Vol. 21, No.

2 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION April, 2022

Earthq Eng & Eng Vib (2022) 21: 341-358 DOI: https://doi.org/10.1007/s11803-022-2089-0

Probability-based analytical model for predicting the post-earthquake

residual deformation of SDOF systems
Zhang Qin1, 2†, Gong Susu1‡, Gong Jinxin3§, Zhang Guanhua2* and Xi Guangheng2*

1. College of Civil and Transportation Engineering, Hohai University, Nanjing 210098, China
2. Highway Maintenance Technology Research and Development Center of Liaoning Provincial Communication
Planning & Design Insitute Co. Ltd., Shenyang 110111, China
3. Department of Civil Engineering, Dalian University of Technology, Dalian 116014, China

Abstract: A probability-based analytical model for predicting the seismic residual deformation of bilinear single-degree-
of-freedom (SDOF) systems with a kinematic/Takeda hysteretic model is proposed based on a statistical analysis of the
nonlinear time history response, and the proposed model explicitly incorporates the influence of record-to-record variability.
In addition, the influence of primary parameters such as the natural vibration period, relative yield force coefficient, stiffness
ratio and peak ground acceleration (PGA) on the seismic residual/maximum deformation ratio (dR/dm) are investigated. The
results show that significant dispersion of the dR/dm ratio is observed for SDOF systems under different seismic ground motion
records, and the dispersion degree is influenced by the model parameters and record-to-record variability. The statistical
distribution of the dR/dm results of SDOF systems can be described by a lognormal distribution. Finally, a case study for
seismic residual deformation and reparability assessment of the bridge structure designed with a single pier is carried out to
illustrate the detailed analytical procedure of the probability-based analytical model proposed in this study.

Keywords: residual deformation; probability model; repairability assessment; single-degree-of-freedom system;

nonlinear seismic response; record-to-record variability

1 Introduction determining the collapse-resisting capacity of structures.

The indexes of the collapse resistance capacity and
Post-earthquake reconnaissance indicates that repairability both need to be emphasized in newly
seismic losses are triggered not only by building collapse proposed performance-based seismic design philosophy
but also by structural rehabilitation and rebuilding after of structures; that is, both the maximum deformation and
an earthquake, and the cost for repairing and rebuilding residual deformation play key roles in determining the
damaged structures could be enormous (Hsu and Fu, properties of structures subjected to seismic excitation
2004; Decanini et al., 2004; Hashimoto et al., 2005; (Lee et al., 2010; Hatzigeorgiou et al., 2011; Saidi et al.,
Zhang et al., 2017). Therefore, there are some limitations 2012). It is well known that a number of damaged RC
of the traditional performance assessment method based columns and structures have to be demolished due to the
on the maximum deformation, which is the key index in large residual deformations that arise in the aftermath
of many earthquake events, such as the Michoacan
earthquake in 1985 (Rosenblueth and Meli, 1986), the
Correspondence to: Zhang Qin, College of Civil and Transportation Kobe earthquake in 1995 (Fujino et al., 2005) and the
Engineering, Hohai University, Nanjing 210098, China Wenchuan earthquake in 2008 (Han et al., 2013; Zhang
Tel: +86-13913899030 et al., 2017). Therefore, the limitations of traditional
E-mail: qinzhang8190@gmail.com
performance assessment methods based on the maximum
†
Associate Professor (PhD); ‡Postgraduate Student; §Professor;
deformation must be addressed, and a more reasonable
*
Professorate Senior Engineer
Supported by: Natural Science Foundations of China under
method based on both the maximum deformation and
Grant Nos. 51508154, 51978125 and 51678104, the Natural residual deformation should be provided.
Science Foundation of Jiangsu Province under Grant To evaluate the residual deformation, extensive
No. BK20211206, the Fundamental Research Funds for the research has been conducted from a practical perspective.
Central Universities under Grant No. B210202033, China MacRae et al. (1997) and Kawashima et al. (1998)
Postdoctoral Science Foundation under Grant No. studied the post-earthquake residual deformation of
2020M670787, and the Priority Academic Program bilinear single-degree-of-freedom (SDOF) systems
Development of Jiangsu Higher Education Institutions with given displacement ductility demands based on the
Received May 1, 2020; Accepted April 6, 2021 results of nonlinear time history analysis. They defined
342 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

the residual deformation response spectrum (rR) as the the probabilistic distribution of the ratio of residual and
ratio of the residual deformation to maximum possible peak displacements of idealized SDOF systems with
residual deformation with variations in a stiffness ratio non-degrading behavior and proposed a probabilistic
to predict the residual deformations of bilinear SDOF model for predicting the residual deformation of the
systems. In their studies, the trend of residual deformation systems.
with variation in natural vibration period was not clearly From the literature review, note that residual
identified, and the influence of seismic level, record- deformation is considered an important design indicator
to-record variability and maximum deformation (i.e., for restorability and is of great concern to many seismic
ductility ratio) were neglected. However, Ruiz-Garcia researchers and engineers. However, the existing
et al. (2006) showed that the residual deformation of empirical models of residual deformation primarily
a SDOF system is not only affected by the stiffness depend on deterministic models established based on
ratio but also strongly dependent on the seismic levels, the mean statistical results of the nonlinear time history
record-to-record variability, natural vibration period in response of SDOF systems, and the influence on the
the short period region (e.g., T  0.5 s), lateral strength uncertainty of record-to-record variability and structural
ratios and types of hysteretic behavior. Guerrero et al. model parameters are usually not considered. Although
(2017) studied the residual displacement demands of only a few probabilistic models of residual deformations
SDOF systems subjected to earthquake ground motions of SDOF systems have been proposed in previous studies,
recorded at soft soil sites of Mexico City, and also the existing models are mainly aimed at systems with
found that the significant parameters affecting residual non-degrading hysteretic behaviors. Obviously, systems
displacements are stiffness ratio, hysteretic response, with degrading hysteretic behavior are more suitable
ductility level and so on. Their research also highlighted for characterization of RC structures subjected to cyclic
that the dispersion is very high in predicting residual loading, and the residual deformations of these systems
displacement demands for SDOF systems subjected to should also be studied. In addition, the existing models
earthquake ground motions recorded in soft soil sites and of residual deformation are generally oriented to systems
in firm soil sites. In addition, Pampanin et al. (2003) and with a constant-ductility demand (i.e., the displacement
Christopoulos et al. (2003, 2004) studied the residual ductility demand of the system is known in advance), but
deformation of SDOF and multiple-degree-of-freedom in fact the displacement ductility demand of the system,
(MDOF) systems through time history analyses and especially for existing structures, is generally unknown,
indicated that the hysteretic characteristics, post-yielding and an appropriate residual deformation model is needed
stiffness, intensity of the seismic input, and maximum for systems with initial properties (e.g., the period, post-
ductility greatly influenced the residual deformations. yielding stiffness ratio, relative yield force coefficient
Considering the influence of the main parameters and so on) that are known in advance. Therefore, the
mentioned above, some different empirical models for main purpose of this study is to discuss the influence of
predicting the residual deformation response of SDOF record-to-record variability and model parameters on the
systems have been investigated by several researchers, residual deformation of SDOF systems with degrading
including Gong et al. (2011), Hao et al. (2013) and Hu and non-degrading hysteretic behavior and with initial
et al. (2015). Although these models were established properties known in advance. A probability-based model
for simplicity and practical application, the uncertainty incorporating the uncertainty of seismic input records and
in estimating residual deformations induced by record- structural model parameters for estimating the residual
to-record variability and structural parameters are deformation is proposed based on the statistical analysis
ignored, and the prediction results may be inconsistent of nonlinear time history results of SDOF systems.
with actual situations. In view of this, Ruiz-Garcia and Meanwhile, the evaluation of residual deformation
Miranda (2010) proposed a probabilistic procedure and repairability for a bridge structure designed with a
for computing residual deformation demand for multi- single RC pier is taken as an example to illustrate the
story regular frame structures, which considered the application of the proposed model in the performance-
uncertainty of record-to-record variability and ground based seismic assessment methodologies. Note that this
motion intensity. Similarly, a post-earthquake damage study ignores the soil-structure interaction (SSI) effects
assessment model of RC structures considering the and P‒∆ effects, and the results given herein should be
residual deformation attained during the earthquake interpreted from this perspective.
based on Bayesian analysis was presented by Yazgan
and Dazio (2011). Recently, studies on the uncertainty
and probabilistic models of residual deformations have 2 Nonlinear time history analysis of a SDOF
received more attention from researchers. Dai et al. system
(2017) employed the probabilistic approach to establish
the relationship between the maximum deformation and 2.1 Analytical model
residual deformation of the buildings to consider the
uncertainties of the model error dispersion and record-to- As shown in Fig. 1, a reinforced concrete bridge
record variability. Guo and Christopoulos (2018) studied pier can be equivalent to a SDOF system, and residual
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 343

displacement (dR) of the structures can be predicted by energy dissipation capacity and no cumulative damage.
nonlinear time history analysis of the SDOF system Therefore, the corresponding unloading stiffness ku is
with the given ground motion records. The structural expressed as the initial stiffness ki (i.e., ku = ki) of the
characteristics of the SDOF system are mainly related system. For the T-model, the hysteretic behavior is
to the design parameters such as the initial stiffness characterized by stiffness degeneration, pinching effects
(ki), the concentrated mass ( m ) and the period ( T ). and an imperfect energy dissipation capacity. Therefore,
Assuming that the concentrated mass ( m ) is equal to a the unloading stiffness (ku) is expressed as ku  ki /  
unit mass, the natural vibration period of the model can by the initial stiffness (ki) and the displacement ductility
be calculated as follows: coefficient (  ) of the system, and the coefficient  
is related to the unloading stiffness and can be taken
as 0.4 (Takeda and Nielsen, 1970). From the above,
m (1)
T  2 the T-model can describe the degradation and pinching
ki behaviors well for RC structures subjected to cyclic
loading, but the K-model is more efficient and simpler
where ki is the initial stiffness of the model and defined to use in the seismic response analysis of structures
as the secant stiffness at the yield point. (Kawashima et al., 1998; Gong et al., 2011). Therefore,
In this study, two typical bilinear hysteretic these two models are adopted in the analysis of residual
models (i.e., the kinematic hysteretic model and the deformations of SDOF systems in this study.
modified Takeda hysteretic model shown in Fig. 2) To illustrate the effects of the hysteretic model on
with different post-yielding stiffnesses were adopted to the nonlinear seismic response of the structures, a case
reasonably reflect the concrete structure characteristics. study of the nonlinear seismic response analysis of
For convenience, these two hysteretic models are SDOF systems with the K-model and T-model under
abbreviated as the K-model and T-model, respectively, in the same ground motion record was conducted by using
the text. For the K-model, the hysteretic behavior of the SAP2000 software. In the case study, the influences of
system is characterized by non-degradation, excellent the system period (T) and post-yielding stiffness ratio (r)

ki  i

dR dR

Fig. 1 Bridge pier and SDOF system

(a) Kinematic hysteretic model (b) Takeda hysteretic model

Fig. 2 Bilinear hysteresis models
344 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

were also considered. The period was taken as 0.6 s or residual/maximum deformation ratio of different SDOF
1.2 s, and the post-yielding stiffness ratio (r) was taken systems with the K-model and T-model.
as 0.05 or 0.1. The El Centro accelerogram listed in In view of the above, the influences of the system
Appendix Table A.1 was selected as the ground motion period (T) and the post-yielding stiffness ratio (r)
input, and the seismic ground motion was normalized on the seismic response of the SDOF systems is also
by the peak ground acceleration (PGA) of 0.25 g. Figure 3 considered in this study. The variation range of the
presents the nonlinear time history analysis results of period ( T ) is 0.2‒3.0 s, and the values are selected at
different SDOF systems with the K-model and T-model an increment of 0.2 s. Eight post-yielding stiffness ratios
under the same normalized seismic ground motion, and (r), 0.0, 0.01, 0.05, 0.1, 0.15, 0.2, 0.25 and 0.3, are
the corresponding time history response curves of these selected for the analysis.
two models are represented by dark lines and light lines,
respectively. From the figure, it is noted that the time 2.2 Selection of earthquake ground motion records
history response curves of the SDOF systems with the
K-model and T-model have significant differences even According to the study conducted by Cheng et al.
though the system parameters and ground motion input (2013), the seismic response of SDOF systems with
are the same. Undoubtedly, there is also a difference bilinear hysteresis models is related not only to the
in the residual deformation (dR) and the maximum system period ( T ) and post-yielding stiffness ratio (r)
deformation (dm) results of the SDOF system determined but also to the relative yield force coefficient (η) and
by the two different hysteresis models, and the degree of peak ground acceleration (αpg). Where, the relative
difference may be related to the period (T) and stiffness yield force coefficient (η) is defined as the ratio of the
ratio (r). In general, the difference of the residual yield load to gravity load (i.e., η = Fy /mg), which is to
deformations with these two models is more obvious reflect the elastic-plastic capacity of the systems, and the
when the stiffness ratio (r) is small and that of the time details can be found in research of Cheng et al. (2013).
history response curve is more obvious when the system To consider the effects of the parameters mentioned, the
period (T) is large. This suggests that the hysteretic relative yield force coefficients (η) are assumed to be 0.1,
properties (or hysteretic models) have different effects 0.2 and 0.3. and the peak ground accelerations (αpg) are
on the time history response and residual deformations assumed to be 0.2 g, 0.4 g and 0.6 g for normalization of
of SDOF systems with different parameters. Therefore, seismic records to simulate the different seismic levels.
it is necessary to consider the influence of the hysteretic Furthermore, a total of 100 strong ground motion
model in residual deformation predictions. For this records of different seismic events and site conditions were
purpose, several statistical analyses are conducted in this selected from the PEER (Pacific Seismic Engineering
study to determine the distribution characteristics of the Research Center) strong earthquake database as the

dm
dm
dR
dR
d (mm)

d (mm)

dR

dR

dm
dm

t (s) t (s)
(a) r=0.05, T=0.6 s (b) r=0.05, T=1.2 s

d
dmm
dR dR
d (mm)

d (mm)

dR
dR

dm
dm

t (s) t (s)
(c) r=0.1, T=0.6 s (d) r=0.1, T=1.2 s
Fig. 3 Time history analyses of SDOF systems with different hysteretic models
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 345

seismic input of nonlinear time history analysis for decreases with the increase of the stiffness ratio (r) of the
considering the influence of record-to-record variability systems. In addition, the influence of the period (T) on
on residual deformations of SDOF systems. According the dispersion of β is also associated with the stiffness
to the previous studies (e.g., Liossatou and Fardis, 2016; ratio (r) and hysteretic model. The dispersion decreases
Guerrero et al., 2017; Guo and Christopoulos, 2017), the with the increase in period (T) when the stiffness ratio (r)
100 ground motions with a large variation in parameters is large, while the dispersion may remain consistent with
such as frequency content, duration, site condition and the increase in period (T) when the stiffness ratio (r) is
pulse-characteristics were selected because these are key small. Additionally, the influence trend of the period (T)
influence parameters for the the residual deformations on the dispersion of β is relatively prominent when the
of the SDOF systems. Of the 100 ground motions, 88 system is based on the K-model.
records that were recorded from magnitude 5.3 to 8.1 Note that the statistical results of the deformation
events at source-to-site distances of less than 20 km, ratio (β) are also affected by the relative yield load
and the remaining records obtained from magnitude factor (η) and the peak ground acceleration (apg). This
5.8 to 7.5 events at the distances ranging from 20 km can be explained by the fact that the seismic residual
to 50 km. Note that this study mainly focus on the deformation may occur in the post-yielding response
influences of randomness and uncertainty of earthquake phase, and the yielding of the structures is mainly related
ground motions on the residual deformations of the to the two parameters mentioned above (i.e., η and apg).
SDOF systems triggered comprehensively by these Therefore, these parameters play an important role in
characteristics mentioned above. So, the influences the residual deformation analysis of the SDOF systems.
of the earthquake ground motions characterized by a Furthermore, according to the variation trends of the
certain variable such as the site condition, duration and mean values of the residual/maximum deformation ratio
so on were not considered specially. The ground motion (β) with the periods (T) for the SDOF systems shown in
records parameters are shown in the appendix Table A1. Fig. 4, the effects of the hysteretic model, stiffness ratio
(r) and period (T) on the seismic response may seem
3 Statistical analysis of the residual more obvious than those affected by the peak ground
deformation of a SDOF system acceleration (apg) and the relative yield force coefficient (η).

3.2 Analytical method and statistical parameters

3.1 Dispersion analysis and parametric study
The detailed procedure of the probabilistic model for
The requirements of modern seismic design predicting the residual deformations of SDOF systems
philosophy indicated that not only the capacity of collapse in this study is as follows: (1) 100 ground motion
resistance but also repairability should be required for records selected from different earthquake events were
designing structures subjected to strong earthquake normalized by the intensity index of the peak ground
excitation. Therefore, the ratio β (= dR/dm) of the residual acceleration apg=0.2 g, 0.4 g, or 0.6 g. Note that the SDOF
deformation to maximum deformation was taken as the systems studied herein are given the initial properties
statistical parameter in this analysis, and the influences (e.g. relative yield force coefficient) in advance, so the
of the model parameters, record-to-record uncertainty, yield states of these structures under different seismic
hysteretic models and seismic intensity levels were also levels need to be concerned, which is vital for the
considered. The variation trend of the deformation ratio study of the residual deformations. The study herein is
(dR/dm) with the increase in period (T) for the different obviously different from previous studies on the residual
bilinear SDOF systems under each of the 100 normalized deformations of SDOF systems that have a known
ground motion records is shown in Fig. 4, and the target ductility demand, which implies that the yield
corresponding mean values of deformation ratio (β) deformation of the system under each seismic ground
with the variation in period (T) are also presented. From motion needs to be determined by iteration according
Fig. 4, the obvious dispersion of the deformation ratio to the given target ductility demand (Kawashima et al.,
(β) is observed for SDOF systems with different ground 1998). Therefore, it is reasonable to conduct amplitude
motion records, although the peak ground acceleration normalization for the ground motions using the peak
is the same, which indicates that the seismic response ground acceleration. (2) The deformation ratios (β) of
results of the systems are also significantly affected by different bilinear SDOF systems under each of the 100
the spectrum characteristics and duration of the ground normalized ground motion records was obtained from
motion records. Certainly, the degree of dispersion is also statistical analyses of the nonlinear time history seismic
related to the model parameters such as the hysterical response. In this study, the determination of the residual
model, stiffness ratio (r) and natural period of vibration deformation of the system by the residual/maximum
(T). The analytical results of β (i.e., dR/dm) of the systems deformation ratio (β) is based on the premise that the
based on the T-model are relatively concentrated two deformations mentioned above are independent
because the degrading hysteretic behavior is considered, of each other. Although the residual deformation is
showing a smaller dispersion than the results of the affected by the maximum deformation, there is no
K-model. The dispersion of analytical results obviously clear correlation between the two, because the system
346 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

T (s) T (s) T (s)

(Ⅰ) apg = 0.4 g, r = 0.0, η = 0.1 (Ⅱ) apg = 0.4 g, r = 0.0, η = 0.2 (Ⅲ) apg = 0.4 g, r = 0.2, η = 0.2

T (s) T (s) T (s)

(Ⅳ) apg = 0.6 g, r = 0.0, η = 0.1 (Ⅴ) apg = 0.6 g, r = 0.0, η = 0.2 (Ⅵ) apg = 0.6 g, r = 0.2, η = 0.2

(a) Based on K-model

T (s) T (s) T (s)

(Ⅰ) apg = 0.4 g, r = 0.0, η = 0.1 (Ⅱ) apg = 0.4 g, r = 0.0, η = 0.2 (Ⅲ) apg = 0.4 g, r = 0.2, η = 0.2

T (s) T (s) T (s)

(Ⅳ) apg = 0.6 g, r = 0.0, η = 0.1 (Ⅴ) apg = 0.6 g, r = 0.0, η = 0.2 (Ⅵ) apg = 0.6 g, r = 0.2, η = 0.2

(b) Based on T-model

Fig. 4 Residual/maximum deformation Ratios of different SDOF systems under 100 seismic motions

unloads from a post-peak random point instead of the ground motion is β = dRi/dmi, and the corresponding
maximum deformation point (see the nonlinear time computational expressions of the mean value (  ),
history analytical results of SDOF systems shown in Fig. standard deviation (   ) and the coefficient of variation
3). Similar results and conclusions have also been drawn (   ) of the deformation ratios are as follows:
in previous studies (e.g., Kawashima et al., 1998; Guo
and Christopoulos, 2018). (3) the statistical parameters 2
(i.e., mean value and coefficient of variation) and 1 Ni
d Ri 1 N i  d Ri 
probability distribution model of (β) were established 
N

i 1 d mi
,    
N i  1 i 1  d mi
 ,

from the statistical analyses of the nonlinear time history
seismic response. (4) the probability distribution model 
  (2)
of the residual deformation (dR) was derived from the 
proposed probability-based model of the deformation
ratio (β). Note that the ratio of residual deformation to To determine the mean values of the deformation
maximum deformation of SDOF systems under a single ratio for the systems with different hysteretic behavior,
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 347

Figs. 5 and 6 show the statistical results of  vs. T for  vs. T is relatively distinct in the short period region
the different bilinear SDOF systems, respectively, based from 0 to 0.6 s and generally becomes stable in the long
on the K-model and T-model with a light line. As shown period region from 0.6 to 3 s. The mean values of the
in Fig. 5, the mean deformation ratios (  ) of the SDOF deformation ratio (  ) decrease as the system stiffness
systems based on the K-model decrease as the period ratio (r) and relative yield force coefficient (η) increases
(T) increases, and the decreasing tendency is associated and increase as the peak ground acceleration (apg)
with parameters such as the stiffness ratio (r), relative increases. Note that the peak ground acceleration (apg)
yield force coefficient (η) and peak acceleration (apg) of has little effect on the mean values of the deformation
the ground motion. Overall, the decreasing tendency of ratio (  ) when the stiffness ratio ( r ) and the relative

T (s) T (s) T (s)

(a) η = 0.1, r = 0.0 (b) η = 0.2, r = 0.0 (c) η = 0.3, r = 0.0

T (s) T (s) T (s)

(d) η = 0.1, r = 0.2 (e) η = 0.2, r = 0.2 (f) η = 0.3, r = 0.2
Fig. 5 Mean values of residual/maximum deformation ratio of SDOF systems based on K-model

T (s) T (s) T (s)

(a) η = 0.1, r = 0.0 (b) η = 0.2, r = 0.0 (c) η = 0.3, r = 0.0

T (s) T (s) T (s)

(d) η = 0.1, r = 0.2 (e) η = 0.2, r = 0.2 (f) η = 0.3, r = 0.2
Fig. 6 Mean values of residual/maximum deformation ratio of SDOF systems based on T-model
348 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

yield force coefficient (  ) are large. Similarly, the For the T-model, Q1 and Q2 can be estimated as
mean value of the residual deformation ratios of SDOF follows:
systems based on the T-model tends to decrease as period
( T ) increases, as shown in Fig. 6. The dispersion of the 0.05
mean values of the residual deformation ratio of SDOF Q1  0.2025apg 2  0.2357apg  
0.1946  r
systems based on the T-model is less than that based 1 (6)
on the K-model because the hysteretic characteristics 0.2218  0.3277  r  2  0.1301
of stiffness degradation of the systems are considered
with the T-model. Compared with the statistical results 2
of the systems with the K-model, the relationship of  Q2  0.9135apg 2  1.5088apg  0.5628r 5 
vs. T of the systems with the T-model is hardly affected 1 (7)
by the peak ground acceleration, especially when the 2.4614 10  2.7461r  1.3161
stiffness ratio and peak ground acceleration are large.
Significantly, some data points of mean values (  ) based
on the K-model and T-model with long period regions The mean values of the deformation ratio of SDOF
( T ) and large relative yield force coefficients (  ) are systems estimated by Eq. (3) are also shown in Figs. 5
not shown in Figs. 5 and 6, which is because the vast and 6, which are described with the dark line. The results
majority of 100 normalization seismic motion inputs of the proposed Eq. (3) are found to correlate well with
with small seismic levels (e.g., PGA = 0.2 g) cannot the statistical data points described with the light line
enable these SDOF systems with given initial properties mark.
(e.g., large relative yield force coefficient) to enter To determine the coefficient of variation (   ) of the
the yielding stage; that is, pi  10% , where pi is the deformation ratio for the systems with different hysteretic
yielding ratio between the effective number ( N i ) and behaviors, Figs. 7 and 8 illustrate the statistical results
the total number (Nt = 100) of ground motion records. of the coefficient of variation of the deformation ratio
In other words, the SDOF systems in this situation can for the different bilinear SDOF systems based on the
be assumed to not yield and remain in an elastic state, K-model and T-model with the light line, respectively.
and the maximum seismic deformation is smaller than As shown in Fig. 8, a linear relationship between the
the corresponding yielding deformation. Therefore, the coefficient of variation (   ) and the period ( T ) of SDOF
residual deformation prediction for SDOF systems in systems is observed for the K-model, which is related to
this situation may be unnecessary. the stiffness ratio ( r ) and relative yield force coefficient
According to the tendency and influences of the (  ); that is, the linear proportionality coefficients for
stiffness ratio, relative yield force coefficient and peak the relationship are influenced by the parameters r and
ground acceleration on the   T relationship for SDOF  of the systems. However, note that the relationship
systems with different parameters, the mean values of the between   and T is barely affected by the peak
deformation ratio of the systems based on the T-model ground acceleration (apg). This indicates that the uniform
and K-model can be consistently expressed based on the linear function can be used to describe the relationship
optimized regression of the statistical results as follows: of   - T of the systems subjected to different seismic
excitation levels (i.e., apg =0.2 g, 0.4 g and 0.6 g). Similar
to Fig. 7, the linear relationship between the coefficient
 T   d R d m  Q1T Q2
 0.2 s  T  3.0 s  (3) of variation (   ) and the period ( T ) of SDOF systems
based on the T-model is shown in Fig. 8. The linear
where Q1 and Q2 are coefficients related to the stiffness relationship is significantly affected by the stiffness ratio
ratio, relative yield force coefficient and peak ground ( r ) and relative yield force coefficient (  ) but not by
acceleration, and can be determined according to the the peak ground acceleration (apg).
statistical results illustrated in Figs. 5 and 6. According to the influence trends of the stiffness ratio
For the K-model, Q1 and Q2 can be estimated as and relative yield force coefficient on the relationships of
follows:   - T for SDOF systems with different parameters, the
coefficient of variation (   ) of the deformation ratio of
0.026 the systems based on the T-model and K-model can be
Q1  0.2386apg 2  0.32apg   consistently expressed based on the optimized fitting of
0.0765  r the statistical results as follows:
1 (4)
0.3519  0.4263  r  2  0.0842
δβ(T) = Q1′ T + Q2′ (0.2 s ≤ T ≤ 3.0 s) (8)

5
Q2  0.0354apg 2  0.2684apg  13.5289r 2  where Q1 and Q2 are coefficients related to the stiffness
1 ratio and relative yield force coefficient and can be
1.2096 3  7.633r  0.2663 (5) determined according to the statistical results illustrated
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 349

T (s) T (s) T (s)

(a) η = 0.1, r = 0.0 (b) η = 0.2, r = 0.0 (c) η = 0.3, r = 0.0

T (s) T (s) T (s)

(d) η = 0.1, r = 2.0 (e) η = 0.2, r = 2.0 (f) η = 0.3, r = 2.0

Fig. 7 Coefficient of variation (δβ) of deformation ratio of the systems based on K-model

T (s) T (s) T (s)

(a) η = 0.1, r = 0.0 (b) η = 0.2, r = 0.0 (c) η = 0.3, r = 0.0

T (s) T (s) T (s)

(d) η = 0.1, r = 2.0 (e) η = 0.2, r = 2.0 (f) η = 0.3, r = 2.0

Fig. 8 The coefficient of variation (δβ) of deformation ratio of the systems based on T-model

in Figs. 7 and 8. For the K-model, Q1 and Q2 can be For the T-model, Q1 and Q2 can be estimated as
estimated as follows: follows:

Q1  0.24486r 3  0.8317r  2.6773 3  Q1  0.9306r 2  0.5306r  2.0701 3 

9
0.7299  0.7895  r 
3
 0.0571 (9) 3
0.7623 2  0.1243  r  5  0.0059
20
(11)
3 1
Q2  0.5259r 2  0.2016 2 
1
0.6981 r  3  0.7022 Q2  0.872r  0.077  1.7005r  0.6088 (12)
(10)
350 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

The coefficients of variation (   ) of the deformation 3.3 Probability distribution

ratios for SDOF systems were estimated by Eqs. (8)‒
(12) and are shown in Figs. 7 and 8, as described with To determine the probability distribution of the
the dark line. The predicted results correlate well with residual/maximum deformation ratio (  ) of the
the statistical data points described with the light line. SDOF systems, typical histograms and corresponding
Certainly, it is noted that the residual deformation ratio lognormal distribution fitting curves of  for the
(  ) is not only affected by the peak ground acceleration SDOF systems based on the K-model and T-model are
(apg), period ( T ), stiffness ratio ( r ), relative yield illustrated in Fig. 9. From the figures, the probability
force coefficient (  ) and hysteresis characteristics, but distribution of  approximatively obeys a lognormal
also by the uncertainty and discreteness of the seismic distribution, which is also consistent with the results of
response of the SDOF systems subjected to different the Kolmogorov-Smirnov (K-S) test. The probability
seismic ground motions. Although the influences of density function and distribution function of  can be
all the parameters mentioned above on the residual written as follows:
deformation ratios are comprehensively considered, the   ln    2 
1
f    exp   
ln 
empirical formulas (i.e., Eq. (3) and Eq. (8)) still have (13)
some unavoidable errors.  2 ln   2 ln2  
 

(Ⅰ) apg = 0.4 g, η =0.1, r = 0, T = 0.6 s (Ⅱ) apg = 0.4 g, η = 0.1, r = 0.1, T = 1.2 s (Ⅲ) apg = 0.4 g, η = 0.2, r = 0.1, T =1.2 s

(Ⅳ) apg = 0.6 g, η =0.1, r = 0.0, T = 0.6 s (Ⅴ) apg = 0.6 g, η = 0.1, r = 0.1, T = 1.2 s (Ⅵ) apg = 0.6 g, η = 0.2, r = 0.1, T =1.2 s

(a) For K-model

(Ⅰ) apg = 0.4 g, η = 0.1,r = 0, T = 0.6 s (Ⅱ) apg = 0.4 g, η = 0.1, r = 0.1, T = 1.2 s (Ⅲ) apg = 0.4 g, η = 0.2, r = 0.1, T =1.2 s

(Ⅳ) apg = 0.6 g, η = 0.1, r = 0, T = 0.6 s (Ⅴ) apg = 0.6 g, η = 0.1, r = 0.1, T = 1.2 s (Ⅵ) apg = 0.6 g, η = 0.2, r = 0.1, T =1.2 s
(b) For T-model
Fig. 9 Histograms and probabilistic distributions of deformation response ratio of SDOF systems
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 351

 ln   ln   predicting the maximum deformation of the systems due

F      (14) to the effects of record-to-record variability (Ruiz-Garcia
 2 2 and Miranda, 2006; Cheng et al., 2013). The statistical
 ln  
relationships between the pushover method and the
where nonlinear time history analysis method in predicting the
maximum deformations of SDOF systems have been
  
 ,  ln   ln 1    
ln   ln  2 established in the literature by Cheng et al. (2013), and
1  2 (15) the probability-based model of the maximum possible
  
deformations of SDOF systems subjected to seismic
loading predicted by the pushover method is shown as
ln  and  ln  are the mean value and standard follows:
deviation of the deformation ratio (  ) after taking the
logarithm, respectively.  and   are the mean value
and coefficient of variation of  , respectively, and can d tm  exp  ln    1  P  ln   d pm (18)
be estimated according to Eqs. (3) and (8). In Fig. 9, f 
is the frequency of the samples of  falling into each in which
interval of the histogram.

ln   ln( ) (19)
1   2
4 Establishment of an analytical model for the
residual deformation of a SDOF system

The seismic deformation ratio (  ) of SDOF  ln   ln(1   2 ) (20)

systems can be assumed to be a random variable
because of the uncertainty of ground motion records,
and the possibility of  exceeding a certain value (e.g., where dtm is the maximum deformation predicted by
dR/dm) can be expressed by the probability P . Since  the nonlinear time history analysis method; dpm is the
obeys a lognormal distribution, P can be written as the maximum deformation predicted by the pushover
following expression: analysis method; P is the given exceeding probability;
ln  and  ln  are, respectively, the mean value and
standard deviation of the maximum deformation ratio 
 dR   ln  d R d m   ln   (i.e., the ratio of the maximum deformations predicted
P  P          (16) by the nonlinear time history analysis and pushover
 dm    ln  
 analysis) after taking the logarithm; and  and  are
the mean and coefficient of variation of  . The detailed
where    is the standard normal distribution function. calculation formulas can be found in Cheng et al. (2013).
Assuming that the probability ( P ) is known, the
seismic residual deformation of SDOF systems can be
derived according to Eq. (17) as follows: 5 Case study for residual deformation analysis
and repairability evaluation of SDOF systems

d R  exp  ln    1  P  ln   d m (17) To illustrate the method proposed herein for

  evaluating the residual deformation and repairability of
SDOF systems, a case study of a seismic repairability
where  1   is the inverse of the standard normal assessment for a bridge structure with a single RC pier
distribution function. Note that the rationality of Eq. (17) illustrated in Fig. 1 was conducted. The column pier is
is only based on the independent relationship between designed with the following properties: bridge height
the residual deformation and the maximum deformation; H is 6000 mm; concentrated mass m of the bridge
the corresponding discussion on the correlation between structure is 3.8×105 kg; cross-sectional dimensions are
the two deformations mentioned above have been made 1500 mm×1500 mm; twenty longitudinal reinforcing
in previous studies (e.g., Kawashima et al., 1998; Guo bars with a diameter of 32 mm, yielding strength of 360
and Christopoulos, 2018) and it has been concluded that MPa and ultimate strength of 500 MPa are distributed
they can be regarded as independent of each other. uniformly on four sides of the pier; transverse reinforcing
According to the proposed model (i.e., Eq. (17)), the bars with a diameter of 10 mm, yielding strength of 300
maximum deformation (dm) should be known in advance MPa and spacing of 100 mm are arranged in the pier.
to predict the seismic residual deformation (dR) of SDOF It was assumed that the bridge is located in a category
systems. dm can be approximately predicted by the Ⅲ site and corresponds to the second design seismic
pushover method from a practical perspective. However, group according to the Chinese Seismic Design Code
there are some differences between the pushover method (GB 50011-2010, 2010). To illustrate the differences in
and the nonlinear time history analysis method in seismic residual deformations of the bridge pier with
352 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

different probability levels, the residual deformations deformation requirements under a given seismic intensity
with exceeding probabilities of 16%, 50% and 84% level and that the residual deformation and repairability
were evaluated by the model proposed in this study, and of a bridge should be considered in terms of a nonfatal
a preliminary analysis of the repairability of the bridge collapse.
pier was carried out. According to the analytical results mentioned above,
According to Eq. (17), the maximum nonlinear the related parameters of the bridge pier for the K-model
deformation (dpm) of the bridge pier should be estimated and T-model can be conveniently determined, such as
first by the pushover method. In the pushover procedure, the natural period ( T ), relative yield force coefficient
the capacity curve (i.e., monotonic load-displacement (η), stiffness ratio ( r ) and ground peak acceleration
curve) of the bridge can be obtained by sectional analysis (apg). Then, combining Eqs. (3), (8), (13) and (14), the
technology and the plastic hinge method (Zhang et al., mean value ( ln  ) and standard deviation (  ln  ) of the
2011). The nonlinear demand spectrum of the bridge logarithm of the residual deformation ratio (  ) can be
based on the T-model and K-model is obtained from obtained. Additionally, the seismic residual deformation
a 5% design spectrum, which is recommended by the (dR) and the corresponding residual deformation drift
Chinese Seismic Design Code (GB 50011-2010), ratio (θR) of the bridge pier with exceeding probabilities
reduced by the equivalent viscous damping considering ( P ) of 16%, 50% and 84% can be obtained according
the hysteretic energy dissipation. The equivalent viscous to Eq. (17), as listed in Table 2.
damping ratios (ζeq) for the T-model and K-model can be From Table 2, it can be seen that the residual
evaluated as follows: deformation drift ratios (  R ) based on the K-model
and T-model are all less than the threshold limit value
(i.e., [θR] = 1.75%), which is recommended by the
2 Fy d m  d y Fd Design Specifications of Highway Bridge (Part V:
 eq  0.05 
 Fd d m , for K-model (21)

1 F d  
0.6

 eq  0.05  1  d  m  
  Fy  d y   , for T-model (22)
 

The maximal displacement responses (dpm) of the

bridge are 146.74 mm and 170.01 mm for the K-model ζeq
and T-model, respectively, according to the capacity
ζeq
spectrum method, and the corresponding equivalent
viscous damping ratios are 46.44% and 22.5%, as shown
in Fig. 10. In this figure, point Y is the pier column
yield point, which corresponds to the first yielding of ζeq
the reinforcing bars in tension.
Assuming that the exceeding probability (Pα) is 50%, ( )
the maximum deformation (dtm) and the maximum drift Fig. 10 Seismic performance points of bridge pier
ratio (  m ) of the top of the bridge pier can be obtained for
the K-model and T-model by Eq. (17), and the predicted
results are listed in Table 1. It can be seen from the table
that the maximum drift ratio (  m ) for the K-model and Table 1 Predicted results of the dtm and θm of the bridge
T-model are all less than the threshold limit value (i.e., Hysteretic models dtm (mm) θm
[  m ] = 1/30) recommended in the Chinese Seismic K-model 180.70 1/33
Design Code (GB 50011-2010). This suggests that the
T-model 172.59 1/35
designed bridge pier meets the maximum nonlinear

Table 2 Predicted results of related parameters of the bridge

Predicted dR (mm) (θR)

T (s) η r apg (g) μlnβ σlnβ
results Pβ=16% Pβ=50% Pβ=84%
K-model 0.85 0.21 0.037 0.41 ‒2.017 0.672 46.90 24.04 12.32
(0.78%) (0.40%) (0.21%)
T- model 0.85 0.21 0.035 0.41 ‒2.379 0.620 29.63 15.99 8.64
(0.49%) (0.27%) (0.14%)
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 353

Seismic Design) (2002) of Japan. This indicates that the reasonable and suitable, which is more consistent with
reparability capacity of the designed bridge in the case the performed-based design philosophy. In addition, a
study is well under the given probability levels. Note that simplified residual deformation model was also provided
the predicted seismic residual deformation decreases in the Seismic Performance Assessment of Buildings
and the corresponding repairability probability of the (FEMA P58, 2012), and the residual deformations of the
bridge under the given seismic excitation increases systems depend on the repairable probability, maximum
with the increase in the given exceeding probability. displacement and yielding displacement. According to the
Furthermore, the analytical results based on the model suggested in FEMA P58, the residual deformation
K-model with nondegrading hysteretic behavior are (dR) of the system is equal to the maximum deformation
larger than those based on the T-model with degrading (dm) minus three times the yielding displacement (3dy)
hysteretic behavior; that is, the analytical results of the when the displacement ductility coefficient is larger
repairability for the structures with the K-model are than 4.0. Hence, the residual deformation of the bridge
more conservative. described in the case study is 69.14 mm, which is a
To further illustrate the reasonability of the proposed deterministic value and is approximately equal to the
model, the K-model based residual deformation model predicted result of Eq. (17) based on the K-model when
(i.e., Eq. (17) with K-model) is compared with the the given transcendence probability P(  ) is 6%. This
method of residual deformation spectrum cR presented comparison also suggests that the proposed model (i.e.,
by Kawashima and MacRae et al. (1998). Their model Eq. (17)) is more reasonable and robust, and is more
was established based on the bilinear hysteretic model suitable for simplified performance assessment and
with non-degrading hysteretic behaviors (i.e., K-model), design for considering the seismic repairability.
and was also adopted by the Design Specifications of
Highway Bridge (Part V: Seismic Design) (2002) of
Japan. The equation of the model is as follow: 6 Conclusions

The elastoplastic seismic responses of different

d R  cR    1 r  1 d y for r (   1)  1 (23)
SDOF systems based on kinematic/Takeda hysteretic
models under a series of normalized ground motion
where cR is defined as the ratio of residual deformation records were statistically analyzed, and the effects of the
to theoretical maximum residual deformation, which uncertainty of the model parameters and seismic records
is mainly related to the stiffness ratio (r) and can be on the residual deformation of SDOF systems were
determined by the residual deformation response studied. The conclusions are as follows.
spectrum suggested by the Japanese code; and  is (1) There is obvious dispersion of the analytical
the displacement ductility coefficient. Substituting the results of the ratios of seismic residual deformation to
relative parameters (i.e., r, dy, dtm and cR) of the bridge maximum deformation (dR/dm) of SDOF systems under
structure described in the case study into Eq. (23), the different seismic ground motion records. The dispersion
predicted residual deformation dR is 49.75 mm. This degree of the residual deformation ratio (dR/dm) is mainly
result is approximately equal to the predicted result influenced by the stiffness ratio, hysteretic model, natural
of Eq. (17) based on the K-model when the given vibration period, relative yield load coefficient and peak
transcendence probability P    is 16%, as listed in ground acceleration.
Table 2. This means that the prediction of residual (2) The statistical distribution of the residual
deformation of SDOF systems by using the proposed deformation ratio (dR/dm) of SDOF systems based on the
model (i.e., Eq. (17)) has a high reliability. However, bilinear K/T hysteretic models can be described by the
note that the residual displacement spectrum cR given lognormal distribution function. A power function and
in the Japanese code is primarily determined based on linear function can be adopted for characterizing the
the mean values of the nonlinear time history analysis relationship between the mean value and the structural
results which ignore the effects of dispersion and period ( T ) and that between the coefficient of variation
uncertainty of the results. Therefore, the upper and and T , respectively.
lower limits of the residual displacement spectrum cR (3) The proposed model based on a lognormal
with plus or minus one time standard deviation from the distribution function can be used to predict the
mean values were also suggested in the Japanese code residual deformations of SDOF systems under the
to consider the prediction dispersion. In this case, the given probability requirements and seismic levels. The
range of residual displacement prediction of the bridge analytical results of residual deformation based on the
described in the case study is 12.44 mm ‒ 96.74 mm K-model is more conservative than those based on the
according to the Japanese code. The wide margins of T-model.
the results indicate that the residual displacements of (4) The proposed probabilistic model of residual
the structures should be predicted considering the effect deformation of SDOF systems in this study is focused
of dispersion, and the residual displacement model on the known maximum deformation, which can be
proposed from the viewpoint of probability may be more estimated by the pushover method or other methods
354 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

reported in the literature. Additionally, the proposed GB 50011-2010 (2010), Code for Seismic Design of
procedure can predict the residual deformation when Buildings, The Ministry of Housing and Urban-Rural
the displacement ductility demand of a SDOF system is Construction of the People′s Republic of China, China
unknown. Architecture & Building Press, Beijing, China. (in
Chinese)
Acknowledgment Gong Jinxing, Cheng Lin and Zhang Qin (2011),
“Statistical Relationship Between Results of Static
This study was funded by the Natural Science Nonlinear Analysis and Elasto-Plastic Time-History and
Foundations of China (Grant Nos. 51508154, 51978125 Calculation of Residual Deformation for Structures,”
and 51678104), the Natural Science Foundation of Jiangsu Journal of Building Structures, 32(12): 224‒234. (in
Province (BK20211206), the Fundamental Research Chinese) https://doi.org/10.14006/j.jzjgxb.2011.12.026
Funds for the Central Universities (B210202033), China Guerrero H, Ruiz-García J and Ji T (2017), “Residual
Postdoctoral Science Foundation (2020M670787), and Displacement Demands of Conventional and Dual
the Priority Academic Program Development of Jiangsu Oscillators Subjected to Earthquake Ground Motions
Higher Education Institutions. The authors wish to Characteristic of the Soft Soils of Mexico City,” Soil
express their gratitude for this support. Dynamics and Earthquake Engineering, 98: 206‒221.
https://doi.org/10.1016/j.soildyn.2017.04.014
References Guo JWW and Christopoulos C (2018), “A Probabilistic
Framework for Estimating the Residual Drift of Idealized
Cheng Lin, Gong Jinxin and Zhang Qin (2013), SDOF Systems of Non-Degrading Conventional
“Investigation on the Accuracy of Pushover Procedures and Damped Structures,” Earthquake Engineering
for SDOF,” Advances in Structural Engineering, & Structural Dynamics, 47(2): 479‒496. https://doi.
16(11): 1957‒1971. https://doi.org/10.1260/1369- org/10.1002/eqe.2975
4332.16.11.1957 Han Qiang, Qin Lushan and Wang Piguang (2013),
Christopoulos C, Pampanin S and Priestley MJN “Seismic Failure of Typical Curved RC Bridges
(2003), “Performance-Based Seismic Response of in Wenchuan Earthquake,” International Efforts
Frame Structures Including Residual Deformations. in Lifeline Earthquake Engineering-Sixth China-
Part I: Single-Degree of Freedom Systems,” Journal Japan-US Trilateral Symposium on Lifeline
of Earthquake Engineering, 7(1): 97–118. https://doi. Earthquake Engineering, Reston, Virginia: American
org/10.1080/13632460309350443 Society of Civil Engineers, 425‒432. https://doi.
org/10.1061/9780784413234.055
Christopoulos C and Pampanin S (2004), “Towards
Performance-Based Design of MDOF Structures with Hao Jianbing, Wu Gang and Wu Zhisheng (2013),
Explicit Consideration on Residual Deformations,” “A Study on Constant-Relative-Strength Residual
ISET Journal of Earthquake Technology, 41(1): 53‒73. Deformation Ratio Spectrum of SDOF System,” China
https://www.researchgate.net/publication/29486873 Civil Engineering Journal, 46(10): 82‒88. (in Chinese)
https://doi.org/10.15951/j.tmgcxb.2013.10.021
Dai Kaoshang, Wang Jianze, Li Bowei and Hong HP
(2017), “Use of Residual Drift for Post-Earthquake Hashimoto S, Fujino Y and Abe M (2005), “Damage
Damage Assessment of RC Buildings,” Engineering Analysis of Hanshin Expressway Viaducts During 1995
Structures, 147: 242‒255. https://doi.org/10.1016/j. Kobe Earthquake: Ⅱ: Damage Mode of Single Reinforced
engstruct.2017.06.001 Concrete Piers,” Journal of Bridge Engineering-ASCE,
10(1): 54‒60. https://doi.org/10.1061/(ASCE)1084-
Decanini LD, Sortis AD, Goretti A, et al. (2004),
0702(2005)10:1(54)
“Performance of Reinforced Concrete Buildings During
the 2002 Molise, Italy, Earthquake,” Earthquake Spectra, Hatzigeorgiou GD, Papagiannopoulos GA and Beskos DE
20(1): 221‒255. https://doi.org/10.1193/1.1765107 (2011), “Evaluation of Maximum Seismic Displacements
of SDOF Systems from Their Residual Deformation,”
FEMA (2012), Seismic Performance Assessment of
Engineering Structures, 33(12): 3422‒3431. https://doi.
Buildings Vol 1—Methodology, Redwood City, CA:
org/10.1016/j.engstruct.2011.07.006
Applied Technology Council. https://www.fema.gov/
medialibrarydata/13964950198480c9252aac91dd18 Hsu YT and Fu CC (2004), “Seismic Effect on Highway
54dc378feb9e69216/FEMAP-58_Volume1_508.pdf. Bridges in Chi Chi Earthquake,” Journal of Performance
Accessed 01/08/2016 of Constructed Facilities-ASCE, 18(1): 47‒53. https://
doi.org/10.1061/(ASCE)0887-3828(2004)18:1(47)
Fujino Y, Hashimoto S and Abe M (2005), “Damage
Analysis of Hanshin Expressway Viaducts During 1995 Hu Xiaobin and He Huigao (2015), “Study on Equal-
Kobe Earthquake. I: Residual Inclination of Reinforced Strength Residual Displacement Ratio Spectrum,”
Concrete Piers,” Journal of Bridge Engineering-ASCE, Engineering Mechanics, 32(1):163‒167. https://doi.
10(1): 45‒53. https://doi.org/10.1061/(ASCE)1084- org/10.6052/j.issn.1000-4750.2013.07.0704
0702(2005)10:1(45) Japan Road Association (2002), Design Specifications of
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 355

Highway Bridge: Part V: Seismic Design, Tokyo, Japan: www.researchgate.net/publication/285119647 The 1985
Maruzen Publishing Co., Ltd. earthquake Causes and effects in Mexico City
Kawashima K, MacRae GA, Hoshikuma J and Ruiz-Garcia J and Miranda E (2006), “Residual
Nagaya K (1998), “Residual Displacement Response Displacement Ratios for Assessment of Existing
Spectrum,” Journal of Structural Engineering-ASCE, Structures,” Earthquake Engineering and Structural
124(5): 523‒530. https://doi.org/10.1061/(ASCE)0733- Dynamics, 3(3): 315‒336. https://doi.org/10.1002/
9445(1998)124:5(523) eqe.523
Lee WK and Billington SL (2010), “Modeling Residual Ruiz-Garcia J and Miranda E (2010), “Probabilistic
Displacements of Concrete Bridge Columns Under Estimation of Residual Drift Demands for Seismic
Earthquake Loads Using Fiber Elements,” Journal Assessment of Multi-Story Framed Building,”
of Bridge Engineering, 15(3): 240‒249. https://doi. Engineering Structures, 32(1): 11‒20. https://doi.
org/10.1061/(ASCE)BE.1943-5592.0000059 org/10.1016/j.engstruct.2009.08.010
Liossatou E and Fardis MN (2016), “Near-fault Saiidi MS and Ardakani SMS (2012), “An Analytical
Effects on Residual Displacements of RC Structures,” Study of Residual Displacements in RC Bridge Columns
Earthquake Engineering & Structural Dynamics, 45(9): Subjected to Near-Fault Earthquakes,” Bridge Structures,
1391‒1409. https://doi.org/10.1002/eqe.2712 8(1): 35‒45. https://doi.org/10.3233/BRS-2012-0036
MacRae GA and Kawashima K (1997), “Post-Earthquake Takeda T, Sozen M and Nielsen NN (1970), “Reinforced
Residual Displacements of Bilinear Oscillators,” Concrete Response to Simulated Earthquakes,” Journal
Earthquake Engineering and Structural Dynamics, of the Structural Division-ASCE, 96(12): 2557‒2573.
26(7): 701‒716. http://dx.doi.org/10.1002/(SICI)1096- https://www.mendeley.com/catalogue/1d5a57fe-aaf4-
9845(199707)26:7<701::AID-EQE671>3.0.CO;2-I 3e61-b35a-79b4f7fca200/
Pacific Earthquake Engineering Research Center (2017), Yazgan U and Dazio A (2011), “Simulating Maximum
PEER Strong Motion Database, California: Berkley. and Residual Displacements of RC Structures: II.
https://peer.berkeley.edu/peer-strong-ground-motion- Sensitivity,” Earthquake Spectra, 27(4): 1203‒1218.
databases https://doi.org/10.1193/1.3650478
Pampanin S, Christopoulos C and Priestley MJN Zhang Qin, Wang Na and Gong Jinxing (2017), “State of
(2003), “Performance-Based Seismic Response of the Art of Seismic Performance Including Shear Effects
Frame Structures Including Residual Deformations: and Failure Modes of Reinforced Concrete Columns,”
Part II: Multi-Degree of Freedom Systems,”Journal of Journal of Building Structures, 38(08): 1‒13. (in
Earthquake Engineering, 7(1): 119‒147. https://doi. Chinese) https://doi.org/10.14006/j.jzjgxb.2017.08.001
org/10.1142/S1363246903000900 Zhang Qin, Gong Jinxing, Jiang Fengjiao and Zhu Jichao
Rosenblueth E and Meli R (1986), “The 1985 Mexico (2011), “Analysis Method for Nonlinear Properties of
Earthquake: Causes and Effects in Mexico City,” Reinforced Concrete Columns,” Journal of Civil and
Concrete International (ACI), 8(5): 23–34. https:// Environmental Engineering, 33(06): 51‒58. https://doi.
org/10.1016/S1005-0302(11)60029-7

Appendix

Table A1 Characteristics of recorded earthquake ground motions

No. Earthquake event Year Station and component Duration (s) PGA (g) Ms Depth (km)
1 Cape-Mendocino 1992 RIO Dell Overpass-FF 270 36.0 0.385 7.10 9.9

2 Cape-Mendocino 1992 Shelter Cove Airport SHL000 36.0 0.229 7.10 9.9

3 Cape-Mendocino 1992 Shelter Cove Airport SHL090 36.0 0.189 7.10 9.9

4 Chi-Chi 1999 TCU-042 N 90.0 0.199 7.60 17.0

5 Chi-Chi 1999 TCU-068 N 90.0 0.462 7.60 17.0

6 Chi-Chi 1999 TCU-095 N 90.0 0.712 7.60 17.0

7 Chi-Chi 1999 TCU-117 V 90.0 0.088 7.60 17.0

8 Imperial Valley 1940 El Centro Site 270 53.72 0.357 7.20 10.0

9 Morgan Hill 1984 Coyote Lake Dam-CA 195 59.98 0.653 6.15 8.2
356 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

Table A1 Continued

No. Earthquake event Year Station and component Duration (s) PGA (g) Ms Depth (km)
10 Morgan Hill 1984 Coyote Lake Dam-CA 285 59.98 1.161 6.15 8.2

11 Morgan Hill 1984 Coyote Lake Dam-CA VERT 59.98 0.384 6.15 8.2

12 Imperial Valley 1979 Cerro Prieto 147 63.74 0.169 7.20 10.0

13 Imperial Valley 1979 Cerro Prieto 237 63.74 0.157 7.20 10.0

14 Kern County 1952 Taft Lincoln School UP 54.16 0.109 7.70 6.0

15 Kern County 1952 Taft Lincoln School 021 54.16 0.156 7.70 6.0

16 Kern County 1952 Taft Lincoln School 111 54.16 0.178 7.70 6.0

17 Kobe 1995 Kakogawa 000 40.96 0.251 7.20 21.9

18 Kobe 1995 Kakogawa 090 40.96 0.345 7.20 21.9

19 Kobe 1995 Shin-Osaka 000 40.96 0.243 7.20 21.9

20 Kobe 1995 Shin-Osaka 090 40.96 0.212 7.20 21.9

21 Yunnan Lancang 1988 Zhutang S00E 25.33 0.553 7.20 13.0

22 Yunnan Lancang 1988 Zhutang S90E 25.43 0.529 7.20 13.0

23 Yunnan Lancang 1988 Zhutang VERT 25.49 0.372 7.20 13.0

24 Landers 1992 Yermo Fire Station UP 44.00 0.136 7.40 22.0

25 Landers 1992 Yermo Fire Station 270 44.00 0.245 7.40 22.0

26 Landers 1992 Yermo Fire Station 360 44.00 0.152 7.40 22.0

27 Livermore 1980 Hayward Csuh Stadium 146 24.00 0.072 5.80 11.0

28 Loma Prieta 1989 47125 Capitola CAP090 39.98 0.399 7.10 17.2

29 Loma Prieta 1989 57007 Corralitos CLS090 39.98 0.479 7.10 17.2

30 Loma Prieta 1989 58135 UCSC/Lick Lab LOB090 39.98 0.410 7.10 17.2

31 Loma Prieta 1989 58065 Saratoga Aloha Ave. 40.00 0.323 7.10 17.2
STG090
32 Loma Prieta 1989 57563 Santa Teresa Hills SJ225 50.00 0.275 7.10 17.2

33 Mexico 1985 Guerrero 02-077 V90E 41.20 0.075 8.10 16.0

34 Mexico 1985 Guerrero 02-083 EW 20.98 0.010 8.10 16.0

35 Mexico 1985 Michoacan 01-001 NS 48.86 0.141 8.10 16.0

36 Mexico 1985 Michoacan 01-026 EW 25.60 0.049 8.10 16.0

37 Northbridge 1994 Montebello Bluff 206 21.83 0.179 7.10 18.0

38 Northbridge 1994 24464 North Hollywood 000 59.98 0.317 7.10 18.0

39 Northbridge 1994 72450 Sylmar Hospital Parking 29.98 0.843 7.10 18.0
NS
40 Northbridge 1994 24514 Sylmar Hospital 090 40.00 0.604 7.10 18.0

41 Northbridge 1994 24514 Sylmar Hospital 360 40.00 0.843 7.10 18.0

42 ParkField 1966 Cholame #2 DWN 43.69 0.255 6.20 14.0

43 San Fernando 1971 Castaic Old Ridge Route 291 30.00 0.268 6.55 13.0

44 San Fernando 1971 Castaic Old Ridge Route –DWN 30.00 0.171 6.55 13.0
No. 2 Zhang Qin et al.: Probability-based analytical model for predicting the post-earthquake residual deformation of SDOF systems 357

Table A1 Continued

No. Earthquake event Year Station and component Duration (s) PGA (g) Ms Depth (km)
45 San Fernando 1971 Hollywood Storage P.E. Lot 180 79.48 0.171 6.55 13.0

46 San Fernando 1971 Hollywood Storage P.E. Lot 90 79.48 0.211 6.55 13.0

47 San Fernando 1971 Lake Hughes #12 021 36.60 0.366 6.55 13.0

48 San Fernando 1971 Lake Hughes #12 DWN 36.60 0.167 6.55 13.0

49 Tianjin 1976 Tianjin Hospital NS 19.20 0.149 7.50 23.0

50 Tianjin 1976 Tianjin Hospital EW 19.20 0.106 7.50 23.0

51 Tangshan 1976 M0303 Qianan EW 21.92 0.099 7.60 10.0

52 Tangshan 1976 M0303 Qianan NS 22.02 0.135 7.60 10.0

53 Tangshan 1976 M0303 Qianan UP 23.19 0.081 7.60 10.0

54 Wenchuan 2008 Mianzhu EW 160.0 0.819 8.00 10.0

55 Wenchuan 2008 Mianzhu NS 160.0 0.841 8.00 10.0

56 Western 1949 Olympia Hwy Test Lab N86E 89.15 0.280 6.90 15.4
Washington
57 Western 1949 Olympia Hwy Test Lab N04E 89.16 0.165 6.90 15.4
Washington
58 Western 1949 Olympia Hwy Test Lab DOWN 89.00 0.092 6.90 15.4
Washington
59 Whittier Narrows 1987 El-Monte CA N00W 28.26 0.133 6.10 15.5

60 Whittier Narrows 1987 El-Monte CA N90W 28.28 0.237 6.10 15.5

61 Whittier Narrows 1987 Inglewood Union Oil 000 39.99 0.299 6.10 15.5

62 Whittier Narrows 1987 Inglewood Union Oil 090 39.99 0.247 6.10 15.5

63 Westmoreland 1981 West 2588 Westmoreland SOUT 88.48 0.444 5.80 22.0

64 Westmoreland 1981 West 2588 Westmoreland EAST 88.44 0.361 5.80 22.0

65 Westmoreland 1981 West 2588 Westmoreland UP 88.46 0.546 5.80 22.0

66 Whittier Narrows 1987 LA-Wonderland Ave 075 18.16 0.039 6.10 15.5

67 Whittier Narrows 1987 LA-Wonderland Ave 165 18.16 0.047 6.10 15.5

68 Whittier Narrows 1987 LA-Wonderland Ave UP 18.16 0.024 6.10 15.5

69 Oroville 1975 1051 Oroville Seismo 037 12.20 0.092 5.70 5.0

70 Oroville 1975 1545 Oroville Airport 180 12.03 0.036 5.70 5.0

71 Oroville 1975 1546 Up & Down Cafe 000 12.98 0.034 5.70 5.0

72 Yunnan 1988 Gengma S00E 16.56 0.092 7.60 8.0

73 Yunnan 1988 Gengma S90E 16.84 0.094 7.60 8.0

74 Yunnan 1988 Gengma VERT 16.40 0.033 7.60 8.0

75 Coalinga 1983 36456 Parkfied-Fault Zone 090 40.00 0.274 6.70 9.6

76 Coalinga 1983 46314 Parkfied-Cantua Creek 40.00 0.281 6.70 9.6

School 360
77 Coalinga 1983 36228 Parkfied-Cholame 2WA 090 40.00 0.114 6.70 9.6
78 Coalinga 1983 1162 Pleasant Valley PP-Yard 135 9.84 0.22 6.70 9.6
358 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 21

Table A1 Continued

No. Earthquake event Year Station and component Duration (s) PGA (g) Ms Depth (km)
79 Coalinga 1983 1607 Anticline Ridge PAD 270 9.86 0.452 6.7 9.6

80 Loma Prieta 1989 47006 Gavilan College, Phys. Sci. 40.00 0.356 7.1 17.2
Bldg. 067
81 Loma Prieta 1989 47379 Gavilan College, Water 39.98 0.442 7.1 17.2
Tank 090
82 Loma Prieta 1989 47380 HWY 101/Bolsa RD. Motel 39.98 0.323 7.1 17.2
090
83 Loma Prieta 1989 47381 Giloy Sewage Plant 090 39.98 0.369 7.1 17.2

84 Loma Prieta 1989 57217 Coyote Lake Dam- 39.98 0.481 7.1 17.2
Southwest Abutment 285
85 Trinidad 1980 1498 Rio Dell Overpass, FF 270 19.67 0.147 7.3 15.0

86 Trinidad 1980 1498 Rio Dell Overpass, E Ground 21.99 0.163 7.3 15.0
000
87 Trinidad 1983 1498 Rio Dell Overpass West, 000 21.45 0.166 7.3 15.0

88 Trinidad 1983 1498 Rio Dell Overpass East, 000 21.45 0.194 7.3 15.0

89 Taiwan Smart1(45) 1986 Smart1 C00 NS 40.00 0.153 7.8 15.0

90 Taiwan Smart1(45) 1986 Smart1 O01 NS 44.00 0.174 7.8 15.0

91 Taiwan Smart1(5) 1981 Smart1 C00 NS 20.00 0.114 5.7 11.1

92 Taiwan Smart1(5) 1981 Smart1 M01 NS 24.00 0.178 5.7 11.1

93 Anza 1980 5160 Anza Fire Station 225 10.30 0.065 5.3 20.0

94 Anza 1980 5044 Anza-Pinyon Flat 135 10.30 0.131 5.3 20.0

95 Round Valley 1984 1661 Mcgee Creek Surface 360 6.86 0.128 5.8 7.23

96 Round Valley 1984 1661 Mcgee Creek Surface 270 6.86 0.088 5.8 7.23

97 Borah Peak 1983 CPP-610 South 32.38 0.086 7.3 10.0

98 Borah Peak 1983 CPP-610 East 32.38 0.067 7.3 10.0

99 Borah Peak 1983 CEM 000 24.00 0.025 7.3 10.0

100 Borah Peak 1983 HAU 090 28.00 0.033 7.3 10.0