Engineering Structures 77 (2014) 143–160

Contents lists available at ScienceDirect

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

Experimental and numerical developing of reduced length

buckling-restrained braces
Seyyed Ali Razavi Tabatabaei ⇑, Seyyed Rasoul Mirghaderi, Abdollah Hosseini
School of Civil Engineering, College of Engineering, University of Tehran, Iran

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

Article history: Buckling-restrained braced frames (BRBFs) have been widely used as an efficient seismic load resisting
Received 18 February 2013 system in recent years mostly due to their symmetric and stable hysteretic behavior and significant
Revised 11 July 2014 energy dissipation capacity. However, buckling-restrained braces (BRBs) are heavier and more expensive
Accepted 25 July 2014
in comparison to other concentric bracing systems. In order to facilitate the use of BRBs, the idea of reduc-
Available online 23 August 2014
ing the length of the core and the encasings which can result in lighter and more replaceable BRBs is pro-
posed and experimentally investigated in this paper. Two relatively similar all-steel reduced length BRBs
Keywords:
(RLBRBs) are designed, detailed and constructed using a special debonding and stopper mechanism. The
Buckling-restrained brace
Low-cycle fatigue
design and construction procedure is accomplished by paying special attention to low-cycle fatigue (LCF).
Test The specimens were tested under the quasi static loading protocol, and withstood high axial strains of 4–
Finite element analysis 5% without any global or local failure. The hysteretic responses of the specimens were stable and sym-
metric. Moreover, numerical models were developed and nonlinear cyclic analyses were performed to
provide better insight into the core and encasing performance as well as application of the RLBRB in
the brace configuration.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction
1.1. Background and research motivation
In comparison to other lateral resisting systems, buckling-
restrained braced frames (BRBFs) have both high stiffness and duc-
tility. To reach an optimum seismic design in which most of the
energy dissipating potential is utilized, different elements with
various strengths and stiffnesses are required. The strength and
stiffness of a buckling-restrained brace (BRB) can be increased
independently by selecting a larger cross section and reducing
the BRB length. Reducing the length of BRB and placing it in series
with an elastic brace has several advantages. The length of BRB
directly effects the amount of material used in BRB, including core,
encasing, filler, and the debonding material, as well as the capacity
of the facilities utilized in the production and handling of the BRB.
Hence, producing BRBs can be assumed as a function of their
length. Although reducing the length of BRB might increase its
stiffness, some studies comparing buildings designed with reduced
⇑ Corresponding author. Address: Structural Laboratory, School of Civil Engineer-
ing, College of Engineering, University of Tehran, 16 Azar St., Enqelab Sq., Tehran,
Iran. Tel.: +98 21 227 334 82; fax: +98 21 66403808.
E-mail addresses: arazavi@ut.ac.ir (S.A. Razavi Tabatabaei), rmirghaderi@ut.ac.ir
(S.R. Mirghaderi), hosseiniaby@ut.ac.ir (A. Hosseini).
length BRBs (RLBRBs) and conventional BRB show that reducing
the length of BRB does not change the required core cross section
majorly [1,2]. In such cases a shorter BRB could naturally be more
economical than a longer one. Moreover, the replaceability and
handling of a fuse-type BRB located in a limited space of the frame
after an earthquake is simpler than a full-length heavy BRB
occupying the whole frame. From the analytical point of view,
the comparison of the performance of structures equipped with
conventional BRBs and RLBRBs shows that the RLBRB systems have
better seismic performance in terms of uniform plasticity through
the height of the structure as well as less residual and maximum
drifts [1–3].
The core strain demand of BRB (ec) is dependent upon the max-
imum story drift (Dmax), the BRB length (L), the ratio of yielding
core to the total length of brace (Lc/L) (a), and the BRB angle (h):
Dmax cos h
ec ¼ ð1Þ
aL
The strains of the core in a long BRB are less than those of a
RLBRB. Naturally, the BRBs used in very long spans are hardly
capable of entering the plastic zone, and therefore, the ductility
capacity of the BRB may not be used efficiently. In such cases,
reducing the BRB length can effectively provide the core plasticity
and energy dissipating demands of the brace. For instance,
considering a long span with 10 m width and 3 m height, a

http://dx.doi.org/10.1016/j.engstruct.2014.07.034
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
144 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160

Nomenclature

A brace cross sectional area h BRB horizontal angle

Ac core cross sectional area l displacement ductility of BRB
A(x) core cross sectional area as a function of distance (x) Dby brace axial displacement corresponding to yielding
b total span of the frame Dbm brace axial displacement corresponding to design story
c fatigue ductility exponent drift
E Young’s modulus Dm frame design story drift
E0 tangent modulus Dmax maximum story drift
E(D) energy dissipated by a BRB through a complete cycle Drby RLBRB axial displacement corresponding to yielding
FDI fatigue damage index Drbm RLBRB axial displacement corresponding to design story
FS factor of safety drift
h height of story Dpl RLBRB plastic axial displacement
K axial stiffness of BRB De strain amplitude of the core
L total length of brace ec average axial strain of the core
Lb bolt spacing e0 f fatigue ductility coefficient
Lc length of core eL local axial strain of the core along the length
Lw higher mode buckling wave length eT local strain of the core along the thickness
Nf number of cycles to fatigue failure eV volumetric strain
ni number of cycles eW local strain of the core along the width
Pcr critical buckling load of the brace g ratio of cumulative plastic displacement to yield
Py yielding load of the brace displacement
a the ratio of yielding core to the total length of brace x strain hardening adjustment factor of BRB
(Lc/L)
b compression strength adjustment factor of BRB
d total deformation of the brace

maximum story drift of 2%, and yielding ratio of 80% corresponding
to conventional BRBs, the core strain demand will be 0.68% which
is corresponding to limited BRB ductility of 5.6 (for ST 37-2 steel).
However, if the yielding length is reduced to 30% in the form of
RLBRB, the strain demand easily increases to 1.81% corresponding
to a reasonable BRB ductility of 15 which is a common practical
value according to BRB literature [4].
In addition, from the architectural point of view, two RLBRBs
can be placed in a single frame in an X-pattern configuration which
is not applicable in normal BRBs.
The steel material is capable of resisting axial strains up to 20–
30%; however, in most of BRBs the strain amplitudes are limited to
1–2% [5]. Considering this advantage of steel material, in this study
a detailing for an all-steel RLBRB, designed to withstand axial
strain of about 4.0–5.0% without fatigue failure, is presented. The
all-steel BRB can be lighter than conventional concrete filled BRBs,
and have higher replaceability potential. Besides, the construction
of all-steel BRBs can take less time due to the elimination of con-
crete curing.
1.2. Relevant prior researches
As mentioned in previous section, an improvement to the
detailing of BRBs is the presentation of all-steel BRBs which sand-
wich the BRB core. This solution helps to replace the damaged core
easily using detachable encasings which might be used during sev-
eral earthquakes. Several detailings have been proposed for the
development of all-steel sandwiched BRBs [6–10]. D’Aniello et al.
tested two detachable all-steel BRBs consisting of a rectangular
steel plate and a restraining steel sleeve which formed two omega
shapes bolted together. The all-steel BRB showed symmetric
response up to the story drift range of ±1.5% which corresponded
to core strain of 2.5% [6]. An all-steel bolted BRB was tested by
Mazzolani et al. to upgrade a non-ductile two-story reinforced con-
crete (RC) structures. The buckling-restraining action was given by
two rectangular steel tubes. The two restraining tubes were joined
together by means of bolted stiffened elements. The BRB showed a
good ductility of 15, though some additional improvement was
required at the core to gusset plate connection [7]. In a recent
attempt to develop extra light weight BRBs, Dusicka et al. proposed
and investigated the behavior of a BRB made up of aluminum core
and bundled glass fiber-reinforced polymer pultruded tubes for the
buckling restraint which reduced the weight of the BRB remark-
ably [11].
Chou and Chen [12] presented a specific type of sandwiched
BRBs which eliminated the use of unbonded material. The enca-
sings were composed of a welded steel channel to a flat plate
and finally filled with 48–58 MPa concrete. The encasings were
connected together using A490 bolts. The authors proposed design
guidelines for providing the global stability of the whole brace, as
well as the local rigidity of the encasings, and the axial demand of
the connecting bolts based on the higher mode buckling wave-
length. The results of the tests showed that in case of providing
sufficient flexural rigidity for the encasings, the proposed BRBs
exhibit stable hysteretic response to a lateral design story drift of
2.4% and proper cumulative plastic ductility [12]. Moreover, a
frame with the same BRB was used and safely tested to a lateral
design story drift of 2.5% [13]. The stopping mechanism which is
responsible for constraining the motion of the encasings to the core
center, was provided by an outward projection in the core where
the tensile fracture initiated. Since concrete is assumed to contrib-
ute merely in providing global stability and can be replaced by
steel, eliminating concrete can be advantageous to reduce the
weight of BRB and expedite the production process. The specimens
tested by Chou and Chen experienced a maximum axial strain of
2.6%. If the length of BRB decreases the plastic strain demand on
the BRB increases and consequently the normal force exerting on
the encasings and the contact points of the core and encasing rise
leading to amplification in the friction force as observed by Tremb-
lay et al. [5]. Higher friction force is detrimental to the fatigue life
of the BRB.
Most of the BRB members developed up to now have a long
encasing which extends through the whole brace. The ratio of
yielding core to the total length of brace (Lc/L) for common BRBs
normally varies from 0.6 to 0.8 [5,14,15] and the axial strain ampli-
tude is less than 3%. Some limited studies have assessed BRBs with
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
small yielding segments with Lc/L of 0.2–0.4 [2,3,5,16–19]. How-
ever, the behavior of short BRBs which are advantageous to the
BRB industry has not been studied in detail yet. Developing BRBs
with strain amplitudes of higher than 3% requires modifications
to the conventional debonding and stopping mechanism.
1.3. Research plan
In this paper the provisions considered in the designing and
detailing of RLBRB which can sustain axial strains of 4–5% as well
as the construction procedure are described. The experimental pro-
gram for testing two RLBRB specimens is described and the results
are presented. The performance of RLBRB specimen is evaluated by
conducting finite element analysis. The numerical studies end with
models in which the RLBRB is applied in the whole brace configu-
ration to investigate the effect of the location of RLBRB in the brace
and to make a comparison between the energy dissipation of the
RLBRBs and conventional BRBs. It should be noted that the focus
of this article is on the short yielding part of the brace or RLBRB.
2. Introducing the reduced length BRB (RLBRB)
A typical application of the RLBRB is depicted in Fig. 1. When the
seismic load is applied to a RLBRB frame, it passes through the
yielding core of RLBRB and then enters the elastic part via a serial
path. The elastic part is designed not to yield and not to buckle
under the design earthquake while the fuse (RLBRB) is functioning.
In other words, neither the second part yields in tension nor does
the whole brace buckle in compression due to the maximum
expected load in the core (considering a safety factor). Using cross
sections with high radii of gyration for the non-yielding part is rec-
ommended to achieve desirable levels of buckling load. Placing
RLBRB at one end of the brace (Fig. 1) helps to achieve a higher crit-
ical buckling capacity [18]. For more information about the seismic
design procedure of RLBRB structures refer to Razavi et al. [2].
The development of RLBRB includes several designing steps
such as providing global stability for the whole brace and restrain-
ing the local buckling of the core. Besides that, the low-cycle fati-
gue (LCF) failure and the energy dissipating capacity of RLBRB
are of great importance, which will be discussed in the following
sections together with the design and detailing of the specimens.
2.1. Low-cycle fatigue (LCF)
BRBs in the form of structural fuses can be used as sacrificial
members dissipating the seismic energy, while preserving the
integrity of other main components [4]. The idea of fuse-type
Fig. 1. Application of RLBRB in frame.
145
LC1
Py1 Py1
δ1 Ordinary BRB δ1
LC2
Py2 Py2
Elastic Part
δ2 δ2
RLBRB
δ1=δ2 ,Py1=Py2 , Py2<Pcr2/FS, εc2>εc1
Fig. 2. Comparison of normal BRB with RLBRB.
BRB is most efficient when the RLBRB has the least possible length.
However, by decreasing the core length, the plastic strains, which
are inversely dependent on the yielding length (Eq. (1)) increase
(Fig. 2). This might raise concerns about low-cycle fatigue (LCF)
failure. Nonetheless, BRBs with strains higher than 2% have also
exhibited appropriate performance [10,15,17,20,21].
From the metallurgical point of view, the fatigue life of steel can
be evaluated by the Coffin–Manson equation based on small axial
tests [22]:
De
¼ e0f ð2Nf ÞC ð2Þ
2
where De is the strain amplitude of the specimen, e0 f is the fatigue
ductility coefficient, Nf is the number of cycles to failure, and c is
fatigue ductility exponent.
This equation defines the number of cycles (Nf) with constant
uniaxial strain amplitude (De) to occurrence of LCF failure. The
state of strain in BRB core is affected by axial actions as well as
in-plane and out-of-plane actions, however, Coffin–Manson type
equations for different BRBs have been developed by many
researchers [23–28]. Generally, the LCF studies on BRBs show that
fatigue capacity depends on various factors such as; the shape of
the core, the stiffness of the buckling-restraining mechanism, the
width to thickness ratio of the axial member, and the friction
between the core and the restraining mechanism [2,14,29]. More-
over, it was found that the variation of steel grade has a minimal
effect on the fatigue life of BRBs [30,31]. The presence of any dis-
continuities, geometric changes, tack welds, and attachments in
core segment which is identified as a protected zone in AISC
341-10 [32] severely degrade the LCF life [17,28,33,34]. Using sec-
tions other than flat plates, like cruciform sections for the core,
decreases the fatigue life [23,29]. Attention must be paid to these
findings in developing the detailing of the RLBRB.
According to Razavi et al. [16] among the several BRB fatigue
equations developed so far, the equation developed by Nakamura
146 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
et al. [23] is rather conservative for the range of strains below 7%.
Hence, the following equation will be used in this research for esti-
mating the fatigue life of the RLBRB [23]:
De ¼ 0:2048N 0:490
f ð3Þ
Since the strain amplitudes exerted on a BRB are variable, the
cumulative fatigue damage index (FDI) is used. FDI can be evalu-
ated based on the Miner equation [35]:
X ni
FDI ¼ ð4Þ
Nfi
where ni is the number of cycles and Nfi is fatigue capacity in terms
of the number of cycles to failure at each constant amplitude (i).
Failure occurs when the FDI equals 1. Accordingly, in this study,
the core length is estimated in such a way that the FDI falls below
1 when the RLBRB is subjected to test loading protocol.
2.2. Energy dissipating capacity
Reducing the BRB length causes the energy dissipation to be
locally concentrated in a limited length of the brace. Although
the volume of steel material participating in energy dissipation
through plastic strains decreases in this case, the energy dissipa-
tion capacity of the brace is not affected. By taking the kinematic
hardening rule into account, the energy dissipated by a BRB
through a complete cycle is [16]:
4Py
EðDÞ ¼ ðE  E0 ÞðDbm  Dby Þ ð5Þ
E length for the core of RLBRB to withstand the AISC 341-10 loading
where E and E0 are the Young’s and tangent modulus, Py and Dby are
the brace yield force and displacement, respectively, and Dbm is the
brace axial displacement corresponding to the design story drift. Eq.
(5) shows that basically at the same story drifts, the dissipated
energy by BRBs with the same steel grades and cross section areas
does not depend on the length of BRB. Nevertheless, if parameters
such as isotropic hardening, the friction force between the core
and encasing, and the non-uniform distribution of strain through
the core length are to be considered, the judgment concerning
energy dissipation capacity will not be easy. An experimental study
showed that as the BRB length decreases, the energy dissipation
capacity increases [36]. In Section 5.3 a numerical comparison
between the energy dissipation of RLBRB and conventional BRB is
presented, which confirms the finding of the mentioned experimen-
tal study.
2.3. Design and detailing of the reduced length BRB (RLBRB)
Two RLBRB specimens are designed to be tested. The two spec-
imens are identical in all details except the loading protocol and
the encasing stiffeners which will be discussed later in the paper.
The steel sections used for constructing all parts of specimen are
among the steel sections locally available. A bolt connection is used
to connect the two encasings together. This helps to disassemble
the specimen after the test to closely investigate the core. Further-
more, the encasing and fillers can be reused for other specimens
after replacing the core. Due to the concern for LCF, a novel
detailing for the stopping mechanism and a special scheme for
debonding are implemented. The connections of RLBRB to the adja-
cent elements (setup end plates in the case of this study) are
designed to be rigid. Although hinged connection has shown better
seismic performance in BRB [37,38], the rigid connection facilitates
a shorter length for the RLBRB in the whole brace. The details of
different parts of the RLBRB are explained in the following sections.
2.3.1. Core
A maximum tensile capacity of 500 kN is considered for the test
setup. The cross section of the core is defined based on the setup
capacity. A flat shape is selected for the core due to its better fati-
gue behavior in comparison with cruciform sections as noted
before. Hence, a 80  10 mm plate made of ST 37-2 (DIN 17100)
with nominal yield strength of 235 MPa and ultimate strength of
365 MPa is considered for the core. Two tensile tests were per-
formed on two machined coupons of core plate according to ASTM
A 270 [39]. The coupons were cut parallel to the longitudinal direc-
tion of the RLBRB core. An average yield strength of 248 MPa was
obtained for the coupons using the offset method of 0.2% strain,
while the average ultimate strength was 402 MPa leading to a
strain hardening factor of 62%. The average elongation of coupon
at necking was 25%. By multiplying the yield strength of core plate
by its cross section, the brace yielding force would be 198 kN.
Other parts of the specimen are designed according to the maxi-
mum strength developed in the specimen.
It is important to make the core segment long enough to limit
strain amplitudes and consequently prevent LCF failure. The length
of the core is defined so that the specimen can sustain two stan-
dard loading protocols, i.e.: AISC 341-10 [32] and ASCE 7-10 [40].
AISC 341-10 [32] specifies a stepwise increasing loading protocol
for qualifying cyclic tests on BRB specimens (section K3). More-
over, ASCE 7-10 [40] defines five cycles with constant displace-
ment amplitudes corresponding to the maximum considered
earthquake (MCE) for testing seismic energy damping devices.
Using the equation of Nakamura et al. (Eq. (3)) and Miner rule
(Eq. (4)), and assuming a design story drift ratio of 1.5% for a typical
3 m height, 5 m width bay with a chevron pattern, the required
protocol will be estimated 1100 mm. Supposing the story drift at
MCE level to be 1.5 times the design story drift, the RLBRB with this
length is capable of withstanding 7 full cycles at MCE which satis-
fies the ASCE 7-10 qualifying test for damping devices. Further
details about calculating the required length of BRB is described
by Razavi et al. [2].
The end of core is stiffened by two trapezoidal 200  80  10
plates (Fig. 3). Two 300  300  30 end plates are used to transfer
the load to the core. Complete joint penetration (CJP) groove welds
connect the core and stiffeners to the end plates (Fig. 3). End plates
are connected to adjacent connection with 8 high strength Euro-
pean grade 8.8 bolts (BS 3692) 22 mm in diameter (Fig. 3). The
thickness of end plates and the bolt patterns are selected in such
a way as to make the plate stiff enough in an effort to prevent
out-of-plane deformation. As such, the total length of the
R=200
CJP Groove Weld Core Plate
(TH=10 mm)
mm
160
0.9Fy
80
Fu
3
1
Copper Plate
160 120
(TH=2 mm)
1100 120 160
(a) Top view
8 M22 (8.8) End Plate
(TH=30 mm)
CJP Groove Weld Fillet Weld
80
Core Stiffener
120 80 (TH=10 mm)
1660
(b) Side view
Fig. 3. Detailing of the core of RLBRB (Unit: mm).
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
specimen, considering the core and transition zones, would be
1660 mm.
To reduce stress concentration, the width of the core section is
gradually increased with a rate of 1:3 with 200 mm radius in the
transition zones so that the core stress is reduced from failure
strength (Fu) in the core to 0.9Fy at the beginning of end stiffeners
(Fig. 3(a)).
2.3.2. Encasings
The encasings are responsible for preventing global buckling of
the core as well as limiting local buckling. These members are
designed for the forces exerted by the core at the contact points
of the core and encasing. Most similar to the concept of the sand-
wiched BRBs [5,9,12], the encasing is a 180  10 mm plate from the
same steel as the core and is reinforced by an A-ST 52 (DIN 17135)
rectangular hollow structural section (HSS) of 60  60  5 for out-
of-plane action. The nominal yield strength and ultimate strength
for the ST 52 HSS section is 355 MPa and 510 MPa, respectively.
The rectangular HSS is intermittently welded to the encasing to
provide the expected shear demand between the two sections.
In normal BRBs, the encasing is also checked to satisfy the
Watanabe criteria [41] for preventing global elastic buckling. How-
ever, in the case of RLBRB, the brace is composed of the short BRB
and a strong elastic member. Therefore, the flexural stiffness of the
encasing should be checked so that the whole brace does not
buckle when the core reaches its maximum load. Assuming an
elastic part made of circular HSS of 150  10, the global buckling
load of the whole brace to the plastic load of the core is 3.52 and
3.00 for the cases of combined and individual behavior of the
two encasings, respectively.
The two encasings are connected together through two
50  14 filler plates with hand tightened European grade 8.8
(BS 3692) bolts. The spacing of the bolts should be close to the
length of higher mode buckling waves in order to effectively
restrain the local bucklings. D’Aniello et al. [42] proposed expres-
sions to calculate the number and length of buckling wave based
on equaling the higher mode buckling load of the core to the axial
core strength. Using these expressions the number and length of
buckling waves (Lw) of the RLBRB would be 27 and 44 mm,
respectively. Accordingly, the bolts spacing (Lb) is selected
50 mm center to center through the total length of the encasing.
Chou and Chen [12] selected the Lb/Lw ratios varying from 0.3 to
2.2 in the test specimens and recommended a ratio less than 2.1
for design purposes.
A minimum clearance is needed to allow for expansion of the
core due to the Poisson effect to avoid the excessive friction
between the core and the encasing. However, if adequate clearance
is provided between the core and encasing, the normal force acting
on the encasing rises as the clearance increases. This fact is in
accordance with findings of Genna et al. [43] who performed para-
metric study on the gap sizes of all-steel BRBs. Larger normal force
results in higher frictional response of BRB and requires providing
higher stiffness and strength for encasing. The increase in gap –
which might naturally take place due to the elastic deformation
of the encasing during the test – can exert normal forces greater
than anticipated in the bolted BRBs and as a result sufficient con-
servativeness should be considered in the design procedure. In this
study, the fillers are designed to provide a 1 mm gap through thick-
ness and a 3 mm gap through width of the core plate to allow for
expansion in compression cycles.
Chou et al. [12] proposed an equation to calculate normal force
of the buckled core against the encasing based on the axial force in
the core plate, the core to encasing gap, and the buckling
wavelength. Later Genna et al. [44] estimated the normal force of
the core to encasing more precisely based on experimental and
numerical results. The equation presented finally by Genna et al.
147
encompasses several effects such as core plastic hardening, cyclic
loading, and friction. Chou and Chen selected bolt capacity to
demand ratios of 1.5, 2.6, 3, and 7.4 in the test specimens and
suggested a ratio of greater than 1.5 based on analytical results
and findings of the test program with strain levels up to 2.6%.
According to the equation suggested by Chou and Chen the exerted
normal forces to each bolt of RLBRB would be 35 kN which is cor-
responding to a capacity to demand ratio of 2.5 for 10 mm 8.8 bolt.
However, according to the equation suggested by Genna et al.
(wavelength reduction factor a = 0.5 for the final cycle, and a fric-
tion coefficient of l = 0.15) the maximum tensile force initiated in
bolts would be 58 kN which is corresponding to a capacity to
demand ratio of 1.5 The normal force calculated from the latter
study is selected since it considers cyclic loading and friction effect.
Such a normal force imposes two sorts of local bending actions on
the encasing, i.e. longitudinal and transverse. The longitudinal
bending demand is evaluated by assuming a point load in the mid-
dle of a simply supported beam with a span equal to bolt spacing,
while the transverse bending demand is estimated by imposing a
uniform line load to a simply supported beam with span equal to
transverse bolt row distance which varies from 140 mm in the
middle of encasing to 238 mm at the end of encasing. The longitu-
dinal bending demand of encasing resulted from the normal force
is 0.73 kN-m which is corresponding to a demand to capacity ratio
of 12. The critical transverse bending is generated at the end of
encasing exactly after the vertical leg of HSS section (Fig 4(d))
which is the weakest section with maximum bending demand.
The transverse bending demand at the mentioned point is
0.97 kN-m which corresponds to a demand to capacity ratio of
1.36. The maximum compressive load of the specimen considering
the global buckling is 11,300 kN based on the proposed equation
by Chou and Chen [12]. Considering all the design criteria, the
transverse bending of the encasing at the end defines the theoret-
ical maximum capacity of the specimen. The maximum compres-
sive capacity of the RLBRB based on the transverse bending
criteria is 463 kN.
The encasing is projected to the stiffened section of the core
(Fig. 4(a)) in order to prevent the instability experienced in some
BRB specimens [17] due to the plastic hinges at the core ends.
Two gaps are provided between the end stiffeners and encasing
at both ends in the form of longitudinal slots (Fig. 4(a)). Although
the required length of these slots is half of the maximum compres-
sive displacement, they are considered longer to prevent the load
carrying of encasing because of the probable slippage of the spec-
imen during testing. Moreover, the slots are made wider than the
core stiffeners to compensate for the construction tolerances and
to prevent steel to steel bearing and friction caused by the in-plane
deformation of the core.
2.3.3. Debonding mechanism
As mentioned before, one of the crucial criteria governing the
design of RLBRB is to prevent LCF failure. Any stress or strain con-
centration can trigger LCF failure. According to the previous stud-
ies, due to the presence of friction force between the core and
encasing, the compression demands localize on the core ends
and the tensile demands concentrate in the middle part [9,45]. This
non-uniform distribution of strain caused by friction reduces the
fatigue life. Accordingly, the less friction there is, the more fatigue
life is obtained. Furthermore, by reducing the friction, an ideal BRB
with identical strength in compression and tension is achieved.
Therefore, unlike some other types of BRB in which no debonder
is used [5,12], applying a debonding agent is one of the necessities
of the RLBRB.
Ceramic has demonstrated to have minimal friction to steel,
especially when covered with grease [46]. Hosseini and Ansari
[47] used ceramic as a paste on the core element of a normal
148 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160

8 M24 (8.8)
Encasing Plate Filler

1
HSS 60x60x5 (TH=10 mm) (TH=14 mm)

323
180
Longitudinal Slot

1
215 Gap 970 215
2 1400
Projected
Length
(a) Top View

8 M24 (8.8) End Plate

(TH=30 mm)
Through-Bolt

1
2

M10 (8.8)@50 Snug-Tight Bolts

Filler Plate HSS 60x60x5

1
2

(TH=14 mm) Welded

Nut
(b) Side View

HSS 60x60x5 Core Stiffener

M10 (8.8)
Core Plate
Critical Point
Hole=12 mm

Filler Filler
Dobonder

(c) Section 1 (d) Section 2

Fig. 4. Detailing of encasing of the RLBRB (Unit: mm).

BRB to decrease the steel to concrete friction. Although ceramic
could successfully reduce the friction between surfaces, it was
crushed at large compression cycles due to large out-of-plane
deformation of the core. In developing RLBRB, the ceramic is
painted on the encasing which remains elastic and experiences
negligible deformation. On the core side, two layers of 0.5 mm
polyethylene are taped on the core plate to prevent steel to cera-
mic contact. In order to further reduce the friction and protect
the ceramic from crushing and the polyethylene from being demol-
ished, industrial grease is used between the polyethylene tape and
the ceramic.
2.3.4. Stopper
Several detailings have been proposed so far to fix the encasing
on the core. Friction can help the encasing to be held on the core;
however, since the friction is not predictable and the encasing
gradually moves to one side leaving the core without restrainer,
this method cannot be trusted and has shown non-satisfactory per-
formance in previous works [9,47,48]. The restraining member was
connected to the core at one end and the entire required gap was
provided at the other end [5,49]. In such detailing, the local dam-
ages such as concrete crushing are anticipated to be more serious
at the free end as reported by Tremblay et al. [5]. A steel stopper
pin has been placed on one or two points in the middle of the flat
face of the core [23,28,29]. An outward projection of the core in the
middle length formed by computer numerically controlled (CNC)
cut has been utilized in some other schemes [12,50]. The core
has been tack welded to the encasing which brought about prob-
lems in the construction process [9].
AISC seismic provision [32] introduces the core plate as the pro-
tected zone on which no welded attachments are permitted. As
mentioned before, the presence of any discontinuity in the core
reduces the fatigue life. Therefore, in this study, a new detailing
has been developed for connecting the encasing to the core with-
out disturbing the core plate. The best place to fix the encasing is
the middle of the core, since the two end distances of encasing
to end plates will be equal and the gap between encasing and
the core stiffeners is half of the case when encasing is connected
to the core at one end. Moreover, it helps provide a more symmet-
ric condition for the two ends of the core. In order to eliminate dis-
continuity in the core, a 80  100 mm copper plate is placed on the
middle of the core (Fig. 3(a)). The thin (2 mm thickness) copper
plate, with a modulus of elasticity half that of steel and high fric-
tion against steel fills the gap between the core and encasing,
and when pressured by the six middle snug-tight bolts
(Fig. 4(b)), initiates a local favorable friction. By following this
approach the AISC criteria for a protected zone of core is satisfied.
In order to further fix the encasing to the core in the middle of the
encasing, a hole was provided in each of the encasings by drilling
the HSS and encasing plate. A nut was welded to the HSS and
two long bolts passing through the encasings were provided at
the center of RLBRB (Fig. 5). When these bolts are fastened through
the welded nuts, they lie on the two copper plates and exert nor-
mal desirable forces upon them and as a result tie the core to
encasing (Fig. 5).
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160 149

Fig. 5. Detailing of applying through bolts to improve stopping mechanism.

3. Experimental program

3.1. Constructing specimens

The core plates were CNC cut from a single plate using water-jet
technology. The cut was in the direction of plate forming process to
gain higher fatigue performance according to Krawinkler [51]. As
for the encasing, CNC plasma machine with less precision was used
for cutting. The core and the trapezoidal stiffeners were welded to
the end plates by CJP groove welds. When the welding procedures
of the core were finished, the layers of polyethylene tape were
pasted longitudinally on the core and the copper plate was tempo-
rarily fixed to the middle part (Fig. 6(a)).
In order to construct the encasing, the rectangular HSS was fillet
welded to the encasing plate after providing fixity by C-clips on the
plate to prevent the deformation of the flat plate during the weld-
ing procedure. The ceramic paste was brushed and leveled on the
encasing (Fig. 6(b)). The ceramic paste is made of two-component
epoxy adhesive which is suitable in terms of strength and stiffness Fig. 7. Assembled RLBRB specimen placed in the uniaxial setup.
as well as abrasion. When this material hardens it creates a metal-
like, smooth, tough, and shiny surface which has minimal friction
against other flat interfaces especially when lubricated by grease. the brace at yielding of BRB (Dby) and brace axial displacement
The core was placed within the encasings and the bolts were at design story drift (Dbm). The deformation Dby can be calculated
then tightened. Six bolts in the middle of the encasing were tightly from the yielding force of RLBRB core and its stiffness. The core
snugged to activate the copper plate stopping mechanism yielding force was previously estimated in Section 2.3.1. The stiff-
(Fig. 4(b)). In addition, two through-bolts were fastened to fix the ness of an axial member with a variable cross section is calculated
encasing on the core (Fig. 4(b)). Finally, the specimen was placed from the following equation:
in uniaxial setup by fastening the two sets of eight bolts on top
E
and bottom as shown in Fig. 7. K ¼ Rl ð6Þ
dx
0 AðxÞ

3.2. Loading protocol where A(x) is the cross sectional area of the core as a function of dis-
tance (x). By dividing the RLBRB core into seven parts the stiffness of
The loading protocol was based on AISC 341-10. This loading the whole RLBRB will be equal to119 kN/mm. Therefore the defor-
protocol includes cycles which are a multiplier of deformation of mation of RLBRB at yielding (Drby) will be 1.7 mm.

(a) Core with polyethylene film and copper plate

(b) Encasing pasted with ceramic

Fig. 6. Constructed parts of RLBRB.
150 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
The deformation Dbm can be calculated by a simple kinematic
relationship relating the interstory drift to the chevron brace axial
deformation as follows:
b Dpl
Dbm ¼ Dm
2L
where b is the total braced bay span, L is the total length of brace,
and Dm is the deformation corresponding to the design story drift.
AISC 341-10 requires a minimum of 1% for the design story drift
ratio for testing BRB specimens [32]. According to ASCE 7-10, the
allowable story drift ratio for a wide range of new buildings is lim-
ited to 2%. Due to the higher inherent stiffness of RLBRBs in compar-
ison to the normal BRBs, the structures equipped with such a
system are expected to undergo less drift demands predicted by
the design codes. Based on the aforementioned discussion the
design story drift ratio for the case of RLBRB is conservatively con-
sidered to be 1.5%.
Considering the chevron pattern for the bracing of a 3 m high
and 5 m wide frame, the total brace axial displacement at design
story drift (Dbm) is 28.8 mm. If the ratio of the core area to the elas-
tic remnant part of brace (Ac/A) is assumed to be 0.33, by extracting
the displacement of the elastic part of the brace, the axial displace-
ment of RLBRB at the design story drift (Drbm) would be 27.7 mm.
According to AISC 341-10, the loading protocol should consist of
displacement amplitudes 200% of the design story drift in case the
analysis is performed using linear method. This amount represents
the maximum interstory drift of frame during the earthquake of
10% in 50 years [40], which is locally concentrated in some stories
[14]. Results of numerous nonlinear time history analyses per-
formed on BRB structures indicated that the amount of 200% for
the estimation of maximum interstory drift might be conservative
[52]. Furthermore, RLBRB frames have more post-elastic stiffness
and hence produce a more uniform story drift in comparison to
normal BRB frames, leading to a lesser concentration of drifts [2].
Therefore, a value less than 200% is expected for RLBRBs. AISC
341-10 outlines that when nonlinear time-history analyses are
performed, the maximum inelastic drift in the loading protocol
can be taken directly from the analysis results. A series of nonlinear
time history analyses performed on a redesigned benchmark seven
story building [53] by utilizing short BRBs with Lc/L of 20–30%
demonstrated that the maximum interstory drift ratio for seven
selected ground motions was less than 2.25% [54]. This amount
is approximately equal to 150% of the design story drift (Dm).
The results of nonlinear dynamic analyses performed on two three
story structure equipped with short BRB with Lc/L of about 0.2 also
verify the value of 150% [1]. Previously, the amount of 150% was
selected by FEMA 450 [55] for taking the maximum drifts into
account. Hence, in this study, 1.5Dbm was selected as the
maximum amplitude of displacement in the loading protocol of
specimen 1. However, for the sake of safety, two cycles of 2Dbm
were included in the loading protocol of specimen 2. The loading
protocol, as well as the estimated cumulative plastic deformation,
P
are shown in Table 1. In Table 1, ec is the average axial strain, Dpl
is the cumulative sum of plastic deformations of the two cycles at
Table 1
Loading protocol, axial strain, and cumulative plastic deformation.
Cycle amplitude Dby 0.5Dbm
Number of cycles N 2 2
Story drift ratio D/h (%) 0. 08 0.75
Story displacement (mm) 2.6 22.5
RLBRB displacement-Drb (mm) 1.7 13.3
ec (%) 0.2 1.2
P
Dpl (mm) – 93.4
g – 56
each displacement amplitude, and g is the ratio of cumulative
plastic displacement to yield displacement, which is calculated
from the following equation:
P
ð7Þ g¼ ð8Þ
Drby
AISC 341-10 [32] and OSHPD regulations [56] require
cumulative inelastic axial deformation of at least 200 and 350
times the yield deformation, respectively, to be considered in
developing the loading protocol. As shown in Table 1, g is 377 up
to the end of 1.5Dbm cycles, which satisfies both aforementioned
requirements.
3.3. Instrumentations and experimental setup
A uniaxial test setup with a double acting hydraulic cylinder
capable of exerting 3000 kN compressive and 1000 kN tensile
forces with a 1000 kN capacity loadcell was used to perform the
tests. During the test the relative axial displacements of the two
end plates were monitored by two linearly variable displacement
transducers (LVDTs) and the average amount was considered as
the value of displacement controlled loading (LVDT1 and LVDT2).
Moreover, two other LVDTs measured the distance between the
encasing and the end plates at the top (LVDT3) and bottom
(LVDT4) of the specimen to check the stopping mechanism and
the slippage of the encasing (Fig. 8). Three LVDTs were mounted
perpendicular to the encasing to monitor the potential out-of-
plane deformation of the specimen at the top, middle, and bottom
of the specimen (LVDT5, LVDT6, and LVDT7). An LVDT was used to
report the in-plane deformation of the specimen at the middle part
(LVDT8). Two series of three strain gauges were glued longitudi-
nally to the top, middle, and bottom of the two encasing boxes
(Fig. 8) to measure the contribution of the encasings in load
carrying and also monitor their local performance and probable
damages. All the data was collected every second using a digital
data acquisition system. The details of the test setup and instru-
mentation devices are shown in Fig. 8.
4. Experimental results
4.1. Results of specimen 1
The load–displacement of specimen 1 is depicted in Fig. 9. The
hysteresis loops are stable, with smooth curves up to the end of
loading protocol. The force–displacement curve of the second cycle
at each displacement amplitude embraces the first cycle which is
the result of the isotropic hardening of steel combined with fric-
tional cumulative effects. The latter is interpreted by the gradual
intensifying local buckling in the core, which causes the friction
force to rise in each cycle. The local sawtooth parts in the yielded
compressive zone (Fig. 9) are due to the irregular sequence of ini-
tiation and vanishing of local friction between the core and ceramic
paste.
Dbm 1.5Dbm (Specimen 1) 2Dbm (Specimen2)
2 2 2
1.50 2.25 3.00
45.0 67.5 90.0
27.7 42.1 56.6
2.5 3.8 5.1
208.6 323.9 439.1
182 377 642
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160 151

Table 2
Core measured properties for specimen 1.

Cycle ec (%) l x b bx
0.5Dbm 1.2 8 1.01 1.09 1.10
0.5Dbm 1.2 8 1.11 1.11 1.23
Dbm 2.1 14 1.28 1.12 1.43
Dbm 2.3 15 1.36 1.14 1.55
1.5Dbm 3.5 23 1.45 1.17 1.70
1.5Dbm 3.8 25 1.55 1.13 1.75
2Dbm 5.0 33 1.68 NA NA

not only the AISC 341-10 [32], but also the OSHPD regulations
[56]. The maximum displacement ductility (l) at 1.5Dbm cycle is
25. The ratio of maximum compressive force to maximum tensile
force is identified as the compression strength adjustment factor
(b) by AISC Seismic Provision [32]. The maximum strength adjust-
ment factor of specimen 1 at the displacement corresponding to
1.5Dbm, is 1.17%. This amount is below the code recommended
limit and shows that the debonding mechanism has satisfactorily
reduced the friction.
Neither torsional deformation at the two ends, nor significant
in-plane or out-of-plane deformation was witnessed up to the
end of the test. Generally the performance of the specimen was
satisfactory up to the maximum uniaxial strain of 3.8% correspond-
ing to 1.5Dbm. For distinguishing the failure mechanism of speci-
men 1, higher displacement amplitudes were exerted to the
specimen. The specimen withstood the tensile excursion corre-
sponding to 2Dbm; however, with the initiation of compressive
increments the stiffness degraded (Fig. 9) and in the consequent
tensile excursion of the second 2Dbm cycle fracture occurred. The
load–displacement after the compressive 2Dbm cycle to the end
of the test is specified as dotted lines in Fig. 9. The maximum ten-
sile force at the 2Dbm cycle is greater than the nominal yielding
force, leading to the strain hardening adjustment factor (x) of
1.68%. This amount is slightly higher than the hardening parameter
resulted from the coupon tests under monotonic loading. This
effect is due to the cyclic hardening behavior of steel as observed
in cyclic coupon tests [30]. Table 2 presents the measured proper-
ties of specimen 1 at each cycle. As shown in Table 2, as the ductil-
Fig. 8. 3D view of RLBRB test setup. ity demand on the core increases the overstrength of the brace
increases.
The performance of specimen prior to the fracture can be
400 explained as follows. In the compressive cycle corresponding to
300 2Dbm, the out-of-plane elastic deformation of the encasing
200 increased resulting in the gap increase. The increase in the gap size
100
caused the normal force of the core to exceed the estimated
Force (kN)

amount and the transverse bending surpassed the bending capac-

0
ity leading to the yielding of encasing at the critical point
-100
(Figs. 4(d) and 10(b)). Therefore the local buckling of the core at
-200 the top end intensified and the encasing could not restrain the core
-300 buckling anymore. The yielding of the encasing at the section after
-400 the critical leg of HSS section (Fig. 10(b)) was probable since little
Onset of
-500 capacity to demand ratio was selected in the design procedure.
Degradation
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 Further increase in the gap size (Fig. 10(c)) caused amplification
Displacement (mm) in the normal force resulting in the failure of last row bolts. From
this step to the end of cycle, the shortening of the RLBRB took place
Fig. 9. Load–displacement curve of RLBRB specimen 1 (positive force denotes by major out-of-plane bending of the core with less stiffness in
tension).
comparison to axial stiffness, which can be easily seen in the
stiffness degradation in the force–displacement curve (Fig. 9). Sim-
The elastic stiffness of specimen 1 obtained from the initial elas- ilarly, when the specimen was unloaded and experienced further
tic cycles is 117 kN, which is close to the theoretical stiffness. The tensile loads, it yielded sooner than anticipated, and the core frac-
maximum tensile and compressive force corresponding to 1.5Dbm ture occurred at the beginning steps of the second tensile cycle,
cycle is 307 kN and 347 kN, respectively. The ratio of cumulative corresponding to 2Dbm at the top end of the core near the locally
inelastic axial deformation to the yield deformation (g) of speci- buckled areas. The local failure of the encasing and the increase
men 1 after the end of 1.5Dbm cycles is 378, which well satisfies in gap size is shown in Fig. 10.
152 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160

Table 3 40
Core residual stain percentile of specimen 1.

Location eT (%) eW (%) eL (%) 30

End (Bottom) 4.7 2.7 7.4 20

Middle 7.2 6.2 13.4
End (Top) 3.7 3.1 6.8
10

LVDT4 (mm)
0
During the test, the values of LVDT3 and LVDT4, which mea-
-10
sured the distance between the encasing and the end plates at
the top and bottom of the specimen, were almost equal to each -20
other confirming the fact that the encasing was stuck to the core
center and the stopping mechanism performed favorably. Fig. 11 -30
shows the response of LVDT 4 versus LVDT 3 during the test. As
shown in Fig. 11 before the formation of the local defect at the -40
-30 -20 -10 0 10 20 30
top of specimen there is limited difference between the response
LVDT3 (mm)
of the two LVDTs, however, in the additional cycles of loading
when the large local buckling of the core formed, the encasing Fig. 11. Response of LVDT4 versus LVDT 3 (Specimen 1).
experienced slippage.
The core was closely investigated after the specimen was dis-
mantled. The core had experienced residual complete sine in-plane fractured in a brittle manner at the top end of the core, where
buckling as well as numerous out-of-plane local bucklings (Fig. 12). the local bucklings had the highest amplitudes.
The amplitude of local buckling was not uniform in all spots, vary- The average residual transverse components of strain were cal-
ing from 1 mm to 8 mm. The local bucklings at the ends were more culated by measuring the width and thickness of the core in the
intensified than those formed in the middle. There were a total of middle and ends after the test and comparing them to the initial
21 local buckling waves ,and the lengths of waves varied from values. The levels of plastic strain components in the core are
30 mm to 60 mm. As can be seen in Fig. 12, specimen 1 was reported in Table 3. In Table 3, eT and eW are the plastic strains

(b) Specimen 1 after the test (photo shot 1)

(a) Specimen 1 photo shots (c) Specimen 1 after the test (photo shot 2)

Fig. 10. Local failure of encasing in Specimen 1.

 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
(a) Top view
(b) Side view
(c) Top view of fractured spot
Fig. 12. Core of RLBRB specimen 1 after test.
along the thickness and width of the core, respectively. eL is the
plastic longitudinal strain which can be calculated from the other
strain components. In the plastic state, the volume of steel does
not change [57], therefore the volumetric strain would be equal
to zero:
ev ¼ eL þ eT þ eW ¼ 0
where ev is the volumetric plastic strain. Assuming eT and ew to be
equal to the average engineering strains, then the plastic longitudi-
nal strain (eL) can be estimated from Eq. (9).
Reviewing the strain gauge data shows that the strain gauges
located at the middle height recorded the maximum amount
among all others, reaching a value of 0.000152. This is the result
of the increasing participation of the encasing in load carrying
while getting close to the center of the encasing due to the cumu-
lated number of contact points between the core and the encasing.
The maximum strain of the four end strain gauges reached
0.000083 which generally indicates little participation of the enca-
sings in load resisting.
All in all, RLBRB specimen 1 showed satisfactory performance in
terms of the appropriate load–displacement curve, ductility, cumu-
lative inelastic displacement, and the level of friction forces.
4.2. Results of specimen 2
In order to improve the performance of RLBRB specimen, the
end zone of the encasing was stiffened to restrain the large ampli-
tude of local bucklings of the core. This was based on evaluating
the transverse flexural demand on the encasing due to the normal
forces generated by the out-of-plane deformation of the core, con-
sidering a safety factor of three for the transverse bending capacity
to demand ratio. Hence, three rows of 10 mm thick transverse stiff-
eners were added to the encasing (Fig. 13).
The instrumentation and the setup condition of specimen 2
were identical to specimen 1. As noted in Section 3.2, the loading
protocol of specimen 2 included two extra 2Dbm cycles. The
load–displacement curve for specimen 2 is depicted in Fig. 14. As
shown in Fig. 14, specimen 2 had stable performance up to the
end of loading protocol including cycles of 2Dbm without any stiff-
ness degradation unlike what was observed in specimen 1 (Fig. 9).
153
(d) Side view of fractured spot
The maximum tensile and compressive forces corresponding to
the second cycle of 2Dbm are 326 kN and 398 kN, respectively.
After specimen 2 withstood the loading protocol, the test was
continued by cycles with 1.5Dbm amplitude to the failure of spec-
imen. The specimen sustained the first additional 1.5Dbm cycle
without failure; however, it experienced fracture exactly at the
ð9Þ
tension excursion of the second additional 1.5Dbm. The ratio of
cumulative inelastic axial deformation to the yield deformation
(g) at the end of the first additional 1.5Dbm is 666 significantly lar-
ger than the code recommended amount. Table 4 presents the
measured properties of specimen 2 at each cycle.
According to Table 4 the compression strength adjustment
factor (b) is within the acceptable limit. The maximum strain hard-
ening adjustment factor (x) of specimen 2 is 68%. The maximum
displacement ductility achieved is 32 which corresponds to a uni-
form axial strain of 5.0%. Similar to specimen 1, the stopping mech-
anism of specimen 2 worked satisfactorily (Fig. 15). Moreover, no
significant in-plane or out-of-plane deformation was observed.
The deformed shape of the core is shown in Fig. 16.
The global pattern of deformations in specimen 2 was similar to
specimen 1; nevertheless, specimen 2 experienced more severe
residual deformations compared to specimen 1. Specimen 2 was
fractured at the bottom end, near the crest of the fourth buckling
wave. The amplitude of buckling waves varied between 1 mm
and 10 mm. There were 28 buckling waves and the length of the
waves altered from 28 mm to 63 mm. The core plastic strain levels
measured after the test are reported in Table 5.
As can be inferred from Table 5 the strain amplitudes of specimen
2 were higher than those of specimen 1 because of higher displace-
ments exerted on specimen 2. The strain history excerpted from the
strain gauges of specimen 2 indicates a maximum strain of 0.000336
which belongs to the middle height strain gauge proving the elastic
state of the encasing to the end of the test. The maximum strain of
the encasing at the ends was 0.000205. The recorded data from
the strain gauges of both specimens show that when the specimens
were loaded in compression, the axial strains of the encasing
increased gradually to a maximum amount at peak compressive dis-
placement followed by a sudden drop when the specimens were
unloaded. This is a good representation of the formation and dimin-
ishment of friction force in consecutive loading steps.
154 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160

Fig. 13. Transverse stiffeners at encasing end of specimen 2.

400 30
300
200 20
100
Force (kN)

0 10
LVDT4 (mm)

-100
-200
0
-300
-400
-10
-500
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
Displacement (mm) -20

Fig. 14. Load–displacement curve of RLBRB specimen 2.

-30
-30 -20 -10 0 10 20 30
LVDT 3 (mm)

Fig. 15. Response of LVDT4 versus LVDT 3 (Specimen 2).

Table 4
Core measured properties for specimen 2.
The results of specimen 2 show that the modification to the
Cycle ec (%) l x b bx encasing member has effectively improved the cyclic performance
0.5Dbm 1.2 8 1.02 1.15 1.17 of RLBRB. The load–displacement curve of specimen 2 is similar to
0.5Dbm 1.2 8 1.18 1.16 1.37 specimen 1, except for the extra cycles and the level of compres-
Dbm 2.4 16 1.27 1.19 1.51 sion forces which is slightly higher in specimen 2 due to higher
Dbm 2.4 16 1.39 1.19 1.65
axial displacements and greater friction between the steel and
1.5Dbm 3.8 25 1.48 1.24 1.84
1.5Dbm 3.6 24 1.61 1.25 2.01
the plastic metal. In both specimens, the fracture occurred at the
2Dbm 5.0 32 1.61 1.25 1.88 large locally buckled end areas. Hence, it is concluded that limiting
2Dbm 4.8 31 1.68 1.21 2.10 extensive local buckling by means of providing enough transverse
1.5Dbm 3.8 25 1.66 1.22 2.03 stiffness for the encasing member at the ends of the encasing is of
1.5Dbm 3.6 24 1.61 NA NA
great importance for developing properly detailed RLBRBs.
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
(c) Top view of fractured spot
Fig. 16. Core of RLBRB specimen 2 after test.
Table 5
Core residual strain percentile of specimen 2.
Point eT (%) eW (%)
End (Bottom) 7.9 4.5
Middle 12.1 10.6
End (Top) 2.4 7.8
5. Finite element modeling
3D finite element modeling of BRB specimens has been recently
performed by a number of researchers [12,16,45] using ABAQUS
[58]. Korzekwa et al. simulated the seismic test of an all-steel
bolted BRB performed by Tremblay et al. [5]. The simulation incor-
porated half of the length of the all-steel BRB. The core plate and
the restraining members were modeled using 20-node C3D20 qua-
dratic solid elements. The bolts were modeled using infinitely stiff
springs connecting the encasings at bolt locations. The overall hys-
teretic response of the bracing member was well reproduced; how-
ever, the analysis was halted at the fourth cycle of the compression
deformation corresponded to a strain of 1.25%. Modeling the whole
length of BRB leads to better recognition of the BRB performance,
since the core behavior is effected by the friction of the two ends
and neglecting the friction at one side postpones formation of
strain concentration at the middle and end of the core. Chou and
Chen [12] conducted finite element analyses on two sandwich
BRBs specimens (Specimens 1 and Specimen 4). The steel core,
restraining member and concrete infill were modeled using
reduced integration eight-node solid elements, C3D8R. The bolts
were simulated by rigid beams. Only one element was used
through the thickness of the core plate. The braces were monoton-
ically pushed and the analyses correctly predicted the envelope of
the hysteresis behavior of the tested braces. The numerical study
was continued with a parametric study on a series of models with
different parameters such as; length and cross-sectional area of the
core plate, bolt spacing, and moment of inertia of the restraining
member which were monotonically pushed to a strain of 3.2%. Raz-
avi et al. [16] performed a correlation study on an all-steel BRB
155
(a) Top view
(b) Side view
(d) Side view of fractured spot
specimen tested by Eryasar et al. [9]. All the plates were modeled
using C3D8 solid elements. Cyclic loading was imposed on the
specimen according to the test loading protocol with the maxi-
eL (%) mum strain of 2%. The results of the analysis showed agreement
12.4 between the test and FE modeling. In all the aforementioned stud-
22.7 ies, the kinematic hardening model with one backstress was used
10.2 to formulate the kinematic hardening effect, which resulted in a
difference between the analyses hysteresis loops and the test cyclic
response.
In this study the performance of the RLBRB specimen 2 is
numerically investigated for obtaining a more detailed perception
about the specimen using ABAQUS [58]. Furthermore, some
parameters which were not recorded during the test can be
assessed by the finite element model such as the core deformation
history. At the end of this section the RLBRB is applied in the whole
brace configuration for two purposes. The first is to evaluate and
compare the buckling capacity of the whole brace when RLBRB is
placed at the middle and the end of brace. The second is to com-
pare the energy dissipation of the brace including RLBRB and a con-
ventional full-length BRB. The detailed description of the model is
presented in the following sections.
5.1. Description of finite element model of specimen 2
The analyses conducted in this study corporate large displacement
as well as contact analysis at high axial strains for specimen 2. The
model included the core plate, filler plates, encasing members
(plate and box), end stiffeners, and end plates (Fig. 17). Since the
core element was to undergo large plastic deformation and large
higher mode bucklings, it was modeled using C3D20 quadratic
brick elements. This element is proper for large displacement and
each of the 20 nodes has three translational degrees of freedom
[58]. The C3D8 solid element was used to model all other elastic
parts. Selecting C3D20 for the core element and C3D8 for other
parts can reduce the CPU time.
The core plate was divided into 6 elements across its width and
2 over its thickness. The encasing plate had less mesh density than
the core. The model included 7546 elements in total, 1184 of which
were used to model the core part. Mesh sensitivity analysis
156 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
Fig. 17. The parts of FE model of the RLBRB.
showed that larger number of elements would not improve the
accuracy of the results. The finite element (FE) model of the RLBRB
is shown in Fig. 17.
Due to their high axial stiffness, the bolts were modeled as con-
straint points tying the fillers and the encasings. The contact prop-
erties considering both normal behavior without penetration and
tangential actions were defined on 4 surfaces of the core plate
and other related surfaces. Genna et al. [43] presented 0.15 as
the friction coefficient for the lubricated steel to steel interface
based on experimental and numerical results. Eryasar et al. [59]
recommended a frictional coefficient of 0.1 for the greasy polyeth-
ylene to steel interface. The preliminary analyses conducted by the
authors showed that the friction coefficient presented by Genna
et al. leaded in better agreement between the test and finite ele-
ment model results in this study. Thus a frictional coefficient of
0.15 was adopted. The movement of the encasing was constrained
to the center of the core by defining constraint equations.
The Young’s modulus and Poisson’s ratio of steel were selected
200 GPa and 0.3, respectively. The steel material model was sup-
posed to follow the von Mises yielding criterion. Mixed plasticity,
including kinematic and isotropic hardening, was used for the core
steel. The kinematic component presents the change of backstress
according to the plastic strain. The isotropic component defines the
change in the size of the yield surface as a function of equivalent
plastic strain. Based on the results of the tests on ST 37-2 coupons,
the hardening parameters were defined and adjusted to gain the
best agreement with the core cyclic response. To this end two
backstresses with yield stress of 157 MPa and kinematics harden-
ing moduli of C1 = 4600 MPa, C2 = 98,000 MPa and kinematic hard-
ening rate factor of c1 = 25, c2 = 1000 were considered. A
maximum change in yield stress Qb = 110 MPa and a rate factor
b = 4 were adopted for isotropic component.
Initial gaps were left between the core and the encasing in the
in-plane and out-of-plane direction of the core according to the
specimen detailing. An out-of-plane imperfection equal to half of
the gap thickness was imposed on the core to form a half-sine
shape. This initial deformation was imposed on the middle part
of the core in a separate model (imperfection model). The
deformed shape resulted from the imperfection model was set as
an initial condition for the main model so that the core plate does
not experience any stress prior to the analysis. The peak displace-
ments of the test were imposed to the moving end plate in the
form of axial displacement, while the other end plate was assigned
fixed condition in all degrees of freedom. The axial force was
derived from the force resultants on the fixed end plate.
5.2. Results of finite element analysis
The force–deformation curve obtained from the finite ele-
ment analysis of RLBRB is depicted in Fig. 18, which shows
good agreement between numerical simulation and the experi-
mental results. The axial stiffness resulted from the FE model is
approximately equal to that of the test. The maximum com-
pression force is greater than the maximum tension force by
18.5%.
A close monitoring of the performance of the core during the
steps of analysis reflects that in the first cycle of compressive dis-
placement the core buckles in the weak axis and comes into con-
tact with the encasing at both ends. When the axial demand
increases, the core buckles in the strong axis forming a sine shape.
Soon after buckling, the edges of the core find supports both close
and stiff enough to continue load carrying. Later, in higher com-
pressive displacement, as soon as the core yields, the core plate
buckles in higher modes in the out-of-plane direction. The local
bucklings put the core and the encasing into contact and result
in higher load carrying. The pattern of intensifying local bucklings
is shown in Fig. 19 for different cycles.
The lengths of buckling waves are shorter at the core ends sim-
ilar to what was observed after the test. The maximum number of
local buckling is 26 which is close to the theoretical amount and
comparable to that of the test. The final stage of analysis
(Fig. 19(d)) shows that although the displacement has reached
zero, some local bucklings are still present in the core. Fig. 20
shows the maximum principal strains in the core at the maximum
displacement amplitudes.
As illustrated in Fig. 20, the transition zones and the stiffeners
are within the elastic range. The results also show that when the
core is in tension the strain concentrates in the middle part
(Fig. 20(a)) and when the core is in compression the strains con-
centration is at the ends of the core (Fig. 20(b)). Based on the pat-
tern of strain distribution, the core can be divided into two parts,
namely compression-controlled and tension-controlled. The latter
consists of the middle part of the core and the former includes
the two ends (Fig. 20). This means that the compressive force is
more pronounced at the end zones leading to local buckling. In
contrast, tension is supported more severely by the middle of the
core with a higher potential of necking. At the end of the test the
compression-controlled zone has a permanent increase in area,
while the tension-controlled zone has experienced permanent
reduction in area, as was observed in the test. The distinction of
these two parts can be interpreted by considering the frictional
interaction of the core and the encasing. In the initial steps of com-
pression cycles the core leans against the encasing leading to a
500
Test
400
FE Modelling
300
200
Force (kN)
100
0
-100
-200
-300
-400
-500
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
Displacement (mm)
Fig. 18. FE load–displacement curve versus the test results of RLBRB specimen 2.
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160 157

(a) Δ by

(b) Δ bm

(c) 2Δ bm

(d) End of loading

Fig. 19. Local buckling of core in different compressive cycles (30).

(a) 2Δ bm tensile displacement

(b) 2Δbm compressive displacement

Fig. 20. Deformed shape and maximum principal strain of RLBRB.

decrease in axial force in its middle parts. Therefore, the end

regions of the core experience the maximum axial force among
other parts of the core leading to local buckling of core at these
spots. When the core is loaded in tension, a portion of tensile dis-
placement is assigned to leveling the local bucklings formed in pre-
vious step, though the entire curvature is not reversed. In the
meantime, the middle region of the core is merely extended longi-
tudinally. This effect is intensified in higher plastic excursions in
which the two local buckled regions make two stiff and enlarged
parts in the core (compression-controlled parts) placed in series
with a soft and a reduced part (tension-controlled part). When
these parts are forced in tension the soft part tends to be elongated
more than the two stiff parts. The encasing in the compression-
controlled region requires enough transverse stiffness and an effec-
tive debonding mechanism while the tension-controlled parts
should be well detailed to prevent premature necking as a result
of using conventional stopping mechanisms.
The deformed shape of encasing is shown in Fig. 21, in which
the bending is majorly formed in the transverse direction, proving
that the bolt spacing is properly selected to prevent longitudinal
bending. Fig. 21 also shows that the transverse stiffeners have effi-
ciently restricted the encasing flexure in the end zones.
Generally, the numerical simulation is compatible with the
experimental results, and the assumptions as well as modeling
approaches could help in predicting the performance of RLBRB
with strain amplitudes up to 5%.
Fig. 21. Deformed shape of encasing at maximum compressive displacement
(150).
5.3. Applying the RLBRB in the brace configuration
The RLBRB was placed in series with circular HSS of 150  10 to
investigate the effect of the place of RLBRB in the whole brace con-
figuration and to compare the energy dissipation capacity of the
RLBRB and conventional BRB. The buckling load of the brace was
investigated for the two cases. In the first case the RLBRB was
placed in the middle of the brace. In the second case the RLBRB
was placed at one side of the brace. Fig. 22 shows the finite ele-
ment model of the two braces.
The HSS was assumed to be made of ST 37-2 steel following the
same hardening rule as the other parts. The C3D8 solid element
was used to model HSS. Other modeling assumptions are in accor-
dance with Section 5.1. After imposing an initial imperfection of L/
1000 to the mid length of the brace, the braces were monotically
158 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160

Fig. 24. FE model of the conventional BRB.

45
40 RLBRB
(a) RLBRB in the middle of the brace 35 Conventional BRB

Energy (kN-m)
30
25
20
15
10
5
0
0.5Δb -0.5Δ bm 1.5Δ bm -1.5Δ bm
Cycle

Fig. 25. Comparison of the plastic energy time histories of conventional BRB and
RLBRB.

400
(b) RLBRB at one end of the brace RLBRB
300
Fig. 22. FE model of RLBRB in series with elastic part. Conventional BRB
200
Force (kN)

100

600 0
RLBRB at Middle
500 -100
RLBRB at End
-200
400
Force (kN)

-300
300
-400
-50 -40 -30 -20 -10 0 10 20 30 40 50
200
Displacement (mm)
100
Fig. 26. Comparison of load–displacement curve of RLBRB and conventional BRB.
0
0 50 100 150 200
Displacement (mm) BRB owing to the more isotropic hardening effect in RLBRB due
to the higher plastic strains.
Fig. 23. Comparison of load–displacement curve of RLBRB at middle and end. This effect can be easily conceived by comparing the force–
displacement curves of the two braces as shown in Fig. 26. The
force–displacement curve of the RLBRB embraces the conventional
BRB curve which shows greater hardening effect.
pushed to reach their buckling capacity. The force–displacement
curve of the two specimens are shown in Fig. 23. The result of finite
element analysis shows that when the RLBRB is placed at the end 6. Summary and conclusions
of the brace larger buckling capacity is provided.
In order to compare the energy dissipated by the brace includ- Based on the advantages of reducing the length of buckling-
ing RLBRB and conventional BRB, a conventional BRB following the restrained brace (BRB), a type of all-steel reduced length BRB
design detailings presented earlier is formulated as shown in (RLBRB) was developed. The design provisions, detailing, and con-
Fig. 24. Both conventional BRB and brace including RLBRB struction procedure of RLBRB was explained. The findings of Low-
(Fig. 22(b)) are exposed to cycles of inelastic deformations of cycle fatigue (LCF) tests on BRBs were outlined and considered in
0.5Dbm and 1.5Dbm. the detailing and fabrication process. The performance of the pro-
The history of cumulated plastic energy dissipated by the two posed RLBRB was experimentally studied by performing two quasi
braces is shown in Fig. 25. The brace including RLBRB shows static tests. Moreover finite element (FE) analysis was conducted to
greater energy absorption of 1.3 times that of the conventional evaluate the performance of specimen and application of RLBRB in
 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
the brace configuration. The main results of this study are listed as
follows:
 Specimen 1 showed fine hysteretic behavior with cumulative
inelastic deformation of 378 times the yielding displacement
and average axial strain of 3.8% up to end of loading protocol.
 Specimen 2 with the enhanced encasing, performed well with-
out any defect up to the end of two cycles of 2Dbm and one addi-
tional 1.5Dbm cycle. Specimen 2 underwent a large axial strain
of 5.0% with high cumulative inelastic deformation of 666 times
yielding deformation which far exceeds the code requirements.
 Generally specimen 1 and 2 satisfied the provision of AISC 341-
10 regarding the BRB specimens and showed large displace-
ment ductilities of 25 and 32, respectively.
 The proposed stopping mechanism using copper plate and the
presented debonding mechanism including ceramic paste, poly-
ethylene film and grease performed satisfactorily.
 The FE model can appropriately simulate the performance of
RLBRB specimens and the results conformed to the test cyclic
response in terms of elastic stiffness, force–displacement curve,
shape of local bucklings, and friction force developed in the
encasing.
 According to FE analysis results confirmed by the test, the core
can be divided into two parts of compression-controlled and
tension-controlled. Limiting the local buckling amplitudes by
providing sufficient encasing stiffness at compression-con-
trolled parts and applying proper stopping mechanism in ten-
sion-controlled part is necessary for achieving a proper RLBRB.
 The results of FE analysis on applying the RLBRB in the brace
configuration shows that when the RLBRB is placed at the end
of the brace higher buckling capacity is achieved in comparison
to the case the RLBRB is place at the middle of the brace.
 The FE analysis shows that RLBRB exhibit larger energy dissipa-
tion capacity compared to conventional full-length BRB.
Based on the findings of this research the short BRBs with less
weight and simpler handling conditions can be implemented in
structural systems as a relatively cost-effective replaceable struc-
tural fuse. Nevertheless, experimental study on the performance
of the whole brace including RLBRB and the elastic part as well
as detailed numerical comparison is recommended.
Acknowledgement
The authors would like to thank Eng. Mohammad Chamani for
his technical consultation and assistance during the construction
and welding process of the specimens.
References
[1] Tremblay R, Poncet L, Bolduc P, Neville R, DeVall R. Testing and design of
buckling restrained braces for Canadian application. In: Proceeding of 13th
world conference on earthquake engineering; 2004.
[2] Razavi SA, Shemshadian ME, Mirghaderi SR, Ahlehagh S. Seismic design of
buckling restrained braced frames with reduced core length. In: Proceeding of
the structural engineers world congress. Italy; 2011.
[3] Shemshadian ME, Razavi SA, Mirghaderi SR, Hosseini A, Khan Mohammadi M.
The advantages of reducing the length of yielding segment in seismic
performance of buckling restrained braced frames. In: Proceeding of sixth
international conference of seismology and earthquake engineering. Tehran,
Iran; 2011.
[4] El-Bahey S, Bruneau M. Buckling restrained braces as structural fuses for the
seismic retrofit of reinforced concrete bridge bents. Eng Struct
2011;33:1052–61.
[5] Tremblay R, Bolduc P, Neville R, DeVall R. Seismic testing and performance of
buckling-restrained bracing systems. Can J Civ Eng 2006;33:183–98.
[6] D’Aniello M, Corte GD, Mazzolani FM. Experimental tests of a real building
seismically retrofitted by special buckling-restrained braces. In: Proceeding of
2008 seismic engineering conference commemorating the 1908 Messina and
Reggio Calabria Earthquake; 2008. p. 1513–20.
159
[7] Mazzolani F, Della Corte G, D’Aniello M. Experimental analysis of steel
dissipative bracing systems for seismic upgrading. J Civil Eng Manage
2009;15:7–19.
[8] D’Aniello M, Della Corte G, Mazzolani FM. Seismic upgrading of RC building by
buckling restrained braces: experimental results vs numerical modelling. In:
Proceeding of international specialty conference on behaviour of steel
structures in seismic area (STESSA). Yokohama, Japan; 2006.
[9] Eryasar ME, Topkaya C. An experimental study on steel-encased buckling-
restrained brace hysteretic dampers. Earthq Eng Struct Dyn 2010;39:561–81.
[10] Mazzolani F. Innovative metal systems for seismic upgrading of RC structures.
J Constr Steel Res 2008;64:882–95.
[11] Dusicka P, Tinker J. Global restraint in ultra-lightweight buckling-restrained
braces. J Compos Constr 2013;17:139–50.
[12] Chou CC, Chen SY. Subassemblage tests and finite element analyses of
sandwiched buckling-restrained braces. Eng Struct 2010;32:2108–21.
[13] Chou CC, Liu JH, Pham DH. Steel buckling-restrained braced frames with single
and dual corner gusset connections: seismic tests and analyses. Earthquake
Eng Struct Dynam 2012;41:1137–56.
[14] Sabelli R, Mahin S, Chang C. Seismic demands on steel braced frame buildings
with buckling-restrained braces. J Eng Struct 2003;25:655–66.
[15] Tsai KC, Lai JW, Hwang YC, Lin SL, Weng CH. Research and applications of
double-cored buckling restrained braces. In: Proceeding of 13th World
Conference on Earthquake Engineering. 2004.
[16] Razavi SA, Mirghaderi SR, Hosseini A, Shemshadian ME. Reduced length
buckling restrained brace using steel plates as restraining segment. In:
Proceeding of 15th World Conference on Earthquake Engineering. Lisbon,
Portugal: International Association for Earthquake Engineering; 2012.
[17] Ma N, Wu B, Li H, Ou J, Yang W. Full scale test of all-steel buckling restrained
braces. In: Proceeding of 14th World Conference on Earthquake Engineering.
Beijing, China 2008.
[18] Mirghaderi R, Ahlehagh S. Study of seismic behavior of SCBF with ‘‘balanced
bracing’’. In: Proceeding of 14th World Conference on Earthquake Engineering.
Bejing, China: International Association for Earthquake Engineering; 2008.
[19] Di Sarno L, Manfredi G. Seismic retrofitting with buckling restrained braces:
application to an existing non-ductile RC framed building. Soil Dyn Earthq Eng
2010;30:1279–97.
[20] Lin PC, Tsai KC, Wang KJ, Yu YJ, Wei CY, Wu AC, et al. Seismic design and hybrid
tests of a full-scale three-story buckling-restrained braced frame using welded
end connections and thin profile. Earthquake Eng Struct Dynam
2011;41:1001–20.
[21] Di Sarno L, Manfredi G. Seismic retrofitting of existing RC frames with buckling
restrained braces. In: Proceeding of conference on improving the seismic
performance of existing buildings and other structures. ASCE; 2009.
[22] Stephens R, Fuchs H, Fatemi A, Stephens R. Metal fatigue in engineering. 2nd
ed. Wiley-Interscience; 2000.
[23] Nakamura H, Maeda Y, Sasaki T. Fatigue properties of practical-scale unbonded
braces. Architectural Institute of Japan; 2000.
[24] Koetaka Y, Narihara H, Tsujita O. Part 3 low cycle fatigue test on aseismic
braces. The experimental study on buckling restrained braces Architectural
Institute of Japan; 2001.
[25] Isoda K, Satake N, Mase S, Terada T, Horie T. Development of unbonded brace
damper restrained by channel section steel. Part4 low cycle fatigue tests:
Architectural Institute of Japan; 2002.
[26] Narihara H, Tsujlta O, Yasuda S. The experimental study on buckling restrained
braces. Part 4 low cycle fatigue test on aseismic brace with SN400B, SN490B:
Architectural Institute of Japan; 2002.
[27] Horie T, Mase S, Terada T. Development of unbonded brace damper restrained
by channel section steel. Part 5 low cycle fatigue tests on unbonded brace
dampers with SN400B. Architectural Institute of Japan; 2003.
[28] Usami T, Wang C, Funayama J. Low-cycle fatigue tests of a type of buckling
restrained braces. Proc Eng 2011;14:956–64.
[29] Iwata M, Muraim M, Nakagomi T. Experimental study on brittle fracture of
buckling-restrained braces influences of core plate welding specifications and
experimental temperatures. In: Proceeding of international specialty
conference on behaviour of steel structures in seismic area (STESSA).
Santiago, Chile 2012.
[30] Dusicka P, Itani AM, Buckle IG. Cyclic response of plate steels under large
inelastic strains. J Constr Steel Res 2007;63:156–64.
[31] Nip K, Gardner L, Davies C, Elghazouli A. Extremely low cycle fatigue tests on
structural carbon steel and stainless steel. J Constr Steel Res 2010;66:96–110.
[32] AISC. Seismic provisions for structural steel buildings. ANSI/AISC 341–10.
Chicago, Illinois: American Institute of Steel Construction; 2010.
[33] Palazzo G, López-Almansa F, Cahís X, Crisafulli F. A low-tech dissipative
buckling restrained brace. Design, analysis, production and testing. Eng Struct
2009;31:2152–61.
[34] Iwata M, Murai M. Buckling-restrained brace using steel mortar planks;
performance evaluation as a hysteretic damper. Earthquake Eng Struct Dynam
2006;35:1807–26.
[35] Miner MA. Cumulative damage in fatigue. J Appl Mech 1945;12:159–64.
[36] Mirtaheri M, Gheidi A, Zandi AP, Alanjari P, Samani HR. Experimental
optimization studies on steel core lengths in buckling restrained braces. J
Constr Steel Res 2011;67:1244–53.
[37] Wigle VR, Fahnestock LA. Buckling-restrained braced frame connection
performance. J Constr Steel Res 2010;66:65–74.
[38] Vargas R, Bruneau M. Experimental response of buildings designed with
metallic structural fuses. II. J Struct Eng 2009;135:394–403.
160 S.A. Razavi Tabatabaei et al. / Engineering Structures 77 (2014) 143–160
[39] ASTM. Standard test methods and definitions for mechanical testing of steel
products. ASTM A2702002.
[40] ASCE. Minimum design loads for buildings and other structures. ASCE 7–10:
American Society of Civil Engineers; 2010.
[41] Watanabe A, Hitomi Y, Yaeki E, Wada A, Fujimoto M. Properties of brace
encased in buckling-restraining concrete and steel tube. In: Proceeding of 9th
World Conference on Earthquake Engineering. Tokyo-Kyoto, Japan 1988. p.
719–24.
[42] D’Aniello M, Della Corte G, Mazzolani FM. All-steel buckling-restrained braces
for seismic upgrading of existing reinforced concrete buildings. In: Proceeding
of international specialty conference on behaviour of steel structures in
seismic area (STESSA). Philadelphia, US2009.
[43] Genna F, Gelfi P. Analysis of the lateral thrust in bolted steel buckling-
restrained braces. I: Experimental and numerical results. J Struct Eng
2012;138:1231–43.
[44] Genna F, Gelfi P. Analysis of the lateral thrust in bolted steel buckling-
restrained braces. II: Engineering analytical estimates. J Struct Eng
2012;138:1244–54.
[45] Korzekwa A, Tremblay R. Numerical simulation of the cyclic inelastic
behaviour of buckling restrained braces. In: Proceeding of international
specialty conference on behaviour of steel structures in seismic area
(STESSA). Philadelphia, US2009.
[46] Andersson P, Ylöstalo O. The influence of lubrication on ceramic and steel
sliding contacts. Mater Sci Eng, A 1989;109:407–13.
[47] Ansari M. Numerical and experimental study of buckling restrained braces
[Master of Science in Civil Engineering], Uinversity of Tehran; 2010.
[48] Wang CL, 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.
[49] Ju YK, Kim MH, Kim J, Kim SD. Component tests of buckling-restrained braces
with unconstrained length. Eng Struct 2009;31:507–16.
[50] Lin SL, Tsai KC. A study of all metallic and detachable buckling braces. Center
for Earthquake Engineering Research, National Taiwan University; 2003.
[51] Krawinkler H. Guidelines for cyclic seismic testing of components of steel
structures, Applied Technology Council; 1992.
[52] Dehghani M, Tremblay R. Standard dynamic loading protocols for seismic
qualification of BRBFs in eastern and western Canada. In: Proceeding of 15th
world conference on earthquake engineering. Lisbon, Portugal: International
Association for Earthquake Engineering; 2012.
[53] Sabelli R, Lopez W. Steel tips: Design of buckling-restrained braced frames,
Structural Steel Educational Council; 2004.
[54] Shemshadian ME, Razavi SA, Hosseini A, Mirghaderi SR, Khan Mohammadi M.
An analytical study on low cycle fatigue effects in buckling restrained braces.
In: Proceeding of computational methods in structural dynamics and
earthquake engineering (COMPDYN). Corfu, Greece 2011.
[55] BSSC. NEHRP recommended provisions for seismic regulations for new
buildings and other structures (FEMA 450). National Earthquake Hazards
Reduction Program: Building Seismic Safety Council, Federal Emergency
Management Agency; 2003.
[56] Merritt S, Uang CM, Benzoni G. Subassemblage testing of star seismic
buckling-restrained braces. La Jolla, California: Pacific Earthquake
Engineering Research Center, University of California at Berkeley; 2003.
[57] Chen WF, Han DJ. Plasticity for structural engineers. New York: Springer-
Verlag; 1988.
[58] SIMULIA. ABAQUS analysis and theory manual. Karlsson, and Sorensen Inc,
Providence, RI; 2007.
[59] Eryasar ME. Experimenral and Numerical investigation of buckling restrained
braces [Master of Science in Civil Engineering]: Middle East Technical
University; 2009.