You are on page 1of 15

Journal of Constructional Steel Research 137 (2017) 37–51

Contents lists available at ScienceDirect

Journal of Constructional Steel Research


journal homepage: www.elsevier.com/locate/jcsr

Response to seismic sequences of short-period structures equipped with T


Buckling-Restrained Braces located on the lakebed zone of Mexico City
Héctor Guerreroa,⁎, Jorge Ruíz-Garcíab, J. Alberto Escobarc, Amador Terán-Gilmored
a
Universidad Nacional Autónoma de México, Mexico
b
Universidad Michoacana de San Nicolás de Hidalgo, Mexico
c
Universidad Nacional Autónoma de México, Mexico
d
Universidad Autónoma Metropolitana, Mexico

A R T I C L E I N F O A B S T R A C T

Keywords: As part of an ongoing research program, this paper presents the results of a numerical study of steel frames with
Buckling-Restrained Braces and without Buckling-Restrained Braces (BRBs), representative of essential buildings (e.g. hospitals). The aim of
Mainshock-aftershock seismic sequences this study is to understand the effects of seismic sequences on this type of structures when located on the soft soil
Dual systems of Mexico City. For that purpose, three-, six-, and nine-storey frames were designed under diverse criteria and
Lake-bed zone of Mexico City
analysed under artificial sequences to understand the influence of BRBs on the peak and residual inter-storey
Soft soils
drifts demands. The results showed that the effect of aftershocks is very significant when their intensity, in terms
of peak-ground velocity, is similar to that of the mainshock. It is also shown that, although aftershock effects may
not be eliminated, a proper design based on the concept of dual systems could help to mitigate aftershock effects.
Based on the analytical results, expressions to estimate the magnitude of peak and residual inter-storey drift
demands under mainshocks and the sequence mainshock-aftershock are offered as a helpful guide to structural
designers.

1. Introduction conducted by Mahin [8] to understand their effects on the displacement


ductility demands of single-degree-of-freedom (SDOF) oscillators. Since
In recent decades, it has been well understood that man-made then, several studies have been conducted on SDOF oscillators (e.g.
structures located in seismic regions are not only exposed to mainshock [9,10]) and in multi-storey structures (e.g. [11–13]). For instance,
events but also to full sequences, which include foreshocks, the main- Erochko et al. [13] observed that residual drifts increased due to the
shock, and aftershocks [e.g. 1,7]. This phenomenon has led to large aftershock on moment-resisting frames and pinned-connected Buckling-
structural affectations, as it was observed in past seismic events such as Restrained Braces (BRB) frames using an artificial mainshock-after-
the 1985 Michoacan earthquakes [2] that struck Mexico City, and more shock sequence formed with the same acceleration-time history of the
recently in the 2011/2012 New Zealand earthquakes that hit the Can- mainshock as an aftershock (i.e. back-to-back approach [7]). However,
terbury region [3,4]. In some cases the problem has been worsened Ruiz-García and Negrete-Manriquez [12] found that artificial seismic
because aftershocks may present higher seismic intensity than that of sequences assembled with the back-to-back approach could over-
the mainshock [6] (i.e. aftershock peak-ground acceleration is higher estimate the peak inter storey and residual drift demands of multi-
than that of the mainshock, which might be explained by aftershock storey steel frames due to the different frequency contents on the
source-to-site distance shorter than that of the mainshock [7]), al- aftershocks relative to the mainshock.
though they have smaller moment magnitude than their corresponding The motivation of this study relies on the fact that only two studies
mainshock event [5]. Therefore, the effects of seismic sequences should are available in the literature for the particular case of the response of
be taken into account within the design stage of a structure. Un- multi-storey structures under mainshock-aftershock sequences recorded
fortunately, these effects are rarely considered in the design process due on very soft soil conditions (those reported by Ruiz-García et al. [1] and
to the complexity of the phenomenon and the uncertainties associated Díaz-Martínez et al. [14]). Ruiz-García et al. [1] assessed the effects of
to it. mainshock-aftershock seismic sequences on peak and residual inter-
Some of the first studies on the effects of aftershocks were storey drift demands of framed-reinforced concrete (RC) buildings


Corresponding author.
E-mail address: hector.guerrerobobadilla@manchester.ac.uk (H. Guerrero).

http://dx.doi.org/10.1016/j.jcsr.2017.06.010
Received 21 January 2017; Received in revised form 6 June 2017; Accepted 12 June 2017
Available online 16 June 2017
0143-974X/ © 2017 Elsevier Ltd. All rights reserved.
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

case Fig. 1. Schematic representation of Buckling-Restrained


A Braces (BRBs).

P tension
core A’
Conventional
Steel tube Concrete braces
core unbonding d
material
BRBs
A - A’ compression

a) Scheme of a type of BRB b) hysteretic curve

located on the lake-bed zone of Mexico City. They found that the inter- This paper is part of an ongoing research program aimed at quan-
storey drift demands increased as the seismic intensity, in terms of tifying the benefits of using BRBs in steel framed structures located on
peak-ground velocity (PGV), of the aftershock relative to that of the the very soft soil area of Mexico City. In previous studies, it was found
mainshock increased. They also observed that other factors (e.g. the that BRBs are very convenient to be particularly used in short-period
stiffness degradation; the predominant period of the aftershock, relative framed buildings such as low-rise hospitals [21].
to that of the mainshock; and the damaged period of the structure re- The particular objective of this paper is to quantify the effects of
lative to that of the aftershock) also had a significant effect on the re- mainshock-aftershock seismic sequences on short-period structures
sponse demands of the studied structures. Díaz-Martínez et al. [14] equipped with BRBs located in the soft soil of Mexico City. For this
studied two steel-moment resisting frames under artificial seismic se- purpose, three-, six- and nine-storey steel frames were analysed under a
quences, where they found that the case-study buildings experienced set of artificial mainshock-aftershock seismic sequences. Five design
detrimental effects due to aftershocks (i.e. the inter-storey drift de- cases for each frame were considered to understand the influence of
mands significantly increased under the aftershocks) when their fun- BRBs on mitigating peak and residual inter-storey drifts demands. The
damental periods of vibration were longer than the predominant period results showed that the effects of the aftershocks are very significant
of the mainshock. From these studies, it is seen that seismic perfor- when their intensity (in terms of PGV) is similar to that of the main-
mance of steel moment-resisting frames incorporating BRBs under shock. As demonstrated in this paper, although aftershock effects may
mainshock-aftershock seismic sequences has not been examined yet. not be eliminated, a proper design of the case-study frames could help
Therefore, this study is focused on that subject. to mitigate them.
On the other hand, there have been significant developments in the
last few decades in the area of seismic protection systems (SPS), e.g.
[15]. However, damages, economical losses, and fatalities are still 2. Buildings considered in this study
commonly observed in seismic regions. The use of structural fuses, such
as BRBs, may help to reduce the affectations generated by earthquakes. 2.1. Building description
Nonetheless, studies are still needed in order to understand quantita-
tively the benefits and disadvantages of the use of SPS in structures. Three buildings, which represent hypothetical hospitals located in
Several experimental studies have shown that BRBs are effective to the lake-bed zone of Mexico City, having three, six, and nine storeys
dissipate seismic energy (e.g. [16–19]). Although a detailed summary were considered as part of this study. They consist of braced steel
can be found in Ref. [20], a brief summary of BRBs is offered here with frames with six equally-spaced bays of 8 m that are schematically
the help of Fig. 1. Even though there are several types of BRBs, probably shown in Fig. 2. In all buildings, the first-storey height is 4 m, while the
the most popular one is composed of a rectangular steel core and a case remaining storeys have 3 m. It was assumed that the frames incorporate
composed of a concrete-filled steel tube. Note that the core is weaker in Buckling-Restrained Braces (BRB) as part of their lateral-load resisting
the central zone, which concentrates plastic deformations on BRB when system.
subjected to axial deformations, than in the end regions. The main For comparison purposes, five cases were studied corresponding to
purpose of the case is preventing buckling of the core when the device is each hospital building with the aim of assessing the effect of different
subjected to compressive loads which, contrary to the degrading be- contributions of BRBs to the initial cost, lateral load capacity, and
haviour of conventional braces, lead to stable hysteretic performance displacement demands. The studied cases are schematically presented
with similar capacity in tension and compression (Fig. 1b). Another in Fig. 3, normalised with respect to the baseline case, and described as
important component of BRBs is the unbonding material located be- follows:
tween the core and the case, which avoids direct interaction between
Case 0. The reference case. A bare steel frame (i.e. without BRBs) is
them and allows axial and transversal deformation of the core.
designed to fully resist the seismic loads.

Fig. 2. Frame structures studied in this


paper.

38
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 3. Schematic description of cases in terms of:

Displacement demands
a) initial cost, b) lateral load capacity, and c)
lateral displacement demand.

Load capacity

No BRBs
No BRBs

No BRBs
Initial cost

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
Case Case Case
a) b) c)

Case 1. BRBs are provided to the steel frame of Case 0 (i.e. without
Table 2
modifying the original steel beam and column sections). This action Areas, in cm2, of the BRB cores for the first storey.
increases the frame's initial cost, but increases its lateral load capacity
and reduces the lateral displacement demands. Building Case 1 Case 2 Case 3 Case 4

Case 2. In this case, the main steel frame is redesigned to resist only 3-Storey 13.5 18.0 6.84 32.0
gravity loads (i.e. smaller sections of beams and columns are specified). 6-Storey 46.9 105.0 60.0 152.0
9-Storey 64.0 127.0 93.0 190.0
In addition, BRBs are provided to increase the lateral load capacity and
to match the initial cost of Case 0.
Case 3. Similar to Case 2, the main steel frame is designed to support Finally, the height-wise distribution of areas of the steel cores was as-
only gravity loads and BRBs are included in such a way to match the sumed proportional to that of the first storey times the following vec-
level of lateral displacement demands of Case 0. Therefore, Case 3 may tors: (1.0, 0.614, 0.273)T, (1.0, 0.791, 0.626, 0.461, 0.296, 0.132)T, and
have less initial cost than Case 0, but similar dynamic response (1.0, 0.847, 0.738, 0.630, 0.521, 0.412, 0.304, 0.195, 0.087)T for the 3-,
demands. 6-, and 9-storey frames, respectively. The masses of the models are
690 tons in the top floor and 864 tons in the others. A detailed de-
Case 4. Once again, the main frame is designed to support only gravity scription of the design process of each frame and corresponding cases
loads and BRBs are included with larger steel core size so that the can be found in Ref. [22].
lateral load capacity and initial cost are similar to those of Case 1.
For design purposes, the steel of columns and beams was assumed to 2.2. Modelling assumptions
be ASTM A992 (fy = 350 MPa) while that of the cores of the BRBs was
assumed as ASTM A36 (fy = 250 MPa). In particular, the BRBs were The set of fifteen buildings included in this investigation were
assumed to be pinned connected to the host frames, so that the devices modelled as two-dimensional (2D) centreline models using the finite
only work under axial deformation and no moments are transferred to element program for earthquake simulation OpenSees [25]. Since the
the connections or frame elements. To be consistent with that, the columns were fixed at their bases, soil-structure interaction was ne-
structures were designed using the method proposed by Guerrero et al. glected. This may have an effect on the dynamic response of the models,
[23], which assumes that the lateral stiffness and capacity of a structure however, relative comparisons may still be valid assuming that the ef-
equipped with BRBs are composed of the summation of two sub- fect is similar in all the cases. The beams and columns were modelled as
structures, namely: A) a moment resisting frame, and B) a pinned 2D frame elements with distributed plasticity [26], where the cross-
connected frame (whose lateral stiffness and capacity are only provided section of the elements was modelled using fibre elements. For this
by the BRBs). The aforementioned method is based on the control of the purpose, expected rather than nominal properties of the materials were
lateral displacements, which is significant because the effects of the considered. The steel was modelled using the Steel02 uniaxial material,
BRBs can be compared directly, while the effect of lack of control of the which is based in the Menegotto-Pinto hysteretic material [25], avail-
lateral displacements is avoided. The final sizes of the steel sections and able in the OpenSees library. A post-yielding stiffness ratio of 0.3% was
steel core area of the BRB located in the first storey corresponding to considered following the recommendation given in Ref. [28]. Addi-
each case are listed in Tables 1 and 2, respectively. Within the design tional strength and stiffness contribution to the beams from the slabs
process of the BRBs, it should be mentioned that the BRB stiffness were considered as recommended in Ref. [27]. The concrete of the slabs
modification factors, KF (as defined in [24]), were assigned as 1.5 for was modelled using the Concrete02 uniaxial material, which assumes
the six- and nine-storey frames, and 2.0 for the three-storey frame. linear tension softening. Finally, it was assumed a rigid floor system for
analysis purposes.
Table 1 Before performing nonlinear dynamic analysis, conventional modal
Steel sections of the case-study buildings and their corresponding cases. analysis and nonlinear static (pushover) analysis were carried out to
obtain the dynamic and mechanical properties of the three building
Columns (HSS sections, mm) Beams models. A summary of the fundamental period of vibration for each
Building Storey Cases 0 & 1 Cases 2, 3, 4 Cases 0 & 1 Cases 2, 3, 4
case-study frame is presented in Table 3. Additionally, Figs. 5 and 6

3-Storey 1 500 × 25 500 × 19 W21 × 68 W21 × 68 Table 3


2 500 × 25 500 × 13 W21 × 68 W21 × 68 Fundamental period of vibration, in seconds, of each case-study frame.
3 500 × 19 500 × 13 W21 × 62 W21 × 62
6-Storey 1 to 3 600 × 38 500 × 16 W27 × 94 W24 × 68 Structure Case 0 Case 1 Case 2 Case 3 Case 4
4 to 6 600 × 19 500 × 13 W27 × 84 W24 × 68
9-Storey 1 to 3 900 × 38 500 × 25 W27 × 129 W24 × 68 3-Storey 0.85 0.73 0.74 0.84 0.66
4 to 6 900 × 25 500 × 16 W27 × 102 W24 × 68 6-Storey 1.17 0.97 0.97 1.10 0.88
7 to 9 800 × 19 500 × 13 W27 × 84 W24 × 68 9-Storey 1.41 1.20 1.26 1.35 1.15

39
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 4. Fundamental periods of vibration


3-Storeys 6-Storeys 9-Storeys 3-Storeys 6-Storeys 9-Storeys
normalised by the period of Case 0.

0 1 2 3 4 0 1 2 3 4
Case Case

a) Normalised periods b) Normalised participation factors

show these periods (normalised by that of Case 0) and the fundamental inter-storey drift ratio through the height of the structure) of the frame
modal shapes, respectively. From Fig. 5a, it is observed that Cases 0 and models for all the studied cases obtained from pushover analyses. The
3 are more flexible than Cases 1, 2, and 4, which is expected from curves were obtained from a mass-distributed loading profile along the
comparison with Fig. 3c. Fig. 5b shows the modal participation factors, frames' height. Pushover analyses were carried out for a target roof
normalised by the modal shape value at the roof level. Uniform values displacement equal to 2% of the total height of the structure. From the
are seen on each building irrespective of the case (i.e. with or without figure, it can be seen that the capacity of the frames is in agreement
BRBs). From Fig. 6, it is observed that the modal shapes are very similar with the trend shown in Fig. 3b (i.e. Cases 1 and 4 have the highest
for all the studied cases and they present shear-type deformation; capacity, then Cases 0 and 2, and finally Case 3). Particularly, it should
however, the cases with BRBs (Cases 1 to 4) tend to have more linear be noted that the maximum inter-storey drift ratio (IDR) of the frames
modal shapes; which has also been observed by others (e.g. [29]). This at yielding is around 0.01 (e.g. for the capacity curve corresponding to
characteristic is significant, as structures equipped with BRBs tend to Case 0 for the 3-storey frame), while inclusion of BRBs allow yielding at
present better, i.e. more uniform, distribution of the inter-storey drift smaller IDRs as it can be seen in the capacity curves corresponding to
demands through their height. Cases 1 to 4. It can also be observed that the post-elastic stiffness is
Fig. 4 shows a comparison of the capacity curves (i.e. base shear positive, and the transition from elastic to plastic response becomes
normalised by the total weight of the structure versus the maximum smoother as the number of storeys increases. The aforementioned ob-

Fig. 5. Modal shapes for the first mode.


3 Case 4 6 Case 4 9 Case 4
Case 3 Case 3 Case 3
Case 2 Case 2 8 Case 2
Case 1 5 Case 1 Case 1
7
Case 0 Case 0 Case 0
2 4 6
Storey

Storey
Storey

5
3
4
1 2 3
2
1
1
0 0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Amplitude Amplitude Amplitude

a) Three storeys b) Six storeys c) Nine storeys

Fig. 6. Comparison of capacity curves corre-


0.5 0.5 0.5
sponding to each case: a) 3-storey, b) 6-storey,
Base shear / total weight

Base shear / total weight


Base shear / total weight

and c) 9-storey.
0.4 0.4 0.4

0.3 0.3 0.3


Case 0
Case 1
0.2 0.2 0.2
Case 2
0.1 Case 3 0.1 0.1
Case 4
0 0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03 0 0.01 0.02 0.03
Inter-storey drift ratio Inter-storey drift ratio Inter-storey drift ratio
a) b) c)

40
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 7. CDAF records of the 1985 Mexico Earthquakes:


100 500
Accel. [cm/s2]

a) accelerogram of the mainshock, b) pseudo-accel-

Pseudo-accel. [cm/s2]
a) 400 b) eration spectrum of the mainshock, c) accelerogram of
the aftershock, and d) pseudo-acceleration spectrum of
0 300
the aftershock.
200
100
-100 Time [s]
0
0 30 60 90 120 150 0 1 2 3 4 5
Period [s]

100 500

Pseudo-accel. [cm/s2]
d)
Accel. [cm/s2]

c) 400
300
0
200
100
-100 Time [s]
0
0 30 60 90 120 150 0 1 2 3 4 5
Period [s]

servations will become relevant in the discussion given in Section 4. In order to perform nonlinear dynamic analyses, several actual
For dynamic analyses, a damping ratio of 5% was considered for all mainshock-aftershock acceleration records of earthquake ground mo-
the studied cases. Although it is recognised that BRBs increase the tions are needed. Unfortunately, this information is very limited in soft
damping ratio (see Ref. [22]), in this paper, 5% was considered indis- soil sites – as an example, only one seismic sequence was recorded in
tinctly. During the analysis, P-Δ effects were included by means of the soft soil during the September 19 and 20, 1985 earthquakes of Mexico.
corotational theory to represent large displacement effects [28]. Dy- The sequence was recorded at the Central de Abastos (CDAF) station.
namic nonlinear analyses, whose results are presented in the next sec- The East-West components are shown in Fig. 7 along with their re-
tion, were conducted using the well-known Newmark direct integration spective pseudo-acceleration spectra. It is noted in the figure that the
method with a small time step to ensure convergence. spectral shapes are significantly different. Also, that the predominant
period of the motion (Tg) – where Tg is taken as the period at which the
3. Mainshock-aftershock seismic sequences maximum peak of the 5%-damped pseudo-acceleration spectrum occurs
– is around 3 s in the mainshock and 2 s in the aftershock; i.e. the
As described in Ref. [44], populated areas of Mexico City are located mainshock presents longer period than the aftershock. The most re-
partly on an old lake bed that was formed by the Texcoco, the Chalco, levant characteristics of CDAF records (e.g. magnitude (Ms), peak-
and the Xochimilco lakes. This lake-bed zone has relatively thick de- ground acceleration (PGA), peak-ground velocity (PGV) and pre-
posits of lacustrine soft clay, which form the soft zone for seismic zo- dominant period of the motion (Tg)) are presented in the first part of
nation purposes, whose depth varies from 10 m to 60 m. These clay Table 4.
deposits are very deformable and are characterized by: a) very high Analysing the PGV of the CDAF records, it is observed that the in-
water contents that reach more than 400%, b) shear wave velocities as tensity of the aftershock was around 35% of the mainshock's PGV. Also,
low as 40 m/s, and c) high plasticity indexes. that the predominant period of vibration, Tg, of the aftershock is shorter
As it can be seen in Figs. 7b, and 3 in Ref. [45], earthquake ground than that of the mainshock.
motions recorded in this region of Mexico City lead to 5%-damped Taking into account the lack of enough seismic sequences recorded
acceleration response spectra that are characterized by peak spectral in soft soil sites, artificial seismic sequences shall be generated and used
acceleration ordinates that are typically between five and six times the in the analyses. For this purpose, two approaches are commonly em-
peak ground acceleration. These spectral amplifications are approxi- ployed [1]: A) the back-to-back approach, and B) the randomized ap-
mately two times larger than those that, on average, occur for the same proach. While the back-to-back approach consists of repeating the same
level of damping in spectra of ground motions recorded on rock or firm mainshock as aftershock at the same or at scaled amplitude, the ran-
sites. Furthermore, the frequency content of these ground motions is domized approach consists of combining two mainshocks (one as a
characterized by a clearly defined predominant frequency and narrower mainshock and the other as an aftershock at reduced or identical am-
spectra [45]. plitude) from different seismic events. Because the former approach

Table 4
Seismic ground motions used in this study, selected from [30].

Date Ms Station name Compo. PGA (m/s2) PGV (cm/s) Tg (s)

19/09/1985 8.1 Central de Abastos N00E 66.3 26.4 2.0


19/09/1985 8.1 Central de Abastos N90E 95.9 36.7 2.9
20/09/1985 7.6 Central de Abastos N00E 40.1 12.1 1.9
20/09/1985 7.6 Central de Abastos N90E 30.2 9.3 1.8
M1 25/04/1989 6.9 Villa del mar EW 46.5 15.3 2.9
M2 25/04/1989 6.9 Villa del mar NS 49.4 22.0 2.9
M3 25/04/1989 6.9 Jamaica NS 35.2 15.6 3.0
M4 25/04/1989 6.9 Rodolfo Menéndez EW 47.7 18.8 2.3
25/04/1989 6.9 PCC Superficie EW 42.5 15.4 2.3
14/04/1989 7.1 Córdova EW 45.2 11.2 2.3
25/04/1989 6.9 Liverpool EW 40.0 12.4 2.3
14/04/1989 7.1 Roma-B EW 23.6 4.8 1.8

41
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 8. Examples of artificially generated seismic sequences.


200 Set 1
Acceleration, cm/s2

Tg = 2.87 s Tg = 2.93 s Seq. 1_2

Mainshock Aftershock
-200
0 100 200 300 400
200 Set 2
Acceleration, cm/s2

Tg = 2.87 s Tg = 2.93 s Seq. 1_2

Mainshock Aftershock
-200
0 100 200 300 400
200 Set 3
Acceleration, cm/s2

Tg = 2.87 s Seq. 1_2


Tg = 2.93 s
0
Aftershock
Mainshock
-200
0 100 200 300 400
Time, s

tends to produce highly biased response [7,12,31], the latter approach • Set 1: 100% PGV SCT (for the Mainshock) and 100% PGVSCT (for the
is employed in this study. Aftershock)
The second part of Table 4 shows a list of eight earthquake ground • Set 2: 100% PGV SCT (for the Mainshock) and 70% PGVSCT (for the
motions used to generate an artificial set of seismic sequences. This set, Aftershock)
referred to as Set A in Ref. [1], has a total of 28 sequences, which were • Set 3: 100% PGV SCT (for the Mainshock) and 35% PGVSCT (for the
generated as follows: 1) the first four motions (M1 to M4) were selected Aftershock).
as mainshocks; 2) then, all the records (except where the mainshock
and the aftershock were the same) were used as aftershocks to generate Fig. 8 shows three artificially generated sequences for sets 1 to 3,
the sequences. It should be noted that the mainshocks (M1 to M4) have respectively. In the figure it is indicated the predominant period of the
a predominant period of vibration close to 3 s, while the aftershocks mainshock and that of the aftershock – which in this case are similar but
have a predominant period either close to 3 s (M1 to M4) or around 2 s this is not always the case. It should be noted that, especially for Set 1,
(the rest of the records). This characteristic allows having seismic se- the PGA value of the aftershock can be greater than that of the main-
quences in agreement with the as-recorded sequence illustrated in shock – even when they have been scaled to reach the same PGV value.
Fig. 7, i.e. mainshocks with periods close or longer than those of the It is also observed that a zero-acceleration segment of 40 s was
aftershocks. added between the main shock and the aftershock in order to ensure the
Regarding the seismic intensity measure, the PGV was selected be- rest position of the models during the dynamic analyses. In the same
cause it is highly correlated with the seismic energy imposed to struc- way, a 40 s segment was added at the end of the aftershock.
tures located in soft soils [32]. For this investigation and similar to that On the other hand, it is highlighted that the scaling factors required
of Ref. [1], the seismic intensity (i.e. PGV) of the mainshocks was scaled to match the PGVSCT value were very large in some cases – reaching
to match that of the well-known SCT-EW record of the M8.1 Michoacán, values greater than 10. Such large factors may have an effect in the
México Earthquake of 19/09/1985 - i.e. PGVSCT = 61.1 cm/s. Then, the deformation demands of short period structures as studied by Quiroz
intensity of the aftershocks was scaled to reach 100%, 70% and 35% of et al. [33] – however, it is considered that the effect is similar in all the
the mainshock's PGV. With this, the effect of the seismic intensity of the studied cases and the relative comparisons, objective of this investiga-
aftershock, relative to that of the mainshock, is assessed. It should be tion, are still valid.
noted that in total three sets of 28 seismic sequences were generated, Figs. 9 to 11 show the mean 5%-damped elastic and inelastic dis-
namely: placement spectra generated from the artificial seismic sequences of set
1, 2 and 3, respectively. While the demands produced by the

42
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 9. Mean displacement spectra generated


180 Mainshock 25
from the artificial seismic sequences of Set 1.
160 Sequence Mainshock
140 20 Sequence
Displacement, cm

Displacement, cm
120
100 Range of 15
interest
80
60
10
40
5
20
0 3-storeys 6-storeys 9-storeys
0
0 1 2 3 4 5
0.6 0.8 1 1.2 1.4
Period, sec Period, sec

a) Periods between 0.2 and 5 s b) Periods between 0.6 and 1.5 s

Fig. 10. Mean displacement spectra gener-


180 Mainshock 25
ated from the artificial seismic sequences of
160 Sequence Mainshock Set 2.
140 20
Displacement, cm

Sequence
Displacement, cm

120
100 Range of 15
interest
80
60 10
40
5
20
0 3-storeys 6-storeys 9-storeys
0
0 1 2 3 4 5
0.6 0.8 1 1.2 1.4
Period, sec Period, sec

a) Periods between 0.2 and 5 s b) Periods between 0.6 and 1.5 s

mainshocks are represented by continuous lines, those by the seismic demanded by the sequences tend to be larger than those demanded by
sequences are represented by discontinuous lines. The elastic responses the mainshocks; 2) while the increases of the mean displacements de-
are represented by dark lines and the inelastic responses by grey lines. manded by the seismic sequences are very significant for set 1, they
The inelastic spectra were estimated considering a strength reduction may be considered negligible for sets 2 and 3, especially for set 3, where
factor, as defined in [34], of Ry = 4, and elastic-perfectly plastic be- the sequence spectra are the same as those of the mainshock spectra;
haviour. While Figs. 9a, 10a and 11a show a period range between 0.2 and 3) the inelastic displacements in short-period range are sig-
and 5 s, Figs. 9b, 10b and 11b show a more specific period range (i.e. nificantly larger than the elastic displacements – which is in agreement
between 0.6 and 1.5 s) in order to have a better insight into the period with findings by others (e.g. [35]).
range of the structures studied in this paper (see Table 3). Note that the Finally and in order to assess the effects of increased seismic in-
period range of interest is indicated in Figs. 9a, 10a and 11a, and cor- tensity on the dynamic response of the studied models, an additional
responds to short period structures (i.e. structures with period of vi- scenario was analysed by increasing all the seismic sequences by a
bration shorter than the predominant period of the ground motions). factor of 1.5. Although it is recognised that this increase is somehow
Figs. 9b, 10b and 11b show the elastic displacement demands and in- arbitrary, it was included in this study because it may be of interest to
dicate the fundamental periods of vibration corresponding to the three-, some readers, as maximum credible earthquakes could reach ground
six- and nine-storey frames. motion intensities 1.5 times that of design earthquakes. The results are
After analysing Figs. 9 to 11, a few observations are worth to be described at the end of the next section.
highlighted: 1) for sets 1 and 2, the inelastic and elastic displacements

180 25 Fig. 11. Mean displacement spectra gener-


ated from the artificial seismic sequences of
160
Set 3.
140 20
Displacement, cm

Displacement, cm

120
100 Range of 15
interest
80
60
10
40
5
20
0 3-storeys 6-storeys 9-storeys
0
0 1 2 3 4 5
0.6 0.8 1 1.2 1.4
Period, sec Period, sec

a) Periods between 0.2 and 5 s b) Periods between 0.6 and 1.5 s

43
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

1 1 1 Fig. 12. Heightwise distribution of mean IDR


Mainshocks
demands for Case 0: response to Set 1.
Sequences
0.8 0.8 0.8

0.6 0.6 0.6


z/H

z/H

z/H
0.4 0.4 0.4

0.2 0.2 0.2

Set 1 (100%M & 100%A) Set 1 (100%M & 100%A) Set 1 (100%M & 100%A)
0 0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03 0 0.01 0.02 0.03
IDR IDR IDR

a) Three-storey building b) Six-storey building c) Nine-storey building

1 1 1 Fig. 13. Heightwise distribution of the mean


Mainshocks
IDR demands for Case 0: response to Set 2.
Sequences
0.8 0.8 0.8

0.6 0.6 0.6


z/H

z/H

z/H
0.4 0.4 0.4

0.2 0.2 0.2

0 Set 2 (100%M & 70%A) 0 Set 2 (100%M & 70%A) 0 Set 2 (100%M & 70%A)
0 0.01 0.02 0.03 0 0.01 0.02 0.03 0 0.01 0.02 0.03
IDR IDR IDR

a) Three-storey building b) Six-storey building c) Nine-storey building

4. Response under artificial seismic sequences frames also exhibited larger mean IDR due to aftershocks scaled at the
same PGV of the mainshock. Another observation derived from ex-
The seismic performance of the case-study frames is evaluated in amining Fig. 12 is that peak IDR triggered by the mainshocks is around,
terms of the peak inter-storey and residual inter-storey drift demands, or larger, than 0.01, which implies that the frames reached their
which is discussed next. yielding displacement capacity according to Fig. 4. This situation may
result in a small level of damage related to the IO performance level.
However, the IDR demands increase as a consequence of aftershocks,
4.1. Peak inter-storey drift demands
which imply that the hospitals of Case 0 cannot be occupied im-
mediately, and they fail to accomplish the purpose of an essential fa-
In order to provide a context of the frames' performance, FEMA 356
cility to remain operational.
[36] recommendations specify inter-storey drift index thresholds of
In order to assess the effects of BRBs, the maximum IDR through the
0.7%, 2.5%, and 5% for the immediate occupancy (IO), life safety (LS),
height of the frames corresponding to each frame and each case was
and collapse prevention (CP) performance levels, respectively. Figs. 12
recorded for every sequence of Set 1. Thus, Fig. 15 shows the ratio of
to 14 show the heightwise distribution of mean peak inter-storey drift
the IDR demands produced by the sequences (IDRSequence) to those of the
ratios (IDR) corresponding to Case 0 (i.e. without BRBs) of the 3-, 6-,
mainshocks (IDRMainshock), namely IDRSequence/IDRMainshock, versus the
and 9-storey frame models when subjected to the seismic sequences
maximum IDR demands generated by the mainshocks, IDRMainshock.
included in sets 1 to 3. For comparison purposes, the vertical axis shows
Central tendencies corresponding to each case, determined using least
the relative storey height normalised by the building's height, z/H.
squares method, are also shown for clarity purposes. While pairs cor-
From Fig. 12, it is observed that IDR increase significantly when the
responding to Cases 0 to 4 are compared separately in Fig. 15a, they are
aftershock has the same seismic intensity of the mainshock. This is
shown all together in Fig. 15b. From the figure, the following ob-
observation is in good agreement with previous observations related to
servations are reported: 1) the IDRSequence/IDRMainshock ratios are equal
Fig. 9 and the findings reported in Ref. [1], where reinforced concrete

1 1 1 Fig. 14. Heightwise distribution of the mean IDR


Mainshocks
demands for Case 0: response to Set 3.
Sequences
0.8 0.8 0.8

0.6 0.6 0.6


z/H

z/H

z/H

0.4 0.4 0.4

0.2 0.2 0.2


Set 3 (100%M & 35%A) Set 3 (100%M & 35%A) Set 3 (100%M & 35%A)
0 0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03 0 0.01 0.02 0.03
IDR IDR IDR

a) Three-storey building b) Six-storey building c) Nine-storey building

44
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 15. Comparison of IDR demands in all the


4 0 4
IDRSequence / IDRMainshock

studied buildings subjected to Set 1.

IDRSequence / IDRMainshock
3.5 1 3.5
3 2 3
2.5 3 2.5
2
4 2
1.5 1.5
1
1
0.5
0.5
0
0 0 0.005 0.01 0.015 0.02
0 0.005 0.01 0.015 0.02
IDRMainshock IDRMainshock

a) Cases 0 to 4 b) All cases

Fig. 16. Mean of the maximum IDR demands


0.0250 Mainshocks 0.0250 0.0250
for Cases 0 to 4 when subjected to Set 1.
Sequences
0.0200 0.0200 0.0200

Max. IDR
Max. IDR

Max. IDR

0.0150 0.0150 0.0150

0.0100 0.0100 0.0100

0.0050 0.0050 0.0050

0.0000 0.0000 0.0000


0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
Case Case Case

a) Three-storey b) Six-storey c) Nine-storey

to or higher than one, which is in agreement with trends observed in assumed that all the plastic behaviour is concentrated into the BRBs
Figs. 9 and 12; 2) the mean IDR demands due to the effect of the while the frame elements may remain within its linear-elastic even
aftershocks (IDRSequence) tend to reduce with respect to the IDR demands though maximum IDR demands are incremented by the aftershocks.
triggered by the mainshock (IDRMainshock, see horizontal axis) when the Consequently, 3- and 6-storey frames designed under Case 4 criteria
latter increases; and 3) Cases 0 and 3 present, in general, higher might be close to the IDR associated to the IO performance level (in
IDRSequence/IDRMainshock ratios than Cases 1, 2, and 4. The latter ob- blue dashed line), but the other frames designed for other cases, such as
servation can be explained with the help of Table 3 and Fig. 9, since Case 0 and 3, would exceed the IO performance level particularly under
frames designed for Cases 1, 2, and 4 are stiffer than those designed for aftershocks, which is not acceptable for essential facilities in seismic
Cases 0 and 3, which implies that they are farther away of the re- regions.
sonance zone than Cases 0 and 3. This characteristic allows smaller IDR From the above discussion, it can be concluded that even when the
demands due to the mainshock for Cases 1, 2 and 4 than in Cases 0 and aftershocks of Set 1 may produce adverse effects (i.e. increase the IDR
3. demands), frames designed under the Case 4 criteria can mitigate them
A comparison of mean maximum IDR demands along-height cor- and ensure that the main structure remains elastic while all the damage
responding to each case of the three-, six-, and nine-storey frames when is concentrated in the replaceable elements (i.e. the BRBs) to be close to
subjected to the mainshocks and the seismic sequences included in Set 1 the IO performance level.
is shown in Fig. 16a–c, respectively. It can clearly be seen that the
amplitude of IDR demands increase due to the aftershocks irrespective
4.2. Residual inter-storey drift demands
of the design case (i.e. with or without BRBs) in each frame. However, it
can also be observed that frames designed under the criteria given in
Fig. 17 show the amplitude of the residual inter-storey drift ratios
Case 4 exhibit the smallest IDR demands, which are around, or smaller,
(RIDR) along height for the three frames designed under Case 0 cri-
than 0.01 both under the mainshocks and the full sequences. For this
terion when subjected to sets 1. Similar plots corresponding to sets 2
IDR demands, and recalling the capacity curves shown in Fig. 4, it is
and 3 are also shown in Figs. 18 and 19, respectively. From Fig. 17, it is

1 1 1 Fig. 17. Heightwise distribution of the mean


Mainshocks
RIDR demands in Case 0: response to Set 1.
Sequences
0.8 0.8 0.8

0.6 0.6 0.6


z/H

z/H
z/H

0.4 0.4 0.4

0.2 0.2 0.2

0 Set 1 (100%M & 100%A) Set 1 (100%M & 100%A) Set 1 (100%M & 100%A)
0 0
0 0.0025 0.005 0 0.0025 0.005 0 0.0025 0.005
RIDR RIDR RIDR

a) Three-storey building b) Six-storey building c) Nine-storey building

45
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

1 1 1 Fig. 18. Heightwise distribution of the mean


Mainshocks
RIDR demands in Case 0: response to Set 2.
Sequences
0.8 0.8 0.8

0.6 0.6 0.6


z/H

z/H
z/H
0.4 0.4 0.4

0.2 0.2 0.2


Set 2 (100%M & 70%A) Set 2 (100%M & 70%A) Set 2 (100%M & 70%A)
0 0 0
0 0.0025 0.005 0 0.0025 0.005 0 0.0025 0.005
RIDR RIDR RIDR

a) Three-storey building b) Six-storey building c) Nine-storey building

1 1 1 Fig. 19. Heightwise distribution of the mean


Mainshocks
RIDR demands in Case 0: response to Set 3.
Sequences
0.8 0.8 0.8

0.6 0.6 0.6


z/H

z/H
z/H

0.4 0.4 0.4

0.2 0.2 0.2


Set 3 (100%M & 35%A) Set 3 (100%M & 35%A) Set 3 (100%M & 35%A)
0 0 0
0 0.0025 0.005 0 0.0025 0.005 0 0.0025 0.005
RIDR RIDR RIDR

a) Three-storey building b) Six-storey building c) Nine-storey building

observed that the RIDR demands due to the full sequences included in 4.3. Effects of seismic intensity
Set 1 tend to increase, particularly for the three- and six-storey frames,
with respect to those triggered by the mainshocks. Unlike previous As introduced at the end of Section 2, the seismic intensity of the
results, RIDR demands slightly increase due to the sequences of Set 2 sequences was scaled to 1.5 in order to assess its effects on the studied
(Fig. 18), while RIDR demands remained unchanged under sequences structures. For simplicity only the results of Set 1 are presented here.
included in Set 3 (see Fig. 19). Further analysis of Figs. 17 to 19 sug- Fig. 21 shows the number of predicted collapses for all the studied cases
gests that increases of RIDR may only be relevant for aftershocks with of the three-, six- and nine-storey frames, respectively. Collapse was
seismic intensity similar to that of the mainshock. However, this is determined following the recommendations of the FEMA P-58 Project
different from findings by Ruíz-García et al. [1], where increases of [37], i.e. when the model showed numerical instability, or when a small
RIDR were observed in the lower storeys of several concrete buildings increase in the seismic intensity generated a significant increase of
due to the seismic sequences with aftershock intensities of 35% and deformation. From the figure, it is observed that the three-storey
70% that of the mainshock. This is expected because of stiffness de- building does not present collapses, while the nine-storey building
gradation of concrete structures. presents a significant number of collapses. This increase of the number
Now the effects of the BRBs on the RIDR demands are further of collapses with the number of storeys may be attributed to several
analysed for the seismic sequences of Set 1. Fig. 20 shows that, even factors, such as: 1) more contribution of the gravity loads as the number
when the RIDR demands increase due to the aftershocks of Set 1, there of storeys increases; 2) the higher buildings have longer periods of vi-
are some cases where they can be controlled to remain small. In other bration, which in this study are closer to the predominant period of the
words, while Cases 1, 2 and 4 consistently produced smaller RIDR de- soil; and 3) analysing Fig. 4, it is observed that the lateral load capacity
mands, Cases 0 and 3 generated the largest ones. A similar trend was of the models, normalised by their total weight, is slightly smaller in the
previously observed in Fig. 16 for the IDR demands. Therefore, it can be taller buildings. It is of interest to see that most of the collapses oc-
argued that, by constraining the IDR demands, the RIDR demands can curred for Case 0 (i.e. without BRBs), and most of them were triggered
also be constrained to avoid undesirable increases due to aftershocks. by the aftershocks. In other words, the use of BRBs helps significantly to

Fig. 20. Mean of the maximum RIDR demands


0.0030 Mainshocks 0.0030 0.0030
for Cases 0 to 4 when subjected to Set 1.
Sequences

0.0020 0.0020 0.0020


Max. RIDR
Max. RIDR

Max. RIDR

0.0010 0.0010 0.0010

0.0000 0.0000 0.0000


0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
Case Case Case

a) Three storeys b) Six storeys c) Nine storeys

46
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

Fig. 21. Number of predicted collapses for Set 1,


28 Sequences 28 Sequences 28 Sequences
scaled to PGV = 1.5 PGVSCT.
Mainshocks Mainshocks Mainshocks

Number of collapses

Number of collapses
Number of collapses

21 21 21

14 14 14

7 7 7

0 0 0
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
Case Case Case

a) Three storeys b) Six storeys c) Nine storeys

Fig. 22. Dual SDOF oscillator. (For interpretation of the re-


Frame V ferences to color in this figure, the reader is referred to the
kframe web version of this article.)

Vy_total

cframe Vy_frame
1
m
BRBs Vy_BRBs
2
kBRBs
kBRBs
d
cBRBs dy_BRBs dy_frame

a) Oscillator b) Capacity curves

reduce the probability of collapse that aftershocks may trigger. It is also total height of the structure, a is a factor that considers that the IDR
of interest to analyse the mechanism that triggered collapse of the demands are not uniform along the height of the building, and r can be
studied frames. In all these cases, dynamic instability occurred due to estimated using the following equation:
excessive lateral inter-storey deformation of one or more storeys; this is,
once the lateral capacity was exceeded by P-Δ effects, the frames went ∑ mi d i
r= droof
off to collapse. Note on Fig. 21b and c that most of the collapses oc- ∑ mi di2 (2)
curred for Case 0, where this type of collapse mechanism is expected
due to the structure typology (i.e. moment resisting frames without In Eq. (2), mi is the mass at the i-th floor of the building, while di and
braces). On the other hand, flexo-compression failure of columns might droof are the displacements at each floor and at the roof, respectively. It
have been expected for Case 1 (see Fig. 21c). However, flexo-com- should be mentioned that assuming a linear displacement profile is a
pression did not happen because the contribution of the BRBs to the reasonable approximation for the estimation of r for low- to medium-
structure's lateral load capacity was small (less than 20%); which did rise structures according to Refs. [38,39].
not caused significant increases of axial demands on the columns. The displacement demands of SDOF oscillators with equivalent
Therefore, sideway mechanism continued occurring for Case 1. Here, properties to those of the case-study frames described in Section 2 were
the reader is aware that flexo-compression failure of columns might estimated in the second stage of this investigation. Then, along with the
actually be expected in cases where the contribution of the BRBs is large results of Section 4, the mean and dispersion of parameter a are de-
and lateral deformations are significant. However, these conditions did termined. For this purpose, the dual system approach described in Ref.
not happen in this particular case study. [23] is used to model the equivalent SDOF oscillators. For instance,
Fig. 22 shows a dual system and its capacity curve (black line), which is
5. MDOF-to-SDOF demand ratios composed of the summation of two lower curves representing the
equivalent capacity of the main frame (red line) and the BRBs (blue
5.1. Inter-storey drift ratios (IDR) line). The stiffness, yielding displacement, strength capacity and
damping coefficient of the main frame are represented by kframe, dy_frame,
A very useful tool to practicing structural engineers is the estimation Vy_frame and cframe, respectively; while those of the BRBs are kBRBs,
of displacement demands of multi-degree-of-freedom (MDOF) systems, dy_BRBs, Vy_BRBs and cBRBs. Further description of the dual system ap-
such as the case-study frames considered in this study, from the dis- proach can be found in Refs. [23,22].
placement demands of single-degree-of-freedom (SDOF) oscillator with Recommendations given in the FEMA 356 [36] report were used to
equivalent properties of the MDOF system. For instance, maximum IDR obtain equivalent bilinear capacity curves for the frame and the BRBs
demand can be approximately estimated as follows: (also referred as the primary and the secondary parts of the dual system,
respectively). As an example, Fig. 23 shows the equivalent bilinear
IDR = a⋅r (dSDOF HT ) (1)
capacity curves obtained from Cases 0 and 2 of the six-storey frame.
where dSDOF is the peak displacement demand of an SDOF oscillator From them, it is possible to obtain the parameters for the equivalent
with equivalent properties of a MDOF under consideration, HT is the SDOF oscillator. Therefore, a similar procedure was followed for the

47
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

2.0E+07 Fig. 23. Equivalent capacity curves for the six-storey


2.0E+07
building.
Total
1.5E+07
Lateral Load, N

1.5E+07

Lateral Load, N
1.0E+07 1.0E+07 Frame

5.0E+06 5.0E+06 BRBs

0.0E+00 0.0E+00
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Displacement, m Displacement, m

a) Case 0 (NoBRBs)
b) Case 2

restoring forces defined by the equivalent capacity curves of the frame


and the BRBs, respectively (see Fig. 23).
Table 5
Equivalent properties of the SDOF oscillators. The solution of Eq. (3) is conducted using a Matlab® subroutine,
available in Ref. [22], which yields dSDOF in Eq. (1). After that, dSDOF is
Building Case Teq Tframe_eq TBRBs_eq dy_frame dy_BRBs α1 α2 r multiplied by r and divided by HT, with a still unknown in Eq. (1). To
[s] [s] [s] [m] [m] [%] [%]
find a, the quantity r(dSDOF/HT) is set to be equal to the peak IDR value
3-Storeys 0 0.90 0.90 – 0.083 – 14.6 – 1.31 found in Section 4, so that the value of a can be found. The afore-
1 0.76 0.9 1.45 0.083 0.0175 14.4 3.8 1.31 mentioned procedure is repeated for all mainshocks and sequences in-
2 0.77 0.98 1.23 0.083 0.0174 11.2 4.2 1.31 cluded in each set. For instance, Figs. 24a and 25a show the values of
3 0.88 0.98 1.97 0.083 0.0169 11.2 4.1 1.31
the parameter a corresponding to all cases of the 6-storey frame when
4 0.68 0.98 0.94 0.083 0.0179 11.2 4.3 1.31
6-Storeys 0 1.26 1.26 – 0.138 – 10.0 – 1.41
subjected to the mainshocks and the full seismic sequences included in
1 1.02 1.26 1.74 0.138 0.0459 10.0 5.8 1.41 Set 1, respectively. It should be noted that same values of a are seen in
2 1.00 1.70 1.23 0.127 0.0524 10.3 7.0 1.41 Fig. 24a, since only four seismic records were used as mainshocks,
3 1.14 1.70 1.52 0.127 0.0476 10.3 6.7 1.41 while different values of a are observed in Fig. 25a since different
4 0.90 1.70 1.06 0.127 0.0571 10.3 7.3 1.41
seismic records were used as aftershocks.
9-Storeys 0 1.52 1.52 – 0.198 – 14.6 – 1.44
1 1.27 1.52 2.29 0.198 0.0729 14.9 5.0 1.44 From Figs. 24b and 25b, it is noted that Cases 0 and 3 lead to larger
2 1.31 2.03 1.71 0.205 0.0825 12.9 5.3 1.44 mean values of a than those of the other cases, while Case 4 lead to the
3 1.41 2.03 1.95 0.205 0.0777 12.9 5.2 1.44 smallest value, which suggests that Case 4 yields more uniform lateral
4 1.19 2.03 1.47 0.205 0.0910 12.9 5.6 1.44 displacements along the height of the frame (i.e. parameter a is a direct
measure of the uniformity of the IDR demands). Additionally, Figs. 24c
and 25c illustrate the coefficients of variation corresponding to each
remaining cases and case-study frames, where Table 5 reports the
case, which suggests that dispersion of parameter a is not significant.
equivalent parameters (where Tframe_eq, dy_frame, and α1 are the period of
Finally, it is also worth noting that the mean and coefficient of
vibration, the yielding displacement, and the post-yielding stiffness
variation of parameter a tend to be larger when considering the full
ratio of the bare frame, respectively; TBRBs_eq, dy_BRBs, and α2 are the
sequences than those computed from the mainshocks, which might
period of vibration, the yielding displacement and the post-yielding
indicate that the aftershocks tend to decrease the uniformity along-
stiffness ratio of the bracing system, respectively; Teq is the total period
height of the IDR demands.
of the dual system and r was defined in Eq. (2)).
Lastly, the mean value of a is depicted in Fig. 26 against the con-
The displacement demand of the equivalent dual SDOF oscillator
tribution of the main frame, in percentage, to the stiffness of the
under a given earthquake ground motion is found by solving the dy-
structure. Note that the data points are divided in two groups. The first
namic equation of motion for a dual system given by [23]:
group corresponds to Cases 2, 3 and 4 (whose frames were designed
2π f (u, u̇) f (u, u̇) ⎤ under gravity loads only); while the second group corresponds to Cases
u¨ (t ) + 2ξ u̇ (t ) + ⎡ s1 + s2 = −u¨ g (t )
T ⎢ m m ⎥ (3) 0 and 1 (whose frames have more strength capacity). It is clearly ob-
⎣ ⎦
served that the smaller the contribution of the main frames to the
where u (t ), u̇ (t ), and u¨ (t ) are the displacement, velocity, and accel- stiffness, the smaller the average value of a. In other words, the higher
eration of the dual equivalent oscillator for the time t; u¨ g (t ) is the the contribution of the BRBs to the stiffness, the more uniform the
ground acceleration, ξ is the damping ratio, T is the equivalent period of deformations through the height of the frames are.
vibration, m is the first-mode mass, while fs1 (u, u̇) and fs2 (u, u̇) are the

2.00 1.50 0.25 Fig. 24. Estimation of parameter a for Set 1:


Coefficient of variation of a

mainshocks.
1.45 0.20
mean value of a

1.50
1.40
0.15
1.00 1.35
0.10
Case 0 Case 1 1.30
0.50 Case 2 Case 3 1.25 0.05
Case 4
1.20 0.00
0.00
0 1 2 3 4 0 1 2 3 4
0 4 8 12 16 20 24 28
Case Case
Mainshock number

a) Values of a b) Mean c) Coeff. of variation

48
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

1.50 0.25 Fig. 25. Estimation of parameter a for Set 1: full

Coefficient of variation of a
2.00 sequences.
1.45 0.20

Mean value of a
1.50 1.40
0.15
1.35
1.00 0.10
1.30
Case 0 Case 1 0.05
0.50 1.25
Case 2 Case 3
Case 4 1.20 0.00
0.00
0 1 2 3 4 0 1 2 3 4
0 4 8 12 16 20 24 28
Case Case
Seismic sequence number

a) Values of a b) Mean c) Coeff. of variation

Fig. 26. Parameter a for Set 1 vs. stiffness contribution of


1.50 1.50 Case 3 Case 0
the main frame.
Average value of a

Case 3 Case 2
Average value of a
1.40 1.40
Case 0
Case 2
1.30 1.30 Case 4 Case 1
Case 1
1.20 Case 4 1.20
1.10 1.10
1.00 1.00
0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100%
Contribution of main frames to stiffness Contribution of main frames to stiffness

a) Mainshocks b) Full sequences

5.2. Residual inter-storey drift ratios IDR. The introduced procedure is consistent with the method suggested
in FEMA P-58 [37] recommendations, and other approaches (e.g. [40]).
Similarly to the estimation of peak IDR, estimation of residual inter- As noted in Refs. [22,41]., the amplitude of the residual displace-
storey drift ratios (RIDR) is important to assess the seismic performance ment demands is highly dependent on the zone on which the peak
of structures equipped with BRB. Therefore, a simplified procedure is displacement demands are located within the capacity curve of a dual
introduced to estimate the amplitude of RDIR demands from the peak SDOF oscillator, as shown in Fig. 27. Note that, while Zone I corre-
sponds to linear-elastic response on both parts of the dual system, Zone
III corresponds to inelastic response on both of them. For Zone II, the
V Zones main frame exhibits linear-elastic response while the BRBs behave in-
elasticity. Considering that, it can be anticipated that if the peak dis-
(I) (II) (III) placement demands are located within Zone I (represented by a squared
Vy_total
dot), the residual displacement demands are zero because the primary
Total
and the secondary parts of the dual system have enough restoring force
capacity to return the structure to its original (i.e. zero-lateral dis-
Vy_frame frame placement) position. On the other hand, if the peak displacement de-
mands are located in Zone II (represented by a circular dot), residual
Vy_BRBs displacements would exist due to the plastic deformation of the sec-
BRBs ondary part. Finally, if the peak displacements are located in Zone III
(represented by a triangular dot), the residual displacements would be
very large since both the primary and the secondary parts exhibit
plastic deformations.
dy_frame dy_BRBs d From the results obtained during the analysis phase of this study,
two relationships between the RIDR and the IDR demands for all the
Fig. 27. Zones of the capacity curve of a dual system.
studied buildings and cases were plotted in Fig. 28. For convenience,

Fig. 28. RIDR and IDR demands.


0.006 Mean Seqs. Sequences 0.006 Mean Seqs. Sequences
Mean Mainsh. Mainshocks Mean Mainsh. Mainshocks
0.005 0.005
0.004 0.004
RIDR

RIDR

0.003 0.003
0.002 0.002
0.001 0.001
0 0
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
IDR IDR

a) In ZoneII b) In Zone III

49
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

the pairs with IDR demands between 0.003 and 0.01 (i.e. those located oscillators with equivalent properties by using Eq. (1). Where the
within Zone II as defined in Fig. 27) are grouped in Fig. 28a, while those mean values of parameter a, which takes into account the variations
pairs with IDR demands larger than 0.01 (i.e. Zone III) are grouped in of the inter-storey drift demands through the height of the frames,
Fig. 28b. It is observed that the RIDR demands remain significantly were estimated and are provided.
small when IDR demands are in Zone II, while the RIDR demands tend • It is interesting to see in Fig. 26 that the higher the contribution of
to increase significantly when IDR demands are in Zone III. This is in the BRBs to the stiffness, the smaller the values of a were (i.e. more
agreement with the observations reported by Guerrero et al. [22,41] on uniform the deformations through the height of the frames were
dual SDOF oscillators. observed due to increased participation of BRBs).
It is also worth mentioning that any significant difference in the • Although the magnitude of the RIDR demands is sensitive in many
trends of IDR-RIDR pairs was identified while examining Cases 0 to 4 of factors such as the post-elastic stiffness, it can be reasonably well
the case-study frames. Therefore, the trend reported in Fig. 28 is valid estimated using the proposed equation (Eq. (4)); which is valid for
for all the cases. Therefore, the following relationships are proposed in short-period structures located in soft soils. If needed, corrections
order to estimate the peak RIDR demands from the peak IDR demands: could be conducted based on the findings by Guerrero et al. [22,41].

⎧ 0; IDR < IDRy2 ; (Zone I)


⎪ It should be mentioned that the methodology introduced in this
RIDR = b; IDRy2 ≤ IDR ≤ IDRy1; (Zone II) paper to estimate peak and residual inter-storey drift demands is lim-

⎪ m (IDR − IDRy1) + b; IDR > IDRy1; (Zone III) (4) ited to short-period framed-buildings equipped with BRBs located on

the lake-bed zone of Mexico City. Particularly, the estimated para-
where IDRy1 is the inter-storey drift ratio at which the main structure
meters to be used in Eqs. (1) and (4) were obtained from the structural
yields, IDRy2 is the inter-storey drift ratio at which the BRBs yield, and
layouts and earthquake ground motions considered in this study, which
IDR is the peal inter-storey drift ratio of the dual-system, which can be
can vary under other circumstances. To validate the use of Eqs. (1) and
estimated from Eq. (1). Parameters b and m were estimated to fit the
(4) for other type of structures (e.g. long-period structures equipped
mean values of the data points observed in Fig. 28, which correspond to
with BRBs), additional research is under development.
b = 0.0002 and m = 0.15 for the mainshocks, and b = 0.0004 and
m = 0.11 for the full sequences.
Acknowledgments
It should be mentioned that the RIDR demands reported in Fig. 28
are smaller than those computed from the method introduced in the
The first and second authors acknowledges to the Universidad
FEMA P-58 report [37]. This might be attributed to several factors,
Nacional Autónoma de México (UNAM) and Universidad Michoacana
including the level of positive post-yielding stiffness ratios (α1 and α2),
de San Nicolas de Hidalgo (UMSNH) for providing the facilities to
observed in Fig. 23 and reported in Table 5, which tend to significantly
conduct this study.
reduce the RIDR demands (e.g. [42,43]). If zero or negative post-
yielding stiffness ratios are found, corrections could be conducted based
References
on the findings by Guerrero et al. [22,41].
[1] J. Ruiz-García, M.V. Marín, A. Terán-Gilmore, Effect of seismic sequences in re-
6. Conclusions inforced concrete frame buildings located in soft-soil sites, Soil Dyn. Earthq. Eng. 63
(2014) 56–68.
The response of three-, six- and nine-storey case-study frames, re- [2] E. Rosenblueth, R. Meli, The 1985 Mexico earthquake: causes and effects in Mexico
City, Concr. Int. 8 (1986) 23–34.
presentative of essential short-period structures located in the lake-bed [3] W.Y. Kam, S. Pampanin, The seismic performance of RC buildings in the 22
zone of Mexico City were subjected to artificial mainshock-aftershock February 2011 Christchurch earthquake, Struct. Concr. 12 (2011) 223–233.
seismic sequences in order to assess the effects of buckling-restrained [4] W.Y. Kam, S. Pampanin, K. Elwood, Seismic performance of reinforced concrete
buildings in the 22 February Christchurch (Lyttelton) earthquake, Bull. N. Z. Soc.
braces on the inter-storey drift ratio (IDR) and residual inter-storey drift Earthq. Eng. 44 (2011) 239–278.
ratio (RIDR) demands. The following conclusions are offered from this [5] R. Shcherbakov, D.L. Turcotte, J.B. Rundle, Aftershock statistics, Pure Appl.
investigation: Geophys. (2005) 1051–1076.
[6] L. Astiz, H. Kanamori, H. Eissler, Source characteristics of earthquakes in the

• As seen in Table 3 and Fig. 5, the structures without BRBs (i.e. Case
Michoacan seismic gap in Mexico, Bull. Seismol. Soc. Am. 77 (1987) 1326–1346.
[7] J. Ruiz-García, Mainshock-aftershock ground motion features and their influence in
0) were in general more flexible than the structures equipped with building's seismic response, J. Earthq. Eng. 16 (2012) 719–737.
[8] S.A. Mahin, Effects of duration and aftershocks on inelastic design earthquakes,
BRBs (Cases 1 to 4). The latter cases were more convenient in this
Seventh World Conference on Earthquake Engineering, Istabul, 7 1980, pp.
study because they were farther away of the resonance zone (see 677–679.
Fig. 9) – as a consequence, smaller demands are expected in struc- [9] G.D. Hatzigeorgiou, Ductility demand spectra for multiple near- and far-fault
earthquakes, Soil Dyn. Earthq. Eng. 30 (2010) 170–183.
tures equipped with BRBs.

[10] C. Amadio, M. Fragiacomo, S. Rajgelj, The effects of repeated earthquake ground
Figs. 9 to 14 and 17 to 19 show that aftershock ground motions are motions on the non-linear response of SDOF systems, Earthq. Eng. Struct. Dyn. 32
very significant when their intensity is similar to that of the main- (2003) 291–308.
shocks. Therefore, aftershock effects shall be taken into account in [11] G.D. Hatzigeorgiou, A.A. Liolios, Nonlinear behaviour of RC frames under repeated
strong ground motions, Soil Dyn. Earthq. Eng. 30 (2010) 1010–1025.
the design stage when the seismic intensity of the aftershocks is [12] J. Ruiz-García, J.C. Negrete-Manriquez, Evaluation of drift demands in existing
expected to be similar to that of the mainshock. steel frames under as-recorded far-field and near-fault mainshock–aftershock

• Analysis of Figs. 15, 16 and 20 shows that, although aftershock ef- seismic sequences, Eng. Struct. 33 (2011) 621–634.
[13] J. Erochko, C. Christopoulos, R. Tremblay, H. Choi, Residual drift response of
fects cannot be eliminated, they can be mitigated if a proper design SMRFs and BRB frames in steel buildings designed according to ASCE 7-05, J.
is conducted, i.e. constraining the peak IDR demands below the Struct. Eng. 137 (2011) 589–599.
yielding drift of the main frame. [14] G. Diaz-Martinez, J. Ruiz-Garcia, A. Teran-Gilmore, Response of structures to

• On the other hand, Fig. 21 shows that the probability of collapse of


seismic sequences corresponding to Mexican soft soils, Earthquakes and Structures
7 (2014) 1241–1258.
Case 0 (i.e. conventional structures without BRBs) is significantly [15] A. Filiatrault, C. Christopoulos, Principles of Passive Supplemental Damping and
increased when the seismic intensity was increased by 50%. This Seismic Isolation, IUSS Press, 2006.
[16] S. Merrit, C.-M. Uang, G. Benzoni, Subassemblage Testing of Corebrace Buckling-
was not the case of the frames equipped with BRBs (Cases 1 to 4) -
Restrained Braces, Report University of California, San Diego; La Jolla, CA, 2003.
where the use of BRBs helps to mitigate the probability of collapse. [17] C.J. Black, N. Makris, I.D. Aiken, Component testing, seismic evaluation and
• Figs. 24 and 25 show that the peak IDR demands can be estimated characterization of buckling-restrained braces, J. Struct. Eng. ASCE 130 (2004)
880–894.
with good accuracy from the peak displacement demands of SDOF

50
H. Guerrero et al. Journal of Constructional Steel Research 137 (2017) 37–51

[18] R. Tremblay, P. Bolduc, R. Neville, R. DeVall, Seismic testing and performance of 41 (2012) 2311–2330.
buckling-restrained bracing systems, Can. J. Civ. Eng. 33 (2006) 183–198. [32] A. Teran-Gilmore, A. Sanchez-Badillo, M. Espinosa-Johnson, Performance-based
[19] G. Della Corte, M. D'Aniello, R. Landolfo, Field testing of all-steel buckling-re- seismic design of reinforced concrete ductile buildings subjected to large energy
strained braces applied to a damaged reinforced concrete building, J. Struct. Eng. demands, Earthquakes and Structures 1 (2010) 69–91.
141 (2015) D4014004. [33] A. Quiroz-Ramírez, D. Arroyo, A. Terán-Gilmore, M. Ordaz, Evaluation of the in-
[20] C.-M. Uang, M. Nakashima, Steel buckling-restrained braced frames, in: tensity measure approach in performance-based earthquake engineering with si-
Y. Bozorgnia, V.V. Bertero (Eds.), Earthquake Engineering From Engineering mulated ground motions, Bull. Seismol. Soc. Am. 104 (2014) 669–683.
Seismology to, CRC Press, Performance-Based Engineering, 2004. [34] E. Miranda, V.V. Bertero, Evaluation of strength reduction factors for earthquake-
[21] H. Guerrero, T. Ji, J.A. Escobar, On the Upgrading of Hospitals in Mexico City by resistant design, Earthquake Spectra 10 (1994).
Using Buckling-Restrained Braces, Geotechnical and Structural Engineering [35] J. Ruiz-Garcia, E. Miranda, Inelastic displacement ratios for design of structures on
Congress, 2016, pp. 299–313. soft soils sites, J. Struct. Eng. 130 (2004) 2051–2061.
[22] H. Guerrero, Seismic Design and Performance of Hospital Structures Equipped with [36] FEMA-356, Agency FEM (Ed.), Prestandard and commentary for the seismic re-
Buckling-Restrained Braces in the Lakebed Zone of Mexico City (PhD Thesis), The habilitation of buildings, 2000 (Washington, D.C.).
University of Manchester, UK, 2016. [37] FEMA-P58, Seismic Performance Assessment of Buildings, Federal Emergency
[23] H. Guerrero, T. Ji, A. Teran-Gilmore, J.A. Escobar, A method for preliminary Management Agency, Washington, D.C., 2012.
seismic design and assessment of low-rise structures protected with buckling-re- [38] T.J. Maley, T.J. Sullivan, Corte G. Della, Development of a displacement-based
strained braces, Eng. Struct. 123 (2016) 141–154. design method for steel dual systems with buckling-restrained braces and moment-
[24] Kersting RA, Fahnestock LA, López WA. Seismic Design of Steel Buckling-restrained resisting frames, J. Earthq. Eng. 14 (2010) 106–140.
Braced Frames. NIST GCR 15-917-342015. [39] M.J.N. Priestley, G.M. Calvi, M.J. Kowalsky, Displacement-based Design of
[25] Opensees, Open Source Finite Element Platform for Earthquake Engineering Structures, Istituto Universitario di Studi Superiori di Pavia, Pavia, Italy, 2007.
Simulations, Univ. of California, Berkeley: Pacific Earthquake Engineering Center, [40] J. Ruiz-García, C. Chora, Evaluation of approximate methods to estimate residual
2014. drift demands in steel framed buildings, Earthq. Eng. Struct. Dyn. 44 (2015)
[26] F.C. Filippou, G.L. Fenves, Methods of analysis for earthquake-resistant structures, 2837–2854.
in: Y. Bozorgnia, V.V. Bertero (Eds.), Earthquake Engineering from Engineering [41] H. Guerrero, J. Ruiz-García, T. Ji, Residual displacement demands of conventional
Seismology to Performance-based Engineering, CRC Press, 2004. and dual oscillators subjected to earthquake ground motions characteristic of the
[27] J. Pantazopoulou, C.W. French, Slab participation in practical earthquake design of soft soils of Mexico City, Soil Dyn. Earthq. Eng. 98 (2017) 206–221.
reinforced concrete frames, ACI Struct. J. 98 (2001) 479–489. [42] G.A. MacRae, K. Kawashima, Post-earthquake residual displacements of bilinear
[28] P. Uriz, F.C. Filippou, S. Mahin, Model for cyclic inelastic buckling of steel braces, J. oscillators, Earthq. Eng. Struct. Dyn. 26 (1997) 701–716.
Struct. Eng. 134 (2008) 619–628. [43] E. Bojórquez, J. Ruiz-García, Residual drift demands in moment-resisting steel
[29] A. Terán-Gilmore, J. Ruiz-García, Comparative seismic performance of steel frames frames subjected to narrow-band earthquake ground motions, Earthq. Eng. Struct.
retrofitted with buckling-restrained braces through the application of force-based Dyn. 42 (2013) 1583–1598.
and displacement-based approaches, Soil Dyn. Earthq. Eng. 31 (2011) 478–490. [44] J.A. Díaz-Rodríguez, Characterization and engineering properties of Mexico City
[30] MSMD, Mexican Strong Motion Database, Mexican Society of Earthquake lacustrine soils, in: Tan et al. (Ed.), Characterisation and Engineering Properties of
Engineering, 1999. Natural Soils, Swets & Zeitlinger, 2003.
[31] K. Goda, C.A. Taylor, Effects of aftershocks on peak ductility demand due to strong [45] E. Miranda, J. Ruiz-García, Influence of stiffness degradation on strength demands
ground motion records from shallow crustal earthquakes, Earthq. Eng. Struct. Dyn. of structures built on soft soil sites, Eng. Struct. 24 (10) (2002) 1271–1281.

51

You might also like