Journal of Building Engineering 68 (2023) 106110

Contents lists available at ScienceDirect

Journal of Building Engineering

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

Seismic resilience assessment of buildings considering the effects

of mainshock and multiple aftershocks
Jie Hu a, b, Weiping Wen a, b, *, Changhai Zhai a, b, Shunshun Pei a, b, Duofa Ji a, b
a
Key Lab of Structures Dynamic Behavior and Control of the Ministry of Education, Harbin Institute of Technology, Harbin, 150090, China
b
Key Lab of Smart Prevention and Mitigation of Civil Engineering Disaster of the Ministry of Industry and Information Technology, Harbin Institute of
Technology, Harbin, 150090, China

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

Keywords: Seismic resilience assessment is a hot topic for highlighting the functional recovery of structures
Aftershock in post-earthquake environments. The current resilience assessment method generally ignores the
Economic loss effects of aftershocks, even though more than 85% of mainshocks are followed by multiple af­
Infill wall tershocks that further increase functionality loss and recovery time. This study proposes a method
Recovery time to quantify the resilience of structures under mainshocks and multiple aftershocks. The concept of
Seismic resilience
aftershock affecting time window is proposed to assess the effects of aftershocks on the recovery
time of structures. Additionally, the effects of infill walls on the state-dependent fragility results
and resilience of the structure are considered in this method. The proposed method is applied to
two reinforced concrete (RC) frame structures, and the effects of multiple aftershocks and infill
walls on the resilience are quantitatively studied. The results indicate that multiple aftershocks
increase the economic loss by 20%–30% on average, and the aftershock affecting time can ac­
count for more than 20% of the total recovery time. The exclusion of infill walls in the finite
element model of the structure induces a significant bias in the resilience assessment results at a
minimum level of 25%.

1. Introduction
Traditional earthquake engineering aims to ensure personal safety and focuses on reducing the risk of structural collapse caused by
earthquakes. The performance-based seismic design was developed on this basis to further focus on economic losses and societal
impacts due to earthquakes. The earthquake damage from the 2011 Great East Japan Earthquake and 2011 New Zealand Earthquake
demonstrated that the damage and collapse of structures caused by earthquakes were effectively controlled. However, the structures
generally suffered from excessive residual displacement or were difficult to repair, which caused serious economic losses and greatly
limited the post-earthquake functionality of structures and the progress of functional recovery. Therefore, it is important to reduce the
post-earthquake functional loss of structures and improve the functional recovery ability of structures [1]. That is, it is necessary to
improve the seismic resilience of structures to solve the problems of complicated reconstruction and long recovery time after
earthquakes.
More than 85% of mainshocks are followed by numerous aftershocks. Available seismic damage investigations and studies have
shown that the additional damage caused by aftershocks is not negligible [2,3]. Aftershocks further aggravate structural damage [4]
and reduce structural collapse resistance capacity [5,6]. Moreover, aftershocks increase the duration of the interruption of building

* Corresponding author. School of Civil Engineering, Harbin Institute of Technology, Harbin, 150090, China.
E-mail address: [email protected] (W. Wen).

https://doi.org/10.1016/j.jobe.2023.106110
Received 5 November 2022; Received in revised form 19 January 2023; Accepted 12 February 2023
Available online 14 February 2023
2352-7102/© 2023 Elsevier Ltd. All rights reserved.
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

functions, economic losses [7–9], and number of casualties [10], resulting in cumulative damage effects. In addition, aftershocks can
affect the post-earthquake repair work. Therefore, it is necessary to evaluate the seismic resilience of structures by considering the
effects of mainshocks and multiple aftershocks.
The seismic resilience of systems at different scales (buildings, communities, cities, etc.) has been intensively explored since the
concept of seismic resilience was first introduced by Bruneau et al. [11]. The seismic resilience of different types of buildings that
perform various functions and are essential components for maintaining community and city functions has also received extensive
attention. However, no uniform index currently exists for quantifying the seismic resilience of a single building. Pre- and
post-earthquake functionalities are generally used to describe resilience for a single building with well-defined functions, such as a
hospital [11–14] or substation [15]. Economic losses, repair time, and casualties are used to comprehensively evaluate the seismic
resilience of a single building without a specific function [16,17]. This method has also been used to evaluate single buildings with
well-defined functions [18,19]. The standard in China evaluates the seismic resilience of buildings using three indicators: restoration
cost of buildings, repair time of buildings, and casualties [20]. In addition to these three indicators, other indicators are available to
evaluate the seismic resilience of buildings. For example, Anwar et al. [21] proposed a probabilistic framework for assessing seismic
risk, sustainability, and resilience, then calculated repair losses, downtime, and equivalent carbon emissions for a reinforced concrete
(RC) frame structure to evaluate its seismic resilience.
Nonstructural components are important to ensure the function of buildings and have received attention in the assessment of
structural seismic resilience. Most buildings are equipped with infill walls that facilitate space division, sound insulation, and heat
preservation. Buildings with specific purposes, such as hospitals, typically have a large amount of specialized equipment to maintain
their functions. There is an interaction between the infill walls and structural components during an earthquake. In addition, the
damage to nonstructural components leads to the partial loss of the function of the building after an earthquake, thus increasing the
direct and indirect economic losses of buildings and recovery time. The seismic damage to RC frames with and without infill walls
under the action of mainshock-aftershock (MSAS) sequences has been compared [22,23]. The results showed that the two RC frame
structures behaved significantly differently, and that disregarding the effect of infill walls leads to different predictions of the seismic
performance of the structures.
Some studies have focused on the seismic resilience of buildings, but the following two issues have not been considered in previous
studies and require further research: (1) The effects of multiple aftershocks on the functionality loss and recovery time, including a
method that can quantify the resilience of structures under mainshock and multiple aftershocks, and (2) the effects of the inclusion of
infill walls in the finite element model (FEM) of structures on the resilience. Therefore, the seismic resilience of RC frame structures
under the action of MSAS sequences was investigated in this study by considering the effects of infill walls and multiple aftershocks.
The remainder of this paper is organized as follows. The proposed method for the seismic resilience assessment of structures under
mainshock and multiple aftershocks is presented in Section 2. The results of the state-dependent seismic fragility analysis and prob­
ability of damage states after multiple aftershocks obtained using two RC frame structures as case studies are presented in Section 3.
The resilience assessment results are presented and the effects of multiple aftershocks and infill walls are discussed in Section 4.
Finally, the conclusions are summarized in Section 5.

2. Methodology
The seismic resilience assessment process for RC frame structures considering the effects of the mainshock and multiple aftershocks
consists of four steps and is illustrated in Fig. 1. First, the hazard of multiple aftershocks is obtained using aftershock probabilistic
seismic hazard analysis (APSHA). Second, in the state-dependent seismic fragility analysis, the probability of structures or components
with different initial states caused by the mainshock reaching other damage states owing to aftershocks is obtained with a double
incremental dynamic analysis. Third, APSHA and state-dependent fragility are combined to determine the seismic damage

Fig. 1. Seismic resilience assessment process.

2
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

accumulation of multiple aftershocks for a given magnitude and location of the mainshock [24]. The change in the damage state
probability of structures or components owing to multiple aftershocks can be computed for the time interval of concern using the
established Markov model. Fourth, the economic losses of structures under the action of MSAS sequences and the recovery time
considering the aftershock affecting time can be obtained to quantify the resilience loss of structures.
The following sections will specify the details of each step. The notations used are listed in Table 1.

2.1. Aftershock probabilistic seismic hazard analysis

APSHA [25] is used to estimate the possibility of strong ground motion at the site due to aftershocks, which is characterized by
considering the variation in the occurrence rate of aftershocks.
The occurrence rate of aftershocks is directly related to the magnitude of the mainshock. An aftershock reaches its maximum
immediately after the mainshock and then gradually decreases over time. The mean daily rate of aftershocks at time t after the
mainshock of magnitude MM = mM , νA|MM , can be computed based on the modified Omori’s law [26] and is expressed as:

10a+b(mM − m1 ) − 10a
νA|MM (t; mM ) = (1)
(t + c)p

where m1 is the minimum aftershock magnitude considered, which is typically set to 5.0; mM is the maximum aftershock magnitude
considered, which is normally considered to be the mainshock magnitude because the aftershock magnitude generally does not exceed
the mainshock magnitude; a, b, c, and p are four constants, which are obtained by historical aftershock sequences estimated separately
using the maximum likelihood method, and the estimation results of Reasenberg and Jones [26] are used in this study, i.e., a, b, c, and p

Table 1
Notations used in this study.

Notation Description

MM The magnitude of mainshock

mA The magnitude of aftershock
mM The maximum aftershock magnitude considered
m1 The minimum aftershock magnitude considered
a, b, c, p The constants in equation (1)
rA The shortest distance from the site to the aftershock rupture surface
IMA The intensity measure of the aftershocks
λim,A (t) The mean daily rate of aftershocks with intensity exceeding im on day t after the mainshock
EDP The engineering demand parameter of a structure or component
edpi The threshold value of the ith damage state of the structure or components
IM The ground motion intensity measure
n The number of damage states of the structure or components
DSi The ith damage state
Pi,j
A|MM The probability of transitioning the damage state from DSi to DSj (j > i) by the aftershock
(tk− 1 , tk ) The kth unit time interval
Pi,j (k) The probability of damage state transition from DSi to DSj in (tk− 1 , tk )
νA (k) The occurrence rate of aftershocks in (tk− 1 , tk )
[I] Unit matrix
R The resilience loss factor
Q(t) The functionality of buildings
t0 The time at which an earthquake occurred
TR The recovery time
L The direct economic loss of buildings
LS, LD, and LC The loss of the structural damage, building decoration, and indoor property, respectively
α The correction coefficient of direct economic loss
Pi The probability of damage state i
μi The loss ratio corresponding to damage state i
γ1, γ2, and γ3 The correction coefficient of decoration loss, considering the differences in regional economic levels, usage of the buildings, and decoration
grade, respectively
t1 The beginning time of the repair
t2 The time of functional recovery to the original level
Tas The aftershock affecting time
TRE The repair time
Trepair The repair time when reconstruction is not considered
Treplacement The reconstruction time
L(I, t) The normalized loss function that represents the loss at time t
H(t) Heaviside step function
TREi The repair time of damage state i
Crepair The repair cost of a building
Creplacement The replacement cost of a building
Vp The peak shear force of the skeleton curve of masonry infill walls
d40, d85, dp, and The displacements corresponding to 40%Vp, 85%Vp, Vp, and 10%Vp, respectively
d10
dc The displacement corresponding to the shear value of 0

3
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

are − 1.67, 0.91, 1.08, and 0.05, respectively.

When a post-mainshock time window [t, t +T] is provided, the average number of aftershocks with magnitudes between m1 and mM
can be obtained by integrating equation (1), after the occurrence of a mainshock of magnitude mM .
In APSHA, the occurrence of aftershocks within the range of [m1 , mM ] is considered through the probability density function
fMA (mA ; mM ) of the aftershock magnitude mA. Additionally, the influence of the aftershock occurrence area is considered using the
conditional probability density function fRA |MA (rA |mA ) of the shortest distance from the site to the aftershock rupture surface rA.
Subsequently, the median value E[ln(IMA )] and standard deviation σln(IMA ) of ln(IMA ) can be predicted by the ground motion prediction
equation, where IMA is the intensity measure of the aftershocks.
The exceedance probability P(IMA ＞im|mA , rA ) of the aftershock intensity for a given magnitude of aftershocks and site is calcu­
lated. Combined with the occurrence of earthquake magnitude, the exceedance probability P(IMA ＞im) of aftershock intensity for a
given mainshock magnitude and fault distance can be calculated as follows:

βe− β(mA − m1 )
fMA (mA ; mM ) = , m1 ≤ mA ≤ mM , β = b ln 10 (2)
1 − e− β(mM − m1 )
( ⃒ )
ln(im) − E[ln(IMA )]⃒⃒
P(IMA ＞im|mA , rA ) = 1 − Φ ⃒mA , rA (3)
σln(IMA )
∫ ∫ mM
P(IMA ＞im) = P(IMA ＞im|mA , rA )fRA |MA (rA |mA )fMA (mA ; mM )dmA drA (4)
rA m1

The mean daily rate of aftershocks with intensity exceeding im on day t after the mainshock, namely the aftershock hazard, can be
calculated with equation (5).
λim,A (t) = νA|MM (t; mM ) · P(IMA ＞im) (5)

2.2. State-dependent seismic fragility analysis method

Seismic fragility represents the probability of a structure or component exceeding different degrees of damage under the action of
ground motions with different intensities, as shown in equation (6). Incremental dynamic analysis (IDA) [27] is a common method for
obtaining the seismic fragility of structures or components.
Fi (x) = P[EDP > edpi |IM = x], i = 1, 2, …, n (6)

where EDP is the engineering demand parameter of a structure or component, edpi is the threshold value of the ith damage state of the
structure or components, IM is the ground motion intensity measure, and n is the number of damage states of the structure or
components.
In seismic fragility analysis, it is generally assumed that the responses of structures or components under seismic action obey a log-
normal distribution. The seismic fragility function can be expressed as:
( )
ln(edpi ) − ln μEDP|IM
P[EDP > edpi |IM = x] = 1 − Φ (7)
βEDP|IM

where Φ() is standard normal cumulative distribution function, ln μEDP|IM is the logarithmic mean value of EDP, and βEDP|IM is the
logarithmic standard deviation of EDP.
The state-dependent seismic fragility analysis is based on the damage states of structures or components under the mainshock. The
IDA of damaged structures under the aftershocks is further conducted to obtain the probability of damaged structures or components
sustaining more serious damage states owing to aftershocks. This process consists of mainshock and aftershock IDAs and is called
double incremental dynamic analysis (Double IDA) [23]. The detailed procedures are as follows.
(1) Mainshock IDA: the mainshock ground motion is gradually scaled up and the ground motion intensity corresponding to the
damage state DSi of structures or components is denoted as IMDSi, i = 0, 1, 2, …, n.
(2) Aftershock IDA: the aftershock ground motion is gradually scaled up for the mainshock ground motion with a given intensity
IMDSi to obtain the response of the structures.
The spectral acceleration value corresponding to the fundamental period of the structure Sa(T1) is used as the IM in the ground
motion scaling process. The seismic response obtained by the double IDA is used to compute state-dependent seismic fragility results.

2.3. Probability of damage states after multiple aftershocks

Damage to structures or components may have occurred after a mainshock, and each subsequent aftershock may cause severe
damage to structures or components. Iervolino et al. [24] provided a probability matrix of the damage state transfer for experiencing
an aftershock triggered by a mainshock of a given magnitude and the shortest distance from the site to the fault plane, [PA|MM ], as in
equation (8):

4
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

⎡ ⎤
∑
n
A|M A|M A|M
⎢1 − P1,j M P1,2 M … … P1,n M ⎥
⎢ ⎥
⎢ j=2 ⎥
⎢ ⎥
⎢ ∑
n ⎥
[ ] ⎢ 0 1−
A|MM
P2,j … …
A|MM ⎥
P2,n ⎥
PA|MM = ⎢
⎢ j=3 ⎥ (8)
⎢ … … … … … ⎥
⎢ ⎥
⎢ A|MM A|M ⎥
⎢ 0 … 0 1 − Pn− Pn− 1,nM ⎥
⎣ 1,n ⎦
0 … … 0 1

where Pi,j denotes the probability of the aftershock transitioning the damage state from DSi to DSj (j > i).
A|MM

It is known that the intensities of any aftershock triggered by the same mainshock are independent and identically distributed
random variables [4]. Subsequently, Pi,j M can be calculated by equations (9)–(11):
A|M

∫
[ ⃒ ]
(9)
A|M
Pi,j M = P DSj ⃒DSi , IMA = im · fIMA |MM (im|mM ) · d(im)

[ ⃒ ] [ ⃒ ] [ ⃒ ]
P DSj ⃒DSi , IMA = im = PDS ⃒ DS ⃒
i,j EDP > edpj IMА = im − Pi,j+1 EDP > edpj+1 IMA = im (10)
∫ ∫ mM
fIMA |MM (im|mM ) = fIMA |MA ,RA (im|mA , rA ) · fRA |MA (rA |mA ) · fMA (mA ; mM ) · dmA · drA (11)
rA m1

A|M
It should be noted that the damage state cannot be transformed into a better case (i.e., Pi,j M (i＜j) = 0) because repair activities are
not allowed before the aftershock that is not negligible. The diagonal elements of the matrix [PA|MM ] represent the probability that an
∑n
aftershock does not cause the damage state transition, which is represented by 1 − j=i+1 Pi,j .
The probability of the damage state transitioning from DSi to DSj in the kth unit time interval (tk− 1 , tk ) when the occurrence
probability of more than one aftershock in unit time interval is negligible is simply expressed by equation (12) [24]:

Pi,j (k) = νA (k)PA|M

i,j
M
(12)

where νA (k) is the occurrence rate of aftershocks in (tk− 1 , tk ).

The matrix of transition probability between any two damage states within a unit time interval can be obtained based on equation
(12) and is expressed as:
[ ]
[P(k)] = νA (k) · PA|MM + (1 − νA (k)) · [I] (13)

where [I] is a unit matrix.

The probability of the structures or components being in each damage state after k unit time intervals can be calculated from
equation (14):
∏
k
{ P1 (k) P2 (k) … Pn (k) } = { P1 (0) P2 (0) … Pn (0) } [P(j)] (14)
j=1

where { P1 (0) P2 (0) … Pn (0) } is the probability vector of the initial damage state. For MSAS sequences, the probability vector
represents the probability of each damage state for the structures or components caused by the mainshock.

2.4. Seismic resilience assessment

The resilience loss factor R proposed by Wen et al. [28] is used in this study to quantify the seismic resilience of the structures,
which is expressed as:
∫
1 t0 +TR
R= [1.0 − Q(t)]dt (15)
TR t0

where Q(t) is the functionality of buildings, Q(t) ∈ [0, 1.0], with 0 indicating a complete loss of functionality and 1.0 indicating
maintenance of the original functionality, t0 is the time at which an earthquake occurred, and TR is the recovery time.

2.4.1. Functional loss assessment

The functionality of a building is computed based on direct economic loss, as proposed by Cimellaro et al. [29]. The direct economic
loss of a building L typically includes the losses from the structural damage LS, building decoration LD, and indoor property LC, which is
expressed as:
L = α(LS + LD + LC ) (16)

5
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

LS, LD, and LC can be obtained from the cost, the loss ratio of each damage state and its corresponding probability [30,31], and are
respectively expressed as:
∑
ns
LS = Ms · Psi · μsi (17)
i=1

∑
nD
LD = γ1 · γ 2 · γ3 Md · Pdi · μdi (18)
i=1

∑
nC
LC = Mc · Pci · μci (19)
i=1

where Ms , Md , and Mc are the replacement costs of the structure, decoration and indoor property, respectively, α is the correction
coefficient of direct economic loss, which is set to 1.0–1.3 [32], n is the number of damage states, Pi is the probability of damage state i,
and μi is the loss ratio corresponding to damage state i. The loss ratio is the ratio of the cost of repairing a structure or component in a
damage state to the replacement cost of the structure or component, determined by referencing the available literature [30]. γ 1 , γ2 , and
γ3 are the correction coefficients of the decoration loss, taking into account the differences in regional economic levels (1.30 for
developed regions, 1.15 for more developed regions, and 1.0 for general regions) [32], usage of the buildings (1.0 for office, resi­
dential, medical, and educational buildings, 1.1 for commercial buildings, and 0.9 for other buildings) [30], and decoration grade
(1.1–1.2 for high-grade, 1.0 for mid-grade, and 0.8–0.9 for ordinary) [30], respectively.

2.4.2. Recovery model

The recovery process is complex, involving the influence of the temporal dimension, spatial dimension, and interrelationships
among the different departments involved in the recovery process. The functional recovery model should be determined according to
the recovery process. Simplified recovery models are typically used to describe the recovery process of buildings. There are three types
of commonly used functional recovery functions: linear [19], trigonometric [33], and exponential [19]. The linear recovery function is
adopted in this study and used when there is no information about available resources and organization, as shown in equation (20).
The linear recovery function fits the case scenario of this study.
t − t1
frec (t, t1 , TRE ) = 1 − (20)
TRE

where TRE is the repair time and t1 is the time to start the repair.

2.4.3. Consideration of the impact of aftershocks

The effect of aftershocks on the resilience of a structure includes two considerations: (1) aftershocks aggravate the functionality
loss, thus making the recovery longer and more difficult, and (2) the occurrence of aftershocks delays the beginning of recovery work,
thus making the recovery longer. For the second point, it is necessary to determine when the effects of aftershocks can be ignored, and
the recovery work can begin. The time of i days from the mainshock is set as the aftershock affecting time Tas. The recovery work can
begin on (i+1)th day when the direct economic loss increases on ith day because the aftershocks is negligible (using 0.05% of the
replacement cost of the building as the threshold). It should be noted that the functionality of the structure may decline further owing
to aftershocks in the aftershock affecting time window.
The occurrence time of the mainshock is denoted as t0, beginning time of the recovery is denoted as t1, time of functional recovery to
the original level is denoted as t2, and the aftershock affecting time window is denoted as [t0 ,t1 ], where t1 = t0 + Tas . The following two

Fig. 2. Seismic resilience of structures under the action of MSAS sequence.

6
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

scenarios are considered in the recovery process:

(1) The recovery time is determined by the damage degree of the structures and recovery process if the functional loss of the
structures has not yet reached the point where reconstruction needs to be considered in the aftershock affecting time window,
which is denoted as Trepair and shown in Fig. 2 (a).
(2) Reconstruction is initiated at the end of the aftershock affecting time window and the recovery time is controlled by the
reconstruction time if the functional loss of the structures has reached or exceeded the minimum acceptable functional
threshold in the aftershock affecting time window, which is denoted as Trepacement and shown in Fig. 2 (b). Owners typically
prefer to rebuild rather than repair a building when the repair cost of a building is approximately 40% of the replacement cost
[34]. This is the minimum acceptable functional threshold.
It should be noted that R is set to 1.0 when the reconstruction of a structure is considered because resilience will no longer be
meaningful for the single structure once the reconstruction is initiated [28].
Q(t) can be calculated by equation (21), which is proposed by Cimellaro et al. [35]:
{
1 − L(I, t), t1 > t ≥ t0
Q(t) = (21)
1 − L(I, t1 ) ·[H(t − t1 ) − H(t − t2 )] · frec (t, t1 , TRE ), t ≥ t1

where frec (t, t1 , TRE ) is the functional recovery function, TRE is the repair time, corresponding to Trepair or Treplacement , L(I, t) is the
normalized loss function that represents the loss at time t, which is equal to L (equation (16)) divided by the replacement cost, I is the
earthquake intensity, t2 is the time to end the repair (t2 = t0 + Tas + TRE ), and H(t) is the Heaviside step function.
The recovery time is defined as the sum of the aftershock affecting and the repair times, which is expressed as:
TR = Tas + TRE (22)
The repair time TRE is calculated as the sum of the product of the probability Pi of damage state i and the corresponding repair time
TREi , which is expressed as:
⎧ ∑ n
⎨T TREi · Pi ，Crepair < 40% · Creplacement
repair =
TRE = i=1
(23)
⎩
Treplacement ，Crepair ≥ 40% · Creplacement

where Crepair and Creplacement are the repair and replacement costs of a building, respectively.

3. Structural information and fragility analysis results

3.1. Structural information and model
3.1.1. Structural information
Two reinforced concrete (RC) frame structures of four and eight stories were adopted as case studies. The structures were designed
according to the Chinese Code for Design of Concrete Structures (GB 50010-2010) [36] and Chinese Code for Seismic Design of
Buildings (GB 50011-2010) [37]. The seismic precautionary intensity of the four- and eight-story RC frame structures was VIII degree.
The corresponding design basic acceleration of ground motion was 0.20 g. The load information for the two structures is listed in
Table 2, and the dimensions and material information of the beams and columns are listed in Table 3. Specific information for the RC
frame structures is illustrated in Fig. 3.
The infilled frame (IF) was obtained by arranging infill walls at the side spans of the RC frame, as shown in Fig. 4. An RC frame
without infill walls was denoted as a bare frame (BF). The infill walls were made of concrete single-row hollow blocks with strength
grades of block MU10 and mortar Mb5. Lastly, the size of the block was 390 mm × 190 mm × 190 mm.

3.1.2. Cost information

Two types of buildings were considered in this study: office and medical. Through the investigation of typical hospitals, the
departmental arrangement and key medical equipment included in four- and eight-story hospitals are shown in Table 4 and Table 5,
respectively.
The unit cost was determined by consulting the China Construction Engineering Cost Information Network [38], China Engineering
Budget Network [39], and engineering personnel, considering the local price level and labor costs, as shown in Table 6.

Table 2
Load information of RC frames.
2 2
Structure Position Dead load/kN⋅m− Live load/kN⋅m− Slab thickness/mm

The four-story RC frame Room 1.5 2.0 100

Corridor 1.5 3.5 100
Roof 5.5 0.4 120
The eight-story RC frame Room 2 2.5 120
Corridor 2 3.5 120
Roof 3 2 120

7
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Table 3
Beam-column dimensions and material information of RC frames.

Structure Floor Sectional dimension Materials Stirrup ratio/% Concrete cover depth/mm

Columns Beams Concrete Rebar Columns Beams

The four-story RC frame 1–4 0.5 × 0.5 0.4 × 0.3 C30 HRB400 0.5 0.3 30
The eight-story RC frame 1 0.8 × 0.8 Side span: C50 HRB400 0.8 0.3 20
2–3 0.8 × 0.8 0.75 × 0.3 C45
4 0.8 × 0.8 Mid-span: C30
5–8 0.65 × 0.65 0.5 × 0.5 C30

Fig. 3. RC frame structures (mm).

Fig. 4. Four-story RC infilled frame.

The decoration ratio is the ratio of the decoration cost to the cost of the structure. The property ratio is the ratio of the indoor
property cost to the cost of the structure. The decoration and property ratios were used to calculate the decoration cost and indoor
property values, respectively. The decoration and property ratios of the RC buildings were determined from Refs. [30–32,40], as shown
in Table 7.
The replacement costs of each part of the four- and eight-story RC frames were calculated based on the unit costs, decoration ratio,
property ratio, and medical equipment prices, as shown in Table 8. The total replacement cost is the sum of the replacement costs of

8
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Table 4
Departmental arrangement of medical buildings.

Departments Key rooms The four-story RC frame The eight-story RC frame

Number Floor Number Floor

Outpatient department Consulting room 3 1 3 1

Emergency department Debridement room 1 1 1 1
Emergency treatment room 1 1 1 1
Medical imaging department NMR room 1 1 2 1
CT room 1 1 2 1
B-ultrasound room 1 1 2 1
Laboratory department Medical laboratory 2 2 3 2
Pharmacy department Pharmacy 1 2 2 2
Blood transfusion department Blood storage room 1 2 2 3
Blood-issuing room 1 2 2 3
Blood-matching room 1 2 2 3
Central supplying department Central sterile supply department 3 2 4 4
Operation department Operation room 2 3 4 4
Intensive care unit ICU ward 2 4 10 8
General ward Ward 17 3 40 5–7
Others Doctor’s office 5 2–4 10 2–8

Table 5
Key medical equipment included in medical buildings.

Equipment Unit price Number of equipment

(ten thousand RMB) The four-story RC frame The eight-story RC frame

NMR 300 1 2
CT 200 1 2
B Ultrasound machine 8 1 2
X-ray equipment 7 1 2
Surgical bed 1 2 4
Anesthesia machine 3 2 4
Ventilator 2 4 8
Desktop computer 0.4 5 15
Office table 0.05 10 30
Electrocardiogram monitor 0.6 4 10
Analyzers 0.3 2 5
Filing cabinet 0.04 10 25
Disinfection storage cabinet 5 6 15
Patient bed 0.06 36 80

Table 6
Unit costs for structural and nonstructural components.

Name RC frame structures Infilled walls Suspended ceilings

Unit costs 1500 RMB/m2 170 RMB/m3 40 RMB/m2

Table 7
Decoration and property ratios of RC buildings.

Building types Decoration ratio Property ratio

Medical building 0.24 –

Office building 0.29 0.5

each part of the building, as listed in Table 8.

The repair time TREi of the building was determined for the structure in different damage states based on the HAZUS geographic
information system-based natural hazard analysis tool, developed and freely distributed by the Federal Emergency Management
Agency (FEMA) [41], as shown in Table 9. The reconstruction time Treplacement was 219 and 329 days for four- and eight-story buildings,
respectively.

3.1.3. Structural finite element model

Two-dimensional finite element models of RC frame structures were established using OpenSees software [42]. Concrete was
simulated using the Concrete02 material [43] and divided into two parts (unconfined and confined concrete) for modeling considering

9
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Table 8
Replacement costs.

Name The four-story RC frame The eight-story RC frame

Project amount Cost (ten thousand RMB) Project amount Cost (ten thousand RMB)

RC frame structures 1223 m2 183.45 6998 m2 1049.73

Infilled walls 349 m3 5.93 1529 m3 26.00
Suspended ceilings 893 m2 3.57 5773 m2 23.09
Indoor property of offical building – 91.73 – 524.86
Decoration of offical building – 55.04 – 304.42
Indoor property of hospital – 569.06 – 1157.80
Decoration of hospital – 45.86 – 251.93
Offical building – 339.72 – 1928.10
Hospital – 807.87 – 2508.55

Table 9
Repair time TREi [41].

Damage state DS1 DS2 DS3 DS4

Four-story buildings (days) 7 30 112 219

Eight-story buildings (days) 12 48 170 329

the confinement effect of stirrups on concrete in RC frame structures [44]. The core and cover concrete were modeled using confined
and unconfined concrete, respectively. The effect of stirrups on the strength and deformation capacity of concrete was considered using
the modified Kent-Park model [44]. Steel02 material [45] was used to model the steel reinforcement, where the strain-hardening ratio
b was set to 0.01. The beam-column sections were simulated using the fiber section model [46] and were specifically divided into three
parts: unconfined cover concrete fibers, confined core concrete fibers, and steel fibers. The beam-column components of the RC frame
structures were simulated using a nonlinear beam-column element [47].
The infill wall was simulated in OpenSees using a single equivalent compression bar and the Pinching4 uniaxial material model
[48]. The skeleton curve of the infill walls was defined with reference to the work of Liberatore [49], and the tensile skeleton curve of
uniaxial materials was defined using near-zero values because of the generally low level of tensile resistance of masonry materials. The
theoretical skeleton curve is shown in Fig. 5 (a). The skeleton curve is defined by four characteristic points with different shear values:
40%Vp, 85%Vp, Vp, and 10%Vp, where Vp is the peak shear force. The displacements corresponding to the four shear values are d40, d85,
dp, and d10, respectively. dc is the displacement corresponding to the shear value of 0. Liberatore et al. [49] estimated the displacements
of characteristic points using regression analysis and proposed a simplified prediction formula for the peak shear force to calculate the
skeleton curves of the infill walls of four- and eight-story RC frame structures, as shown in Fig. 5 (b). The parameters of Pinching4 were
determined based on the obtained skeleton curves and with reference to the studies of Blasi et al. [50] and Kumar et al. [51]. These are
expressed as:
[ rDispP fForceP uForceP ] = [ rDispN fForceN uForceN ] = [ 0.5 0.1 − 0.05 ]

Fig. 5. Skeleton curve of masonry infill walls.

10
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

⎡ ⎤ ⎡ ⎤
gK1 gK2 gK3 gK4 gKLim 0.5 0.2 0.3 0.2 0.9
⎢ gD1 gD2 gD3 gD4 gDLim ⎥ ⎢ 0.1 0.1 2.0 2.0 0.5 ⎥
⎢
⎣ gF1 gF2
⎥=⎢ ⎥ (24)
gF3 gF4 gFLim ⎦ ⎣ 0.1 0.1 1.0 1.0 0.9 ⎦
gE 10

where rDispP and rDispN denote the deformation ratios at which reloading occurs to the maximum and minimum historic deformation
demands, respectively. fForceP and fForceN denote the ratios of the force at which reloading begins, corresponding to the maximum
and minimum historic deformation demands, respectively. uForceP and uForceN denote the strength ratios that developed from
unloading the negative load to the maximum and minimum strength developed under monotonic loading, respectively. gK1, gK2, gK3,
gK4, and gKLim are the related parameters of the unloading stiffness degradation. gD1, gD2, gD3, gD4, and gDLim are the related
parameters of reloading stiffness degradation. gF1, gF2, gF3, gF4, and gFLim are the related parameters of strength degradation.
The same 2D models were established using SAP2000 software to verify the correctness of the OpenSees models of the four RC
frame structures. The fundamental periods of structures that were obtained are listed in Table 10. The calculation errors of the two
software packages were less than 10%, which verified the correctness of the OpenSees models.

3.2. Aftershock probabilistic seismic hazard analysis results

The structures were located near a fault in the Sichuan-Yunnan region of China. The shortest distance from the site to the fault (fault
distance) was 20 km. It was assumed that the mainshocks and aftershocks occurred on the fault. The length of the earthquake fault was
calculated from the empirical relationship between the earthquake magnitude and rupture length [52]. The median value and standard
deviation of the IM of the aftershocks were predicted using the SY model [53], which is a ground motion attenuation model based on
records from the Sichuan-Yunnan region of China. The aftershock hazards of the structure at different mainshock magnitudes (6.0, 6.5,
7.0, and 7.5) on the different days after the mainshock were obtained using equation (5).
The aftershock hazard curves of the BF4 model (four-story bare-frame structure) on the 1st, 5th, 10th, 15th, 20th, 25th, and 30th
days after a mainshock of magnitude 7.0 are shown in Fig. 6. The aftershock hazard decreased sharply with an increase in the number
of days after the mainshock and was most obvious in the first 10 days after the mainshock.
The selected IM, Sa(T1), was closely related to the fundamental period of vibration of structure T1. As such, the aftershock hazards of
the four structures with different T1 on the 1st day after the mainshock with a magnitude 7.0 were also compared, as shown in Fig. 7.
The results show that there was a significant difference in the aftershock hazard of structures with large differences in the fundamental
periods. Additionally, the smaller the fundamental period, the greater the aftershock hazard.

3.3. State-dependent seismic fragility analysis

3.3.1. Damage states of structural and nonstructural components
The inter-story drift ratio (IDR) can effectively measure the seismic damage degree of RC frame structures and was selected as the
EDP. The IDR threshold values corresponding to each damage state were determined using the HAZUS analysis tool [41] and are listed
in Table 11.
Infill walls are the most common displacement-sensitive nonstructural components. The most used form in Chinese building
structures is a non-bearing wall [54]. The limit values of IDR corresponding to the damage states of the infill wall based on the existing
studies are listed in Table 12 [54,55].
Acceleration-sensitive nonstructural components, including suspended ceilings and furniture, are susceptible to sliding and tipping
during earthquakes, with damage caused primarily by the floor acceleration response of the structures. Suspended ceilings, key
medical equipment, and indoor properties were considered in this study. The division of damage states and limit values of suspended
ceilings were based on the Standard for Seismic Resilience Assessment of Buildings (GB/T 38591-2020) [20]. The key medical
equipment in hospitals consists of X-ray machines, B Ultrasound machines, nuclear magnetic resonance apparatus, medicine cabinets,
and patient beds, all of which are acceleration-sensitive [56]. The damage states and corresponding limit values of key medical
equipment were determined based on the existing literature [56]. Descriptions of the damage states and seismic fragility parameters
for some indoor properties were provided by FEMA P-58, including office furniture, fragile items, and home entertainment equipment,
all of which have only one damage state [57,58]. The limit values of the damage state of the aforementioned property were averaged as
the limit value for indoor properties based on FEMA P-58. The damage states and corresponding limit values of the
acceleration-sensitive nonstructural components are listed in Table 13.

3.3.2. Selected ground motions

A total of 50 MSAS ground motions were selected from the Pacific Earthquake Engineering Research Center Ground Motion
Database [59], European Strong-Motion Database [60], Center for Engineering Strong Motion Data [61], and China Earthquake
Networks Center [62], as listed in Table A1 in Appendix A. The mainshock and aftershock records in each MSAS sequence were

Table 10
Fundamental periods of structures (s).

Software BF4 model IF4 model BF8 model IF8 model

OpenSees 0.6415 0.1717 1.3082 0.4966

SAP2000 0.6415 0.1852 1.3474 0.5210

11
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Fig. 6. Aftershock hazards of the BF4 model for different days after the mainshock (MM = 7.0).

Fig. 7. Aftershock hazards of structures with different fundamental periods on the 1st day after the mainshock (MM = 7.0).

Table 11
Damage state threshold values of RC frames.

Structure Slight damage Moderate damage Severe damage Complete damage

(DS1) (DS2) (DS3) (DS4)

Four-story RC frame 0.0033 0.0058 0.0156 0.0400

Eight-story RC frame 0.0025 0.0043 0.0117 0.0300

Table 12
Damage state threshold values of infilled walls.

Damage state Description of damage state Limit value of

IDR

Intact The wall is in elastic state, a small amount of corner wall skin peeling off –
Slight damage The crack width of the wall is not more than 1.5 mm, the fine cracks increase, the cracks expand further inward and 0.0010
(DS1) diagonal cracks appear
Moderate damage The crack width of the wall is greater than 1.5 mm, and the diagonal cracks expand and gradually penetrates 0.0018
(DS2) continuously until the wall is severely cracked and crushed cavities appear at the corners
Severe damage The crack width of the wall is greater than 3.0 mm, and the wall is severely cracked and crushed, with holes in the wall 0.0050
(DS3) surface
Complete damage Multiple obvious penetrating cracks appear in the wall, and the wall collapses when the inter-story displacement angle 0.0060
(DS4) further increases

12
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Table 13
Damage state threshold values of acceleration-sensitive nonstructural components.

Components Damage state Description of damage state Limit value of PFA/g

Suspended ceilings DS1 5% of suspended ceilings are damaged 1.385

DS2 30% of suspended ceilings are damaged 1.765
DS3 Suspended ceilings are damaged completely 1.945
Key medical equipment DS1 Equipment is not available 2.160
Indoor properties DS1 Items are unusable due to collapse 0.868

recorded by the same station, and a time gap of 30 s was added between the mainshock and aftershock to ensure that the structures
damaged from the mainshock had sufficient time to stop vibrating before the aftershock.

3.3.3. Fragility analysis results

The IDA method was used to obtain the fragility parameters of the structural and nonstructural components under the mainshock,
as listed in Table 14. The average maximum IDR of all floors of the structure was used as the EDP of displacement-sensitive
nonstructural components, considering that the IDR and PFA of each floor of the structure were different and there were multiple
damage states of nonstructural components. Similarly, the average PFA of all floors was used as the EDP of acceleration-sensitive
nonstructural components to reasonably reduce the calculation volume.
A nonlinear time history analysis was carried out following the Double IDA method described in Section 2.2 to obtain the response
of the structural and nonstructural components under the MASA sequence, Subsequently, the response results were further fitted to
obtain the fragility parameters. The fragility parameters of the structural and nonstructural components of each model with different
initial damage states are listed in Table A2 in Appendix A.

3.4. Results of probability of damage states after multiple aftershocks

The probability matrix of the damage state transfer of the structure and nonstructural components owing to the aftershock can be
obtained based on the results of the state-dependent seismic fragility and probability density function of the magnitude of aftershocks.
The results for the BF4 model show that the structure was most likely to be maintained in its initial damage state (damage state of the
structure after the mainshock) after one aftershock, as shown in Table 15. The second highest probability was transferring to the more
severe adjacent damage state, and there was very little probability of directly transferring across the adjacent damage state to the most
severe damage state.
The variations in the probability of the BF4 model remaining in the initial damage state after one aftershock with different
mainshock magnitudes are shown in Fig. 8. The probability that the structure would remain in the initial damage state after an
aftershock decreased with an increase in the mainshock magnitude. Meanwhile, the probability of maintaining the original damage
state after one aftershock was lower if the initial damage was lighter, indicating that structures with light damage after the mainshock
were more likely to be affected by aftershocks.
According to the state-dependent fragility, the probabilities of each damage state of the BF4 model from the 0th day to the 10th day
with a mainshock of magnitude 7.0 are shown in Fig. 9, where the mainshock occurred on 0th day. After the mainshock, the probability
order of the BF4 model being in each damage state was DS0 > DS2 > DS1 > DS3 > DS4. The probability order changed after two days of
aftershocks to DS2 > DS0 > DS1 > DS3 > DS4, indicating that the structure underwent cumulative damage due to the aftershock
sequences in the first two days after the mainshock. In addition, the change in the transition probability of the damage states decreased
with an increase in the number of days after the mainshock. Taking the change in the probability of DS1 as an example, the increases in
the first three days after the mainshock were 9.96%, 1.27%, and 0.6%, respectively. The increase decreased to less than 0.5% from the
4th day and to approximately 0.1% on the 10th day, indicating that the effects of the aftershocks were concentrated in the first few
days after the mainshock.

4. Resilience assessment results

4.1. Effects of multiple aftershocks
The total economic losses are the sum of the direct and indirect economic losses. The ratios of indirect to direct economic losses for
structures with different seismic fortification criteria are provided in the existing literature combining earthquake disaster data from
China [63]. The ratios of indirect to direct economic losses of structures in different damage states considering the uses and types of
buildings are listed in Table 16.
The ratios of the total economic losses caused by the MSAS sequences to those caused only by the mainshock are presented in
Fig. 10. Based on the minimum acceptable functional threshold in Section 2.4.3, it was determined that the upper limits of direct
economic loss for the maintenance of the four-story office building and hospital were 1,358,880 and 3,231,480 RMB, respectively. The
corresponding values were 7,712,400 and 10,034,200 RMB for the eight-story office building and hospital, respectively. For the case
where the reconstruction threshold was not reached in the aftershock affecting time window, the average ratios of the total economic
losses caused by the MSAS sequences to that caused only by mainshock was 1.2–1.3. This ratio can be large when structures need to be
reconstructed owing to damage caused by aftershocks, which is the case for the IF4 models under the MSAS sequences with mainshock
magnitude of 7.5. These indicate that the total economic losses are significantly underestimated when only the effect of the mainshocks
is considered, and the degree of underestimation increases with an increase in the mainshock magnitude.

13
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Table 14
Fragility parameters of components under the mainshock.

Structural model Component DS1 DS2 DS3 DS4

Sa σln Sa Sa σln Sa Sa σln Sa Sa σln Sa

BF4 model Structure 0.18 0.25 0.28 0.33 0.64 0.34 1.56 0.39
Infilled walls 0.07 0.18 0.12 0.24 0.29 0.33 0.35 0.33
Suspended ceilings 1.06 0.55 1.41 0.58 1.67 0.56 – –
Key medical equipment 1.98 0.68 – – – – – –
Indoor properties 0.65 0.54 – – – – – –
IF4 model Structure 0.73 0.71 1.13 0.79 2.31 0.92 3.98 0.94
Infilled walls 0.39 0.66 0.53 0.65 1.10 0.70 1.35 0.75
Suspended ceilings 0.70 0.58 0.87 0.52 1.01 0.50 – –
Key medical equipment 1.22 0.50 – – – – – –
Indoor properties 0.48 0.56 – – – – – –
BF8 model Structure 0.03 0.77 0.12 0.36 0.26 0.46 0.83 0.45
Infilled walls 0.04 0.41 0.07 0.37 0.18 0.41 0.20 0.43
Suspended ceilings 0.55 0.83 0.67 0.85 0.74 0.87 – –
Key medical equipment 0.90 1.08 – – – – – –
Indoor properties 0.33 0.70 – – – – – –
IF8 model Structure 0.40 0.36 0.68 0.36 1.55 0.48 3.01 0.60
Infilled walls 0.27 0.28 0.38 0.33 0.94 0.40 1.08 0.41
Suspended ceilings 1.10 0.46 1.53 0.49 1.74 0.50 – –
Key medical equipment 2.14 0.63 – – – – – –
Indoor properties 0.65 0.44 – – – – – –

Table 15
Transition probability of structural damage states for BF4 model after one aftershock.

Mainshock magnitude Initial damage state DSi Damage state after one aftershock DSj

DS0 DS1 DS2 DS3 DS4

6.0 DS0 0.9921 0.0061 1.72E-03 7.82E-05 3.20E-07

DS1 0 0.9977 2.19E-03 9.34E-05 4.15E-07
DS2 0 0 0.9999 1.05E-04 4.52E-07
DS3 0 0 0 1.0000 9.85E-07
DS4 0 0 0 0 1
6.5 DS0 0.9746 0.0166 8.03E-03 8.35E-04 5.75E-06
DS1 0 0.9895 9.48E-03 9.63E-04 7.35E-06
DS2 0 0 0.9989 1.06E-03 7.97E-06
DS3 0 0 0 1.0000 1.63E-05
DS4 0 0 0 0 1
7.0 DS0 0.9607 0.0225 1.44E-02 2.32E-03 2.07E-05
DS1 0 0.9809 1.64E-02 2.63E-03 2.62E-05
DS2 0 0 0.9971 2.86E-03 2.84E-05
DS3 0 0 0 0.9999 5.66E-05
DS4 0 0 0 0 1
7.5 DS0 0.9545 0.0245 1.75E-02 3.40E-03 3.38E-05
DS1 0 0.9764 1.97E-02 3.83E-03 4.27E-05
DS2 0 0 0.9958 4.14E-03 4.63E-05
DS3 0 0 0 0.9999 9.11E-05
DS4 0 0 0 0 1

The ratio of indirect to direct economic losses caused by MSAS sequences is shown in Fig. 11. Direct economic losses for office
buildings account for a major portion of the total economic losses when the reconstruction threshold is not reached. Conversely, it is
mainly the indirect economic losses for hospitals that account for the majority of the total economic losses, and the ratio of indirect to
direct economic losses shows an overall increasing trend with an increase in the mainshock magnitude. The ratio is equal to the results
in Table 16 when the structure must be reconstructed under the impact of aftershocks. These results indicate that earthquake-induced
shutdowns are more likely to have a more severe impact on hospitals than on office buildings.
The recovery time is composed of the aftershock affecting time and the repair time. The aftershock affecting time was determined
using the method described in Section 2.4.3. The repair or reconstruction of the structures was decided according to the minimum
acceptable functional threshold after the aftershock impact was over and the repair time was determined, as shown in Fig. 12. The IF4
models reached the reconstruction threshold in the aftershock affecting time window under the action of MSAS sequences with a
mainshock magnitude of 7.5. The times to reach this threshold for office buildings and hospitals were 1.178 and 28.562 days,
respectively. Comparing the aftershock affecting time and total recovery time, it was found that the ratio of aftershock affecting time to
total recovery time for office buildings was 28.99% on average, with the maximum reaching 50%. The ratio for hospitals was 23.61%
on average, with the maximum reaching 50%. Overall, the aftershock affecting time exceeded 20% of the total recovery time, which
further indicates the need to consider aftershocks.

14
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Fig. 8. Probability of the BF4 model structure remaining in the initial damage state after one aftershock.

Fig. 9. Variation of the occurrence probability of damage states for the BF4 model under the action of MSAS sequences (MM = 7.0).

Table 16
Ratio of indirect to direct economic losses.

Damage state Intact DS1 DS2 DS3 DS4

Hospital 0 0 0.75 4.5 32

Offical building 0 0 0.5 2 6

The resilience loss factors of the four models under the MSAS sequences were calculated using the method described in Section 2.4,
as shown in Fig. 13. The resilience loss factor of the office building was greater than that of the hospital under each operating condition,
indicating that the models in this study achieved better seismic resilience when used as hospitals. This is mainly because the expected
value of the ratio of aftershock affecting time to total recovery time was higher for office buildings than for hospitals, resulting in
hospitals being able to start repairs earlier.

4.2. Effects of infills

The ratios of the total economic losses caused by MSAS sequences to those caused by mainshocks for the IF models were larger than
those for the BF models for office buildings and hospitals, indicating that the economic losses were greater for the RC frame structures
with infill walls under the MSAS sequences than for those without infill walls, as shown in Fig. 10.
The aftershock affecting time considers the additional effect of aftershocks on economic losses. A smaller aftershock affecting time
indicates a shorter time for the structure to reach its maximum direct economic loss. The BF models started repairs earlier than the IF
models under 97% of the operating conditions, indicating that the IF models were more likely to be affected by aftershocks for a longer

15
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Fig. 10. Ratios of the total economic losses caused by the MSAS sequences to that caused only by mainshock.

Fig. 11. Ratio of indirect to direct economic losses under the action of MSAS sequences.

Fig. 12. Recovery time under the action of MSAS sequences.

16
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Fig. 13. R of office buildings and hospitals under the action of MSAS sequences.

time compared to that of the BF models, as shown in Fig. 12. Additionally, there was no uniform relationship between the repair times
of the BF and IF models, except for the conditions in which the structures needed to be reconstructed owing to aftershock effects. If the
ratio of the IF to BF model repair times was NRT, the average NRT was 2.952 for the four-story office building, 2.952 for the four-story
hospital, 0.315 for the eight-story office building, and 0.312 for the eight-story hospital. The BF models of the four-story structures
sustained less damage than the IF models. The BF models of the eight-story structures sustained more severe damage than the IF
models. The differences in the results for the four-story and eight-story structures could be predicted from their damage state
probabilities.
The ratios of the resilience loss factor R of the IF models to that of the BF models are shown in Fig. 14. The ratios of R for the four-
story structure were greater than 1.0 for an office building or hospital. Conversely, the ratios of R for an eight-story structure were less
than 1.0. This phenomenon indicates that the effect of having infill walls or not on the seismic resilience of RC frame structures varies
with a different number of stories. The structures without infill walls achieved better seismic resilience than those with infill walls for
four-story RC frame structures. Conversely, the structures without infill walls demonstrated less seismic resilience than those with infill
walls for eight-story RC frame structures. This result is related to the damage to the structures caused by earthquakes. The above
phenomena indicate that the exclusion of infill walls in the FEM of the structure would induce significant bias in the resilience
assessment results. In addition, the minimum amplitudes of bias were 83% and 25% for the four- and eight-story case study structures,
respectively.

5. Conclusions
This study proposed a method for assessing the seismic resilience of structures under MSAS sequences. The method was applied to

Fig. 14. Ratio of R for the IF and the BF models.

17
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

two case studies of RC frame structures. The effects of multiple aftershocks and infill walls on the economic losses, recovery time, and
seismic resilience loss factor were quantitatively studied. The following conclusions were reached.
(1) It is essential to consider the effects of multiple aftershocks on the seismic resilience assessment. The total economic losses were
significantly underestimated when only the mainshock was considered, and the degree of underestimation increased with an
increase in the mainshock magnitude. If aftershocks did not result in the reconstruction of structures, the total economic loss
under the MSAS sequence was on average 20%–30% larger than that corresponding to the mainshock-only case. This value
reached 90% if aftershocks resulted in the reconstruction of structures. The average aftershock affecting time for each operating
condition accounted for more than 20% of the total recovery time.
(2) Compared to RC frame structures without infill walls, structures with infill walls had greater economic losses under MSAS
sequences and were more likely to be affected by aftershocks for a longer time. The exclusion of infill walls in the FEM of the
structure induced significant bias in the resilience assessment results, and the minimum amplitudes of bias were 83% and 25%
for four- and eight-story case study structures, respectively.
(3) The same RC frame structure for different applications demonstrated different seismic resilience performances. Unlike office
buildings, indirect economic losses accounted for most of the economic losses to hospitals caused by the MSAS sequence. This
indicates that there is a greater probability that shutdowns caused by earthquakes will have a more severe impact on hospitals.
However, the resilience loss factor of office buildings in each operating condition was greater than that of hospitals, indicating
that the models in this study achieved better seismic resilience when used as hospitals.
This study can provide theoretical support for the seismic resilience design of RC frame structures by considering the effects of
MSAS sequences. On one hand, the seismic capacity of structural and nonstructural components should be improved to reduce the
economic losses and repair time of buildings damaged by multiple aftershocks. On the other hand, it should be recognized that multiple
aftershocks increase the total recovery time by delaying the repair work of buildings while increasing the economic losses. In the
process of recovery of communities and cities after earthquakes, accurate estimation of the aftershock affecting time can facilitate
reasonable scheduling of the restoration of buildings and improve recovery efficiency.
The effects of nonstructural components, multiple aftershocks, and functional loss during the aftershock affecting time were
considered in the resilience quantification process. However, there are still contents that need to be further investigated. The bare and
infilled frame models are considered in this study, but only the results of four- and eight-story RC frame structures were analyzed. The
infill walls were observed to have an impact on the seismic resilience performance, but it was not sufficient to provide a clear influence
law. Therefore, this needs to be further explored by increasing the number of structural examples. The effects of multiple aftershocks
on the economic losses and recovery time of buildings under different sites, structure types, and building usages can also be further
analyzed for the seismic resilience design of structures considering the effects of MSAS sequences. These would allow for the devel­
opment of appropriate pre- and post-earthquake plans to improve the level of seismic resilience of buildings, including how much to
invest in improving the seismic capacity of components and how to develop recovery schedules.

Author statement
Jie Hu: Investigation, Formal analysis, Writing - Original Draft, Software, Visualization; Weiping Wen: Conceptualization, Writing -
Review & Editing, Funding acquisition; Changhai Zhai: Funding acquisition, Project administration, Resources; Shunshun Pei:
Conceptualization, Writing - Review & Editing; Duofa Ji: Resources, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability

No data was used for the research described in the article.

Acknowledgements
This investigation is supported by the National Natural Science Foundation of China (No. 52222811, 52008142 and U1939210),
Natural Science Foundation of Heilongjiang Province (No. LH2021E075), the Fundamental Research Funds for the Central Universities
(No. HIT. BRET.2022010). These supports are greatly appreciated.

18
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

APPENDIX A

Table A.1
Information of selected MSAS ground motions

No. Earthquake MSAS ground motions Mainshock Aftershock

Mw PGA/g Mw PGA/g

1 ChiChi ChiChi-ms-as2_TCU067-N 7.6 0.325 6.2 0.206

2 ChiChi-ms-as2_TCU078-E 7.6 0.444 6.2 0.471
3 ChiChi-ms-as2_TCU122-E 7.6 0.220 6.2 0.201
4 ChiChi-ms-as2_CHY080-E 7.6 0.968 6.2 0.473
5 ChiChi-ms-as3_CHY101-E 7.6 0.353 6.2 0.178
6 ChiChi-ms-as4_CHY024-E 7.6 0.278 6.2 0.239
7 ChiChi-ms-as4_HWA019-E 7.6 0.127 6.2 0.125
8 ChiChi-ms-as4_TCU056-N 7.6 0.134 6.2 0.102
9 ChiChi-ms-as4_TCU059-N 7.6 0.173 6.2 0.139
10 ChiChi-ms-as4_TCU138-N 7.6 0.225 6.2 0.152
11 Darfield, NZ2010-ms-as_NZCACS-N40E 7.0 0.182 6.2 0.186
12 New Zealand NZ2010-ms-as_NZCCCC-N26W 7.0 0.202 6.2 0.367
13 NZ2010-ms-as_NZCHHC-N01W 7.0 0.198 6.2 0.336
14 NZ2010-ms-as_NZCMHS-N10E 7.0 0.238 6.2 0.397
15 NZ2010-ms-as_NZDFHS-S17E 7.0 0.459 6.2 0.459
16 NZ2010-ms-as_NZHPSC-N04W 7.0 0.156 6.2 0.149
17 NZ2010-ms-as_NZHVSC-S64E 7.0 0.619 6.2 1.187
18 NZ2010-ms-as_NZKPOC-S75E 7.0 0.300 6.2 0.213
19 NZ2010-ms-as_NZREHS-N02E 7.0 0.247 6.2 0.366
20 NZ2010-ms-as_NZRHSC-S04W 7.0 0.219 6.2 0.250
21 Wenchuan WenCh-ms-as_051JYD-NS 7.9 0.467 5.8 0.338
22 WenCh-ms-as_051LXM-EW 7.9 0.329 6.3 0.222
23 WenCh-ms-as_051LXM-NS 7.9 0.289 6.3 0.189
24 WenCh-ms-as_051LXS-EW 7.9 0.225 5.9 0.158
25 WenCh-ms-as_051LXT-EW 7.9 0.343 6.3 0.210
26 WenCh-ms-as_051LXT-NS 7.9 0.347 6.3 0.202
27 Imperial Valley IV-ms-as_A-E04140 6.5 0.485 5.0 0.262
28 IV-ms-as_A-E05230 6.5 0.379 5.0 0.239
29 IV-ms-as_A-E06230 6.5 0.439 5.0 0.366
30 IV-ms-as_A-HVP225 6.5 0.253 5.0 0.127
31 IV-ms-as_A-HVP315 6.5 0.221 5.0 0.211
32 Northridge, USA NR-ms-as2_R03000 6.7 0.303 5.3 0.117
33 NR-ms-as2_E-RRS228 6.7 0.825 5.3 0.652
34 NR-ms-as2_E-RRS318 6.7 0.487 5.3 0.430
35 NR-ms-as2_FAR090 6.7 0.242 5.3 0.122
36 NR-ms-as2_HOLLY360 6.7 0.358 5.3 0.289
37 NR-ms-as2_JEN022 6.7 0.411 5.3 0.213
38 NR-ms-as2_PACOI090 6.7 0.301 5.3 0.152
39 NR-ms-as2_PACOI360 6.7 0.433 5.3 0.228
40 NR-ms-as2_TUJ262 6.7 0.163 5.3 0.134
41 Whitter, USA RSN589_WHITTIER.A_A-ALH180_ms-as 6.0 0.290 5.3 0.181
42 RSN590_WHITTIER.A_A-ALT000_ms-as 6.0 0.307 5.3 0.271
43 RSN595_WHITTIER.A_A-JAB207_ms-as 6.0 0.229 5.3 0.139
44 RSN639_WHITTIER.A_A-OBR270_ms-as 6.0 0.428 5.3 0.344
45 RSN663_WHITTIER.A_A-MTW090_ms-as 6.0 0.180 5.3 0.155
46 RSN691_WHITTIER.A_A-SMA270_ms-as 6.0 0.138 5.3 0.147
47 RSN598_WHITTIER.A_A-TUJ352_ms-as 6.0 0.207 5.3 0.174
48 RSN614_WHITTIER.A_A-BIR090_ms-as 6.0 0.232 5.3 0.160
49 RSN620_WHITTIER.A_A-GLP267_ms-as 6.0 0.176 5.3 0.150
50 RSN648_WHITTIER.A_A-NYA180_ms-as 6.0 0.149 5.3 0.133

Table A.2
State-dependent seismic fragility parameters

Structural model Component Initial damage state Final damage state

DS1 DS2 DS3 DS4

Sa σln Sa Sa σln Sa Sa σln Sa Sa σln Sa

BF4 model Structure Intact 0.16 0.38 0.28 0.37 0.68 0.33 1.81 0.32
DS1 – – 0.26 0.38 0.65 0.33 1.79 0.33
DS2 – – – – 0.63 0.33 1.77 0.33
DS3 – – – – – – 1.69 0.36
(continued on next page)

19
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

Table A.2 (continued )

Structural model Component Initial damage state Final damage state

DS1 DS2 DS3 DS4

Sa σln Sa Sa σln Sa Sa σln Sa Sa σln Sa

Infilled walls Intact 0.08 0.17 0.13 0.25 0.32 0.36 0.38 0.35
DS1 – – 0.12 0.29 0.31 0.38 0.37 0.37
DS2 – – – – 0.28 0.40 0.35 0.39
DS3 – – – – – – 0.31 0.44
Suspended ceilings Intact 0.74 0.65 0.97 0.69 1.12 0.71 – –
DS1 – – 0.81 0.76 0.97 0.79 – –
DS2 – – – – 0.83 0.77 – –
Key medical equipment Intact 1.26 0.76 – – – – – –
Indoor properties Intact 0.46 0.59 – – – – – –
IF4 model Structure Intact 0.97 0.56 1.55 0.64 3.44 0.74 7.10 0.85
DS1 – – 1.35 0.69 3.38 0.74 6.99 0.88
DS2 – – – – 3.16 0.82 6.91 0.94
DS3 – – – – – – 6.20 0.95
Infilled walls Intact 0.52 0.42 0.74 0.50 1.60 0.62 1.84 0.63
DS1 – – 0.63 0.55 1.54 0.65 1.82 0.64
DS2 – – – – 1.48 0.67 1.78 0.66
DS3 – – – – – – 1.43 0.71
Suspended ceilings Intact 0.87 0.36 1.07 0.37 1.20 0.39 – –
DS1 – – 0.87 0.46 0.98 0.47 – –
DS2 – – – – 0.90 0.47 – –
Key medical equipment Intact 1.39 0.39 – – – – – –
Indoor properties Intact 0.57 0.37 – – – – – –
BF8 model Structure Intact 0.08 0.16 0.10 0.46 0.18 0.67 0.85 0.44
DS1 – – 0.08 0.51 0.17 0.60 0.83 0.43
DS2 – – – – 0.14 0.70 0.81 0.43
DS3 – – – – – – 0.79 0.44
Infilled walls Intact 0.02 0.25 0.07 0.30 0.13 0.55 0.16 0.52
DS1 – – 0.06 0.35 0.12 0.50 0.15 0.50
DS2 – – – – 0.11 0.45 0.14 0.50
DS3 – – – – – – 0.12 0.55
Suspended ceilings Intact 0.20 1.24 0.25 1.31 0.28 1.33 – –
DS1 – – 0.18 1.35 0.21 1.46 – –
DS2 – – – – 0.20 1.46 – –
Key medical equipment Intact 0.34 1.58 – – – – – –
Indoor properties Intact 0.14 1.07 – – – – – –
IF8 model Structure Intact 0.47 0.36 0.78 0.32 2.08 0.41 4.92 0.57
DS1 – – 0.74 0.33 2.05 0.41 4.88 0.57
DS2 – – – – 1.97 0.44 4.85 0.58
DS3 – – – – – – 4.83 0.60
Infilled walls Intact 0.29 0.30 0.45 0.33 1.11 0.38 1.36 0.34
DS1 – – 0.41 0.40 1.10 0.37 1.33 0.35
DS2 – – – – 1.04 0.40 1.28 0.37
DS3 – – – – – – 0.99 0.53
Suspended ceilings Intact 0.94 0.55 1.20 0.60 1.36 0.63 – –
DS1 – – 1.04 0.73 1.26 0.70 – –
DS2 – – – – 1.06 0.85 – –
Key medical equipment Intact 1.59 0.64 – – – – – –
Indoor properties Intact 0.54 0.53 – – – – – –

References
[1] S. Mahin, Lessons from recent earthquakes: the need for more resilient cities, Building Structure 42 (2012) 1–8.
[2] N. Theodoulidis, C. Karakostas, V. Lekidis, K. Makra, B. Margaris, K. Morfidis, C. Papaioannou, E. Rovithis, T. Salonikios, A. Savvaidis, The Cephalonia, Greece,
January 26 (M6.1) and February 3, 2014 (M6.0) earthquakes: near-fault ground motion and effects on soil and structures,, Bull. Earthq. Eng. 14 (2016) 1–38.
[3] K. Goda, A. Pomonis, S.C. Chian, M. Offord, K. Saito, P. Sammonds, S. Fraser, A. Raby, J. Macabuag, Ground motion characteristics and shaking damage of the
11th March 2011 Mw9.0 Great East Japan earthquake, Bull. Earthq. Eng. 11 (2013) 141–170.
[4] I. Iervolino, E. Chioccarelli, A. Suzuki, Seismic damage accumulation in multiple mainshock–aftershock sequences, Earthq. Eng. Struct. Dynam. 49 (2020)
1007–1027.
[5] M. Raghunandan, A.B. Liel, N. Luco, Aftershock collapse vulnerability assessment of reinforced concrete frame structures, Earthq. Eng. Struct. Dynam. 44 (2015)
419–439.
[6] P. Oggu, K. Gopikrishna, Assessment of three-dimensional RC moment-resisting frames under repeated earthquakes, Structures 26 (2020) 6–23.
[7] M. Shokrabadi, H.V. Burton, Building service life economic loss assessment under sequential seismic events, Earthq. Eng. Struct. Dynam. 47 (2018) 1864–1881.
[8] M. Parker, D. Steenkamp, The economic impact of the Canterbury earthquakes,, Reserve Bank of New Zealand Bullet. 75 (2012) 13–25.
[9] X. Li, W. Wen, C. Zhai, A probabilistic framework for the economic loss estimation of structures considering multiple aftershocks, Soil Dyn. Earthq, Eng. Times
133 (2020) 106121.

20
J. Hu et al. Journal of Building Engineering 68 (2023) 106110

[10] R. Han, Y. Li, J. van de Lindt, Impact of aftershocks and uncertainties on the seismic evaluation of non-ductile reinforced concrete frame buildings, Eng. Struct.
100 (2015) 149–163.
[11] M. Bruneau, S. Chang, R. Eguchi, G. Lee, T. O’Rourke, A. Reinhorn, M. Shinozuka, K. Tierney, W. Wallace, D. Winterfeldt, A framework to quantitatively assess
and enhance the seismic resilience of communities, Earthq, Spectra 19 (2003) 733–752.
[12] P. Yu, C. Zhai, J. Liu, S. Pei, Z. Song, L. Xie, Resilience assessment methods for hospitals subjected to earthquakes: state of the art, Earthquake Eng. Resilience 1
(2022) 40–59.
[13] P. Yu, W. Wen, D. Ji, C. Zhai, L. Xie, A framework to assess the seismic resilience of urban hospitals, adv. Civ, Eng. Times 2019 (2019) 7654683.
[14] J. Liu, C. Zhai, P. Yu, A probabilistic framework to evaluate seismic resilience of hospital buildings using bayesian Networks, Reliab. Eng. Syst. Saf. (2022)
108644.
[15] J. Li, T. Wang, Q. Shang, Probability-based seismic resilience assessment method for substation systems, Struct. Infrastruct. Eng. 18 (2022) 71–83.
[16] T.K. Singhal, Civil Infrastructure Resilience Assessment Framework (CIRAF), in, University of Toronto, Canada, 2018.
[17] Y. Dong, D.M. Frangopol, Performance-based seismic assessment of conventional and base-isolated steel buildings including environmental impact and
resilience, Earthq. Eng. Struct. Dynam. 45 (2016) 739–756.
[18] Q. Shang, T. Wang, J. Li, A quantitative framework to evaluate the seismic resilience of hospital systems,, J. Earthq. Eng. 26 (2020) 3364–3388.
[19] G.P. Cimellaro, A.M. Reinhorn, M. Bruneau, Seismic resilience of a hospital system, Struct. Infrastruct. Eng. 6 (2010) 127–144.
[20] Standard for seismic resilience assessment of buildings, State Administration for Market Regulation; Standardization Administration of china, 2020, p. 68.
[21] G.A. Anwar, Y. Dong, C. Zhai, Performance-based probabilistic framework for seismic risk, resilience, and sustainability assessment of reinforced concrete
structures, Adv. Struct. Eng. 23 (2019) 1454–1472.
[22] A. Furtado, H. Rodrigues, H. Varum, A. Arêde, Mainshock-aftershock damage assessment of infilled RC structures, Eng. Struct. 175 (2018) 645–660.
[23] F. Di Trapani, M. Malavisi, Seismic fragility assessment of infilled frames subject to mainshock/aftershock sequences using a double incremental dynamic
analysis approach, Bull. Earthq. Eng. 17 (2018) 211–235.
[24] I. Iervolino, M. Giorgio, E. Chioccarelli, Markovian modeling of seismic damage accumulation, Earthq. Eng. Struct. Dynam. 45 (2016) 441–461.
[25] G.L. Yeo, C.A. Cornell, A probabilistic framework for quantification of aftershock ground-motion hazard in California: methodology and parametric study,
Earthq. Eng. Struct. Dyn. 38 (2009) 45–60.
[26] P.A. Reasenberg, L.M. Jones, Earthquake Hazard After a Mainshock in California 243 (1989) 1173–1176.
[27] D. Vamvatsikos, C.A. Cornell, Incremental dynamic analysis, Earthq. Eng. Struct. Dynam. 31 (2002) 491–514.
[28] W. Wen, M. Zhang, C. Zhai, W. Liu, Resilience loss factor for evaluation and design considering the effects of aftershocks, Soil Dynam. Earthq. Eng. 116 (2019)
43–49.
[29] G.P. Cimellaro, A.M. Reinhorn, M. Bruneau, Framework for analytical quantification of disaster resilience, Eng. Struct. 32 (2010) 3639–3649.
[30] S. Zheng, Z. Shang, J. He, H. Zheng, J. Dong, Method and application of economic loss assessment for earthquake disasters, J. Catastrophol. 35 (2020) 94–101.
[31] H. Chen, Study on Earthquake Damage Loss Assessment of Urban Buildings’ Decorations, in, Institute of Engineering Mechanics, China Earthquake
Administration, 2008.
[32] Post-Earthquake Field Works—Part 4:Assessment of Direct Loss, in, Administration of Quality Supervision,Inspection and Quarantine;Standardization
Administration of China, 2011, p. 48. Beijing.
[33] S. Chang, M. Shinozuka, Measuring improvements in the disaster resilience of communities, Earthq. Spectra 20 (2004) 739–755.
[34] L. Martins, V. Silva, M. Marques, H. Crowley, R. Delgado, Development and assessment of damage-to-loss models for moment-frame reinforced concrete
buildings, Earthq. Eng. Struct. Dynam. 45 (2016) 797–817.
[35] G. Cimellaro, A. Reinhorn, M. Bruneau, Quantification of seismic resilience, Proc. 8th Nat. Conf. Earthquake Eng. 8 (2006) 1–10.
[36] Code for design of concrete structures, Administration of Quality Supervision;Inspection and Quarantine, 2010, p. 440. B445.
[37] Code for Seismic Design of Buildings, 2010, p. 499. A494.
[38] China Construction Engineering Cost Information Network. http://www.cecn.org.cn/.
[39] China Engineering Budget Network. http://de.yusuan.com/index.html.
[40] K. Bi, Vulnerability Analysis of Group Buildings and Earthquake Loss Fast Estimation, in, Dalian University of Technology, 2009.
[41] M.H. Fema, 2.1 Earthquake Model Technical Manual, 2012.
[42] Open System for Earthquake Engineering Simulation. https://opensees.berkeley.edu/.
[43] Concrete02 Material, Linear tension softening. https://OpenSees.berkeley.edu/wiki/index.php/Concrete02_Material_–_Linear_Tension_Softening.
[44] B.D. Scott, R. Park, M.J. Priestley, Stress-strain behavior of concrete confined by overlapping hoops at low and high strain rates, J. Proc. (1982) 13–27.
[45] Steel02 Material, Giuffré-menegotto-pinto model with isotropic strain hardening. https://opensees.berkeley.edu/wiki/index.php/Steel02_Material_–_Giuffr%
C3%A9-Menegotto-Pinto_Model_with_Isotropic_Strain_Hardening.
[46] Fiber Section. https://opensees.berkeley.edu/wiki/index.php/Fiber_Section.
[47] Force-Based Beam-Column Element. https://opensees.berkeley.edu/wiki/index.php/Force-Based_Beam-Column_Element.
[48] Pinching4 Material. https://opensees.berkeley.edu/wiki/index.php/Pinching4_Material.
[49] L. Liberatore, F. Noto, F. Mollaioli, P. Franchin, In-plane response of masonry infill walls: comprehensive experimentally-based equivalent strut model for
deterministic and probabilistic analysis, Eng. Struct. 167 (2018) 533–548.
[50] G. Blasi, F. De Luca, M.A. Aiello, Brittle failure in RC masonry infilled frames: the role of infill overstrength, Eng, Structure 177 (2018) 506–518.
[51] M. Kumar, Ductility Reduction Factors for Masonry Infilled RC Frames - Effect of Hysteretic Model Parameters, 2010.
[52] D. Wells, K. Coppersmith, New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement, Bull. Seismol.
Soc. Am. 84 (1994) 974–1002.
[53] X. Li, C. Zhai, W. Wen, L. Xie, Ground motion prediction model for horizontal PGA, 5% damped response spectrum in sichuan-yunnan region of China, J. Earthq.
Eng. 24 (2020) 1829–1866.
[54] X. Xie, L. Zhang, Z. Qu, Reparability-based fragility analysis of masonry infills in buildings, J. Build. Struct. 39 (2018) 159–167.
[55] W. Yang, J.-P. Ou, Finite element simulation and analysis of performance of infill walls of aseismic structures, Huanan Ligong Daxue Xuebao/J. South China
Univ. Technol. (Nat. Sci.) 38 (2010) 140–144.
[56] Q. Shang, T. Wang, J. Li, Seismic resilience assessment of emergency departments based on the state tree method, Struct. Saf. 85 (2020) 101944.
[57] Fema, Seismic performance assessment of buildings, methodology, in, Federal Emerg. Manag. Agency (1) (2012). Washington, DC.
[58] Fema, Seismic Performance Assessment of Buildings Volume 2, Federal Emergency Management Agency, Washington, DC, 2012 implementation guide, in.
[59] Pacific Earthquake Engineering Research Center. https://peer.berkeley.edu/nga/.
[60] The European Strong-Motion Database. http://www.isesd,hi,is/ESD_Local/frameset.htm.
[61] Center for Engineering Strong Motion Data. http://www.strongmotioncenter.org/.
[62] China Earthquake Networks Center. https://www.csndme.ac.cn/newweb/data.htm.
[63] Y. Ma, G. Zhao, Seismic Hazard Risk Analysis and Management, China, Science Publishing & Media Ltd, Beijing, 2008.

21