See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303914222

A method for preliminary seismic design and assessment of low-rise

structures protected with buckling-restrained braces

Article  in  Engineering Structures · September 2016

DOI: 10.1016/j.engstruct.2016.05.015

CITATIONS READS

23 539

4 authors, including:

Hector Guerrero Tianjian Ji

Universidad Nacional Autónoma de México The University of Manchester
22 PUBLICATIONS   102 CITATIONS    89 PUBLICATIONS   1,129 CITATIONS   

SEE PROFILE SEE PROFILE

Amador Teran
Metropolitan Autonomous University
99 PUBLICATIONS   955 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Displacement-based preliminary seismic design of tall diagrid systems View project

Elasticity of structures using analytical, semi-analytical and numerical methods View project

All content following this page was uploaded by Tianjian Ji on 19 December 2017.

The user has requested enhancement of the downloaded file.

 Engineering Structures 123 (2016) 141–154

Contents lists available at ScienceDirect

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

A method for preliminary seismic design and assessment of low-rise

structures protected with buckling-restrained braces
Hector Guerrero a,⇑, Tianjian Ji a, Amador Teran-Gilmore b, J. Alberto Escobar c
a
The University of Manchester, UK
b
Universidad Autonoma Metropolitana, Mexico
c
Institute of Engineering, Universidad Nacional Autonoma de Mexico, Mexico

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

Article history: This paper proposes a method for preliminary Performance-Based Seismic Design (PBSD) of low-rise
Received 3 October 2015 structures protected with Buckling-Restrained Braces (BRBs). It is assumed that a frame structure pro-
Revised 10 May 2016 tected with BRBs, termed as a dual structure, is rationally represented by a dual single-degree-of-
Accepted 11 May 2016
freedom (SDOF) oscillator whose parts yield at different displacement levels. The formulation of the
method is presented for SDOF structures. This simplification is validated using a case study example. A
comparison of the responses between conventional and dual structures shows that, when designing dual
Keywords:
structures, the common practice of using conventional design spectra may lead to biased designs. One of
Buckling-restrained braces
Performance-based seismic design
the main advantages of the method is that, during its application, information useful for preliminary and
Performance assessment quick assessment of structures is generated, facilitating the application of the PBSD philosophy. A case
Seismic design method study example is conducted to show its applicability and its potential for preliminary assessment of
Dual systems structures. Regarding its limitations, the method is valid for low-rise regular buildings with rigid
Structural fuse in-plane diaphragms, and whose dynamic response is dominated by their fundamental mode of vibration.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction To reduce damage, protection technologies such as Buckling-

There is great concern about the levels of economical damage
observed after recent earthquakes in structures designed with code
procedures because losses have surpassed expectations by a large
amount [1]. As a result, some strategies have been proposed (a)
to estimate expected damage in a more reasonable way and (b)
to reduce it.
To estimate damage in a realistic and reliable way,
Performance-Based Seismic Design (PBSD) has been proposed to
predict and evaluate the performance of buildings (or facilities)
with clear understanding of risk [1,2]. PBSD is superior to code pro-
visions because it is able to predict different types of losses for dif-
ferent shaking intensities in a probabilistic manner, while codes
mainly intend to provide resistance to avoid collapse without a
clear understanding of risk of collapse or extension of damage
and repair cost [2]. However, implementation of PBSD is often
reserved for critical facilities only, due to the required increase of
engineering design involvement. Procedures that facilitate its
implementation are still needed.
⇑ Corresponding author.
Restrained Braces (BRBs) have been developed and implemented
in structures because they are very effective to dissipate energy
[3,4] and help to control lateral displacements and inter-storey
drifts [5]. Moreover, they can be used as structural fuses; i.e.
devices that concentrate damage and are easy to replace while
the main structure remains undamaged [6]. An attractive solution
is achieved when they are combined with moment-resisting
frames because they allow the reduction of inter-storey drifts [5]
and permanent (or residual) deformations [7,8].
In order to design structures protected with BRBs, methods
based in the control of the response have been proposed recently.
Most of these methods were proposed only for BRB frames, defined
as systems whose lateral resistance is only provided by BRBs while
the contribution of the frame is neglected [9–11]. However, the
contribution of the main structure may represent a significant
amount of capacity and should be taken into account when design-
ing and assessing structures equipped with BRBs. In this regard,
Maley et al. [5], Lin et al. [12] and Sutcu et al. [13] have proposed
methods based in the Direct Displacement-Based Design method-
ology [14]. In this approach, the hysteretic damping provided by
inelastic deformation of the BRBs is replaced by an equivalent vis-
cous damping to convert the nonlinear system into an equivalent
linear system. Vargas and Bruneau [6] proposed a method based

http://dx.doi.org/10.1016/j.engstruct.2016.05.015
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
142 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
on a parametric formulation that considered the contribution of
the BRBs and the main structure. Unfortunately, some key param-
eters such as maximum displacement ductility, stiffness ratio and
strength ratio were selected arbitrarily; which led to limited con-
trol of the design process. Teran-Gilmore and Virto-Cambray [15]
proposed a displacement-based method; which considered the
contribution of the BRBs and the main structure to the capacity.
This method, applicable for structures not significantly affected
by higher modes and flexural behaviour, provides a basis for the
development of the method proposed here. On the other hand, cur-
rent methods in the literature use spectra as design input; which
normally are generated based on elastic-perfectly plastic oscilla-
tors, referred hereafter as conventional oscillators. Since they
behave differently to dual oscillators, this may lead to biased
designs (as will be seen in the next section).
In this paper, a method for seismic design of buildings equipped
with BRBs is proposed. To control the lateral displacement
demands, it takes into account explicitly the parameters affecting
the behaviour of structures equipped with BRBs such as the rela-
tive contribution of the BRBs and the main structure to the lateral
capacity of the dual system. Instead of using design spectra gener-
ated from conventional oscillators, it uses seismic records to solve
the dynamic equation of motion for dual oscillators; which leads to
better estimation of the response. Key parameters such as ductility
are not arbitrarily selected but estimated at the beginning of the
design process as a function of the geometric and mechanical prop-
erties of the structural members. The method facilitates the imple-
mentation of PBSD because it generates statistics of the response
that allows a rapid assessment of the performance. This benefits
designers and stakeholders in making more intelligent decisions
based not only on initial construction costs but also in life-cycle
considerations at the beginning of the design process [1].
The structure of this paper is as follows. Section 2 presents the
idea of designing single-degree-of-freedom (SDOF) structures with
BRBs and compares the response of conventional and dual SDOF
oscillators. Section 3 presents a definition of multi-degree-of-
freedom (MDOF) structures equipped with BRBs and presents a
method for their design. Design and assessment of an example
building are conducted in Sections 4 and 5. Discussion and conclu-
sions are presented, respectively, in Sections 6 and 7.
2. BRBs and SDOF structures
2.1. Buckling-restrained braces
The properties of BRBs have been well documented elsewhere
(e.g. [4]). In this paper only a brief summary is offered. Two types
of BRBs are commonly available: all-steel BRBs (e.g. [16]) and
unbonded BRBs (e.g. [17]). For illustration, a typical unbonded
BRB is shown in Fig. 1a, which consists of two parts: a core and a
case. The core is commonly made of a steel plate which is weaker
in the central zone in order to concentrate the plastic deformation
there. The case is normally made of a steel tube filled with mortar
to restrain the core and avoid buckling due to compressive loads.
An unbonding material is located between the core and the mortar
to avoid direct interaction. As observed in Fig. 1b, when a BRB is
subjected to cyclic axial loads, a stable hysteretic behaviour is
appreciated with slightly higher capacity in compression than in
tension. For simplicity and for illustration purposes, an equivalent
bilinear hysteretic model (Fig. 1c) is used for the BRB to develop
the parameters of the proposed method. However, diverse hys-
teretic characteristics can be used during the design process.
The load capacity of a BRB can be estimated as Pye = fyeA; being
fye and A the expected yielding stress of the composing material of
the core and the corresponding cross-sectional area of the weaker
part. It shall be noted that expected rather than nominal properties
are considered. The yielding displacement can be estimated from:
1 L 1 f ye
dye ¼ P ye ¼ L ð1Þ
f k AE fk E
where E is the modulus of elasticity; L is the total length of the BRB;
and fk is a factor that takes into account the geometry of the core,
that can be established from catalogues of the manufacturers, and
that can take values between 1.2 and 2 or even higher. For the sam-
ple BRB of Fig. 1a, fk can be estimated as [17]: fk = 1/[g(1  c) + c];
being g = A/Aend and c = Ly/L.
2.2. The idea of designing frames with BRBs
First, the differences between frames and frames with BRBs
need to be distinguished. When the beams of a frame yield, a bi-
linear behaviour is usually exhibited. If the columns of that same
frame yield at a later time, it may develop a tri-linear behaviour.
In this situation, the beam and column are structural members of
the frame, and the frame can be treated as a SDOF oscillator with
bi-linear behaviour (in exceptional cases, tri-linear behaviour
may be observed). For a frame with a BRB, the BRB is not necessar-
ily part of the frame. Thus the BRB can be designed in parallel with
the frame because it does not add a new degree of freedom, and the
frame and the BRB form two sets of bi-linear curves – which
together are called dual system.
For convenience, moment resisting frames (MRFs) are referred
hereafter as conventional structures (Fig. 2a) while MRFs equipped
with BRBs are referred as dual structures or dual systems (Fig. 2b).
When a conventional structure is subjected to a major earth-
quake, the structure may be damaged due to large deformation
demands, as shown in Fig. 2a. If a BRB is installed into the structure
(to form a dual structure), it is expected that the BRB would absorb
a good amount of energy and be damaged while the structure
remains in its elastic range of deformation (Fig. 2b). This requires
a rational design of the BRB or of the frame and the BRB. This sub-
section provides the basic idea for such a design.
While the response of a conventional structure can be usually
modelled using a bi-linear single degree-of-freedom (SDOF) oscil-
lator with mass m, damping coefficient c1 and stiffness k1, that of
a dual structure can be modelled using a dual SDOF oscillator con-
sisting of the SDOF oscillator and a secondary part (representing
the device) with damping coefficient c2 and stiffness k2, as shown
in Fig. 3a. Fig. 3b illustrates the load and deformation capacities
of the dual oscillator in which the primary part starts to yield at
dy1 when subjected to a force Vy1 and the secondary part yields
at dy2 when it experiences a force of Vy2. The combined capacity
of the dual oscillator is illustrated in a dashed line in Fig. 3b. Then,
the question is how to select properly the properties of the main
structure (dy1 and k1, and hence Vy1) and those of the BRB (dy2
and k2, and hence Vy2) that allow controlling satisfactorily the dis-
placement demands induced by earthquake actions while ensuring
that the BRB yields first.
In this paper, it is proposed that the moment resisting frame is
initially designed under the condition of gravity loads. This pro-
vides the initial values of the primary part (i.e. dy1 and the lower
limit of k1). Then, one of the next approaches may be followed to
control the displacement demands: (1) by fixing the initial values
of dy1 and k1, find the values of dy2 and k2; or (2) for a desired pro-
portion of the BRB to the load capacity, find the properties of both
the primary and secondary parts. Both approaches are addressed in
the following subsections.
 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154 143

Steel tube
Lend Ly Lend
C
Aend Aend mortar
A
unbonding
core case C’ core material
C-C’
(a) A typical BRB
P
Ptu P
Pye Pye

d
d
dye
dye

|Pcu|>Ptu
(b) Typical load-displacement response (c) Assumed hysteretic model
Fig. 1. A typical BRB.

Δ d<Δ

α
h h
üg(t) üg(t)
L L

(a) Conventional structure (b) Dual structure

(damage in the structure) (damage focused in the BRB)
Fig. 2. Conventional vs. dual structures.

V
Primary characteristic is convenient for designing purposes because any
k1 Total capacity k2 and Vy2 can be provided without affecting dy2.
c1 (VyT = Vy1 + Vy2 ) Now, for the primary and secondary parts of a dual SDOF oscil-
VyT lator, let the load capacity factors b1 and b2 be, as defined by
m Vy1 Fig. 3b:
Secondary
k2 k1 V y1 V y2
c2
b1 ¼ and b2 ¼ ð3Þ
Vy2 V yT V yT
k2
d and the capacity ratio
u g (t ) dy2 dy1 du
b2 V y2 k2 dy2 dy1 b1 k2
(a) A dual SDOF oscillator (b) Capacity curves of the ¼ ¼ or ¼ ð4Þ
b1 V y1 k1 dy1 dy2 b2 k1
dual oscillator
where b1 + b2 = 1. Since dy1 and k1 are considered being known and
Fig. 3. A dual SDOF oscillator and its behaviour curves. dy2 is determined using Eq. (2), the stiffness of the secondary part,
k2, is calculated with Eq. (4) as long as its relative contribution
(b2) is known. Since b2 is unknown, an iterative process is required
2.2.1. Designing BRBs for known properties of the primary structure to find its value that allows controlling the lateral displacement
First, the lateral yielding displacement of a BRB (dy2), as demands.
installed in a frame, can be calculated as follows [15]: The iterative process starts with the estimation of the ductility
factors of the primary and secondary parts, when the dual oscilla-
1 fy h tor experiences the maximum displacement dmax:
dy2 ¼ ð2Þ
f k E cos a sin a
dmax dmax
l1 ¼ and l2 ¼ ð5Þ
dy1 dy2
where the parameters h and a are defined in Fig. 2 and are consid-
ered constant. All other parameters have already been defined in then, using Eq. (4), the ratio of the ductility factors is
Eq. (1). It should be noted that, if for selected values of fk and the
l2 dy1 b1 k2
material properties (fy and E), dy2 were higher than dy1, increasing ¼ ¼ ð6Þ
fk may be very effective to ensure that the BRB yields first (i.e. to
l1 dy2 b2 k1
achieve the fuse concept). Eq. (6) can be re-written as:
By analysing Eq. (2), it can be observed that once fk and
k2 l2 b2
the material properties are selected, the yielding displacement of a¼ ¼ ð7Þ
the BRB, dy2, is constant and independent of k2 and Vy2. This k1 l1 b1
144 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
Eq. (7) provides the formula to determine the stiffness ratio of
the secondary part to the primary part, a. This equation is simple
for a dual SDOF oscillator, but it presents the idea that the stiffness
of a BRB can be estimated from the stiffness of the frame structure
once the value of a is known. This is possible if a value of b2 is
selected and the ductility factors (l1 and l2) are calculated based
in a displacement threshold (dmax) and the yielding displacements
of the primary and secondary parts (dy1 and dy2). Note that if
l2 > l1, then the fuse concept is guaranteed because the BRB yields
first as illustrated in Fig. 3b.
For convenience, the stiffness ratio (a) is also used to relate the
periods of the primary part (T1) and the secondary part (T2), as
follows:
k1 2 T 21
T 22 ¼ T ¼
k2 1 a
Therefore, once a and T1 are known, the period T2 can be deter-
mined. In consequence, the total period of the dual SDOF oscillator
(T) is calculated as:
1 1 1
¼ þ
T2 T 21 T 22
Then, the period T is used as input to estimate the dynamic
response of dual oscillators using the dynamic equation of motion,
which is as follows:


2p _
f ðu; uÞ _
f ðu; uÞ
€ ðtÞ þ 2ðn1 þ n2 Þ
u _
uðtÞ þ s1 þ s2 € g ðtÞ
¼ u
T m m nations of the parameters of Rows 1 and 2 in Fig. 4. Comparison
where u; u; _ and u€ are the displacement, velocity and acceleration;
€ g is the acceleration of the ground, n1 and n2 are the damping ratios
u
of the primary and secondary parts of the dual oscillator; T is the
period estimated with Eq. (9); m is the modal mass; f s1 ðu; uÞ _ and
_ are the restoring forces of each part of the dual oscillator
f s2 ðu; uÞ
(they depend on the history of displacements, the sign of the veloc-
ity, and the yielding load capacity, which may be determined using
Eqs. (11) and (12)). In this part, it should be noted that the restoring
forces do not necessarily have to follow the bilinear hysteretic
model presented in Fig. 1c. While a bilinear model is recommended
to model the behaviour of the secondary part (i.e. the BRB), a cus-
tomised model may be required for the primary part (e.g. Takeda
model for concrete frames, bilinear model for steel frames, and
flag-shape model for structures with self centring capacity). Fur-
thermore, P–D effects can be taken into account in the force–dis-
placement relationship of the primary part. This will be further
described later. Note that the restoring forces and capacity loads
are divided by the modal mass; therefore, as in conventional oscil-
lators, m does not need to be known to solve Eq. (10). Also note that
_ are removed from
for the conventional oscillator, n2 and f s2 ðu; uÞ
Eq. (10).
 2
V y1 2p
¼ dy1
m T1
 
V y2 V y1 b2
¼ ð12Þ
m m b1
The response of dual oscillators is highly affected by b1 and b2
(i.e. the relative contribution of the primary and secondary parts
to the load capacity). However, b1 = 1  b2. Therefore, by trying dif-
ferent values of b2, the maximum displacement demands (umax)
estimated with Eq. (10) can effectively be controlled. Once the dis-
placement demands are deemed satisfactory, the value of k2 and
other parameters (e.g. Vy2, T2 and T) are readily available from
Eqs. (4)–(9).
Note from Eqs. (7)–(9) that not only b2 but also other parame-
ters (such as the ductility factor of the BRB, l2) affect the response
of a dual oscillator. Therefore, umax can also be controlled by pro-
viding different values of l2; which, in turn, can be varied by
selecting different values of fk and (if commercially available) prop-
erties of the materials. Thus, there is a significant range of options
to control umax. However, special attention shall be paid to avoid
very large values of l2 that could result in unexpected failures of
BRBs. Form experimental tests, a condition of l2 < 10 might be
deemed appropriated as summarised by Fahnestock et al. [18].
A summary of the parameters that affect the response of dual
oscillators is shown in Fig. 4. The input parameters that can be
selected to control the response are seen in Row 1 while Rows 2
and 3 show the parameters that are considered fixed in this sub-
section. Note that by solving Eq. (10) not only umax but also other
response parameters, useful for preliminary assessment of struc-
ð8Þ tures, are estimated. This benefit is addressed in Section 5.
2.2.2. Designing BRBs for a desired contribution to the load capacity
It can be appreciated from Fig. 4 that, in order to control the
maximum displacement demands (umax), when a fixed contribu-
tion of the BRB to the load capacity (b2) is required, different values
ð9Þ of fk and (if available) fy and E may be tried. However, if these
parameters do not allow a satisfactory control of the response,
the parameters of Row 2 (dy1 and T1) can also be modified by pro-
viding more capacity to the elements of the primary part. An iter-
ative process, similar to that in the previous subsection, may be
required up to find a satisfactory control of umax. Furthermore, dif-
ferent design scenarios can be obtained by using different combi-
ð10Þ
between scenarios might help designers to achieve efficiency, e.g.
designs with similar displacement demands and less construction
cost, or designs with similar construction cost but with better
behaviour under the same seismic demand.
2.3. Conventional vs. dual SDOF oscillators
Since current practices of design mostly use design spectra
developed from elastic-perfectly plastic SDOF oscillators (referred
hereafter as conventional oscillators), an extensive parametric
study was carried out on a conventional and a dual oscillators to
understand the differences of their response and to show how
the use of input generated from conventional oscillators may, in
many cases, lead to biased designs. Moreover, it will be also shown
that the proposed design method can be very useful to control the
displacement response of dual structures, for example, to maintain
the primary structure elastic while the BRB yields.
The oscillators were subjected to a number of earthquake
ground motions recorded in the lakebed zone of Mexico City. How-
ever, for simplicity only representative results are presented. It is
highlighted that both oscillators had the same period of vibration
and damping ratio; and their response was determined for a target
ductility of seven. The conventional oscillator, which had elastic-
ð11Þ perfectly plastic behaviour, had the following properties: T = 0.5 s
and n = 0.05. The dual oscillator had also a total damping of
n1 + n2 = 0.05 and a period of T = 0.5 s, which included the period
of the primary part (T1 = 0.87 s) and that of the secondary part
(T2 = 0.61 s). The other design parameters were chosen as
b1 = 0.8, b2 = 0.2 and a = 2 so that, according to Eq. (7), l2 = 7 and
l1 = 0.87 (i.e. the primary part shall remain elastic). It should be
noted that, even when the conventional and dual oscillators had
the same period of vibration, their post-elastic behaviour was sig-
nificantly different because the parts of the dual oscillator yielded
at different displacement levels.
The response of the oscillators to the SCT-EW record of the 1985
Michoacan, Mexico Earthquake is shown in Fig. 5; where the dis-
placement and strength demands are compared. It can be observed
from the figure that:
 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154 145

Primary structure Secondary structure Dual SDOF system

Row 1: b2 fk fy E

parameters
Input
Row 2: T1 dy1

Row 3: ξ1 ξ2 dmax h, α u g (t )

1 fy h
d y2 = (2)
f k E cos α sin α
d max μ2 b2
μ1 = (5a) d max a= (7)
d y1 μ2 = (5b) μ1 b1
d y2
Intermediate
calculations

b1 = 1 − b2 T22 = T12 / a (8)

2
⎛ 2π ⎞
V y1 Vy 2 ⎛ Vy1 ⎞ b2 1 1 1
=⎜ ⎟ d y1 (11) =⎜ ⎟ (12) = + (9)
m ⎝ T1 ⎠ m ⎝ m ⎠ b1 T 2 T12 T22

2π ⎡ f (u , u ) f s 2 (u, u ) ⎤
u (t ) + 2(ξ1 + ξ 2 ) u (t ) + ⎢ s1 + = −u g (t ) (10)
T ⎣ m m ⎥⎦
response

Displacement demand: umax

Output

Other demand parameters: umax umax uresidual μ1 μ2

Fig. 4. Parameters affecting the response of dual oscillators and design for BRBs.

60
Dual Coventional
displacement,
mm

Difference 53%
Time, sec
-60
20 30 40 50 60 70 80 90 100
2.5
Force demand/ m,

Dual Coventional
N/kg

Difference 49% Time, sec

-2.5
20 30 40 50 60 70 80 90 100

Fig. 5. Responses in the time domain of a conventional and a dual SDOF oscillator.

 Even when the oscillators had the same period of vibration and experience large inelastic deformation when subjected to an earth-
ductility demand, the peak displacement of the conventional quake ground motion.
oscillator is 53% larger than that of the dual oscillator.
 The conventional oscillator experienced a larger residual dis- 3. BRBs and MDOF structures
placement while the dual oscillator almost returned back to
the original position. 3.1. Definition
 To reach l = 7 and l2 = 7, respectively, the conventional oscilla-
tor required 49% less force capacity than the dual oscillator. Similar to a one-storey dual structure, a low-rise MDOF struc-
ture equipped with BRBs can be modelled as a dual SDOF oscillator
The results of the comparative study indicate that: (a) conven- when its response is dominated by its fundamental mode of vibra-
tional and dual oscillators behave differently – therefore, the use of tion. Fig. 6a illustrates a rigid frame structure with BRBs installed,
conventional oscillators to generate input design spectra may lead which can be decomposed into two independent substructures: a
to biased designs; and (b) through an appropriate design of the MRF without braces (Fig. 6b) and a pin-connected braced frame
stiffness ratio (a) and resistance distributions (b1 and b2) in a dual (Fig. 6c), in which the lateral stiffness is completely contributed
oscillator, the response can effectively be controlled for example, by the brace members. The two decomposed substructures can
to maintain the primary part elastic while the secondary part can be represented as a dual oscillator (Fig. 6d).
146 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154

( ) rigid connection ( ) pinned connection

Primary
k1
h3
c1
α3

h2 = + m
α2 Secondary
k2
c2
h1
α1 u g (t )
(a) Dual MDOF structure (b) Primary sub-structure (c) Secondary sub-structure (d) Dual SDOF structure
(structure equipped (structure designed (it is assumed that the lateral (Simplistic representation
with hysteretic dampers) without devices) resistance is only provided by of an equipped structure)
the devices)

Fig. 6. A MDOF structure equipped with BRBs represented by a dual SDOF oscillator.

The representation of a structure by two independent substruc- independent linear displacement profile has been assumed, which
tures is based on the superposition principle; in which, the stiff- may be reasonably acceptable for regular, low-rise MRFs that pre-
ness matrix (K) of the structure in Fig. 6a can be estimated as the sent beam-sway mechanism [14]. This profile has been selected
summation of the stiffness matrix of the MRF (KMRF) in Fig. 6b here for preliminary design of MRFs equipped with BRBs because,
and the stiffness matrix of the braced pinned frame (KBS) in Fig. 6c: as expressed in [5], the MRF is expected to dictate the displacement
shape due to relatively low-yielding displacements and low post-
K ¼ KMRF þ KBS ð13Þ
yielding stiffness of BRBs. For cases where the expected displace-
The fundamental periods of the system, T, the MRF, T 1 , and the ment profile may vary significantly with increasing lateral deforma-
braced system, T 2 , have the following relation: tion, the floor displacements (di) shall be estimated accordingly. The
displacement profile obtained in Step 2 by pushover analysis may
1 1 1
2
¼ þ ð14Þ provide a good approximation.
T T 21 T 22 Step 4: Decide the requirement for BRBs. For this purpose, a con-
It can be noted that the dual oscillator converted from the ventional SDOF oscillator with period T1 and equivalent yielding
MDOF structure is the same as that of the previous section. How- load capacity of the primary structure (see Eq. (11)) is subjected
ever, a strategy for PBSD of low-rise buildings needs to be devel- to possible ground motions. If, from the pushover analysis of Step
oped for designing the dual oscillator and the actual structure. 2, it is observed that the post-yielding stiffness ratio (r) is signifi-
cantly different from zero, an approximate r must be used here.
Then, the maximum displacement demands and thresholds are
3.2. Proposed procedure for preliminary design of low-rise buildings
compared, which determines whether BRBs are needed. If, for
example, the mean plus one standard deviation of the demands
A procedure for preliminary PBSD of low-rise buildings is
is larger than their respective thresholds, BRBs need to be designed
proposed and summarised in Fig. 7. The key steps are explained
according to the subsequent steps; otherwise, the process is fin-
as follows:
ished. It is worth to mention that the mean plus one standard devi-
Step 1: Select the objectives of design. For each seismic intensity
ation demand is selected arbitrarily to illustrate the application of
level, a performance level should be established in terms of the
the methodology; however, any other demand level could be
inter-storey drift threshold (e.g. drift limit for: Fully Operability,
selected. It is also recognised that inter-storey drifts are not uni-
hFO; Operability, hOp; Life Safety, hLS; and Collapse Prevention, hCP).
form through the height of the building; thus, the displacements
Step 2: Design the primary structure (moment resisting frame
demands should be amplified by a factor that takes into account
without braces) under gravity loads. Thus, the fundamental period
this matter. Similar to [15] and based on observations by [20], a
of vibration (T1) of the frame can be calculated and the yielding dis-
factor of 1.20 is recommended for structures equipped with BRBs
placement (dy1) of the frame at the top floor may be obtained from a
and 1.50 is for structures without BRBs. These factors change to
pushover analysis. If P–D effects are not negligible (as defined in
1.50 and 2.0 respectively for ductility demands higher than 2.0.
FEMA-356 [19]), they have to be taken into account in the pushover
Step 5: Estimate l1max and l2max. This is explained later in
analysis.
Section 3.2.1.
Step 3: Determine the displacement thresholds. There should be
Step 6: Select the relative participation of the secondary sub-
one threshold for each performance level under consideration.
structure (0 6 b2 6 1). Then, the required periods of vibration of
First, the displacement at the ith floor, di, is estimated as the pro-
the dual system (T2 and T) and the force capacity of the secondary
duct of the inter-storey drift threshold established in Step 1 and
substructure (Vy2/m) are determined using the period of vibration
the height of that floor. Thus, the displacement threshold, dmax,
of the primary substructure (T1) determined in Step 2, the ductility
for an equivalent SDOF structure is estimated as given by Priestley
demands estimated in Step 5, and Eqs. (7)–(9) and (12). It is recog-
et al. [14]:
, nised that T2 may be affected by the axial deformation in the col-
X
N
2
X
N umns of the bracing system, in particular for a high participation
dmax ¼ mi di mi di ð15Þ of the secondary substructure. This effect is to be discussed in
i¼1 i¼1
Section 3.2.3.
where N is the number of storeys; mi is the mass in the ith storey. Step 7: Estimate the maximum displacement demands (e.g.
Eq. (15) converts the displacement thresholds for all storeys into d50%/50y, d20%/50y, d10%/50y, d2%/50y) of an equivalent dual SDOF oscil-
a single value for the dual SDOF oscillator. Note that an intensity- lator using Eq. (10). As in Step 4, the displacement demands are
 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154 147

1. Select the objectives of 5. Estimate μ1max and μ2max from 3.2.1

design (θFO, θOp , θLS , θCP)

6. Select b2 and determine T2 , T, and

Vy2 /m
2. Design the main structure
for gravity loads and
determine T1 and dy1 7. Using records, determine the displa-
cement demands (d50%/50y, d20%/50y,
d10%/50y, d2%/50y) on a dual SDOF model
3. Determine the displace-
ment thresholds (dFO, dOp ,
dLS , dCP) 8.
Is the response No
satisfactory?

4. Are Yes Yes

BRBs 9. Determine the required cross-sectional
needed? areas of the BRBs from section 3.2.2

No
END

Fig. 7. Algorithm for PBSD of structures equipped with BRBs.

estimated as the mean plus one standard deviation of the
responses obtained from multiple actual or artificial hazard-
compatible records. Also, the displacements demands should be
amplified by the non-uniform inter-storey drift factor (see Step 4).
Step 8: A few iterations between Steps 6 and 7 may be required
to achieve the most convenient balance between the relative par-
ticipation of the protection substructure (b2) and the displacement
demands. If the response is still not satisfactory, recommendations
of Sections 2.2.1 and 2.2.2 can be useful.
Step 9: Once the response is deemed satisfactory, determine the
required cross-sectional areas of the BRBs. This is detailed in
Section 3.2.2.
Some of the steps are explained further in the following
sub-sections.
3.2.1. Estimation of maximum ductility ratios in the dual structure
The maximum ductility on the primary substructure is defined
as the ratio of the top-floor displacement at the maximum thresh-
old (corresponding to the highest performance level, e.g. Collapse
Prevention) to the yielding displacement estimated in Step 2, i.e.:
dN
l1max ¼
dy1
The maximum ductility on the secondary substructure (l2max)
is estimated as the average of all inter-storey ductility ratios on
the BRBs for the highest performance level, i.e.
1X N
1XN
Dmax i
l2max ¼ l ¼
N i¼1 i N i¼1 Dy2i
where li is the inter-storey ductility ratio at the ith storey and
Dmax i ¼ hmax hi
1 fy hi
D2yi ¼
f ki E cos ai sin ai
being hmax the maximum allowed inter-storey drift. Note that
Eq. (19) is similar to Eq. (2); however, Eq. (19) takes into account
the properties at each storey of the building. It is also of interest
to note that the ductility at each storey cannot be supposed and
assigned directly to the structure because it depends on, and shall
be estimated from, the yielding displacement capacity of the BRBs.
3.2.2. Estimation of the required cross-sectional areas of the BRBs
The cross-sectional areas of the yielding zone of the BRBs are
determined in two steps: (1) estimating the relative areas of all
BRBs through a static analysis, and (2) calculating the absolute val-
ues through achieving the required period of vibration of the sec-
ondary substructure (T2).
1. Relative areas: lateral loads proportional to the floor mass are
applied to the braced pin-connected model (Fig. 6c). Thus the
axial loads (Nj) in the braces of the statically determinate frame
can be easily determined. Then, with an initial area of 1.0 cm2
assigned to the braces in the ground storey, cross-sectional
areas are provided to all the braces proportionally to the axial
loads. These areas are denoted Aassu. For practical applications,
some braces in consecutive storeys may have the same areas.
2. Absolute areas: The response of the pin-connected substructure
ð16Þ is considered being contributed by the deformations of the
beams, columns and braces. However, deformations in
the beams may be neglected due to rigid-diaphragm action of
the floor system. Thus, as schematically represented in Fig. 8
on a single-storey single-bay pinned frame, the total lateral
deformation of the structure may be considered to consist of
ð17Þ two components: (1) the lateral deformation produced by the
axial deformation of the columns while the beams and braces
are infinitely stiff (Fig. 8b); and (2) the lateral deformation due
to braces while the beams and columns are infinitely stiff
ð18Þ (Fig. 8c). Note that, for convenience, the stiffnesses of the braces
and columns are described by two springs in series (Fig. 8d).
First, consider a case in which the columns are infinitely stiff in
ð19Þ such a way that their axial deformation is negligible. In this case,
the response of the structure is contributed only by the braces,
and the corresponding fundamental natural period (T2,assu) of
the artificial frame with assumed areas for the BRBs (Aassu) can
be determined.
148 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
u ucols
= +
(a) Total (b) Contribution (c) Contribution
deformation of columns
Fig. 8. Deformation of pin-connected structure as contributed by columns and braces.
As the stiffness of the artificial frame is controlled by the areas
of the bracing members, the areas should be modified to reach the
desired period T 2 , using the following equation:
K2 T 2assu
ABRB;i ¼ Aassu;i ¼ Aassu;i
K 2;assu T 22
Now, the updated areas of BRBs will make the secondary sub-
structure to have the required period of vibration T 2 . For fabrica-
tion, the above areas of the yielding zone of the BRBs and the
initially-selected effective stiffness factors (fki) are provided to
the manufacturer of the devices to produce the actual geometry
and shapes.
3.2.3. Correction for axial deformation in columns
In the previous subsection, the effect of axial forces and defor-
mation in the columns of the secondary structure is not consid-
ered. Considering that the stiffnesses or flexibilities of the
columns and braces of the structure can be represented by two
springs in series (Fig. 8d), the flexibility of one of the two can be
evaluated by treating the other infinitively stiff. In other words,
the total flexibility of the secondary substructure can be repre-
sented by:
T 22 ¼ T 2cols þ T 2braces
Now, making the bracing members infinitely stiff and calculat-
ing the natural period of the revised secondary substructure, T col ,
the flexibility of the columns is reflected.
Then, the following ratio of the periods can be calculated:
ratio ¼ T 2cols =T 22
If the ratio is less than a given tolerance (say 5%), this test is fin-
ished and the procedure may continue. Otherwise,
1. The ductility of the secondary part (l2max) is updated by multi-
plying it by the factor:
T 2braces
T 2cols þ T 2braces
which is less than unity when the flexibility of the columns is
not negligible. T 2cols is the fundamental period of the pin-
connected substructure when the braces are treated infinitively
stiff and represents the flexibility of the columns.
T 2braces ¼ T 22  T 2cols is the fundamental period of the pin-
connected substructure when the columns are treated as infini-
tively stiff and represents the flexibility of the braces alone. Note
that the contribution of beams has not been considered because
they are treated as infinitively stiff due to the rigid diaphragm
action of the floor system.
It is highlighted that, for safety purposes, the flexibility of
the braces should be higher than that of the columns, i.e.
T 2braces > T 2cols . If this condition is not satisfied, re-designing of
the columns is advisable considering higher cross-sectional
areas. Teran-Gilmore and Coeto [21] recommend a maximum
ratio T 2cols =T 2braces of 1/3.
ubrace
kcols kbraces m u(t)
(d) Spring model
of brace
2. The period of vibration of the bracing system (T2) is estimated
using the updated ductility demand of the secondary part
(l2max) and Eqs. (7) and (8). A few iterations are required
between the previous step and this one up to T2 stabilises.
ð20Þ
3. Then, the fundamental period of vibration of the whole dual
structure is found using Eq. (9). The procedure may continue
normally.
4. Example of design
In order to verify the applicability of the proposed method, a
five-storey building is designed with buckling-restrained braces.
It is assumed that the structure is located in the lakebed zone of
Mexico City and will locate a hospital.
4.1. Description and requirements of design
The 2D steel framed building is shown in Fig. 9 and is designed
according to the following requirements: for each objective of
design the maximum inter-storey drifts are shown in Table 1;
and the primary frame must remain elastic for the Life Safety per-
ð21Þ
formance level. The masses are 461 t on the top floor and 576 t on
the others. The steel of beams and columns is ASTM A992
(fy = 350 MPa) and that of the braces is ASTM A36 (fy = 250 MPa).
Since the proposed method requires seismic records as input,
the 30 records used by Bojórquez and Ruiz-García [22] selected
from the Mexican Database of Strong Motions [23] are used here
ð22Þ because they were recorded in the assumed location of the build-
ing. The records are pga-scaled in order to account for different
seismic hazard levels. For a seismic hazard with probability of
50% of being exceeded within 50 years, it is assumed that
pga = 0.05g; for 20% in 50 years, pga = 0.10g; for 10% in 50 years,
pga = 0.20g; and for 2% in 50 years, pga = 0.30g. Although the pga
may not be considered the best parameter to indicate the seismic
ð23Þ intensity levels, this will not affect the objective of the study.
4.2. Design with the proposed procedure
The steps proposed in Section 3.2 are followed and summarised
as follows:
5@3m
4m
6@8m
Fig. 9. Five-storey framed building equipped with BRBs.
 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
Table 1
Maximum inter-storey drifts for each objective of design of the study example.
Objective of design 1 2
Performance F. Operability
Seismic intensity, pga 0.05g
Max. drift 0.0025
Table 2
Properties of the design options.
Option b2 (%) dy1 (m) T1 (s) T2 (s) Tcols (s)
a 60 0.15 1.74 0.90 0.48
b 30 0.12 1.15 1.08 0.32
c 40 0.13 1.29 0.97 0.35
Step 1. Selection of the objectives of design. They are shown in
Table 1 in terms of inter-storey drifts limits, namely:
hFO = 0.0025; hOp = 0.005; hLS = 0.01; and hCP = 0.02.
Step 2. Designing the primary structure under gravity loads. The
primary structure was initially dimensioned for gravity loads.
The resultant T1, dy1 and steel profiles are shown in the first row
of Tables 2 and 3 (option a).
Step 3. Calculation of displacement thresholds. According to Eq.
(15), they were dFO = 0.029 m; dOp = 0.058 m; dLS = 0.115 m; and
dCP = 0.230 m.
Step 4. Determination of requirement for BRBs. A conventional
SDOF oscillator, with period of T1 = 1.74 s and Vy1/m = 1.38 N/kg,
was subjected to the 30 records scaled to the pgas of Table 1. Typ-
ical damping ratios were considered, namely: n = 3% for pga = 0.05g
and 5% for the others. Fig. 10a shows that the structure requires
BRBs because displacement demands (represented by dots) were
larger than the displacement thresholds (dash-dot lines).
Step 5. Estimation of threshold ductility factors. BRBs with an
effective stiffness factor of fk = 1.50 were introduced. The ductility
ratios, corresponding to the Collapse Prevention performance level,
were: l1max = 2.18 and l2max = 7.48. It shall be noted that these
ductility ratios are thresholds (i.e. maximum allowed), the actual
ductility demands will be determined later in subsequent steps.
Step 6. Selecting a relative participation of the secondary substruc-
ture. A value of b2 = 30% was selected initially as a first trial.
Step 7. Estimation of the maximum displacement demands. Mean
plus one standard deviation of the maximum displacements were
estimated on an equivalent dual SDOF oscillator using Eq. (10)
and the 30 ground motions previously described. Higher values
of damping ratio were considered here for dual oscillators because
experimental evidence suggests that BRBs increase the damping
ratio significantly due to friction produced by the interaction
between the core and case of the BRBs (e.g. [24,25]). They were
n = 5%, 8%, 10%, and 12% for the seismic intensities of Table 1. Note
that values higher than 10% were used for the larger seismic inten-
sities because large inelastic deformations are expected at those
seismic levels. The results showed that the displacement demands
were larger than their corresponding thresholds.
Step 8. Determining if the design is satisfactory. Since b2 = 30% was
insufficient, the method allowed different options. Three of them
were developed here to find a satisfactory response, namely:
option (a) for fixed properties of the primary structure (T1 and
dy1), b2 was increased to 60% (Fig. 10b); option (b) for a fixed con-
tribution of the protection structure of b2 = 30%, the capacity of the
primary structure was increased; and option (c) a combination of
the previous options (i.e. b2 = 40%). The results are summarised
in Tables 2 and 3. It is observed that a participation of the BRBs
of b2 = 60% requires much less capacity of the primary structure
(Vy1) and less usage of structural steel than b2 = 40% and 30%.
149
3 4
Operability Life safety Collapse prevention
0.10g 0.20g 0.30g
0.005 0.010 0.020
Tbrbs (s) T (s) Vy1/m (N/kg) Vy2/m (N/kg) VyT/m (N/kg)
0.76 0.80 1.38 2.07 3.45
1.03 0.79 2.62 1.12 3.74
0.90 0.77 2.22 1.48 3.70
Step 9. Determination of required cross-sectional areas of the BRBs.
They were estimated according to Section 3.2.2 and are shown in
Table 3 for the first storey. Those of the other storeys are propor-
tional to the following vector: (1.0, 0.76, 0.56, 0.36, 0.16)T.
4.3. Validation of the design by nonlinear analyses
4.3.1. Static analysis
For validation purposes, a pushover analysis (with constant
gravity loads and mass-distributed lateral loads) of the structure
designed in the previous subsection was conducted with and with-
out BRBs using Opensees [26]. All the beam-column connections
were modelled as rigid and the connections of the BRBs to the
frame were modelled as pinned. The model was fully fixed to the
base. The gravity loads were modelled by means of distributed
loads in beams. The floors were considered as rigid diaphragms.
The structural elements (columns, beams and BRBs) were mod-
elled using distributed plasticity elements, i.e. elements with
force-based formulation with five points of integration. These ele-
ments were used because they remove some limitations of other
elements [27] and are time-efficient. Flexural and shear deforma-
tion of the elements was considered. The cross-sections were mod-
elled with fibre elements. Although the stability coefficient,
determined using the recommendations of FEMA-356 [19], sug-
gested that P–D effects could be neglected, they were considered
using the concept of the corotational formulation [27,28]. A con-
crete slab with effective thickness of 90 mm and width of 2 m
was included. Regarding the materials, expected rather than nom-
inal resistances were modelled. The concrete of the slab possessed
an expected resistance of f0 c = 31.25 MPa and a modulus of elastic-
ity of Ec = 21,925 MPa. The steel was modelled using the Steel02
material (as defined in Opensees [26]). This material corresponds
to the Giuffre–Menegotto–Pinto model with strain-hardening.
Similar to [28], the strain-hardening ratio was considered equal
to 0.3%. The other parameters controlling the transition from elas-
tic to plastic response were R0 = 20, cR1 = 0.925, cR2 = 0.15, a1 = 0,
a2 = 1, a3 = 0 and a4 = 1. For the BRBs, a calibration of the parame-
ters a1 and a3 was carried out to account for their behaviour
observed in experiments [3]. They resulted a1 = 0.07 and
a3 = 0.05. Comparison of the calibration is shown in Fig. 11. The
experimental and numerical responses are very similar.
The results of the pushover analysis are shown in Fig. 12 for
b2 = 60% (i.e. design option (a) of the previous section), where the
horizontal axis indicates the displacement at the top floor and
the vertical axis represents the shear capacity normalised by the
total weight (WT) of the structure. By comparing
Figs. 12 and 10b, it can be said that the equivalent dual SDOF oscil-
lator and the MDOF structure equipped with BRBs display good
agreement. Notwithstanding the similarities, some particularities
150 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154

Table 3
Structural elements of the design options.

Option b2 (%) Columns Storey location Beams Storey location Steel weight (kg) ABRB (cm2)
a 60 HSS16  16  1/2 1 and 2 W14  61 1–5 35,635 100
HSS16  16  3/8 3–5
b 30 BOX500  25 mm 1–3 W18  65 1–3 57,283 61
BOX500  13 mm 4 and 5 W14  61 4 and 5
c 40 BOX500  19 mm 1 and 2 W16  67 1 and 2 48,952 78
BOX500  13 mm 3–5 W14  61 3–5

Capacities: Primary Secondary Total

Thresholds: FO limit Op limit LS limit CP limit
Demands: 50/50 20/50 10/50 2/50
120%

Relative participation
0.2 100%
0.2 80%
V1 / m, N/kg

0.1 60%
40%
0.1
20%
0.0
0.00 0.20 0.40 0.60 0.80 0%
Displacement, m 0 0.05 0.1 0.15 0.2 0.25
Displacement, m
(a) Conventional SDOF oscillator (Step 4) (b) Dual SDOF oscillator (Step 8)
Fig. 10. Displacement demands.

are observed in Fig. 12: (1) the transition from elastic to plastic
response is smooth; and (2) the post-yielding stiffness ratio is dif-
2.0 Experimental
Numerical ferent from zero. These differences may be however considered
1.5 negligible for preliminary design purposes.
Normalised axial force

1.0
0.5 4.3.2. Dynamic analysis
0.0 The model described in the previous subchapter was subjected
to the same seismic records used during the design. For consis-
-0.5
tency, they were scaled to the same intensities of design (or pga)
-1.0
of Table 1. In total, the model was subjected to 120 analyses, in
-1.5 which damping ratios of n = 5%, 8%, 10% and 12% were considered
-2.0 for each seismic intensity level analysed, respectively. It should
-0.03 -0.02 -0.01 0 0.01 0.02 0.03 be noted that, to be consistent with the design, these damping
Brace strain, m/m ratios are the same values used in Step 7 to estimate the response
of an equivalent dual SDOF oscillator. Also note that damping
Fig. 11. Calibration of Giuffre–Menegotto–Pinto steel model to model BRBs.
ratios higher than 10% were considered for the larger seismic
intensities because large inelastic deformations were observed
during the design stage and here. The damping ratios are in agree-
ment with experimental evidence that BRBs increase the damping
ratios and they are intensity-dependent (e.g. [24,25]). As recom-
Total
mended by [29], viscous damping was assigned proportional to
0.4 Primary tangent stiffness in columns, beams, and braces in order to avoid
Secondary artificial damping forces generated during nonlinear analyses.
P–D effects were considered as described in the previous section.
0.3
The Newmark method of integration was used assuming constant
V/ WT

average acceleration.
0.2 First and for comparison purposes, Fig. 13 shows the displace-
ment response at the top floor of the MDOF model and that of an
0.1 equivalent dual SDOF oscillator to one of the simulated ground
motions (i.e. the SCT-EW record scaled to pga = 0.3g). The response
of the dual oscillator is amplified by a factor of (dN/dmax) as defined
0 in section 3.2 so that they can be compared. It is observed that the
0 0.05 0.1 0.15 0.2 0.25 0.3
responses are very similar.
Displacement at top floor, m
Now, considering the 30 ground motions, Fig. 14 shows the
Fig. 12. Pushover analysis of the designed structure equipped with BRBs for maximum drift demands at each storey of the model at the four
b2 = 60%. seismic intensities considered in the design stage (see Table 1).
 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154 151

0.5

Reaction / weight
0.2 MDOF structure w/BRBs

Displacement, m
Dual SDOF oscillator
0
0

Time, s -0.5
-0.2 -0.2 0 0.2
20 40 60 80 100 Displacement, m
(a) Response in the time domain (b) Hysteretic loops
Fig. 13. Response of the model and an equivalent dual SDOF oscillator to the SCT-EW record.

5 5
pga=0.05g pga=0.10g
4 4

3 3
Storey

Storey
2 2

1 1

0 0
0 0.001 0.002 0.003 0.004 0.005 0 0.0025 0.005 0.0075 0.01
Inter-storey drift, m/m Inter-storey drift, m/m

5 5
pga=0.20g pga=0.30g
4 4

3 3
Storey

Storey

2 2

1 1

0 0
0 0.005 0.01 0.015 0.02 0 0.01 0.02 0.03
Inter-storey drift, m/m Inter-storey drift, m/m

Fig. 14. Maximum inter-storey drift demands from nonlinear dynamic analyses.

The mean and mean plus and minus one standard deviation of the
demands are also shown with dark lines. It is observed that: (1) the
heightwise distribution of the inter-storey drifts for linear-elastic
response (pga = 0.05g and 0.10g) was different to that of nonlinear
response (pga = 0.20g and 0.30g); and (2) as the level of the nonlin-
earity increased, the mean and dispersion of the drifts increased
more significantly in the lower floors. These observations justify
the values of the non-uniform inter-storey drift factors suggested
in Step 4 of the proposed procedure. Fig. 14 shows that the mean
plus one standard deviation demands were smaller than the corre-
sponding thresholds, in such a manner that it can be concluded
that the design is acceptable.
5. Preliminary assessment
One attractive characteristic of the proposed method is the
additional information it generates; which may be useful to con-
duct preliminary assessments of the building. Since the nonlinear
dynamic equation of motion for the dual SDOF oscillator, i.e.
Eq. (10), is solved for each ground motion and intensity considered,
information such as peak and residual displacements, floor veloci-
ties and accelerations, is available and can be transformed to
approximate the response of a MDOF structure to conduct the
assessment without the necessity of generating a detailed and
time-consuming model. Based on the good agreement observed
in Fig. 13, a procedure to transform the statistics of the response
of a dual SDOF oscillator to those of a MDOF structure is presented
in [24]. The reader is prevented to be cautious here because the
resultant assessment will only be an approximation and the level
of approximation will depend on how much the dual SDOF model
represents the behaviour of the corresponding MDOF structure fit-
ted with BRBs. In this regard, only comparisons may be made in
relative terms, for example, to determine the most convenient
option of design amongst different options. Other parameters that
are also available and may provide valuable information are the
maximum and cumulative ductility demands. With them, the
remaining useful life of BRBs can be determined and compared
to its respective expected capacity; which may be estimated as
suggested by Takeuchi et al. [30].
In order to conduct a probabilistic assessment of the case study
example of the previous section, the Monte Carlo simulation proce-
dure proposed by the ATC-58 Project [2] was adopted here. For
illustration purposes, an intensity-based assessment was con-
ducted for option (a) (b2 = 60%), as defined in Section 4. The assess-
ment procedure consists of four probabilistic analyses, which are
seismic hazard analysis, dynamic response analysis, damage state
analysis, and loss analysis. In the pursuit of simplicity, only the
results of the latter are presented in the following paragraphs
and the reader is referred to Ref. [24] for more details.
In order to conduct a loss analysis, a decision variable must be
selected first. In this paper, total repair cost has been chosen. Then,
the results of the dynamic response analysis (i.e. statistics of the
152 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
1 1
P(Repair cost ≤ Cr)
0.75
0.5
0.25
0 0
0 0.25
Repair / Replacement Cost (Cr)
(a) CDFs for different seismic intensities
Fig. 15. Cumulative distribution functions of repair cost for design option (a).
structural response and collapse fragility) and the damage state
analysis (fragility data of components) are used to estimate the
expected repair cost of the building when subjected to a given
pga. The procedure can be summarised as follows: for a given seis-
mic intensity (e.g. pga = 0.20g), a uniform random number between
zero and unity is generated. If the random number is higher than
the probability of collapse (estimated in the dynamic response
analysis), a collapse is considered to have occurred and the total
repair cost is that corresponding to replacement cost (including
demolition and clearance cost). On the other hand, if the random
number is less that the probability of collapse, the total repair cost
is estimated as follows: (a) a vector of statistically consistent sim-
ulated demands is generated according to [2] using the statistics of
the dynamic response analysis; (b) a random damage state is gen-
erated for every component of the building (damage state analy-
sis); and (c) the associated repair cost is obtained by adding all
the products of unit cost and quantity of every component. It is
also considered that, if the structure is deemed irreparable (as
defined in [2]), the total replacement cost (including demolition
and clearance cost) shall be considered. This procedure is repeated
many times (hundreds or thousands) to generate a smooth distri-
bution of possible damage states and repair cost. Yang et al. [31]
have found that as few as 200 repetitions, termed as realisations,
can provide stable cost estimates.
For the case study example, the total repair cost was deter-
mined for the four design intensities of Table 1. Fig. 15a shows
the lognormal fit of the cumulative distributed function (CDF) of
the total repair cost, normalised by the replacement cost. Since col-
lapse or irreparability were detected at the higher intensities, total
replacement cost was required; therefore, a step in the CDF curves
is observed at normalised cost of unity. The median repair cost (i.e.
50% of probability) were 4%, 11%, 29% and 60% for pga = 0.05g,
0.10g, 0.20g and 0.30g, respectively. As expected, it is observed that
the higher the seismic intensity, the higher the repair cost.
On the other hand and as a matter of interest, the sensitivity of
the repair cost to variations in the areas of the BRBs is assessed. For
that purpose, the areas of the BRBs calculated previously in Sec-
tion 4.2 for the design option (a), i.e. b2 = 60%, were scaled by 0.7
and 1.5. Fig. 15b shows the CDF determined for pga = 0.20g. In
the figure, it is observed that repair costs change significantly as
the areas of the BRBs are varied. The mean values were 47%, 29%
and 17%, respectively. It is appreciated that increasing the areas
of the BRBs is an effective way of reducing the expected losses in
this case study example.
Finally, although not shown here, similar assessments were
conducted for options (b) and (c) of the case study example, i.e.
b2 = 30% and b2 = 40% as defined in Section 4.2. For typical structural
steel and BRBs cost, the probable repair cost, normalised by the
replacement cost, resulted similar for the three cases; however,
the constructional cost of option (a) was 10% and 8% smaller than
P(Repair cost ≤ Cr)
0.75
0.5
Participation
0.25 Intensity:
of BRBs:
pga=0.20g
b2 =60%
0.5 0.75 1 0 0.25 0.5 0.75 1
Repair / Replacement Cost (Cr)
(b) Variations in the areas of the BRBs
those for options (b) and (c). Therefore, the designer may select
option (a) as the most efficient in this particular case study
example.
6. Discussion
Significance of this research. Performance-Based Seismic Design
(PBSD) philosophy is regarded as superior to code provisions
because it is able to predict and evaluate, in a realistic and reliable
way, the performance of buildings (or facilities) with clear
understanding of risk [1,2]. However, implementation of PBSD is
often reserved for critical facilities, due to the required increase
of engineering design involvement. Therefore, the proposed
method for designing buildings equipped with BRBs facilitates
the implementation of PBSD in such type of structures. In this
way, trade-offs based not only on initial construction costs but also
in life-cycle considerations can be made by decision makers at the
initial stage of the design process [1]. Rapid application of PBSD is
achieved by assuming that a structure equipped with BRBs can
rationally be represented by a dual SDOF oscillator (as described
by Eq. (10) and validated in Section 4.3).
Justification of using Eq. (10). It is observed in Fig. 5 that the dis-
placement and resistance demands of conventional and dual oscil-
lators may be significantly different, even when they have the
same period of vibration and ductility demands. This behaviour
is attributed to the fact that the parts of dual oscillators yield at dif-
ferent levels of displacement. Therefore, the practice of designing
structures equipped with BRBs using spectra constructed from
conventional oscillators may lead to structures that do not behave
as expected. Thus, in the proposed method the dynamic equation
of motion for dual oscillators (Eq. (10)) is solved for possible earth-
quake motions. In this way, the hysteretic characteristic of each
part of the dual oscillator can be explicitly considered in the solu-
tion. If significant, P–D effects are also taken into account in the
force–displacement relationship of the primary part of the dual
SDOF oscillator (see Step 2 of Section 3.2). Another advantage of
solving Eq. (10) is that useful information is generated, so that pre-
liminary performance-based assessments of the structure can be
conducted. Besides, the time required to solve the equation is
deemed negligible using current computer systems.
Limitations. Like many other methods, this proposed method
has limitations:
 The method is valid for regular, low-rise structures. Two aspects
are discussed here. First, a dual SDOF oscillator with dynamic
properties equivalent to those of a MDOF structure equipped
with BRBs is used, thus, the method is valid for structures with
response dominated by the first mode of vibration. In structures
with significant participation of higher modes, such as high-rise
buildings, further extension of the method is required. Second, a
 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
linear displacement profile is assumed in Step 3 of Section 3.2 to
determine the displacement thresholds of dual SDOF oscillators.
This may be reasonably acceptable for regular and low-rise
structures with beam-sway mechanism [14]. However, for
other cases, the floor displacements (di) shall be estimated
accordingly. The displacement profile obtained in Step 2 by
pushover analysis may provide a good approximation.
 Additional data and a computer subroutine are required to apply
the method. Despite the advantages of the proposed method, it
is recognised that more information should be collected before
its application: (1) seismic records, actual or artificial, compat-
ible with the local hazard are required – they may be obtained
readily from databases most of the time available online or from
professional associations of earthquake engineering; (2) if
assessment is also being conducted, information of components
(structural, non-structural and contents) and their damageabil-
ity are required – this step has been facilitated by the ATC-58
Project [2] where typical and normative information has been
compiled; and (3) a computer subroutine is required to solve
Eq. (10) because without it the method would be impractical
– this subroutine has been developed and is available in [24].
Further application of the proposed method. Another important
application of the proposed method is in the retrofitting and
upgrading of existent structures. In this particular case, the charac-
teristic of the existent structures are fixed, and the designer has to
select a relative participation of the BRBs to control the global
response. If desired, low participation from the BRBs might be
selected in order to avoid damage to existent connections and
foundations. As previously observed, this is straightforward to do
using the proposed method along with preliminary assessments
to decide if retrofitting is convenient and to which level.
7. Conclusions
A method for preliminary design of low-rise buildings equipped
with BRBs has been proposed. An example building was designed
to show its applicability. Taking advantage of the information that
is generated during the application of the method, an intensity-
based assessment of the example building was also conducted.
The following conclusions are formulated:
1. Considering the assumptions discussed in this paper, applica-
tion of the PBSD philosophy can be facilitated using the pro-
posed method. As a consequence, diverse options of design
can now be compared to find efficient structures, based on
trade-offs between initial costs and life-cycle considerations.
2. Fig. 5 showed that conventional and dual structures behave dif-
ferently, even when they have the same period of vibration and
ductility demand. Therefore, the use of dual SDOF oscillators
may be more rational to design structures equipped with BRBs
than the use of conventional oscillators in order to avoid
designs that do not behave as intended.
3. The proposed method realises the structural fuse concept
because the ductility factors for the frame and BRBs can be esti-
mated at the beginning of the design process, based on the geo-
metric and mechanical properties of their members.
4. An example of application was developed to show the applica-
bility of the method. From nonlinear static and dynamic analy-
ses of the example structure, it was appreciated that dual SDOF
oscillators represent well the response of MDOF structures
equipped with BRBs. As a consequence, the additional informa-
tion obtained during the solution of Eq. (10) can be transformed
to approximate the response of the MDOF structure and to con-
duct a rapid and preliminary assessment of the performance.
153
5. Also from the assessment of the example structure, it was
appreciated that, for this particular structure and conditions,
the cross-sectional area of the BRBs is very effective to control
expected losses. Since a different conclusion may be found
under different conditions, rapid and preliminary assessments
of the structural performance result very convenient to find effi-
cient designs.
Acknowledgments
The first author acknowledges the sponsorship by CONACYT
and SEP from Mexico.
References
[1] Krawinkler H, Miranda E. Performance-based earthquake engineering. In:
Bozorgnia Y, Bertero VV, editors. Earthquake engineering from engineering
seismology to performance-based engineering. CRC PRESS; 2004.
[2] FEMA-P58. Seismic performance assessment of buildings. Washington, D.C.:
Federal Emergency Management Agency; 2012.
[3] Merrit S, Uang C-M, Benzoni G. Subassemblage testing of core brace buckling-
restrained braces. La Jolla (California): University of California, San Diego;
2003.
[4] Uang C-M, Nakashima M. Steel buckling-restrained braced frames. In:
Bozorgnia Y, Bertero VV, editors. Earthquake engineering from engineering
seismology to performance-based engineering. CRC Press; 2004.
[5] Maley TJ, Sullivan TJ, Della Corte G. Development of a displacement-based
design method for steel dual systems with buckling-restrained braces and
moment-resisting frames. J Earthq Eng 2010;14:106–40.
[6] Vargas R, Bruneau M. Analytical response and design of buildings with metallic
structural fuses. I. J Struct Eng – ASCE 2009;135:386–93.
[7] Pettinga D, Christopoulos C, Pampanin S, Priestley MJN. Effectiveness of simple
approaches in mitigating residual deformations in buildings. Earthq Eng Struct
Dynam 2007;36:1763–83.
[8] Kiggins S, Uang C-M. Reducing residual drift of buckling-restrained braced
frames as a dual system. Eng Struct 2006;28:1525–32.
[9] Kim J, Seo Y. Seismic design of low-rise steel frames with buckling-restrained
braces. Eng Struct 2004;26:543–51.
[10] Sahoo DR, Chao S-H. Performance-based plastic design method for buckling-
restrained braced frames. Eng Struct 2010;32:2950–8.
[11] Ragni L, Zona A, Dall’Asta A. Analytical expressions for preliminary design of
dissipative bracing systems in steel frames. J Constr Steel Res 2011;67:102–13.
[12] Lin YY, Tsai MH, Hwang JS, Chang KC. Direct displacement-based design for
building with passive energy dissipation systems. Eng Struct 2003;25:25–37.
[13] Sutcu F, Takeuchi T, Matsui R. Seismic retrofit design method for RC buildings
using buckling-restrained braces and steel frames. J Constr Steel Res
2014;101:304–13.
[14] Priestley MJN, Calvi GM, Kowalsky MJ. Displacement-based design of
structures. Pavia, Italy: Istituto Universitario di Studi Superiori di Pavia; 2007.
[15] Teran-Gilmore A, Virto-Cambray N. Preliminary design of low-rise buildings
stiffened with buckling restrained braces by a displacement-based approach.
Earthq Spectra 2009;25:185–211.
[16] Della Corte G, D’Aniello M, Landolfo R. Field testing of all-steel buckling-
restrained braces applied to a damaged reinforced concrete building. J Struct
Eng 2015;141:D4014004.
[17] 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.
[18] Fahnestock L, Ricles J, Sause R. Experimental evaluation of a large-scale
buckling-restrained braced frame. J Struct Eng 2007;133:1205–14.
[19] FEMA-356. Prestandard and Commentary for the Seismic Rehabilitation of
Buildings. In: Agency FEM, editor. Washington, D.C.; 2000.
[20] Sabelli R, Mahin S, Chang C. Seismic demands on steel braced frame buildings
with buckling-restrained braces. Eng Struct 2003;25:655–66.
[21] Teran-Gilmore A, Coeto G. Displacement-based preliminary design of tall
buildings stiffened with a system of buckling-restrained braces. Earthq Spectra
2011;27:153–82.
[22] Bojórquez E, Ruiz-García J. Residual drift demands in moment-resisting steel
frames subjected to narrow-band earthquake ground motions. Earthq Eng
Struct Dynam 2013;42:1583–98.
[23] MSMD. Mexican strong motion database. Mexican Society of Earthquake
Engineering; 1999.
[24] Guerrero H. Seismic design and performance of hospital structures equipped
with buckling-restrained braces in the lakebed zone of Mexico City PhD
Thesis. UK: The University of Manchester; 2016.
[25] Kasai K, Ito H, Ooki Y, Hikino T, Kajiwara K, Motoyui S, et al. Full-scale shake
table tests of 5-story steel building with various dampers. In: Joint conference
7CUEE & 5ICEE. Tokyo, Japan: Tokyo Institute of Technology; 2010.
154 H. Guerrero et al. / Engineering Structures 123 (2016) 141–154
[26] Opensees. Open source finite element platform for earthquake engineering
simulations. Berkeley: Pacific Earthquake Engineering Center, Univ. of
California; 2014.
[27] Filippou FC, Fenves GL. Methods of analysis for earthquake-resistant
structures. In: Bozorgnia Y, Bertero VV, editors. Earthquake engineering from
engineering seismology to performance-based engineering. CRC Press; 2004.
[28] Uriz P, Filippou FC, Mahin S. Model for cyclic inelastic buckling of steel braces. J
Struct Eng 2008;134:619–28.
View publication stats
[29] Charney FA. Unintended consequences of modeling damping in structures. J
Struct Eng – ASCE 2008;134:581–90.
[30] Takeuchi T, Ida M, Yamada S, Suzuki K. Estimation of cumulative deformation
capacity of buckling restrained braces. J Struct Eng 2008;134:822–31.
[31] Yang T, Moehle J, Stojadinovic B, Der Kiureghian A. Seismic performance
evaluation of facilities: methodology and implementation. J Struct Eng
2009;135:1146–54.