SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL

Vlado Gicev (vgicev@gmail.com)

Department of Computer Science,
Goce Delcev University, Toso Arsov St. 14, 2000 Stip,
F.Y.R. of Macedonia

Abstract. A two-dimensional (2-D) model of a building supported by a semi-circular flexible

foundation embedded in nonlinear soil is analyzed. The building, the foundation, and the soil
have different physical properties. The model is excited by a half-sine SH wave pulse, which
travels toward the foundation. The results show that the spatial distribution of permanent,
nonlinear strain in the soil depends upon the incident angle, the amplitude, and the duration
of the pulse. If the wave has a large amplitude and a short duration, a nonlinear zone in
the soil appears immediately after the reflection from the half-space and is located close to
the free surface. This results from interference of the reflected pulse from the free surface
and the incoming part of the pulse that still has not reached the free surface. When the wave
reaches the foundation, it is divided on two parts—the first part is reflected, and the second part
enters the foundation. Further, there is separation of this second part at the foundation-building
contact. One part is reflected back, and one part enters the building. After each contact of the
part of the wave that enters the building with the foundation-building contact, one part of the
wave energy is released back into the soil. This process continues until all of the energy in
the building is released back into the soil. The work needed for the development of nonlinear
strains spends part of the input wave energy, and thus a smaller amount of energy is available
for exciting the building.

Keywords: soil-structure interaction; non-linear wave propagation; energy distribution

1. Introduction

Field reconnaissance of the effects of many earthquakes has provided numer-

ous examples of different types of soil failure and permanent deformations
caused by strong shaking. Examples include settlement of cohesionless soils,
liquefaction of saturated sands, flow slides due to liquefaction of cohesionless
soils, bulkhead failures due to backfill liquefaction, slides caused by lique-
faction of thin sand layers, failures of fills on weak foundations, and lateral
movement of bridge abutments. Many structures settle, tilt, or overturn on
liquefied soil. Some of the best-known examples of this occurred during the
1964 Alaska and 1964 Niigata earthquakes (Seed, 1970). The sequence of the

T. Schanz and R. Iankov (eds.), Coupled Site and Soil-Structure Interaction Effects 151
with Application to Seismic Risk Mitigation, NATO Science for Peace and Security
Series C: Environmental Security, 
c Springer Science+Business Media B.V. 2009
152 V. GICEV

soil-structure interaction (SSI) phenomena, which led to the overturning of

apartment buildings in Kawagishi-cho during the Niigata earthquake, is com-
plicated, and its complete modelling and analysis are still a major challenge
for any nonlinear numerical simulation. It probably started with development
of nonlinear strain zones in the soil close to the foundation, which in turn
expanded the trapped nonlinear energy to initiate liquefaction. We are assum-
ing here that the large energy of earthquake waves trapped in the zones of
strain localization initiated liquefaction (Trifunac, 1995), which then spread
all around the foundation, causing the buildings to tilt and overturn. Analysis
of this sequence is well beyond the scope of this paper, however. We will
describe only the early stages, which involve the creation of the nonlinear
zones of soil response.
Trifunac (1972) presented the analytical solution for interaction of the
wall sitting on an embedded semi-circular rigid foundation. Wong and Tri-
funac (1975) studied the wall–soil–wall interaction with the presence of two
or more shear walls, and Abdel-Ghaffar and Trifunac (1977) studied the soil–
bridge interaction with a semi-cylindrical rigid foundation and an input plane-
SH wave. Other studies have been conducted to analyze the influence of the
shape of a rigid foundation on the interaction. Wong and Trifunac (1974)
solved the interaction of the shear wall erected on an elliptical rigid foun-
dation for shallow and deep embedment, and Westermo and Wong (1977)
studied different boundary models for the soil-structure interaction of an em-
bedded, semi-circular, rigid foundation. They concluded that without a trans-
mitting boundary all of the models develop resonant behaviour and that the
introduced damping in the soil cannot adequately model the radiation damp-
ing. Luco and Wong (1977) studied a rectangular foundation welded to an
elastic half-space and excited by a horizontally propagating Rayleigh wave.
Lee (1979) solved a 3-D interaction problem consisting of a single mass sup-
ported by an embedded, hemispherical, rigid foundation for incident plane
P, SV, and SH waves in spherical coordinates. The recent publications deal
with a flexible foundation. Todorovska et al. (2001) solved an interaction of a
dike on a flexible, embedded foundation, and Hayir et al. (2001) described the
same dike but in the absence of a foundation. Aviles et al. (2002) analyzed the
in-plane motion of a 4-degrees-of-freedom model and Gicev (2005) studied
the soil-flexible foundation-structure interaction for incident-plane SH waves
with a numerical model using finite differences.
The soil-structure interaction phenomenon includes several features,
among them wave scattering, radiation damping, damping in the structure,
and the presence of different frequencies (system frequency, apparent fre-
quency, rocking frequency, horizontal frequency, and fixed-base frequency).
In this paper, in the presence of the interaction, the development of the
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 153

nonlinear zones in the soil is studied for incident pulses representing the near-
field destructive strong ground motion. The problems that must be addressed
in the numerical study of the nonlinear soil-structure interaction include
heterogeneities and discontinuities in the medium, the modelling of the free
surface, artificial boundaries, and keeping track of the nonlinear constitutive
law at each point in the soil. According to Moczo (1989) and Zahradnik
et al. (1993), the computational FD schemes that are used in applications
of wave propagation can be divided into homogenous and heterogeneous.
Alterman and Karal (1968) used the homogeneous formulation to solve
elastic wave propagation in layered media, and Boore (1972) proposed the
heterogeneous scheme. Tsynkov (1998) reviewed the existing global and
local artificial boundaries. The global boundaries are perfect absorbers, but
they cannot be readily applied in “marching-in-time” procedures because of
their non-locality, both in time and space. The main advantage of the local
(imperfect) artificial boundaries is that they are local in space and time and
are not frequency dependent.

2. Model

During the wave passage, the soil, the foundation, and the superstructure
undergo nonlinear deformations and permanent strains. Because the aim of
this paper is to study the nonlinear zones in the soil only—for simplicity—
only the soil is modeled as nonlinear, while the foundation and the building
are assumed to remain linear. The model is shown in Fig. 1. The incoming
wave is a half-sine pulse of a plane SH wave. A dimensionless frequency
2a a
η= = (1a)
λ β s · td0
is introduced as a measure of the pulse duration, where a is the radius of the
foundation, λ is the wavelength of the incident wave, β s is the shear-wave
velocity in the soil, and td0 is duration of the pulse.
To set up the grid spacing, the pulse is analysed in space domain (s), and
the displacement in the points occupied by the pulse is
π·s
w(s) = A sin , (1)
β s · td0
where A is the amplitude of the pulse and s is the distance of the considered
point to the wave front in initial time in the direction of propagation. Using
the fast Fourier transform algorithm, the half-sine pulse Eq. (1) is transformed
in wave number domain (k) as follows:
w(k) = F(w(s)). (2)
154 V. GICEV

ρb, βb

Hb
0 x
a ρf, βf

Hs = 5a
ρs, βs

Lm = 10a

Figure 1. Soil-flexible foundation-structure system.

The maximum response occurs for k = 0 (rigid-body motion). As k increases,

the response decreases and goes asymptotically to zero as k approaches infin-
ity. We selected the largest wave number, k = kmax , for which the k-response
is at least 0.03 of the maximum response (dashed lines in Fig. 2a). Then,
for this value of kmax , the corresponding frequencies and the corresponding
wavelengths are computed:
2π 2πβ
λmin = = . (3)
kmax ωmax
It can be seen from Fig. 2a that ωmax ≈ 245 rad/s for η = 0.5, while ωmax ≈
980 rad/s for η = 2.
A measure of the numerical accuracy of the grid is related to the ratio
between the numerical and physical velocity of propagation, r = c/β, which
ideally should be 1. The parameters that influence this accuracy are:
• the density of the grid m = λ/Δx (m is the number of points per wave-
length λ and Δx is the spacing between the grid points);
• the Courant number, χ = β s Δt/Δx;
• the angle of the wave incidence, θ.
It has been shown (Alford et al., 1974; Dablain, 1986; Fah, 1992) that the
error increases when m decreases, χ decreases, and θ is close to 0 or π/2. For
second-order approximation, the above authors recommend m = 12.
To compare hysteretic energies and the nonlinear zones in the soil, the soil
box should have the same dimensions for any dimensionless frequency of the
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 155

a) 1.0 1.0

h = 0.5 h=2
0.8 0.8

0.6 0.6
F(w) / Fmax(w)

F(w) / Fmax(w)
0.4 0.4

0.2 0.2

0.3 0.3
0.0 0.0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
w (rad/s) w (rad/s)

b)
u(m) u(m)
0.05 0.05

0.04 h = 0.5 0.04 h=2

0.03
0.03

0.02
0.02

0.01
0.01

0.00

0.00
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
t(s) t(s)
Figure 2. (a) Normalized one-side frequency response: η = 0.5 (left), η = 2 (right);
(b) filtered pulse: η = 0.5 (left), η = 2 (right).
156 V. GICEV

pulse, η. For that reason, we chose a rectangular soil box with dimensions
Lm = 10 · a and H s = Lm /2 = 5 · a (Fig. 1). Also, for merely practically
reasons, the maximum number of space intervals in the grid in the horizontal
(x) direction is set at 250 and in the vertical (y) direction at 400 (125 in the
soil box and 275 in the building). The minimum spatial interval for this setup
is Δxmin = Lm /250 = 95.5/250 = 0.382 m. For a finer grid, the computational
time increases rapidly. Having this limitation in mind, from Eq. (3) and for
η = 2 (ωmax = 980 rad/s), the shortest wavelength is λmin = 1.603 m, and the
finest grid density for this wavelength is m = λmin /Δxmin = 1.603/0.382 ≈ 4
points/λmin < mmin .
Our numerical scheme is O Δt2 , Δx2 , so from the above recommenda-
tions we should have at least m = 12 points/λmin to resolve for the shortest
wavelength, λmin . This implies that the pulse should be low-pass filtered. A
cut-off frequency ωc = 200 rad/s was chosen, and the pulse was low-pass
filtered (Fig. 2b). This implies that λmin = 7.854 m and then the grid density is
λmin 7.854
m= = ≈ 20 points/λmin > mmin . (4)
Δxmin 0.382
It can be seen in Fig. 2a (dotted lines) that for η = 0.5 only a negligible
amount of the total power is filtered out, while for η = 2 a considerable
amount is filtered out. Also, it can be seen in Fig. 2b that for η = 2 the
amplitude of the filtered pulse is smaller than the amplitude of the non-filtered
pulse, which is A = 0.05 m, while for η = 0.5 the amplitude is almost equal
with the amplitude of the non-filtered pulse. From numerical tests, it has been
shown that the viscous absorbing boundary rotated toward the centre of the
foundation reflects only a negligible amount of energy back into the model
(Gicev, 2005).
For 2-D problems, the numerical scheme is stable if the time increment
(Mitchell, 1969) is: ⎛ ⎞
⎜⎜⎜ ⎟⎟⎟
⎜⎜⎜ 1 ⎟⎟⎟
Δt ≤ min ⎜⎜⎜  ⎟⎟⎟ . (5)
⎜⎝ ⎟
β Δx1 2 + Δy1 2 ⎠
Further, we assume that the shear stress in the x direction depends only upon
the shear strain in the same direction and is independent of the shear strain
in the y direction (and vice versa for shear stress in the y direction). The
motivation for this assumption comes from our simplified representation of
layered soil, which is created by deposition (floods and wind) into more or
less horizontal layers. The soil is assumed to be ideally elastoplastic, and the
constitutive σ − ε diagram is shown in Fig. 3. Further, it is assumed that the
contacts remain bonded during the analysis and the contact cells C, D, E, F, G,
and H in Fig. 4 remain linear, as does the zone next to the artificial boundary
(the bottom four rows and the left-most and right-most four columns).
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 157

s (KPa)
m=0

m = ms m = ms
em e
s

m = ms

Figure 3. The constitutive law, σ − ε, for the soil.

Lb = 2a
y
Hb

x
H D
E
G T
4′
1′ 3′
B y′ 4 x′
2′ 3
S 1F
A
2G C

Figure 4. Numerical model with nonlinear soil. The points A, S, B, 1, 2, 1 , and 2 can
undergo permanent strains.

For our problem, the system of three partial differential equations (for u,
v, and w) describing the dynamic equilibrium of an elastic body is reduced to
∂
the third equation only (because u = v = ∂z = 0). Neglecting the body forces
in the z direction (Fz = 0), this equation is:
 
∂2 w ∂τ xz ∂τyz
ρ 2 = + . (6)
∂t ∂x ∂y
158 V. GICEV

Introducing the new variables v = ∂w/∂t, ε xz = ∂w/∂x, and εyz = ∂w/∂y, and
dividing (5) by ρ, the order of (6) is reduced to the system of three first-order
partial differential equations (PDE)

U,t = F, x +G, y, (7)

where
⎧ ⎫ ⎧1 ⎫ ⎧1 ⎫
⎪
⎪ v⎪ ⎪
⎪ τ xz ⎪
⎪ ⎪
⎪ τyz ⎪
⎪
⎨ ⎪
⎪ ⎪
⎬ ⎨ρ ⎪
⎪
⎪ ⎪
⎬ ⎨ρ ⎪
⎪
⎪ ⎪
⎬
U=⎪
⎪ε xz ⎪ , F = F(U) = ⎪ v ⎪ , G = G(U) = ⎪ 0 ⎪ . (8)
⎩ε ⎪
⎪ ⎪
⎭ ⎪
⎩ 0 ⎪
⎪
⎪ ⎪
⎪
⎭
⎪
⎩ v ⎪
⎪
⎪ ⎪
⎪
⎭
yz

The first equation in (7) represents the dynamic equilibrium of forces in the z
direction with neglected body force Fz , while the second and third equations
give the relations between the strains and the velocity. The abbreviations ε x =
ε xz , σ x = τ xz , εy = εyz , and σy = τyz are used later in the text. The Lax–
Wendroff computational scheme (Lax and Wendroff, 1964) is used for solving
Eq. (7) (Gicev, 2005).

3. Energy and Permanent Strain Distribution

As a test example, the properties of the Holiday Inn hotel in Van Nuys, Cal-
ifornia in the east–west direction are considered (Blume and Assoc., 1973).
A question arises about how to choose the yielding strain εm (Fig. 3) to study
permanent strain distribution. The displacement, the velocity, and the linear
strain in the soil (β s = 250 m/s) during the passage of a plane wave in the
form of a half-sine pulse are:
(  )
π s
w = A sin t− , (9)
td0 βs
π πt
v = ẇ = A cos , (10)
td0 td0
vmax πA
|ε| = = . (11)
βs β s td0
If, for a given input plane wave, we choose the yielding strain εm given
by (11) multiplied by some constant between 1 and 2, the strains in both
directions will remain linear before the wave reaches the free surface or
the foundation. This case can be called “intermediate nonlinearity”. If
we want to analyze only the nonlinearity due to scattering and radiating
from the foundation, we should avoid the occurrence of the nonlinear
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 159

strains caused by reflection from the half-space boundary. Then we may

choose εm = max(2πA sin γ/β s td0 ; 2πA cos γ/β s td0 ). We call this case “small
nonlinearity”.
If the soil is allowed to undergo permanent strains only due to wave pas-
sage of incident waves in the full space, then we may choose the maximum
strain εm < max (πA sin γ/β s td0 ; πA cos γ/β s td0 ). This condition guarantees
that in either the x or y direction the soil will undergo permanent strains
during the passage of the plane wave. Generally, the yielding strain can be
written as
vmax πA
εm = C =C , (12)
βs β s td0
where C is a constant that controls the yielding stress (strain) in the soil. We
then consider the following cases of nonlinearity, depending upon C:
1. C ≥ 2: Small nonlinearity. Permanent strain does not occur until the wave
hits the foundation with any angle of incidence.
2. 1 ≤ C < 2: Intermediate nonlinearity. Permanent strain does not occur
until the wave is reflected from the free surface or is scattered from the
foundation, for any angle of incidence. Permanent strain will or will not
occur after the reflection of the incident wave from the free surface, de-
pending upon the angle of incidence.
3. C < 1: Large nonlinearity. Permanent strain occurs after reflection from
the free surface. Permanent strain may or may not occur before the wave
reflects from the foundation surface, depending upon the angle of inci-
dence.

4. Energy Distribution in the System

The energy flow through a given area can be defined, in terms of a plane-wave
approximation (Aki and Richards, 1980), as:
td0
a
Ein = ρ s · β s · A sn v2 · dt, (13)
0

where ρ s and β s are density and shear-wave velocity in the soil and v is a
particle velocity, which, for the excitation considered in this paper, is given
by Eq. (10). Asn is the normal area through which the wave is passing. For
our geometrical settings of the soil (Fig. 1), the area normal to the wave
passage is:

A sn = 2 · H s · sin γ + Lm · cos γ = Lm · (sin γ + cos γ). (14)

160 V. GICEV

Inserting Eqs. (10) and (14) into (13) and integrating, the analytical solution
for the input wave energy into the model is
 2
π·A td0
Ein = ρ s · β s · Lm · (sin γ + cos γ) ·
a
· . (15)
td0 2
As can be seen from Eq. (15), for the defined size of the soil island, Lm ,
and the defined angle of incidence, γ, the input energy is reciprocal with the
duration of the pulse and is a linear function of the dimensionless frequency η
(Eq. (1a)). Because the short pulses are low-pass filtered up to ωc = 200 rad/s
(Fig. 2b), the analytical and the numerical solutions (13) for input wave en-
ergy do not coincide (Fig. 5). Since our system is conservative, the input
energy is balanced by:
• Cumulative energy going out from the model, Eout , computed using
Eq. (13); cumulative hysteretic energy (energy spent for creation and
development of permanent strains in the soil), computed from:

"
T end "
N
Ehys = Δt · σ xi (Δε xpi + 0.5 · Δε xei ) + σyi (Δεypi + 0.5 · Δεyei ) ,
t=0 i=1
(16)

Einp(KJ)

40000

30000

20000

10000

0
0 1 2
η
Figure 5. Input energy in the model: from analytic half-sine pulse (dashed line); from
low-pass filtered half-sine pulse (solid line).
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 161

where N is the total number of soil points; σ xi , σyi are the stresses at
the point i in the x and y directions, respectively; Δε xpi = εt+Δt
xpi − ε xpi is
t

the increment of the permanent strain in the x direction at point i; and

Δεypi = εt+Δt
ypi − εypi is the increment of the permanent strain in the y
t

direction at point i
• Instantaneous energy in the building, consisting of kinetic and potential
energy, which can be computed from:

"
N
Eb = Ek + E p = 0.5 · Δx · Δyb · ρ · v2i + μ · (ε2x + ε2y ) . (17)
i=1

In Fig. 6, this balance is shown for a pulse with η = 1.5, for incident angle
γ = 30◦ , and a yielding strain defined by C = 1.5 (Eq. (12)).
To study the effect of scattering from the foundation only, the building is
considered to be high enough so that the reflected wave from the top of the
building cannot reach the building-foundation contact during the analysis.
The analysis is terminated when the wave completely exits the soil island. In
this study, the hysteretic energy in the soil and the energy in the building are
the subjects of interest. In Fig. 7, these two types of energy are presented as

E (KJ)
20000
out
inp
E

+E
hys

ut
Eb + E

Eo

15000

10000

5000

Ehys

Eb
0
0.0 0.2 0.4 0.6
t (s)
Figure 6. Energy balance in the model for γ = 30◦ and η = 1.5.
162 V. GICEV

E (KJ)
β
s =
50
0m
β /s
s =1
00
0m

γ=
/s
1500 γ=

30
βs
60

=
10
γ = 3 βs =
0

00
25

m
0m

/s
γ = 60 /s

γ=
/s
0m

30
50

1000 βs =
=

2
s
β

50m
/s
γ=
60

βs = 2
50m/s
500 γ=3
βs = 2 0
50m/s
γ=
60

γ = 30 βs = 500m/s
γ = 60 γ = 60 β = 1000m/s
γ = 30 s
0
0 1 2
h
Figure 7. Hysteretic energy (solid lines) and energy entering the building (dashed lines) vs.
dimensionless frequency for intermediate nonlinearity C = 1.5.

functions of the dimensionless frequency η. Considering the energy entering

the building (dashed lines), the results confirm the expectations that as the
foundation becomes stiffer, a larger part of the input energy is scattered and
less energy enters the building. In contrast, the results for hysteretic energy
in the soil are not so straightforward. For an angle of incidence γ = 60◦ ,
the results are as would be expected—e.g., the hysteretic energy increases as
the foundation stiffness increases. However, this is not the case for the angle
of incidence γ = 30◦ . It can be seen that at some frequency intervals the
hysteretic energy can be larger for softer foundations. For example, for η ≤
0.7 and an angle of incidence γ = 30◦ , as the foundation becomes softer the
hysteretic energy becomes larger (the softest foundation we considered had
β f = 250 m/s). Similarly, a foundation with medium stiffness, β f = 500 m/s,
for η > 0.8 and an angle of incidence γ = 30◦ , causes the largest hysteretic
energy in the soil, where the maximum occurs at η = 1.5. This unexpected
behaviour of the soil can be explained by the destructive interference that may
occur with stiffer foundations, which decreases the released hysteretic energy
in the soil.
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 163

5. Distribution of the Permanent Strain in the Soil

Considering Fig. 8, and starting from dynamic equilibrium of the differential

pentahedron shown in the figure, we can find the principal stress at a point
and its direction as:
τzy
τzp = τzx cos γ + τzy sin γ and γ = tan−1 .
τzx
The principal permanent strain in√the soil is illustrated in Fig. 9a, b, c, for the
case of small nonlinearity (C = 3) for two angles of incidence, θ = 30◦ and
60◦ , and for three foundation stiffnesses, β f = 2500, 500, and 1000 m/s. This
value of C guarantees that for angles of incidence 30◦ ≤ γ ≤ 60◦ there is no
occurrence of permanent strain until the wave hits the foundation.
The principal permanent strain is illustrated in Fig. 10a, b, c, for the case
of intermediate nonlinearity (C = 1.5) for the same angles of incidence,
θ = 30◦ and θ = 60◦ and for three foundation stiffnesses, β f = 250, 500,
and 1000 m/s. In this case, permanent strain occurs before the wave hits the
foundation but after it reflects from the free surface.
For long pulses η = 0.1, it can be seen from Fig. 9a that for an angle of
incidence γ = 30◦ there is a small, permanent strain for the stiffest foundation
(β f = 1000 m/s) only, while for softer foundations the soil remains linear after
the pulse has left the model. For intermediate nonlinearity, shown in Fig. 10a,
for an angle of incidence γ = 30◦ it can be seen that after the creation of
nonlinear zones the effect of the interaction is negligible compared with the
effects of interference of the incoming wave and the reflected wave from the

τzp
1

τzx τzy

γ s
γ z
in
s

ss
co

x
s γ

y
Figure 8. Orthogonal and principal shear stresses on differential pentahedron.
164 V. GICEV

Figure 9. Principal permanent strain in the soil for: (a) η = 0.1; (b) η = 0.5; (c) η = 1, two
angles of incidence, and three foundation stiffness. Small nonlinearity in the soil C = 1.73.
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 165

Figure 10. Principal permanent strain in the soil for: (a) η = 0.1; (b) η = 0.5; (c) η = 1, two
angles of incidence, and three foundation stiffness. Small nonlinearity in the soil C = 1.5.
166 V. GICEV

free surface. This is not the case for γ = 60◦ . It can be concluded from Figs. 9a
and 10a that for stiffer foundations the effect of interaction is more dominant
than the effect of the interference. For the softest considered foundation, the
effect of the interaction on creation of nonlinear strains is small.
The observations are similar for a five-times-shorter pulse η = 0.5. It can
be seen from Figs. 9b and 10b that for the softest foundation the effect of
the interaction is negligible and that as the foundation becomes stiffer the
nonlinear zones are created and developed in the soil next to the front of the
foundation.
As the pulse becomes shorter, η = 1, it can be seen that nonlinear zones
are also formed behind the foundation. This can be explained by the inter-
ference of waves reflected from the free surface and diffracted around the
foundation. Again, the permanent strain in front of the foundation increases
as the stiffness of the foundation increases.

6. Conclusions

Numerical methods are powerful tools for studying nonlinear soil-structure

interaction problems. Because of grid dispersion, the selection of the grid
spacing must be done carefully. Short waves cannot be reconstructed even
with very fine grids, and the incident wave (pulse) should be low-pass filtered
to utilize numerical methods effectively.
In the presence of a foundation and small angles of incidence (close
to vertical incidence), the permanent strains in the y direction are domi-
nant, while for large angles of incidence (close to horizontal incidence) the
permanent strains in the x direction are dominant.
For long waves and small angles of incidence (Figs. 9a and 10a for θ =
30◦ , the effect of the interaction on the nonlinear response in the soil is small.
For soft foundations, β f = 250 m/s, and small incident angles (the top left
plots in Figs. 9a, b, c and 10a, b, c), the effect of the interaction on the non-
linear response of the soil is also small. As the foundation becomes stiffer,
zones of large permanent strains develop around the foundation. For stiff
foundations, short waves (η = 1 and η = 2, and large incidence angles, a zone
of permanent strains develops behind the foundation, which appears to be
due to the concentration of rays associated with diffraction of the waves from
the foundation. The zones of large permanent strains illustrated in Figs. 9a,
b, c and 10a, b, c are responsible for the damage and failures in the shallow
infrastructure (water and gas pipes, underground cables, etc.) that accompany
large earthquakes and cause interruptions of gas and water supplies (Trifunac
and Todorovska, 1997, 1998a, 1998b).
 SOIL-STRUCTURE INTERACTION IN NONLINEAR SOIL 167

As the large and permanent strains develop along the foundation-soil

interface, the effective foundation compliances are reduced, which decreases
the equivalent rocking stiffness of the foundation-structure system. With si-
multaneous action of in-plane wave motions, which are always present in
3-D settings during earthquake excitation and which will excite the in-plane
rocking of the model we studied in this paper, it is easy to see how the non-
linear zones in the soil (as illustrated in Figs. 9 and 10) will take the structure
one step closer to overturning or even eventual collapse, as in the examples
mentioned in the introduction.

References

Abdel-Ghaffar, A. M. and Trifunac, M. D. (1977) Antiplane dynamic soil–bridge interaction

for incident plane SH-waves. In Proc. 6th World Conf. Earthquake Eng., Vol. 2, New
Delhi, India.
Aki, K. and Richards, P. (1980) Quantitative Seismology, Theory and Methods, San Francisco,
W. H. Freeman & Co.
Alford, R. M., Kelly, K. R., and Boore, D. M. (1974) Accuracy of finite-difference modeling
of the acoustic wave equation, Geophysics 39, 834–842.
Alterman, Z. and Karal, F. C. (1968) Propagation of elastic waves in layered media by finite
difference methods, Bull. Seism. Soc. of Am. 58(1), 367–398.
Aviles, J., Suarez, M., and Sanchez-Sesma, F. J. (2002). Effects of wave passage on the relevant
dynamic properties of structures with flexible foundation, Earthq. Eng. Struct. Dynam. 31,
139–159.
Blume J. A. and Assoc. (1973) Holiday Inn. In L. M. Murphy (ed.), San Fernando, Cali-
fornia, Earthquake of February 9, 1971, U.S. Dept. of Commerce, National Oceanic and
Atmospheric Administration, Washington, D.C., pp. 359–394.
Boore, D. M. (1972) Finite difference methods for seismic wave propagation in heterogeneous
materials. In B. A. Bolt (ed.), Seismology, Surface Waves and Earth Oscillations, Vol. 11
of Methods in Comp. Physics, New York, pp. 1–37, Academic Press.
Dablain, M. A. (1986) The application of high-order differencing to the scalar wave equation,
Geophysics 51(1), 54–66.
Fah, D. J. (1992) A hybrid technique for the estimation of strong ground motion in sedimentary
basins, Ph.D. Dissertation, Swiss Federal Institute of Technology, Zurich, Switzerland.
Gicev, V. (2005) Investigation of soil-flexible foundation–structure interaction for inci-
dent plane SH waves, Ph.D. Dissertation, Dept. of Civil Engineering, Univ. Southern
California, Los Angeles, California.
Hayir, A., Todorovska, M. I., and Trifunac, M. D. (2001) Antiplane response of a dike
with flexible soil-structure interface to incident SH waves, Soil Dynam. Earthq. Eng. 21,
603–613.
Lax, P. D. and Wendroff, B. (1964) Difference schemes for hyperbolic equations with high
order of accuracy, Comm. Pure Appl. Math. XVII, 381–398.
Lee, V. W. (1979) Investigation of three-dimensional soil-structure interaction, Report No. CE
79-11, Dept. of Civil Engineering, Univ. of Southern California, Los Angeles, CA.
Luco, J. E. and Wong, H. L. (1977) Dynamic response of rectangular foundations for Rayleigh
wave excitation. In Proc. 6th World Conf. Earthq. Eng., Vol. 2, New Delhi, India.
168 V. GICEV

Mitchell, A. R. (1969) Computational Methods in Partial Differential Equations, New York,

Willey.
Moczo, P. (1989) Finite-difference technique for SH-waves in 2-D media using irregular
grids—application to the seismic response problem, Geophys. J. Int. 99, 321–329.
Seed, H. B. (1970) Soil problems and soil behavior. In R. L. Wiegel (ed.), Earthquake
Engineering, Englewood Cliffs, N.J., Prentice Hall, pp. 227–251.
Todorovska, M. I., Hayir, A., and Trifunac, M. D. (2001) Antiplane response of a dike
on flexible embedded foundation to incident SH-waves, Soil Dynam. Earthq. Eng. 21,
593–601.
Trifunac, M. D. (1972) Interaction of a shear wall with the soil for incident plane SH waves,
Bull. Seism. Soc. Am. 62(1), 63–83.
Trifunac, M. D. (1995) Empirical criteria for liquefaction in sands via standard penetration
tests and seismic wave energy, Soil Dynam. Earthq. Eng. 14(6), 419–426.
Trifunac, M. D. and Todorovska, M. I. (1997) Northridge, California, earthquake of 17 Jan-
uary 1994: Density of pipe breaks and surface strains, Soil Dynam. Earthq. Eng. 16(3),
193–207.
Trifunac, M. D. and Todorovska, M. I. (1998a) Nonlinear soil response as a natural passive
isolation mechanism—the 1994 Northridge, California earthquake, Soil Dynam. Earthq.
Eng. 17(1), 41–51.
Trifunac, M. D. and Todorovska, M. I. (1998b) The Northridge, California, earthquake of
1994: Fire ignition by strong shaking, Soil Dynam. Earthq. Eng. 17(3), 165–175.
Tsynkov, S. V. (1998) Numerical solution of problems on unbounded domains: A review,
Appl. Numerical Math. 27, 465–532.
Westermo, B. D. and Wong, H. L. (1977) On the fundamental differences of three basic soil-
structure interaction models, Proc. 6th World Conf. of Earth. Eng., Vol. 2, New Delhi,
India.
Wong, H. L. and Trifunac, M. D. (1974) Interaction of a shear wall with the soil for incident
plane SH waves: Elliptical rigid foundation, Bull. Seism. Soc. Am. 64(6), 1825–1884.
Wong, H. L. and Trifunac, M. D. (1975) Two-dimensional antiplane, building-soil–building
interaction for two or more buildings and for incident plane SH waves, Bull. Seism. Soc.
Am. 65(6), 1863–1885.
Zahradnik, J., Moczo, P., and Hron, F. (1993) Testing four elastic finite-difference schemes for
behaviour at discontinuities, Bull. Seism. Soc. Am. 83, 107–129.