You are on page 1of 7

Bull Eng Geol Environ (2010) 69:287–293

DOI 10.1007/s10064-009-0254-6

ORIGINAL PAPER

Effect of nonlinearity on site response and ground motion


due to earthquake excitation
Haydar Arslan • Bilge Siyahi

Received: 18 November 2008 / Accepted: 12 November 2009 / Published online: 8 January 2010
Ó Springer-Verlag 2009

Abstract Ground motions under earthquake loading are scientists trying to find the most appropriate ways of
influenced by the local soil conditions, but the non-linearity reducing the magnitude of earthquake-related hazards.
of the soil behavior makes it difficult to estimate the site Interpreting strong ground motions is one of the most
response. The paper reports a non-linear analysis developed important problems in geotechnical earthquake analysis,
to elucidate the effect of non-linearity on site response, which is undertaken to determine the site response and
illustrating its applicability with reference to a petroleum estimate the free-field ground shaking during an earth-
storage tank site subjected to earthquake excitation. quake. Although seismic waves generally travel tens of
kilometers in rock but \100 m in soil, the soil plays a very
Keywords Earthquake  Nonlinear  Site response  important role in determining the characteristics of ground
Ground motion motion (Kramar 1996).
A number of different techniques have been developed
Résumé Les mouvements du sol résultant de sollicita- for site response analysis since the 1920s. To approximate
tions sismiques sont influencés par les conditions locales, the actual non-linear, inelastic response of soils, an
mais la non-linéarité du comportement des sols rend dif- equivalent linear approach was proposed by Schnabel et al.
ficile l’estimation de la réponse du site. L’article présente (1972). In this approach, linear analyses are performed
une analyse non-linéaire développée afin d’élucider les with soil properties that are iteratively adjusted to be
effets de la non-linéarité sur la réponse d’un site, illustrant consistent with an effective level of shear strain induced in
son applicabilité par référence à un site de stockage de the soil. Yoshida (1994), Huang et al. (2001) and Yoshida
réservoirs de pétrole soumis à des sollicitations sismiques. and Iai (1998) showed that equivalent linear analysis
exaggerates the peak acceleration because the method
Mots clés Séisme  Non-linéarité  Réponse de site  calculates acceleration in the high frequency range.
Mouvement du sol

Previous studies on site response analysis


Introduction
The importance of site effects on seismic motion has been
Earthquake engineering deals with the effects of earth- realized since the 1920’s and many studies have been
quakes on people and environments, with engineers and undertaken. The amplification due to sediments is well
understood in terms of linear elasticity for the weak ground
H. Arslan (&) motion accompanying small earthquakes, but there has
University of Colorado, Boulder, USA been a debate regarding the amplification associated with
e-mail: arslan@colorado.edu the strong ground motions produced by large earthquakes.
As Field et al. (1997, 1998) explained, the view of geo-
B. Siyahi
Gebze Institute and Technology, Gebze, Turkey technical engineers, based largely on laboratory studies, is
e-mail: bilge.siyahi@gyte.edu.tr that Hooke’s law (linear elasticity) breaks down at larger

123
288 H. Arslan, B. Siyahi

strains causing a reduced (non-linear) amplification. Seis- Borja et al. (1999) developed a fully non-linear finite-
mologists, on the other hand, have tended to remain element (FE) model to investigate the impact of hysteretic
skeptical of this non-linear effect (Field et al. 1997), mainly and viscous material behavior on the downhole motion
because the relatively few strong-motion observations recorded by an array at a large-scale seismic test site in
seemed to be consistent with linear elasticity. Lotung, Taiwan, during the earthquake of 20 May 1986.
Since the 1970s, quantitative studies have been under- The constitutive model was based on a three-dimensional
taken using strong-motion array data. Several methods bounding surface plasticity theory with a vanishing elastic
have been proposed for evaluating site effects using region and accounts for shear stiffness degradation at the
ground-motion data, such as soil-to-rock spectral ratios onset of loading. The accuracy of the method proposed by
(e.g., Borcherdt 1970), a generalized inversion (e.g., Iwata Borja et al. (1999) is good, although the peak values were
and Irikura 1988; Boatwright et al. 1991), and horizontal- slightly under-estimated.
to-vertical spectral ratios (e.g., Joyner and Chen 1975; Lam Rodriguez-Marek et al. (2001) proposed an empirical
et al. 1978; Lermo and Chavez-Garcia 1993; Field and geotechnical seismic site response procedure that accounts
Jacob 1995; Yamazaki and Ansary 1997; Nakamura 1988; for the non-linear stress–strain response of earth materials
Bardet et al. 2000; Bardet and Tobita 2001; Arslan and under earthquake loading. In this study, the primary effects
Siyahi 2006; Siyahi and Arslan 2008). of material non-linearities are the increase of site period-
Analytical methods for site response analysis include icity and material damping as the intensity of ground
many parameters that could affect earthquake ground motion increases. The larger damping ratio is observed in
motions and corresponding response spectra. It is important lower spectral amplifications for all periods. However, the
to investigate the effect of these parameters on site effect of damping is pronounced for high frequency
response analysis in order to make confident evaluations of motion. Thus, soil damping significantly affected the peak
earthquake ground motions at the site. Seed and Idriss acceleration.
(1970), Joyner and Chen (1975) and Hwang and Lee In summary, there has been much research on site
(1991) investigated the effects of site parameters such as response analysis of the ground under earthquake loading.
secant shear modulus, low-strain damping ratio, types of The equivalent linear approach (Schnabel et al. 1972) is
sand and clay, location of water table and depth of bedrock. widely used for site response analysis. The paper compares
The parametric studies have shown that the secant shear the equivalent linear approach and the non-linear approach
modulus, depth of bedrock and types of sand and clay have proposed by Kramar (1996) to illustrate similarities and
a significant effect on the results of site response analysis. differences of the two methods.
However, the low-strain damping ratio and variations in
water table have only a minor influence.
For application in site response analysis, two basic Non-linear site response analysis
approaches have commonly been employed to represent
soil behavior during cyclic loading. The first, in which the The linear site response approach has been used because it
soil is modeled by a series of springs and frictional ele- is computationally convenient and provides reasonable
ments (Iwan model), uses Masing’s rules to establish the results for some practical cases (Kramar 1996). However,
shape of the cyclic, hysteresis curves (Seed et al. 1972). the non-linear and inelastic behavior of soil is well estab-
This model does not normally simulate the degradation lished in geotechnical engineering. For dynamic analysis,
observed due to cyclic loading of soils, nor does it provide the non-linearity means that the shear modulus of the soil is
a good simulation of the observed strain dependence of the constantly changing while the inelasticity means that the
shear modulus and damping ratio. Furthermore, application soil unloads along a different path than its loading path,
of Masing’s rules does not provide an adequate approxi- thereby dissipating energy at the points of contact between
mation for shear modulus and damping ratio simulta- particles. Both time domain and frequency-domain analy-
neously. In the second approach, damping is modeled as a ses are used to account for the non-linear effects in site-
viscous, rather than frictional, effect. This approach uses a response problems. Non-linear and equivalent-linear
pseudo-linear treatment and applies an iterative procedure methods are used in the time and frequency domain
in order to account for the strain dependence of modulus respectively for the one-dimensional analysis of shear
and damping (Schnabel et al. 1972). The main shortcoming wave propagation in layered soil media. When compared
of the linear method is its inability to take account of the with earthquake observation, non-linear analysis has been
strong strain dependence observed experimentally for shear shown to agree with the observed record better than the
modulus and damping ratio. The best that can be done with equivalent-linear analysis.
the linear model is to apply the method of iteration and to Kramar (1996) developed a non-linear approach
set values of shear. whereby a non-linear inelastic stress–strain relationship is

123
Effect of nonlinearity on site response and ground motion 289

followed in a set of small, incrementally linear, steps. The


soil medium is divided into sub-layers with absolute dis-
placements; uj, defined at the jth sub layer interface; and
shear stress (sj) defined at the mid-points of each interface.
As Kramar (1996) explained, the response of a soil deposit
under dynamic loading is governed by the equation of
motion:
os o2 u
¼q 2 ð1Þ
oz Dt
The differentiation for a soil divided into N sub-layers of
Fig. 1 Bedrock-halfspace interface representation for two layer soil
thickness Dz and for the small time increment (Dt) is
medium
calculated by using the finite difference method as:
os siþ1  si1
¼ ð2Þ
oz Dz
o2 u u_ i;tþDt  u_ i;t
¼ ð3Þ
ot2 Dt
o2 u
where u_ ¼ ou
ot is the velocity of the motion and ot2 ¼ ootu_ is
the acceleration.
Combining Eqs. 10, 11 and 12;
siþ1  si1 u_ i;tþDt  u_ i;t
¼q ð4Þ
Dz Dt
which can be simplified as:
Dt
u_ i;tþDt ¼ u_ i;t þ ðsiþ1;t  si;t Þ ð5Þ
qDz Fig. 2 Strain dependent shear modulus ratio
It must be noted that a shear stress of zero for the soil
surface and the appropriate boundary conditions for each
sub-layer must be satisfied. Joyner and Chen (1975) gravel (GM) is 100 m thick, with a bulk density of 195 kPa
proposed an equation for soil rock boundaries as: and a shear wave velocity Vs2 of 580 m/s. The shear wave
sr;t ¼ qr vsr ð2u_ r ðt þ DtÞ  u_ Nþ1;tþDt Þ ð6Þ velocity of the half-space interface is 1,000 m/s. The shear
modulus–strain relation is plotted in Fig. 2.
Using Eqs. 14 and 15 would satisfy the boundary Figure 3 shows the equivalent-linear approach which
conditions. Kramar (1996) gave the shear stress for each can be compared to the proposed non-linear solution.
layer as: SHAKE was used for the equivalent-linear solution and the
oui;t uiþ1;t  ui;t non-linear formulation was implemented into MATLAB.
ci;t ¼  ð7Þ As Figs. 3, 4 and 5 illustrate, the equivalent-linear
oz Dz
approach gives a higher acceleration as this site response
As can be seen from the above equations, the shear
analysis calculates acceleration in the high frequency
stress is related to the current shear strain and stress–strain
range.
history ðsi;t ¼ Gi ci;t Þ. Thus the proposed method satisfies
the non-linear and inelastic behavior of soil under cyclic
loading.
Non-linear site response simulation of an industrial
region
Numerical simulation
A non-linear site response analysis was undertaken for a
The effect of non-linearity has been investigated for a two- petroleum storage tank site between the Kucukcekmece
layer soil deposit as shown in Fig. 1. The total thickness of and Buyukcekmece Lakes, west of Istanbul and some
the soil deposits is 200 m. The upper silty sand (SM) has a 10 km from the ‘‘Main Marmara Fault’’ (Fig. 6). The effect
thickness of 70 m, a bulk density ðq1 Þ of 20 kPa and a shear of pile foundations on the site response analysis was
wave velocity (Vs1) of 410 m/s. The lower layer of silty investigated under earthquake loading.

123
290 H. Arslan, B. Siyahi

Fig. 3 Acceleration time history relation for the bottom soil layer Fig. 5 Acceleration response spectra for the top layer

Fig. 4 Acceleration time history relation for the top layer

Fig. 6 North Anatolian Fault-construction site


The petroleum storage tank area is underlain mainly by
Tertiary sedimentary rocks. Here the Cekmece Formation
consists of alternating layers of gravel and gravelly silty were chosen to represent an earthquake stress of 100 bars,
clayey sand, clay, marl, tuff-tuffite, claystone and sand- density of 2.8 g/cm3, Vs of 3.6 km/s, the attenuation in the
stone, increasing in depth toward the Marmara Sea. These propagation path geology (Kappa = 0.0035) and the
lagoonal lithologies often vary both horizontally and ver- probable ground response (NEHRP B/C boundary) condi-
tically. The Istanbul seismic zone is very distinct from the tions. The simulated horizontal acceleration based on the
neighboring tectonic units in its stratigraphy, absence of mean and mean ?1 standard deviation theoretical Fourier
metamorphism and lack of major deformation as the Intra- Amplitude Spectra (FAS) are presented in Figs. 7 and 8,
Pontide Suture of Late Triassic–Early Jurassic age sepa- respectively. As can be seen from the figures, the peak
rates the Istanbul and Sakarya Zones. ground accelerations are 0.53 g and 0.78 for the mean and
The ground acceleration which would be created by a mean ?1r FAS, respectively.
Mw = 7.4 earthquake on the Main Marmara Fault was The two-dimensional non-linear finite difference soft-
simulated. The parameters used in the simulation process ware Fast Lagrangian Analysis of Continua (FLAC 2000)

123
Effect of nonlinearity on site response and ground motion 291

was used for the site response analysis which considered the silty sand layer with a unit weight of 17 kPa and shear
coupled mechanical deformation of the pile structure where wave velocity of 225 m/s. Below is a high plasticity clay
each element behaves according to a prescribed linear or (CH) layer with unit weight of 20.0 KN/m3 and shear wave
non-linear stress/strain law in response to the applied forces velocity of 500 m/s.
or boundary restraints. The material may yield and flow The piles were modeled using beam elements. The
accordingly. The explicit Lagrangian calculation scheme viscous boundary conditions are specified along the side
and the mixed-discretization zoning technique used in FLAC boundaries of the model to take account of the wave
ensure that plastic collapse and flow are modeled accurately. radiation and minimize the wave reflections. Input earth-
quake motion is applied at the bottom boundaries to sim-
ulate the ground motion in Figs. 7 and 8. The simulation
Two-dimensional model presented in Fig. 7 is based on the mean theoretical Fourier
amplitude spectrum; the effects of standard deviation on
Based on the borehole data, two soil profiles were selected the analysis are shown in Fig. 8. The peak ground accel-
to reflect both the topographic and local soil effects. Both erations (0.53 g and 0.78 for mean and mean ?1r FAS) are
consist of a fill layer with a unit weight of 17.0 kPa and compatible with both the probabilistic and attenuation
shear wave velocity approximately 150 m/s overlying a based values.

Fig. 7 Simulated horizontal


acceleration time history

Fig. 8 Simulated acceleration


time histories considering
standard deviation

123
292 H. Arslan, B. Siyahi

plotted to determine the most critical regions under given


excitation. It can be seen that the maximum permanent
displacement is around 400–650 m while the maximum
deformation of the piles is about 5%.

Conclusions

This paper presents an overview of what is currently


known about the linear and non-linear site response
analysis.
A non-linear analysis was developed to investigate the
effect of non-linearity on site response analysis and com-
pared with the equivalent-linear method of analysis. It has
been shown that the effect of non-linear soil behavior is
one of the key factors for response spectra.
Maximum acceleration distribution along depth and
spectrum ratios has shown that equivalent-linear analysis
gives larger peak accelerations. The non-linear approach
was used to investigate the seismic behavior of an indus-
Fig. 9 Displacement–time history on the soil surface (without pile)
trial region with and without pile foundations. Two-
dimensional non-linear analysis showed that the interaction
between the components of motion and the pile/soil during
the earthquake affects the seismic soil response.

References

Arslan H, Siyahi B (2006) A comparative study on linear and


nonlinear site response analysis. Environ Geol 50(8):1193–2000
Bardet JP, Tobita T (2001) NERA: a computer program for nonlinear
earthquake site response analyses of layered soil deposits.
Department of Civil Engineering, University of Southern
California, Los Angeles, CA, 43 pp
Bardet JP, Ichii K, Lin CH (2000) EERA: a computer program for
equivalent-linear earthquake site response analyses of layered
soil deposits. Department of Civil Engineering, University of
Southern California, Los Angeles, CA, 37 pp
Boatwright J, Fletcher JB, Fumal TE (1991) A general inversion
scheme for source, site, and propagation characteristics using
multiply recorded sets of moderate-sized earthquakes. Bull
Seism Soc Am 81:1754–1782
Borcherdt RD (1970) Effects of local geology on ground notion near
San Francisco Bay. Bull Seism Soc Am 60:29–81
Borja RI, Chao HY, Montans FJ, Lin CH (1999) Nonlinear ground
response at Lotung LSST site. J Geotech Geoenviron Eng
Fig. 10 Displacement time history at the soil surface (with pile) 125(3):187–197
Fast Lagrangian Analysis of Continua (FLAC) (2000) Itasca
consulting group., Inc., Minneapolis
Maximum permanent displacements throughout the Field EH, Jacob KH (1995) A comparison and test of various site
response estimation techniques, including three that are not
profiles have been evaluated. The input accelerations at the reference-site dependent. Bull Seism Soc Am 85:1127–1143
model boundary resulting from ongoing wave propagation Field EH, Johnson PA, Beresnev IA, Zeng Y (1997) Nonlinear
vary slightly along the model boundary but are effectively ground-motion amplification by sediments during the 1994
equivalent to the original input motion. The time histories- Northridge earthquake. Nature 390:599–602
Field EH, Kramer S, Elgamal A-W, Bray JD, Matasovic N, Johnson
displacements at the top of the profiles without pile and PA, Cramer C, Roblee C, Wald DJ, Bonilla LF, Dimitriu PP,
with pile stiffness are shown in Figs. 9 and 10, respec- Anderson JG (1998) Nonlinear site response: where we’re at.
tively. Displacement time histories for different points are Seismol Res Lett 69:230–234

123
Effect of nonlinearity on site response and ground motion 293

Huang HC, Shieh CS, Chiu HC (2001) Linear and nonlinear Schnabel PB, Lysmer J, Seed HB (1972) SHAKE A computer
behaviors of soft soil layers using Lotung downhole array in program for earthquake response analysis of horizontally layered
Taiwan. Terr Atoms Ocean Sci 12:503 sites. Report no. EERC72-12, University of California, Berkeley
Hwang HHM, Lee CS (1991) Parametric study of site response Seed HB, Idriss IM (1970) Soil moduli and damping factors for
analysis. Soil Dyn Earthq Eng 10(6):282–290 dynamic response analysis. Report no. EERC70-10, University
Iwata T, Irikura K (1988) Source parameters of the 1983 Japan Sea of California, Berkeley
earthquake sequence. J Phys Earth 36:155–184 Seed HB, Whitman RV, Dezfulian H, Dobry R, Idriss IM (1972) Soil
Joyner WB, Chen ATF (1975) Calculation of nonlinear ground conditions and building damage in the 1967 Caracas earthquake.
response in earthquakes. Bull Seismol Soc Am 65:1315–1336 J Soil Mech Found Div ASCE 98:787–806
Kramar SL (1996) Geotechnical earthquake engineering. Prentice- Siyahi B, Arslan H (2008) Earthquake induced deformations of earth
Hall, New Jersey dams. Bull Eng Geol Environ 67(3):397–403
Lam I, Tsai CF, Martin GR (1978) Determination of site dependent Yamazaki F, Ansary MA (1997) Horizontal-to-vertical spectrum ratio
spectra using nonlinear analysis. Second international confer- of earthquake ground motion for site characterization. Earth-
ence on microzonation, San Francisco, CA quake Eng Struct Dyn 26:671–689 JSSMFE:14–31
Lermo J, Chavez-Garcia FJ (1993) Site effects evaluation using Yoshida N (1994) Applicability of conventional computer code
spectral ratios with only one station. Bull Seism Soc Am SHAKE to nonlinear problem. In: Proceedings of symposium on
83:1574–1594 amplification of ground shaking in soft ground
Nakamura Y (1988) On the urgent earthquake detection and alarm Yoshida N, Iai S (1998) Nonlinear site response analysis and its
system (UrEDAS). In: Proceeding of 9th world conference on evaluation and prediction. 2nd International symposium on the
earthquake engineering, Tokyo-Kyoto, Japan effect of surface geology on seismic motion, Yokosuka, Japan,
Rodriguez-Marek A, Bray JD, Abrahamson N (2001) An empirical pp 71–90
geotechnical seismic site response procedure. Earthq Spectra
17(1):68–88

123

You might also like