local buckling of the restraining plates. Local buckling at the tran-
sition part is not a problem if the maximum ductility demand is
below a certain value, e.g., 3%, where the premise is that the details
at the transition part are well designed. For a restraining plate with
a large width-to-thickness ratio, local buckling of the restraining
plate can occur under compression. This failure mode can be
avoided by specifying a limit value for the width-to-thickness ratio
of the restraining plate. Global buckling is related with the slender-
ness ratio of the whole BRB cross section. Likewise, global buckling
of the whole member can be avoided by specifying the limit value
of the slenderness ratio of the whole BRB cross section. Meanwhile,
the aforementioned failure modes (4)–(6) are all correlated with
the maximum ductility of a conventional BRB. The maximum com-
pressive force commonly increases as the maximum ductility
increases, and a large maximum ductility corresponds to a higher
demand on both the local and global buckling loads. Utilizing the
stoppers can greatly improve the local buckling load of the transi-
tion part, since it can reduce the unrestrained length of the transi-
tion part at the instant when the deformation changes from
 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119 107

Fig. 2. Proposed fish-bone buckling-restrained brace.

tension to compression. Application of the stoppers can commonly
increase the maximum ductility, which also makes a BRB more vul-
nerable to local buckling of the restraining plates and global buck-
ling of the whole member.
In this study, a novel all-steel fish-bone shaped BRB (FB-BRB)
aiming to obtain high ductility capacity was proposed. Experimen-
tal studies utilizing three small-scale FB-BRBs were conducted to
investigate its failure modes and corresponding mechanisms, such
as ductile fracture and core plate buckling. A conventional BRB was
also experimentally investigated to compare with the seismic per-
formance of the FB-BRBs. Numerical fracture simulation of FB-BRBs
was also carried out to investigate the deformation and failure
mechanisms, where a micro-mechanics based ductile fracture
model and a cyclic plasticity model proposed by one of the authors
[37–40] were employed.

2. Research program

2.1. Scope of this study

A series of proof tests to develop a new type of high-

Fig. 3. Deformation of a coupon under monotonic tension for mild steel.
performance all-steel FB-BRBs were conducted in this study, aim-
ing to investigate the deformation mechanisms and failure modes.

Fig. 4. Expected deformation and rupture mechanisms of conventional and fish-bone buckling-restrained braces.
108 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119

Fig. 5. Configurations of specimens in this study.

 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119 109

Table 1
Dimensions of specimens.

No. Specimens L (mm) d (mm) d0 (mm) S Es (%) D (mm)

1 Common-BRB 670 1 2 0 – 0
2 FB-BRB-S5-E8-D0 670 1 2 5 8 0
3 FB-BRB-S2-E8-D0 670 1 1 2 8 0
4 FB-BRB-S2-E6-D3 670 1 1 2 6 3

Note: L = length of yielding portion of the core plate; d = gap between the core plate and restraining plate; d0 = gap between the core plate and restraining plate; S = number of
stoppers; Es = allowable maximum elongation of the segment between two adjacent stoppers; D = half of the reduced width at the center of each segment.

Table 2
Mechanical properties of steel.

Component E (GPa) ry (MPa) rt (MPa) d (%) t (mm) Direction

Core plate 212 289 433 32.1 10 Perpendicular
211 276 432 29.6 10 Rolling
Filling plate 212 283 433 33.0 12
Restraining plate 212 269 425 33.4 16

Note: E = Young’s modulus, ry = yield stress, rt = tensile strength, d = elongation (a gauge length of 200 mm), t = thickness of steel plate.

Fig. 6. Improved details for stiffener details of conventional buckling-restrained braces.

110 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119

An FB-BRB includes the following main components as illustrated
in Fig. 2: (1) fish-bone shaped core plate with multiple stoppers;
(2) restraining plates; (3) unbonding material; (4) filling plates
which are thicker than the core plate to create gaps between the
core plate and the restraining plates. In this study, butyl rubber
was employed as the unbonding material. The concept of the FB-
BRB was motivated by a tension coupon test result of steel shown
in Fig. 3. Deformation capacity of a material is commonly repre-
sented by elongation of a tension coupon, which can be divided
into two parts, i.e., pre-necking and post-necking. Necking is com-
monly assumed to initiate at the peak load shown in the figure. For
mild steel such as SS400 or SN400, the post-necking deformation
takes a large portion of the whole deformation capacity. Generally,
failure of a structural member is defined at the instant when
decrease of the load is up to 5% or 10%. It can also be found from
Fig. 3 that the load decreases slowly after necking initiation, imply-
ing that the post-necking deformation can be employed to enlarge
the deformation capacity of steel components such as BRBs. Failure
modes due to buckling of BRBs can be solved in most cases through
proper details and design parameters, and the ductility capacity is
generally dominated by the deformation capacity of the core plate
at tensile side. For conventional BRBs, there is always only a single
neck within the core plate as illustrated in Fig. 4(a), which makes
deformation of the whole specimen concentrate at a localized
region during the post-necking stage. This deformation mechanism
limits the ductility capacity of a conventional BRB, and there is a
potential to fully utilize the post-necking deformation if there are
several necks occurring within the core plate of a BRB. Based on
this concept, the FB-BRB was proposed, where several necks were
expected to develop before rupture of the core plate at tensile half
cycles. The deformation mechanism of a conventional BRB before
rupture and the expected one of an FB-BRB are shown in Fig. 4(b).
2.2. Configuration of specimens
Three small-scale FB-BRBs shown in Fig. 5(a) and (c) were
designed and manufactured, where the restraining plates were of
the same configuration. A conventional BRB without stoppers
shown in Fig. 5(a) and (b) was also experimentally tested. The
yielding portions of the core plate for the two types of specimens
were with a same length of 670 mm, and the length of the whole
specimens was 940 mm. Geometrical parameters of the FB-BRBs
and the conventional BRB are shown in Fig. 5(b) and (c), and also
listed in Table 1. The numbering of the FB-BRBs is illustrated by
Specimen FB-BRB-S2-E6-D3, where ‘‘S” = number of stoppers, S,
‘‘E” = allowable maximum elongation of the segment between
two adjacent stoppers, Es, ‘‘D” = half of the reduced size at mid-
width of core plate. The possible numbers of necks before rupture

Fig. 7. Configuration of assembling process for fish-bone buckling-restrained braces (FB-BRB-S5-E8-D0).

Fig. 8. Test setup for specimens.

 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
of the specimens were determined by S, and 3 values for S were
designed. Proper selection of the allowable maximum elongation,
Es, is critical for the success of the experiments. Preliminary
numerical fracture simulations using validated cyclic plasticity
and ductile fracture model were carried out to determine the value
of Es, which was finally set to 6% and 8%. For BRBs, a gap between
the core plate and the filling plate, d0, is required to avoid transfer
of shear forces, which ranges from 1 mm to 2 mm in this study
[41]. A distance of 1 mm was adopted for the gap between the core
plate and the restraining plate for all the specimens. Cracks initiat-
ing at edges of stoppers are reported in previous studies, e.g., [14].
To avoid this unfavorable failure mode, one of the core plate width
was reduced to make necking and rupture occur at the centers of
the segments, and the strain localization close to the stoppers
can thus be mitigated. The width of the core plate was reduced
by a value of 2D as shown in Fig. 5(c).
Fig. 9. Loading history of specimens.
Fig. 10. Cyclic average stress-average strain curves of specimens.
111
To exclude the effect of material difference, all the steel plates
with the same thickness were respectively cut from the same plate.
The material was made of SS400, which is a mild steel widely
employed in structural engineering in Japan. Coupon tests cut
along the rolling direction of the steel plates were conducted for
the 3 types of steel plates, and those along the direction perpendic-
ular to the rolling direction were tested for the 10 mm thickness
plate, which were employed as the core plates. The coupon test
results are listed in Table 2.
Besides, details of the stiffener employed in the present study
are distinguished from previous studies. Cracking at the weld toe
of stiffener-to-core-plate joint shown in Fig. 1(a) is reported, which
is mainly due to strain concentration and poor fracture-resistant
properties of the heat affected zone of the fillet welds. A new detail
for the stiffeners illustrated in Fig. 6(b) was employed by all the
specimens in this study to prevent this failure mode. The edge of
the stiffener was located within the transition portion shown in
Fig. 6(b), where the strain level is much lower than that in Fig. 6
(a). The problem can also be solved by post-weld treatment [24],
but the newly proposed detail is simpler and cost-effective. In
addition, it can also be found from Fig. 6 that the yielding portion
of the BRB with the newly proposed detail is longer than that with
the previous one.
The assembling process for the FB-BRB is illustrated by Speci-
men FB-BRB-S5-E8-D0 shown in Fig. 7. The core plate, filling plates
and restraining plates are respectively manufactured using a 10, 12
and 16 mm thickness steel plate. The dimensions of the specimens
were designed with the consideration of the capacity of the loading
system, and the overall buckling was prevented using proper
design criterion in terms of a safety index, mF , given in the following
equation [42],
1
mF ¼   P 3:0 ð3Þ
Py Py L
PRE
þ M Ry
 aþdþe
L
where Py = yield load of the core plate; PRE = buckling load of the
restraining plate; MRy = yield moment of the restraining plates;
112 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
a = initial imperfection at the mid-length of a BRB; d = gap between
the core plate and the restraining plates; L = length of yielding por-
tion of the core plate; e = eccentricity of the compressive force of
the core plate. For an FB-BRB, the restraining plates have to not only
restrain the buckling of the core plate, but also limit the maximum
elongation of the segments of the core plate subsequently. If the
maximum compressive deformation is limited, the restraining
plates will be subjected to additional compressive forces transferred
through the stoppers. Besides, the maximum deformation capacity
of an FB-BRB is larger than that of a conventional one, which may
make a value satisfying Eq. (3) not enough to prevent premature
global buckling. Since the main concern in this study is to realize
the expected deformation mechanism of the core plate, the cover
plates were thus designed with a conservative index for the afore-
mentioned reasons, and a value of 18.3 was obtained for mF if the
a and e of L/1000 are postulated. Actually, the design of the cover
plates was dominated by local buckling of the cover plates, which
will be numerically studied in a subsequent section in this paper.
2.3. Testing setup
All the specimens were experimentally tested utilizing an MTS
loading system shown in Fig. 8, where the load and displacement
capacity are respectively 500 kN and ±75 mm. The two ends of
the specimens were clamped to the loading system, where the
top loading head was movable, and the bottom one was fixed.
Table 3
Ductility capacity and energy dissipation of specimens.
No. Specimen lmax
1 Common-BRB 26.1
2 FB-BRB-S5-E08-D0 29.9
3 FB-BRB-S2-E08-D0 29.9
4 FB-BRB-S2-E06-D3 33.6
Fig. 11. Failure mode of conventional buckling-restrained brace.
The tests were performed at room temperature with a quasi-
static speed under displacement control. Incremental cyclic load-
ing history shown in Fig. 9, was employed, and the increment
was set to 0.5% (average strain with related to the yielding portion
of the core plate) for all the cycles, which equals 3.35 mm in this
study. Seven displacement transducers shown in Fig. 8 were
employed in the tests to accurately capture the relative deforma-
tion of the yielding port of the core plate, and monitor the defor-
mation of the restraining plates. Thirty-four high-strength bolts
with a nominal diameter of 12 mm (M12, F10T) were employed.
The maximum internal forces of the bolts were calculated at the
instant of global buckling of the BRB by assuming a distributed
force between the core plate and the restraining plates. A load-
carrying capacity larger than the calculated internal force was
designed for the bolts. Since the expected ductility capacity of
the FB-BRBs is higher than conventional ones, and the postulation
of uniformly distributed forces can lead to a non-conservative
design, a safety index of 9.7 was employed for the design of the
bolts.
3. Experimental results
3.1. Hysteretic properties
The hysteretic stress-strain curves of the tested specimens are
presented in Fig. 10, where the stress data are the average values
lc CID (%) REp (kJ)
437 58.5 80.8
484 64.9 93.9
488 65.4 94.6
591 79.2 115.0
 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119 113

of the corresponding minimum cross section of the core plate, and
the strain data the average values of the yielding portion of the
core plate. The conventional BRB sustained a maximum average
tensile strain of 3.5% at a compressive half cycle as shown in
Fig. 10(a). Specimen FB-BRB-S5-E8-D0 achieved a maximum aver-
age tensile strain of 4.0% at a tensile half cycle as shown in Fig. 10
(b), and the curve is almost the same as that of Specimen FB-BRB-
S2-E8-D0 shown in Fig. 10(c), where experiment of the latter was
terminated soon after the load decreased during the last half cycle,
i.e., from 4% to +4.5%. Specimen FB-BRB-S2-E6-D3 achieved the
largest maximum average tensile strain of 4.5% in all the four spec-
imens, which failed at a tensile half cycle from 4.5% to +5%. All the
BRBs exhibited stable hysteretic properties, and neglectable distur-
bances due to frictional forces among the core plate, the restraining
plates and the filling plates, can be observed at the compressive
half cycles. The maximum disturbance for the load due to friction
under compressive loading is less than 2 kN, which is less than
1% of the peak load. The hysteretic curves of the FB-BRBs are
almost the same as that of the conventional BRB, where differences
occur at the last several half cycles.
3.2. Ductility and energy dissipation
Besides the cumulative ductility index, lc, there is also an alter-
native index, termed cumulative inelastic deformation (CID), which
can be employed to evaluate the cumulative ductility capacity. CID
is given by the following equation,
X
CID ¼ ep ð4Þ
where ep = average plastic strain of the BRB yielding portion. It is
interesting to note that lc is dependent on the yield strength of
the steel core plate, while CID is independent of the steel grade,
though both of the indices are non-dimensional. There are a number

Fig. 12. Failure mode of FB-BRB-S5-E8-D0.

Fig. 13. Failure mode of FB-BRB-S2-E8-D0.

114 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119

of studies aiming to investigate the cumulative ductility capacity of
BRBs in both building and bridge engineering, and it is important to
investigate which one is more suitable to fairly evaluate the cumu-
lative ductility capacity. It was found by one of the authors [21] that
CID is more suitable to evaluate the cumulative ductility capacity of
BRBs in steel arch bridges. The values of the maximum and cumu-
lative ductility related parameters of the specimens are presented
in Table 3. The accumulated plastic energy dissipation data are also
given in the table, which is defined as the enveloped area under the
load-displacement curve. The largest value of lmax for the FB-BRBs
is 33.6, which is larger than the best one reported in the literature
by 34%. The CID of FB-BRB-S2-E6-D3 is larger than 70% [23,25],
which can be deemed as a high performance BRB. The value of lc
for FB-BRB-S2-E6-D3 can reach a value of 591. Commonly a large
value of lmax leads to a small value of lc , and this value indicates
an excellent cumulative ductility capacity for a lmax of 33.6.

Fig. 14. Failure mode of FB-BRB-S2-E6-D3.

Fig. 15. Failure process of FB-BRB-S2-E6-D3.

 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
3.3. Failure modes
The conventional BRB failed due to out-of-plane local buckling
at the movable end as shown in Fig. 11. Apparent increase of the
average compressive stress at the last compressive half cycle can
be found in Fig. 10(a), which is probably due to a large frictional
force occurring induced by wearing of the core plate at the fixed
end shown in Fig. 11(b). The wearing problem was avoided suc-
cessfully through grinding the edges of the U-shaped openings of
the restraining plates illustrated in Fig. 2 for the other specimens.
This indicates that the sharp U-shaped edge of the restraining plate
can lead to the wearing of the core plate.
The failure mode of Specimen FB-BRB-S5-E8-D0 is presented in
Fig. 12, and half of the stoppers ruptured during the cyclic loading,
where they are all at the side close to the movable end. Since it is
not possible to observe what occurred inside the FB-BRB, the fail-
ure process can just be inferred based on the cyclic stress-strain
curve shown in Fig. 10(b) and the deformed shape of the specimen
shown in Fig. 12. The variation of the load-carrying capacity at the
last tensile half cycle shown in Fig. 10(b) is mainly due to contact
and rupture of the stoppers subsequently as the tensile deforma-
tion of the core plate increases. The probable failure process can
be divided into the following six steps as marked in Fig. 10(b):
(1) necking initiation close to the central stoppers;
(2) crack initiation at the edge of a central stopper;
(3) crack propagation through the whole cross-section of the
core plate;
(4) shear failure of the central stopper;
(5) shear failure of the stoppers close to the central stopper;
(6) shear failure of the stoppers close to the movable loading
head.
It can be found that this failure mode was induced by necking
initiation close to the central stoppers, and strain concentration
then occurs there, which eventually leaded to cracking at the edges
of the central stoppers. The in-plane buckling of the core plate
induced force concentration at the lower side central stopper as
shown in Fig. 12(b), which made a crack first occur there.
Fig. 16. Force transferred by stoppers.
Fig. 17. Details of FE model.
115
Failure mode of Specimen FB-BRB-S2-E8-D0 shown in Fig. 13 is
similar to that of Specimen FB-BRB-S5-E8-D0. The testing was ter-
minated when the load began to decrease, and it is thus possible to
observe the failure mechanism clearly. Necking first initiated at the
central segment of the core plate adjacent to the stoppers at right
side shown in Fig. 13(a), and cracks formed at the edges of the
stoppers due to strain concentration. Finally, the load-carrying
capacity decreased due to the crack propagation.
The details of Specimen FB-BRB-S2-E6-D3 were modified to
mitigate the strain concentration close to the stoppers by reducing
the width at the mid-length of each segment by 6 mm (2D), and
the allowable maximum elongation of the segment between two
adjacent stoppers, Es, was also changed from 8% to 6%. The appear-
ance of the core plate after testing is shown in Fig. 14, and the fail-
ure mechanism can be inferred based on the figure and the
hysteretic stress-strain curve. The load-displacement curve at the
last cycle of the specimen is plotted in Fig. 15, and the failure prob-
ably proceeded in the following sequence: (1) load-carrying capac-
ity decreases due to necking deformation development; (2)
increase of the load-carrying capacity for the contact of the central
pair of stoppers with the filling plates; (3) decrease of the load-
carrying capacity due to further necking development at the cen-
tral segment and/or cracking of the top pair of stoppers; (4) rup-
ture of the core plate at mid-length of the central segment due
to excessive necking development, followed by cracking and rup-
ture of the top stoppers shown in Fig. 14(a).
3.4. Performance of multiple stoppers
As illustrated in Fig. 16, the left segment had a smaller width of
60 mm, and the segment at right side had a width of 62 mm. When
the stoppers contact with the filling plates, the total force of the
Fig. 18. Hysteretic curves of test and numerical results for FB-BRB-S02-E6-D3.
116 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
right segment, Fc, i+1, will equal the sum of the total force of the left
segment, Fc, i, and the shear force of the stoppers, Fs. A parameter
termed stopper strength index, gs, defined as the ratio of the yield
shear strength of the stopper to the tensile maximum strength of
the core plate can be given by the following equation:
2bs tsy 2bs sy
gs ¼ ¼ ð5Þ
b c t rt b c rt
where bs = width at root of the stopper; t = thickness of the plate;
sy = yield shear stress of the material; bc = width of the minimum
cross section of the core plate (e.g., 60 mm as shown in Fig. 16);
rt = tensile strength of the material. In this study, gs approximately
equals 0.26 in this study. The yield shear stress is employed in Eq.
(5) to ensure the stoppers be elastic when the core plate achieves
its tensile strength. In addition, since the shear strength of the stop-
pers was designed to resist the difference of forces between two
adjacent segments of the core plate, gs can thus be less than 1.0.
Cracking of the stoppers was observed in all the three FB-BRBs,
and the stoppers failed to effectively control the maximum average
strain of the central segment due to the premature rupture for FB-
BRB-S5-E8-D0 and FB-BRB-S2-E8-D0. However, the stoppers in FB-
BRB-S2-E6-D3 successfully limited the maximum strain of the cen-
tral segment within the Es (6%) at the half cycle from 4.0% to
+4.5%, and failed due to cracking at the subsequent tensile half
cycle from 4.5 to +5.0%.
3.5. Discussions
The premature failure of the FB-BRBs is either due to cracking at
the location of strain concentration close to the stoppers, or rup-
ture of the central segment due to excessive development of the
necking deformation. Thus, the following 3 issues are of significant
importance to obtain high seismic performance FB-BRBs: (1) miti-
gation of strain concentration at the edges of the stoppers to
improve the ductile fracture-resistant properties under cyclic large
plastic straining; (2) proper target setting of the maximum allow-
able elongation for each segment, Es; (3) enough shear strength of
the stoppers to effectively control the deformation of the segments.
A value of 0.26 for gs is found to be not enough, and a higher value
less than 1.0 is required, where further experimental study should
be conducted. The experimental results also indicate that the dis-
tribution of shear forces for the same pair of stoppers can be
uneven due to in-plane buckling of the core plate under compres-
sion, and one side of the stopper is thus subjected to a larger por-
tion of shear force, which can lead to asymmetric failure mode of
the stoppers.
Fig. 19. Contact between stoppers and filling plates.
4. Numerical simulation
4.1. Numerical modeling
To investigate the failure mechanism of FB-BRBs, three-
dimensional solid model for Specimen FB-BRB-S02-E6-D3 was ana-
lyzed using Abaqus 6.10 [43], since the specimen succeeded in con-
trolling the maximum average strain of the central segment during
the experiment. Reduced integration elements, C3D8R, were
employed to improve the computational efficiency of the high
inelastic problems, where cyclic plasticity, contact and ductile frac-
ture are all simulated in the analysis. The mesh and boundary con-
ditions of the finite element (FE) model are shown in Fig. 17, where
one of the restraining plates is not shown to make the internal core
plate and filling plates visible. The left end of the core plate was
fixed, and the out-of-plane freedoms of the right end were con-
strained, where a displacement loading history the same as that
employed in the experiment was applied to the right end. The Cou-
lomb friction model was employed between the restraining com-
ponents and the core plate. The small value of 0.075 [25] was
employed considering the fact that there is unbonding material
between the restraining components and the core plate. However,
the frictional coefficient can increase under large deformation
loading due to thinning and slip of the unbonding material. The
bolts of the FB-BRB were simulated by coupling all the freedoms
of nodes at the surface of each hole with a reference node at the
center of each hole. A validated plasticity model with a combined
hardening rule [39], was utilized to describe the stress-strain
behaviors under cyclic large plastic straining. Meanwhile, a
micro-mechanics based cyclic ductile fracture model [37,38,40]
was also adopted to simulate the fracture behaviors and failure
mechanism of the specimen. All the parameters of the plasticity
and cyclic ductile fracture models were calibrated using corre-
sponding coupon test results of the core plate along the rolling
direction under monotonic tension. To investigate the mesh-
dependency effect, 3 mesh sizes for the central part of the core
plate were analyzed, and the mesh configuration of Fig. 17 was
finally determined through convergence study of the predicted
average stress-average strain curve and instant at crack initiation.
4.2. Comparison between experimental and numerical results
The predicted average stress-average strain curve was com-
pared with the experimental one in Fig. 18, where rupture of the
specimen occurred at a cycle earlier than the experimental results.
The hysteretic properties of the FB-BRB were well evaluated, and
the reversal of load-carrying capacity shown in Fig. 15 was also
successfully reproduced by the FE result. Contact between the
stoppers and the filling plates as illustrated in Fig. 19 occurred at
 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119 117

a cycle preceding crack initiation at the core plate. This indicates
that the stoppers of Specimen FB-BRB-S02-E6-D3 succeed in con-
trol the maximum elongation of the central segment of the core
plate at the instant. The failure process of the specimen obtained
from the numerical study is illustrated in Fig. 20, which can be
divided into the following stages: (1) load-carrying capacity
decreases due to necking deformation development at the central
segment; (2) increase of the load-carrying capacity for the contact
between the filling plates and the two pairs of stoppers; (3)
decrease of the load-carrying capacity due to further necking
development at the central segment and cracking of the core plate;
(4) cracking and rupture of core plate at mid-length of the central
segment due to excessive necking development; (5) failure of the
stoppers. The numerical result also shows that strain concentration
occurs at mid-length of the central segment under large plastic
straining as shown in Fig. 20(b).

Fig. 20. Failure process of numerical simulation.

118 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
Fig. 21. Local buckling failure of FB-BRB with a cover plate thickness of 6 mm.
4.3. Local buckling of cover plates
Local buckling of the cover plate can also occur in the flat cover
plates of the FB-RBRs due to the large width-to-thickness ratio. The
width-to-thickness ratio in this study was defined as ratio of the
bolt hole spacing to the cover plate thickness. To investigate the
optimal design value for the cover plate, a parametric analysis
was conducted, and premature local buckling can be effectively
avoided when the cover plate thickness is larger than 7 mm, corre-
sponding to a width-to-thickness ratio of 20.6. In the analysis, only
the thickness of the cover plate was changed. An FB-BRB failed due
to premature local buckling is illustrated in Fig. 21, where the
cover plate thickness is 6 mm, corresponding to a width-to-
thickness ratio of 24. This indicates that a width-to-thickness ratio
no larger than 20.6 can effectively prevent premature local buck-
ling of the cover plates. In the experimental studies, a width-to-
thickness ratio of 9 was employed for the cover plates, which is
stiff enough to avoid the premature local buckling. Meanwhile, a
T-shaped cross section can be employed for the cover plate to
improve the local buckling strength and reduce its thickness to
achieve an economical design. Besides, the values of the mF for
the FB-BRBs with cover plate thickness of 6 mm and 7 mm are
respectively equal to 3.3 and 4.2, indicating that premature global
buckling can be effectively avoided according to Eq. (3).
5. Conclusions
In this study, a new type of light-weight all-steel fish-bone
shaped buckling-restrained braces (FB-BRBs) were proposed, aim-
ing to achieve high seismic performances such as large maximum
ductility, cumulative ductility and energy dissipation capacities.
The core plate of the FB-BRB was divided into several segments
by multiple pairs of stoppers. The characteristic of the FB-BRBs is
that the post-necking straining capacity of a material, e.g., steel,
can be more effectively employed through generating several
necks in several segments. The multiple stoppers were employed
to control the development of the maximum necking deformation
of each segment, and lead to necking and restraining of the seg-
ments subsequently. Based on the aforementioned mechanism,
high seismic performance of the FB-BRBs can be obtained.
Four scaled BRBs with different details were experimentally
studied, where a conventional BRB was utilized to compare with
the other three FB-BRBs. Experimental results show that the FB-
BRBs can have a favorable seismic performance than the conven-
tional BRB. Strain concentration close to the stoppers, excessive
necking deformation, lack of shear strength of the stoppers and
premature cracking at the edges of the stoppers were observed
in the test results of the FB-BRBs, indicating that there is a potential
to further improve their seismic performance. Numerical fracture
simulation using validated cyclic plasticity and ductile fracture
models was also conducted, which well evaluated the hysteretic
average stress-average strain curve and explicitly showed the fail-
ure mechanism. Based on numerical simulation results, it is found
that a width-to-thickness ratio no larger than 20.6 can avoid pre-
mature local buckling of the cover plates, though further experi-
mental validation is required. Meanwhile, more experimental
and numerical studies are required to study the effects of main
parameters, such as number of stoppers, allowable elongation
and shear strength of the stoppers, on seismic performance of
the newly proposed FB-BRBs. In addition, the lateral thrust forces
[28,44,45] due to buckling of the core plate should be further stud-
ied to facilitate design of the cover plates and the bolts.
 L.-J. Jia et al. / Engineering Structures 138 (2017) 105–119
Acknowledgements
The financial support from the Advanced Research Center for
Natural Disaster Risk Reduction, Meijo University, which is sup-
ported by Ministry of Education, Culture, Sports, Science, and Tech-
nology (MEXT), Japan, is greatly appreciated. This study is also
partially supported by National Nature Science Foundation of
China (51508401).
References
[1] Kimura K, Yoshizaki K, Takeda T. Tests on braces encased by mortar in-filled
steel tubes. In: Summaries of technical papers of annual meeting. Architectural
Institute of Japan; 1976. p. 1041–2.
[2] Wakabayashi M, Nakamura T, Kashibara A, Morizono T, Yokoyama H.
Experimental study of elasto-plastic properties of precast concrete wall
panels with built-in insulating braces. In: Summaries of technical papers of
annual meeting. Architectural Institute of Japan; 1973. p. 1041–4.
[3] Yoshino T, Karino Y. Experimental study on shear wall with braces (Part 2). In:
Summaries of technical papers of annual meeting. Architectural Institute of
Japan; 1971. p. 403–4.
[4] Mochizuki S, Murata Y, Andou N, Takahashi S. Experimental study on buckling
of unbonded braces under axial forces: Part 2 experimental results and
discussions. In: Summaries of technical papers of annual
meeting. Architectural Institute of Japan; 1979. p. 1625–6.
[5] Mahin SA. Lessons from damage to steel buildings during the Northridge
earthquake. Eng Struct 1998;20:261–70.
[6] FEMA-350. Recommended seismic design criteria for new steel moment-frame
buildings. Washington, D.C.: Federal Emergency Management Agency; 2000.
[7] AIJ. Fracture in steel structures during a severe earthquake. Tokyo:
Architectural Institute of Japan; 1995.
[8] Tsai K, Hsiao CP, Bruneau M. Overview of building damages in 921 Chi-Chi
earthquake. Earthq Eng Eng Seismol 2000;2:93–108.
[9] Zhao B, Taucer F, Rossetto T. Field investigation on the performance of building
structures during the 12 May 2008 Wenchuan earthquake in China. Eng Struct
2009;31:1707–23.
[10] Okazaki T, Lignos DG, Midorikawa M, Ricles JM, Love J. Damage to steel
buildings observed after the 2011 Tohoku-Oki earthquake. Earthq Spectra
2013;29:S219–43.
[11] Fahnestock LA, Sause R, Ricles JM, Lu L-W. Ductility demands on buckling-
restrained braced frames under earthquake loading. Earthq Eng Eng Vibr
2003;2:255–68.
[12] Fahnestock LA, Ricles JM, Sause R. Experimental evaluation of a large-scale
buckling-restrained braced frame. J Struct Eng (ASCE) 2007;133:1205–14.
[13] Black C, Makris N, Aiken I. Component testing, seismic evaluation and
characterization of buckling-restrained braces. J Struct Eng (ASCE)
2004;130:880–94.
[14] Chou C-C, Chen S-Y. Subassemblage tests and finite element analyses of
sandwiched buckling-restrained braces. Eng Struct 2010;32:2108–21.
[15] Piedrafita D, Cahis X, Simon E, Comas J. A new perforated core buckling
restrained brace. Eng Struct 2015;85:118–26.
[16] Wang C-L, Usami T, Funayama J, Imase F. Low-cycle fatigue testing of extruded
aluminium alloy buckling-restrained braces. Eng Struct 2013;46:294–301.
[17] Zhao J, Wu B, Li W, Ou J. Local buckling behavior of steel angle core members
in buckling-restrained braces: cyclic tests, theoretical analysis, and design
recommendations. Eng Struct 2014;66:129–45.
[18] Razavi Tabatabaei SA, Mirghaderi SR, Hosseini A. Experimental and numerical
developing of reduced length buckling-restrained braces. Eng Struct
2014;77:143–60.
[19] Piedrafita D, Cahis X, Simon E, Comas J. A new modular buckling restrained
brace for seismic resistant buildings. Eng Struct 2013;56:1967–75.
[20] AISC. Seismic provisions for structural steel buildings, ANSI/AISC 341-10.
Chicago, IL: American Institute for Steel Construction; 2010.
119
[21] Chen X, Ge HB, Usami T. Seismic demand of buckling-restrained braces
installed in steel arch bridges under repeated earthquakes. J Earthq Tsunami
2011;5:119–50.
[22] Motosaka M. Lessons of the 2011 Great East Japan Earthquake focused on
characteristics of ground motions and building damage. In: Proceedings of the
international symposium on engineering lessons learned from the 2011 Great
East Japan Earthquake. Tokyo, Japan; 2011. p. 1–4.
[23] Usami T. Developing high-performance damage control seismic dampers. In:
10th symposium on ductile design method for bridges: JSCE; 2007. p. 11–22.
[24] Wang C-L, Usami T, Funayama J. Improving low-cycle fatigue performance of
high-performance buckling-restrained braces by toe-finished method. J Earthq
Eng 2012;16:1248–68.
[25] Wang C-L, Usami T, Funayama J. Evaluating the influence of stoppers on the
low-cycle fatigue properties of high-performance buckling-restrained braces.
Eng Struct 2012;41:167–76.
[26] Ma N, Wu B, Zhao J, Li H, Ou J, Yang W. Full scale test of all-steel buckling
restrained braces. In: Proceedings of 14th world conference on earthquake
engineering. Beijing, China; 2008. p. Paper No. 208.
[27] Hoveidae N, Rafezy B. Overall buckling behavior of all-steel buckling
restrained braces. J Constr Steel Res 2012;79:151–8.
[28] Genna F, Gelfi P. Analysis of the lateral thrust in bolted steel buckling-
restrained braces. I: experimental and numerical results. J Struct Eng (ASCE)
2012;138:1231–43.
[29] Judd JP, Marinovic I, Eatherton MR, Hyder C, Phillips AR, Tola Tola A, et al.
Cyclic tests of all-steel web-restrained buckling-restrained brace
subassemblages. J Constr Steel Res 2016;125:164–72.
[30] Hoveidae N, Tremblay R, Rafezy B, Davaran A. Numerical investigation of
seismic behavior of short-core all-steel buckling restrained braces. J Constr
Steel Res 2015;114:89–99.
[31] Hosseinzadeh S, Mohebi B. Seismic evaluation of all-steel buckling restrained
braces using finite element analysis. J Constr Steel Res 2016;119:76–84.
[32] Corte GD, D’Aniello M, Landolfo R. Field testing of all-steel buckling-restrained
braces applied to a damaged reinforced concrete building. J Struct Eng (ASCE)
2015;141:D4014004.
[33] Usami T, Wang C, Funayama J. Low-cycle fatigue tests of a type of buckling
restrained braces. Procedia Eng 2011;14:956–64.
[34] Eryasßar ME, Topkaya C. An experimental study on steel-encased buckling-
restrained brace hysteretic dampers. Earthq Eng Struct Dynam
2010;39:561–81.
[35] Zhao J, Wu B, Ou J. A novel type of angle steel buckling-restrained brace: cyclic
behavior and failure mechanism. Earthq Eng Struct Dynam
2011;40:1083–102.
[36] Usami T, Sato T. Low-cycle fatigue tests and verification method for a steel
buckling-restrained brace. J Struct Eng (JSCE) 2010;56A:486–98.
[37] Jia L-J, Koyama T, Kuwamura H. Experimental and numerical study of
postbuckling ductile fracture of heat-treated SHS stub columns. J Struct Eng
(ASCE) 2014;140:04014044.
[38] Jia L-J, Kuwamura H. Ductile fracture simulation of structural steels under
monotonic tension. J Struct Eng (ASCE) 2014;140:04013115.
[39] Jia L-J, Kuwamura H. Prediction of cyclic behaviors of mild steel at large plastic
strain using coupon test results. J Struct Eng (ASCE) 2014;140:04013056.
[40] Jia L-J, Kuwamura H. Ductile fracture model for structural steel under cyclic
large strain loading. J Constr Steel Res 2015;106:110–21.
[41] Usami T, Sato T, Kasai A. Developing high-performance buckling-restrained
braces. J Struct Eng (JSCE) 2009;55A:719–29.
[42] Usami T, GE HB, Kasai A. Overall buckling prevention condition of buckling-
restrained braces as a structural control damper. In: Proceedings of 14th world
conference on earthquake engineering. Beijing, China; 2008. p. Paper No. 3950.
[43] ABAQUS. ABAQUS standard manual (Version 6.10). Hibbitt. Pawtucket (RI,
USA): Karlsson & Sorensen Inc.; 2010.
[44] D’Aniello M, Corte GD, Landolfo R. Finite element modelling and analysis of
‘‘all-steel” dismountable buckling restrained braces. Open Constr Build
Technol J 2014;8:216–26.
[45] Genna F, Gelfi P. Analysis of the lateral thrust in bolted steel buckling-
restrained braces. II: engineering analytical estimates. J Struct Eng (ASCE)
2012;138:1244–54.