Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Vol. 18, No.

3 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION July, 2019

Earthq Eng & Eng Vib (2019) 18: 535-554 DOI: https://doi.org/10.1007/s11803-019-0520-y

Effect of site amplification on inelastic seismic response


Adhikary S1† and Singh Y2‡
1. Department of Civil Engineering, Visvesvaraya National Institute of Technology, Nagpur-440010, Maharashtra, India

2. Department of Earthquake Engineering, Indian Institute of Technology, Roorkee-247667, Uttarakhand, India

Abstract: The available models for effective periods of site and structure are reviewed in context of frequency tuning in
the inelastic seismic response of soil-structure system. The effect of seismic intensity and ductility demand, on the effective
periods, is investigated, and inelastic site amplification is shown to be strongly correlated to the normalized effective period.
Two non-dimensional parameters, analogous to the conventional site amplification factors in codes, are defined to quantify
the inelastic site amplification. It is shown that the inelastic site amplification factor (i.e. ratio of constant ductility spectral
ordinates at soil site to those at rock outcrop) is able to represent the site effects more clearly, as compared to the inelastic
site amplification ratio (i.e. ratio of inelastic spectral ordinates at soil site to the corresponding elastic spectral ordinates at
rock outcrop). Further, the peak in the amplification factor corresponding to the effective site period diminishes rapidly with
increasing ductility demand.

Keywords: site amplification; normalized response spectra; effective period; inelastic seismic response; inelastic site
amplification factor

1 Introduction of structures. Miranda (1993, 2000) and Ruiz-Garcia


and Miranda (2004) evaluated site-dependent inelastic
Two approaches have been commonly used, in seismic design spectra and the inelastic displacement
literature, to estimate inelastic response from the ratio (Cμ) and concluded that the shape of elastic and
elastic response spectrum. In the first approach, inelastic spectra is quite different. They also pointed out
inelastic displacement ratio, Cμ (defined as the ratio that the predominant period of ground motion has an
of the peak displacement of an inelastic oscillator to important role in inelastic seismic response of structures.
its corresponding linear response) is used to directly Zhai et al. (2007) studied the effect of soil condition on
multiply with the elastic displacement response to get the inelastic displacement ratio spectra for near-fault type
inelastic displacement response (Veletsos and Newmark, ground motions and reported that local soil condition
1960; Veletsos et al., 1965; Newmark, 1982; Miranda, have little influence on inelastic displacement ratio
1993; FEMA 273, 1997; FEMA 356, 2000; FEMA 440, spectra. Mollaioli and Bruno (2008) also studied the
2005). In the second approach, a substitute or equivalent influence of soil type on inelastic displacement ratio and
elastic system with appropriate period and damping is drift demand. They concluded that the inelastic response
used to estimate the inelastic response (Rosenblueth and is strongly affected by site conditions in addition to the
Herrera, 1964; Gulkan and Sozen, 1974; Iwan, 1980; ductility ratio and the fundamental period of structure.
Kowalsky, 1994; ATC 40, 1996; Grant et al., 2005; Athanassiadou et al. (2011) studied elastic and inelastic
Priestley et al., 2007; Dwairi et al., 2007). Presently, the spectra for earthquake records on different soil types in
effect of site amplification on total (inelastic) response Greece and noted that the effect of ground conditions
of structure is considered in the elastic response is less pronounced on both types of spectra if plotted
spectrum itself, and hence is independent of the ductility against normalized period. Assimaki et al. (2012)
demand on the structure. Recently, some researchers studied the effect of soil nonlinearity on the inelastic
have explored the effect of soil on nonlinear response response of SDOF systems, using synthetic ground
motion simulations, in terms of variability of the inelastic
Correspondence to: Adhikary S, Department of Civil displacement ratio. Their results showed high bias and
Engineering, Visvesvaraya National Institute of Technology, scatter in the inelastic displacement ratio predicted by
Nagpur-440010, Maharashtra, India using the empirical Next Generation Attenuation (NGA)
Tel: +91-9561458832 relationships and linear elastic site response models, for
E-mail: shrabonyeq@gmail.com periods close to the fundamental period of the site. The

Assistant Professor; ‡PhD study by Assimaki et al. (2012) also showed that, due
Received August 31, 2017; Accepted April 16, 2018 to soil nonlinearity, the amplitude and the frequency
536 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

content of the ground motion are increasingly modified gets flattened with increasing ductility demand. Since Cμ
for higher PGA. is defined as the ratio of the peak displacement of an
Adhikary et al. (2014) showed that the effect of inelastic oscillator to its corresponding linear response,
soil amplification on inelastic response of structures is this flattening of peak in the inelastic response spectrum
not as pronounced as in the case of elastic response, is manifested as a dip in Cμ corresponding to the
and the code provisions are not as non-conservative in predominant site period.
estimating inelastic response, as they appear to be in Adhikary et al. (2014) also concluded that there
case of elastic response. This finding was mainly due to is a need to develop site amplification models for the
the fact that a dip in the Cμ is observed corresponding inelastic response spectrum, as all the available models
to the predominant site period. This dip in Cμ has also have been derived for elastic response. Conventionally,
been observed by Miranda (1993) and Ruiz-Garcia and the fundamental period corresponding to initial (elastic)
Miranda (2004) for soft soil, and by Athanassiadou et stiffness is considered as the characteristic period, in the
al. (2011) for all site classes. This dip coincides with case of site as well as structure, which is applicable only
the peak in the amplification factor (F) as demonstrated under low amplitude motions.
more clearly in Fig. 1. The figure has been drawn for the This study attempts to identify effective periods
KGWH02 site (Stewart et al., 2008) for two assumed of site and structure under strong earthquake motions,
depths of bedrock, (a) for 9 m and in the second case and their influence on inelastic response in context of
(b) for 29 m. Each figure contains 4 graphs as shown in performance-based seismic design (PBSD).
the legend. The amplification factor (F) and the inelastic In the ensuing, some of the major studies concerning
displacement ratio (Cμ) have been plotted for two levels frequency tuning between site and structure and its
of seismic intensity, expressed as effective peak ground influence on structural response have been summarized.
acceleration at Rock Outcrop, EPGARO equal to 0.05 g The effect of nonlinearity on characteristic periods of
and 0.36 g. The engineering model for the KGWH02 site and structure is also discussed. The phenomenon
site has been adopted from Stewart et al. (2008), and of frequency tuning between the soil and structure is
equivalent linear analysis has been performed in illustrated using a numerical study, and the effect of
DEEPSOIL for the experimentally obtained material seismic intensity and ductility demand on effective
curves, as referred to in Stewart et al. (2008). It can be periods of soil and structure, respectively, are
seen from the figure that the peak in F shifts to longer investigated. Subsequently, the combined effect of
periods with increase in depth of bedrock and intensity soil and structure nonlinearity on response of SDOF
of shaking. Interestingly, the dip in the Cμ also shifts to oscillators is presented by comparing the numerically
the same period and always coincides with the peak in F. obtained inelastic spectral amplification factors with
The cause of this dip in the Cμ plot is that the peak in the those obtained using the conventional approach.
inelastic spectrum diminishes with increasing ductility
demand, and the shape of the inelastic spectrum is very 2 Site-structure frequency tuning and
different from the corresponding elastic spectrum. normalized response spectra
The effective stiffness, and hence the period of
vibration, of an inelastic structure varies in a range Among the various parameters known to affect
between initial (elastic) and secant period, depending the site amplification, the site period, in addition to
on the instantaneous displacement. As a result, the effect the average shear wave velocity and the amplitude of
of frequency tuning between the ground motion and ground motion, has been identified as the most crucial
the oscillator is less prominent in the case of inelastic parameter, by many researchers. Seed et al. (1976)
response, and the peak in the response spectrum demonstrated the importance of site-dependent response
corresponding to the predominant ground motion period spectrum for earthquake resistant design and noted




i
i

Fig. 1 Amplification factor (F) and inelastic displacement ratio (Cμ) plotted for the KGWH02 site, for two levels of seismic intensity
considered in the present study: (a) Bedrock considered at 9 m, (b) Bedrock considered at 29 m. The thin curves
correspond to EPGARO = 0.05 g, and the thick curves correspond to EPGARO = 0.36 g. The inelastic displacement ratio is
shown for μ = 4.
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 537

the occurrence of a pronounced peak in the response (2006) showed that normalizing the structural period
spectrum near the fundamental period of site. They also by a dominant period of motion before statistically
noted that the spectral amplification factors of individual processing the recorded spectral response values
sites may be significantly greater than the mean value for produces “completely different shape and size of spectral
a considered suite of sites. Trifunac and Brady (1976) ordinates”. When plotted against the normalized period,
have also noted the frequency-dependence of response T/Tg, the average spectrum exhibits a “characteristic
spectra due to local site effects. Seed et al. (1985) and peak at values of T/Tg not far from 1”.
Dickenson (1994) carried out extensive studies with Ziotopoulou and Gazetas (2009, 2010) carried out an
strong ground motion records from the 1985 Mexico extensive analytical study (considering a total of 1009
City earthquake and the 1989 Loma Prieta earthquake cases) for ASCE 7-10 (2010) site class ‘D’ with three
and showed that there can be significant amplification shear wave velocities (VS,30 = 180 m/sec, 260 m/sec, 360
near the site period, and that structures having period m/sec), different shear wave velocity distributions, two
in this range are more susceptible to damage. Stewart impedance contrast (bedrock to soil) ratios, two depths
et al. (2008) noticed a local “bump” in the response of bedrock, and different PGA values; they established
spectrum near the elastic site period. A large subset of that the bi-normalized spectrum is practically unique,
KiK-net data has been considered by Cadet et al. (2012) as observed by Xu and Xie (2004). They demonstrated
to develop empirical correlations between amplification that BNS is a more rational form of design spectrum
factors and simple site parameters, such as VSZ (with Z as compared to the “spurious flat spectrum” in design
equal to 5, 10, 20 and 30 m) and the fundamental site codes, and that it has a peak value of Sa /A close to 3.75
frequency f0. Observations similar to those of Stewart (rather than 2.5) in a narrow range of normalized period.
et al. (2008) were also made by Cadet et al. (2012) Gazetas (2012) reported that the conventional design
regarding the site amplification prediction equation spectrum may be unconservative, especially when
(SAPE), which represent a “bell-shaped amplification structures have period near the predominant period of
curve with a peak at the fundamental site frequency”. ground motion, and it may lead to erroneous and unsafe
They also observed that “thicker and/or softer sites could results in the case of soil-structure interaction. Pavel
include larger contributions from higher resonance et al. (2013, 2014) have also emphasized adopting bi-
modes.” In the present study, the numerical results have normalized response spectrum accompanied by “site-
been compared with the empirical results with the help dependent control period and/or mean period values”
of the site amplification prediction equations. in design codes. However, all these researchers have
It has been observed by Miranda and Bertero (1991) highlighted the difficulty in estimating the predominant
that the shape of elastic and inelastic spectra varies ground motion period for a given site and seismic event,
significantly with the change in predominant period of doubted the reliability of the “expected relevant periods”
the ground motion. Miranda (1993) computed constant both for soils and for buildings, and identified this as the
ductility inelastic response spectra, also known as main obstacle in adopting BNS in design, at present.
strength demand spectra, for earthquakes recorded on
rock, alluvium and soft soil sites. In the case of soft soils,
the inelastic response spectra were plotted as a function 3 Effective period of a soil site
of normalized period T/Tg, where Tg is the “predominant
period” of ground motion, defined as the period where Conventionally, the fundamental period of a soil
5% damped velocity response spectrum attains its deposit (T0) (i.e. the period corresponding to the
maximum. The study concluded that both the strength fundamental mode of vibration of the soil deposit) is
and deformation demand of structures on soft soils is used as the characteristic site period to relate with the
strongly dependent on the proximity of the period of the dynamic response of sites. It depends on the thickness,
structure to the predominant period of the site. density and shear wave velocity of the soil deposit. This
In the case of elastic spectra, Xu and Xie (2004) site period provides a useful indication of the period of
introduced the concept of a bi-normalized spectrum vibration at which the most significant amplification can
(BNS), which is a plot of Sa /A with T/Tg; where Sa is be expected as a result of frequency tuning between the
the pseudo spectral acceleration, A is the peak ground site and the structure. Kramer (1996) stated that for a
acceleration, and Tg is the “spectral predominant period” uniform, damped soil deposit on rigid rock, the greatest
(actually denoted as Tp by Xu and Xie) of ground amplification factor occurs approximately at the lowest
motion as defined by Miranda (1993). It has also been natural frequency or the characteristic site period, given
termed as “effective ground period” by Mylonakis and as
Gazetas (2000) and Gazetas (2006). Xu and Xie plotted 2 4H
BNS for 206 strong-motion records of the 1999 Chi- T0   (1)
0 Vs
Chi earthquake and concluded that the BNS is unique
and is least affected by the soil category, nature of In the case of a layered soil deposit, Vs can be
seismic excitation, and method of wave propagation calculated using the weighted average of shear wave
analysis. Mylonakis and Gazetas (2000) and Gazetas velocities of the individual layers. If Hi is the uniform
538 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

thickness of the ith layer, and its average shear wave overestimation of the resonance frequency.
velocity is Vsi, then the equivalent shear wave velocity In the present study, the period corresponding to the
of the deposit consisting of ‘n’ layers is calculated based peak site amplification (i.e. ratio of spectral response
on the travel time as at site to that at rock outcrop) obtained using nonlinear
(equivalent linear in the present study) 1-D wave
H n
H propagation is considered as the ‘effective’ period of
= i (2) the site (Ts,eff), as it takes into account the effects of soil
Vs i 1 Vsi
nonlinearity as well as the possibility of a higher mode
dominating the site response.
Substituting the value of H/Vs in the previous
equation, the fundamental natural period of the deposit
can be computed as the summation of the natural periods 4 Effective period of an inelastic structure
of each of the layers in the deposit, as reported by
Vijayendra et al. (2010): Structures exhibit inelastic behavior under strong
ground shaking; and the effective stiffness and hence
n
Hi n
the effective period of vibration changes with inelastic
T0 = 4 =  Ti (3)
i 1 Vsi i 1 displacement. In the case of nonlinear time history
analysis, the change in stiffness (and hence in the effective
Several closed form solutions as well as approximate period of vibration) is inherently considered. However,
methods (Dobry et al., 1976) for estimating fundamental definition of the effective period of vibration is a crucial
site period, have been developed over the years. The issue in estimating seismic response of structures using
two commonly used methods are the Rayleigh solution response spectra. It is customary to use either the initial
(Clough and Penzien, 1975) and Madera’s procedure, (elastic) period of vibration, Ti or the secant period
also known as the two-layer solution (Dobry et al., corresponding to the peak displacement Tsec as the
1976). reference period. Accordingly, two sets of approaches
The fundamental site period may also be estimated have evolved for estimating inelastic response of
empirically by using the horizontal-to-vertical noise structures. In the first approach, commonly known as
ratio from ambient vibrations (Nakamura, 1989) or displacement modification method (DMM) (Veletsos
through the ratio of horizontal-to-vertical components and Newmark, 1960; Veletsos et al., 1965; Newmark,
of the earthquake recordings (Lermo and Chávez- 1982; Miranda, 1993; FEMA 273, 1997; FEMA 356,
García, 1993; Zhao et al., 2006). Cadet et al. (2012) 2000; FEMA 440, 2005) an inelastic displacement ratio,
have defined the ‘fundamental resonance frequency’ Cμ is used to multiply with the elastic response, to get
of a site as the frequency corresponding to the first the inelastic response. The inelastic displacement ratio is
(i.e. lowest frequency) peak of the surface to borehole obtained from a ‘Constant Ductility’ inelastic (or yield)
(S/B) spectral ratio, where a peak is defined as a spectrum (Veletsos and Newmark, 1960), in which the
“specific local maximum with amplitude larger than 2 initial (elastic) period is used as the reference period for
in a statistical sense”. Régnier et al. (2013) have used all ductility levels. In the second approach, known as
a slightly different definition of characteristic frequency Equivalent Linearization (EL) Approach (Rosenblueth
of a site. They defined the ‘predominant frequency’ of a and Herrera, 1964; Gulkan and Sozen, 1974; Iwan,
site as the frequency for which the site-response curve 1980; Kowalsky, 1994; ATC 40, 1996; Grant et al., 2005;
(borehole Fourier spectral ratio) has its maximum value Priestley et al., 2007, Dwairi et al., 2007) an equivalent
(not necessarily the peak corresponding to the lowest linear system with appropriate period and damping is
frequency). They illustrated that the presence of a major used to obtain the response of the inelastic system. The
impedance contrast in the strata near the surface can secant period corresponding to peak displacement is
trigger soil nonlinearity even at low PGA, signifying the commonly used as the reference period, in this approach.
importance of predominant period of the site. Gofrani et Rosenblueth and Herrera (1964) have provided the
al. (2013) have shown that the fundamental frequencies following expression for estimating the secant period
of sites as obtained from empirical surface to borehole corresponding to ductility demand, μ:
cross-spectral ratios, have a very strong correlation with
the theoretically predicted fundamental frequencies by
using 1-D wave propagation analysis, as observed by 
Tsec  Ti (4)
other researchers (Cadet et al., 2012; Steidl et al., 1996). 1    
Recently, Chenari and Taleshani (2016) evaluated the site
response of heterogeneous natural deposits for harmonic
excitation and concluded that considering only the top where, α is post yield stiffness ratio.
30 m of soil for analysis and neglecting the impedance Sullivan et al. (2004) and Miranda and Ruiz-Garcia
contrast of shear wave propagation velocity profile will (2002) have compared the two approaches and concluded
lead to underestimation of the resonance amplitude and that either of the two approaches can be used to estimate
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 539

the inelastic response with sufficient accuracy for the in inelastic response estimated by different equivalent
PBD. As both approaches utilize empirically obtained linearization and displacement modification methods
calibration factors, any period (even other than Ti and and observed that the Iwan (1980) method resulted in
Tsec) can be used as reference period for estimating smaller errors, as compared to the other considered
response with reasonable accuracy. However, in the methods, which were based on secant period. However,
present study, the effective period of structure is required Iwan’s method also underestimated displacement in
for the purpose of identifying the period at which tuning short period range. A rigorous statistical analysis was
with the soil vibrations occurs and it is perceived that performed under the FEMA 440 (2005) project, for
neither Ti nor Tsec can serve this purpose and the actual minimization of error in the estimated inelastic response,
effective period is somewhere in between the two values. and improved optimal equivalent linear periods (Eqs.
Eser and Aydemir (2011) investigated the inelastic 7-9) have been estimated for different hysteretic models
displacement ratios and ductility demands for SDOF and post-yield stiffness ratios, α.
systems with elastoplastic behavior considering soil
structure interaction. They modeled the soil structure Teff = [G (   1) 2  H (   1)3  1]Ti , for μ < 4 (7)
systems with effective period, effective damping and
effective ductility values different from fixed-base case.
For inelastic time history analyses, Newmark method Teff = [ I  J (   1)  1]Ti , for 4.0 ≤ μ ≤ 6.5 (8)
for step-by-step time integration was adapted in an in-
house computer program, and they proposed a new
equation for determining inelastic displacement ratio of   (   1)  
interacting system as a function of structural period of Teff =  K   1  1 Ti , for μ > 6.5 (9)
interacting system, strength reduction factor, and period   1  L(   2)  
lengthening ratio.
Some efforts have been made in the past to identify where G, H, I, J, K, and L are empirical coefficients,
an optimal effective period of equivalent linear SDOF given in Table 1.
system to minimize the error in the empirical procedures It is to be noted that the original references differ
for estimating inelastic response. Based on extensive slightly in definition, notation, and nomenclature of
time-history analyses of a number of SDOF hysteretic these periods, but used these for the same purpose,
systems in the mid-period range of 0.4-4.0 sec, subjected i.e. characterization of the equivalent linear elastic
to twelve earthquake time histories, Iwan (1980) SDOF system, to result in optimal estimate of the peak
proposed the following empirical formula for estimating displacement of a nonlinear structure. In the present
the optimal effective period of equivalent linear system: study, these have been collectively referred to as
“effective period” and their suitability for representing
the frequency tuning between the nonlinear soil and
Teff =[1  0.121(   1)0.939 ] Ti (5)
structure has been examined.
In a similar study, Kwan and Billington (2003)
proposed a more refined relationship for effective 5 Numerical study
(equivalent) period, considering six different hysteresis
models, period range of 0.1 to 1.5 sec, and four ductility
ratios in the range of 2 to 8. 5.1 Methodology

In the present study, the combined effect of soil


Teff  [0.8 C1 ]Ti (6) and structural nonlinearity on the total (inelastic)
response of SDOF systems has been studied. Four
where, the coefficient C1 varies between 0.43 to 0.55 for well defined geotechnical array sites, viz. KGWH02,
different hysteresis types and equals to 0.5 for elasto- KiK-net, Japan (Stewart et al., 2008), La Cienega,
plastic, moderately degrading, and slightly degrading Los Angeles (Stewart et al., 2008), Lotung, Chinese
hysteretic systems. Taiwan (Stewart et al., 2008), and Treasure Island near
Miranda and Ruiz-Garcia (2002) compared the error San Francisco bay area (Lopez, 2002), for which soil

Table 1 Values of empirical coefficients used in Eqs. 5-7 for estimating optimal equivalent
linear period of vibration of inelastic systems (FEMA 440, 2005)
Coefficient
Hysteresis type
G H I J K L
Bilinear 0.11 -0.018 0.09 0.14 0.77 0.05
Stiffness degrading 0.18 -0.037 0.15 0.16 0.92 0.05
540 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

layer properties are available from literature, have depth, obtained from the models by Darendeli (2001)
been considered. All the sites belong to ASCE 7-10 (as referred to in Stewart et al., 2008) have been used.
site class ‘D’ (VS,30 ranging from 181.5-260.8 m/sec) The effect of varying confining pressure on modulus
and represent a practical range of impedance contrast reduction behavior of low- (PI=0) and high-plasticity
( i 1VS, i 1 / iVS, i , where ρi and VS,i are density and shear (PI=10) soil has also been considered. For the La Cienega
wave velocity, respectively of the ith layer) between site, 13 material curves have been used for different
different layers. As shown in Table 2, a number of depths and soil type as suggested by Stewart et al. (2008)
strata have been generated from these sites by assuming based on laboratory testing of the site material. For the
bedrock at four different depths to study the effect of soil Lotung site, three material curves for different depths as
depth. The actual depths of the soil profiles at KGWH02, provided by Stewart et al. (2008), based on the work of
La Cienega, Lotung and Treasure Island are 200 m, Zeghal et al. (1995), have been used. The material curves
251.5 m, 47 m, and 122 m, respectively. The other soil vary with depth to account for the influence of effective
profiles have been generated by varying the depth of the state of stress on the shape of the modulus reduction
bedrock while retaining the thickness and shear wave and damping curves. As reported by Lopez (2002), only
velocity of individual layers, and stratification from partial information on the properties and condition of the
the actual sites. Several studies have been conducted soil at Treasure Island are available in the literature. The
by researchers to determine the effect of deep soil on soil profile and shear wave velocity profile have been
seismic response (Jakka et al., 2015); however, this selected based on the works of De Alba et al. (1994) and
study aims to investigate the response of deep layered Gibbs et al. (1992), respectively. However, the response
soils as encountered in real site condition as compared spectra at the surface of the Treasure Island are available
to uniform soil deposit. for Loma Prieta Earthquake (1989), which was used to
The shear wave velocity models for KGWH02, La validate the ground response model assuming different
Cienega, and Lotung sites are adopted from Stewart et dynamic properties of the soil profile. A trial and error
al. (2008), while for the Treasure Island site, the shear sensitivity analysis procedure has been used, which
wave velocity model is adopted from Lopez (2002). produces the closest surface response spectra when
The shear wave velocity at the reference rock outcrop compared to the instrumental observation.
site has been considered as 760 m/sec, consistent with The response of all the sixteen strata (four sites with
the lower limit for ASCE 7-10 site class ‘B’. For higher four assumed bedrock depths) has been obtained from
shear wave velocities at reference rock outcrop, the equivalent linear 1-D wave propagation analysis using
estimated amplification factors are expected to increase the software “DEEPSOIL” (Hashash et al. 2011). The
marginally. For the KGWH02 site, 15 material curves equivalent linear capability of the software is based
representing different soil material properties along the on the pioneering work of Idriss and Seed (1968), and

Table 2 Comparisons of fundamental and effective site periods of different soil profiles considered for the numerical study

Bedrock Effective Site Period (Ts,eff) at


Site Depth Considered Fundamental Site Period (T0) EPGARO =
(m) 0.05 g 0.36 g
KGWH02 9 0.22 0.28 0.57
(Stewart et al., 2008) 29 0.51 0.55 0.92
56 0.84 0.87 1.21
200 0.69 0.89 1.21
La Cienega 9.9 0.2 0.2 0.27
(Stewart et al., 2008) 29.4 0.42 0.4 0.57
60.7 0.68 0.63 0.79
251.5 1.90 0.63 0.79
LOTUNG 9 0.26 0.28 0.55
(Stewart et al., 2008) 29.4 0.62 0.66 1.01
59.6 1.05 1.11 1.65
200 3.23 4.25 5.99
Treasure Island 11 (first stiff layer removed) 0.28 0.34 0.66
(Lopez, 2002) 28 0.61 0.69 1.21
88 1.34 1.45 1.65
200 2.44 2.56 3.92
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 541

Seed and Idriss (1970) as employed in the widely used S1. In the present study, Ss have been considered as 2.5
program SHAKE (Schnabel et al., 1972) and its more times EPGARO, and S1 equal to EPGARO, for the purpose
current version SHAKE91 (Idriss and Sun, 1992). of generating the target response spectrum on rock
Similar programs such as EERA (Bardet et al., 2000), outcrop). Although spectrum compatible motions are
NERA (Bardet and Tobita, 2001) and DTASSA (Sahin, unrealistic and have frequency distortions, these have
2015 a, b) can also be used for one-dimensional ground certain useful features such as distinct corner period and
response analysis. In the study, only vertical incoherency smooth spectral shape, which help in identifying some
is studied and spatial incoherency is ignored. Recently, important characteristics of site amplification which may
Tavakoli et al. (2016) carried out the seismic not be clearly identifiable in recorded time histories due
microzonation study of Babol, Iran and demonstrated a to large variability in spectral shape and corner period.
good agreement between the proposed ranges of natural Ergun and Ates (2014) demonstrated that the record-to-
periods and alluvial amplification ratios obtained from record variability of structural response caused by scaled
the 1-D analysis using DEEPSOIL and the experimental synthetic earthquake records is less than that caused
microtremor studies. by real earthquake records in the seismic analysis of a
To estimate the ground response on chosen sites, six storied building. Recently, Gkatzogias and Kappos
seven rock outcrop motions, made compatible with (2015) have used a set of spectrum compatible motions
ASCE 7-10 spectrum for site class ‘B’, have been for nonlinear dynamic analysis of concrete bridges.
considered. According to ASCE 7-10, site class B is To minimize the distortion, the seven ground
defined as reference “Rock” site with shear wave velocity motions were chosen from the PEER database (2011) to
in the range of (760-1500 m/s), and for this site class, the best match with the targeted spectra and modified using
coefficients Fa and Fv equal to 1 are recommended. In the wavelet-based spectral matching procedure (Mukherjee
present study, the shear wave velocity at the reference and Gupta, 2002; Hancock et al., 2008) to fine tune with
rock outcrop site has been considered as 760 m/s the ASCE 7-10 site class B spectra. Table 3 presents the
consistent with the lower limit for ASCE site class B. The details of the original ground motions used for spectrum
lower limit has been chosen because most of the ground compatibilization. Adhikary et al. (2014) have shown
motions on rock outcrop, used in the present study, that although the response spectra for the two types of
were recorded on sites with shear wave velocity close ground motions (recorded and spectrum compatible)
to this value. To illustrate the effect of soil nonlinearity, are significantly different, there is a good match in the
two values of target effective peak ground acceleration corresponding amplification factors.
(EPGARO), 0.05 g and 0.36 g, with corresponding long The ground motions obtained for the sixteen strata
period corner periods of 4 sec and 6 sec, respectively, (four depth and for the four sites, viz., KGWH02, La
were chosen. (Actually, ASCE 7-10, does not provide Cienega, Lotung and Treasure Island), have been used
response spectra in terms of EPGARO, but in terms of to obtain constant ductility spectra, using the software
spectral ordinates at short period, Ss and at 1 sec period, “BISPEC” (Hachem, 2004). Seven spectrum compatible

Table 3 Earthquake ground motions made compatible with the ASCE 7-10 site class ‘B’ spectrum
NGA Event Station Date Ms VS,30 (m/sec) Comp.
For EPGARO = 0.05 g
788 Loma Prieta Piedmont Jr high (CDMG station 58338) 10/18/89 6.93 895.4 045
315
797 Loma Prieta Rincon hill, 000 (CDMG station 58151) 10/18/89 6.93 873.1 000
090
804 Loma Prieta So. San Francisco, Seirra pt. 10/18/89 6.93 1020.6 115
(CDMG station 58539) 205
946 Northridge Antelope buttes, (CDMG station 24310) 1/17/94 6.69 821.7 000
For EPGARO = 0.36 g
284 Irpinia, Italy Auletta 11/23/80 6.9 1000 000
270
285 Irpinia, Italy Bagnoli Irpino 11/23/80 6.9 1000 000
270
797 Loma Prieta Rincon hill, 000 (CDMG station 58151) 10/18/89 6.93 873.1 000
090
1021 Northridge Lake Hughes #4 - Camp Mendenhall 1/17/94 6.69 821.7 000
542 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

rock outcrop motions corresponding to each EPGARO This parameter can be used to obtain the inelastic
(0.05 g and 0.36 g), two hysteresis models (the bilinear response of the structure with a given ductility demand,
and the stiffness degrading modified Clough models), directly from the elastic design spectrum. It can be seen
for four different ductility levels (μ = 2, 4, 6, and 8) with that the inelastic site amplification ratio, Aμ, represents
5% material damping and post-yield stiffness ratio α = the combined effect of the conventional elastic site
0.05, have been used. The total damping for an inelastic amplification factors F (= Fμ , for μ = 1) and inelastic
system consists of material and hysteretic damping. displacement ratio, Cμ.
Identical material (or viscous) damping equal to 5%
has been considered for all the systems, whereas the A = F C (12)
hysteretic damping is inherently taken into consideration
through the nonlinear time history analysis by using the where Cμ is defined (Miranda, 2000; Ruiz-Garcia and
bilinear and the stiffness-degrading modified Clough Miranda, 2004) as the ratio of the inelastic spectral
hysteresis models. In the bilinear model, the stiffness and displacement to the elastic spectral displacement, for a
yield strength remain constant under the cyclic loading given initial period, Ti. It is conventionally used to obtain
and do not show any degradation or deterioration. The the inelastic response of the structure from the elastic
elasto-plastic model is a special case of Bilinear model design response spectrum of the codes. The Miranda
in which the post-yield stiffness ratio α is considered (2000) model for Cμ is given as
as zero. In the stiffness-degrading modified Clough
model, stiffness degradation is an important parameter,   1    C1 Ti   C2  
1

which represents the behaviour of reinforced concrete C = 1    1 e  (13)


structures. The reduction in unloading stiffness Kr with    
displacement is in the form of the expression:

where the coefficients C1 and C2 are defined as functions
D  of the displacement ductility ratio, μ, and site class. For
Kr  Ky  m  sites with average shear wave velocity higher than 180
D
 y  m/sec (firm sites), the values of coefficients C1 and C2
where β is the unloading stiffness degradation parameter, provided by Miranda (2000) are 12 and 0.8, respectively.
Ky is the initial elastic stiffness, and Dm is the previous For soft soil sites, as referred to in Ruiz-Garcia and
maximum displacement. In total, 224 cases of ground Miranda (2004), the inelastic displacement ratio Cμ is a
response analysis and 1904 cases of structural response function of ductility level μ, the normalized period (T/Tg),
(constant ductility spectra) analysis have been performed. and the various constants such as θ1, θ2, θ3, which depend
on the hysteretic behaviour of the system and the ground
5.2 Non-dimensional parameters motion ensembles considered in their study.

Two non-dimensional parameters have been 5.3 Results and discussion


defined in the present study to quantify the inelastic
site amplification effects. The first parameter, termed as Table 4 presents a comparison of the values of
inelastic site amplification factor, Fμ, is defined as the effective periods estimated using the four models
ratio of the inelastic spectral displacement at site to the described earlier. In the case of the FEMA 440 model, the
inelastic spectral displacement at rock outcrop, for a effective periods have been estimated for two hysteresis
given initial (elastic) period, Ti. types – Bilinear (BL), and Stiffness Degrading (SD) –
with a constant post-yield stiffness ratio, α = 0.05. It can
be observed from the table that the secant period (obtained
Sdsite (Ti ,  )
F = (10) from Rosenblueth and Herrera model) is much longer
Sdro (Ti ,  ) as compared to the effective periods obtained from the
other three models, which result in closer estimates for
This parameter is analogous to the site amplification low ductility demand (μ = 2, 4), but for larger values of
factors used in codes, with a difference that it is defined ductility demand, the Iwan (1980) model results in much
for inelastic response (for given ductility demand, lower values. The FEMA 440 and Kwan and Billington
μ), whereas the amplification factors in codes are for (2003) models result in close estimates for all the
elastic response. The second parameter, inelastic site values of ductility demand. Further, the effective period
amplification ratio, Aμ , is defined as the ratio of the depends significantly on the hysteresis type considered
inelastic spectral displacement at site to the elastic in the analysis, and this may be one of the reasons for the
spectral displacement at rock outcrop, for a given initial differences observed in different models. As the FEMA
(elastic) period, Ti. 440 model is based on a rigorous statistical analysis and
considers different values of α and different hysteresis
Sdsite (Ti ,  ) (11) types, the same has been considered in the present study
A = ro for further investigation.
Sd (Ti ,  = 1)
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 543

Table 4 Comparison of effective periods estimated by different models


Teff/Ti
Models Considered μ=2 μ=4 μ=6 μ=8
BL SD BL SD BL SD BL SD
FEMA 440 (α = 0.05) 1.09 1.14 1.51 1.63 1.79 1.95 2.07 2.21
Iwan (1980) 1.12 1.34 1.54 1.75
Kwan and Billington (2003) 1.13 1.60 1.95 2.26
Rosenblueth and Herrera (1964) 1.38 1.86 2.19 2.43
(Secant period)
Note: BL - Bilinear hysteresis, SD - Stiffness degrading hysteresis

Table 2 shows a comparison of effective period of It is interesting to observe that when plotted against
the site (Ts,eff) with the fundamental site period, T0 as normalised effective period, the peaks in all the plots of
calculated from Eq. 1, for the sites of varying depth and Fμ converge to Teff/Ts,eff close to unity. This demonstrates
subjected to motions with varying EPGARO. For low that the FEMA 440 relationships are reasonably accurate
amplitude motions (EPGARO = 0.05 g), Ts,eff is usually to estimate the effective period of inelastic structures, and
close to the fundamental site period, except in some cases, the site amplification effects in inelastic response have
where participation of higher modes in site response good correlation with the normalized effective period
is higher or comparable with that of the fundamental (Teff/Ts,eff), which can serve as a basis for development
mode. This happens in the case of soil strata having of future site amplification models. Here, amplification
sharp impedance contrast at an intermediate or shallow ratios for La Cienega site with 251.5 m strata and for
level. For example, in the case of the La Cienega site, Lotung site with 200 m deep strata, need some closer
there is a sharp impedance contrast at a depth of 60.7 m. examination. These deep strata have sharp impedance
As a result of this impedance contrast, the second mode contrasts at intermediate levels (at 60.7 m and 29.4 m,
of vibration in the case of 251.5 m deep strata, which has respectively). As a result, contribution of higher modes
the same period as the first mode of vibration in the case is quite significant in these cases and the peak in the Fμ
of the 60.7 m deep strata, dominates the site response. curve is not visible as prominently as in other cases. (In
For stronger motions (EPGARO= 0.36 g), Ts,eff elongates the case of the La Cienega site, the peaks corresponding
away from T0, illustrating the effect of soil nonlinearity to the 2nd, 3rd and 4th modes are even higher than that
on the effective site period. corresponding to the fundamental mode).
Figures 2-5 show the mean inelastic (arithmetic It is observed from Figures 2-5 that the effect of
mean for the seven spectrum compatible time histories) ductility demand on site amplification is more significant
site amplification factor Fμ for the sixteen strata, as for Teff/Ts,eff close to unity, whereas, the effect of ductility
functions of the initial period (Ti) and also as a function is much smaller for Teff/Ts,eff >1.5. Another interesting
of normalised effective period (Teff/Ts,eff), for the two observation is that plots of Fμ vs. Teff/Ts,eff for different
levels of EPGARO (=0.05 g and 0.36 g). As mentioned ductility demands intersect at Teff/Ts,eff close to 0.7 and 1.5,
before, the Fμ plot has been developed for different in almost all the cases, indicating a reversal in the trend
ductility levels (μ =1, 2, 4, 6, and 8) for both bilinear of variation with ductility demand, in different period
and stiffness degrading hysteresis types, however, only ranges. In the range 0.7< Teff/Ts,eff <1.5, Fμ decreases with
the results for the stiffness degrading hysteresis type are increased ductility demand, whereas outside this range,
presented here for brevity. Results for bilinear hysteresis Fμ increases with ductility demand.
type are similar and available in Adhikary (2014). The Table 5 presents the peak inelastic site amplification
inelastic site amplification factor plots show distinct factor (Fμmax) and corresponding period (Tpeak) for all the
peaks close to the effective site period, which elongates strata described in Table 2. It can be observed from the
with increasing depth of bedrock and EPGARO. However, Table that the peak inelastic site amplification factor
with increase in the ductility demand, the peak in the reduces with increase in ductility demand for all the
Fμ plot diminishes and shifts towards the left (shorter strata and both the values of EPGARO. Similarly, the
period). Due to frequency tuning between the site and peak inelastic site amplification factor also reduces
the structure (represented by equivalent single degree with increased EPGARO for all the strata and ductility
of freedom oscillator in response spectrum), the peak demands. As expected, the period corresponding to
amplification occurs when the effective period of site, peak inelastic site amplification factor increases with
Ts,eff, coincides with effective period of the structure depth, except in the case of the last two strata of the
Teff. As the ductility demand on the structure increases, La Cienega site. The periods corresponding to the peak
the Teff corresponding to a given Ti elongates, and as a amplification for these two strata match for μ = 1 but
result, the peak in Fμ vs. Ti plot shifts towards the left. differ for higher ductility demands. As higher modes
544 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

Table 5 Peak mean inelastic site amplification factor, Fμ and corresponding period, Tpeak, for different sites
considered in the present study

Site H (m) EPGARO (g) μ=1 μ =2 μ=4 μ=6 μ=8

Tpeak Fμmax Tpeak Fμmax Tpeak Fμmax Tpeak Fμmax Tpeak Fμmax
KGWH02 9 0.05 0.28 4.42 0.24 3.21 0.19 2.88 0.15 2.63 0.14 2.53
0.36 0.57 2.86 0.53 2.47 0.47 1.95 0.3 1.83 0.35 1.8
29 0.05 0.55 5.13 0.49 3.92 0.33 3.08 0.25 2.96 0.22 2.89
0.36 0.92 3.38 0.74 3.01 0.51 2.65 0.44 2.47 0.41 2.4
56 0.05 0.87 4.56 0.74 3.83 0.55 3.39 0.44 3.08 0.37 2.97
0.36 1.21 3.92 1.23 3.36 0.9 2.79 0.79 2.7 0.56 2.65
200 0.05 0.89 4.67 0.74 3.98 0.55 3.58 0.48 3.29 0.36 3.17
0.36 1.33 4.01 1.22 3.63 0.9 2.94 0.71 2.85 0.62 2.76
La Cienega 9.9 0.05 0.2 3.01 0.17 2.49 0.15 2.23 0.05 2.16 0.05 2.16
0.36 0.27 2.68 0.24 2.21 0.18 2.02 0.15 1.94 0.13 1.91
29.4 0.05 0.4 2.68 0.32 2.31 0.19 2.09 0.15 2.05 0.2 2.01
0.36 0.57 2.55 0.52 2.43 0.47 1.9 0.37 1.83 0.34 1.83
60.7 0.05 0.63 2.31 0.53 2.2 0.19 2 0.18 1.93 0.1 1.89
0.36 0.79 2.28 0.71 2.19 0.47 2 0.39 1.87 0.34 1.86
251.5 0.05 0.63 1.97 0.56 1.86 0.22 1.8 0.2 1.8 0.19 1.76
0.36 0.79 1.87 0.67 1.76 0.5 1.67 0.67 1.72 0.54 1.73
Lotung 9 0.05 0.28 2.82 0.25 2.43 0.19 2.21 0.15 2.02 0.14 1.95
0.36 0.55 2.27 0.52 2.01 0.34 1.73 0.27 1.67 0.34 1.65
29.4 0.05 0.66 2.78 0.55 2.58 0.47 2.1 0.39 2.02 0.35 1.99
0.36 0.66 2.78 0.55 2.58 0.47 2.1 0.39 2.02 0.35 1.99
59.6 0.05 1.11 2.36 1.09 2.33 1.07 2.14 0.92 2 0.85 1.95
0.36 1.65 2.13 1.66 2.04 1.38 1.94 1.38 1.81 1.26 1.78
200 0.05 4.25 2.17 4.63 1.99 3.91 2.06 3.67 2 3.35 2.1
0.36 5.99 2 5.57 2.02 4.53 1.81 4.83 1.76 4.25 1.78
Treasure Island 11 0.05 0.34 3.6 0.29 3.09 0.19 2.86 0.18 2.73 0.15 2.62
0.36 0.66 2.65 0.53 2.42 0.47 2.05 0.43 1.95 0.35 1.93
28 0.05 0.69 3.7 0.56 3.31 0.48 2.77 0.41 2.65 0.31 2.6
0.36 1.21 2.76 1.22 2.57 0.91 2.03 0.79 2.03 0.56 1.99
88 0.05 1.45 3.38 1.1 3.3 1.08 2.81 0.73 2.72 0.64 2.66
0.36 1.65 2.91 1.63 2.58 1.39 2.48 1.33 2.33 1.26 2.24
200 0.05 2.56 2.6 2.08 2.6 1.52 2.37 1.24 2.44 0.97 2.43
0.36 3.92 2.1 3.19 1.79 2.76 1.84 2.35 1.8 2.09 1.75

contribute significantly in the case of the 251.5 m deep of ASCE 7-10, and SAPE, for estimation of inelastic
strata, several peaks of more or less the same amplitude response has been examined by studying the inelastic
are observed in this case, and the curve is rather flat for site amplification ratio Aμ, obtained as a product of
higher ductility demands. In such a case, identifying the (elastic) site amplification factor, F, and inelastic
period corresponding to Fμmax becomes difficult and is displacement ratio, Cμ, and compared with the results of
not very reliable. the numerical study. For estimating Aμ using SAPE, and
As discussed earlier, conventionally, the effect of site ASCE 7-10, the Miranda (2000) model for Cμ has been
amplification is considered through multiplication factors considered. Adhikary et al. (2014) observed that SAPEf
0
applied on the elastic response spectrum. Adequacy of results in amplification factors closer to those obtained
the site amplification model (site coefficients Fa and Fv) numerically, as compared to SAPEVs,30,f for the selected
0
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 545

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Fig. 2 Mean inelastic site amplification factor, Fμ plotted against initial period, Ti and normalised effective period, Teff/Ts,eff for
KGWH02 site. The first two columns correspond to the EPGARO equal 0.05 g and the last two for 0.36 g. The four
rows correspond to the four depths of strata, i.e. 9 m, 29 m, 56 m, and 200 m, respectively

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Fig. 3 Mean inelastic site amplification factor, Fμ plotted against initial period, Ti and normalised effective period, Teff/Ts,eff for
La Cienega site. The first two columns correspond to the EPGARO equal 0.05 g and the last two for 0.36 g. The four
rows correspond to the four depths of strata, i.e. 9.9 m, 29.4 m, 60.7 m, and 251.5 m, respectively
546 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Fig. 4 Mean inelastic site amplification factor, Fμ plotted against initial period, Ti and normalised effective period, Teff/Ts,eff for
Lotung site. The first two columns correspond to the EPGARO equal 0.05 g and the last two for 0.36 g. The four
rows correspond to the four depths of strata, i.e. 9 m, 29.4 m, 59.6 m and 200 m, respectively

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Ti (sec) Teff/ Ts,eff Ti (sec) Teff/ Ts,eff

Fig. 5 Mean inelastic site amplification factor, Fμ plotted against initial period, Ti and normalised effective period, Teff/Ts,eff for
Treasure Island site. The first two columns correspond to the EPGARO equal 0.05 g and the last two for 0.36 g. The four
rows correspond to the four depths of strata, i.e. 11 m, 28 m, 88 m, and 200 m, respectively
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 547

sites, and therefore, only SAPEf has been considered for obtained numerically, for all the considered strata and two
0
the present comparison. values of EPGARO (=0.05 g and 0.36 g). The numerical
Figures 6-9 compare the amplification ratio Aμ, results shown in the figure have been obtained using
obtained from ASCE 7-10 and SAPE models, with those stiffness-degrading modified Clough hysteresis model.

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Fig. 6 Mean inelastic site amplification ratios for the KGWH02 site, obtained numerically, using the ASCE 7-10 amplification
factors, and from SAPE for varying ductility and EPGARO equal 0.05 g (left bin) and 0.36 g (right bin). The four rows
correspond to the four depths of strata, i.e. 9 m, 29 m, 56 m and 200 m, respectively
548 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Fig. 7 Mean inelastic site amplification ratios for the La Cienega site, obtained numerically, using the ASCE 7-10 amplification
factors, and from SAPE for varying ductility and EPGARO equal 0.05 g (left bin) and 0.36 g (right bin). The four rows
correspond to the four depths of strata, i.e. 9.9 m, 29.4 m, 60.7 m, and 251.5 m, respectively

It can be observed from the figure that there is a sharp increasing ductility demand. As a result, the numerically
peak in the elastic amplification ratio, corresponding to obtained amplification ratios for μ = 1 are significantly
the effective site period, which diminishes rapidly with higher than the site coefficients specified in ASCE 7-10
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 549

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Fig. 8 Mean inelastic site amplification ratios for the Lotung site, obtained numerically, using the ASCE 7-10 amplification factors,
and from SAPE for varying ductility and EPGARO equal 0.05 g (left bin) and 0.36 g (right bin). The four rows
correspond to the four depths of strata, i.e. 9 m, 29.4 m, 59.6 m and 200 m, respectively

in the vicinity of effective site period, but the difference numerical results follow a pattern similar to the SAPE
diminishes rapidly with increasing ductility demand. For model, but for other sites and higher EPGARO (=0.36 g)
the KGWH02 site subjected to EPGARO = 0.05 g, the the difference increases significantly. One of the reasons
550 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Ti (sec) Ti (sec)

Fig. 9 Mean inelastic site amplification ratios for the Treasure Island site, obtained numerically, using the ASCE 7-10 amplification
factors, and from SAPE for varying ductility and EPGARO equal 0.05 g (left bin) and 0.36 g (right bin). The four
rows correspond to the four depths of strata, i.e. 11 m, 28 m, 88 m, and 200 m, respectively

for this difference may be due to the ground motion (only 3% of data with PGA beyond 0.1 g)”. Whereas, the
database in the SAPE model (as indicated by Cadet et larger difference for other sites indicates high variability
al. 2012 b) being “dominated by moderate motion data of the site amplification effects with the local site (soil
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 551

strata) characteristics. are significantly higher than those obtained using ASCE
Further, due to high values of Cμ in the short period 7-10 site coefficients. However, the difference reduces
range (Figure 1), Aμ is overshadowed by the inelastic rapidly with increasing ductility demand. An elastic
effects in the structure, and the effect of frequency oscillator having period close to the effective period of
tuning between the site and structure is not visible as a site is in a ‘quasi-resonance’ condition with the site,
prominently as in the case of Fμ. In the short period resulting in a sharp peak in the elastic response spectrum
range, the value of Aμ increases rapidly with decreasing corresponding to the effective site period.
periods, and it is difficult to clearly identify a peak On the other hand, period of an inelastic oscillator
corresponding to effective site period. varies in a range between the initial and secant periods,
“depressing” (in words of Gazetas, 2012) the resonance.
This variance is manifested in terms of a gradual
6 Conclusions flattening of peak in the inelastic response spectra with
increasing ductility, and a corresponding dip in the
The effect of site amplification on inelastic seismic inelastic displacement ratio.
response of structures has been examined using a The results of the present study for the KGWH02 site
numerical study. A suite of spectrum-compatible ground subjected to low seismic intensity (EPGARO= 0.05 g) are close
motions was used to obtain response of 16 soil strata at to those of the SAPE model and follow a similar trend,
four well defined sites. The main focus of the study was whereas the difference increases for the other sites and
the phenomenon of frequency tuning between the soil higher values of EPGARO. To draw definitive conclusions
and the structure. For this purpose, different definitions of about the adequacy of current code provisions about site
characteristic periods of site and structure were examined amplification, the present study needs to be supported
with due consideration to their nonlinear behavior. It is by more results, including empirical studies. The study,
widely recognized in the literature that nonlinearity has however, clearly demonstrates the role of frequency
significant effect in both soil as well as structure, and tuning and ductility demand in the site amplification
the effective periods of both elongate with the intensity effects in inelastic structural response.
of shaking. Further, a higher mode may dominate the
response of a site having sharp impedance contrast at
an intermediate level. Accordingly, the effective period References
of a soil site has been chosen as the one corresponding
to the peak site to rock-outcrop spectral ratio, as it Adhikary S (2014), “Effect of Soil Conditions on the
takes into account the effects of soil nonlinearity as Seismic Response of Structures,” Ph.D. Dissertation,
well as the possibility of a higher mode dominating the Department of Earthquake Engineering, Indian Institutre
site response. In the case of structure (inelastic SDOF of Technology Roorkee, Roorkee, India.
oscillator) the effective period proposed in FEMA 440
Adhikary S, Singh Y and Paul DK (2014), “Effect of Soil
has been used. Although, it was originally conceived
Depth on Inelastic Seismic Response of Structures”, Soil
with a slightly different purpose (for minimization of
Dynamics and Earthquake Engineering, 61: 13–28.
error in equivalent linearization method of estimating
inelastic response), the present study shows that it ASCE 7-10, (2010), Minimum Design Loads for
serves well as the characteristic period to represent the Buildings and other Structures, American Society of
frequency tuning between site and structure. For low Civil Engineers, Reston, VA.
amplitude motions (EPGARO=0.05 g), the effective site Assimaki D, Li W and Fragiadakis M (2012), “Site
period Ts,eff is quite close to the fundamental period, Effects in Structural Response Predictions of Inelastic
T0, approximated as four times the average shear wave SDOF Oscillators,” Earthquake Spectra, 28(3): 859–
travel time from bedrock ground surface, whereas for 883.
stronger motions (EPGARO= 0.36 g), Ts,eff is much longer ATC 40 (1996), Seismic Evaluation and Retrofit of
than T0, illustrating the effect of soil nonlinearity on the Concrete Buildings, Applied Technology Council,
effective site period. Redwood city, California, USA.
To study the site amplification effects, two non-
Athanassiadou C J, Karakostas CZ, Margaris BN
dimensional parameters, Fμ and Aμ have been defined.
and Kappos AJ (2011), “Displacement Spectra and
The effect of frequency tuning between the site and
Displacement Modification Factors, Based on Records
structure is visible in both the parameters, which
from Greece,” Soil Dynamics and Earthquake
show a peak corresponding to the normalised effective
Engineering, 31(12): 1640–1653.
period, Teff/Ts,eff. However, Fμ is able to represent the
site amplification effects in inelastic structural response Bardet JP and Tobita T (2001), “NERA: A Computer
more clearly than Aμ, which is overshadowed by the large Program for Nonlinear Earthquake Site Response
effects of structure inelasticity in the short period range. Analyses of Layered Soil Deposits,” Department of
Due to frequency tuning between the site and the Civil Engineering, University of Southern California,
structure, in the vicinity of effective site period, the Los Angeles, CA, USA.
elastic amplification ratios obtained in the present study Bardet JP, Ichii K and Lin CH (2000), “EERA: A
552 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

Computer Program for Equivalent-Linear Earthquake Gazetas G (2012) “Some Presumptions on the Nature of
Site Response Analyses of Layered Soil Deposits,” Base Excitation May Erroneously Affect the Response
Department of Civil Engineering, University of Southern of Strongly Inelastic Systems”, Proceedings of the 15th
California, Los Angeles, CA, USA. World Conference on Earthquake Engineering, Sept.
Cadet H, Bard PY and Rodriguez-Marek A (2012a), 24–28, Lisbon, Paper No. 69.
“Site Effect Assessment Using KiK-Net Data: Part 1. Ghofrani H, Atkinson GM and Goda K (2013),
A Simple Correction Procedure for Surface/Downhole “Implications of the 2011 M9.0 Tohoku Japan Earthquake
Spectral Ratios”, Bulletin of Earthquake Engineering, for the Treatment of Site Effects in Large Earthquakes,”
10(2): 421–448. Bulletin of Earthquake Engineering, 11(1): 171–203.
Cadet H, Bard PY, Duval AM and Bertrand E (2012b), Gkatzogias KI and Kappos AJ (2015), “Deformation-
“Site Effect Assessment Using KiK-Net Data: Part 2-Site Based Seismic Design of Concrete Bridges,”
Amplification Prediction Equation Based on f0 and Vsz,” Earthquakes and Structures, 9(5): 1045–1067.
Bulletin of Earthquake Engineering, 10(2): 451–489. Grant DN, Blandon CA Priestley MJN (2005),
Chenari RJ and Bostani Taleshani SA (2016), “Site Modelling Inelastic Response in Direct Displacement-
Response of Heterogeneous Natural Deposits to Based Design, Report 2005/03, IUSS Press, Pavia, Italy,
Harmonic Excitation Applied to more than 100 Case p. 104.
Histories,” Earthquake Engineering and Engineering Gulkan P and Sozen MA (1974), “Inelastic Responses of
Vibration, 15: 341. https://doi.org/10.1007/s11803- Reinforced Concrete Structure to Earthquake Motions”,
016-0326-0. ACI Journal Proceedings (71: ACI).
Darendeli M (2001) “Development of a New Family of Hachem MM (2004), “BISPEC: Interactive Software
Normalized Modulus Reduction and Material Damping for the Computation of Unidirectional and Bidirectional
Curves,” Ph.D. Dissertation, Department of Civil Nonlinear Earthquake Spectra,” (ASCE).
Engineering, University of Texas, Austin.
Hancock J, Bommer JJ and Stafford P J (2008), “Numbers
Dickenson SA (1994), “The Dynamic Response of of Scaled and Matched Accelerograms Required for
Soft and Deep Cohesive Soils During the Loma Prieta Inelastic Dynamic Analyses,” Earthquake Engineering
Earthquake of October 17, 1989,” Ph.D. Dissertation, and Structural Dynamics, 37(14):1585–1607.
University of California, Berkeley.
Hashash YMA, Groholski DR, Phillips CA, Park D and
Dobry R, Oweis I and Urzua A (1976), “Simplified Musgrove M (2011), DEEPSOIL 4.0, User manual and
Procedures for Estimating the Fundamental Period of tutorial.
a Soil Profile,” Bulletin of the Seismological Society of
Idriss IM and Seed HB (1968), “Seismic Response of
America, 66(4): 1293–1321.
Horizontal Soil Layers,” Journal of the Soil Mechanics
Dwairi HM, Kowalsky MJ and Nau JM (2007), and Foundations Division, ASCE, 94(SM4): 1003–1031.
“Equivalent Damping in Support of Direct Displacement-
Idriss IM and Sun JI (1992), “SHAKE91: A Computer
Based Design,” Journal of Earthquake Engineering,
Program for Conducting Equivalent Linear Seismic
11(4): 512–530.
Response Analyses of Horizontally Layered Soil
Ergun M and Ates S (2014), “Comparing of the Effects Deposits,” User’s Guide, University of California,
of Scaled and Real Earthquake Records on Structural Davis, California, pp.13.
Response,” Earthquakes and Structures, 6(4): 375–392.
Iwan WD (1980), “Estimating Inelastic Response
Eser M and Aydemir C (2011), “The Effect of Soil- Spectra from Elastic Spectra,” Earthquake Engineering
Structure Interaction on Inelastic Displacement and Structural Dynamics, 8(4): 375–388.
Ratio of Structures,” Structural Engineering and
Jakka R S, Hussain M, and Sharma M L (2015), “Effects
Mechanics, 39(5): 683–701.
on Amplification of Strong Ground Motion Due to Deep
FEMA 273 (1997), NEHRP Guidelines for the Seismic Soils”, Geomechanics and Engineering, 8(5): 663–674.
Rehabilitation of Buildings, Federal Emergency
Kramer SL (1996), Geotechnical Earthquake
Management Agency ,Washington, D.C., USA.
Engineering, Pearson Education, Indian LPE, New
FEMA 356 (2000), Prestanard and Commentary for the Delhi.
Seismic Rehabilitation of Buildings, Federal Emergency
Kowalsky MJ (1994), “Displacement-Based Design-a
Management Agency,Washington, D.C., USA.
Methodology for Seismic Design Applied to RC Bridge
FEMA 440 (2005), Improvement of Nonlinear Static Columns,” Master’s Thesis, University of California at
Seismic Analysis Procedures, Federal Emergency San Diego, La Jolla, California.
Management Agency, Washington, D.C., USA.
Kwan WP and Billington SL (2003), “Influence
Gazetas G (2006), “Seismic Design of Foundations of Hysteretic Behavior on Equivalent Period and
and Soil-Structure Interaction,”, Proceedings of the 1st Damping of Structural Systems,” Journal of Structural
European Conference on Earthquake Engineering and Engineering, 129(5): 576–585.
Seismology, Paper no. K7.
Lermo J and Chávez-García FJ (1993), “Site Effect
No. 3 Adhikary S et al.: Effect of site amplification on inelastic seismic response 553

Evaluation Using Spectral Ratios with Only One Regnier J, Cadet H, Bonilla LF, Bertrand E and Semblat
Station,” Bulletin of the Seismological Society of JF (2013), “Assessing Nonlinear Behavior of Soils in
America, 83(5): 1574–1594. Seismic Site Response: Statistical Analysis on KiK-
Lopez F (2002), “Does Liquefaction Protect Overlying net Strong-Motion Data,” Bulletin of the Seismological
Structures from Ground Shaking?” M.Sc. Dissertation, Society of America, 103(3): 1750–1770.
European School of Advanced Studies in Reduction of Rosenblueth E and Herrera I (1964), “On a Kind of
Seismic Risk, Rose School, p.105. Hysteretic Damping,” Journal of Engineering Mechanics
Miranda E (1993), “Site-Dependent Strength-Reduction Division ASCE, 90: 37–48.
Factors,” Journal of Structural Engineering, 119(12): Ruiz-Garcia J and Miranda E (2004) “Inelastic
3503–3519. Displacement Ratios for Design of Structures on Soft
Miranda E (2000), “Inelastic Displacement Ratios Soils Sites,” Journal of Structural Engineering-ASCE,
for Structures on Firm Sites,” Journal of Structural 130(12): 2051-2061.
Engineering, 126(10): 1150-1159. Sahin A (2015a), “Dynamic Simulation Models for
Miranda E and Bertero VV (1991), “Evaluation of Seismic Behavior of Soil Systems-Part I: Block
Structural Response Factors Using Ground Motions Diagrams,” Geomechanics and Engineering, 9(2): 145–
Recorded during the Loma Prieta Earthquake,” 167.
CSMIP-1991. Sahin, A (2015b), “Dynamic Simulation Models for
Miranda E and Ruiz-Garcia J (2002), “Influence of Seismic Behavior of Soil Systems-Part II: Solution
Stiffness Degradation on Strength Demands of Structures Algorithm and Numerical Applications,” Geomechanics
Built on Soft Soil Sites,” Engineering Structures, 24(10): and Engineering, 9(2): 169–193.
1271–1281. Schnabel PB, Lysmer J and Seed HB (1972) “SHAKE:
Mollaioli F and Bruno S (2008), “Influence of Site A Computer Program for Earthquake Response Analysis
Effects on Inelastic Displacement Ratios for SDOF and of Horizontally Layered Sites,” Report No. EERC 72-12,
MDOF Systems,” Computers and Mathematics with Earthquake Engineering Research Center, University of
Applications, 55(2): 184–207. California, Berkeley, California.
Mukherjee S and Gupta VK (2002), “Wavelet-Based Seed HB and Idriss IM (1970), “Soil Moduli and
Generation of Spectrum-Compatible Time-histories”, Damping Factors for Dynamic Response Analyses.”
Soil Dynamics and Earthquake Engineering, 22(9): Report No. EERC 70-10, Earthquake Engineering
799–804. Research Center, University of California, Berkeley,
California, pp.40.
Mylonakis G and Gazetas G (2000), “Seismic Soil-
Structure Interaction: Beneficial or Detrimental?” Seed HB, Romo MP, Sun JI, Jaime A and Lysmer J
Journal of Earthquake Engineering, 4(3): 277–301. (1988), “The Mexico Earthquake of September 19,
1985—Relationships Between Soil Conditions and
Nakamura Y (1989), A Method for Dynamic
Earthquake Ground Motions,” Earthquake Spectra,
Characteristics Estimation of Subsurface Using
4(4): 687–729.
Microtremor on the Ground Surface, Railway Technical
Research Institute, Quarterly Reports. Seed HB, Ugas C and Lysmer J (1976), “Site-Dependent
Spectra for Earthquake-Resistant Design, Bulletin of the
Newmark NM and Hall WJ(1982), Earthquake Spectra
Seismological Society of America, 66: 221–243.
and Design, Earthquake Engineering Research Institute,
Berkeley, CA, 1982. Steidl JH, Tumarkin AG and Archuleta RJ (1996),
“What is a Reference Site?” Bulletin of the Seismological
Pavel F, Vacareanu R and Lungu D (2013), “Bi-
Society of America, 86(6): 1733–1748.
Normalized Response Spectra for Various Frequency
Content Ground Motions,” Journal of Earthquake Stewart JP and Center, Pacific Earthquake Engineering
Engineering, 18(2): 264–289. Research (2008), Benchmarking of Nonlinear
Geotechnical Ground Response Analysis Procedures,
Pavel F, Vacareanu R, Neagu C and Andrei P
Pacific Earthquake Engineering Research Center.
(2014), “Bi-Normalized Response Spectra and
Seismic Intensity in Bucharest for 1986 and 1990 Sullivan TJ, Calvi GM and Priestley MJN (2004), “Initial
Vrancea Seismic Events,” Earthquake Engineering Stiffness Versus Secant Stiffness in Displacement Based
and Engineering Vibration, 13: 125. https://doi. Design,” Proceedings of the 13th World Conference on
org/10.1007/s11803-014-0217-1. Earthquake Engineering, 1–6.
PGMD. PEER Ground Motion Database (BETA version Tavakoli HR, Amiri MT, Abdollahzade G, and Janalizade
as on October 10, 2011) (http://peer.berkeley.edu/peer_ A (2016), “Site Effect Microzonation of Babol, Iran,”
ground_motion_database); 2011. Geomechanics and Engineering, 11(6): 821–845.
Priestley MJN, Calvi GM and Kowalsky MJ (2007), Trifunac MD and Brady AG (1976), “Correlations of
Displacement-Based Seismic Design of Structures, IUSS Peak Acceleration, Velocity and Displacement with
Press, Pavia, Italy. Earthquake Magnitude, Distance and Site Conditions,”
554 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Vol. 18

Earthquake Engineering & Structural Dynamics, 4(5): Zeghal M, Elgamal AW, Tang HT and Stepp JC
455–471. (1995), “Lotung Downhole Array, II: Evaluation of
Veletsos AS and Newmark NM (1960), “Effect of Soil Nonlinear Properties,” Journal of Geotechnical
Inelastic Behavior on the Response of Simple Systems Engineering, 121(4): 363–378.
to Earthquake Motions,” Proceedings of the 2nd World Zhao JX, Irikura K, Zhang J, Fukushima Y, Somerville
Conference on Earthquake Engineering, 2: 895–912, PG, Asano A et al. (2006), “An Empirical Site-
Tokyo, Japan. Classification Method for Strong-Motion Stations in
Veletsos AS, Newmark NM and Chelapati CV (1965), Japan using H/V Response Spectral Ratio,” Bulletin of
“Deformation Spectra for Elastic and Elastoplastic the Seismological Society of America, 96(3): 914–925.
Systems Subjected to Ground Shock and Earthquake Zhai C, Li S, Xie L and Sun Y (2007), “Study on Inelastic
Motions,” Proceedings of the 3rd World Conference on Displacement Ratio Spectra for Near-Fault Pulse-
Earthquake Engineering, 2: 663–682, New Zealand. Type Ground Motions,” Earthquake Engineering and
Vijayendra KV, Prasad SK and Nayak S (2010), Engineering Vibration, 6: 351. https://doi.org/10.1007/
“Computation of Fundamental Period of Soil s11803-007-0755-x.
Deposit: A Comparative Study,” Oxford, UK, Second Ziotopoulou A and Gazetas G (2009), Non-Linear
Edition, Indian Geotechnical Conference, December Seismic Response Analysis of Soil Deposits and Piles and
16–18. Proposed Unique Bi-Normalized Spectrum, Research
Xu L and Xie L (2004), “Bi-Normalized Response Report LSM, NTUA-09-01.
Spectral Characteristics of the 1999 Chi-Chi Ziotopoulou A and Gazetas G (2010), “Are Current
Earthquake,” Earthquake Engineering and Engineering Design Spectra Sufficient for Soil-Structure Systems
Vibration, 3(2): 147–155. on Soft Soils?” Advances in Performance-Based
Earthquake Engineering (Springer), 79–87.

You might also like