Soil Dynamics and Earthquake Engineering 69 (2015) 148–162

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

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

Correlation between strong motion durations and damage measures

of concrete gravity dams
Gaohui Wang a,b,n, Sherong Zhang c, Chuangbing Zhou d, Wenbo lu a,b
a
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China
b
Key Laboratory of Rock Mechanics in Hydraulic Structural Engineering, Ministry of Education, Wuhan University, Wuhan 430072, China
c
State Key Laboratory of Hydraulic Engineering Simulation and Safety, Tianjin University, Tianjin 300072, China
d
Nanchang University, Nanchang 330000, China

art ic l e i nf o a b s t r a c t

Article history: Strong motion duration affects the cumulative damage of structures significantly. There are more than 30
Received 1 August 2014 different definitions of strong motion duration. This study describes numerically, the interdependency
Received in revised form between several different definitions of strong motion duration and structural accumulated damage
27 October 2014
indices, and the aim is to determine the definitions of strong motion duration that exhibit the strongest
Accepted 6 November 2014
influence on structural damages. For this purpose, 20 as-recorded accelerograms with a wide range of
durations, which are modified to match a 5% damped target spectrum, are considered in this study, and
Keywords: several different definitions of strong motion duration, such as significant duration, bracketed duration
Strong motion duration and uniform duration are proposed for measuring these durations. On the other hand, nonlinear seismic
Concrete gravity dam
analyses of concrete gravity dams subjected to earthquake motions with different strong motion
Correlation coefficient
durations are conducted according to the Concrete Damaged Plasticity (CDP) model including the strain
Earthquake ground motion
Damage indices hardening or softening behavior. Peak displacement, local damage index, global damage index and
Accumulated damage damage energy dissipation are established for characterizing the influence of strong motion duration on
the dynamic response of concrete gravity dams. The degree of the interrelationship between strong
motion durations and damage measures is provided by correlation coefficients. Comparison of the
correlation between the different durations of the ground motion and different damage measures
reveals that strong motion durations calculated from different definitions have no significant influence
on damage measure based on the peak displacement response of the dam, but are positively correlated
to the accumulated damage measures such as the local damage index, global damage index and damage
energy dissipation for events with similar response spectrum.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction amplitude and frequency content has been universally recognized.

As is well known, earthquake ground motion can be character-
ized by amplitude, frequency content and strong motion duration
[1], each of which reflects some particular feature of the shaking.
Amplitude is generally characterized by the peak ground accel-
eration (PGA), the peak ground velocity (PGV) and the peak
ground displacement (PGD). The frequency content is generally
described by the Fourier spectrum of the ground motion. None-
theless, both amplitude and frequency distribution can be
described by the widely accepted response spectrum (in terms of
acceleration, velocity, or displacement). The importance of the
n
Corresponding author at: State Key Laboratory of Water Resources and Hydro-
power Engineering Science, Wuhan University, Wuhan 430072, China.
Tel.: þ 86 27 68772221; fax: þ86 27 68772310.
E-mail address: wanggaohui@whu.edu.cn (G. Wang).
However, the conclusions with regard to the relevance of strong
motion duration to structural response differ widely, ranging from
null to significant, which remains a topic of considerable debate.
This is mainly because the influence of strong motion duration
on structural response and damage depends on many factors
including the type of structure examined, the construction model,
the other parameters used to characterize the ground motion,
the measure of structural damage employed, and the large
number of widely differing duration definitions that have been
proposed [2–4].
There are more than 30 different definitions of strong motion
duration [5]. While there is no unanimous view regarding which
of the definitions of strong motion duration is to be preferred,
which probably reflects the fact that different definitions may be
more or less suitable for different applications. Although a large
number of definitions of strong motion duration have been

http://dx.doi.org/10.1016/j.soildyn.2014.11.001
0267-7261/& 2014 Elsevier Ltd. All rights reserved.
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162
presented in the literature, the available definitions can be
grouped into four different categories: (a) bracketed duration
[6,7]; (b) uniform duration [8]; (c) effective duration [5], and;
(d) significant duration [9], and then classified by whether the
amplitude or energy thresholds used for their measurement are
absolute or relative to the peak value in the recording. Subse-
quently, several new definitions and prediction models of strong
motion duration have been put forward. For example, Taflampas
et al. [10] proposed a new definition of strong motion duration
combining the alternative bracketed and significant duration
definitions based on the time integral of the absolute ground
velocity, and their presented bracketed-significant duration was
found to be well correlated with the strong motion part of the
records. Montejo and Kowalsky [11] proposed a procedure for
estimation of frequency dependent strong motion duration based
on the continuous wavelet transform and the decomposition of
the earthquake record. Arjun and Kumar [12] developed a neural
network approach for estimation of strong motion duration based
on earthquake records and site characteristics. Yaghmaei-Sabegh
et al. [13] presented a simple and effective empirical model for
predicting the significant duration of ground motions based on
recorded earthquake events in Iran.
Experiences from a number of earthquakes show that a ground
motion with moderate peak ground acceleration and a long
duration may cause greater strength and stiffness degradation
than a ground motion with a large acceleration and a small
duration [5]. The duration of strong motion may significantly
affect the damage of structures and plays an important role in
assessing the damage potential of earthquake ground motions.
However, current approaches for the earthquake-resistant design
and structural analysis based on the response spectrum have not
yet considered the influence of the ground motion duration. There
are many studies reporting that link structural damage to para-
meters related either directly or indirectly to strong motion
duration. However, the relevance of strong motion duration to
structural response remains an open question, with some research
indicating no effect [14,15] and other research indicating a
possible correlation [16,17]. At least part of the reason that
researches have differing conclusions on the importance of strong
motion duration is the use of different duration definitions,
structural models and damage metrics. For example, Hancock
and Bommer [2] presented a summary and critical review of the
literature with regard to the influence of strong motion duration
on structural demand, and concluded that those studies employ-
ing damage measures related to cumulative energy usually found a
positive correlation between strong motion duration and struc-
tural damage, while those using damage measures, such as
maximum response parameters, generally found little or no
correlations between duration and damage.
In order to investigate the influence of the strong motion
duration on structural response and damage, a substantial amount
of research has been carried out over the past decades. Youd et al.
[18] clearly recognized that the strong motion duration has
profound effects on the behavior of saturated soils. Mahin [19]
found that strong motion duration might play an important role in
the inelastic deformation and energy dissipation demands of short
period structures. Bommer et al. [4] showed that the duration of
strong motion can make a significant influence on the strength
degradation of masonry structures. Chai and his co-workers
[20,21] found that long duration will increase inelastic design
base shear. Iervolino et al. [22] addressed the question of which
nonlinear demand measures are sensitive to ground motion
duration by statistical analyses of several case studies. The results
led to the conclusion that duration of ground motion does not
have a significant influence on displacement ductility and cyclic
ductility demand. Hancock and Bommer [23] revealed that
149
duration of strong motion has no influence on damage measures
employing the peak response such as inter-storey drift, but if
cumulative parameters are used to measure the damage, the
duration of strong motion is found to have a significant influence
on the inelastic structural response. It should be noted that few
studies have focused their attention on the nonlinear dynamic
response and seismic damage of concrete gravity dams subjected
to earthquake motions with different strong motion durations. For
example, Zhang et al. [24] investigated the effects of strong motion
duration on the dynamic response and accumulated damage of
concrete gravity dams based on the definition of significant
duration. Their result showed that strong motion duration is
insignificant to peak displacement response assessment. While
studies employing damage measures using local and global
damage indices showed that strong motion duration is positively
correlated to the accumulated damage for events with similar
response spectrum. Léger and Leclerc [25] suggested that short
duration analytic records should not be used as a substitute for
other types of more appropriate records in the earthquake safety
evaluation of concrete dams.
The objective of this paper is to provide a method for quantifying
the interrelationship between strong motion durations and damage
measures of concrete gravity dams. 20 as-recorded accelerograms
with a wide range of durations, which are scaled and matched to
match a 5% damped target spectrum, are selected in this study.
Three different definitions of significant duration, bracketed dura-
tion and uniform duration are presented for measuring strong
motion durations. Local damage index, global damage index, peak
displacement, and damage energy dissipation are employed as the
measures of structural damage. A Concrete Damaged Plasticity
(CDP) model including the strain hardening or softening behavior
is selected for the concrete material. Nonlinear dynamic response
and seismic damage analyses of Koyna gravity dam under different
strong motion durations are conducted to furnish the structural
damage status. The interrelationships between the different strong
motion durations and the damage measures are given.
2. Strong motion duration-related measure used in this study
2.1. Definitions of strong motion duration
Any attempt to study on the correlation between strong motion
durations and structural damage levels immediately faces the
problem that there is currently no universally accepted definition
of strong motion duration. Several researches in the past have
been conducted for the quantification of strong motion duration
[8,10,12,26], and there are more than 30 different definitions of
strong motion duration [5]. There is no clear consensus as to
which of the multiple definitions of duration is to be preferred,
which probably reflects the fact that different definitions may be
more or less suitable for different applications. In the past, a
number of researchers have proposed procedures to compute
strong motion duration of an earthquake record. In general, the
available definitions can be classified in four different groups:
(a) bracketed duration (TB) [6,7] in which the duration is defined as
the time interval between the first and the last exceedances of a
particular threshold of acceleration (usually 0.05g); (b) uniform
duration (TU) [8], which rather than a continuous time window are
defined as the sum of time intervals during which the record
exceeds a particular acceleration threshold; (c) effective duration
(TE) [5] that defines the duration of strong motion as the time
interval between two particular thresholds of the Arias intensity,
and; (d) significant duration (TS) [9], it is defined to be the interval
between the times at which a given percentage of Arias intensity
of the record is reached. Most of the proposed definitions are
150 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162

applied directly to recordings of the ground motion but a few
apply one of these generic definitions to dynamic response of
structures. Within the four generic definitions, distinction, which
is related to whether the thresholds of acceleration or energy are
absolute values or defined relative to the maximum value attained
in the accelerogram, can also be found. Absolute definition
includes bracketed duration (TB), uniform duration (TU) and
effective duration (TE), and relative definition contains significant
duration.
Among different seismic parameters, attention is focused on
those which have a high level of correlation with the examined
damage indices. The definition of effective duration is not con-
sidered in this study because it is very difficult to ascertain the
particular thresholds of the Arias intensity for the effective
duration. Therefore, the bracketed duration, the uniform duration,
and the significant duration according to Trifunac/Brandy are
chosen in this study.
The uniform duration only considers the intervals for which the
ground acceleration is above the threshold, which differs from the
bracketed duration; therefore uniform duration is always shorter
than bracketed duration for a given acceleration record, as shown
in Fig. 1. For these two definitions, the absolute threshold such as
0.05g or 0.10g can be used.
It is well known that the destructiveness of a seismic excitation
can be described using several intensity parameters. The Arias
intensity [27] is an energy-related seismic index which is designed
to reflect the total energy content of a seismic excitation, and is
defined by the following relation:
Z HðT 1 Þ ¼ 15%
π T0 2
I0 ¼ a ðtÞdt ð1Þ
2g 0
where I0 is the Arias intensity, T0 is the total duration of the record
and a is the ground acceleration.
Trifunac and Brady [9] have defined the significant duration of
strong ground motion as the time interval between 5% and 95% of
the Arias intensity. For this definition, the Husid diagram [28] is
used. The Husid diagram is the time history of the seismic energy
content scaled to the total energy content and it is given by
R
π =2g 0t a2 ðtÞdt
HðtÞ ¼ ð2Þ
I0
where a is the ground acceleration and H(t) is the Husid diagram
as a function of time t. The Husid diagram of a natural accelero-
gram is shown in Fig. 2. In this figure strong motion duration after
Tricunac/ Brandy can be computed easily.
Because the thresholds of Arias intensity used in this study are
relative to the total Arias intensity of the record, the symbol TS
(significant duration) is used, with the subscript indicating the
relative threshold. Two definitions of TS (70%) (15–85%) [29] and TS
(90%) (5–95%) [9] are used, as shown in Fig. 2. The first strong
motion duration is defined as the time interval between 15% and
85% of the Husid diagram and the second strong motion duration
is defined as the time interval between 5% and 95% of the Husid
diagram. The former definition is intended to capture the energy
from the body waves whereas the latter includes the full wave
train. TS is defined by the following relation:
TS ¼ T2 T1 ð3Þ
( (
HðT 1 Þ ¼ 5%
or ð4Þ
HðT 2 Þ ¼ 85% HðT 2 Þ ¼ 95%

0.3
0.2
Acceleration (g)

0.1
0.0
-0.1
-0.2
-0.3
0 2 4 6 8 10
Time (s)
Absolute acceleration (g)

0.3

0.2

0.1

0.0
0 2 4 6 8 10
Time (s)
Absolute acceleration (g)

0.3

0.2

0.1

0.0
0 2 4 6 8 10
Time (s)

Fig. 1. The diagram of bracketed and uniform durations of a natural accelerogram. (a) A natural accelerogram. (b) The diagram of bracketed duration. (c) The diagram of
uniform duration.
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162 151

where TS is the significant duration of ground motion, T2 is the
time at the 85% or 95% of the Husid diagram and T1 is the time at
the 15% or 5% of the Husid diagram.
2.2. Accelerogram selection and strong motion duration prediction
In order to identify the relationship between the accumulated
damage and strong ground motion duration, 20 real earthquake
records used in this study have been selected from the databases
of the COSMOS [30] and PEER [31] with a broad distribution of
durations. Table 1 shows the data of the acceleration time histories
with different earthquake magnitudes (i.e. Mw 6.0–7.8) that have
occurred in the United States, Taiwan and India. The data sample
includes worldwide well known earthquakes with strong seismic
activity, such as the Loma Prieta (1989), the Northridge (1994), and
the Chi-Chi (1999) events.
0.3
Acceleration (g)
For the purpose of analyzing strong motion duration effects on
nonlinear demand, it is necessary to minimize the influence of
spectral amplitude and other factors. The ground motion records
are normalized to have the same peak ground acceleration (PGA)
equal to 0.3g. Besides, the accelerograms are scaled and matched
to the target spectrum at a given damping level using the
SeismoSignal. The SeismoSignal uses wavelets to adjust the accel-
erograms to match the response spectra, while minimizing
changes to the other ground motion characteristics. The adjusted
accelerograms have a good match on average to the target
spectrum, which ensures that the main difference between the
records is the strong motion duration. The modified records with
different range durations are illustrated in Fig. 3, and the corre-
sponding response spectrum is shown in Fig. 4.
In order to develop vector predictions of duration in conjunction
with other ground-motion parameters related to amplitude and
energy content, three different levels of peak ground acceleration
(PGA) are considered for the input motions: 0.25g, 0.30g and 0.35g.
Among different strong motion durations, attention is focused on
those which have a high level of correlation with the examined

damage measures. Therefore, the definitions of the significant,

bracketed, and uniform duration are used in this study. According
0.0
to these definitions, strong motion durations for all seismic excita-
tions described in Table 1 are calculated, as shown in Table 2. The
duration parameters of the non-modified real accelerograms are
also presented in Table 1. It can be seen from Table 2 that these
-0.3
0 10 20 30 40 50 60 earthquakes with larger PGA will increase absolute measures like
Time (s) bracketed duration (TB) and uniform duration (TU) whereas the
100 relative measures (e.g. significant duration: TS (70%) (15–85%) and
85% 95% TS (90%) (5–95%)) are not influenced by the increased amplitude.
Arias intensity (%)

80 Fig. 5 shows the correlation between total duration and

T70%
60
different definitions of duration of the modified records with a
T90% PGA of 0.30g, and Fig. 6 illustrates a case of correlation between
40 H(t) significant duration (15–85% Arias intensity) and other different
T70%=18.73s definitions of duration of the modified records with a PGA of 0.30g.
20 T90%=26.04s As shown, it is known that although durations calculated using
5% 15%
0 different definitions are generally poorly correlated with the total
0 10 20 30 40 50 60 duration, there is an approximate correlation between significant
duration (TS (70%) (15–85%)) and other different definitions of
Time (s)
duration for these modified records used in this study. This is
Fig. 2. The Husid diagram of a natural accelerogram. because the records have been scaled to have the same PGA, and

Table 1
Strong motion database (non-modified records).

No. Earthquake Date Distance to fault (km) Station location Mw Comp. PGA (cm/s2) TS (s) TB (s) TU (s)

TS (70%) TS (90%)

1 Koyna 11-12-1967 13.0 Koyna Dam 6.5 0 480.20 3.08 5.02 9.07 3.96
2 Northridge 17-01-1994 35.8 Los Angeles Terrace #24592 6.7 180 310.10 5.47 10.28 16.21 5.51
3 Northridge 17-01-1994 15.7 Epiphany Litheran Church #5353 6.7 196 404.20 7.26 10.26 16.86 8.56
4 Loma Prieta 18-10-1989 17.2 South St and Pine Dr #47524 7.0 90 174.50 12.26 23.60 24.72 4.70
5 Chi-Chi 20-09-1999 8.3 Taichung, Taiwan #TCU078 7.6 0 301.80 15.84 22.91 30.16 12.08
6 Petrolia 17-08-1991 11.7 General Store #89156 6.0 90 488.70 2.10 3.27 6.66 3.26
7 Northridge 17-01-1994 18.4 Los Angeles Mulholland Dr #5314 6.7 35 641.90 4.46 7.64 13.22 6.90
8 Loma Prieta 18-10-1989 15.9 Capitola, CA #47125 7.0 90 390.80 8.50 13.19 20.18 7.80
9 Chi-Chi 20-09-1999 7.9 Taichung, Taiwan #TCU072 7.6 360 360.10 15.32 23.65 30.98 16.84
10 Landers 28-06-1992 28.6 Whitewater Canyon #5072 7.3 270 124.92 23.65 31.41 32.30 4.73
11 Northridge 17-01-1994 8.6 Los Angeles Dam #2141 6.7 334 419.10 4.23 6.50 9.98 5.21
12 Loma Prieta 18-10-1989 11.9 Santa Teresa Hills #57563 7.0 315 223.40 6.60 9.38 15.18 5.54
13 Superstition Hills 24-11-1987 13.0 Westmorland, CA #11369 6.6 180 203.56 9.02 19.48 22.98 6.92
14 Kern County 21-07-1952 36.2 Lincoln School #1095 7.5 111 175.95 13.78 28.57 15.58 3.80
15 Landers 28-06-1992 10.0 Joshua Tree #22170 7.3 90 278.40 21.46 25.98 31.46 12.76
16 San Simeon 22-12-2003 14.8 Caltrans Bridge Grnds #37737 6.5 90 175.00 6.60 11.37 8.49 2.03
17 Parkfield 28-09-2004 14 USGS Parkfield Dense #02 6.0 360 170.00 6.47 11.69 12.99 2.93
18 Imperial Valley 15-10-1979 9.8 Casa Flores Mexicali #6619 6.5 0 238.10 6.56 10.58 15.69 5.54
19 Northridge 17-01-1994 12.5 Coldwater Canyon School #5309 6.7 180 313.30 8.18 15.82 18.36 6.44
20 Chi-Chi 20-09-1999 8.9 Taichung, Taiwan #TCU050 7.6 0 127.50 18.37 26.63 29.86 5.60
152 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162

1 2 3 4 5 6

7 9 10 11
8

0.6 g 12 13 15
14

16 17 18 19 20

20 s

Fig. 3. Acceleration time histories for the modified strong motion records used in this study. For ground motion information see Table 1.

3.0 the displacement at the elastic limit); Δx1,alw ¼ x1,alw  xy ¼ultimate

plastic excursion.
2.5 Apart from the already mentioned index of ductility, the most
Amplification factor

2.0
1.5
1.0
0.5
0.0
0 1 2 3
Period (s)
Fig. 4. Acceleration spectrum for scaled earthquake records.
have been matched to the target spectrum
damping level.
3. Seismic accumulated damage indices
Among many parameters of the structural response, attention
is focused on those which are able to reflect the structural damage.
The available parameters in the literature which quantify the
seismic damage can be classified into ductility based, modal and
energy based damage indices [32]. There have been considerable
efforts to represent the damage level using damage indices that
are capable of quantifying numerically the degree of damage. The
force based approach and the displacement based approach [33]
are introduced in current seismic provisions and are deemed to
ensure the prescribed performance objectives. However, the
inelastic strength spectra and the inelastic displacement spectra
are significant only on the maximum demand of ductility and/or
displacement and they do not provide any information about the
effects of the cumulative damage [34,35]. To account for the effects
of cyclic loading, various noncumulative damage indices [36,37]
including the stiffness and strength degradation are presented.
The seismic-resistant assessment of structures that are not sensi-
tive to the cumulative damage phenomena can be performed
using the well-known criterion of ductility [38]
    tures and RC structures (1.6–1.8); in a damage analysis, sometimes
Dμ ¼ μ  1 = μ1;alw 1 ¼ Δxmax =Δx1;alw o 1 ð5Þ
where Dμ is the ductility index; Δxmax ¼xmax  xy ¼maximum
plastic excursion (xmax is the maximum displacement and xy is
used widely damage indices are the Park and Ang damage index,
the index of hysteretic energy, and the low-cycle fatigue index
[35]. But the proposed formulation is applicable only to the
systems that show negligible or low strain hardening in the
nonlinear field. The analytical expression of the Park and Ang
index DP&A [39] is
xmax EH
DP&A ¼ þβ ð6Þ
x1;alw F y x1;alw
where xmax is the maximum displacement; x1,alw is the ultimate
4 displacement; β is an empirical constant that depends on struc-
tural characteristics (the experimental values of β ranges between
about  0.3 to þ1.2, with a medial of about 0.15 [34]); EH is the
hysteretic energy; and Fy is the elastic strength. Both maximum
plastic displacement and plastic dissipated energy can be taken
at a given into account in the index. However, the experimental assessment
of the parameter β is very difficult and the methodology is not
well stated; another limitation is the linear combination of
ductility and energy in a highly nonlinear problem.
Cumulative damage indices can be divided in deformation
based [40] or hysteresis based [41,42] formulations and methods
that generalize the linear law of low-cycle fatigue of metals
through a hypothesis of linear damage accumulation [43]. Another
approach is based on the accumulation of damage due to cyclic
loading that is usually modeled by introducing the low-cycle
fatigue law. Sucuoǧ lu and Erberik [44] presented low-cycle fatigue
damage models for deteriorating systems on the basis of test data
and analysis. Kamaris et al. [45] proposed a new damage index
which takes into account the phenomenon of low-cycle fatigue
and the interaction between the axial force and bending moment.
The analytical expression of the low-cycle fatigue index DF [46] is
!b  b
n μi  1 n Δx i
DF ¼ ∑ ¼ ∑ ð7Þ
i¼1 μ1;alw  1 i¼1 Δx1;alw
where b is an empirical damage parameter. The value assumed by
this damage index is defined through the only constant b, which
depends on the structural typology and material, and the number
of different plastic displacements. Typical values of b can be
obtained through experimental data, mainly regarding steel struc-
a conservative value of 1.5 is assumed [34]. However, it is very
difficult to get the experimental parameter b for dam structures.
In addition, Massumi and Moshtagh [47] offer a new damage index
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162 153

Table 2
Strong motion durations obtained from modified records.

No. Earthquake T0 (s) Significant duration TS (s) Bracketed duration TB (s) Uniform duration TU (s)

TS (70%) TS (90%) PGA ¼0.30g PGA ¼0.25g PGA ¼0.35g PGA ¼0.30g PGA ¼0.25g PGA ¼0.35g

1 Koyna 10.00 3.08 5.02 6.92 6.92 8.98 2.62 2.16 3.11
2 Northridge 20.00 5.49 10.27 14.31 12.03 16.21 4.82 3.72 5.94
3 Northridge 20.00 7.14 10.12 13.74 11.12 16.58 6.22 5.12 7.20
4 Loma Prieta 40.00 13.00 23.70 33.18 25.98 33.22 10.60 8.08 12.74
5 Chi-Chi 60.00 18.73 26.04 31.31 31.28 31.33 12.24 10.05 14.27
6 Petrolia 10.00 2.00 3.26 3.76 3.70 5.14 2.36 1.90 2.74
7 Northridge 25.00 4.46 7.62 9.28 8.56 9.28 3.76 3.12 4.44
8 Loma Prieta 39.82 8.48 13.16 14.62 14.46 14.62 5.74 4.88 6.62
9 Chi-Chi 40.00 15.24 23.55 30.52 25.02 30.56 12.38 10.53 14.40
10 Landers 56.55 23.59 31.36 43.04 40.18 49.16 20.00 16.84 22.58
11 Northridge 26.55 4.06 6.67 7.02 6.66 9.01 3.08 2.43 3.62
12 Loma Prieta 25.00 6.60 9.32 16.28 15.18 19.46 6.62 5.74 7.60
13 Superstition Hills 59.82 8.82 17.04 23.24 23.20 23.24 7.44 5.76 8.78
14 Kern County 54.22 13.86 28.86 33.84 32.98 35.34 8.50 6.44 10.74
15 Landers 40.00 21.48 26.08 31.44 31.10 36.68 14.14 11.88 16.48
16 San Simeon 25.00 5.86 10.19 10.97 9.14 11.66 3.28 2.39 4.09
17 Parkfield 20.00 6.41 11.63 16.58 14.86 18.68 5.93 4.81 6.99
18 Imperial Valley 16.95 6.65 10.56 16.27 15.69 16.69 7.02 5.87 8.03
19 Northridge 21.86 8.20 15.86 18.82 18.80 19.04 6.30 5.06 7.66
20 Chi-Chi 50.00 17.62 25.77 33.52 32.55 34.02 18.61 16.05 20.51

TS(70%) (15~85%)
40 40
Strong motion duration (s)
Strong motion duration (s)

TS(90%) (5~95%)
TB (0.05g)
30 30
TU (0.05g)

20 20

10 10

0 0
0 10 20 30 40 50 60 0 5 10 15 20 25

Total duration (s) Significant duration TS(70%) (15~85%) (s)

Fig. 5. Correlation between total duration and different definitions of duration Fig. 6. Correlation between significant duration (TS (70%) (15–85%)) and different
(PGA ¼0.3g). definitions of duration (PGA ¼ 0.3g).

for seismic damage assessment of RC buildings based on the
variation of the nonlinear fundamental period. Hatzigeorgiou
and Beskos [48] proposed a direct damage controlled method for
the determination of the damage of concrete structures.
We evaluate the effects of mainshock–aftershock seismic
sequences on accumulated damage based on the ultimate state
of the dam. As a new indicator that means the soundness of a dam
against crack penetration failure, both local and global damage
indices are proposed to assess the accumulated damage of struc-
tures quantitatively. In this model, the global damage is obtained
as a weighted average of the local damage at the ends of each
element, with the dissipated energy as the weighting function.
The local damage index (Fig. 7) is given by the following relation:
 
lDi
DI Li ¼
Li
where DILi is the local damage index for crack path i, Li is the total
length to which crack path i is expected to grow, and lDi is the
length of the damage path in crack path i. The damage of a crack at
an element integration point is indicated by shading the related
area with red color. As shown in Fig. 7, crack paths along the
damaged elements are obtained. The residual crack paths are
established a-priori, and assumed to propagate approximately
horizontally toward the opposite face, with a cracking pattern
that extends completely across the section.
In the present case, the global damage index is a weighted
average of the local damage indices and the damage dissipation is
chosen as the weighting function. The global damage index is
given by
∑ni¼ 1 DILi Ei
DIG ¼ ð9Þ
∑ni¼ 1 Ei
where DIG is the global damage index, DILi is the local damage
index, Ei is the damage dissipation energy at the crack path i (the
energy dissipated in the whole (or partial) model by damage), n is
the number of crack paths at which the local damage is computed.
4. Seismic damage analysis of Koyna dam
ð8Þ
4.1. Description of model of Koyna gravity dam used for evaluation
The Koyna concrete gravity dam in India, 103 m high and
70.2 m wide at its base, which is one of a few concrete dams that
have experienced a destructive earthquake, is selected as an
application. Finite element model for the tallest section of the
dam is shown in Fig. 8. The mesh of the dam is adequately refined
at the base and near the changes in the slope of the downstream
face, in which crack propagations are expected. The reason is that
damage due to tensile stresses is expected to initiate near stress
154 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162
The length of the residual path in crack
path 1 where cracking did not occur
l
l
Fig. 7. The establishment of the crack paths.
Fig. 8. Finite element model of Koyna dam.
concentrations in those zones. The foundation of the dam is taken
as being rigid.
The material parameters for the Koyna dam concrete are as
follows: the elasticity modulus E ¼3.1  104 MPa, the Poisson's
ratio ν ¼0.2, the mass density ρ ¼2643 kg/m3, the fracture energy
is 250 N/m. The tensile and compressive strength of the dam are
2.9 and 24.1 MPa, respectively. A dynamic magnification factor of
1.2 is considered for the tensile strength to account for strain rate
effects. The energy dissipation of the monolith is considered by the
Rayleigh damping method with 5% damping ratio. The maximum
reservoir water level of 96.5 m is considered. Applied loads include
self-weight of the dam, hydrostatic, uplift, hydrodynamic, and
earthquake forces. The static solutions of the dam due to its gravity
loads and hydrostatic loads are taken as initial conditions in the
dynamic analyses of the system. Westergaard virtual mass [49] is
employed to include the hydrodynamic effect, the value of the
Westergaard virtual mass m0i at node i on the upstream surface of
the dam is
qffiffiffiffiffiffiffi
7 of the downstream face. The top cracking profiles are almost either
m0i ¼ hyi ρw ðbi1 þ bi2 Þ=2
8 nearly horizontal or sloping downward from the downstream face
L
The length of the residual path in crack
path 2 where cracking did not occur
L
where h is the depth of water; yi is the distance from node i to the
water surface, ρw is the mass density of water; and bi1 and bi2 are
the length of the edges of the quadrilateral constant-strain
elements beside node i on the upstream surface of the dam.
4.2. Strong motion duration effects on accumulated damage
of concrete gravity dams
Earthquakes often lead to stiffness and strength deterioration
of structures. For two earthquake ground motions with similar
spectral amplitude but of different duration, the motion of longer
duration would be expected to be more damaging. Hence, the
duration of earthquake ground motion should be considered an
important parameter in addition to the maximum amplitude
and frequency content for adequately characterizing the effect
of earthquake ground motion on seismic damage of structures.
In order to investigate the effects of strong motion duration on the
accumulated damage of concrete gravity dams, the dynamic
damage analyses of Koyna gravity dam under selected real earth-
quakes are conducted employing the Concrete Damaged Plasticity
(CDP) model developed by Lublinear et al. [50] and modified by
Lee and Fenves [51] (see [52] for a detailed description of the
constitutive model). The integration time step used in the analysis
is 0.01 s. Only the horizontal component of the seismic input is
considered in these analyses, which are conducted considering 20
different earthquake records, with records selected to represent a
wide range of intensities from relatively weak motions to very
strong shaking. These records are modified to match a 5% damped
target spectrum.
The accumulated damage profiles of Koyna dam during the
modified real accelerograms with a PGA of 0.3g are shown in Fig. 9.
The figures depict the damage predicted for different range
durations of real ground motions considered in this study. The
shaded areas related to red color indicate those elements that
experienced some level of tensile damage over the duration of the
analysis. As shown, the crack propagation process and failure
modes are obtained. From the cracking profiles shown in Fig. 9
(a)–(t), it can be observed that the failure mechanism is formed of
two main damage zones, one at the base and one in the upper part
of the dam. In almost all the analyses, the cracking is always
initiated at the dam heel which may be due to stress concentra-
tion, and then progresses a long way from the upstream face to the
downstream face. The cracking profiles in the upper part of
the dam are always initiated near the discontinuity in the slope
ð10Þ
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162 155

Fig. 9. Cracking profiles for Koyna dam under modified real accelerations with a PGA of 0.30g. (a) Koyna, (b) Northridge #24592, (c) Northridge #5353, (d) Loma Prieta
#47524, (e) Chi-Chi #TCU078, (f) Petrolia #89156, (g) Northridge #5314, (h) Loma Prieta #47125, (i) Chi-Chi #TCU072, (j) Landers #5072, (k) Northridge #2141, (l) Loma
Prieta #57563, (m) Superstition Hills, (n) Kern county #1095, (o) Landers #22170, (p) San Simeon #37737, (q) Parkfield #02, (r) Imperial Valley #6619, (s) Northridge #5309,
and (t) Chi-Chi #TCU050.
156 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162

toward the upstream face at approximately a 451 angle to the
vertical. The crack trajectory curves decrease due to the compres-
sive stresses resulting from rocking of the top block. But in some
analyses, cracks are predicted to initialize near the middle of the
upstream face or the downstream face, and extend into the dam.
Because of the thrust of impounded water which is opposing the
tendency of the top section to slide along the crack in the
upstream direction, the computed crack profiles in the upper part
of the dam can be considered neutral or favorable conditions to
maintain stability.
For the seismic motions associated with a short duration, the
actual response of the dam will exhibit some tensile cracking, and
some small damage is identified. But it will not drastically affect
the results of the dynamic behavior of the dam. On the other hand,
the cases associated with a moderate duration, some moderate
damage is found but it does not seem to reach a level that could
compromise the integrity of the section. However, the results
corresponding to the input motions with a long duration clearly
show indications of significant strength degradation in the dam,
with a cracking pattern that extends completely across the
upper section. Longer duration will lead to greater accumulate
damage to which aseismic design of the dam should be given
attention.
5. Correlation study between strong motion durations
and damage measures
The influence of strong motion duration on nonlinear structural
response, however, remains a topic of much debate and universal
conclusions are unlikely to reach since the resolution of the issue
is complicated by the variety of definitions of duration and the
variety of structural behavior, as well as the difficulty of decou-
pling the specific effect of duration from other features of the
ground motion. In order to analyze the interrelationships between
the different strong motion durations and the damage measures,
both relative (15–85% significant duration and 5–95% significant
duration) and absolute (0.05g bracketed duration and 0.05g
uniform duration) definitions are used to determine the influence
of ground motion duration on different damage measures such as
peak displacement, local damage index, global damage index and
damage energy dissipation. The seismic response analyses are
carried out by scaling each input record, to progressively increase
the intensity of ground shaking by increments of about 0.05g. Real
earthquake records are scaled to different intensity levels in order
to produce responses ranging from elastic to large nonlinear
responses of the dam. A further aspect which has been taken into
consideration is the expected damage potential of strong motion
duration on concrete gravity dams to be analyzed.
After the nonlinear dynamic response analyses of the examined
concrete gravity dam for all the selected accelerograms at different
intensity levels, the peak displacement and damage energy dis-
sipation are obtained, and the local and global damage indices are
computed according to Eqs. (8) and (9) for each seismic excitation.
The damage index values for each used accelerogram are pre-
sented in Table 3. The influence of strong motion duration on the
damage of concrete gravity dams is illustrated by using different
symbols for different levels of duration on plots of average spectral
accelerations versus damage indices (Figs. 10–12). Correlations are
also made between strong motion duration and peak displace-
ment and damage energy dissipation (Figs. 13 and 14).
As mentioned above, the failure mechanism is formed of two
main damage zones, one at the base and one in the upper part of
the dam. Hence, two local damage indices are calculated. To analyze
the effects of strong motion duration on the accumulated damage of
concrete gravity dams, Figs. 10–12 are generated by plotting the
accumulated damage of the dam imparted by the 20 records for a
given level of intensity in terms of local and global damage indices.
Trend lines (straight lines) for each level of intensity are displayed
with the aim of identifying a general tendency (if any). This also
applies for the other figures of this type in the paper. However,
S-form trend lines are fitted to the plots to show the general trends
for the correlation between strong motion duration and local
damage index of the dam heel. In some cases, the S-form trend
lines are also used to identify the correlation between the uniform
duration and damage measures. The trend lines show that, as
would be expected, the local and global damage indices of the
examined dam during the modified real accelerograms with longer
duration are generally greater than those under shorter events for
the same level of spectral acceleration. It can also be found that the

Table 3
Damage index values (for ground motion information see Tables 1 and 2).

No. Earthquake Local damage index Global damage index

The dam heel The upper part of the dam

PGA (0.25g) PGA (0.30g) PGA (0.35g) PGA (0.25g) PGA (0.30g) PGA (0.35g) PGA (0.25g) PGA (0.30g) PGA (0.35g)

1 Koyna 0.08 0.09 0.14 0.08 0.21 0.28 0.08 0.12 0.18
2 Northridge 0.14 0.19 0.22 0.08 0.26 0.32 0.12 0.21 0.25
3 Northridge 0.19 0.25 0.28 0.26 0.48 0.65 0.21 0.31 0.38
4 Loma Prieta 0.20 0.27 0.31 0.40 0.61 0.79 0.25 0.36 0.44
5 Chi-Chi 0.20 0.24 0.28 0.58 0.90 1.00 0.29 0.40 0.46
6 Petrolia 0.09 0.14 0.17 0.05 0.10 0.13 0.08 0.13 0.16
7 Northridge 0.10 0.14 0.17 0.18 0.30 0.38 0.12 0.18 0.23
8 Loma Prieta 0.17 0.22 0.25 0.25 0.42 0.64 0.19 0.27 0.35
9 Chi-Chi 0.17 0.22 0.25 0.38 0.63 0.86 0.22 0.32 0.40
10 Landers 0.23 0.26 0.31 0.69 1.00 1.00 0.35 0.44 0.48
11 Northridge 0.05 0.08 0.11 0.02 0.15 0.21 0.04 0.10 0.14
12 Loma Prieta 0.13 0.17 0.22 0.17 0.28 0.41 0.14 0.20 0.27
13 Superstition Hills 0.16 0.22 0.27 0.23 0.44 0.55 0.18 0.28 0.34
14 Kern County 0.20 0.23 0.27 0.36 0.64 0.90 0.25 0.34 0.43
15 Landers 0.22 0.25 0.30 0.59 0.90 1.00 0.31 0.42 0.48
16 San Simeon 0.20 0.25 0.28 0.30 0.32 0.54 0.23 0.27 0.35
17 Parkfield 0.14 0.20 0.23 0.24 0.31 0.55 0.17 0.23 0.32
18 Imperial Valley 0.14 0.17 0.22 0.21 0.43 0.55 0.16 0.24 0.31
19 Northridge 0.16 0.19 0.22 0.34 0.44 0.67 0.20 0.25 0.34
20 Chi-Chi 0.22 0.26 0.31 0.40 0.75 1.00 0.27 0.39 0.50
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162 157

0.3 0.3

Local damage index

Local damage index

0.2 0.2

0.1 0.1
PGA=0.30g PGA=0.30g
PGA=0.25g PGA=0.25g
PGA=0.35g PGA=0.35g
0.0 0.0
0 5 10 15 20 25 0 5 10 15 20 25 30 35
Duration (s) Duration (s)

0.3 0.3
Local damage index

Local damage index

0.2 0.2

0.1 0.1
PGA=0.30g PGA=0.30g
PGA=0.25g PGA=0.25g
PGA=0.35g PGA=0.35g
0.0 0.0
0 10 20 30 40 50 0 5 10 15 20 25

Duration (s) Duration (s)

Fig. 10. The influence of strong motion duration on the local damage index measure for the dam heel. (a) TS (70%) (15–85%); (b) TS (90%) (5–95%); (c) TB (0.05g); (d) TU (0.05g).

1.0 1.0
Local damage index
Local damage index

0.8 0.8

0.6 0.6

0.4 0.4

PGA=0.30g PGA=0.30g
0.2 0.2
PGA=0.25g PGA=0.25g
PGA=0.35g PGA=0.35g
0.0 0.0
0 5 10 15 20 25 0 5 10 15 20 25 30 35
Duration (s) Duration (s)

1.0 1.0
Local damage index

Local damage index

0.8 0.8

0.6 0.6

0.4 0.4

PGA=0.30g PGA=0.30g
0.2 0.2
PGA=0.25g PGA=0.25g
PGA=0.35g PGA=0.35g
0.0 0.0
0 10 20 30 40 50 0 5 10 15 20 25
Duration (s) Duration (s)

Fig. 11. The influence of strong motion duration on the local damage index measure for the upper part of the dam. (a) TS (70%) (15–85%); (b) TS (90%) (5–95%); (c) TB (0.05g);
(d) TU (0.05g).

ground motions with greater peak ground acceleration (PGA) would PGA (Fig. 10(d), Fig. 11(d), and Fig. 12(d)). The damage demand
cause greater damage to concrete gravity dams than smaller events. examined by the significant duration (15–85% significant duration
However, the correlation is relatively weak with significant overlap and 5–95% significant duration) is proportional to the peak ground
between uniform duration and damage measures with different acceleration of ground motions. The results from the bracketed
158 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162

0.5 0.5

Global damage index

0.4 0.4
Global damage index

0.3 0.3

0.2 0.2

0.1 PGA=0.30g 0.1 PGA=0.30g

PGA=0.25g PGA=0.25g
PGA=0.35g PGA=0.35g
0.0 0.0
0 5 10 15 20 25 0 5 10 15 20 25 30 35
Duration (s) Duration (s)

0.5 0.5

Global damage index

0.4 0.4
Global damage index

0.3 0.3

0.2 0.2

PGA=0.30g PGA=0.30g
0.1 0.1
PGA=0.25g PGA=0.25g
PGA=0.35g PGA=0.35g
0.0 0.0
0 10 20 30 40 50 0 5 10 15 20 25
Duration (s) Duration (s)

Fig. 12. The influence of strong motion duration on the global damage index measure for the dam. (a) TS (70%) (15–85%); (b) TS (90%) (5–95%); (c) TB (0.05g); (d) TU (0.05g).

6 6
Peak displacement (cm)
Peak displacement (cm)

5 5

4 4

3 3

2 2
PGA=0.30g PGA=0.30g
1 PGA=0.25g 1 PGA=0.25g
PGA=0.35g PGA=0.35g
0 0
0 5 10 15 20 25 0 5 10 15 20 25 30 35
Duration (s) Duration (s)

6 6
Peak displacement (cm)

5 5
Peak displacement (cm)

4 4

3 3

2 2
PGA=0.30g PGA=0.30g
1 PGA=0.25g 1 PGA=0.25g
PGA=0.35g PGA=0.35g
0 0
0 10 20 30 40 50 0 5 10 15 20 25
Duration (s) Duration (s)

Fig. 13. The influence of strong motion duration on peak displacement of the dam. (a) TS (70%) (15–85%); (b) TS (90%) (5–95%); (c) TB (0.05g); (d) TU (0.05g).

duration and uniform duration measures are not very similar. This is
because the relative measures (significant duration) are not influ-
enced by the increased amplitude.
As can be seen from Figs. 10–12, studies employing damage
measures using local and global damage indices show that strong
motion durations calculated from different definitions (bracketed
duration, uniform duration and significant duration) are all posi-
tively correlated to the accumulated damage for events with
similar response spectrum. The damage measures such as the
local damage index for the dam upper zone and global index for
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162 159

80 80

Damage energy dissipation (kN.m)

Damage energy dissipation (kN.m)

PGA=0.30g PGA=0.30g
PGA=0.25g PGA=0.25g
60 PGA=0.35g 60 PGA=0.35g

40 40

20 20

0 0
0 5 10 15 20 25 0 5 10 15 20 25 30 35
Duration (s) Duration (s)
Damage energy dissipation (kN.m)

Damage energy dissipation (kN.m)

80 80
PGA=0.30g PGA=0.30g
PGA=0.25g PGA=0.25g
60 PGA=0.35g 60 PGA=0.35g

40 40

20 20

0 0
0 10 20 30 40 50 0 5 10 15 20 25

Duration (s) Duration (s)

Fig. 14. The influence of strong motion duration on the damage energy dissipation measure for the dam. (a) TS (70%) (15–85%); (b) TS (90%) (5–95%); (c) TB (0.05g);
(d) TU (0.05g).

0.5
1.0
0.4
Global damame index
Local damame index

0.8

0.3
0.6

0.2 TS(70%) (15~85%)

0.4 TS(70%) (15~85%)
TS(90%) (5~95%) TS(90%) (5~95%)
0.2 0.1 TB (0.05g)
TB (0.05g)
TU (0.05g) TU (0.05g)
0.0 0.0
0 10 20 30 40 0 10 20 30 40
Duration (s) Duration (s)
Damage energy dissipation (kN.m)

30

20

TS(70%) (15~85%)
10 TS(90%) (5~95%)
TB (0.05g)
TU (0.05g)
0
0 10 20 30 40
Duration (s)

Fig. 15. Comparison of the correlations between strong motion duration with different definitions and (a) the local damage index measure for the dam upper part; (b) the
global damage index measure; (c) damage energy dissipation measure.

the dam are consistently greater for ground motions with longer for the upper zone of the dam is more sensitive to ground motion
duration. While the accumulated damage on the dam heel is not duration, which gives more importance to the dissipated energy
very sensitive to strong motion duration. The accumulated damage during the hysteretic behavior of the structure since the high
160 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162

seismic response zone is mainly located in the upper zone of the
dam. After the long duration earthquake, more significant accu-
mulated damage remains in the upper zone of the dam due to
plastic strain during cyclic loadings.
To analyze the effects of strong motion duration on the damage
of concrete gravity dams, Fig. 13 is generated to identify whether
the peak displacement response of the dam is related to strong
motion duration, and Fig. 14 is generated by plotting the damage
energy dissipation of the dam imparted by the 20 records for a
given level of intensity. From Fig. 13, it can be observed that there
is no significant relationship between any of the three duration
definitions (bracketed duration, uniform duration and significant
duration) under consideration and the damage measure based on
the peak displacement of the structure. However, the results from
the damage energy dissipation measure are similar to that from
the local and global index measures. Stronger duration depen-
dence might be found for higher loading levels, which would
cause greater deformation and damage energy dissipation in the
dam. From Fig. 14, it can also be noticed that the bracketed and
uniform durations significantly increase the scatter of damage
measures around the trend lines.
The above analyses show that strong motion duration has
significant influence on the local damage of the dam upper part,
the global damage, and the damage energy dissipation. Hence,
Fig. 15 mainly shows the comparison of the correlations between
strong motion duration and these damage measures during the
modified real accelerograms with a PGA of 0.3g. The trend lines in
Fig. 15 show that the events with longer duration generally cause
greater damage on the dam than the shorter events for the same
level of spectral acceleration. The findings do indicate that the
seismic performance of concrete gravity dams would be improved
by taking account of the expected duration of earthquake ground
motion. The comparison of the scatter of the observations around
the trend lines illustrated in Fig. 15 indicates that the significant
duration obtained based on the time between 15% and 85% of the
Arias intensity is the duration measure which correlates slightly
more with accumulated damage. This is to be expected, as this
measure is the best indication to assess the potential damage
energy of earthquake ground motion, as noted earlier. However,
the duration measures with 5–95% significant duration, 0.05g
bracketed duration and 0.05g uniform duration generally increase
the scatter of the damage measures around the trend lines. In order
to justify this assertion and emphasize the grade of interrelation
between strong motion duration and damage measures, the corre-
lation coefficient (R) has been used, which is defined by the
following relation:
 
∑N i ¼ 1 ðxi  xÞ yi  y
R ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi ð11Þ
 2
∑N ð x  x Þ 2
∑ N y  y
i¼1 i i¼1 i
where xi and x are the actual and their average values; yi and y are
the predicted and their average values; and N denotes the number
of data in the analysis. The values (R) of correlation coefficients
between strong motion durations with different definitions and
damage measures are tabulated in Table 4. The correlation coeffi-
cients between absolute durations with different thresholds (0.05g
and 0.10g bracketed durations, 0.05g and 0.10g uniform durations)
and damage measures are also compared in Table 4.
Through the correlation coefficients presented in Table 4, it can
be seen that the strong motion duration as defined by Trifunac
and Brady has a very high correlation with accumulated damage
and damage energy dissipation. This is due to the fact that
their definition does take into account the seismic energy content.
The two kinds of significant durations (TS (70%) (15–85%) and TS (90%)
(5–95%)) provide almost the same interrelation grade between the
examined damage measures. It must be pointed out that the
duration measure predicted by the 15–85% significant duration
has shown a slightly higher correlation with accumulated damage
when compared with the 5–95% significant duration, while both of
them have higher correlation when compared with the absolute
durations. Comparing the correlations between damage measures
and absolute durations with different thresholds (0.05g and 0.10g),
very high correlation between damage measures and absolute
durations with 0.05g thresholds can be seen. On the other hand, a
very poor correlation can be observed between damage measures
and absolute durations with 0.10g thresholds. Among the examined
definitions of strong motion duration, the 15–85% significant dura-
tion has the strongest correlation with the local damage indices.
The analyses presented in this paper show that for ground motions
with comparable modified real records, those with longer duration or
greater magnitude generally cause greater damage, as measured by
local damage index and global damage index, in concrete gravity
dams. This is simply an empirical verification of expected structural
behavior, which suggests that shorter strong motion duration records
should not be used as the seismic motion input in the earthquake
safety evaluation of concrete gravity dams. It should be noted that the

Table 4
The values (R) of correlation coefficient between strong motion duration and damage measures.

Damage measures Duration measures The value (R) of correlation coefficient

PGA (0.25g) PGA (0.30g) PGA (0.35g)

The local damage index measure for the dam upper part TS (70%) (15–85%) 0.942 0.980 0.940
TS (90%) (5–95%) 0.894 0.938 0.939
TB (0.05g) 0.884 0.926 0.896
TU (0.05g) 0.831 0.916 0.886
TB (0.10g) 0.851 0.926 0.887
TU (0.10g) 0.766 0.843 0.815
The global damage index measure for the dam TS (70%) (15–85%) 0.908 0.929 0.888
TS (90%) (5–95%) 0.892 0.917 0.886
TB (0.05g) 0.872 0.909 0.877
TU (0.05g) 0.814 0.872 0.860
TB (0.10g) 0.522 0.371 0.307
TU (0.10g) 0.476 0.345 0.285
The damage energy dissipation measure for the dam TS (70%) (15–85%) 0.920 0.930 0.913
TS (90%) (5–95%) 0.880 0.913 0.898
TB (0.05g) 0.865 0.881 0.884
TU (0.05g) 0.823 0.813 0.828
TB (0.10g) 0.486 0.429 0.437
TU (0.10g) 0.447 0.374 0.382
 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162
correlation grade between the examined strong motion durations and
the damage indices presented in this study is valid for the specific case
of spectrum compatible accelerograms as presented in Section 2.2.
Further investigations will be carried out using the abundant strong
motion records.
6. Conclusions
Researchers have shown increasing interest in relating the
structural damage caused by earthquakes to parameters associated
with ground motion duration. In this paper a methodology for
quantifying the interrelationship between strong motion dura-
tions and damage measures of concrete gravity dams is presented.
The questions which definition of strong motion duration is the
most useful indicator of earthquake damage potential and which
damage measure is the most useful index of structural perfor-
mance are evaluated. The nonlinear dynamic response and seismic
damage process of concrete gravity dams during ground motions
with different durations are conducted according to the Concrete
Damaged Plasticity (CDP) model. The local damage index, global
damage index, peak displacement response and damage energy
dissipation are considered as the damage measures.
Twenty real earthquake records with a wide range of durations,
which are scaled to match the target spectrum at a given damping
level, are considered in this study. The definitions of significant
duration, bracketed duration and uniform duration are used for
characterizing the strong motion duration. From the calculated
durations with different definitions, it can be found that durations
are generally poorly correlated with the total duration. While
there is an approximate correlation between significant duration
(TS (70%) (15–85%)) and other different definitions of duration for
these modified records.
The results show that, for concrete gravity dams, the influence
of strong motion duration (measured by different definitions)
depends on both the peak ground acceleration and the damage
measure employed. Although measure using peak response is
predominantly used in design and assessment applications
because of their conceptual simplicity, the damage measure based
on the peak displacement response exhibits a poor correlation
with the duration of the ground motion. Strong motion duration
has been found to be significant to the cumulative damage of the
dam. Damage measures such as local damage index, global
damage index and damage energy dissipation are consistently
greater for ground motions with longer duration. Among the
examined definitions of strong motion duration, the 15–85%
significant duration has the strongest correlation with the accu-
mulated damage, as this duration definition is intended to assess
the potential damage energy of earthquake ground motions.
TS (70%) (15–85%) displays the highest correlation coefficient with
the local damage index for the dam upper part. Good correlation
can be observed between damage measures and absolute dura-
tions with 0.05g thresholds, whereas absolute durations with 0.10g
thresholds provide poor or fair correlation with damage measures.
The influence of duration depends on a number of different
factors including the definition of strong motion duration, the
damage measure, the strong motion record and the structural
model. These promising results are the starting point for further
exploration, of more seismic records and of other types of dams.
Acknowledgments
The authors gratefully appreciate the supports from the
National Natural Science Foundation of China (Nos. 51125037
and 51379141).
161
References
[1] Kramer SL. Geotechnical earthquake engineering. Upper Saddle River, NJ:
Prentice-Hall; 1996.
[2] Hancock J, Bommer JJ. A state-of-knowledge review of the influence of strong-
motion duration on structural damage. Earthq Spectra 2006;22(3):827–45.
[3] Bommer JJ, Hancock J, Alarcón JE. Correlations between duration and number
of effective cycles of earthquake ground motion. Soil Dyn Earthq Eng 2006;26
(1):1–13.
[4] Bommer JJ, Magenes G, Hancock J, et al. The influence of strong-motion
duration on the seismic response of masonry structures. Bull Earthq Eng
2004;2(1):1–26.
[5] Bommer JJ, Martínez-Pereira A. The effective duration of earthquake strong
motion. J Earthq Eng 1999;3(2):127–72.
[6] Ambraseys NN, Sarma SK. The response of earth dams to strong earthquakes.
Geotechnique 1967;17(3):181–213.
[7] Bolt BA. Duration of strong ground motion. In: Proceedings 5th World
conference on earthquake engineering. Rome; 1973. p. 1304–13.
[8] Bommer JJ, Stafford PJ, Alarcon JE. Empirical equations for the prediction of the
significant, bracketed, and uniform duration of earthquake ground motion.
Bull Seismol Soc Am 2009;99(6):3217.
[9] Trifunac MD, Brady AG. A study on the duration of strong earthquake ground
motion. Bull Seismol Soc Am 1975;65(3):581–626.
[10] Taflampas IM, Spyrakos CC, Koutromanos IA. A new definition of strong
motion duration and related parameters affecting the response of medium-
long period structures. Soil Dyn Earthq Eng 2009;29(4):752–63.
[11] Montejo LA, Kowalsky MJ. Estimation of frequency-dependent strong motion
duration via wavelets and its influence on nonlinear seismic response. Comput
Aided Civil Infrastruct Eng 2008;23(4):253–64.
[12] Arjun CR, Kumar A. Neural network estimation of duration of strong ground
motion using Japanese earthquake records. Soil Dyn Earthq Eng 2011;31
(7):866–72.
[13] Yaghmaei-Sabegh S, Shoghian Z, Neaz Sheikh M. A new model for the
prediction of earthquake ground-motion duration in Iran. Nat Hazards
2014;70(1):69–92.
[14] Shome N, Cornell CA, Bazzurro P, et al. Earthquakes, records, and nonlinear
responses. Earthq Spectra 1998;14(3):469–500.
[15] Cornell CA. Does duration really matter? In: Proceedings of the FHWA/NCEER
workshop on the national representation of seismic ground motion for new
and existing highway facilities; 1997. p. 125–33.
[16] Reinoso E, Ordaz M, Guerrero RUL. Influence of strong ground-motion
duration in seismic design of structures. In: Proceedings of the 12th World
conference on earthquake engineering; 2000. p. 1151.
[17] Hancock J, Bommer JJ. The influence of phase and duration of earthquake
damage in degrading structures. In: Proceedings 13th world conference on
earthquake engineering. Vancouver (B.C. Canada); 2004. p. 1990.
[18] Youd TL, Idriss IM, Andrus RD, et al. Liquefaction resistance of soils: summary
report from the 1996 NCEER and 1998 NCEER/NSF workshops on evaluation of
liquefaction resistance of soils. J Geotech Geoenviron Eng 2001;127
(10):817–33.
[19] Mahin SA. Effects of duration and aftershocks on inelastic design earthquakes.
In: Proceedings of the seventh world conference on earthquake engineering.
Istanbul; 1980. p. 677–80.
[20] Chai YH, Fajfar P. A procedure for estimating input energy spectra for seismic
design. J Earthq Eng 2000;4(4):539–61.
[21] Kunnath SK, Chai YH. Cumulative damage-based inelastic cyclic demand
spectrum. Earthq Eng Struct Dyn 2004;33(4):499–520.
[22] Iervolino I, Manfredi G, Cosenza E. Ground motion duration effects on
nonlinear seismic response. Earthq Eng Struct Dyn 2006;35(1):21–38.
[23] Hancock J, Bommer J. Using spectral matched records to explore the influence
of strong-motion duration on inelastic structural response. Soil Dyn Earthq
Eng 2007;27(4):291–9.
[24] Zhang S, Wang G, Pang B, et al. The effects of strong motion duration on the
dynamic response and accumulated damage of concrete gravity dams. Soil
Dyn Earthq Eng 2013;45:112–24.
[25] Léger P, Leclerc M. Evaluation of earthquake ground motions to predict
cracking response of gravity dams. Eng Struct 1996;18(3):227–39.
[26] Shoji Y, Tanii K, Kamiyama M. A study on the duration and amplitude
characteristics of earthquake ground motions. Soil Dyn Earthq Eng 2005;25
(7):505–12.
[27] Arias A. A measure of earthquake intensity. In: Hanson RJ, editor. Seismic
design for nuclear power plants; 1970. p. 438–83.
[28] Husid RL. Analisis de terremotos: analisis general. Rev IDIEM 1969;8
(1):21–42.
[29] Takizawa H, Jennings PC. Collapse of a model for ductile reinforced concrete
frames under extreme earthquake motions. Earthq Eng Struct Dyn 1980;8
(2):117–44.
[30] COSMOS Virtual Data Center.: 〈http://strongmotioncenter.org/vdc/scripts/
default.plx〉.
[31] PEER Strong Motion Database.: 〈http://peer.berkeley.edu/peer_ground_mo
tion_database/〉.
[32] Elenas A, Meskouris K. Correlation study between seismic acceleration
parameters and damage indices of structures. Eng Struct 2001;23(6):698–704.
[33] Calvi GM, Kingsley GR. Displacement-based seismic design of multi-degree-
of-freedom bridge structures. Earthq Eng Struct Dyn 1995;24(9):1247–66.
162 G. Wang et al. / Soil Dynamics and Earthquake Engineering 69 (2015) 148–162
[34] Cosenza E, Manfredi G. Damage indices and damage measures. Prog Struct Eng
Mater 2000;2(1):50–9.
[35] Cosenza E, Manfredi G, Polese M. A simplified method to include cumulative
damage in the seismic response of SDOF systems. J Eng Mech 2009;135
(10):1081–8.
[36] Banon H, Veneziano D. Seismic safety of reinforced concrete members and
structures. Earthq Eng Struct Dyn 1982;10(2):179–93.
[37] Roufaiel MSL, Meyer C. Analytical modeling of hysteretic behavior of R/C
frames. J Struct Eng 1987;113(3):429–44.
[38] Kelly A, Tyson WR, Cottrell AH. Ductile and brittle crystals. Philos Mag 1967;15
(135):567–86.
[39] Park Y, Ang AH. Mechanistic seismic damage model for reinforced concrete.
J Struct Eng 1985;111(4):722–39.
[40] Stephens JE, Yao JTP. Damage assessment using response measurements.
J Struct Eng 1987;113(4):787–801.
[41] McCabe SL, Hall WJ. Assessment of seismic structural damage. J Struct Eng
1989;115(9):2166–83.
[42] Bracci J, Reinhorn A, Mander J. Deterministic model for seismic damage
evaluation on reinforced concrete structures. (Technical report NCEER 89-
0033). Buffalo: State University of New York; 1989.
[43] Krawinkler H, Zohrei M. Cumulative damage in steel structures subjected to
earthquake ground motions. Comput Struct 1983;16(1):531–41.
[44] Sucuoǧ lu H, Erberik A. Energy-based hysteresis and damage models for
deteriorating systems. Earthq Eng Struct Dyn 2004;33(1):69–88.
[45] Kamaris GS, Hatzigeorgiou GD, Beskos DE. A new damage index for plane steel
frames exhibiting strength and stiffness degradation under seismic motion.
Eng Struct 2013;46:727–36.
[46] Miner MA, et al. Cumulative damage in fatigue. J Appl Mech 1945;12
(3):159–64.
[47] Massumi A, Moshtagh E. A new damage index for RC buildings based on
variations of nonlinear fundamental period. Struct Des Tall Spec Build 2013;22
(1):50–61.
[48] Hatzigeorgiou GD, Beskos DE. Direct damage-controlled design of concrete
structures. J Struct Eng 2007;133(2):205–15.
[49] Westergaard HM. Water pressures on dams during earthquakes. Trans ASCE
1933;98:418–32.
[50] Lubliner J, Oliver J, Oller S, et al. A plastic-damage model for concrete. Int J
Solids Struct 1989;25(3):299–326.
[51] Lee J, Fenves GL. Plastic-damage model for cyclic loading of concrete
structures. J Eng Mech 1998;124(8):892–900.
[52] Zhang S, Wang G, Sa W. Damage evaluation of concrete gravity dams under
mainshock–aftershock seismic sequences. Soil Dyn Earthq Eng 2013;50:16–27.