Soil Dynamics and Earthquake Engineering 15 (1996) 29-43

Copyright © 1995 Elsevier Science Limited

Printed in Great Britain. All rights reserved
0267-7261(95)00027-5 0267-7261/96/$15.00
ELSEVIER

Nonlinear' response of single piles under lateral

inertial and seismic loads
Deepak Badoni & Nicos Makris
Department of Civil ,Engineering and Geological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA

(Received 17 April 1995; accepted 21 July 1995)

A macroscopicmodel that consists of distributed hysteretic springs and frequency

dependent dashpots is utilized to model the lateral soil reaction and a practical
method based on one-dimensional finite element formulation is developed to
compute the nonlinear response of single piles under dynamic lateral loads. The
model is physically motivated, adequate for cohesive and cohesionless soils, and
involves standard geotechnical parameters. Only two parameters have to be
calibrated by fitting experimental data. Hysteretic and radiation damping are
modeled realisticallywithin the practical range of amplitudes and frequencies.The
model is calibrated and validated against five well instrumented full-scaleexperi-
ments and typical values for the range of the model-parameters are provided.
Subsequently, the developed model is utilized to study the nonlinear seismic
response of single piles. Finally, the developed method and the calibrated model
are used to predict the inertial and seismicresponse of one of the piles used in the
foundation of the Ohba bridge near Tokyo, Japan.

INTRODUCTION have been conducted with the finite element method. 6

The finite element method, which requires discretization
In the last two decades, several numerical and analytical of the pile and the surrounding soil, although powerful,
methods have been developed to compute the dynamic is computationally very expensive; and therefore cannot
stiffness and seismic response factors of pile-foundations be efficiently combined with structural dynamics codes
accounting for pile-soil-pile interaction. Most of these that compute the response of the superstructure. In
methods assumed linear viscoelastic response of the contrast, the Winkler foundation model is a versatile
surrounding soil. Nevertheless, under moderate and and economical approach, since the problem of pile-
strong seismic loading, pile-foundations undergo large soil interaction becomes one-dimensional. A Winkler
displacements and the behavior of the soil-pile system foundation model was adopted by Nogami et al. 7 who
can be strongly nonlinear. also studied the nonlinear pile response under dynamic
Realistic modeling of tile nonlinear dynamic response lateral loading. They proposed a six-parameter general-
of the pile-soil system is very important, especially for ized Kelvin model together with a lumped 'consistent'
the dynamic analysis of highway-bridges, overcrossings mass that involves an arbitrary value of some radius, in
and ramps, several of which experienced significant order to model the distributed soil reaction on the pile.
damage in the recent 199'4 Northridge earthquake. 1 In Recently, E1 Naggar & Novak 8 presented a similar shaft
most cases the bents and end-abutments of the above- model that combines springs, sliders and dashpots to
mentioned elevated structures are supported on pile approximate the behavior of the pile-soil interface, the
groups. Although we know little about the significance 'inner' field and the 'outer' field. Their model is restricted
of the pile-foundation-dynamics to the total response of to vertical motion only, and requires the value of the
a freeway overcrossing,2 there is ample documentation of radius of the 'inner' field. Furthermore, the ascending
pile distress and failure in the literature. Mizuno 3 reports and descending branches of the resulting hysteretic loop
on 28 cases involving seismic pile failure in Japan, are described by two different equations. Any reversal of
CNEL-ENEL4 documents pile rupture under two the loading must be detected, and the proper equation
bridges during the 1976 Fruili, Italy, earthquake and relating force to displacement must be used, resulting in a
Ross et al. 5 describe numerous failures of piles support- quite complex procedure.
ing bridges during the 1964 Alaska earthquake. The most comprehensive study on the nonlinear
Studies on the nonlinear dynamic response of piles response of piles using a Winkler foundation model is
29
30 D. Badoni, N. Makris

probably that ofTrochanis et al. 9 who utilized a model of embedded in a layered soil and subjected to lateral
viscoplasticity, better known as the Boue-Wen motion. The pile is considered as a linear elastic beam
model, 1°-12 to describe the force-displacement relation with circular cross-section area, Ap, diameter, d, moment
of distributed nonlinear springs that approximate the soil of inertia, Ip, Youngs modulus, Ep, and mass density pp.
reaction on the pile. Their study concentrated on the The surrounding soil is considered to be an inhomo-
static and quasi-static (zero-frequency limit) loading of geneous deposit having shear wave velocity, Vs, density,
piles, and for this type of loading it was shown that the ps, shear strength, S, or angle of internal friction q~, that
Bouc-Wen model predicts well the response of a variety vary with depth. The soil-pile interface is modeled as a
of soil-pile systems. Winkler foundation interacting with the pile through
Herein the distributed nonlinear spring described by continuously distributed nonlinear springs and linear
the Bouc-Wen model is combined with a distributed dashpots as sketched in Fig. 1.
viscous dashpot placed in parallel. The presence of the The problem of developing a macroscopic model that
damper makes the model very efficient for the prediction suitably predicts the response of a structural or mechan-
of the pile response under dynamic loads since it ical system does not have a unique solution. Different
accounts realistically for the energy that radiates out- constitutive models might approximate, within some
ward. It should be noted that in the original formulation tolerance, the response of a system for some range of
of his model, Trochanis 13 had included the presence of the frequency-spectrum and amplitude-intensities if their
the dashpot. In this paper, a physically motivated parameters are appropriately calibrated. Nevertheless,
formulation is developed so that most of the parameters constitutive models can vary from arbitrary mathe-
of the proposed nonlinear frequency-dependent Kelvin matical expressions that happen to fit experimental
model are readily associated with the physical soil data to physically motivated force-displacement rela-
parameters. For instance, the Young's modulus, the tions. Also, some constitutive models are more attractive
undrained shear strength (in the case of cohesive soils), than others because of their simplicity, the physical
and the angle of internal friction (in the case of cohesion- meaning of some of their parameters, their validity
less soils) are explicit parameters of the proposed model. over large frequency-spectra and load-intensities, and
Furthermore, while hysteretic damping is accounted for their consistency at limiting cases.
by the Bouc-Wen model, the distributed damping The above reasons have motivated for the formulation
coefficient that models radiation damping is provided of the constitutive model proposed herein, which is a
through a realistic frequency dependent expression 14 continuously distributed nonlinear spring in parallel with
which has been modified to allow for some reduction a frequency dependent linear dashpot. This model is the
when nonlinear behavior prevails. Accordingly, the extension at large deflections of the linear Kelvin model,
proposed model at the limit of small frequencies and the extension at finite frequencies of the hysteretic-
collapses to the solitary nonlinear spring studied by type nonlinear spring utilized by Trochanis et al. 9 The
Trochanis et al., 9 and at the limit of small amplitudes force resulting from the nonlinear spring alone is given
collapses to the linear frequency-dependent Kelvin by
model which has been used with success for the pre- Fs = a K y + (1 - a ) K y o ( (1)
diction of the linear inertial and seismic response of piles
and pile-groups. 15 18 where y is the pile deflection at the location of the spring,
In this paper a systematic examination of the per-
formance of the proposed model in predicting the lateral M(t)
response of single piles under inertial and seismic loading
is presented. Five well instrumented full-scale experi-
ments are selected where the single pile response was
recorded beyond the elastic limit. The proposed model is
shown to be capable of predicting the experimental data
with desirable accuracy, and the developed model is
subsequently used to compute the nonlinear seismic
response of single piles. Finally, the proposed method-
ology is used to compute the nonlinear dynamic response
of one of the piles used in the foundation of the Ohba
bridge near Tokyo, Japan.

PARAMETERS OF THE MACROSCOPIC

CONSTITUTIVE MODEL
Fig. 1. Schematic illustration of dynamic nonlinear pile-soil
The problem studied herein is that of a single pile interaction.
 Nonlinear response of single piles 31

K is some reference stiffness, a is a parameter that value A = 9 that was proposed by Broms 21 after experi-
controls the post yielding stiffness, Y0 is the value of mental and theoretical studies.
pile deflection that initiates yielding in the spring, and ~ is At shallow depths, the plane-strain assumption is
a hysteretic dimensionless quantity that is governed by inappropriate because of the non-zero vertical deforma-
the following equation tion of the soil during lateral motion of the pile. Accord-
ingly, near the surface parameter ), assumes smaller
Yo~ + 7[)[~[~[ n-' +/3)1(I n - A) = 0 (2)
values. Broms 21 proposed a value of A = 2 near the
In the above equation/3, 7, n and A are dimensionless surface whereas Matlock 2z used A = 3. Between the
quantities that control the shape of the hysteretic loop. surface and large depths the value of A increases pro-
The hysteretic model of (1) and (2) was originally gressively as plane-strain conditions prevail. Matlock 22
proposed by Bouc l° for n = 1, and subsequently assumed a simple form for the increment of A between
extended by Wen 11'12, and used in random vibration the minimum value )~min = 3 and )~max = 9:
studies of inelastic systems. The expressions given by O"z Z 6d
(1) and (2) are preferred to similar expressions used by A(z) = 3 + s Z< w (7)
%d
Trochanis et al., 9 since in the present formulation the S(z) + J
hysteretic parameter ff is dimensionless whereas in Tro-
chanis' formulation, ff has dimensions of length. 6d
It can be easily shown 19 that when A = 1 in (2), A(z)=9, z_> ~7s (8)
parameter K in (1) becomes the small-amplitude elastic S(z) + s
distributed stiffness, and parameter a becomes the ratio The second term in eqn (7) gives the increase with depth
of the post to pre-yielding stiffness. Based on this due to overburden pressure, and the third term repre-
observation, in all the subsequent analysis A = 1. By sents the restraint that even a weightless soil around a
eliminating the time variable in eqn (2) one obtains pile would provide against upward flow of the soil. The
d¢ value of parameter J in (7) must be determined empiri-
Y0 d)---~= 1 + (/7 4- 7)( n (3) cally. Matlock 2 reported that studies on the Sabine
profile indicated that a value of J of approx. 0.5 is
in which the (-4-) sign depends on whether the values o f ) satisfactory, whereas a lower value of about 0"25 fits
and ( are positive or negative. For a monotonic displace- the Lake Austin data somewhat better.
ment input the maximum value of ( is obtained when Broms 23 also proposed an analytical expression for the
d ( / d y = 0, and by virtue of (3) this maximum takes the ultimate value of Fs,max for cohesionless soils
value
1 + sin q~
( 1 ~ l/n Fs'rnax(Z) = #Tsd 1 - sin~b z (9)
~max = ~ - - ~ ] (4)
where 7s = Psg is the specific weight of the soil, d is the
pile diameter, 4~is the angle of internal friction measured
Returning to eqn (1) we note that the maximum value
from drained triaxial or direct shear test, z is vertical
of the spring reaction, Jq, is reached when y and
distance from the surface, and # is a parameter for which
take their maximum value. For example, during virgin
Broms 23 assumed the value # = 3. In the case where the
initial loading, F~,max is reached at the first loading-
deposit is saturated, the buoyant specific weight should
reversal-point (d~/dy = 0)
be used.
/ 1 \ll, Following Broms work, 23 other researchers elaborated
Fs,ma x = o z K y m a , + ( 1 - - o ~ ) K y o t - ~ - ' ~ ) (5) on the mechanisms that develop when the soil surround-
ing the pile fails. Reese et al., 24 using soil mechanics
The maximum force given by (5) must be equal to the theory and a wedge-type failure near the surface, pre-
ultimate lateral soil reaction per unit length of pile. For sented expressions more sophisticated than eqn (9) to
cohesive soils, theoretical studies based on the theory of approxima{e the ultimate soil resistance near the ground
plasticity 2° give that under plane-strain conditions the surface and at a depth. Nevertheless, in this study the
lateral soil reaction is expression proposed by Broms 23 is preferred because of
its simplicity, since it involves only one parameter
Fs,max(Z) =/kg(z)d (6)
compared to the expressions proposed by Reese
where S(z) is the shear strength of the soil at depth z, d is et al., 24 which involve three parameters. However, the
the pile diameter and ), is a dimensionless quantity formulation presented herein is not restricted to the use
9 - 1 4 2 < A < 11.940, where the lower limit 9.142= of a specific expression that approximates soil resistance,
6 + 7r is for a perfectly smooth pile and the upper limit and the best possible available expression should be used.
11.940 = 4x/2 + 27r is for a perfectly rough pile. The The plastic soil behavior (no hardening) at large pile-
theoretical values for Aderived by Randolph & Houlsby 2° deflections indicates that the ultimate post yielding
for plane-strain conditions do not differ much from the stiffness of the soil is zero; and therefore, the parameter
32 D. Badoni, N. Makris

a in eqn (5) is set equal to zero. Furthermore, from eqn Accordingly, for this zero-frequency-limit loading the
(4) which gives the upper bound of ¢ we see that if nonlinear distributed spring alone utilized by Trochanis
fl + 7 = 1 then [¢[ _< 1. This is a convenient formulation et al. 9 is sufficient to model the response.
extensively used in base-isolation studies. 19 Accordingly, Under dynamic loading (finite frequencies), part of the
with a -- 0,/3 + 7 = 1 and the combination of (5) with input energy is dissipated through hysteresis in the
(6) and (9), the force resulting from the nonlinear spring vicinity of the pile-soil interface, and the remaining
and given by (1) reduces to energy radiates outwards. In this case, the nonlinear
Fs(z ) = A(z)S(z)d¢, for cohesive soils (10) spring alone is not sufficient to model realistically the
response, since the radiation damping is of viscous type.
This insufficiency motivated the addition of the viscous
Fs(z ) = #%d 11Z+ sin~z(,
si--~¢
for cohesionless soils dashpot in parallel with the nonlinear spring as shown in
Fig. 1, to model radiation damping. Of course, one could
(11)
adjust the parameters of the Wen's model at some
The value # = 3 reported by Broms 21 is an assumption frequency to incorporate the energy loss due to radia-
based on experience. Herein, n and J in (10) or n and # in tion into the hysteretic loop generated by the nonlinear
(11) are parameters of the model to be calibrated from spring. Nevertheless, this will result in poor modeling,
experimental data. Furthermore the value of parameter since the radiation damping is frequency dependent; and
fl in eqn (2) has minor significance and is taken equal to therefore, at some other frequency the parameters of the
0"4 throughout this analysis. nonlinear spring should be re-adjusted. Moreover, the
Under lateral loading, the small-amplitude stiffness K, incorporation of all the energy loss into one hysteretic
of the soil reaction in eqns (1) and (5) can be approxi- loop implies that all of the energy is dissipated at the
mated realistically 14'17'18 by pile-soil interface with no waves propagating outwards.
K ~ l'2Es (12) Such a crude approach removes from the model the
capability to study the pile-soil-pile-interaction effect
where Es is the Young's modulus of the soil. and eventually the nonlinear group response. Accord-
The dimensionless hysteretic quantity ~, in (10) is ingly, for the case of dynamic loading the presence of a
provided by (2), in which Y0 (value of pile-deflection at frequency dependent dashpot is needed in the model.
which yielding initiates in the spring), is given by A good candidate for the expression that describes the
.,, s(z) . force resulting from the dashpot also has to be valid for
yo(z) = At z ) ~ a , for cohesive soils (13) small-amplitude motions. At this linear range, theoreti-
cal studies on the damping coefficient of single piles 14-16
%d 1 + sin q5 show that the resulting force stemming from radiation
yo(z) = # l'2Es(z) 1 - sin~bz' for cohesionless soils damping should be of the form
(14) F a = [Oao°25ps Vsd]j~ (15)
In the proposed model (eqns (10) or (11) and (2)) the where the term within the brackets is the distributed
parameters to be determined are n and J for cohesive frequency dependent damping coefficient c( ao) = Qao °'25
soils, or n and # for cohesionless soils. The remaining ps Vs d, ao = ~ d / V s is a dimensionless frequency, and the
parameters, %, Es, S, ~b, are standard geotechnical coefficient Q is given by the expression
parameters obtained from field or laboratory tests.
[ 3-4 ] 1"25/7r)0"75 (16)
Parameters n and J or # are calibrated by fitting
Q=2 1 + 7r(1 ~us) \4]
experimental data from static monotonic, or zero-
frequency cyclic tests. This is an interesting result, since where us is the Poisson's ratio of the soil.
in our formulation all of the model-parameters of the The expression for Q given by (16) is derived from a
nonlinear spring are frequency independent and can be simple 'cone' model 14'25 which resembles Novak's
calibrated readily from quasi-static monotonic tests. model 16 but does allow for some non-zero vertical
For the static and quasi-static (zero-frequency limit) deformation of the soil during lateral motion, as is
loads studied by Trochanis et al. 9 the Bouc-Wen model appropriate due to the presence of the stress-free surface
given by (1) and (2) was shown to be capable of and the non-uniformity with depth of pile deflections. At
predicting well the nonlinear response of a variety of shallow depths, (say less than three diameters) the
soil-pile systems. Under strong static loading, energy is expression given by (16) can be replaced simply by
lost in the vicinity of the pile-soil interface because of the Q = 3 (see Ref. 14).
hysteretic behavior of the soil, whereas radiation damp- Under harmonic vibrations with frequency, Lo,eqn (15)
ing is zero since the frequency is zero. The same energy- yields a force proportional to wy. Nevertheless, when the
loss mechanism prevails under quasi-static cyclic loading response becomes nonlinear (y > Y0), the force generated
since again radiation damping is practically zero and all from radiation damping does not increase considerably
the induced energy is dissipated through hysteresis. compared to the maximum damping force that develops
 Nonlinear response of single piles 33

under linear conditions. Accordingly, in order to account predicting the experimentally measured single pile
for large pile deflections under harmonic excitation, eqn response. Five well instrumented full-scale experiments
(15) is refined to are selected where the pile response was recorded beyond
the elastic limit.
Fd = Oao 0"25psVsdw (Y) (17)
where (A) Cohesive soils
(Y)=Y0 fory>y0 (18)
and Dynamic response of pile in saturated peat
Kramer et al. 27 conducted lateral monotonic and cyclic
(y)=y fory<y0 (19) tests on a cylindrical steel pile installed in the Mercer
With this formulation, the nonlinear spring eqn (10) or Slough peat near the eastern shore of Lake Washington
(11), and the dashpot eqn (17), of our constitutive model in Bellevue, Washington. Subsequently, Crouse et al. 28
are consistent with the viscoelastic analysis at the limit of completed the experimental investigation on the same
small-deflections and the plasticity analysis at the zero- pile by conducting lateral dynamic harmonic and quick-
frequency limit. The performance of this model in release tests. The subsurface material of the Mercer
predicting the measured dynamic nonlinear lateral Slough consisted of 14"9m thick peak deposit over
response of a variety of soil-pile systems is investigated medium dense sand. The soil properties and pile
in the sequel. characteristics are summarized in Table 1. For more
The proposed method, developed herein for lateral information the reader is referred to the original
pile motion, can be used without difficulty to model the references. 27,28
nonlinear response of piles under axial vibration. The Figure 2 (top) compares the predicted head-force-
bearing capacity equations (equivalent to eqns (6) and displacement curve based on eqn (10) against experi-
(9)) for axial pile motion are available in the geotechnical mental data from monotonic lateral loading. 27 The
literature and are also summarized by El Naggar & agreement is very good. The profile of the Young's
Novak. 8 modulus, E~, and that of the undrained shear strength,
Su, are also illustrated in Fig. 2, where the values shown
are those reported from experimental measurements.
FINITE ELEMENT FORMULATION FOR P I L E - The computed response (solid line) is obtained with
SOIL SYSTEM n = 1 and J = 0.5. The calibrated model was then used
to predict the dynamic response of the single pile. Figure
With the macroscopic nonlinear model developed in the 2 (center and bottom) compares the prediction of the
previous section, the dynamic pile-soil interaction prob- model for the dynamic stiffness and damping coefficients
lem reduces to the analysis of a beam supported on a of the pile against experimental data reported by Crouse
nonlinear Winkler foundation. Under dynamic head et al. 28 The damping coefficient of the pile was computed
loading, dynamic equilibrium of the pile gives via the expression

Eplp y(z, t) + m - ~ y ( z , t) - Fs(z , t) - Fa(z, t) = 0

WD
c(cv) -- 7r;vy2ax (21)
(20)
where WD is the area of the force-displacement loop at
where z is the spatial coordinate along the pile, and t is
the pile head and Ymax is the maximum pile-head dis-
time. Fs and Fd are the spring and dashpot reactions on
placement. This area corresponds to the sum of the
the pile given by (10) or (11) and by (17), respectively.
hysteretic and radiation damping. It should be noted
With the developed finite element formulation, the
however, that the amplitude, P0 = 1.5 kN, of the applied
solution of (20) which is coupled with (2), is possible
load during the harmonic test was very low compared to
by considering the variation of soil properties along the
the lateral load needed to generate nonlinear response
pile-length. Furthermore, the portion of the pile that
(P ~ 20 kN); and therefore, the response shown in Fig. 2
protrudes above the ground surface together with head
(center and bottom) is linear. Accordingly, in this case
and tip boundary conditions can be modeled in a simple
WD is the energy lost primarily due to radiation. The
and realistic manner. Details of the solution procedure of
storage stiffness of the pile, K1, was subsequently
the discretized equations can be found elsewhere. 26
obtained from the expression

PREDICTION OF FULl, SCALE EXPERIMENTAL (22)

RESULTS \ k,Ymax,/ ~,Try2ax.])

The efficiency of the proposed model is demonstrated by where P0 is the amplitude of the imposed force.
34 D. Badoni, N. M a k r &

Table 1. Constitutive model parameters for single pile

Test/Year Soil properties Pile characteristics Model
parameters
Cohesive soils J n
Kramer et al., 1990 us = 0'49 L= 14"9m 0"5 1.0
Crouse et al., 1993 V~(tip) = 30 m/s d= 0.20 m
p~ = 1.12mg/m 3 t= 0-0064m
Eplp = 3.81 × 103 kN-m2
Brown et al., 1987 us = 0-49 L= 13.4m 0"15 0"25
Blaney & O'Neill, 1986 V~(tip) = 115 m/s d= 0.273 m
p~ = 2-1 mg/m 3 t= 0-0093 m
Eplp = 7.3 x 10 3 kN-m 2
Cohesionless soils # n
Brown et al., 1988 us = 0"48 L = 13.1 m 5"0 3'0
(sand) VAtip) = 160 m/s d = 0.273 m
Ps 1.6 mg/m 3 t = 0.0093 m
Eplp = 7.3 × 103 kN-m 2
Jennings et al., 1984 us = 0'49 L = 6.75 m 5"0 3"0
Vs(tip) = 125 m/s d = 0.45 m
Ps l'6mg/m3 t = 0.010m
Eplp = 0-8 x 105 kN-m 2
Ting et al., 1987 us = 0"48 L = 9.75m 3'0 2"0
V~(tip) = 70m/s d = 0.61 m
p~ = 2.0 mg/m 3 t = 0.013 m
Eplp = 2.2 × 105 kN-m 2

The computed dynamic stiffness and damping of the Dynamic response o f pile in stiff clay
pile (at the ground surface) are plotted with dotted lines Full-scale static and dynamic experiments on a cylindri-
on Fig. 2. At low frequencies the storage stiffness reaches cal steel pile embedded in a stiff overconsolidated clay
a value of approx. 750 kN/m, which is in agreement with have been conducted by Blaney & O'Neill 29 at the
the static stiffness reported by K r a m e r et al. (630 kN/m). University of Houston. Details on the upper and lower
At moderate frequencies (2-3 Hz) the computed values values of undrained shear strength and Young's modulus
of the dynamic stiffness and damping are approximately of the deposit can be found in the original reference.
twice the experimental values reported by Crouse et al. 28 Herein, the average distribution of the undrained shear
These experimental values have been extracted from strength reported by Brown et al. 3° in a subsequent study
acceleration records recorded on a steel platform sup- on the same profile, is used. This average distribution of
ported by the pile at 0"712 m above the ground surface. the shear strength, is depicted in Fig. 3 (top) together
Crouse et al. reported that in order to extract the with the average value used for the Young's modulus
dynamic stiffness of the pile at the ground surface, they which is equal to Es = 600Su. This relatively small value
assumed that the above ground portion of the pile of Young's modulus (for soil dynamics studies) is
responded as a rigid body. The solid lines on Fig, 2 are actually in good agreement with experimental data,
the values of the stiffness and damping computed by since it corresponds to values of the Young's modulus
taking into account the flexibility of the protruded greater than those measured via the self-boring pressure-
portion of the pile above the ground surface. In eqn meter (SBP) test, but smaller than those obtained with
(22), P0 is the amplitude of the horizontal force applied the cross-hole test.
on the pile by the mass-shaker and Ymax is the amplitude The prediction of the proposed model for monotonic
of the total pile displacement at the platform level. loading is compared in Fig. 3 (top) against experi-
Evidently, the flexibility of the protruded portion of the mental data. The computed response (solid line) was
pile above the surface reduces the stiffness of the pile-soil obtained with n = 0.25 and J = 0.15. Subsequently, the
system resulting in better agreement with the values calibrated model was used to predict the dynamic
reported by Crouse et al. 28 response of the single pile. The only available experi-
The dashed line on Fig. 2 (center and bottom) plots the mental result on the dynamic response of the pile is a
equivalent linear dynamic stiffness and damping gener- frequency response curve (transfer function). For a
ated when a harmonic force with amplitude P0 = 20 k N load controlled test with amplitude P0, the transfer
is applied at the level of the mass shaker. For this function is defined as
amplitude of loading the pile response is nonlinear;
resulting in a reduced equivalent linear stiffness and an _ y _ 1 (23)
augmented equivalent-linear damping. H(w) Po K1 + iK2 - m ~
 Nonlinear response o f single piles 35

30 lOO

9o
t~ 25
<
r.rJ
,-r-
~: ro

i,- l S ~ sc
~" 4c
~ lO
3(

<
~1c

f
o o~,~ o11 oA5 o:~ 0.25 0.005 0.01 0.015 0.02 0.025 0. 3 .035
DISPLACEMEr~TATTHEPILEHEAD(m) D I S P L A C E M E N T AT T H E PILE H E A D (m)

0,014

I000] , , , 'Po= 1.5~N at g~und su~ace ' '...... 0.012 ~ ~2.95m\

8001........................................................................................ o o 4cg I--2.21m\
0.01 - - 1.43m

:°oL o o ~0.82m I
,., | ...... _8o o o_ _ _o_ % o _o..o_ _~_ o_ _ _ _% _ _ 0.008
200 i" - - r%~, ~ O-k.,fi -
/
0| I i i i i i i i i o.oo,
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2,6 2,8
FREQUENCYO"h)
0.004 0 0

25I.........
~ 2o " ~ - "~ . . . . . . . . . . . .. . . . . . . . . ... . . . . . . . . ... . . . . . . .
P o = 1.SkN at g r o u n d surface t 0.002 0

,5| .............. Po~20~_. . . . _j 01

0
i
0,5
i
1
i
1.5
J
2
i
2.5
l
3 3,5
I
4
'
415

~0~__ Po= LSk~ ..... o ~ o l FREQUENCY(Hz)

5~ oo . . . . 0 1
Fig. 3. Top: Comparison of predicted head-force-displace-
0 1
1
i
1.2
1
1.4
i
1.6
i
1.8
i ,
2
i
2.2
i
2.4
i
2.6
i
2.8
i
3
ment curve (solid line) with the experimental data (points) from
F R E Q U E N C Y fHz) Brown et al. 3° Bottom: Comparison of predicted transfer
function (solid line) with recorded data (points) from Blaney
Fig. 2. Top: Comparison of predicted head-force-displace- & O'Neill. 29
ment curve (solid line) with the experimental data (points) from
Kramer et al. 27 Bottom: Comparison of the predicted dynamic
stiffness and damping coefficients with the experimental data
(points) from Crouse et al. 28 low frequencies up to 1.5Hz the prediction of the
response is good. Nevertheless, the prediction shows a
resonance frequency at 1.7Hz, whereas the recorded
where y is the complex-valued displacement recorded resonance frequency is at 2"1 Hz. It should be mentioned
when the force reaches its maximum. K] is the storage that the pile protruded 0.82 m above the ground and the
stiffness of the pile (real part) given by (22),/(2 -- wc(w), location of the center of mass of the pile-cap and the
is the loss stiffness of the pile (imaginary part), and m is carriage that supported the vibrator was at 2.21 m above
the above-ground-mass of the cap. Using the time the ground surface. Furthermore, the horizontal driving
domain formulation pre,;ented herein under harmonic force was apparently measured at 2.95m above
loading with frequency w, the resulting amplitude of an the ground surface. Sensitivity studies based on our
equivalent linear transfer function is analysis showed that the response of the pile-soil
system is much more sensitive to the values of the
J
above ground stiffness and the location of the point of
IH(~)} : 1 I application of the driving force, rather than to the values
of the mechanical parameters of the distributed soil
~(y~)2 // Wo-~2 i ._Z_~- springs and dashpots.
- k--~max j - m w 2 + 7rymax

(24)
(B) Cohesionless soils
Again, in this experiment,, the amplitude of the excitation
force is only 2-67 kN, and the response of the pile is The majority of well instrumented, nonlinear, full scale
restricted in the linear range. Figure 3 (bottom) com- experiments on single piles embedded in sands are static
pares the predicted response from (24) against the or quasi-static, with the exception of the work conducted
measured transfer function by Blaney & O'Neill. At by Ting. 31 Nevertheless, before presenting the single pile
36 D. Badoni, N. Makris

response under true dynamic loading, the performance p-y C U R V E AT z = . 9 0 9 m p-y C U R V E AT z = l . 2 2 m

of our model is examined against the data from two well

instrumented quasi-static full scale experiments.
12C
'20I
I OC / o 1oo
/
Monotonic loading o f pile in medium dense sand 3 /
/

sol
Brown et al. 32 tested a hollow pile with 27.3 cm outer I
diameter and 0.93cm wall thickness. The pile was Z
~ 6c 6o}
initially driven in stiff clay which was excavated to a
depth of 2-9 m and sand was backfilled around the pile.
Because the sand extends to a depth of slightly more than
10 pile diameters, the response of the pile is expected to
depend only on the presence of the sand. The average dry
density after compaction was about Ps = 1"57g/cm3.
2C
4o}

2o~

ol
/
0.005 0.01 0.005 0.01
Results from direct shear tests indicated that the com- D E F L E C T I O N (m) D E F L E C T I O N (m)

pacted sand has an angle of internal friction, ~b = 38.5 °.

For more details the reader is referred to the original Fig. 5. Comparison of predicted p - y curves (solid line), with the
experimental data (points) of Brown et al. 32 At two pile depths.
reference. The value of the Young's modulus was not
reported, and a typical linearly increasing distribution
was assumed. data. The agreement of our predictions with recorded
Figure 4 (top) compares the prediction of the present data are very good. To further demonstrate the capabil-
model (eqn (11)) for the head force-displacement curve ities of our model, Fig. 5 illustrates the distributed soil
against experimental data from monotonic lateral load- resistance (force/length) as a function of pile deflection
ing. The computed response (solid line) is obtained with at two distinct depths: z - - - 0 . 9 0 9 m = 3.33d, and z =
n = 3 and # = 5. Figure 4 (bottom) compares the result- 1-22 m - - 4 . 4 7 d . The agreement of this prediction with
ing distributions of bending moments against recorded the measured values is also very encouraging.

12o
n= 3 .
g5
. . .
~o
. .
_
Cyclic loading o f pile in saturated silty sand
Jennings et al. 33 tested two 45"0 cm steel hollow piles with
1"0 cm wall thickness and 8.25 m length. Both piles were
embedded 6"50 m below the ground level. The piles were
installed in the flood plain of the Hunt River about 1 km

i
f,1
upstream from where the river discharges into Well-
~ 60
ington Harbor, New Zealand. Two consolidated
~4o undrained triaxial tests with pore pressure measure-
ments were performed at a strain rate of 0" 1% min and
indicated zero cohesion and an angle of internal friction
z=13.1m I 1 2 0 138.5
of q~ = 35 °. The reported dry density is % = 1.65 g/cm 3.
/ i / More details about field investigation and laboratory
0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
DISPLACEMENT AT THE PILE HEAD (m) measurements can be found in the paper by Jennings
et al. 33
o
250

n=3
p,=5
200

E iso f o

E s (MPa) ~)o

100 35
r~

-2
so

I
"100 ~o
- 0
,
50
.
100 150 35
BENDING MOMENT (kN-m) ," i , i i i
°o o.oos 0.01 o.01s 0.02 o.oas 0.o3 o.03s 0.~ o.04s
DISPLACEMENTATGROUNDLEVEL(m)
Fig. 4. Comparison of predicted (solid line) head-force dis-
placement curve (top), and static bending moment profile Fig. 6. Comparison of predicted head-force displacement curve
(bottom) against the experimental data (points) of Brown (solid line) with the experimental data (points) from Jennings
et al. 32 e t al. 33
 Nonlinear response of single piles 37

"' P=4OkN "' ~P--40kN f=2.1 Hz

o o
1 .1 I -1 Es (MPa)
z--0m_ I "~ o
2 .2}
0t
3 31
,= -4~
11
s -S[ .2¸ 2t
(; 3 31
g
-o.o= -o.b, O O.O1 0.02 -IOO -s'o o so oo 41:9.75m~30 ~ l
4
" P=120kN "~[P=120kN u.] n=2
>. 5 51 g=3
11 6
Z 71 ' 0=31°
© .7
¢3 al
> 41
.8
©
<~ sl 9
e~
-0.01 -0.;05 0.(105 0.01 -100 -50 0 50 100
-0.04 -0.02 o 0.02 o.o= I -=do o aoo DEFLECTION (m) BENDING MOMENT (kN-m)

-' "P=200kN P=200kN Fig. 8. Comparison of predicted dynamic deflection and bend-
ing moment profiles (solid line) with the experimental data of

./..-
il
Ting. 31 The applied force on the pile-head is 11 kN.

computed response (solid line) is obtained for n -- 2 and

#=3.
To further investigate the capabilities of our method,
-o.b~ ! o.bs -s~o o soo
the soil pressure (Fs + Fd)/d is computed at different
DEFLECTION (m) BENDING MOMENT (kN-m)
depths. Figure 9 compares the predicted soil pressure vs
Fig. 7. Comparison of predicted static deflection and bending pile deflection loops against the recorded data. Although
moment profiles (solid line) with the experimental data (points) our model cannot capture the apparent gapping effect,
of Jennings e t al. 33 the agreement is indeed very encouraging since both
stiffness and damping are well predicted even at the
Figure 6 compares the predicted head force-displace- local scale. It is interesting to note that although pile
ment curve against experimental data. The profile of the deflection decreases with depth, the resulting pressure at
Young's modulus used is also shown in the figure. The two diameters depth (right plot) is larger than the
computed response (solid line) is obtained with n = 3
and/z = 5. Figure 7 shows experimental data and pre-
dicted values for the displacement profile (left) and f=2.1 Hz
bending moments (right), due to a nominal lateral p-y LOOP AT z= 0.61 m p-y LOOP AT z= 1.22m
force of 40, 120 and 200kN. The recorded data are 40 40r
from two different piles (o: East pile; *: West pile). The ?
proposed model predicts very well the response for all 30F
three levels of loading.
2O 20[
i / i
/ .
Dynamic response of pile in saturated sand 10}
Ting 31 conducted a detailed study on the dynamic
response of a single pile driven into a dense, saturated, 0l
silty fine sand located at Seal Beach, California.
-113 -101
Standard penetration tests blowcounts at the site i ,
¢,0
averaged 13. Details on the experimental setup can be -213 -201 ¢ ®

found in the original paper. The soil Young's modulus is i I"

assumed to increase linearly with depth having a value of -3C -301
30 MPa at depth of 9.75 m (location of the pile tip). The
-4(3 -402
angle of internal friction for the sand is reported to be 0 5 0
DEFLECTION (mm) DEFLECTION (mm)
4 ~ = 3 1 °.
Figure 8 compares the computed dynamic deflections Fig. 9. Comparison of predicted (solid line) dynamic soil
and bending moments :along the pile against experi- reaction at frequency f = 2.1 Hz against the recorded data
mental data. The comparison is indeed excellent. The from Ting. 31
38 D. Badoni, N. Makris

resulting pressure at one diameter depth (left plot). This & Novak 35 based on the earlier work developed by
is because the Young's modulus increases linearly with Kaynia & Kausel. 36 In the case of strong loading,
depth and at large deflections the response becomes where the deformation level near the pile-soil interface
nonlinear and little incremental pressure develops. The is large and linear elasticity fails to realistically represent
proposed model captures these local details with the displacement and stress fields, the problem becomes
accuracy. even more complicated, and rigorous three-dimensional
Table 1 summarizes the soil properties, pile character- dynamic formulation using theories of viscoplasticity
istics and the values of the calibrated model-parameters becomes a formidable task. Work in the context of
from the five full scale experiments studied herein. The dynamic viscoplasticity has been conducted by Cheng
purpose of Table 1 is also to provide a realistic range for et al., 37 where it was shown that even for a two-
the values of n, J and #. Recall that parameters n dimensional plane-strain case, the computational effort
controls the transition from the linear to the nonlinear is indeed very large. In contrast, the Winkler foundation
range and parameters J and # are associated with the model is a versatile and economical approach, since
strength of a cohesive and cohesionless soil respectively the problem of pile-soil interaction becomes one-
(see eqns (10 and (11)). It is observed that for cohesive dimensional.
soils n _< 1 and 0.15 < J _< 0.5, whereas for cohesionless The developed formulation is used herein to compute
soils 2 < n < 3 and 3 < # < 5. It is interesting to note the nonlinear seismic response of the single pile. The free-
that the values of parameter J obtained by Matlock 22 field soil displacement is imposed as a support motion to
after fitting experimental data were: J = 0.25 for the the distributed nonlinear springs and frequency depend-
Lake Austin profile and J = 0.50 for the Sabine profile. ent dashpots; and the pile response is computed using the
These values reported by Matlock lie within the range of developed, one dimensional finite element formulation.
values obtained from this study. Of course, more Details on the implementation of the method are pre-
experiments are needed to establish with confidence the sented by Badoni. 26
range of values for the model parameters. Nevertheless, The effects of the nonlinear soil-pile kinematic inter-
the present study provides some useful values which are action are portrayed in the form of two kinematic
used for the prediction of the nonlinear dynamic response factors
response of a pile embedded in the Ohba-Ohashi soil
profile. l y(0)l and I~ ]J(0)ld (25)
Iy-- u(O) ' u(O)

NONLINEAR SEISMIC RESPONSE plotted as functions of the frequency factor, a 0 -- ~d/Vs,

where y(0) is the pile head displacement and u(0) is the
While an extremely flexible pile might simply follow the free-field surface displacement.
seismic motion of the ground, real piles with finite When the pile-soil response is nonlinear, the number
bending rigidity 'resist' the induced soil deformation, of physical and geometrical parameters that affect the
generating additional soil strains in their vicinity. As a system response is large, and it is more practical to study
result, the incident seismic waves are scattered and the individual cases. Herein, the nonlinear kinematic
seismic excitation to which the pile-foundation is response factors are computed for different levels of
effectively subjected differs from that of the free field nonlinearity, for both free-head and fixed-head piles.
motion. Analysis of the linear kinematic response of The level of nonlinearity is measured through the ratio
single pile and pile groups has been studied by many Ug/d where Ug is the amplitude of the input motion at
researchers (see Refs 17, 18, 33-35, and references the pile tip.
therein). However, linear analysis is limited to small Figures 10 and l l depict values of the kinematic
displacement gradients, and fails to describe the stress- response factors Iy and I¢ for fixed- and free-head piles
and displacement-fields in the vicinity of piles when embedded in the cohesive profiles reported by Kramer et
excited by strong seismic motions. al. 27 and Crouse et al., 28 for three values of the ratio
The problem studied herein is that of a single pile Ug/d = 0.01, 0.1 and 1"0. The linear seismic response of
subjected to strong lateral motion because of the passage the piles is also computed with the developed method
of vertical shear waves. The seismic loading considered (solid line) and is compared with the 'rigorous' solution
produces horizontal harmonic oscillations at a 'free-field' using the formation developed by Kaynia & Kause136
point of the deposit, i.e. at a point unaffected by the (stars).
presence of the piles. Around the piles the 'perturbed' Herein, for the computation of the seismic free-field
wave field is a complicated combination of incident motion the profile of cohesive soils was assumed homo-
waves reflected at the surface, and waves diffracted by geneous, although the shear modulus near the surface
the piles. At the limit of linear elastodynamics a rigorous assumes smaller values than its value at a few diameters
method of solution to this three-dimensional dynamic depth. For a homogeneous soil deposit excited at the
boundary value problem has been presented by Kaynia base with motion Ug = Ug exp(iwt), the amplitude of the
 Nonlinear response o f single piles 39

n=l ..•. BEM

n=l __ LINEAR
I=0.5 .r--o.5 _ _ _
1
1.6 • ... t ~ da-- l"
0.8 l • .i

,"-.~ 0.6!

0.4
0.6 J

0.2

o~ 0.1 0.T- 0.3 0.4 0.5 0,1 0.2 0.3 0.4 0.S
rod rod
a0 = V a0= -V-S
s

n--0.25 n=l ,, N BEM

•" LINEAR
JM).15 o.25 r--o.5 ___ u=.,~.o~.
tr~/d= O. l
...... [¥d= 1
0.8 0.2

~ . ~ 0.6 -.e-o.1~ : / - - 7 2 : ~
M m ~ BEN
0.4 --- g,__%o, 0.1 , ~ ~3~,' , ,/'~

0.2
~d: ,
o5 o.1 o.2 ola o14 o.s G( 0.1 0.2 0.3 0.4 0.5
rod rod
a0= ~-
S
ao= V-S
Fig. 10. Displacement kinematic response factors for fixed-head Fig. ll. Displacement(top)and rotation(bottom)kinematic
pile embedded in soft peat, Kramer et al., 27 (top), and medium response factors for free-head pile embedded in soft peak,
stiffclay, Blaney & O'Neill, 29 (bottom). Kramer et al. 27

free-field motion is approximated by together with Table 1, the nonlinear dynamic response
of a single pile embedded in the Ohba-Ohashi soil profile
near Tokyo, Japan is computed.
The Ohba bridge is located in Fujisawa City near
u(z) = Uecos(_:_0vL~ (26) Tokyo. It is supported by eleven piers and is 485m
\v, / long and 10.8m wide. The bridge piers are supported
on pile foundations. For instance, pier 6 is supported on
In Figs 10 and 11, the maximum head response of the an 8 × 8 = 64 group with steel piles (32 batter and 32
pile surrounded by yielding soil exceeds the response of vertical). The piles have diameter = 0.6 m, length = 22 m
the same pile surrounded by linear elastic soil. The
kinematic displacement :factor tends to fluctuate with SOIL TYPE V~(m~) Ps (Mg/m3) S~ (kN/m 2)
frequency and in some cases the pile head deflection
exceeds the free-field surface displacement. This behavior 50

is reminiscent of the linear seismic response of piles

embedded in soil deposits containing a thin, soft top 50

layer. 34 This similarity in behavior should not be surpris-

70
ing, since the yielding portion of the soil near the surface
behaves to some extent as an equivalent linear soft top
layer.

P R E D I C T I O N OF PILE R E S P O N S E IN THE
OHBA-OHASHI SOIL PROFILE
!m
Using the method developed in the previous section Fig. 12. Profile of the Ohba-Ohashi soil deposit.
40 D. Badoni, N. Makris

and wall thickness = 0.009 m (for the vertical piles) and Po=400 kN
J=0.6 J=0.1
0"012 m (for the batter piles). The river runs between pier 50C 5O0

f= 0.01 H; f= 0.01 H z
6 and pier 7 and the soil profile obtained from a bore hole 40C 400

30C

near pier 6 is shown in Fig. 12. The top layers through

5J
20C

which the piles penetrate consist of soft alluvial strata of I0C

humus and silty clay with shear wave velocity ranging

from 50 to 60 m/s. The total thickness of the alluvium is -20C -200

-30C -300
about 22m. The underlying substratum of alluvial -40C -400

deposits consists of stiff clay and sand with shear wave -50_~
-0.05 0.05 -50_8
~OC
velocity of about 400 m/s. 400 f=0.1 Hz f= 0.1 Hz
Figure 13 plots the head force-displacement curves 30~

predicted with the developed model using the soil

2oa ~ ~~ 20~
10c
loo
properties summarized in Fig. 12, parameter n = 1 and o - -
-1 oc -10
parameter J -- 0.1 and 0.6. Parameter n was taken equal
to one, since the soft deposit of the Ohba-Ohashi area
-zoo

-300
-°12
-400
resembles the Mercer Slough peat, 27 for which the -5o_~
-5°-8~- -0.05 o o.o~
calibration of the model resulted: n = 1 and J = 0.5. A 500
f= 1.0 H z 400 f= 1.0 H z
single experimental value reported from a full-scale test

/1
300 3OO

conducted by Shimizu Corporation is also depicted. This 200 200 i

experimental point was taken at y ( 0 ) = 0.02m and noo

provides little information on the nonlinear behavior of -1 oo

the pile. 200

-300
-200

-300

Figure 14 plots computed force-displacement loops at -400

the pile head for three selected frequencies ( f = 0.01, 0' 1 -5o8'~ ~.06 0 0.0~ 0.~ -5? 8 -o.o~ o o

D I S P L A C E M E N T (m) D I S P L A C E M E N T (m)
and 1.0 Hz), amplitude of applied force, P0 = 400kN,
with J--- 0.6 (left)and J = 0.1 (right). Figure 15 portrays Fig. 14. Dynamic head-force-displacement loops under har-
the resulting force-displacement loops due to a harmo- monic force with amplitude P0 = 400kN of a single pile
nic force with amplitude P0 = 800 kN. It is interesting to embedded into the Ohba-Ohashi soil deposit.
note that in both Figs 14 and 15 the shape of the loop
tends towards that of an ellipse as frequency increases.
This is because as frequency increases, radiation damp- deflections are getting smaller with increasing frequency,
ing increases and the associated linear response prevails the apparent (secant) stiffness of the distributed spring
over the nonlinear hysteretic response. Furthermore, as increases. This explains the slight increment in the slope
the radiation damping force increases, the amplitude of of the force-displacement loops as frequency increases
the pile deflection decreases (for the same amplitude of (see Fig. 15).
applied force) and nonlinear effects are further sup- Figure 16 plots the resulting equivalent storage
pressed. In addition, in this case where n = 1, there stiffness (eqn (22)) and damping coefficient (eqn (21))
is a continuous transition in the nonlinear response even as a function of frequency for three values of the
at small deflections (see Fig. 13). Accordingly, as pile amplitude of the applied force, P0 = 10,400 and
800kN. Referring to Fig. 14, the force amplitude
P0 --- 10kN was selected in order to obtain the limiting
I09C

900 j~ case of linear response. The results obtained with our

method (continuous line) are compared with the
'rigorous' solution obtained with the formulation
developed by Kaynia & Kause136 (points). The results
from the two methods are in good agreement. When the
force amplitudes P0 = 400kN and P0 = 800kN, the
resulting pile response is nonlinear. It is interesting to
40C note that as frequency increases pile deflection
decreases and the storage stiffness, K1 increases, while
the damping coefficient decreases, tending toward the
values obtained from linear analysis.
N
Figure 17 plots the nonlinear kinematic response
~ o.os 0.1 o.~s 9.2 0.2s o.3 o.35 o.4 9.,~ 9.5 factors of a fixed-head pile embedded in the Ohba
DISPLACEMENT AT THE PILE HEAD (m)
profile for three values of the ratio Ug/d = 0.01, 0"1
Fig. 13. Predicted monotonic head-force-displacement curves and 1"0, and two values of J = 0.1 and 0"6. The com-
for two values of soil strength, with n = 1. puted linear response of the pile (solid line) is compared
 Nonlinear response of single piles 41

Po=800 kN
J--0.6
J=0.6 J=0.1 1 m , .........
. . f = .0 . 0 1.H z !. . .
,,o<. ,,oo f = 0 0. 1 Hz
0.8
400 ~ 400
=O0 2OO

o 0
-200 -200
o., u d .o, :
~oo -4O0
4oo -600 ~d- I
-eoo -BOO 0.2 •
_I oo.~ -0.:~ 0 O.=S 05 -I 0 .C"~'.5 .0.25 0 0.25 0.5
1o o o =00
0.5
eoo f=0.1Hz ~ 1 :°° f = 0 " l H f ~ , O~ 0.1 0.2 0.3 0.4

eoo ~od
"
4oo

2oo
o
-2oo
/ °:
::o:
r
/ "
a0 ----- -~-~s

C) -400
n=l
~soo
J=0.1
I - ::-LL _ : _ . : : ......
. e ( ~ '

-"°°8:n -o.=n o.~,5 0.5 ° -°8.5 ,o.ks o o.=5 o.~

1ooc 1ooo
..o,: f= 1:OH z H .I of--1o .0 z
.oo [ "~1 ,.-Z" o.e .,,.,, "'- :2L-.,
0.4 ~ Udd--O.Ol _ ~ "

.'20 ii -=oo .~°' / 0.2

.... ~,i=b.f
....... I ~ d = l
"
.
N
.~oc -eoo.
-800 ~ . e o o
011 0:2 0:3 0:4 0.5
- I oo- 8 .0.=6 0 ==S 0.5 "IO__,S -0." 6 0.25 0.5

DISPLACEMENT(m) DISPLACEMENT
(rn) rod
a 0 = VZ
Fig. 15. Dynamic head-force-displacement loops under har-
monic force with amplitude P0 = 800kN of a single pile Fig. 17. Displacement kinematic response factors for fixed-head
embedded into the Ohba-Ohashi soil deposit. pile embedded in the Ohba-Ohashi soil profile.

with the rigorous solution by Kaynia & Kause136 (stars). nonlinear behavior amplifies the values of kinematic
Finally, Fig. 18 plots the nonlinear displacement and response factors at moderate frequencies.
rotation kinematic response factors for a free-head pile
for three values of the ratio Ug/d -- 0.01, 0.1 and 1.0 and
J = 0.6. The general trend in both Figs 17 and 18 is that
CONCLUSIONS

A macroscopic viscoelastoplastic model which consists

J=0.6 J---O. 1 of distributed hysteretic springs and frequency depen-
40 • • 40
P~=I0 kN
dent dashpots in conjunction with one-dimensional finite
x element formulation was found capable of predicting the
~ 20 20
dynamic nonlinear response of single piles under lateral
loads. The model is physically motivated and involves
standard geotechnical parameters. Only two parameters
5 10 15
have to be calibrated by fitting experimental data. The
0 5 10 15
predictions of the calibrated model are in very good
agreement with a variety of measured values from five
10
full-scale experiments. The developed method and the
p0=800 kN
calibrated model were used to predict the inertial and
seismic response of the single pile in the Ohba-Ohashi
~mmw soil profile, Tokyo, Japan. At low frequencies, the non-
L) linear pile response prevails. However, as frequency
5 10 15
increases the component of the soil reaction due to
5 10 15
FREQUENCY (Hz) F R E Q U E N C Y (Hz) radiation damping increases and the nonlinear charac-
teristics of the response are less pronounced. Nonlinear
Fig. 16. Dynamic stiffness and damping coefficients of a single behavior amplifies the kinematic response factors at
pile embedded into the Ohba-Ohashi soil deposit for three
different values of load amplitudes. moderate frequencies.
42 D. Badoni, N. Makris

n=l interaction model for dynamic lateral motion. J. geotech

J=0.6 engng, ASCE, 1992, 118, 89-106.
8. E1 Naggar, M. H. & Novak, M. Non-linear model for
1.5 dynamic axial pile response. J. geotech engng, ASCE, 1992,
118, 89-106.
9. Trochanis, A., Bielak, J. & Christiano, P. Simplified model
1:
for analysis of one or two piles. J. geotech engng, ASCE,
1994, 120, 308-29.
0.5 _ _ LINEAR 10. Bouc, R. Modele mathematique d' hysteresis. Acustica,
1971, 21, 16-25 (in French).
I" 11. Wen, Y.-K. Approximate method for nonlinear random
0'.I 0'.2 0'.3 014 -b.5
vibration. J. engng mech. div. ASCE, 1975, 101, EM4, 389-
401.
oJd
12. Wen, Y.-K. Method for random vibration of hyster-
ao = V-
S etic systems. J. engng mech. div. ASCE, 1976, 102, EM2,
249-63.
13. Trochanis, A. A three-dimensional nonlinear study of piles
0.5 n=l
leading to the development of a simplified model. Thesis
J=0.6
0.4 ~ BEM
presented to Carnegie Mellon University, Pittsburgh, PA,
_ _ LINEAR ....... >
-" r .
:'. in partial fulfillment of the requirements for the degree of
_ _ _ U g d - - O . O i

u~dd= o.l . -

• Doctor of Philosophy, 1988.

o.a trod= ~ . .,' . . . . - - - --. ~.~..~-: 14. Dynamic head-force-displacement loops under harmonic
force with amplitude P0 = 400kN of a single pile
0.2 '~ embedded into the Ohba-Ohashi soil deposit.
15. Novak, M. Dynamic stiffness and damping of piles.
O.1 " ~ Canadian geotech. J., 1974, 11, 574-98.
16. Novak, M., Nogami, T. & Aboul-Ella, F. Dynamic soil
reaction for plain strain case. J. engng mech. div ASCE,
% 0.1 0'.2 o'.a 014 0.5
1978, 104, 1024-41.
rod 17. Makris, N. & Gazetas, G. Dynamic pile-soil-pile inter-
a°= ~s action. Part II: Lateral and seismic response. Earthq.
Engng. Struct. Dynam., 1992, 21, 145-62.
Fig. 18. Displacement (top) and rotation (bottom) kinematic 18. Makris, N. Soil-pile interaction during the passage of
response factors for free-head pile embedded in the Ohba- Rayleigh waves: An analytical solution. Earthq. Engng.
Ohashi soil profile. Struct. Dynam., 1994, 23, 153-67.
19. Constantinou, M. C. & Adnane, M. A. Evaluation of two
models for yielding systems. Report to NSF, Dept. Civil
ACKNOWLEDGEMENTS Engrn, Drexel University, PA, 1987.
20. Randolph, M. F. & Houlsby, G. T. The limiting pressure
Partial financial support for this project has been on a circular pile loaded laterally in cohesive soil. Geotech-
provided by Shimizu Corporation, Japan (Grant No. nique, 1984, 34, 613-23.
21. Broms, B. B. Lateral resistance of piles in cohesive soils. J.
NCEER/RF150-7014A to N C E E R ) , and by the Depart- soil mech Fdns Div. ASCE, 1964, 90, SM2, 27-63.
ment of Civil Engineering and Geological Sciences, 22. Matlock, H. Correlations for design of laterally loaded
University of Notre Dame. piles in soft clay. Proc. Offshore Technology Conf., Paper
No. OTC1204, pp. 1-577-594, 1970.
23. Broms, B. B. Lateral resistance of piles in cohesionless soils.
J. Soil mech Fdns Div. ASCE, 1964, 90, SM3, 123-56.
REFERENCES 24. Reese, L. C., Cox, W. R. & Koop, F. D. Analysis of
laterally loaded piles in sand. Proc. Offshore Technology
1. Moehle, J. P. Preliminary report on the seismological and Conf., Paper No. OTC 2080 pp. 473-84, 1974.
engineering aspects of the January 17, 1994 Northridge 25. Wolf, J. P. Cone models for pile foundation. Piles Under
earthquako~ Report No UCB/EERI-94/01 1994. Dynamic Loads. ed. S. Praakash, GT Div. ASCE, New
2. Makris, N., Badoni, D., Delis, E. & Gazetas, G. Prediction York, 1992.
of observed bridge-response with soil-pile-structure inter- 26. Badoni, D. Nonlinear analysis of piles under inertial and
action. J. Struct. Eng. ASCE, 1994, 2992-3011. seismic loading. Master thesis, University of Notre Dame,
3. Mizuno, H. Pile damage during earthquakes in Japan. In IN, 1995.
Dynamic Response of Pile Foundations, ed. T. Nogami, 27. Kramer, S. L., Satari, R. & Kilian, A. P. Evaluation of in
ASCE pp. 53-78, 1987. situ strength of a peat deposit from laterally loaded pile test
4. CNEL-ENEL, Contribution of the study of the Fruili results. Transportation Research Record, No 1278,
earthquake of May 1976, Rome, Italy, (1976). Transp. res. board, Washington, D.C., pp. 103-109, 1992.
5. Ross, A. G., Seed, B. H. & Migliaccio, R. Bridge founda- 28. Crouse, C. B., Kramer, S. L., Michell, R. & Hushmand, B.
tions in the Alaska earthquake. J. soil mech. found engng. Dynamic test of pipe in saturated peat. J. geotech, engng,
ASCE, 1969, 95. ASCE, 1993, 119, 1550-67.
6. Angelides, D. & Roesset, J. M. Nonlinear lateral dynamic 29. Blaney, G. W. & O'Neill, M. W. Measured lateral response
stiffness of piles. J. geotech, engng., ASCE, 1981, 107, of mass on single pile in clay. J. geotech, engng, ASCE,
1443-60. 1986, 112, 443-57.
7. Nogami, T., Otani, J. &Chen, H. L. Nonlinear soil-pile 30. Brown, D. A. Reese, L. C. & O'Neill, M. W. Cyclic lateral
 Nonlinear response of single piles 43

loading of a large scale pile group. J. geotech, engng, ASCE, Piles under Dynamic Loads, ed. S. Prakash. Geotechnical
1987, 113, 1326-43. Special Publication No 34, pp. 56-93, 1992.
31. Ting, J. M. Full-scale cyclic dynamic lateral pile responses. 35. Kaynia, A. M. & Novak, M. Response of pile foundations
J. geotech, engng, ASCE, 1987, 113, 30-45. to Rayleigh waves and to obliquely incident body waves.
32. Brown, D. A., Morrison, C. & Reese, L. C. Lateral load Earthq. Engng. Struct. Dynam., 1992, 21, 303-18.
behavior of pile group in sand. J. geotech, engng, ASCE, 36. Kaynia, A. M. & Kausel, E. Dynamic stiffness and seismic
1988, 144, 30-45. response of pile groups. Research Report, Dept. of Civil
33. Jennings, D. N., Thurston, S. J. & Edmonds, F. D. Static Engrg., MIT, Cambridge, MA, 1982.
and dynamic lateral loading of two piles. Proc. 8th WCEE, 37. Cheng, F.-P., Roesset, J. M. & Tassoulas, J. L. Dynamic
San Francisco, CA, (3), pp. 561-68, 1984. response of circular foundations in an elastoplastic
34. Gazetas, G., Fan, K., Tazoh, T., Shimizu, K., Kavvadas, medium. Geotech, Engng. Report, GR 86-3, Dept. Civil
M. & Makris, N. Seismic pile-group-structure interaction. Engrg. Univ. Texas Austin, Austin, TX, 1986.