Engineering Structures 211 (2020) 110450

Contents lists available at ScienceDirect

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

Fragility analysis of reticulated domes subjected to multiple earthquakes T

a,b a,b,⁎ a,b
Yingnan Zhang , Xudong Zhi , Feng Fan
a
Key Lab of Structure 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 Disasters of Ministry of Industry and Information Technology, Harbin Institute of Technology, Harbin
150090, China

A R T I C LE I N FO A B S T R A C T

Keywords: In this paper, single-layer and double-layer reticulated domes were taken as two typical research objects.
Reticulated domes Through the dynamic time history analysis of reticulated domes in full process domain, log-linear regression was
Main shock-aftershock sequences performed with a large number of example results to compare the degrees of correlation and efficiency about 11
Damage levels common ground motion intensity measures, and PGA was selected as the most applicable one for fragility
Aftershocks
analysis of reticulated domes with aftershock influence. Then, considering that main shock usually caused
Fragility analysis
structural damage, with incremental dynamic analysis method, fragility analysis of reticulated domes under
multiple earthquakes was conducted from two aspects as follows. One part was main shock-aftershock sequence
fragility analysis of intact structures. The other part was aftershock fragility analysis of damaged structures with
four different damage levels caused by main shocks. Based on the fragility curves, the aftershock influence on
fragility of reticulated domes and the aftershock collapse probability of damaged reticulated domes with four
different damage levels caused by main shocks were investigated. It was found that when the damage level
reached severe damage under main shocks, the aftershock collapse probability of reticulated domes would in-
crease significantly. The results expanded the application scope of traditional fragility analysis in seismic en-
gineering and realized the prediction and evaluation of seismic performance of reticulated domes under after-
shocks.

1. Introduction intensities, which has become the most commonly used method to
predict structural states under earthquakes nowadays. Mathematically,
A complete earthquake process contains a main shock, aftershocks it refers to the conditional probability that structural damage level
and related swarm sequences [1]. Historical earthquake disasters in- exceeds a certain limit state under an earthquake with a certain ground
dicated that not only main shocks could affect structural normal func- motion intensity. The expression is:
tion, but also aftershocks could cause structural damage, even make Sd
structures collapse. For example, in the 2008 Wenchuan earthquake, Pf = P [ ⩾ 1]
Sc (1)
the largest magnitude and frequency of aftershocks were 6.4 Ms and
more than 70 thousand times respectively, which leaded about seventy with Pf the structural failure probability for a certain limit state, Sd the
thousand houses collapse under aftershocks [2]. Because of the non- structural seismic responses with a certain ground motion intensity, and
negligible influence of multiple earthquakes on structures, so far, many Sc the structural seismic responses for a certain limit state. In addition, a
scholars have performed a series of researches about multiple earth- large number of previous statistical results showed that both Sd and Sc
quake influence on seismic performance of some conventional structure followed lognormal distribution [6–8], so that Eq. (1) could also be
forms, such as concrete structures and steel-frame structures [3–5]. As a expressed as Eq. (2):
result, they found that aftershocks had an obvious influence on struc- Sd ln(1) − λ λ
tural seismic responses, and current seismic design codes could not Pf = P [ ⩾ 1] = 1 − Φ( ) = Φ( )
Sc ξ ξ (2)
guarantee structural safety adequately only considering main shocks.
Nowadays, in seismic engineering, seismic fragility analysis is where Φ (●) meant cumulative distribution function (CDF) with stan-
commonly used to characterize the failure probability of structures dard normal distribution, then λ and ξ referred to the median and
S
reaching different limit states under different ground motion standard deviation of ln( Sd ) respectively, obtained by statistical
c

⁎
Corresponding author at: Space Structures Research Center, School of Civil Engineering, Harbin Institute of Technology, China.
E-mail address: zhixudong@hit.edu.cn (X. Zhi).

https://doi.org/10.1016/j.engstruct.2020.110450
Received 2 July 2019; Received in revised form 10 December 2019; Accepted 28 February 2020
Available online 07 March 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

regression analysis of many numerical example results. dynamic time history analysis, the representative value of gravity load
In recent years, except reticulated domes, fragility analysis of some of reticulated domes was suggested to take the standard value of con-
other common structure forms subjected to multiple earthquakes has stant loading plus the 50% standard value of snow loading. In the finite
been conducted gradually. For concrete structures, D'Aragona et. al element models of reticulated domes, the representative value of
investigated aftershock collapse fragility of RC buildings using incre- gravity load was converted to concentrated mass applied to the nodes of
mental dynamic analysis method and found that the collapse prob- reticulated domes. Moreover, finite element model and seismic re-
ability obviously increased for higher return periods of main shocks [9]. sponse analysis of reticulated domes were established and conducted by
Abdelnaby evaluated seismic fragility of different types of RC frames the finite element software ABAQUS.
with and without multiple earthquake influence respectively and came
to a conclusion that multiple earthquakes had significant effects on the 2.1. Finite element model of reticulated domes
fragility of RC frames [10]. Pang and Wu also did the fragility analysis
of RC bridges under main shock-aftershock sequences and got the si- The design process was divided into two parts: static design and
milar conclusion that aftershocks would increase structural collapse dynamic design. During the static design process, considering geometric
probability [11]. For steel-frame structures, Silwal and Ozbulut ex- and material nonlinearity, SD-R8-S40 was designed with the Kiewitt-8
plored the seismic fragility of two kinds of steel frames and the results single-layer reticulated dome form. The Beam31 element was chosen as
showed that with the intensity of main shocks higher, aftershocks each member, the cross section of which used the PIPE section [14].
caused more damage for the SMRF, but less damage for the SVD frame With members linked by rigid connection, hinge supports were set as
[12]. In addition, Park et. al established an assessment model for the boundary condition of the structure. The length of members and the
structural aftershock fragility and verified the applicability and ra- number of structural rings were obtained according to the limitation of
tionality of the model [13]. slenderness ratio of members in related Chinese design codes [22].
For reticulated domes, on account of the large service space, they Considering the damage accumulation effect, the material of members
were often used as temporary emergency shelters in earthquake- was made of Q235 steel with a 2.1 × 105 N/mm2 elastic modulus. The
stricken areas, so how to ensure structural safety under multiple section dimension of each member was determined based on the related
earthquakes was an urgent problem to be solved with great social sig- Chinese design code [22] to meet general design requirements with the
nificance. Based on single earthquake, lots of remarkable research re- assurance of structural stability under static loads by employing a safety
sults about seismic performance of reticulated domes have been factor of 2.0. During the dynamic design process, the damping model
achieved, including the researches of failure mechanism [14,15], fra- applied the Rayleigh damping, with Rayleigh mass proportional
gility analysis [6,16] and joint behavior [17]. However, compared to damping factor α and Rayleigh stiffness proportional damping factor β.
other conventional structure forms, the researches of multiple earth- The damping ratio was 0.02 according to the suggestion in related
quake influence on reticulated domes are in their infancy. Zhi and design codes and references [14,23]. In addition, the influence of initial
Zhang [18] explored the response law and failure mechanism of single- bending of members was considered and the deformation of the first-
layer reticulated domes under multiple earthquakes and defined five order buckling mode was taken as the initial imperfection with 1/300
different levels of aftershock influence on seismic responses. They also of the span.
proposed the seismic design method considering aftershocks that by There existed several differences in finite element model of SD-R8-
controlling the values of 1P ratio and structural damage factor at the S40 and DD-R12-S60 (Fig. 1), that the Truss element was chosen as
end of main shock, the safety of single-layer reticulated domes under each member and the linking form of members and boundary condition
aftershocks could be guaranteed. Nevertheless, fragility analysis of re- are both hinge connection for DD-R12-S60. Besides, other aspects of
ticulated domes with multiple earthquakes has not been performed yet, DD-R12-S60 were the same with those of SD-R8-S40. The above finite
which restricted the development of reticulated domes in seismic re- element modeling method has been investigated and supported by
gions and had influence on structural function under earthquakes. In several shaking-table tests and references [14,23,24].
view of this backgrounds, with single-layer and double-layer reticulated
domes as examples, this paper determined the aftershock ground mo- 2.2. Selection and input method of main shock-aftershock sequences
tion intensity measure applicable to fragility analysis of reticulated
domes under multiple earthquakes at first. Then, based on incremental In seismic engineering, considering the applicability of seismic
dynamic analysis method, from the perspective of uncertainty, fragility analysis, Ref [18] provided four criteria for selection of main shock-
analysis of reticulated domes considering main shock-aftershock se- aftershock sequences as follows. (i) Main shock and its corresponding
quences was conducted to discuss structural probability of exceedance aftershocks should occur within 100 days. (ii) Longitude and latitude of
related to different limit state. Finally, aftershock collapse fragility epicentres of main shock and its corresponding aftershocks should
curves of damaged reticulated domes with different damage levels were differ by less than 0.6 degrees. (iii) Main shock and aftershock ground
obtained, which provided reference and support for the seismic re- motions should recorded by the same seismic station. (iv) Premised on
habilitation of reticulated domes. meeting (i), (ii) and (iii), aftershock with the largest magnitude should
be selected. According to the above criteria, 342 pairs of main shock-
2. Finite element model and input method of multiple aftershock sequences were collected from the Pacific Earthquake En-
earthquakes gineering Research Center (PEER), shown in Table 1. For convenience
of representation, the abbreviations of event names were given in
Numerical analysis is one of the most commonly used methods to brackets.
solve structural engineering problems [16,18,19]. Due to the limited In Ref [18], the input method of main shock-aftershock sequences in
space, in this paper, the reticulated domes with only two parameters ABAQUS were constructed that displacement time histories of main
were taken as typical research objects. For the convenience of ex- shock and its corresponding aftershock were linked end to end with a
planation, the reticulated domes were denoted by numbers. In detail, 30 s zero-amplitude interval. Fig. 2 showed the input displacement time
SD-R8-S40 referred to the single-layer reticulated dome, with a 40 m history curves recorded by CHY024 Station in the 1999 Chi-Chi
span and a 1/5 rise-span ratio. DD-R12-S60 referred to the double-layer earthquake. Ref [18] has proved the rationality of this input method
reticulated dome, with a 60 m span and a 1/5 rise-span ratio. The roof and the length of time of a 30 s zero-amplitude interval, which could
weight was 180 kg/m2, contained constant loading (about 150–160 kg/ achieved the vibration balance state that structural vibration would be
m2) and snow loading (about 50 kg/m2). According to the related essentially attenuated to zero before aftershocks. Furthermore, the
provisions in Chinese structural design codes [20,21], in seismic ground motions were applied simultaneously at three directions and the

2
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 1. Finite element model of SD-R8-S40 and DD-R12-S60.

amplitude scaling ratios adopted the amplitude ratios of the actual PGA = max |a (t )| (3)
original ground motions at three directions.
where a (t ) is earthquake acceleration time histories

(2) Peak ground velocity (PGV):

3. Selection of appropriate aftershock ground motion intensity
measure PGV = max |v (t )| (4)
where v (t ) is earthquake velocity time histories.
Prior to seismic fragility analysis, choosing the appropriate ground
motion intensity measure is the basis for obtaining reliable and rea- (3) Peak ground displacement (PGD):
sonable results. For reticulated domes, Zhong proposed the dominant
vibration mode-based ground motion intensity measure Sa, dom(Td 1, Td PGD = max |d (t )| (5)
2, …, Td i, …, Td N, ξ), and evaluated the applicability of this parameter where d (t ) is earthquake displacement time histories.
on seismic fragility analysis from several aspects [25,26]. However, Sa,
dom(Td 1, Td 2, …, Td i, …, Td N, ξ) was put forward based on main (4) Arias intensity (IA):
shocks. Since the dominant vibration modes of damaged reticulated
domes under main shocks would change compared with those before As characterization of total seismic intensity, IA was proposed by
main shocks, Sa, dom(Td 1, Td 2, …, Td i, …, Td N, ξ) was no longer Arias in 1969 to represent the energy consumed by a unit particle in a
suitable for the fragility analysis under main shock-aftershock se- simple elastic particle system during the whole seismic process [27].
quences. In view of this, this paper found a new rational parameter for The expression is:
fragility analysis of reticulated domes subjected to multiple earth-
π Td
quakes, from 11 commonly used ground motion intensity measures in IA =
2g
∫0 a2 (t ) dt
(6)
seismology and seismic engineering as follows.
where g means the acceleration of gravity (9.81 m/s2), Td represents the
(1) Peak ground acceleration (PGA): total duration of an earthquake.

Adopted by structural seismic design codes in most countries in the (5) Cumulative absolute velocity (CAV):
world, PGA is the most widely used ground motion intensity measure in
seismology and seismic engineering. The expression is:

Table 1
Collection of 342 pairs of main shock-aftershock sequences.
Event name Station number Main shock magnitude (Ms) Aftershock magnitude (Ms) Site condition Main shock time Aftershock time

Rock Soil

Chalfant Valley (CV) 5 5.77 6.19 – 5 1986.07.20 14:29 1986.07.21 14:42

Chi-Chi, Taiwan (CT) 269 7.62 6.3 134 135 1999.09.20 1999.09.25
Coalinga (CO) 1 6.36 5.77 – 1 1983.05.02 23:42 1983.07.22 02:39
Hollister (HO) 1 5.6 5.5 – 1 1961.04.09 07:23 1961.04.09 07:25
Irpinia, Italy (II) 10 6.9 6.2 8 2 1980.11.23 19:34 1980.11.23 19:35
Livermore (LI) 6 5.8 5.42 2 4 1980.01.24 19:00 1980.01.27 02:33
LuShan (LS) 8 7 5.4 1 7 2013.04.20 08:02 2013.04.20 08:07
Mammoth Lakes (ML) 2 6.06 5.94 – 2 1980.05.25 16:34 1980.05.27 14:51
Managua, Nicaragua (MN) 1 6.24 5.2 – 1 1972.12.23 06:29 1972.12.23 07:19
Northridge (NO) 17 6.69 6.05 8 9 1994.01.17 12:31 1994.01.17 12:32
Northwest China (NC) 1 5.9 6.1 – 1 1997.04.05 23:46 1997.04.11 05:34
Superstition Hills (SH) 1 6.22 6.54 – 1 1987.11.24 05:14 1987.11.24 13:16
WenChuan (WC) 9 8 6.4 2 7 2008.05.12 14:28 2008.05.25 16:21
Whittier Narrows (WN) 11 5.99 5.27 4 7 1987.10.01 14:42 1987.10.04 10:59
Total 342 – – 159 183 – –

3
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 2. Input displacement time history curves corresponding to CHY024 Station.

n
Table 2 1
Damage factor of reticulated domes with different damage levels.
Sa, avg = ∏ (Sa (Ti )) n
1 (10)
Damage No Damage Slight Damage Medium Significant Collapse
Levels Damage Damage
where Ti is the value of a certain structural period.

DIE 0–0.30 0.30–0.50 0.50–0.70 0.70–1.00 1.00 (11) Ground motion intensity measure considering Sa(T1) and Sa,avg
(INp):

Table 3 INp = Sa (T1) N pα (11)

30 pairs of main shock-aftershock sequences for the evaluation of ground mo-
Sa, avg
tion intensity measures. Np =
Sa (T1) (12)
Abbreviation of Shear Site Shear Shear Site
seismic station wave condition wave wave condition where T1 means structural fundamental periods, the range of α is 0–1.
velocity of China velocity velocity of China Existing researches showed that the relationship between structural
(m/s) (m/s) (m/s)
seismic demand D and ground motion intensity measure IM usually
HWA003 1525.85 I MSJ 367.6 II obeyed power exponent function under main shocks [7]. The expres-
BAG 1000 I DVD 338.5 II sion is:
ILA063 996.5 I ING 316 II
TTN042 845.3 I MAN 288.8 II D = a × (IM )b × ε (13)
TTN040 728 I NEW 269.1 II
TTN025 705 I PVP 257.4 III where a and b are regression coefficients, ε is error term. By logarithmic
HWA038 642.7 I CHY063 246.9 III transformation of Eq. (13), the following expression was obtained:
LAK 602.1 I ELI 234.9 III
ln D = ln a + b ln IM + ln ε (14)
PAL 551.6 I TAP010 226.4 III
APE 517.1 I CHY069 215 III which indicated that ln D and ln IM were subjected to a linear function.
PKC 508.1 II WLA 207.5 III
CHY081 473.9 II CHY044 194.5 III
However, for the study on seismic performance of reticulated domes
ANA 446 II CHY059 191.1 III under multiple earthquakes, different damage levels of reticulated
TTN041 418.2 II CHY054 172.1 III domes at the end of main shocks should be considered necessarily. Ref
HWA033 395.6 II CHY078 160.7 III [18] has indicated that even though aftershock ground motions are the
same, if reticulated domes were with different damage levels at the end
Td
of main shocks, the influence of aftershocks on reticulated domes were
CAV = ∫0 |a (t )| dt (7) also different. So, in this study, reticulated domes at different damage
levels under main shocks should be made, which should be discussed
(6) Spectrum intensity of acceleration (SIA): separately. According to Ref [28], the damage levels of reticulated
domes and the corresponding damage factors were shown in Table 2.
0.5
SIA = ∫0 Sa (T , ξ ) dT (8)
Then, according to Eq. (13), structural seismic demand under
aftershocks D2 and aftershock ground motion intensity measure IM2
where Sa is earthquake acceleration response spectra, ξ means damping appear the following relationship:
ratio (equals 0.02 for reticulated domes).
Dt = D1 + D2 = D1 + a2 × (IM2)b2 × ε2 (15)
(7) Spectrum intensity of velocity (SIV): where D1 and D2 are structural damage caused by main shocks and
2.5 aftershocks respectively. By logarithmic transformation of Eq. (15), the
SIV = ∫0 Sv (T , ξ ) dT (9) following expression was obtained:
where Sv is earthquake velocity response spectra. ln D2 = ln a2 + b2 ln IM2 + ln ε2 (16)
which indicated that lnD2 and lnIM2 were also subjected to a linear
(8) Acceleration response spectrum values corresponding to structural function.
fundamental periods (Sa(T1)) Based on the above linear function of Eq. (16), the rationality of
(9) Acceleration response spectrum values corresponding to structural ground motion intensity measures would be evaluated. There exist lots
sensitive frequencies (Sa(fp)) of evaluation indexes to appraise the applicability of ground motion
(10) Average values of acceleration response spectrum corresponding intensity measures. In this paper, two most commonly used indexes
to several structural periods (Sa,avg): were selected, correlation index and efficiency index.

4
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 3. Linear fitting results of ln(IM2) and ln(DIEp) for SD-R8-S40.

Correlation index used Pearson’s correlation coefficient ρ [25], to stronger the correlation of two parameters is. The expression is:
evaluate the degree of linear correlation between lnD2 and lnIM2, the
range of which was usually between −1 and 1. The larger the absolute
value of ρ is, the greater the degree of linear correlation is and the

5
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 3. (continued)

6
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 4. Linear fitting results of ln(IM2) and ln(DIEp) for DD-R12-S60.

7
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 4. (continued)

8
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Finally, by least square method, 2812 pairs of example results

(ln IM2i , ln D2i ) were fitted linearly, where D2 was expressed by DIEp. The
fitting results of SD-R8-S40 and DD-R12-S60 were shown in Figs. 3 and
4 respectively.
The results indicated that for reticulated domes with no damage and
some damage under main shocks respectively, the degrees of correla-
tion and efficiency about 11 aftershock ground motion intensity mea-
sures were significantly different, but also existed certain laws. For SD-
R8-S40, with no damage under main shocks, the degrees of correlation
and efficiency of PGA、IA、CAV、SIA appeared better, with ρ all above
0.94 and β all below 0.35. By contrast, the degrees of correlation and
efficiency of PGD、Sa(fp)、Sa,avg appeared worse. Besides, with some
damage (slight damage, medium damage and significant damage)
under main shocks, the degrees of correlation and efficiency of PGA、
IA、SIA、Sa(T1) appeared better, which of PGD、SIV、Sa,avg appeared
worse. For DD-R12-S60, with no damage under main shocks, the de-
grees of correlation and efficiency of IA、CAV、SIA appeared better,
with ρ all above 0.94 and β all below 0.34, and the degrees of corre-
lation and efficiency of PGD、Sa,avg、INp appeared worse. Furthermore,
Fig. 5. Probability of occurrence for different PGA ratio of main shock-after-
shock sequences. with some damage (slight damage, medium damage and significant
damage) under main shocks, the degrees of correlation and efficiency of
n PGA、IA、SIA、Sa(T1) appeared better, which was consistent with the
¯ 2 )(ln D2 − ln ¯D2 )
∑ (ln IM2i − ln IM results of SD-R8-S40, and the degrees of correlation and efficiency of
i
ρ= i=1 PGD、Sa,avg、INp still appeared worse.
n n
¯ 2 )2 ∑ (ln D2 − ln ¯D2 )2
∑ (ln IM2i − ln IM With summary of the above analysis results, the degrees of corre-
i
i=1 i=1 (17) lation and efficiency of PGA、IA、SIA appeared better than other
aftershock ground motion intensity measures, and there existed no
Efficiency index used linear fitting variance β [29], to evaluate the
significant differences among them. Considering the applicability and
dispersion degree between lnD2 and lnIM2. The larger the value of β is,
universality of theoretical analysis and practical engineering seismic
the greater the dispersion degree is and the worse the efficiency is. The
design, PGA is the most convenient to obtain and use. PGA is also the
expression is:
ground motion intensity measure adopted in the current Chinese
n seismic code [21] for seismic analysis and seismic design, so that the
∑ (ln D2 − ln ¯D2 )2 selection of PGA can be consistent with the related seismic codes. Fi-
i=1
β= nally, PGA was selected as the most suitable aftershock ground motion
n−2 (18)
intensity measure for the analysis of reticulated domes under multiple
Based on the above evaluation methods and indexes, seismic re- earthquakes.
sponses of SD-R8-S40 and DD-R12-S60 were analyzed in full process In addition, through the above results, it was found that the ground
domain under main shocks and aftershocks in sequence, to investigate motion intensity measures with better correlation degree were also with
the applicability of the above 11 common ground motion intensity better efficiency degree. Compared with no damage, slight damage and
measures for reticulated domes. Damage factor used the damage index medium damage, these three damage levels under main shocks, for the
based on seismic demand DIE [28]. According to site conditions, with reticulated domes with significant damage, ground motion intensity
different shear wave velocities, the input ground motion records were measures were often with the worse correlation degree, but there was
selected from Table 1, shown in Table 3. With Ref [30], at least 15 no obvious change in the degree of efficiency.
ground motion records could satisfy the requirements of the number of
ground motion records for seismic analysis of reticulated domes.

Fig. 6. Main shock-aftershock sequence fragility curves with different PGA ratio following Gaussian distribution.

9
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 7. Main shock-aftershock sequence fragility curves of SD-R8-S40 with 0–1.2 PGA ratio.

4. Fragility analysis considering main shock-aftershock sequences part was fragility analysis of intact reticulated domes under main
shock-aftershock sequences and main shock-aftershock sequence fragi-
There existed two general methods for seismic fragility analysis. lity curves of structures were obtained for seismic design of reticulated
One was an empirical analysis method based on a large amount of domes considering multiple earthquake effects before earthquakes. The
historical earthquake disaster data. The other one was a theoretical other part was fragility analysis of damaged reticulated domes with
analysis method based on incremental dynamic analysis. Because the different damage levels caused by main shocks under aftershocks and
earthquake disaster data about reticulated domes was difficult to collect aftershock fragility curves of damaged structures were obtained for
with insufficient quantity, this paper used the theoretical analysis evaluation of seismic performance of damaged reticulated domes under
method to discuss the seismic fragility of reticulated domes considering aftershocks.
multiple earthquake effects.
There also existed two representation methods for seismic fragility
4.1. Fragility analysis under main shock-aftershock sequences
analysis. One was continuous curves. This method performed a linear
fitting regression on a large number of example results and obtained a
In this section, the reticulated domes with no damage were taken as
series of continuous curves with ground motion intensity measures as
research objects, incremental dynamic analysis of which under main
abscissa and probability of exceedance as ordinate, to describe the
shock-aftershock sequences was performed. Besides, it should be spe-
probability of structures reaching different damage levels under dif-
cially mentioned that incremental dynamic analysis step involved not
ferent magnitude of ground motion intensity measures. The other one
only PGA of main shocks, but also PGA ratio of main shock-aftershock
was discrete matrix, which could rapidly obtain the probability that
sequences (PGA of aftershocks/PGA of main shocks) to investigate the
structures would reach different damage levels under a certain seismic
influence of different aftershock intensities on structural seismic fragi-
intensity. Discrete matrix was suitable for the description of structural
lity. Hence, PGA ratio of main shock-aftershock sequence records in
seismic fragility under some particular amplitudes and continuous
Table 1 were calculated and it was found that probability of occurrence
curves was more all-sided and comprehensive, using for the description
of different PGA ratio of main shock-aftershock sequences showed a
of structural seismic fragility under all amplitudes in full process do-
significant Gaussian distribution in Fig. 5. Then, based on the above
main. In order to observe the influence of aftershocks on seismic fra-
Gaussian distribution, the Monte Carlo method was used to randomly
gility of reticulated domes intuitively, in this paper, continuous curves
generate main shock-aftershock sequence ground motions with dif-
were used to describe the seismic fragility of reticulated domes.
ferent PGA ratio. By adjusting PGA of main shocks, the incremental
The fragility analysis in this paper mainly included two aspects. One
dynamic analysis of reticulated domes was performed under main

10
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 8. Main shock-aftershock sequence fragility curves of DD-R12-S60 with 0–1.2 PGA ratio.

Table 4 95% (a collapse probability of 5%), for SD-R8-S40 and DD-R12-S60

Representative values of DIE with different damage levels. under main shock-aftershock sequences, the maximum amplitudes of
Damage Levels No Damage Slight Medium Significant PGA of main shocks were 0.39 g and 0.54 g respectively. However,
Damage Damage Damage under main shock ground motions with these PGA values, structures
usually did not collapse according to the large amount of numerical
Representative 0.15 0.40 0.60 0.85
examples provided by references [18,24,31], which indicated that
values of DIE
aftershocks had influence on structural collapse probability. For further
investigation of the influence, the detailed discussion was carried out as
shock-aftershock sequence ground motions. The input ground motion follows.
records were shown in Table 3. Finally, the example results were sta- As it was shown in Fig. 5, with a guarantee rate of 95%, the range of
tistically analyzed by Eq. (2) to obtain the main shock-aftershock se- PGA ratio of main shock-aftershock sequences was 0–1.2 g. In view of
quence fragility curves of reticulated domes, which of SD-R8-S40 and this, with 0.2 as incremental dynamic analysis step of PGA ratio, by
DD-R12-S60 were shown in Fig. 6. adjusting PGA of main shocks, incremental dynamic analysis of re-
According to structural design experiences, safety demands usually ticulated domes was performed under main shock-aftershock sequence
required a guarantee rate of 95%. In Fig. 6, with a guarantee rate of ground motions within the range of PGA ratio as 0–1.2 g. According to

Fig. 9. Different possible structural states of SD-R8-S40 with slight damage level (With deformation scale factor as 100).

11
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Table 5 which indicated that aftershocks had a significant impact on seismic

Predominant frequency of simulated aftershock ground motions. fragility of reticulated domes. In addition, it could be also found that
Number of Predominant frequency (Hz) the distance between the fragility curve of main shock-aftershock se-
earthquakes quence with PGA ratio of 1.2 and the fragility curve of main shock
East-west North-south Up–down would increase with the increase of damage level of limit state, which
direction direction direction illustrated that the degree of the above aftershock impact would in-
Main shock 3.491 3.198 0.757 crease with the increase of damage level of limit state. With a guarantee
Aftershock 1 4.447 3.598 1.672 rate of 95% (a collapse probability of 5%), for SD-R8-S40 and DD-R12-
Aftershock 2 3.442 3.756 2.034 S60, the PGA values corresponding to main shocks and main shock-
Aftershock 3 3.694 3.590 1.652 aftershock sequences with 1.2 PGA ratio were 0.66 g, 0.29 g and 0.74 g,
Aftershock 4 4.136 3.298 2.080
0.35 g respectively. It could be seen that the PGA corresponding to main
Aftershock 5 4.758 3.373 2.824
Aftershock 6 3.394 3.822 2.108 shock-aftershock sequences with PGA ratio of 1.2 was about 1/2 of the
Aftershock 7 2.753 2.865 1.686 PGA corresponding to main shocks, which indicated that aftershocks
Aftershock 8 3.319 4.346 1.927 had an obvious influence on structural collapse probability and only
Aftershock 9 4.463 3.881 1.788
considering main shocks could not guarantee the safety of reticulated
Aftershock 10 4.501 4.405 0.671
Aftershock 11 4.539 4.129 1.248 domes under main shock-aftershock sequences.
Aftershock 12 4.166 4.274 2.301 In this section, the above obtained fragility curves could be used to
Aftershock 13 4.367 3.004 1.487 predict seismic performance of reticulated domes under multiple
Aftershock 14 3.944 3.301 1.889 earthquakes, and guide seismic risk assessment and practical seismic
Aftershock 15 4.803 3.669 0.869
design of reticulated domes considering multiple earthquake influence.

5577 pairs of example results, the main shock-aftershock sequence

4.2. Fragility analysis under aftershocks
fragility curves of SD-R8-S40 and DD-R12-S60 were shown in Figs. 7
and 8 respectively.
Different with Section 4.1, the research content of this section was
Through the observation of fragility curves in Figs. 7 and 8, it was
concerned with the failure probability of reticulated domes under
found that probability of exceedance with the same PGA would increase
aftershocks, which have damaged under main shocks. Firstly, the in-
with the increase of PGA ratio of main shock-aftershock sequences,
cremental dynamic analysis of reticulated domes under main shocks

Fig. 10. Aftershock fragility curves of SD-R8-S40.

12
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Fig. 11. Aftershock fragility curves of DD-R12-S60.

Fig. 12. Aftershock collapse fragility curves of SD-R8-S40 and DD-R12-S60.

was performed to make structures reach different damage levels before Therefore, it was necessary to input main shock ground motion records
aftershocks. Table 2 gave the ranges of damage factor DIE corre- with significant difference in characteristic parameters of ground mo-
sponding to each damage level of reticulated domes. In order to facil- tions for the construction of reticulated domes with different damage
itate the construction of aftershock fragility curves, the damage factor states of each damage level. Considering the computation efficiency,
values of each damage level were unified and the median values of each with random selection method, the input ground motion records in this
damage level were taken as representative values, shown in Table 4. In section used 15 main shock ground motion records from Table 3, which
particular, the reticulated domes with the same damage level under could meet the requirements of the number of ground motion records
main shocks could have different damage states probably (See Fig. 9). [30].

13
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

Then, for considering the influence of the difference of aftershock aftershock ground motion intensity measure for the research of
ground motions on seismic analysis results, it could not meet the re- multiple earthquake influence on reticulated domes.
quirement that only one aftershock ground motion record contained in (2) With incremental dynamic analysis method, fragility analysis of
the collected actual main shock-aftershock sequence ground motion reticulated domes was performed under main shock-aftershock se-
record was input. From this point, in this section, an artificial after- quence ground motions and aftershock ground motions respec-
shock ground motion method was used to synthesize 15 aftershock tively, and main shock-aftershock sequence fragility curves and
ground motion records corresponding to the each main shock ground aftershock fragility curves were constructed. In addition, aftershock
motion record, totaling 225 aftershock ground motion records. The collapse probability curves of reticulated domes with different da-
above artificial aftershock ground motion method was constructed with mage levels under main shocks were obtained and the application
the trigonometric series method based on the correlations of ampli- scope of traditional fragility analysis was expanded.
tudes, frequency amplitude spectra, durations and energies between (3) The fragility analysis results indicated that aftershocks had an ob-
main shocks and aftershocks [32]. As an example, Table 5 showed the vious influence on fragility analysis of reticulated domes and the
predominant frequency of 15 simulated aftershock ground motions influence degree could increase with the increase of the damage
corresponding to the main shock ground motion recorded by Elizabeth levels of limit states. The aftershock collapse probability of re-
Lake Station in the 1994 Northridge earthquake, with obvious differ- ticulated domes could also increase with the increase of the struc-
ence about frequency components. Finally, the incremental dynamic tural initial damage levels. Especially, with the initial damage level
analysis of the above-mentioned reticulated domes with different da- as significant damage, the aftershock collapse probability could
mage levels under main shocks was performed under aftershocks. Ac- increase obviously. When the PGA of aftershocks was less than
cording to 11,020 pairs of example results, the data were statistically 0.2 g, if the initial damage level was medium damage or below for
analyzed by Eq. (2) to obtain the aftershock fragility curves of re- the examples in this paper. Therefore, in practical seismic design,
ticulated domes with different damage levels under main shocks, which the damage level of reticulated domes before aftershocks should be
of SD-R8-S40 and DD-R12-S60 were shown in Figs. 10 and 11 respec- guaranteed to be medium damage or below, otherwise, reticulated
tively. The aftershock fragility curves could provide reference for post- domes should be banned from further use immediately under
earthquake seismic performance assessment and structural seismic re- aftershocks.
habilitation of reticulated domes, and provide protection for the con-
tinued use of reticulated domes after earthquakes. CRediT authorship contribution statement
According to the above results of aftershock fragility results, the
probability of exceedance of SD-R8-S40 and DD-R12-S60 with different Yingnan Zhang: Conceptualization, Methodology, Software,
initial damage levels reaching each damage level of limit states under Formal analysis, Data curation, Writing - original draft, Writing - review
aftershocks was obtained. However, in practical seismic design, en- & editing. Xudong Zhi: Conceptualization, Methodology, Writing -
gineers tended to pay more attention to whether the damaged re- original draft, Writing - review & editing, Project administration. Feng
ticulated domes under main shocks would collapse under aftershocks. Fan: Conceptualization, Resources, Supervision, Funding acquisition.
Fig. 12 gave the aftershock collapse fragility curves of SD-R8-S40 and
DD-R12-S60 with different initial damage levels, which indicated that Declaration of Competing Interest
the aftershock collapse probability would increase with the increase of
initial damage levels of reticulated domes. In particular, compared with The authors declare that they have no known competing financial
the other three initial damage levels, when the initial damage level of interests or personal relationships that could have appeared to influ-
reticulated domes was significant damage, the aftershock collapse ence the work reported in this paper.
probability was obviously improved. Furthermore, for SD-R8-S40 and
DD-R12-S60, when the PGA of aftershocks was less than 0.2 g, if the Acknowledgement
initial damage level of reticulated domes was medium damage or
below, aftershocks would not cause structures collapse. This paper is supported by the National Nature Science Foundation
In summary, for the reticulated domes in this section, SD-R8-S40 of China (No. 51778183) and the National Science Foundation for
and DD-R12-S60, in practical seismic design, it should be ensured that Distinguished Young Scholars of China (No. 51525802). These supports
the damage level of reticulated domes under main shocks was medium are extremely appreciated.
damage or below, so as to guarantee the safety of structures under
aftershocks to the greatest extent. If the damage level of reticulated References
domes was significant damage before aftershocks, it should be banned
from further use immediately to avoid unnecessary casualties and [1] Kisslinger C. Aftershocks and fault-zone properties. Adv Geophys 1996;38(38):1–36.
property losses under aftershocks. [2] Bian GB. The statistical property of mainshock-aftershock sequences of Wenchuan
earthquake and its impact on the structures. Dalian: Dalian University of
Technology; 2012. [in Chinese].
5. Conclusions [3] Li Q, Ellingwood BR. Performance evaluation and damage assessment of steel frame
buildings under main shock-aftershock earthquake sequence. Earthquake Eng Struct
Dyn 2007;36:405–27.
Based on incremental dynamic analysis method, taking single-layer [4] Ruiz-Garcia J, Marin VM, Teran-Gilmore A. Effect of seismic sequences in reinforced
and double-layer reticulated domes as research objects, this paper concrete frame buildings located in soft-soil sites. Soil Dynamics and Earthquake
analyzed a large number of example results in statistics and performed Engineering 2014;63:56–68.
[5] Jeon JS, DesRoches R, Lowes LN, Brilakis I. Framework of aftershock fragility as-
the fragility analysis of reticulated domes under multiple earthquakes,
sessment-case studies: older California reinforced concrete building frames.
for the investigation of the aftershock influence on the probabilities of Earthquake Eng Struct Dyn 2015;44:2617–36.
exceedance of reticulated domes with different initial damage levels. [6] Gao GY. Research of seismic probabilistic risk assessment for single-layer reticulated
domes. Harbin: Harbin Institute of Technology; 2012. [in Chinese].
The main conclusions were as follows:
[7] Cornell CA, Jalayer F, Hamburger RO, Foutch DA. probabilistic basis for 2000 SAC
federal emergency management agency steel moment frame guidelines. J Struct Eng
(1) With dynamic time history analysis method in full process domain 2002;128(4):526–33.
for reticulated domes, a large number of example results were dealt [8] Ellingwood BR, Celik OC, Kinali K. Fragility assessment of building structural sys-
tems in Mid-America. Earthquake Eng Struct Dyn 2010;36(13):1935–52.
with by log-linear regression to compare the degrees of correlation [9] D'Aragona MG, Polese M, Elwood KJ, Shoraka MB. Aftershock collapse fragility
and efficiency of 11 common ground motion intensity measures curves for non-ductile RC buildings: a scenario-based assessment. Earthquake Eng
synthetically, and PGA was selected as the most applicable Struct Dyn 2017;46:2083–102.

14
Y. Zhang, et al. Engineering Structures 211 (2020) 110450

[10] Abdelnaby AE. Fragility curves for RC frames subjected to tohoku mainshock- Architecture Industry Press 2010 [in Chinese].
aftershocks sequences. J Earthquake Eng 2018;22(5):902–20. [22] Ministry of Housing and Urban-Rural Development of the People’s Republic of
[11] Pang YT, Wu L. Seismic fragility analysis of multispan reinforced concrete bridges China. Technical specification for space frame structures (JGJ 7-2010). Beijing:
using mainshock-aftershock sequences. Mathematical Problems in Engineering China Architecture Industry Press 2011 [in Chinese].
2018;2018:1–12. [23] Yang FY, Zhi XD, Fan F. Effect of complex damping on seismic responses of a re-
[12] Silwal B, Ozbulut OE. Aftershock fragility assessment of steel moment frames with ticulated dome and shaking table test validation. Thin-walled Structures
self-centering dampers. Eng Struct 2018;168:12–22. 2019;134:407–18.
[13] Park SW, Park HS, Oh BK, Choi SW. Fragility assessment model of building struc- [24] Zhang TM. Research on rapid assessment of seismic damage for spatial latticed
tures using characteristics of artificial aftershock motions. Comput-Aided Civ structures. Harbin: Harbin Institute of Technology; 2017. [in Chinese].
Infrastruct Eng 2018;33:691–708. [25] Zhong J, Zhi XD, Fan F. A dominant vibration mode-based scalar ground motion
[14] Nie GB, Zhi XD, Fan F, Dai JW. Failure pattern of single-layer reticulated dome intensity measure for single-layer reticulated domes. Earthquake and Structures
under severe earthquake and its shaking table test. International Journal of Steel 2016;11(2):245–64.
Structures 2014;14(2):399–409. [26] Zhong J, Zhang JP, Zhi XD, Fan F. Identification of dominant modes of single-layer
[15] Fan F, Sun MH, Zhi XD. Simplified design method and seismic performance of space reticulated shells under seismic excitations. Thin-walled Structures
trusses with consideration of the influence of the stiffness of their lower supporting 2018;127:676–87.
columns. Earthquake Engineering and Engineering Vibration 2016;15(2):401–9. [27] Campbell KW, Bozorgnia Y. A comparison of ground motion prediction equations
[16] Zhong J, Zhi XD, Fan F. Sensitivity of seismic response and fragility to parameter for arias intensity and cumulative absolute velocity developed using a consistent
uncertainty of single-layer reticulated domes. International Journal of Steel database and functional form. Earthquake Spectra 2012;28(3):931–41.
Structures 2018;18(5):1607–16. [28] Zhong J. Probabilistic seismic fragility analysis of reticulated domes. Harbin:
[17] Ma YY, Ma HH, Fan F, Yu ZW. Experimental and theoretical analysis on static be- Harbin Institute of Technology; 2016. [in Chinese].
havior of bolt-column joint under in-plane direction bending in single-layer re- [29] Giovenale P, Cornell AC, Esteva L. Comparing the adequacy of alternative ground
ticulate shells. Thin-walled Structures 2019;135:472–85. motion intensity measures for the estimation of structural responses. Earthquake
[18] Zhi XD, Zhang YN, Fan F, Shen SZ. Responses of single-layer reticulated domes Eng Struct Dyn 2004;33(8):951–79.
subject to multiple earthquakes. Engineering Mechanics 2018;35(9):107–13. [in [30] Nie GB. Failure mechanism and performance evaluation of reticulated shells under
Chinese]. severe seismic motion based on damage cumulative constitutive model. Harbin:
[19] San BB, Waisman H, Harari I. Analytical and numerical shape optimization of a Harbin Institute of Technology; 2012. [in Chinese].
class of structures under mass constraints and self-weight. J Eng Mech 2020;146(1). [31] Fan F, Zhi XD, Shen SZ. Failure mechanism of reticulated shells under earthquake.
[20] Ministry of Housing and Urban-Rural Development of the People’s Republic of Beijing: Science Press; 2014. [in Chinese].
China. Load code for the design of building structures (GB 50009-2012). Beijing: [32] Zhang YN, Zhi XD, Fan F. Study of the correlations between main shocks and
China Architecture Industry Press 2012 [in Chinese]. aftershocks and aftershock synthesis method. Earthquake Engineering and
[21] Ministry of Housing and Urban-Rural Development of the People’s Republic of Engineering Vibration 2019;18(4):759–75.
China. Code for Seismic Design of Buildings (GB 50011-2010). Beijing: China

15