International Journal of Steel Structures (2019) 19(4):1055–1069 Online ISSN 2093-6311

https://doi.org/10.1007/s13296-018-0187-7 Print ISSN 1598-2351

Effect of Loading History and Restraining Parameters on Cyclic

Response of Steel BRBs
Ahmad Fayeq Ghowsi1 · Dipti Ranjan Sahoo1

Received: 9 September 2017 / Accepted: 5 November 2018 / Published online: 14 November 2018
© Korean Society of Steel Construction 2018

Abstract
This study presents a numerical modelling and prediction of cyclic response of all-steel buckling-restrained braces (BRBs)
under different loading histories using a finite element software ABAQUS. The numerical models are validated by compar-
ing the predicted hysteretic response, core fracture, energy dissipation and displacement ductility with the results of a past
experimental study. Later, the validated numerical model is used in a parametric study to investigate the influence of coef-
ficient of friction and clearance between the steel core and restrainers, the length of yielding core segment and the position
of stoppers on core segments on the cyclic response of steel BRBs. The coefficient of friction is varied in the range of 0–0.5,
whereas the gap value is varied in the range of 0–5 mm. The main parameters studied are the hysteretic response, axial resist-
ance, instance of core fracture and energy dissipation response. Based on the analysis results, the optimum values of friction
coefficient, gap, length of yielding core and stopper orientation have been recommended for steel BRBs.

Keywords Axial strength · Buckling-restrained braces · Finite element analyses · Fatigue fracture · Hysteretic energy ·
Numerical modelling

1 Introduction core element typically consists of central yielding segments

of rectangular or cross-shaped sections and projected con-
Buckling-restrained braces (BRBs) are designed to yield nection as well as transition segments of the larger sectional
in both tension and compression without global flexural area at both ends. These connection and transitional seg-
buckling under cyclic axial loading in order to achieve sig- ments are designed to remain elastic without local buckling.
nificant inelastic deformations. This is accomplished by A gap is maintained between the yielding core plate and
an inner steel core plate that resists the axial load and an buckling-restraining system of BRBs by using unbonding
outer casing that provides restraining mechanism against the material at the interface. Extensive studies have been con-
global buckling of core plate. The outer casing consists of ducted at the component and frame levels to understand the
a square, rectangular, or round hollow structural steel sec- seismic performance of conventional BRBs. Black et al.
tions and is typically filled with mortar, concrete, or other (2002) conducted component testing of BRBs to character-
materials to provide buckling-restraining mechanism (Wu ize their hysteretic response. Kim et al. (2015) experimen-
and Mei 2015). The constrained inner core plate undergo- tally studied the cyclic performance of BRBs with H-shaped
ing local high-mode buckling results in similar hysteretic yield core. BRBs with short yielding core and detachable
behavior under both compression and tension loading, and casings have shown to exhibit the significant energy dis-
the development of significant energy dissipation capacity sipation (Mirtaheri et al. 2011; Pandikkadavath and Sahoo
during earthquakes (Wu et al. 2014; Clark et al. 1999). The 2017) and to mitigate the excessive residual drift response
of braced frames (Pandikkadavath and Sahoo 2016; Chao
* Dipti Ranjan Sahoo et al. 2013). Extensive numerical studies (Sabelli et al. 2003;
drsahoo@civil.iitd.ac.in Fahnestock et al. 2007; Ghowsi and Sahoo 2013, 2015) have
Ahmad Fayeq Ghowsi also conducted to verify the seismic response mitigation
faieq.ghowsi@gmail.com potential of buckling-restrained braced frames (BRBFs).
BRBs have been applied to improve the performance of
1
Department of Civil Engineering, Indian Institute
of Technology Delhi, Hauz Khas, New Delhi 110016, India

13
Vol.:(0123456789)
1056 International Journal of Steel Structures (2019) 19(4):1055–1069

seismically-deficient existing buildings and bridges (Celik used to study the influence of magnitude of frictional resist-
and Bruneau 2009; Khampanit et al. 2014). ance and clearance between the core and the casing, core
All-steel welded or bolted systems have been developed, slenderness ratio, and position of stopper on the hysteretic
to date, for providing buckling-restraining mechanisms response, energy dissipation, core fracture, and displacement
to the yielding core plates of BRBs. In an all-steel BRB, ductility response of BRBs.
the inner core is sandwiched by a restraining system made
entirely of steel components (Chou and Chen 2010). An all-
steel BRB provides several advantages, in terms of light- 3 Numerical Study
weight, cost saving, ease in maintenance and inspection,
and faster fabrication and production (Dusicka and Tinker In this study, nonlinear finite element analysis has been con-
2013; Della Corte et al. 2015). An experimental study by ducted on steel BRBs of different configurations under vari-
Tremblay et al. (Tremblay et al. 2006) demonstrated the ous loading histories. The details of geometric and material
potential of the all-steel BRB system for adequate ductile properties of BRBs, modelling technique, and damage evo-
seismic response. Providing a direct contact between the lution and initiation model considered in fracture analysis
core and the outer casing may be adequate for achieving are discussed in the following sections:
stable hysteresis at the high strain levels (Eryasar and Top-
kaya 2010). However, friction between the core plate and 3.1 Geometric Properties of BRBs
the restraint member of all-steel BRB may lead to the larger
compression overstrength and result in the global flexural Six all-steel BRBs of a constant core length of 1600 mm
buckling of the outer casings (Khampanit et al. 2014). A gap and varying core cross-sections of 160 mm × 20 mm,
between the inner core and outer casing may be provided 200 mm × 20 mm, and 200 mm × 25 mm are considered
to cater the Poisson’s expansion of the core in compression in this study. These BRBs are referred to as W160T20-1,
and to reduce the frictional resistance. A significant amount W160T20-2, W200T20-1, W200T20-2, W200T25-1, and
of research work has been performed in Japan and U.S. and W200T25-2, in which “W” and “T” represent the width
elsewhere in Asia over the last few decades for the devel- and thickness of brace core segments, respectively. Fig-
opment of BRBs. Numerical investigations (Korzekwa and ure 1 shows the various components, restrainer assembly,
Tremblay 2009; Hoveidae et al. 2015) indicated a complex and dimensions of BRB models considered in this study.
interaction developed between the inner core and the outer The stoppers are provided on both sides of strong surfaces
casing of all-steel BRBs. (i.e., parallel to the wider dimensions). The clearance (gap)
between the core and the restrainers are considered as 1 mm
and 2 mm along the strong and weak surfaces, respectively.
2 Research Objectives Single rectangular cross-section as the core segments and
cruciform-shaped connection and transition segments are
Since the outer casings of all-steel BRBs are relatively considered for all BRBs. These BRBs have been tested ear-
lighter in weight, intermediate stoppers are often placed lier under the slow-cyclic loading by Wu et al. (2014). In this
along the length of core plates to keep the casings in posi- study, these BRBs are modelled numerically to predict their
tion for providing the effective buckling-restraining mech- cyclic performance. The results of experimental study (Wu
anism. The location of stoppers along the core segments et al. 2014) have been used to validate the proposed numeri-
may influence the overall cyclic performance of steel BRBs. cal model. As shown in the Fig. 1, the weak-axis restrainers
It is necessary to study the influence of location of stop- completely covered the transition segments, whereas the
pers and type of unbonding material (i.e., frictional resist- strong-axis restrainers extended over the full length of core
ance) between the core and the restrainer elements in order segments only. The details of finite element modelling of
to develop uniform strain demand in the brace core (Chen BRBs are discussed in the following section. In the para-
et al. 2016). In addition, there is a need to investigate the metric study (discussed later), the thickness of plates in the
role of slenderness ratio of core plates and type of load- transition and end connection zones are varied as per the
ing history on the hysteretic behavior, energy dissipation requirements based on the core slenderness ratio. Further,
potential and core fracture of all-steel BRBs. An attempt has end stiffeners are provided symmetrically at both ends to
been made in this study to address these issues numerically limit the inelastic yielding at the central region of BRBs.
using finite element analysis approach. A parametric study
has been conducted to predict the cyclic response of all-steel 3.2 Finite Element Modelling
BRBs under different loading protocols. The results of finite
element analysis are compared with the findings of a past Finite element (FE) analysis of BRB models has been con-
experimental study. The validated numerical model is later ducted using a commercial software (ABAQUS CAE 2010).

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1057

(a) (b)

300 Wc 100 25

350 180 Lc = 1600 180 350

10
60 tc

350 150 1660 150 350

(c) (d)

Fig. 1  Schematic representation of all-steel BRB: a various components, b assembled configuration, c dimension of core plate, d cross-section at
mid-span (all dimensions are in millimetres)

General-purpose eight-node first-order hexahedral (C3D8R) the use of shell elements, instead of solid elements, may pro-
elements with reduced-integration technique and Hourglass vide the same results with the lesser computational efforts.
control are adopted to generate the finite element mesh. Each High-strength bolts connecting the restrainers as used in
node of C3D8R elements have three degrees of freedoms. the experiment (Wu et al. 2014) are not modelled explicitly
Hexahedral elements not only provide the solution of equiva- in this study for simplicity. It is worth-mentioning that the
lent accuracy at less cost, they have a better convergence rate main function of restrainer elements is to provide sufficient
and the lesser mesh sensitivity than the tetrahedral elements. restraints against the buckling of BRB core elements prior to
Volumetric strain is also constant throughout the hexahe- their compression yielding. Therefore, a perfect contact with
dral element. Reduced integration technique uses the lower- tie constraints has been assumed at the restrainer interfaces.
order integration to form element stiffness, thereby, reducing The similar modelling approach has also been adopted in
the computational cost as compared to the full-integration various past numerical studies on BRBs (e.g., Korzekwa
technique, and prevents mesh “locking” for the compres- and Tremblay 2009; Hoveidae et al. 2015). Figure 2 shows
sive material response. Hourglass control prevents the shear the finite element modelling of different components and the
locking as well as volumetric control. However, C3D8R ele- assembled configurations of BRBs. The surface-to-surface
ments tend to be not stiff enough in bending. Since stresses, contact with a friction coefficient of 0.15 in the tangential
strains, etc. are computed at the integration points which are direction (Bondonet and Filiatrault 1997) and a hard contact
located in the middle of the reduced integration element, the in the normal direction is considered between the core plates
smaller mesh size is required to capture a stress concentra- and restrainer assembly. Instead of enforcing a strict hard
tion at the boundary of a structure. Second-order reduced- pressure-closure constraint behavior, a nonlinear penalty
integration elements generally yield more accurate results method has been adopted for hard contact in the normal
than the corresponding fully-integrated elements. However, behavior between the surfaces. Some degree of penetration
the accuracy for first-order elements with full versus reduced is allowed in this case and the contact force is proportional to
integration is largely dependent on the nature of the prob- the penetration distance. Figure 3 shows the nonlinear pres-
lem. Solid elements have also been used in the modelling of sure-overclosure relationship and the variation of penalty
different types of BRBs in past studies (Pandikkadavath and stiffness with overclosure. The contact pressure between the
Sahoo 2016; Jia et al. 2017a, b). It is worth-mentioning that surfaces is zero in the inactive contact regime (i.e., clearance

13
1058 International Journal of Steel Structures (2019) 19(4):1055–1069

Fig. 2  Finite element model

of all-steel BRB: a various com-
ponents, b assembled configura-
tion

(a) (b)

Fig. 3  a Nonlinear pressure-overclosure relationship and b variation of penalty stiffness with overclosure

value greater than C0). A linear variation in the contact pres- models and is considered as 360 MPa for both W200T25
sure is assumed to be linear with a slope, Ki, for penetration models. Although the isotropic hardening model represents
values lying within points C0 and e. Initial penalty stiffness, the growth of yield surface in the multi-dimensional stress
Ki, is equal to the stiffness of underlying element. Similarly, space, it does not consider the cold working anisotropic
the contact pressure varied linearly with a slope equal to the effects that are introduced during cyclic loading (Halama
final penalty stiffness, Kf, for penetrations greater than d as et al. 2012). Further, the movement of dislocation in one
shown in Fig. 3a. The ratio of initial stiffness (Ki) to the final direction generating local back stresses that causes the
stiffness (Kf) is to be 1%. For penetrations in the range of e reduction in yield stress (Bauschinger effect) is also not
to d, the variation of the contact pressure is assumed to be integrated in the isotropic hardening models (Armstrong
quadratic and the penalty stiffness linearly increases Ki to Kf and Frederick 1966). In order to determine the functional
as shown in Fig. 3b. The value of d is assumed as 3% of the dependence of yield stress on the plastic loading history,
characteristic length representing a typical facet size. The nonlinear combined isotropic and kinematic hardening
ratio (e-C0)/(d-C0) representing the lower quadratic limit is model has been used in the material modelling in this study.
taken as 33.3%. The desired gaps as mentioned earlier are Nonlinear combined hardening model consists of two
provided in the respective directions of the core plates an components: (i) a kinematic hardening component that rep-
initial global imperfection of total length/1000 is assigned resents the translation of the yield surface in stress space
based on the first mode of buckling scaling to the BRB mod- through the backstress (α), and (ii) an isotropic hardening
els (Zhao et al. 2011; Kumar et al. 2015). component describing the change in the size of the yield
surface. Backstress represents a constant kinematic shift of
3.3 Material Properties the yield surface, which is very useful for modelling the
effects of residual stresses without considering them in the
Elastoplastic material properties with combined isotropic equilibrium solution. The kinematic hardening component
and kinematic hardening behavior have been assumed for ( 𝛼̇ ) is expressed as the combination of a linear kinematic
the modelling of core plates and stoppers. All restrainers are term and a nonlinear relaxation term (Chaboche et al. 1979)
modelled as perfectly elastic elements. The material yield as follows:
stress is taken as 330 MPa for W160T20 and W160T20

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1059

3.4 Damage Evolution Model

1
𝛼̇ = C (𝜎 − 𝛼)𝜖̄̇ p − 𝛾𝛼 𝜖̄̇ p (1)
𝜎0
A damage model has been explicitly considered in this
where σ0 is the initial yield stress, σ is the second-order study to predict the core fracture of BRBs under the fatigue
stress, 𝜖̄̇ p signifies the equivalent plastic strain rate, C repre- loading. In general, cyclic fatigue damage of steel material
sents the initial kinematic hardening modulus, and γ deter- involves two primary process, namely, damage initiation and
mines the rate at which the kinematic hardening modulus damage evolution. Two prominent mechanisms used for the
decreases with the increasing plastic deformation. The iso- damage initiation and evolution of the ductile metals are
tropic hardening component is expressed as a direct func- (i) nucleation, growth, and coalescence of voids, and (ii)
tion of the equivalent plastic strain, 𝜖̄p, which can consider shear fracture due to shear band localization. Most of the
both the hardening and softening of the material as follows fracture phenomenon is based on equivalent strain at frac-
(Lemaitre and Chaboche 1990): ture as a function of stress triaxiality (η). Stress triaxiality is
( p) expresses as follows:
𝜎 0 = 𝜎 0 (𝜖̄p ) = 𝜎|0 + Q∞ 1 − e−b𝜖̄ (2)
where 𝜎|0 is the yield stress at zero plastic strain, Q∞ is the 3𝜎m 𝜎 1 + 𝜎2 + 𝜎 3
𝜂= =√
maximum change in the size of the yield surface, and b 𝜎eq (3)
𝛼12 + 𝛼22 + 𝛼32 − 𝜎1 𝜎2 − 𝜎2 𝜎3 − 𝜎3 𝜎1
defines the rate at which the size of the yield surface changes
as plastic straining develops. The material parameters, Q∞
where σ1, σ2 and σ3 are the principal components of the
and b must be calibrated from the cyclic test data. Since the
stress tensor, σm = (σ1 + σ2 + σ3)/3, σeqis the equivalent stress,
cyclic tests are not carried out in the study, cyclic harden-
and α is the ratio of minor principal strain (rate) to major
ing parameters adopted in past studies are considered as the
principal strain (rate), In this study, the former mechanism
reference values. Korzekwa and Tremblay (2009) conducted
is used to predict the onset of damage in BRBs as shown in
a numerical study to investigate the cyclic performance of
Fig. 4a–d. This is based on the assumption that the equiva-
BRBs in which the values of C1, b and Q∞ were considered
lent plastic strain, 𝜖̄D, at which the damage initiates is a func-
p
as 8 GPa, 4, and 110 MPa, respectively. These values are
tion of the stress triaxiality (η) and the plastic strain rate (𝜖̄̇ p)
suitably modified using trial-and-error method to match the
as follows:
predicted hysteretic response with the test results. Accord-
p( )
ingly, the following values are considered as the material
p
𝜖̄D = 𝜖̄D 𝜂, 𝜖̄̇ p (4)
properties in this study: C1 = GPa, γ 1 = 52, Q∞ = 25 MPa The norm for damage initiation is met with the satisfac-
and b = 10. Korzekwa and Tremblay (2009) considered the tion of the following relationship:
modelling of surface-to-surface contact between the core
and restraining elements assuming infinite stiffness in the d𝜖̄p
transverse direction along with the tangential coulomb fric- ∫ (5)
𝜔D = p( ) =1
𝜖̄D 𝜂, 𝜖̄̇ p
tional behavior. However, a more realistic surface-to-surface
contact modelling considering that some degree of penetra- where ωD= a state variable that increases monotonically
tion and nonlinear variation of contact pressure with over- with plastic deformation. The determination of fracture
closure between the surfaces. strain and stress triaxiality experimentally is a complex and

Fig. 4  Fracture mechanism

depicting a void nucleation, b
necking between voids, c void
growth and d void coalescence,
and e damage evolution based
on the effective plastic displace-
ment

13
1060 International Journal of Steel Structures (2019) 19(4):1055–1069

tedious process. Hooputra et al. (2004) developed analyti- 3.5 Loading Histories

cal expressions for aluminium material based on the lim-
ited experimental results. These parameters are iteratively Six different loading histories, originally considered by
modified in this study so as to match the predicted damage Wu et al. (2014) for the experimental investigation, have
initiation with the experimental results. Since the material been selected in this study for the chosen BRB models. As
stress–strain relationship does not hold good once the dam- shown in Fig. 5, these loading histories include the possible
age is initiated, a stress-displacement approach (Hillerborg seismic effects on BRBs, such as, the gradually increasing,
et al. 1976) is considered in this study. During the analysis, decreasing or constant displacements. Loading history of
stress tensor in the material at any given time is given by the W160T20-1 represents the gradually increasing symmetrical
scalar damage equation as follows: core displacements till a maximum core strain of 3.5% in an
interval of 0.5% followed by constant fatigue cycles corre-
𝜎 = (1 − D)𝜎̄ (6)
sponding to 2% core strain. The difference between the load-
where D = overall damage variable, σ = effective (or undam- ing histories for W160T20-1 and W160T20-2 is that the core
aged) stress tensor computed in the current increment, displacements are gradually decreased from a core strain of
𝜎̄ = stress that would exist in the material in the absence of 3.5% to a value of 0.5% in the latter case. Loading histories
damage. The overall damage variable is computed as the of BRBs W200T20-1 and W200T20-2 consist of sudden
maximum of individual damage variables. A material is said increase or decrease of symmetrical cyclic displacements
to be completely fractured when the value of D reaches unity with maximum core strains of 3.0% and 4.5%, respectively,
as shown in Fig. 3b. followed by constant fatigue cycles. Loading history of
In this study, a relationship between the fracture strain W200T25-1 represents the compression-dominated loading
and stress triaxiality has been established by trail-and with a maximum tension core strain of 1.5%. The compres-
error method at a plastic strain rate of 0.0001 by compar- sion core strain reached a maximum value of 4.5% followed
ing the predicted cyclic response with the test results. by 3% in the subsequent cycles. Finally, loading history of
The same relationship is used for all simulation models. W200T25-2 consists of only fatigue cycles of constant core
Damage variable equal to 0.6 is assumed at an equivalent strain magnitude of 2%.
plastic displacement of 1.2. An element deletion option is
also used in the numerical models to indicate the failure
of an element if it reaches a maximum strain of 22%. 4 Results and Validation

The main parameters evaluated in this study are hyster-

etic response, energy dissipation potential, displacement
ductility, instance of core fracture, and equivalent viscous

60 60
Displacement (mm)

W160T20-1 60 W200T25-1
W200T20-1
30 30 30

0 0 0

-30 -30 -30

-60
-60 -60
60
60
Displacement (mm)

W160T20-2 60 W200T20-2 W200T25-2

30 30 30

0 0 0

-30 -30
-30
-60 -60
-60 0 5 10 15 20 25
0 10 20 30 40 0 5 10 15 20 25
Time steps Time steps Time steps

Fig. 5  Loading histories considered for the all-steel BRB models

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1061

damping of BRBs under the imposed loading histories. The and post-yield stiffness. A higher post-yield stiffness in ten-
predicted hysteretic response of BRBs is compared with the sion regions is noted for W200T25-1 BRB under the com-
experimental findings reported by Wu et al. (2014) for the pression-dominated loading. This is due to the Poisson’s
validation of finite element models. effect under the progressive compressive loading leading to
the increase in cross-sectional area of core segments result-
4.1 Hysteretic Response ing in the increased tension strengths as observed in the
hysteretic response. However, both models of same cross-
Figure 6 shows the comparison of hysteretic behavior of sectional area exhibited exactly same compression post-yield
steel BRBs under different load protocols. The predicted stiffness. Further, nearly-parallel post-yield stiffness indicate
hysteretic response of all BRBs matched very well with the the same strength adjustment factors and post-yield stiffness
experimental results. Some minor variation in the hyster- ratios of BRBs.
etic response is noted in the loading cycle corresponding to
the initiation of yielding of core segments in some numeri- 4.2 Energy Dissipation Potential
cal models. However, the overall response, the peak values
of axials load, and the hardening behavior in each cycle of The energy dissipated by a member under cyclic loading
loading and unloading of BRBs are well captured in the can be computed from the area enclosed under the corre-
numerical models. Figure 7a shows the comparison of back- sponding hysteretic loops. Figure 7b shows the cumulative
bone curves of hysteretic response of BRBs. BRBs of same energy dissipation potential of BRBs under the selected
cross-sectional area showed the exactly same elastic stiffness loading protocols. Both W160T20-1 and W160T20-2 BRBs

3000
W160T20-1 W200T20-1 W200T25-1
2000
Axial force (kN)

1000

-1000

-2000 Expt Expt Expt

FEM FEM FEM
-3000
3000
W160T20-2 W200T20-2 W200T25-2
2000
Axial force (kN)

1000

-1000

-2000 Expt Expt Expt

FEM FEM FEM
-3000 -80 -60 -40 -20 0 20 40 60 80
-60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60
Axial displacement (mm) Axial displacement (mm) Axial displacement (mm)

Fig. 6  Comparison of predicted and observed hysteretic response of all-steel BRBs

Fig. 7  Compression of a back- 3000

Energy dissipation (kNm)

bone curves, b cumulative 3000

energy dissipation 2000 2500
Axial force (kN)

1000 2000
0 1500
W160T20-1 W160T20-1
-1000 W160T20-2 W160T20-2
W200T20-1
1000 W200T20-1
-2000 W200T20-2 W200T20-2
W200T25-1 500 W200T25-1
-3000 W200T25-2 W200T25-2
0
-80 -60 -40 -20 0 20 40 60 80 0 4 8 12 16 20 24
Axial displacement (mm) Number of cycles
(a) (b)

13
1062 International Journal of Steel Structures (2019) 19(4):1055–1069

dissipated the same magnitude of cumulative energy at 16th Table 1  Comparison of instance of core fractures and cumulative dis-
cycle of loading corresponding to the core strain value of placement ductility
3%. Whereas an exponential variation in energy dissipation BRB model Instance of core fracture Cumulative displacement
is noted for W160T20-1 under the gradually increasing dis- in fatigue cycles ductility
placement cycles, the nature of the curve is reversed when Experiment Simulation Experiment Simulation
the loading cycles are gradually reduced from an initial peak
value. The fatigue cycles of constant axial strain resulted in W160T20-1 – 2nd 477 481
the linearly increasing energy dissipation beyond 16th cycle. W160T20-2 9th – 750 340
BRB W200T20-2 exhibited the higher energy dissipation at W200T20-1 39th 29th 1429 1072
the smaller number of loading cycles primarily because of W200T20-2 4th 4th 456 464
the higher core strain demand (4.5%) followed by the fatigue W200T25-1 6th 14th 1029 604
cycles of 3% core strain. Thus, the loading history signifi- W200T25-2 31st 36th 865 878
cantly influences the energy dissipation demand on BRBs
without affecting their strength adjustment factors.
Cumulative displacement ductility is an important
4.3 Core Fracture and Displacement Ductility acceptance criteria for the suitability of a BRB in seismic-
resistance applications. Table 1 also shows the comparison
Fatigue cycles of constant core strain are repeated at the end of cumulative displacement ductility of BRBs which is
of the loading history to study the fracture of BRB cores. computed as the ratio of cumulative plastic displacement
Figure 8 shows the comparison of hysteretic response of to the corresponding yield displacements. In most cases,
BRBs undergoing the core fracture. The predicted hysteretic the predicted values of cumulative displacement ductil-
response corresponding to core fracture matched very well ity matched with the past experimental results. Numeri-
with the experimental results. Table 1 shows the instance cal study predicted early fracture leading to the smaller
of core fracture noted in the numerical analysis and experi- cumulative ductility in W160T20-2 and W200T20-1,
ments. Though the number of fatigue cycles corresponding whereas the fracture loading history applied in analysis
to the core fracture in the analysis did not exactly match is somewhat different from that used in testing. Figure 9
with the test results, this difference seems to be acceptable shows the location of facture in BRBs as obtained from
considering the complexity in the modelling of fracture, the numerical analysis. As the core slenderness of BRBs
variation in the material properties in the post-ultimate stage is decreased, the core fractures are noted at the multiple
and initial imperfections which are difficult to capture in the locations.
numerical modelling.

3000
W160T20-1 W200T20-1 W200T25-1
2000
Axial force (kN)

1000

-1000

-2000

-3000

3000
W160T20-2 W200T20-2 W200T25-2
2000
Axial force (kN)

1000

-1000

-2000

-3000 -80 -60 -40 -20 0 20 40 60 80

-60 -40 -20 0 20 40 60 -80 -60 -40 -20 0 20 40 60 80
Axial displacement (mm) Axial displacement (mm) Axial displacement (mm)

Fig. 8  Predicted hysteretic response of all-steel BRBs undergoing core fracture

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1063

Fig. 9  Prediction of fracture of

core segments of all-steel BRBs

4.4 Equivalent Viscous Damping and the core cross-section of W160T20-1 model has been
considered. The same mesh arrangement and boundary con-
Equivalent viscous damping ratio (βeq) of BRBs at any axial dition as discussed earlier have been used in this parametric
core strain level is computed using the following expression study.
(FEMA 2000):
5.1 Frictional Resistance Between the BRB Core
1 Eloop and Restrainer
𝛽eq =
2𝜋 Keq (D+ − D− )2 (7)
Past studies (Tremblay et al. 2006; Eryasar and Topkaya
where Eloop = Dissipated energy per hysteretic loop at a 2010) have concluded that the frictional resistance between
drift cycle, Keq = Effective stiffness of hysteretic loop, D+ the core and restrainer assembly not only changes the com-
and D− are the peak positive and negative displacements of pression load-carrying capacity of BRBs, but also influ-
the loop. Table 2 summarizes the equivalent viscous damp- ences the higher-mode buckling of brace cores. This may
ing ratios of BRBs at different core strain level. It is noted induce the global flexural buckling of outer casings. To
that equivalent viscous damping ratio of BRBs is varied in investigate this issue further, the coefficient of friction (μ)
the range of 30–50%. No significant increase in the damp- between the core and restrainers of W160T20-1 BRB is
ing potential of BRBs is noted at the higher core strain in varied in the range of 0–0.5 in this study, keeping all other
excess of 1%. parameters constant. The restrainers are modelled as rigid
elements with a gap of 1 mm on strong surfaces and 2 mm
gap on weak surfaces. Figure 10a shows the comparison of
hysteretic response of the BRB with the varying values of
5 Discussion frictional resistance. Stable and nearly-symmetrical hyster-
etic response of BRBs is noted for all values of coefficient
A parametric study has been conducted using the validated of friction except for a value of 0.5 in which the compres-
numerical model to investigate the influence of frictional sion capacity of BRB is increased beyond the core strain
resistance and clearance between the core plate and restrain- value of 2.5%. This indicates the significant contribution
ers, length of yield core segments, and position of stoppers of the restrainer to the compression resistance of BRB. In
of all-steel BRBs. For this purpose, the loading protocol fact, the compression resistance of BRBs is increased if the

Table 2  Computation of BBR model 0.25% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.5%
equivalent viscous damping
ratio of BRBs at different core W160T20-1 0.328 0.441 0.504 0.478 0.485 0.497 0.508 0.483 –
strain levels
W160T20-2 0.330 0.445 0.481 0.483 0.480 0.478 0.497 0.503 –
W200T20-1 0.318 – 0.489 – 0.506 – 0.471 – –
W200T20-2 0.329 – – 0.495 – – 0.509 – 0.508
W200T25-1 – 0.316 – 0.476 – – – – –
W200T25-2 0.318 – – – 0.507 – – – –

13
1064 International Journal of Steel Structures (2019) 19(4):1055–1069

Fig. 10  Comparison of a 3000

hysteretic response, and b core µ= 0
fracture of W160T20-1 BRB 2000

Axial force (kN)

with varying coefficient of fric- 1000
tion between the core plate and
restrainer 0
-1000
-2000
-3000
3000
µ = 0.05
2000

Axial force (kN)

1000
0
-1000
-2000
-3000
3000
µ = 0.1
2000
Axial force (kN)

1000
0
-1000
-2000
-3000
3000
µ = 0.15
2000
Axial force (kN)

1000
0
-1000
-2000
-3000
3000
µ = 0.2
2000
Axial force (kN)

1000
0
-1000
-2000
-3000
3000
2000 µ = 0.5
Axial force (KN)

1000
0
-1000
-2000
-3000
-4000
-5000
-60 -40 -20 0 20 40 60
Axial displacement (mm)
(a) (b)

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1065

coefficient of friction exceeds a value of 0.10. For example, therefore necessary to investigate the influence of this gap
the compression strength of BRB for the coefficient of fric- on the cyclic performance of steel BRBs. A parametric study
tion as 0.2 is increased by 30% at a core strain of 3.5%. As has been conducted on the W160T20-1 BRB model with
expected, the tension resistance of BRBs is found to be same different values of gap between the core and restrainer. For
for all BRBs at all core strain levels. It can be concluded that this selected BRB, the minimum gap requirements along the
the coefficient of friction between the core and restrainer has strong and weak axes of core cross-sections are computed as
no significant influence on the energy dissipation response of 0.2 mm and 1.6 mm, respectively, corresponding to an axial
BRBs designed to sustain the core strain demand less than strain of 3%.Therefore, in this study, the gap magnitudes are
2.5%. However, the value of coefficient of friction in excess varied from 0 to 5 mm on both axes, which represents the
of 0.2 may significantly influence the behavior of outer cas- direct contact to more than six times of the required value.
ing in the higher core strain levels. Figure 10b compares the In all cases, a coefficient of friction is assumed as 0.15.
core fracture of the BRB under the fatigue loading cycles. Figure 11 shows the comparison of hysteretic response of
For very low values of coefficient of friction (< 0.1), the BRBs with different magnitudes of gap between the core
core fracture is noted near the transition regions due to the and restrainers. As expected, the tensile resistance of BRBs
high unconstrained rotational demand under the compressive remain unchanged for the selected range of clearance val-
loading. For the value of coefficient of friction greater than ues. However, the initiation of increase in the compression
0.1, the core fracture takes places near the central region. resistance of BRBs is noted from the initial loading cycles
Table 3 shows the comparison of cumulative displace- for zero clearance and from 2% core strain for a gap value
ment ductility and number of fatigue cycles at fracture of of 0.5 mm. In the latter case, the requirement of gap con-
the selected BRB. It can be clearly seen that the maximum sidering the Poisson’s effect is found to be higher than the
number of fatigue cycles applied to the BRB prior to the provided value of 1.0 mm beyond a core strain of 2%. No
core fracture is noted for the coefficient of friction as 0.15 significant difference is noted in the hysteretic response of
beyond which the core fracture is noted at an early stage. BRBs for gap values in the range of 1–2 mm (total gap of
The maximum value of cumulative displacement ductility 2–4 mm). However, some reduction in the compression
is computed for the coefficient of friction equal to 0.15. If resistance is observed in the BRB model for a gap value of
the frictional coefficient value is increased to 0.5, the core 5 mm on both sides.
fracture is observed even before the application of fatigue Figure 12 shows the deformed configuration and core
cycles. In addition, flexural deformation of end connection fractures, if any, of the BRB W160T20-1 with varying val-
segment of BRB also noted for this case may impose addi- ues of gaps between the core and restrainers. For the BRB
tional demand on the beam and columns of the braced frame. with no gap between the core and restrainer, the core seg-
Hence, the coefficient of friction should be considered as ment is stressed to a lesser extent as compared to the end
0.15 for the optimal cyclic behavior of steel BRB. connection segments due to lateral resistance provided by
the restrainer. This resulted in the axial and flexural defor-
5.2 Gap Between the BRB Core and Restrainer mations of unrestrained end connection segments under the
axial loading leading to the higher axial compression resist-
Theoretically, a minimum gap between the BRB core and the ance of BRB. For a gap of 0.5 mm on both sides of core,
restrainer is necessary to accommodate the Poisson’s expan- the fracture of BRB is noted at both ends near the transition
sion of core segment under axial compressive loading. A zones. The higher mode compression buckling along with
higher clearance (gap) between the core and restrainer facili- the fracture at the central region of core segment is noted
tates the higher mode compression buckling of BRBs. It is for BRBs with the magnitude of total gap exceeding a value
of 1.6 mm. For the BRB with a gap value of 5 mm on both
Table 3  Cumulative displacement ductility and fatigue cycles of
sides (i.e., 625% of the gap requirement as per the Poisson’s
BRBs with varying values of coefficient of friction between core and effect), the buckling of central core segment caused a reduc-
restraining elements tion in the compression resistance of BRBs as depicted in
Friction coefficient (µ) Cumulative displacement Number of the hysteretic response.
ductility fatigue cycles Figure 13a shows the comparison of backbone curves
of hysteretic response of BRBs with varying values of
0 572 5
gap between the core and the restrainer. BRB with zero
0.05 663 8
gap between the core and restrainer exhibited the signifi-
0.1 542 4
cantly higher compression resistance as compared to other
0.15 785 12
BRBs. Some minor increase in the tensile axial capacity is
0.2 603 6
also noted in this case. For a gap value of 0.5 mm on both
0.5 421 0
sides, the compression resistance is increased beyond the

13
1066 International Journal of Steel Structures (2019) 19(4):1055–1069

3000
Gap = 0 Gap = 0.5 Gap = 1
2000
Axial force (kN) 1000

-1000

-2000

-3000
3000
Gap = 1.5 Gap = 2 Gap = 5
2000
Axial force (kN)

1000

-1000

-2000

-3000
-60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60
Axial displacement (mm) Axial displacement (mm) Axial displacement (mm)

Fig. 11  Comparison of hysteretic response of W160T20-1 BRB model with varying magnitudes of gap between the core plate and restrainer

Fig. 12  Comparison of
deformed configurations and
core fracture of W160T20-1
BRB with varying magnitudes
of gap between the core plate
and restrainer

Fig. 13  Comparison of a back- 3000

Energy disipation (kN-m)

bone curves and b energy dissi- 3000 Gap = 0

pation response of W160T20-1 2500 Gap = 0.5
2000
Axial force (kN)

Gap = 1.0
BRB model with varying 2000
1000 Gap = 1.5
magnitudes of gap between the Gap = 2.0
core plate and restrainer 0 Gap =0 1500 Gap = 5.0
Gap = 0.5
-1000 Gap = 1.0 1000
-2000 Gap = 1.5
Gap = 2.0 500
-3000 Gap = 5.0
-80 -60 -40 -20 0 20 40 60 80
0
0 4 8 12 16 20 24
Axial displacement (mm) Number of cycles
(a) (b)

axial displacement of 32 mm (i.e., core strain of 2%) with 1–2 mm. For BRB with gap of 5 mm, some minor reduction
no substantial increase in the axial capacity in tension. No in the peak tension and compression resistance is noted at
difference in the axial resistance in both tension and com- each axial core strain levels. Figure 13b shows the com-
pression is noted for BRBs with gap varying the range of parison of cumulative energy dissipation potential of BRBs

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1067

with varying gap values. The higher energy dissipation is energy dissipation at each core strain level. Further, BRB
noted for BRB with zero gap between the core and restrainer, with the shorter yield core sustained the higher numbers
whereas the BRB with a higher gap (i.e., 5 mm) exhibited of fatigue cycles as compared to those with longer yield-
the smaller energy dissipation at each loading cycles. How- ing core segments prior to the core fracture. Therefore, the
ever, BRBs with the clearance values ranging between 1 and length of yielding core segment of BRBs should be selected
2 mm exhibited the same cumulative energy dissipation at judiciously in order to improve the axial capacity, energy
all core strain levels. This showed that the gap between the dissipation potential and resistance to fatigue fracture of
core and restrainer may be increased to 2.5 times the mini- core.
mum required value.
5.4 Position of Stoppers
5.3 Yielding Core Lengths
Stoppers are provided on the core segments of BRBs to keep
The influence of core length on the overall cyclic perfor- the restrainer assembly in the desired position. However,
mance of BRB has been studied by considering the three dif- the location of these stoppers may influence the energy dis-
ferent values of core lengths, namely, 1000 mm, 2000 mm, sipation potential and core fracture. Three different cases of
and 3000 mm. Figure 14 shows the comparison of hysteretic stopper position are considered in this study, namely, with
response of BRBs of varying yielding core lengths. BRBs stoppers on strong surface, with stoppers on weak surface,
with the smaller yielding core length exhibited the smaller and without stoppers. For this purpose, BRB W160T20-1
axial deformation under the same loading history and vice has been considered with the same geometric and material
versa. Ductile core fracture is noted in the BRB having a properties as explained earlier. Figure 16 shows the compari-
core length exceeding 2000 mm. Figure 15 shows the back- son of hysteretic response, core fracture and energy dissipa-
bone curves and cumulative energy dissipation potential of tion potential with the varying stopper configurations. The
these BRBs. The short core BRB exhibited the higher axial predicted hysteretic response and instance of core fracture
stiffness as well as strength adjustment factors as compared for the BRB with stoppers on the strong surface matched
to the longer yielding core segments. However, BRB with very well with the experimental results (Wu et al. 2014).
the longer core segment exhibited the higher cumulative Though the stable hysteretic response is noted for all BRBs,

3000
L = 1000 L = 2000 L = 3000
2000
Axial force (kN)

1000

-1000

-2000

-3000
-60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 -80 -40 0 40 80
Axial displacement (mm) Axial displacement (mm) Axial displacement (mm)

Fig. 14  Comparison of core fracture of W160T20-1 BRB model with varying yielding core lengths

Fig. 15  Comparison of a back- 3000

3000
Energy disipation (kN-m)

bone curves and b energy dissi-

pation response of W160T20-1 2500
2000
Axial force (kN)

BRB model with varying yield- 2000

1000
ing core lengths
0 1500
-1000 1000
-2000 L = 1000 L = 1000
L = 2000 500 L = 2000
-3000 L = 3000 L = 3000
0
-80 -60 -40 -20 0 20 40 60 80 0 4 8 12 16 20 24
Axial displacement (mm) Number of cycles
(a) (b)

13
1068 International Journal of Steel Structures (2019) 19(4):1055–1069

Fig. 16  Comparison of hyster- 3000 3000

Expt Stoppers on strong surfaces Without stopers
etic response, core fracture and
2000 FEM 2000
energy dissipation potential of

Axial force (kN)

Axial force (kN)
W160T20-1 BRB model with 1000 1000
different stopper position
0 0

-1000 -1000

-2000 -2000

-3000 -3000
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Axial displacement (mm) Axial displacement (mm)

3000
Stopers on weak surfaces
2000 3000

Energy dissipation (kN-m)

Axial force (kN)

1000 2500
0 2000
-1000 1500
-2000 1000
-3000 500 Stoppers on strong surfaces
-80 -60 -40 -20 0 20 40 60 80 Stoppers on weak surfaces
Axial displacement (mm) 0
Without stopper

0 2 4 6 8 10 12 14 16
Number of cycles

those with stopper on weak surface as well as without stop- ers of BRBs results in the unsymmetrical hysteretic
pers exhibited relatively early core fracture. As shown in the response along with the reduction in the energy dissipa-
energy dissipation curves, BRB with stoppers on the strong tion. A higher frictional resistance results in the higher
surface exhibited the better response due to the delayed core demand on the end connection segments as well as the
fracture. Hence, it may be concluded that using stoppers on restrainers.
the strong surface of core segment enhanced the displace- • The clearance between the BRB core and restrainers
ment ductility of BRBs under cyclic loading. should be in the range of 1.0–2.5 times the value com-
puted as per the requirements of Poisson’s effect in order
to get the same axial resistance and energy dissipation
6 Conclusions potential. While the higher gap induces the higher mode
compression buckling of core over a limited central
The following conclusions can be drawn from this study: regions, the smaller gap results in the additional demand
on the end connections segments and restrainers of steel
• The proposed finite element modelling technique not BRBs.
only successfully predicted the hysteretic response, • The length of yielding core segment of BRBs should be
cumulative displacement ductility and energy dissipa- selected judiciously in order to improve the axial capac-
tion response, but also captured the fatigue fracture of ity, energy dissipation potential and resistance to fatigue
core segments. fracture of core. Further, using stoppers on the strong
• The imposed loading history significantly influences the surface of core segment enhanced the displacement duc-
energy dissipation demand on BRBs. The equivalent tility of BRBs under cyclic loading as compared to those
viscous damping potential of steel BRBs is found to be on the weak surfaces.
about 50% at a core strain of 3% and higher depending • This study is focused on quantifying the influence of
on the loading history. friction, gap, and core length and stopper position on
• The coefficient of friction between the core segment the cyclic performance of steel BRBs without consid-
and the restrainer should be considered as about 0.15 ering the interaction between them. Further studies are
in order to get the better fatigue resistance of core seg- required to investigate the interaction effect of these
ment and cumulative displacement ductility of BRBs. parameters on the hysteretic response and fracture per-
Increasing friction between the core plate and restrain- formance of BRBs.

13
International Journal of Steel Structures (2019) 19(4):1055–1069 1069

References Hillerborg, A., Modeer, M., & Petersson, P. E. (1976). Analysis of

crack formation and crack growth in concrete by means of fracture
mechanics and finite elements. Cement and Concrete Research,
ABAQUS. (2010). ABAQUS user’s manual, Version 6.10. Hibbit,
6, 82–773.
Karlsson, and Sorenson, Inc., Providence.
Hooputra, H., Gese, H., Dell, H., & Werner, H. (2004). A compre-
Armstrong, P. J. & Frederick, C. O. (1966). A mathematical repre-
hensive failure model for crashworthiness simulation of alumin-
sentation of the multi-axial Bauschinger effect. G.E.G.B Report
ium extrusions. International Journal of Crashworthiness, 9(5),
RD/B/N, 731.
449–463.
Black, C. J., Makris, N. & Aiken, I. D. (2002). Component testing,
Hoveidae, N., Tremblay, R., Rafezy, B., & Davaran, A. (2015). Numeri-
stability analysis and characterization of buckling-restrained
cal investigation of seismic behavior of short-core all-steel buck-
unbonded braces. Pacific Earthquake Engineering Research
ling restrained braces. Journal of Constructional Steel Research,
Center, Report No. PEER 2002/08, Berkeley.
114, 89–99.
Bondonet, G., & Filiatrault, A. (1997). Frictional response of PTFE
Jia, L., Ge, H., Maruyama, R., & Shinohara, K. (2017a). Development
sliding bearings at high frequencies. ASCE Journal of Bridge
of a novel high-performance all-steel fish-bone shaped buckling-
Engineering, 2(4), 48–139.
restrained brace. Engineering Structures, 138, 19–105.
Celik, O. C., & Bruneau, M. (2009). Seismic behavior of bidirectional-
Jia, M., Yu, X., Lu, D., & Lu, B. (2017b). Experimental research
resistant ductile end diaphragms with buckling restrained braces
of assembled buckling-restrained braces wrapped with carbon
in straight steel bridges. Engineering Structures, 31, 93–380.
or basalt fiber. Journal of Constructional Steel Research, 131,
Chaboche, J.L., Dang, V.K. & Cordier, G. (1979). Modelization of the
61–144.
strain memory effect on the cyclic hardening of 316 stainless steel.
Khampanit, A., Leelataviwat, S., Kochanin, J., & Warnitchai, P. (2014).
In 5th International Conference on Structural Mechanics in Reac-
Energy-based seismic strengthening design of non-ductile rein-
tor Technology, Division L11/3, (p.1–10). Berlin: Bundesanstalt
forced concrete frames using buckling-restrained braces. Engi-
fur Material prufung.
neering Structures, 81, 22–110.
Chao, S. H., Karki, N. B., & Sahoo, D. R. (2013). Seismic behavior
Kim, D. H., Lee, C. H., Ju, Y. K., & Kim, S. D. (2015). Subassemblage
of steel buildings with hybrid braced frames. ASCE Journal of
test of buckling-restrained braces with H-shaped steel core. The
Structural Engineering, 139(6), 32–1019.
Structural Design of Tall and Special Buildings, 24(4), 243–256.
Chen, Q., Wang, C., Meng, S., & Zeng, B. (2016). Effect of the
Korzekwa, A., & Tremblay, R. (2009). Numerical simulation of the
unbonding materials on the mechanical behavior of all-steel
cyclic inelastic behavior of buckling restrained braces. London:
buckling-restrained braces. Engineering Structures, 111, 93–478.
Taylor and Francis Group. (ISBN 978-0-415-56326-0).
Chou, C. C., & Chen, S. Y. (2010). Sub-assemblage tests and finite
Kumar, P. C. A., Sahoo, D. R., & Kumar, N. (2015). Limiting values
element analyses of sandwiched buckling-restrained braces. Engi-
of slenderness ratio for circular braces of concentrically braced
neering Structures, 32, 21–2108.
frames. Journal of Constructional Steel Research, 115, 35–223.
Clark, P., Aiken, I., Kasai, K., Ko, E. & Kimura, I. (1999). Design pro-
Lemaitre, J., & Chaboche, J.-L. (1990). Mechanics of solid materials.
cedures for buildings incorporating hysteretic damping devices. In
Cambridge: Cambridge University Press.
Proceedings of 68th Annual Convention of Structural Engineers
Mirtaheri, M., Gheidi, A., Zandi, A. P., Alanjari, P., & Samani, H.
Association of California (SEAOC), Santa Barbara.
(2011). Experimental optimization studies on steel core lengths
Della Corte, G., D’Aniello, M., & Landolfo, R. (2015). Field testing
in buckling restrained braces. Journal of Constructional Steel
of all-steel buckling-restrained braces applied to a damaged rein-
Research, 67, 53–1244.
forced concrete building. ASCE Journal of Structural Engineer-
Pandikkadavath, M. S., & Sahoo, D. R. (2016). Cyclic testing of short-
ing, 141(1), D4014004.
length buckling-restrained braces with detachable casings. Earth-
Dusicka, P., & Tinker, J. (2013). Global restraint in ultra-lightweight
quakes and Structures, 10(3), 699–716.
buckling-restrained braces. Journal of Composites for Construc-
Pandikkadavath, M. S., & Sahoo, D. R. (2017). Mitigation of seismic
tion, 17(1), 50–139.
drift response of braced frames using short yielding-core BRBs.
Eryasar, M., & Topkaya, C. (2010). An experimental study on steel-
Steel and Composite Structures, 23(3), 285–302.
encased buckling restrained brace hysteretic damper. Earthquake
Sabelli, R., Mahin, S., & Chang, C. (2003). Seismic demands on steel
Engineering & Structural Dynamics, 39, 81–561.
braced frame buildings with buckling-restrained braces. Engineer-
Fahnestock, L. A., Sause, R., & Ricles, J. M. (2007). Seismic response
ing Structures, 5, 66–655.
and performance of buckling-restrained braced frames. ASCE
Tremblay, R., Bolduc, P., Neville, R., & DeVall, R. (2006). Seismic
Journal of Structural Engineering, 133(9), 204–1195.
testing and performance of buckling-restrained bracing systems.
FEMA 356. (2000). Prestandard and commentary for the seismic reha-
Canadian Journal of Civil Engineering, 33(2), 98–183.
bilitation of buildings. Washington: Federal Emergency Manage-
Wu, A. C., Lin, P. C., & Tsai, K. C. (2014). High-mode buckling
ment Agency.
responses of buckling-restrained brace core plates. Earthquake
Ghowsi, A. F., & Sahoo, D. R. (2013). Seismic performance of buck-
Engineering & Structural Dynamics, 43(3), 93–375.
ling-restrained braced frames with varying beam-column connec-
Wu, B., & Mei, Y. (2015). Buckling mechanism of steel core of buck-
tions. International Journal of Steel Structures, 13(4), 21–607.
ling-restrained braces. Journal of Constructional Steel Research,
Ghowsi, A. F., & Sahoo, D. R. (2015). Fragility assessment of buck-
107, 61–69.
ling-restrained braced frames under near-field earthquakes. Steel
Zhao, J., Wu, B., & Ou, J. (2011). A novel type of angle steel buckling-
and Composite Structures, 19(1), 90–173.
restrained brace: Cyclic behavior and failure mechanism. Earth-
Halama, R., Sedlák, J., Šofer, M. (2012). Phenomenological model-
quake Engineering & Structural Dynamics, 40, 102–1083.
ling of cyclic plasticity. In P. Miidla (Ed.), Numerical Modelling.
IntechOpen. https​://doi.org/10.5772/35902​.

13