You are on page 1of 23

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2003; 32:1749–1771 (DOI: 10.1002/eqe.300)

Soil–structure interaction in yielding systems

Javier Aviles1; ∗; † and Luis E. Perez-Rocha 2


1 Instituto Mexicano de Tecnologa del Agua; Jiutepec 62550; Morelos; Mexico
2 Centro de Investigacion Ssmica; Carretera al Ajusco 203; Tlalpan 14200; Mexico

SUMMARY
The eects of soil–structure interaction in yielding systems are evaluated, including both kinematic and
inertial interaction. The concepts developed previously for interacting elastic systems are extended to
include the non-linear behavior of the structure. A simple soil–structure system representative of code-
designed buildings is investigated. The replacement oscillator approach used in practice to account for
the elastic interaction eects is adjusted to consider the inelastic interaction eects. This is done by
means of a non-linear replacement oscillator dened by an eective ductility together with the known
eective period and damping of the system for the elastic condition. To demonstrate the eciency
of this simplied approach, extensive numerical evaluations are conducted for elastoplastic structures
with embedded foundation in a soil layer over elastic bedrock, excited by vertically propagating shear
waves. Both strength and displacement demands are computed with and without regard to the eect of
foundation exibility, taking as control motion the great 1985 Michoacan earthquake recorded at a site
representative of the soft zone in Mexico City. Results are properly interpreted to show the relative
eects of interaction for elastic and yielding systems. Finally, it is demonstrated how to implement this
information in the context of code design of buildings. Copyright ? 2003 John Wiley & Sons, Ltd.

KEY WORDS: eective damping; eective ductility; eective period; elastic system; yielding system;
soil–structure interaction

INTRODUCTION
firstly equivalen oscilator used by these
Although the eects of soil–structure interaction have been the subject of numerous inves-
tigations in the past, they have generally been examined at the exclusion of the non-linear
behavior of the structure. For elastic systems, the rst studies of interaction using an anal-
ogy with an equivalent single oscillator were made by Jennings and Bielak [1], Veletsos and
Meek [2], and Veletsos and Nair [3] for surface-supported foundations, and by Bielak [4]
for embedded foundations. They showed that the eects of inertial interaction could be su-
ciently approximated by merely modifying the fundamental period and associated damping of

∗ Correspondence to: Javier Avil


es, Instituto Mexicano de Tecnologa del Agua, Jiutepec 62550, Morelos, Mexico.
† E-mail: javiles@tajin.imta.mx

Received 5 September 2001


Revised 4 September 2002
Copyright ? 2003 John Wiley & Sons, Ltd. Accepted 3 January 2003
1750  AND L. E. PEREZ-ROCHA
J. AVILES 

the rigid-base structure. After these investigations, the increase in the natural period and
the change in the damping ratio resulting from the soil exibility and wave radiation have
been extensively studied by several authors, e.g. References [5–7], using as base excitation
the free-eld ground motion. Based on the same analogy, the eects of kinematic inter-
action on the relevant dynamic properties of the structure have also been evaluated [8–10],
considering the foundation input motion induced by dierent types of traveling seismic
waves. Now days, single oscilators are strictly applied
In its present form, the replacement oscillator approach is strictly valid only to account for
the elastic interaction eects. This approach, originally developed by Veletsos [11], forms the
basis of the seismic design provisions currently in use. Even though the non-linear behav-
ior of the structure is not considered, this approach has been implemented in major building
codes [12; 13] for the convenience of using standard free-eld response spectra in combination
with the eective period and damping of the system. Nevertheless, considering that in some
practical cases the eects of interaction may dier between elastic and yielding systems,
the current interaction provisions based on elastic response studies could not be directly ap-
plicable to seismic design of typical buildings, expected to deform considerably beyond the
yield limit during severe earthquakes.
The earliest studies of rigidly supported yielding structures were made by Veletsos and New-
mark [14]. By using pulse-type excitations and broad-band earthquakes, they derived simple
approximate rules for relating the maximum deformation and yield resistance of elastoplastic
structures to the corresponding values of an associated elastic structure. There are no similar
relationships taking the eect of foundation exibility into consideration. A more comprehen-
sive investigation is required to develop a better understanding of the eects of interaction in
non-linear systems, the results of which can provide the basis for the formulation of appro-
priate seismic design provisions for building structures.
The transient response of an elastoplastic surface-supported structure on an elastic half-space
has been examined briey by Veletsos and Verbic [15], who suggested that the structural
yielding increases the relative exibility between the structure and soil and hence decreases
the eects of soil–structure interaction. Based on the harmonic response of a bilinear hysteretic
structure supported on the surface of a viscoelastic half-space, Bielak [16] has shown that the
resonant structural deformation may be signicantly larger than would result if the supporting
soil were rigid. Yet it is not well understood how the displacement ductility demand of a
exible-base structure changes with respect to that developed in the associated rigid-base
structure. Practical solutions should be developed to estimate the yield resistance of a exibly
supported structure that is required to limit the displacement ductility demand to the specied
available ductility.
The eects of soil–structure interaction on the ductility of simple systems have not been
suciently elucidated thus far. This study is aimed at lling such a need. The main objec-
tives are: (1) to formulate an approximate procedure for the simplied analysis of non-linear
building-foundation systems; (2) to evaluate the relative importance of elastic and inelastic
interaction eects, as well as the inuence of the principal parameters involved; and (3)
to develop information that may be used in design for assessing the yield resistance and
maximum deformation of interacting systems through the non-linear analysis of xed-base
systems.
The soil–structure system considered in this study is formed by an elastoplastic one-
story structure placed on a rigid foundation embedded into a layer of constant thickness

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1751

underlain by a homogeneous half-space. This interacting system, although simple, ensures a


wide applicability of results because it satises various requirements suggested by building
codes. The eects of kinematic interaction due to the dierence between the foundation input
motion and the free-eld ground motion are incorporated for the important special case of
vertically incident shear waves. Because of the characteristics of this wave excitation, the
overall motion of the foundation consists of both horizontal and rocking components. Only
in the case of surface-supported foundation is the eective input motion the same as the
horizontal free-eld motion.
The concepts presented in this work are developed by reference to the replacement oscillator
generally used in building codes for assessing the elastic interaction eects. Herein a non-linear
oscillator is proposed, characterized by an eective ductility along with the known eective
period and damping of the system for the elastic condition. It is demonstrated that modifying
both the ductility factor as well as the relevant natural period and damping ratio of the rigid-
base structure supplies a reliable and practical mean to account for the inelastic interaction
eects. Both strength and displacement demands are computed for a large number of system
congurations, using as control motion the great 1985 Michoacan earthquake recorded at a
site representative of the soft zone in Mexico City. It is shown that, in some practical cases,
the eects of interaction for yielding systems may be equally important as those for elastic
systems.

SOIL–STRUCTURE MODEL

The soil–structure system under investigation is illustrated in Figure 1. It consists of a yield-


ing one-story structure supported by a rigid foundation embedded in a viscoelastic stratum of
constant thickness overlying a uniform viscoelastic half-space. The structure is characterized
by the height He , mass Me and mass moment of inertia Je about a horizontal centroidal axis.
The natural period and damping ratio of the structure for the elastic and xed-base conditions
are given by

Me
Te = 2 (1)
Ke
Ce
e = √ (2)
2 Ke Me

where Ce and Ke are, respectively, the viscous damping and initial stiness of the structure
when xed at the base. The foundation is assumed as a circular mat, perfectly bonded to the
surrounding soil, of radius r, depth of embedment D, mass Mc and mass moment of inertia
Jc about a horizontal centroidal axis. The layer is characterized by the thickness Hs , Poisson’s
ratio s , mass density s , shear wave velocity s and hysteretic damping ratio s . The corre-
sponding material properties of the half-space are dened by o , o , o and o . The one-story
structure may be viewed as representative of more complex multistory buildings responding
essentially as a single oscillator in their xed-base condition. In this case, it would be neces-
sary to interpret the parameters of the one-story structure as those of the multistory building
when vibrating in its xed-base fundamental mode.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1752  AND L. E. PEREZ-ROCHA
J. AVILES 

Figure 1. Single yielding structure placed on a cylindrical foundation that is embedded in a stratum
overlying a half-space, under vertically incident shear waves.

For all calculations performed in this study, the following system parameters were xed at
conventional values, namely: Mc =Me = 0:25, Jc =Je = 0:3, s =o = 0:8, s =o = 0:2, e = s = 0:05,
o = 0:03, s = 0:45 and o = 0:33. These values are intended to approximate typical building
structures and site conditions as those prevailing in Mexico City. The remaining system pa-
rameters control the structural response, so they were considered variable as indicated in a
later section.
The soil–structure system is subjected to vertically incident plane shear waves, propagating
along the z-axis with particle motion parallel to the x-axis. The horizontal displacement at
the ground surface generated by the free-eld motion is denoted by Ug . The presence of the
foundation modies the free-eld ground motion by the addition of diracted and scattered
waves, which results in a foundation input motion consisting of the horizontal and rocking
components denoted by Uo and o , respectively. The torsional component is not generated.
The degrees of freedom of the interacting system are the relative displacement of the structure
Ue , the displacement of the foundation Uc relative to the horizontal input motion Uo , and the
rocking of the foundation c relative to the rocking input motion o .

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1753

ELASTIC COMPUTATION OF IMPEDANCE FUNCTIONS AND INPUT MOTIONS

Fundamental steps in the analysis of soil–structure interaction are the elastic computations of
impedance functions and input motions for the foundation. Provided the foundation is rigid
and the soil behaves linearly, these quantities represent: the former, the frequency-dependent
springs and dashpots (Kmn and Cmn ; m; n = h; r) by which the supporting soil is replaced for
each vibration mode of the massless foundation; and the latter, the motion of the massless
foundation when subjected to the seismic excitation, resulting in the horizontal and rocking
input motions at the base.
The impedance functions are obtained by making use of an ecient numerical technique
based on the thin layer element method [17]. In this technique, the base of the stratum is
taken xed. This is not, however, a serious restriction because it is always possible to choose
a depth that is large enough to simulate the presence of an underlying half-space. To account
for the exibility of bedrock, the foundation was assumed here to be embedded in a surface
layer of thickness Hs , which is underlain by a stier stratum with rigid base of thickness 2Hs
representing the bedrock. The practical importance of using a rigorous numerical technique is
that the foundation embedment and layer depth aect considerably the springs and dashpots
by which the supporting soil is replaced. Probably, the most important eect is that, for a soil
layer, a cut-o frequency exists below which the radiation damping is not activated [18].
Once the impedance functions are known, the overall motion of the foundation is deter-
mined by application of the averaging method of Iguchi [19]. With this ecient numerical
technique suitable for embedded foundations, the foundation input motion is obtained by tak-
ing a weighted average of the free-eld displacements along the soil–foundation interface and
adding the displacement and rocking caused by the resultant force and moment associated with
the free-eld tractions along this surface. The overall motion of the foundation so obtained, un-
der harmonic excitation, is expressed by means of the transfer functions Qh (!) = Uo (!)=Ug (!)
and Qr (!) = o (!)=Ug (!), which relate the complex amplitudes of the horizontal and rocking
input motions, respectively, to the complex amplitude of the control motion on the ground
surface. For a stratum over a half-space, we have [6]:
2Ao
Ug = (3)
cos(ks Hs ) + ip sin(ks Hs )
where p = s s =o o is the impedance ratio between the layer and the underlying soil and
ks = !=s is the wave number for the stratum. Without loss of generality, one can assume
that 2Ao is the complex amplitude of the free-eld motion on the rock outcropping, i.e. on
the level of the ground surface but assuming there is no layer. It should be noticed that the
parameter p expresses the geometrical attenuation in the stratum due to the wave radiation
within the bedrock.
For calculations of impedance functions and input motions, material attenuation in the soil
was introduced through the use of complex shear wave velocities, i.e. s;∗ o = s; o (1 + is; o ), in
which the subscripts s and o denote the stratum and bedrock, respectively. Results are quite
sensitive to variations in the soil material damping. Thus, this factor may be essential for the
accurate determination of the structural response. The principal eects of energy dissipation
by hysteretic behavior of the soil are to increase the eective damping of the system and
hence to decrease the deformation of the structure, as demonstrated earlier for interacting
elastic systems [3].

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1754  AND L. E. PEREZ-ROCHA
J. AVILES 

The time histories of the foundation input motion for a specic earthquake are obtained by
formal application of discrete Fourier transforms, using the Fast Fourier Transform technique
and taking due precautions to minimize the error involved in the analysis of non-periodic
excitations. The fundamental steps involved in the Fourier analysis are: (1) to compute the
direct Fourier transform, Ug∗ (!), of the horizontal free-eld ground acceleration, Ug (t); (2) to
calculate the Fourier transforms of the horizontal and rocking components of the input accel-
eration of the foundation as Uo∗ (!) = Qh (!)Ug∗ (!) and   o∗ (!) = Qr (!)Ug∗ (!), respectively;
and (3) to compute the acceleration histories of the foundation input motion, Uo (t) and   o (t),
 ∗  ∗
by taking the inverse Fourier transforms of Uo (!) and o (!).
It should be pointed out that, for both the interacting system and the replacement oscillator,
the control motion is taken to be the purely horizontal free-eld motion on the ground surface,
i.e. the motion of the soil surface in absence of the foundation. The control motion is neither
the input motion nor the bedrock motion. On the other hand, it is necessary that the seismic
excitation used as control motion be consistent with the local site conditions.

ELASTIC DETERMINATION OF EFFECTIVE PERIOD AND DAMPING OF SYSTEM

As is known, the elastic interaction eects may be expressed by an enlargement in the natural
period Te and a change (generally an increase) in the damping ratio e of the structure assumed
with rigid base. The resulting enlarged period T̃e and increased damping ˜e are referred to
as the eective period and damping of the system for the elastic condition. They can be
determined using an analogy between the interacting system excited by the foundation input
motion and a replacement oscillator excited by the free-eld motion on the ground surface.
The mass of this equivalent oscillator is taken to be equal to that of the given structure.
Under harmonic base excitation, it is imposed that the resonant period and peak response
of the interacting system be equal to those of the replacement oscillator. The formulation
and validation of this approach to estimate the modied period and damping of the building
by both kinematic and inertial interaction is published elsewhere [9]. Introducing permissible
simplications, the values of these parameters are determined by the expressions:

T̃e ≈ T̃ei (4)

˜ie
˜e ≈ (5)
|Qh + (He + D)Qr |

where T̃ei and ˜ie are the eective period and damping of the system for inertial interaction only,
which can be obtained by well-established procedures ignoring kinematic interaction. With
this approach, the inuence of kinematic interaction is taken into account by considering the
base excitation to be unchanged (equal to the free-eld motion) while the system damping
is changed; the system period is practically not aected. By this means, the same overall
eect is achieved. It is interesting to notice that kinematic interaction is excluded if Qh = 1
and Qr = 0, with which the system damping for solely inertial interaction is recovered. This
happens with surface-supported foundations, the input motion of which is the same as the
free-eld motion.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1755

In this study, we will use the parameters T̃ei and ˜ie derived by Aviles and Suarez [20] for
structures with embedded foundation in a soil layer, which are given by

T̃ei = (Te2 + Th2 + Tr2 )1=2 (6)


T3 h Th2 r Tr2
˜ie = e e + 2
+ 2
(7)
3
T̃e 1 + 2h T̃e2 1 + 2r T̃e2

in which:

 1=2
Me
Th = 2 (8)
Khh
 1=2
Me (He + D) 2
Tr = 2 (9)
Krr

are the natural periods if the structure were rigid and its base were only able either to translate
or to rock, and

Chh
h = (10)
T̃e Khh
Crr
r = (11)
T̃e Krr

are the damping ratios of the soil for the translational and rocking modes of the foundation.
As the natural periods Th and Tr have to be evaluated at the eective frequency of the system,
!˜ e = 2= T̃e , an iterative process is required for calculating the system period, after which the
system damping is directly computed.

NON-LINEAR ANALYSIS OF SOIL–STRUCTURE MODEL

As has been considered, the non-linear behavior of the interacting system is restricted to the
structure only. The force–deformation relationship for the resisting element of the structure is
presumed to be of elastoplastic type. In this section we will describe the method of non-linear
analysis for the structure supported on the equivalent springs and dashpots and excited by the
foundation input motion. The equilibrium equations of the system in the time domain can be
written in matrix form as:
Ms Us (t) + Ps (t) = − Mo U o (t) − Jo 
 o (t) (12)
where Us = {Ue ; Uc ; c }T is the displacement vector of the system, whereas Mo and Jo are
load vectors and Ms is the mass matrix of the system, given in Reference [9]. Additionally,
Ps = {Pe ; Pc }T is the vector of internal forces of the system, Pc being the vector of interaction

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1756  AND L. E. PEREZ-ROCHA
J. AVILES 

forces of the soil acting on the foundation and Pe = Ce U̇e + Re . Here Re is the restoring force
of the structure limited by the yield resistance Ry = Ke Uy , with Uy being the yield deformation.
The analysis of the interacting system can be conducted either in the frequency domain
using harmonic impedance functions or in the time domain using impulsive impedance func-
tions. The frequency-domain analysis, however, is not practical for structures deforming in the
inelastic range. On the other hand, the time-domain analysis may be accomplished by use of
frequency-independent foundation models, so that constant springs and dashpots are chosen to
represent the foundation soil, as indicated by Wolf and Somaini [21]. With this simplication,
the convolution integral describing the soil interaction forces is avoided, and thus the integra-
tion procedure of the equilibrium equations is carried out as for xed-base yielding systems.
Calculations are usually performed with the values of stiness for zero frequency and the
values of damping for innite frequency. This is known as the doubly asymptotic approxima-
tion [22], since it is asymptotically exact at both low and high frequencies. To improve the
approximation, the equivalent springs and dashpots may be evaluated at other specic frequen-
cies, for example at the system frequency for the elastic condition, as done in this investigation,
or may be averaged over the frequency range of interest. This practical approach works better
for deep strata than for shallow strata, because the impedance functions for the latter vary
remarkably with the exciting frequency, especially for surface-supported foundations.
To compute the step-by-step non-linear response of the elastically supported structure, a
time-integration scheme based on the Newmark method was implemented. The details of
this method applied to xed-base yielding systems are well known and therefore will not be
repeated here. In this study, constant ductility response spectra are computed. They supply
the yield resistance of the exibly supported structure, as a function of the xed-base period,
that is required to limit the ductility demand to the allowable ductility. The ductility demand
developed in the structure for a particular earthquake is a term that refers to the ratio of its
maximum to yield deformations, that is:

Uemax
max = (13)
Uy

Constant ductility response spectra are calculated by iteration on the structural resistance Ry
until the ductility demand given by Equation (13) is the same as the allowable ductility
e = Um =Uy , with Um being the permissible deformation. The iterative process for a xed
value of Ke is stopped when the dierence between the computed and target ductilities is
considered satisfactory for engineering purposes. The tolerance chosen here was 1%. Due
precautions are taken when the ductility demand does not increase monotonically as the yield
strength decreases. This fact was rst observed by Veletsos and Newmark [23] analyzing
xed-base yielding systems. In this case, there is more than one strength corresponding to
the specied available ductility. For earthquake-resistant design only the largest strength is of
interest. This is the minimum strength that needs to be furnished to the structure in order to
limit the ductility demand to the target ductility.
It should be noticed that the non-linear analysis of the structure for some xed-base period
Te requires rst the foundation soil be replaced by constant springs and dashpots evaluated
for an excitation frequency corresponding to the exible-base period T̃e . Then the structure
excited by the foundation input motion is analyzed repeatedly until the yield resistance Ry
corresponding to the prescribed ductility factor e is found.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1757

NON-LINEAR REPLACEMENT OSCILLATOR

The ductile capacity of the structure is now examined by reference to a non-linear replace-
ment oscillator. We start postulating that the required strength for the actual exible-base
structure is similar to that for a modied xed-base structure, having the same mass as the
former and whose initial natural period and damping ratio are dened by the eective period
and damping of the system for the elastic condition. An equivalent ductility factor requires
further to be dened for a complete characterization of this non-linear replacement oscillator.
The practical advantage of this approach is that the required strength with interaction can be
estimated from standard inelastic response spectra for the prescribed free-eld ground mo-
tion, using the natural period, damping ratio and ductility factor modied by soil–structure
interaction.

Eective ductility of system

A model based on the equality of yield strengths and maximum plastic deformations under
monotonic loading is suitable for the derivation of an equivalent ductility factor for the re-
placement oscillator. The capacity of a non-linear structure to dissipate plastic energy under
cyclic loading is in fact larger than under monotonic loading. So, the value of the equiva-
lent ductility factor derived here may be considered to be a lower bound to the real value. The
force–deformation relationships for the resisting elements of the interacting system and
the replacement oscillator are shown in Figure 2. In these resistance diagrams, Uy and Ũy
denote the yield deformations of the given structure and the equivalent oscillator, whereas
Um and Ũm indicate the corresponding permissible deformations. Consequently, the ductility
factors are dened in each case as e = Um =Uy and ˜ e = Ũm = Ũy .

Force

Ry = Ry

Ke
Ke
1
1

Uy Uy Um Um
Displacement

Figure 2. Force-displacement relationships for the resisting elements of the interacting system (solid line)
and the replacement oscillator (dashed line), considering elastoplastic behavior.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1758  AND L. E. PEREZ-ROCHA
J. AVILES 

As the yield strength of the equivalent oscillator, R̃y = K̃ e Ũy , is the same as that of the
given structure, Ry = Ke Uy , it follows that the yield deformation of the latter is related to that
of the former by

K̃ e
Uy = Ũy (14)
Ke

In view of Ke = 4 2 Me =Te2 and K̃ e = 4 2 Me = T̃e2 , this expression can also be written as

Te2
Uy = Ũy (15)
T̃e2

On the other hand, as the equivalent oscillator has the same capacity of plastic deformation
as the given structure, it follows that

Ũm − Ũy = Um − Uy (16)

By using the denitions of e and ˜ e , this expression is transformed into

(˜ e − 1)Ũy = (e − 1)Uy (17)

Substitution from Equation (15) into Equation (17) results in what we shall call the eective
ductility of the system, that is:

Te2
˜ e = (e − 1) + 1 (18)
T̃e2

This reduced ductility factor, that appears to have been rst proposed by Muller and
Keintzel [24], controls the non-linear behavior of the replacement oscillator. It is apparent
from Equation (18) that 16˜ e 6e , implying that the eective ductility of the system is
lower than the allowable ductility of the structure. The eective ductility ˜ e will be equal
to the structural ductility e for innitely-rigid soil (for which T̃e = Te ) and to unity for
innitely-exible soil (for which T̃e = ∞). It is interesting to note that the eective ductility
of the system is independent of the seismic excitation. Nevertheless, it will be shown in the
next section that, for a particular earthquake, the reduction of ductility due to soil–structure
interaction is still eciently predicted by Equation (18). Even if not entirely accurate, this
expression gives results that are adequate for practical purposes.
It is a simple matter to show that this model also satises the equality of maximum plastic
energies. Referring again to Figure 2, during a characteristic cycle of plastic behavior with
ductility demand e for the given structure, or ˜ e for the equivalent oscillator, the dissipated
plastic energy in the former is given by EH = (Um − Uy )Ry , while in the latter by Ẽ H =
(Ũm − Ũy )R̃y . By equating these quantities, considering that Ry = R̃y , the relationship between
˜ e and e given by Equation (18) is conrmed. This seems to be the most rational way of
formulating a replacement oscillator with the same capacity of plastic energy dissipation as
the interacting system.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1759

Relationship between replacement oscillator and interacting system

With the basis on the analogy established between the replacement oscillator and the inter-
acting system, it is recognized that their resisting elements would experience the same yield
strength, that is:
Ry = R̃y (19)
In addition to this, both resisting elements would experience the same maximum plastic de-
formation, but dierent yield deformation due to the reduction from Ke to K̃ e . As the stiness
of the replacement oscillator represents the total stiness of two springs in series simulating
the exibilities of the structure and foundation, part of its yield deformation is developed
in the structural spring and the remainder in the soil spring. By substituting Uy = Um =e and
Ũy = Ũm = ˜ e into Equation (15), we nd that the maximum deformation of the elastically sup-
ported structure, Um , and that of the replacement oscillator, Ũm , are interrelated by
Te2 e
Um = Ũm (20)
T̃e2 ˜ e
Equations (19) and (20) represent the equivalence conditions between the replacement oscil-
lator and the interacting system. The latter condition is indeed a consequence of the former. It
is evident from Equation (20) that Um = (Te2 = T̃e2 )Ũm if e = 1 and therefore ˜ e = 1 according
to Equation (18). This fact has been demonstrated earlier for interacting elastic systems [2].
It is further evident that Um = Ũm when there is no interaction, i.e. T̃e = Te and ˜ e = e .
In view of the preceding considerations, it can be concluded that the actual exible-base
structure with natural period Te , damping ratio e and ductility factor e may be replaced by
a modied xed-base structure with enlarged period T̃e , increased damping ˜e and reduced
ductility ˜ e , determined by the procedure just outlined. It should be pointed out that the latter
structure must be analyzed non-linearly using as base excitation the free-eld ground motion.

NUMERICAL RESULTS

Results presented in this paper are computed for a control motion dened by the 19 September,
1985 Michoacan earthquake (EW component) recorded at site SCT located on the lakebed
zone in Mexico City. The acceleration and displacement traces of this recording are shown
in Figure 3. Also shown in this gure are the normalized, acceleration (Sa=g = Ry =Me g) and
displacement (Sd=Ugmax ) response spectra for elastic and inelastic behavior, at the exclusion of
soil–structure interaction. Here g is the acceleration of gravity and Ugmax is the peak ground
displacement.
In accordance with the one-dimensional wave propagation theory, the predominant period
of the site can be written as
Hs r
Ts = 4 (21)
r s
As the soil prole at site SCT is characterized by Hs = 38 m and s = 76 m=s, we have that
Ts = 2 s. This value is similar to that observed for the prevailing period of the excitation,

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1760  AND L. E. PEREZ-ROCHA
J. AVILES 

Figure 3. Top: acceleration and displacement time histories for the 1985 Michoacan earthquake at station
SCT in Mexico City. Bottom: acceleration and displacement response spectra without interaction for
e = 1 (solid line), 2 (dashed line) and 4 (dotted line).

interpreted as the period corresponding to the peak spectral acceleration for elastic behavior. To
cover a wide range of practical interest, the more important system parameters were assigned
the following values: He =r = 2 and 4, for the slenderness ratio of the structure; D=r = 0 and 1,
for the embedment ratio of the foundation; and Hs =r = 3 and 5, for the normalized thickness
of the stratum. Considering that Ts is xed, it is apparent from Equation (21) that variations
in Hs =r are due to changes in r=s . The latter parameter, known as wave transit time, is often
used to characterize the foundation exibility. Note that increasing values of Hs =r lead to
decreasing values of r=s . At the same time, Hs =r and r=s vary when r changes, since Hs and
s are also xed.

Reduction of ductility

Firstly, we show that the reduction of ductility due to soil–structure interaction may be ade-
quately predicted by the concept of eective ductility. The starting point of the examination
is the fact that the ductility demand for a particular earthquake is related to the yield strength
of the structure. In view of what has been discussed in the previous section, it is recognized
that the yield strength of the replacement oscillator for the eective ductility ˜ e equals that
required by the interacting system for the allowable ductility e . In other words, e = ˜ e rep-
resents the ratio between the ductility demands developed in both structures with the same
yield resistance. This is a consequence of the fact that, although the maximum inelastic ex-
cursion of the replacement oscillator is equal to that of the interacting system, their maximum
absolute deformations are dierent because of the dierence between the corresponding yield
deformations, according to Equation (15).
The reduction of ductility in terms of the relative stiness of the structure and soil, He =s Te ,
can be appreciated in Figure 4 for all combinations of the system parameters considered.
Comparisons are made between the eective ductility of the system and the reduced ductility

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1761

Figure 4. Comparison of the eective ductility of the system (curves) with the reduced ductility of the
replacement oscillator (points) imposed by the SCT recording of the 1985 Michoacan earthquake in
Mexico City. Results correspond to He =r = 2 (solid line and solid symbol) and 4 (dashed line and open
symbol), for e = 2 (lower plots) and 4 (upper plots).

of the replacement oscillator. Results marked with curves were computed by simple appli-
cation of Equation (18), evaluated using the exible-base period T̃e . The steps involved in
the computation of results marked with points are summarized next. For any given value of
He =s Te , the xed-base period Te is obtained, since the values of He =r and r=s are xed along
each graph. From this, the initial stiness Ke = 4 2 Me =Te2 is determined, providing that the
ratio Mc =Me is known and the value of Mc is assumed. Then the interacting system is analyzed
repeatedly until the yield strength Ry corresponding to the specied ductility e is found. Next
the replacement oscillator is assigned the same yield strength as the interacting system, and
it is then analyzed non-linearly considering the reduced stiness K̃ e = 4 2 Me = T̃e2 . Finally, the
ductility demand developed in the replacement oscillator, i.e. the ratio of its maximum to
yield deformations, is plotted as a point.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1762  AND L. E. PEREZ-ROCHA
J. AVILES 

It is shown in Figure 4 that the eective ductility of the system is in reasonable agreement
with the reduced ductility of the replacement oscillator. Although some discrepancies are
observed for relatively large values of He =s Te , they are irrelevant for engineering purposes
because those values correspond to unusual interaction conditions. For a representative story
height of 3:5 m, the ratio He =Te is approximately equal to 25 m=s for many types of buildings,
assuming the eective height as 0.7 of the total height and the xed-base period as 0:1 s of
the number of stories. For this reason, He =s Te = 0:5 would correspond to a very soft soil, with
shear wave velocity as low as s = 50 m=s. On the other hand, the results of Figure 4 provide
further conrmation of the fact that the eective ductility ˜ e equals the structural ductility
e for the limiting rigid-soil condition, s = ∞, and equals unity for the limiting exible-soil
condition, s = 0. It is interesting to note that, irrespective of the system conguration, the
reduced ductility by interaction tends to unity for short natural periods and to the allowable
ductility for long natural periods, as predicted precisely by Equation (18).
From these results it can be concluded that the reduction of ductility is controlled by the
relative stiness of the structure and soil. The eective ductility ˜ e decreases monotonically
with respect to the structural ductility e as the value of He =s Te increases. The inuence of
the other system parameters is relatively less important. It should be noted that the decrease
from e to ˜ e is due only to the reduction from Ke to K̃ e , without any change in the yield
strengths and the plastic deformation capacities of the interacting system and the replacement
oscillator.

Strength and displacement demands

In what follows, the validity of Equations (19) and (20) will be conrmed by comparison
of strength and displacement spectra determined approximately for the replacement oscillator
with those obtained rigorously for the interacting system. Results for elastic and inelastic re-
sponse are displayed in Figures 5 and 6 for He =r = 2 (squat) and 4 (slender), respectively.
To reduce the number of examples, for each structure height two cases are analyzed con-
cerning the foundation embedment and layer depth, namely: D=r = 0 (surface-supported) and
Hs =r = 5 (corresponding to r=s = 0:1 s); and D=r = 1 (embedded) and Hs =r = 3 (corresponding
to r=s = 1=6 s). In these graphs, the abscissas represent the xed-base period and the ordi-
nates represent the normalized, yield resistance and maximum deformation of the structure
interacting with the soil.
The degree of agreement between the response spectra for the replacement oscillator and
the interacting system is adequate for the required accuracy in engineering applications. Some
discrepancies are observed at the short-period region of the strength spectra for slender struc-
tures. They are not, however, of practical signicance because such structures normally fall
in the medium- and long-period spectral regions. This satisfactory agreement conrms the
eciency of the replacement oscillator approach to readily estimate strength and displacement
demands with interaction. To have a measure of its accuracy, we also evaluate the dierence
between the approximate and exact strength spectra, computing the ratio (R̃y − Ry )=Ry for
all cases exhibited in Figures 5 and 6. This relative error is plotted against the period ratio
Te =Ts in Figure 7. It can be seen that the percentage error is in general small, being less than
10% in most cases. Notice that errors are of more signicance when the structure period is
shorter than the site period. The degree of approximation involved in the strength spectra
is the same as in the displacement spectra, since Equations (19) and (20) are identical but

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1763

Figure 5. Acceleration (top) and displacement (bottom) response spectra with interaction for the SCT
recording of the 1985 Michoacan earthquake in Mexico City. Comparison of the exact solution for the
interacting system (thick line) with the approximate solution for the replacement oscillator (thin line).
Results correspond to He =r = 2, for e = 1 (solid line), 2 (dashed line) and 4 (dotted line).

Figure 6. Acceleration (top) and displacement (bottom) response spectra with interaction for the SCT
recording of the 1985 Michoacan earthquake in Mexico City. Comparison of the exact solution for the
interacting system (thick line) with the approximate solution for the replacement oscillator (thin line).
Results correspond to He =r = 4, for e = 1 (solid line), 2 (dashed line) and 4 (dotted line).

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1764  AND L. E. PEREZ-ROCHA
J. AVILES 

Figure 7. Relative error between the approximate and exact solutions for the replace-
ment oscillator and the interacting system, respectively, using all the acceleration response
spectra shown in Figures 5 and 6.

expressed dierently. Recall that the latter equation follows directly from the former by a
simple mathematical manipulation.
It can be seen in Figures 5 and 6 that, regardless of the system conguration, both the
spectral acceleration for very short period as well as the spectral displacement for very long
period are independent of the value of e , as imposed by structural dynamics. The value of
Sd for Te = ∞ is equal to the peak ground displacement. This is similar to what happens with
xed-base systems because soil–structure interaction is negligible at the long-period spectral
region. There are no theoretical indications regarding the value of Sa for Te = 0, but it tends
to be larger than the peak ground acceleration. This occurs if Sa is an increasing function of
the structure period, which is particularly true for highly peaked narrow-band response spectra
when Te ¡Ts .
The dierence between the response spectra of Figures 5 and 6 and those of Figure 3
reects the eects of soil–structure interaction. For the elastic case, the strength spectra with
interaction tend to shift towards the left of that without interaction. In view of the period
lengthening, the resonant response with interaction occurs for a structure period shorter than
the site period, this peak response being higher or lower than the xed-base value depending
on the decrease or increase in damping, respectively. We have found that the importance of
interaction increases with increasing the structure height and with decreasing the foundation
embedment and layer depth, which is in agreement with earlier ndings by other authors for
elastic systems. The eects of interaction may result in large increments or reductions of the
strength and displacement demands, with respect to the xed-base values, depending on the
period ratio of the structure and site. It is interesting to note that, for the xed-base case,
the inelastic displacement demands for natural periods somewhat longer than the site period
are considerably smaller than the elastic displacement demands. This eect is because the

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1765

Figure 8. Comparison of acceleration and displacement response spectra for total soil–structure interac-
tion (thick line) with those for inertial interaction only (thin line), considering the SCT recording of
the 1985 Michoacan earthquake in Mexico City. Results correspond to D=r = 1 and Hs =r = 3, for e = 1
(solid line), 2 (dashed line) and 4 (dotted line).

structural yielding tends to shift the structure period to the long-period spectral region, for
which the equal displacement rule is applied. By the same reasoning, soil–structure interaction
is observed to produce a similar eect, that is more pronounced for elastic than for yielding
systems.
The exact results for the case of embedded foundation given in Figures 5 and 6 for total
soil–structure interaction are also compared in Figure 8 with those obtained when excluding
the eects of kinematic interaction, i.e. analyzing the interacting system using as base excita-
tion the free-eld ground motion. It is conrmed that these eects are not very signicant. In
the whole period range, kinematic interaction reduces the strength and displacement demands
corresponding to inertial interaction only, irrespective of the structural ductility. For this rea-
son, the demands imposed by the free-eld ground motion may be considered to be an upper
bound to those imposed by the foundation input motion, even for slender structures that are
more eectively excited by rocking of the foundation.

Signicance of interaction for elastic and yielding systems

To have a measure of the importance of interaction for elastic and yielding systems, the
interaction factor
Sa(Te ; s ) Sd(Te ; s )
R (Te ) = = (22)
Sa(Te ; ∞) Sd(Te ; ∞)
Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1766  AND L. E. PEREZ-ROCHA
J. AVILES 

Figure 9. Variations of the interaction factor for the SCT recording of the 1985 Michoacan earthquake
in Mexico City. Results correspond to e = 1 (solid line), 2 (dashed line) and 4 (dotted line).

relating exible- to rigid-base response spectra was computed, using the exact results given in
Figures 5 and 6 for certain s and those of Figure 3 for s = ∞. This factor should be used
for assessing the response with interaction starting from that without interaction. The shapes
of R for elastic and yielding systems are compared in Figure 9. As can be seen, there is
not a smooth correlation between the values of R and Te . Despite this irregular behavior, it
is apparent that soil–structure interaction aects the response of slender structures adversely
(R ¿1) for Te ¡Ts , positively (R ¡1) in the period range Ts ¡Te ¡2Ts and negligibly (R = 1)
when Te ¿2Ts . It should be noticed that the high values of R at very short period are not
realistic for slender structures.
A recent study by Rodrguez and Montes [25] has concluded that the eects of interaction
in Mexico City are in general more important for elastic than for yielding systems, and they
are not very signicant in most of the inelastic cases. These conclusions are partially in
contradiction with our results, which reveal that such eects may be of large signicance for
slender structures, whether elastic or plastic. The variation of R with e does not show a
clear tendency indicating which systems are most inuenced by interaction. Unfortunately,
a direct comparison with the results reported in Reference [25] is not possible, since the
normalization used by these authors is inappropriate to represent conveniently the eects of
interaction. Specically, the values taken for the structure height, foundation dimension and
soil exibility, which are crucial parameters dened here by the ratios He =r and r=s , are
unknown. It is clear that the conclusions from that study should be applied only to the type
of building-foundation systems considered for evaluation.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1767

Approximate design procedure

Despite the simplicity of the soil–structure model investigated, it forms the basis of the current
design practice. A design procedure based on a similar model has already been established
by Veletsos [26] for interacting elastic systems, which permits the use of standard xed-base
response spectra. With the information that has been presented, this procedure may be adjusted
for interacting yielding systems. The steps involved in the analysis are summarized as follows:

1. From Equations (4) and (5), evaluate the eective period and damping of the system,
T̃e and ˜e , provided the values of Te and e for the rigid-base structure are known.
2. By use of Equation (18), evaluate the eective ductility of the system, ˜ e , provided the
value of e for the rigid-base structure is known.
3. From strength spectra for the prescribed free-eld ground motion, determine the yield
resistance of the system, Ry , corresponding to T̃e , ˜e and ˜ e , just as though the structure
were xed at the base.
4. If desired, determine the maximum deformation of the structure as Uemax = e Uy , once
the yield deformation Uy = Ry =Ke is obtained.
Let us illustrate the application of the proposed procedure by considering the site-specic
design spectrum specied by the Mexico City building code for the SCT site. A modied
version of this smoothed spectrum is given by:

  
Sa c T
= ao + 
− ao when T ¡Ta (23)
g  Ta
Sa  c
=  when Ta 6T 6Tb (24)
g 
  2  2
Sa  c Tb Tb
=  k + (1 − k) when T ¿Tb (25)
g  T T


where  = 1 + ( − 1)= k is the peak strength-reduction factor and  = (0:05=)0:6 is a scaling
factor used to account for the supplemental damping due to soil–structure interaction. As is
common practice, the nominal damping implicit in the design spectrum is 5%. The spectral
shape depends on the following site parameters: the peak ground acceleration, ao = 0:2; the
peak spectral acceleration, c = 1; the lower and upper periods of the at part of the spectrum,
Ta = 1:3 s and Tb = 2:4 s; and the ratio between peak ground displacement and peak spectral
displacement, k = 0:35. When applying Equations (23)–(25), the natural period T , damping
ratio  and ductility factor  should take the following values: Te , e and e , for the xed-base
case; and T̃e , ˜e and ˜ e , for soil–structure interaction.
Figure 10 depicts comparisons of strength spectra specied by the code (design spectra)
with those obtained from the control motion (response spectra). Results are given only for
the case of slender structure, considering both surface-supported and embedded foundations.
As can be seen, the response spectra with and without interaction are safely enveloped by the
corresponding design spectra in the whole period range. It should be noted that the spectra

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1768  AND L. E. PEREZ-ROCHA
J. AVILES 

Figure 10. Comparison between site-specic response (thin line) and design (thick line) spectra for
He =r = 4 and e = 1 (solid line), 2 (dashed line) and 4 (dotted line).

with interaction are valid for foundations with r=s = 0:1 and 1=6 s. As s = 76 m=s, results
for r=s = 0:1 s are applied to a foundation with r ≈ 7:5 m, whereas results for r=s = 1=6 s
are applied to a foundation with r ≈ 12:5 m. In consequence, the value of He =r = 4 would
correspond to a structure with He ≈ 30 m if r ≈ 7:5 m, of approximately 12 stories and Te ≈ 1:2 s,
or to a structure with He ≈ 50 m if r ≈ 12:5 m, of approximately 20 stories and Te ≈ 2 s. Having
known the values of the xed-base period, the required strengths with and without interaction
are directly obtained by going through the corresponding graphs.
Similar comparisons can be appreciated in Figure 11 between interaction factors relating
exible- to rigid-base design spectra and those obtained from response spectra, using the data

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1769

Figure 11. Comparison between interaction factors obtained from site-specic response (dashed line)
and design (solid line) spectra for He =r = 4, D=r = 1 and Hs =r = 3.

of Figure 10 only for the case of embedded foundation. Results are found to be in some
disagreement at the short-period spectral region. This discrepancy is more evident for elastic
than for yielding systems and is due to the conservatism involved in the ascending branch
of the xed-base elastic spectrum stipulated by the code. Although the representation is not
perfect, the interaction factors for design reproduce satisfactorily the tendencies observed in
those associated with the control motion, i.e. R ¿1 for Te ¡Ts and R 61 when Te ¿Ts . It
should nally be noted that while the greater increments of the strength and displacement

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
1770  AND L. E. PEREZ-ROCHA
J. AVILES 

demands arise in yielding systems, the greater reductions take place in elastic systems. Both
situations are conservative for systems deforming in the inelastic range, satisfying a practical
requirement.

CONCLUSIONS

An evaluation of the eects of soil–structure interaction in yielding systems has been made,
using an analogy with a non-linear replacement oscillator dened by an eective ductility
in addition to the eective period and damping of the system for the elastic condition. The
eciency of this simplied approach to readily estimate strength and displacement demands
was validated by comparison with results obtained rigorously. The eects of both kinematic
and inertial interaction were examined in the conditions of Mexico City, for which soil–
structure interaction is known to produce in some cases eects even more signicant than site
eects. Results were given for a representative site of the lakebed zone, showing that kinematic
interaction reduces the strength and displacement demands corresponding to inertial interaction
only, this reduction being of little practical importance. This is in agreement with earlier
ndings by other authors for dierent site conditions. The combined eects of foundation
exibility and structural yielding were found to be clearly benecial for slender structures with
natural period somewhat longer than the site period, but quite detrimental if the structure period
is shorter than the site period. There is no clear evidence whether elastic or yielding systems
are most inuenced by interaction. Results for other soft sites in Mexico City lead essentially
to the same conclusions. On the other hand, the application of the proposed procedure was
illustrated with a practical example, and some suggestions were made to devise more rational
code interaction provisions. Caution should be taken when using this procedure for pile-
supported structures, since the pile eects decrease the system period and increase the system
damping.

ACKNOWLEDGEMENTS

The authors appreciate the helpful comments of Eduardo Miranda and Daniel Resendiz to the original
manuscript. We are also in debt to the anonymous reviewers for their intelligent suggestions that
contributed to the clarity of this paper.

REFERENCES

1. Jennings PC, Bielak J. Dynamics of building–soil interaction. Bulletin of the Seismological Society of America
1973; 63:9 – 48.
2. Veletsos AS, Meek JW. Dynamic behavior of building–foundation systems. Earthquake Engineering and
Structural Dynamics 1974; 3:121–138.
3. Veletsos AS, Nair VVD. Seismic interaction of structures on hysteretic foundations. Journal of the Structural
Division (ASCE) 1975; 101:109 –129.
4. Bielak J. Dynamic behavior of structures with embedded foundations. Earthquake Engineering and Structural
Dynamics 1975; 3:259 –274.
5. Luco JE. Soil–structure interaction and identication of structural models; Proceedings of the ASCE Specialty
Conference in Civil Engineering and Nuclear Power: Tennessee, 1980.
6. Wolf JP. Dynamic Soil–Structure Interaction. Prentice-Hall: New Jersey, 1985.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771
SOIL–STRUCTURE INTERACTION IN YIELDING SYSTEMS 1771

7. Aviles J, Perez-Rocha LE. Evaluation of interaction eects on the system period and the system damping due
to foundation embedment and layer depth. Soil Dynamics and Earthquake Engineering 1996; 15:11–27.
8. Todorovska MI, Trifunac MD. The system damping, the system frequency and the system response peak
amplitudes during in-plane building–soil interaction. Earthquake Engineering and Structural Dynamics 1992;
21:127–144.
9. Aviles J, Perez-Rocha LE. Eects of foundation embedment during building–soil interaction. Earthquake
Engineering and Structural Dynamics 1998; 27:1523 –1540.
10. Aviles J, Suarez M, Sanchez-Sesma FJ. Eects of wave passage on the relevant dynamic properties of structures
with exible foundation. Earthquake Engineering and Structural Dynamics 2002; 31:139 –159.
11. Veletsos AS. Dynamics of structure-foundation systems. In Structural and Geotechnical Mechanics, Hall WJ
(Ed.). Prentice-Hall: New Jersey, 1977.
12. Applied Technology Council. Tentative Provisions for the Development of Seismic Regulations for Buildings;
ATC-3-06: California, 1984.
13. Federal Emergency Management Agency. Recommended Provisions for the Development of Seismic
Regulations for New Buildings; NEHRP-BSSC: Washington, 1997.
14. Veletsos AS, Newmark NM. Eect of inelastic behavior on the response of simple systems to earthquake
motions; Proceedings of the 2nd World Conference on Earthquake Engineering: Tokyo, 1960.
15. Veletsos AS, Verbic B. Dynamics of elastic and yielding structure-foundation systems; Proceedings of the 5th
World Conference on Earthquake Engineering: Rome, 1974.
16. Bielak J. Dynamic response of non-linear building-foundation systems. Earthquake Engineering and Structural
Dynamics 1978; 6:17–30.
17. Tassoulas JL, Kausel E. Elements for the numerical analysis of wave motion in layered strata. International
Journal for Numerical Methods in Engineering 1983; 19:1005 –1032.
18. Meek JW, Wolf JP. Insights on cuto frequency for foundation on soil layer. Earthquake Engineering and
Structural Dynamics 1991; 20:651– 665.
19. Iguchi M. An approximate analysis of input motions for rigid embedded foundations. Transactions of
Architectural Institute of Japan 1982; No. 315: 61–75.
20. Aviles J, Suarez M. Eective periods and dampings of building-foundation systems including seismic wave
eects. Engineering Structures 2002; 24:553–562.
21. Wolf JP, Somaini DR. Approximate dynamic model of embedded foundation in time domain. Earthquake
Engineering and Structural Dynamics 1986; 14:683–703.
22. Wolf JP. Soil–Structure Interaction Analysis in Time Domain; Prentice-Hall: New Jersey, 1988.
23. Veletsos AS, Newmark NM. Response spectra for single-degree-of-freedom elastic and inelastic systems. Report
RTD-TDR-63-3096; Air Force Weapons Laboratory: New Mexico, 1964.
24. Muller FP, Keintzel E. Ductility requirements for exibly supported antiseismic structures; Proceedings of the
7th European Conference on Earthquake Engineering: Athens, 1982.
25. Rodrguez ME, Montes R. Seismic response and damage analysis of buildings supported on exible soils.
Earthquake Engineering and Structural Dynamics 2000; 29:647 – 665.
26. Veletsos AS. Design concepts for dynamics of soil–structure interaction. In Developments in Dynamic
Soil–Structure Interaction, Gulkan P, Clough RW (Eds). Kluwer: The Netherlands, 1993.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:1749–1771

You might also like