ARTICLE IN PRESS

Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

www.elsevier.com/locate/soildyn

Dynamic analysis of piled foundations in stratified soils

by a BEM–FEM model
L.A. Padrón, J.J. Aznárez, O. Maeso
Instituto Universitario de Sistemas Inteligentes y Aplicaciones Numéricas en Ingenierı´a (IUSIANI), Universidad de Las Palmas de Gran Canaria,
Edificio Central del Parque Cientı´fico y Tecnológico del Campus Universitario de Tafira, 35017 Las Palmas de Gran Canaria, Spain
Received 23 February 2007; received in revised form 4 July 2007; accepted 18 July 2007

Abstract

In this paper, a 3D BEM–FEM coupling model is used to study the dynamic behavior of piled foundations in elastic layered soils in
presence of a rigid bedrock. Piles are modelled by FEM as beams according to the Bernoulli hypothesis, and every layer of the soil is
modelled by BEM as a continuum, semi-infinite, isotropic, homogeneous, linear, viscoelastic medium. First, the main points of the model
are set out. Then, several results of vertical, horizontal and rocking impedances for single piles and 2  2 pile groups embedded in a
stratum resting on a rigid bedrock, are presented. The influence on the dynamic response of stratum depth, soil stiffness and piled
foundation configuration is discussed. Finally, the influence of the stratigraphy on the seismic response of a 3  3 pile group is analyzed,
together with the pile-to-pile kinematic interaction and the wave-scattering phenomena.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Pile groups; Pile–soil interaction; Dynamic impedances; Seismic response; BEM–FEM coupling

1. Introduction instance, parametric studies of the seismic response of

single piles and pile groups to Rayleigh waves and to
The steady-state dynamic response of piled foundations vertically and obliquely incident body waves have been
in elastic soils has been the subject of much research, in reported in [14–17].
both the kinematic and the forced vibration analyses (see On the other hand, due to the fact that piles are
e.g. [1]). For the study of piles and pile groups embedded in commonly used when avoiding shallow soil of low bearing
a homogeneous half-space, several frequency domain capacity and transferring load to deeper soil or rock of
boundary integral formulations in conjunction with the high bearing capacity is needed, a particular case of interest
monodimensional finite element method (FEM) have been in the dynamic analysis of piles is that of piled foundations
used by different authors [2–5]. Linear analyses for non- embedded in a viscoelastic stratum resting on a rigid
homogeneous media or layered soils have also been carried bedrock, for both floating and hinged piles. However, not
out using the same kind of approach [6–9]. Subsequently, many papers have been reported focusing on this topic
more versatile and rigorous linear numerical models have [18,19], though some have dealt with it while studying the
been developed using the boundary element method pile–soil-structure interaction problem [20–23].
(BEM) for both soil and piles [10–13], but with the For this reason, the aim of this paper is to present a
disadvantage of a high computational cost. method for the dynamic analysis of piles and pile groups
Although a great deal of this research has been focused and, making use of it, contribute to the topic discussed
on the forced vibration problem, much of it has also dealt above by investigating, through parametric studies: (a) the
with the kinematic response of piled foundations. For influence of the presence of a rigid bedrock on the dynamic
impedances of piled foundations, and (b) the influence of
Corresponding author. Tel.: +34 928451920; fax: +34 928451921. the stratigraphy on the seismic response of hinged pile
E-mail address: [email protected] (O. Maeso). groups. To this end, a BEM–FEM coupling model previously

0267-7261/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2007.07.005
 ARTICLE IN PRESS
334 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

presented by the authors [24] is used to compute time- by the equation

harmonic dynamic impedances and the seismic response of
piled foundations embedded in viscoelastic zoned-homo- K̄ up ¼ Fext þ Qqp , (1)
geneous layered soils. The model has been validated for
where K̄ ¼ K  o2 M, being K and M the stiffness and mass
both the kinematic and the stiffness problems with several
matrices of the pile, o is the circular frequency of
results taken from the literature (e.g. [2,18–20]), though
excitation, and up the vector of nodal translation and
these comparisons are not presented here for the sake of
rotation amplitudes along the pile. Fext includes the forces
brevity.
at the top Ftop and the axial force at the tip of the pile Fp ,
In this approach, and in the line of a previous static
qp is the vector of tractions along the pile–soil interface,
model developed in [25–27], the pile–soil interaction takes
whereas Q is the matrix that transforms these nodal
place, from the integral equation point of view, through
traction components to equivalent nodal forces. By means
internal forces, as it is assumed that the soil continuity is
of Eq. (1), piles are modelled as vertical Bernoulli beams by
not altered by the presence of the piles. These are modeled
three-node FEM elements on which there are defined 13
by FEM as beams according to the Bernoulli hypothesis,
degrees of freedom: three lateral displacements on each
and every stratum of the soil is modeled by BEM as a
node, and two rotations on each one of the extreme nodes.
continuum, semi-infinite, isotropic, homogeneous, linear,
On the other hand, each stratum of the soil is modelled
viscoelastic medium. The model not only allows the
by BEM as a linear, homogeneous, isotropic, viscoelastic,
dynamic analysis of piled foundations embedded in a
unbounded region. The boundary integral equation for a
half-space or in a stratum resting on a rigid bedrock, but
time-harmonic elastodynamic state defined in a domain Om
also in multilayered soils of generic stratigraphy and
with boundary Gm can be written in a condensed and
topography, including deposits and inclusions.
general form as
Firstly, the main points of the BEM–FEM coupling
model formulation are set out. Secondly, several results of Z Z Z
vertical, horizontal and rocking time-harmonic dynamic c k uk þ p u dG ¼ u p dG þ u X dO, (2)
Gm Gm Om
impedances of single piles and 2  2 pile groups embedded
in a stratum resting on a rigid bedrock, are presented. where ck is the local free term matrix at collocation point
Different depths of the stratum and three foundation ‘k’, X are the body forces in the domain Om , u and p are the
configurations are studied, and the effects associated to displacement and traction vectors, and u and p are
these parameters are discussed. Finally, the influence of the the elastodynamic fundamental solution tensors for a time-
stratigraphy on the seismic response of a 3  3 pile group is harmonic concentrated load at point ‘k’. A hysteretic
analyzed. To this end, displacement transfer functions for damping model is used for the soil through a complex
vertically incident plane time-harmonic shear waves and valued shear modulus m of the type m ¼ Re½mð1 þ 2ixÞ,
response spectra for a particular configuration, under two being x the damping coefficient. Details about BEM
different strong ground motions, are presented for several formulation can be found in [28].
soil profiles. Pile-to-pile kinematic interaction and wave- Generally, body forces X are considered to be zero in
scattering phenomena, are also investigated. most of elastodynamic problems. Nevertheless, in this
Regarding the impedances study, it is shown that the approach, from the integral equation point of view, the
effect of the presence of the rigid bedrock is vital for the pile–soil interaction takes place through internal punctual
horizontal impedance of single piles and pile groups, and forces placed at the geometric piles tip and through load-
also for the rocking behavior of single piles, while it is lines placed along the piles axis, as it is assumed that the
almost negligible for the rocking response of pile groups. soil continuity is not altered by the presence of the piles.
Regarding the vertical impedances of floating piled The load-lines within the soil, the tractions along the
foundations, the effect of the bedrock is not important pile–soil interface acting over the pile and within the soil
except for frequencies below 1.5 times the fundamental (qpj ¼ qsj ), and the internal punctual forces F pj at the tip
natural frequency of the stratum. of the piles, are represented in Fig. 1, where a sketch of the
As for the performed seismic analysis, it can be seen model is shown.
that the presence of a soft layer at the top of a stratum
yields to rapidly decreasing transfer functions and, conse-
quently, to a weakened seismic response at the pile cap.
However, taking into account additional layers with
shear wave velocity increasing with depth is of minor
importance.

2. Pile–soil interaction model

The behavior of a pile submitted to harmonically varying

loads, considering zero internal damping, can be described Fig. 1. Load-lines representation of two piles embedded in a layered soil.
 ARTICLE IN PRESS
L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346 335

Then, Eq. (2) can be written as the scattered wave fields uS and pS , produced by the
Z presence of the piles. Consequently, the total fields are the
k k
c u þ p u dG sum of these two (u ¼ uI þ uS , p ¼ pI þ pS ), while it is only
Gm
2 3 in the scattered wave field that the tractions along the
Z nm
X Z
ll pile–soil interface and the forces at the pile tip exist
¼ u p dG þ 4 u qsj dGpj  dj Yjk F pj 5, ð3Þ (q ¼ qS , Fp ¼ ðFp ÞS ). In this case, the above equations are
Gm j¼1 Gm
p j written for the scattered fields in all regions, and then
where Gm pj is the pile–soil interface along the load-line j
expressed in terms of the total and the incident fields as
within the domain Om ; nm ll is the total number of load-lines
m
nll m
nll
X X
in the domain Om ; dj is equal to one if the load-line j ss s
H u G p  ss s
Gspj qsj þ dj Ysj F pj ¼ Hss usI  Gss psI
contains the tip of a floating pile and zero otherwise; and j¼1 j¼1
Yjk is a three-component vector that represents the (7)
contribution of the axial force F pj at the tip of the jth
load-line. and
Once all boundaries have been discretized, Eq. (3) can be m
nll
X
m
nll
X p
written for each region for all nodes in Gm and Gm upi þ Hpi s us  Gpi s ps  Gpi pj qsj þ dj Ypi j F pj ¼ uI i þ Hpi s usI  Gpi s psI ,
pj to
j¼1 j¼1
obtain two matrix equations of the type
m m
(8)
nll nll
X X where the right hand terms are known.
Hss us  Gss ps  Gspj qsj þ dj Ysj F pj ¼ 0, (4)
j¼1 j¼1
Only the main points of the formulation have been
presented here. However, more details can be found in
and Padrón et al. [24], where the formulation, focused on pile
m
nll m
nll groups embedded in a half-space, has been presented.
X X
pi pi s s
u þH u G p  pi s s
G pi p j s j
q þ dj Ypi j F pj ¼ 0,
j¼1 j¼1 3. Dynamic stiffness of piled foundations in homogeneous
(5) strata

respectively, where H and G are coefficient matrices The dynamic stiffness matrix K ij of a pile relates the
obtained by integration over the elements of the funda- vector of forces (and moments) applied at the pile top to
mental solution times the corresponding shape functions, the resulting vector of displacements (and rotations) at
us and ps are the vectors of nodal displacements and the same point. For a group of piles, it is assumed that the
tractions of boundary elements, and upi is the vector of pile heads are constrained by a rigid pile-cap, and the
nodal displacements of the load-line i. foundation stiffness is the addition of the contributions of
A coupled system of equations, of the type each pile. Fig. 2 illustrates the approached problem for a
Afus ; ps ; qs ; Fp ; up gT ¼ B, (6) usual configuration, where L and d are used to denote the
length and diameter of the piles, s refers to the distance
representing the layered soil–piles system, can be obtained between adjacent piles and H denotes the depth of the
from Eqs. (1), (4) and (5), where equilibrium and stratum.
compatibility fully bonded contact conditions over the The dynamic stiffness terms for a time-harmonic
different interfaces of the problem have been imposed. A excitation are functions of frequency o, and they are
and B are the square matrix of coefficients and the known
vector, respectively, both computed by rearranging the
equations and prescribing the known boundary conditions.
In this work, two different kinds of analyses are carried
out, each one of them with a different set of boundary
conditions, though, in both cases, piles in a group are
considered to be fixedly connected to a rigid massless cap
(not in contact with the soil) and zero-traction free surface
is assumed.
Firstly, dynamic stiffness of piles and pile groups are
computed by prescribing forced vibration at the pile cap.
Secondly, the seismic response of a piled foundation
subject to vertically incident plane time-harmonic shear
waves is analyzed. To this end, the displacement and
traction fields in the layered soil are considered as the
superposition of two: the uniform viscoelastic layered soil Fig. 2. 2  2 pile group embedded in a stratum resting on a rigid bedrock.
fields uI and pI , whose analytic expressions are known; and Problem geometry definition.
 ARTICLE IN PRESS
336 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

usually written as a rigid bedrock are shown in Fig. 4 for ratios between
stratum depth and pile length of H=L ¼ 1 (hinged pile),
K ij ¼ kij þ iao cij , (9)
1.5, 2, 3, 5 and 10, together with the response for the half-
space. It is assumed that the stratum properties are:
where kij and cij are the frequency dependent dynamic
internal damping coefficient bs ¼ 0:05 and Poisson ratio
stiffness and damping coefficients, respectively, ao is the
ns ¼ 0:4. The ratio between densities is rs =rp ¼ 0:7, the
dimensionless frequency
aspect ratio of the pile is L=d ¼ 15, and the pile/soil
od modulus ratio is E p =E s ¼ 103 .
ao ¼ (10) As can be seen from the figures, the presence of a rigid
cs
bedrock below a floating pile has a strong influence on the
and cs is the soil shear-wave velocity. impedances in the frequency band from the static value to
Fig. 3 shows a sketch of the discretizations used to approximately 1.5 times the dimensionless fundamental
obtain the stiffness of different pile groups embedded in natural frequency of the stratum in compression–extension
a viscoelastic homogeneous stratum resting on a rigid mode. Above this frequency, stiffness and damping
bedrock. As the developed software incorporates symmetry coefficient are coincident with those of a floating pile in a
properties, only a quarter of the total geometry of the half-space, which reveals that the main damping mechan-
problem has to be discretized. ism at intermediate and high frequencies is the energy
In what follows, vertical, horizontal and rocking dissipation through surface waves for both the half-space
impedances of different piled foundation configurations and the stratum. Below the first natural frequency, the
are presented. In the case of single piles, the stiffness and damping coefficients are far lower than the ones corre-
damping functions are normalized by the respective static sponding to the floating pile in a half-space, because there
stiffness. As for pile groups, the vertical and horizontal cannot be surface waves in a stratum at low frequencies
impedance functions are divided by the respective single and, thus, the energy is confined in it. As expected, the
pile static stiffness (kszzo and ksxxo ) times the number (N) of static value of the vertical stiffness of a hinged pile is much
piles in the group. Finally, the rocking impedances are higher than the corresponding to a floating pile. More
normalized with respect to the sum of the products of precisely, it is 5.1 times higher than the corresponding to a
the respective single pile vertical static stiffness (kszzo ) times floating pile embedded in a stratum of depth H ¼ 1:5L. On
the square of the distance to the rotation axis (xi ). All the other hand, the fundamental natural frequency of the
results are plotted versus the dimensionless frequency system related to the compression–extension mode is
defined by Eq. (10), and several ratios H=L have been clearly highlighted in each case. Note that the influence
considered. of the presence of the rigid bedrock is still noticeable for a
In the first place, vertical impedances of a single pile stratum depth five times the pile length, while it is hardly
embedded in a homogeneous stratum of depth H resting on significant when the stratum depth is ten times the pile
length.
Horizontal and rocking impedances of a single pile
Free BEM soil embedded in a homogeneous stratum of depth H resting on
surface discretization a rigid bedrock are shown in Figs. 5 and 6, respectively.
The characteristics of piles and soils for these and further
Rigid cases are the ones defined above and, for the sake of clarity,
FEM pile
bedrock only results for ratios between stratum depth and pile
discretization
length of H=L ¼ 1 (hinged pile), 1.5 and 2, together with
Fig. 3. 2  2 pile group and stratum BEM–FEM discretization (only a the response for the half-space, are presented. The range of
quarter of the geometry). interest in each stiffness figure has been enlarged, and the

6.50 7
6.25 H=L
6 H=1.5L
6.00 5 H=2L
H=3L
kzz/(kzzo)

1.25
czz/(kzzo)

4 H=5L
1.00 H=10L
3 half-space
0.75
0.50 2

0.25 1
0.00 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao

Fig. 4. Vertical impedances of a single pile embedded in a homogeneous stratum.

 ARTICLE IN PRESS
L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346 337

ao ao
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
1.2 3

2.5 H=L
1
H=1.5L
2 H=2L
0.8 1.1
half-space

cxx/(kxxo)
kxx/(kxxo)
0.6 1 1.5

0.4 1
0.9
0.2 0.5
0
0 0.1 0.2 0.3 0.4
0 0

Fig. 5. Horizontal impedances of a single pile embedded in a homogeneous stratum.

1.2 1.2
H=L
1 1 H=1.5L
H=2L
0.8 0.8 half-space

cφφ/(kφφο)
kφφ/(kφφο)

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao

Fig. 6. Rocking impedances of a single pile embedded in a homogeneous stratum.

natural frequencies to which every peak is associated have dances fluctuate about the half-space solution, and the
been labeled for the horizontal cases. amplitude of this fluctuation increases with the ratio s=d.
In this case, the resonance effects are associated to both Table 1 presents the comparison between the values of
the shear and compression–extension modes. The influence dimensionless natural frequencies obtained with the pre-
of the presence of the rigid bedrock is significant sented coupling model and those analytically computed for
approximately until the second natural frequency related an undamped stratum. Only the values obtained for the
to the shear mode. In this range, the impedances fluctuate horizontal impedance of a pile group with a ratio s=d ¼ 10
about the half-space solution for both the stiffness and the are presented. As can be seen, the proposed method
damping coefficients. As for the latter, they are small at low predicts adequately the actual fundamental frequencies.
frequencies as discussed above. For higher frequencies, the Finally, rocking impedances of a 2  2 pile group
impedance behavior is similar to the one corresponding to embedded in a homogeneous stratum on a rigid bedrock
a pile embedded in a half-space. are shown in Fig. 9. Only the ratio s=d ¼ 10, for which the
Vertical and horizontal impedances of a 2  2 pile group influence of the rigid bedrock is stronger, has been
embedded in a homogeneous stratum of depth H resting on displayed in order to illustrate the fact that the rocking
a rigid bedrock are shown in Figs. 7 and 8, respectively. impedances of pile groups are little influenced by the
Ratios between pile separation and diameter of s=d ¼ 2, 5 presence of the bedrock.
and 10 are presented. For scale reasons, the peaks
associated to the stratum natural frequencies do not 4. Seismic response of pile groups to vertically incident shear
appear very clearly in the figures, but their magnitudes waves
are proportional to those of the single pile case, and even
increase with the ratio s=d. Besides, the vertical impedances The influence of the stratigraphy on the seismic behavior
at intermediate and high frequencies are equivalent to of a piled foundation is studied in this section. To this end,
those of a floating pile group embedded in a half-space. the seismic response of a hinged pile group under vertically
However, the group effect is predominant over the incident plane S-waves is analyzed. The cases included in
influence of the presence of the rigid bedrock. On the this experiment are sketched in Fig. 10. Three different
other hand, for the horizontal case, the influence of layered soils resting on a rigid bedrock, formed by up to
the rigid bedrock is relatively much more important, even four different layers with shear wave velocity increasing
though the group effect is still predominant. The impe- with depth, are considered. Besides, two limiting cases
 ARTICLE IN PRESS
338 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

8 10
H=L
6 8 s =2 H=1.5L
d H=2L

kGzz /(N kszzo)

cGzz /(N kszzo )

half-space
4 6

2 4

0 2

-2 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao
8 10
s =5 H=L
6 8 d H=1.5L
H=2L
kGzz /(N kszzo)

cGzz /(N kszzo )

4 half-space
6

2 4

0 2

-2 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao

8 10
H=L
6 8 H=1.5L
H=2L
kGzz /(N kszzo)

cGzz /(N kszzo )

half-space
4 6
s =10
2 4 d

0 2

-2 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao

Fig. 7. Vertical impedances of a 2  2 pile group embedded in a homogeneous stratum. s=d ¼ 2, 5 and 10 (from top to bottom).

corresponding to homogeneous profiles with the properties displacement (uff ), for vertically incident plane time-
of the softest and stiffest layers, respectively, as well as the harmonic S-waves. Real and imaginary parts of these
half-space case, are also taken into account. The properties transfer functions are presented in Fig. 11 for the six
of the stiffest layer are taken as those of a reference soil. As profiles defined above, ao being referred to the stiffest soil
it is well known that the number of piles is not a significant for all cases. Comparing reference soil one-stratum and
parameter in the horizontal seismic response of piled two-strata cases, it can be seen that taking into account a
foundations (see e.g. [16]), only the case of a square 3  3 soft layer atop yields to much more rapidly decreasing
hinged pile group is presented. transfer functions (as highlighted before, for instance, in
The non-dimensional properties of the reference soil [16,20]), while the addition of intermediate shear wave
(with shear wave velocity cs ) and pile group are: internal velocity layers in between does not alter significantly the
damping coefficient of the soil bs ¼ 0:05, Poisson ratio response up to ao ¼ 0:4 with respect to the two-strata
of the soil ns ¼ 0:4, ratio between densities rs =rp ¼ 0:7, profile. Also, the difference between the transfer functions
pile/soil modulus ratio E p =E s ¼ 102 , aspect ratio of the for the half-space and the stratum are rather small, only
piles L=d ¼ 15, ratio between stratum depth and pile noticeable at the stratum natural frequencies.
length H=L ¼ 1 (hinged piles), and ratio between pile A phenomenon that can be studied here, in reference to
separation and diameter s=d ¼ 10. Damping coefficient, the difference between homogeneous and stratified strata
Poisson ratio and density are kept constant for all layers. transfer functions, is the horizontal deformation of the free
The first step is obtaining the time-harmonic transfer surface. Assuming S-waves acting along direction y, Fig. 12
functions relating horizontal displacement amplitudes at shows, for the one-layer and two-layer profiles, and several
the pile cap (ucap ) to the free field surface horizontal frequencies, the real part of the horizontal displacements of
 ARTICLE IN PRESS
L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346 339

1.75 5
H=L s =2
1.5 H=1.5L d
4
1.25 H=2L

kGxx /(N ksxxo)

cGxx /(N ksxxo)

half-space
1 3

0.75 2
0.5
1
0.25
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao
1.75 5
1.5 H=L s =5
H=1.5L 4 d
1.25 H=2L
kGxx /(N ksxxo)

cGxx /(N ksxxo)

half-space
1 3

0.75 2
0.5
1
0.25
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao
1.75 5
1.5 s =10
4 d
1.25
kGxx /(N ksxxo)

cGxx /(N ksxxo)

1 3

0.75 2
H=L
0.5 H=1.5L
H=2L 1
0.25 half-space
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao

Fig. 8. Horizontal impedances of a 2  2 pile group embedded in a homogeneous stratum. s=d ¼ 2, 5 and 10 (from top to bottom).

Table 1 the presence of the piles on the incident field is much more
Comparison between natural frequencies for an undamped stratum and stronger, both in amplitude and extension, when a soft soil
for a 2  2 pile group-stratum system stratum exists atop, which is related to the more rapidly
Stratum natural frequencies in varying transfer function of this case. Obviously, the
magnitude and shape of this perturbation is function of the
Shear mode Compression– H=L wave length in the soil. In all cases, the perturbation is not
extension mode
significant at distances from the foundation axis higher
asoð1Þ asoð2Þ apoð1Þ
than 70d.
Undamped stratum 0.11 0.31 0.26 1 On the other hand, Fig. 13 presents the distributions of
Pile group–soil system 0.12 0.33 0.25 incident field and central pile displacements along depth for
Undamped stratum 0.07 0.21 0.17 1.5 the four-layer soil profile. Non-dimensional frequencies ao
Pile group–soil system 0.09 0.23 0.17 ¼ 0.09, 0.20 and 0.30 are shown. It can be seen that the
pile is able to keep approximately to the incident field
Undamped stratum 0.05 0.16 0.13 2
Pile group–soil system 0.07 0.16 0.12 within the deep layers, where the wave length is long
enough. In the upper layers, on the contrary, where wave
s=d ¼ 10. lengths are shorter, the pile is not flexible enough so as to
follow the free field ground motion, which causes the
points on the free surface placed along axis y and x, for difference between pile cap and free field surface motions.
horizontal coordinates from 12 to 82 times the diameter of Additionally, Fig. 14 compares the deformed shapes of all
the piles. It can be seen that the perturbation generated by piles in the 3  3 pile group, and also of a fixed head single
 ARTICLE IN PRESS
340 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

7 4
3.5 H=L
6.5 H=1.5L
3 H=2L

G /(Σx2 ks )
φφ /(Σx i kzzo)

i zzo
6 half-space
2.5
2 s 1.5 2
1.5

cφφ
kG
1
1
0.5 0.5
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ao ao

Fig. 9. Rocking impedances of a 2  2 pile group embedded in a homogeneous stratum. s=d ¼ 10.

Fig. 10. Sketch of studied soil profiles.

1.5
1
1
0.5
Re (ucap/uff)

Im (ucap/uff)

0.5
0
0

-0.5 -0.5

-1 -1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
ao ao

Fig. 11. Time-harmonic displacement transfer functions at the pile cap.

pile (no rotations allowed), for the four-layer soil profile. It group of concrete piles of diameter d ¼ 1:0 m, being worth
can be seen that, for a certain non-dimensional frequency, noting that the resulting properties for the softest layer
all deformed shapes are very close one to another. Thus, correspond to the case of a considerably soft soil. The
there seems to be no significant pile-to-pile interaction system is subjected to the two following simulated earth-
effects under seismic excitation in the problem at hand. quakes, specified at the free surface:
However, as this is only shown for this particular case,
general conclusions would need of further studies.  The N-S component recorded at the Imperial Valley
Acceleration time-histories and response spectra can Irrigation District substation in El Centro, California,
now be obtained for particular cases. The chosen problem, during the Imperial Valley, California, earthquake of
whose parameters are summarized in Table 2, is that of a May 18, 1940.
 ARTICLE IN PRESS
L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346 341

Fig. 12. Horizontal deformation of free surface.

 ARTICLE IN PRESS
342 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

ao=0.30

ao=0.09 ao=0.20

Fig. 13. Distributions of incident field and central pile displacements along depth for the four-layer soil profile.

Fig. 14. Comparison of deformed shapes of all piles in the group, and of a single pile, for the four-layer soil profile.

Table 2 are significantly smaller in the range of periods shown in

Reference soil, pile and pile group properties Figs. 16 and 18. However, there is not a clear tendency
Soil Piles Pile group
between the response spectra for the different profiles as for
to say that the more layers considered, the more
cs ¼ 239 m=s E p ¼ 2:76  1010 N=m2 s=d ¼ 10 conservative estimate, neither the contrary. For instance,
rs ¼ 1750 kg=m3 rp ¼ 2500 kg=m3 L=d ¼ 15 for periods below 0.2 s, the response for the soft stratum is
ns ¼ 0:4 d ¼ 1m H=L ¼ 1 the lowest one, while between 0.25 and 0.4 s, it is the two-
strata profile whose values are the smallest.
Up to now, the same seismic excitation has been
 The S50W component recorded at Bonds Corner specified at the free surface for all soil profiles, but it is
substation in El Centro, California, during the Imperial also worth investigating the system behavior when the
Valley, California, earthquake of October 15, 1979. seismic input is specified at the bedrock. In this case,
different amplification phenomena due to the soil strati-
Figs. 15–18 show the acceleration time-histories and graphy take place in each profile, as can be seen in Fig. 19,
response spectra ðx ¼ 5%Þ, for both earthquake ground which presents the free field and pile cap acceleration
motions, all of them measured at the pile-cap. As expected response spectra ðx ¼ 5%Þ obtained when El Centro (1940)
in view of the transfer functions presented above, it can be earthquake is specified at the bedrock. Indeed, the strong
seen that the response of the piled foundation embedded in ground motion, measured at the free surface, is signifi-
a half-space or in a stratum, both with the reference soil cantly amplified, especially at frequencies near the strata
properties, are almost coincident with the excitation. As for natural frequencies. However, the attenuating effect of the
the rest of cases, the acceleration response spectra values piled foundation seen before, is also evident in this case. As
 ARTICLE IN PRESS
L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346 343

0.3 Free field 0.3 1 stratum

acceleration (g)

acceleration (g)
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
-0.3 pga=0.319 g -0.3 pga=0.316 g

0 2 4 6 8 10 0 2 4 6 8 10
t (s) t (s)
0.3 1 soft stratum 0.3 2 strata
acceleration (g)

0.2

acceleration (g)
0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
-0.3 pga=0.238 g pga=0.222 g
-0.3
0 2 4 6 8 10 0 2 4 6 8 10
t (s) t (s)
0.3 3 strata 0.3 4 strata
acceleration (g)

acceleration (g)
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
-0.3 pga=0.222 g -0.3 pga=0.226 g
0 2 4 6 8 10 0 2 4 6 8 10
t (s) t (s)

Fig. 15. Response acceleration histories for El Centro (1940) earthquake specified at the free surface.

Free field 1 soft stratum 1 stratum 3 strata 5. Conclusions

Halfspace 2 strata 4 strata
1 In this paper, a BEM–FEM coupling model previously
0.9 presented by the authors has been used to compute
0.8 dynamic impedances and kinematic response of piled
acceleration (g)

0.7 foundations embedded in viscoelastic homogeneous strata

0.6 resting on a rigid bedrock. This way, the two first steps of
0.5 the substructuring method are performed.
0.4
Piles have been modelled by FEM as beams according to the
Bernoulli hypothesis, and strata have been modelled by BEM
0.3
as continuum, semi-infinite, isotropic, homogeneous, linear,
0.2
viscoelastic media. The model not only allows the dynamic
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
analysis of piled foundations embedded in a half-space or in a
T (s)
stratum resting on a rigid bedrock, but also in multilayered
Fig. 16. Five per cent-damped acceleration response spectra for El Centro soils of generic stratigraphy, including deposits and inclusions.
(1940) earthquake specified at the free surface. Any kind of geometry can be modelled by boundary
discretizations, thus, more complex stratigraphies and topo-
expected, reference soil one-stratum free field and pile cap graphies than those reported in this paper can be analyzed.
response spectra are again almost coincident, but as for the Firstly, several results of vertical, horizontal and rocking
rest of cases, the presence of the piles always leads to impedances for single piles and three different configura-
smaller responses. tions of 2  2 pile groups embedded in a stratum have been
Since further parametric studies should be carried out presented. Several depths of the stratum have also been
with a greater variety of profiles, including parameters such studied. From the analysis of these results, the following
as densities, damping coefficients and Poisson ratios, conclusions can be drawn:
conclusions from this study may not be generalized. On
the other hand, only the horizontal response has been  The pile–soil system presents peaks associated to the
treated here but, depending on the problem, the rocking stratum natural frequencies that the numerical model is
behavior could also be an important parameter. able to predict adequately.
 ARTICLE IN PRESS
344 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

0.8 0.8
Free field 1 stratum

acceleration (g)

acceleration (g)
0.4 0.4

0 0

-0.4 -0.4
pga=0.778 g pga=0.769 g
-0.8 -0.8
0 2 4 6 8 10 0 2 4 6 8 10
t (s) t (s)
0.8 0.8
1 soft stratum 2 strata
acceleration (g)

acceleration (g)
0.4 0.4

0 0

-0.4 -0.4
pga=0.529 g pga=0.532 g
-0.8 -0.8
0 2 4 6 8 10 0 2 4 6 8 10
t (s) t (s)
0.8 0.8
3 strata 4 strata
acceleration (g)

acceleration (g)
0.4 0.4

0 0

-0.4 -0.4
pga=0.567 g pga=0.566 g
-0.8 -0.8
0 2 4 6 8 10 0 2 4 6 8 10
t (s) t (s)

Fig. 17. Response acceleration histories for El Centro (1979) earthquake specified at the free surface.

Fig. 18. Five per cent-damped acceleration response spectra for El Centro
(1979) earthquake specified at the free surface.

 The vertical impedance functions of a piled foundation

embedded in strata of the analyzed depths are equiva-
lent to those embedded in a half-space at frequencies
above 1.5 times the fundamental natural frequency of
the stratum in the compression–extension mode. Be-
sides, the foundation vertical behavior is only influenced
Fig. 19. Five per cent-damped acceleration response spectra for El Centro
by this natural frequency.
(1940) earthquake, specified at the bedrock.
 The influence of the stratum depth over the horizontal
impedance functions of a piled foundation is significant
over a broader band of frequencies. Thus, several peaks  This influence is even more evident in the damping
associated to both the shear and the compression– functions, which at low frequencies are much smaller
extension modes appear in the foundation response. than the corresponding to the half-space. On the
 ARTICLE IN PRESS
L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346 345

contrary, their values are similar at higher frequencies, resulting acceleration response spectra are significantly
which reveals that the main damping mechanism at lower than those corresponding to the free field.
intermediate and high frequencies is the energy dissipa-  the simplification of the soil profile to just one stratum
tion through surface waves for both the half-space and or to a half-space would lead to over- or under-
the strata. estimating, depending on the chosen properties, the
 The influence of the presence of the rigid bedrock is still excitation at the base of a superstructure in a
noticeable for a stratum depth five times the pile length. substructuring analysis.
 The group effect is predominant over the influence of  these conclusions are valid for earthquake motions
the presence of the rigid bedrock, especially in vertical specified at both the free surface or at the bedrock.
and rocking impedances, in which the peaks associated
to the natural frequencies of the stratum are of small
magnitude. However, the effects of the presence of the Acknowledgements
bedrock on the horizontal behavior are stronger and
become apparent along a broader frequency band. In all The authors would like to thank the reviewers for their
cases, the influence of the rigid bedrock increases with valuable comments, that have contributed to improve the
the ratio s=d. paper. This work was supported by the Ministry of
 The rocking behavior is the less influenced by the Education and Science of Spain through research project
presence of a rigid bedrock. BIA2004-03955-C02-02 and co-financed by the European
Fund of Regional Development. L.A. Padrón is recipient
Also, additional experiments (not shown here for the sake of the FPU research fellowship AP-2004-4858 from the
of brevity) that have been carried out in order to Ministry of Education and Science of Spain. The authors
investigate the role of the ratio E p =E s , show that the would like to thank for this support.
frequency band in which the foundation response is
influenced by the rigid bedrock broadens with the increase References
of the soil stiffness. Also, a larger number of natural
frequencies of the stratum in both the shear and the [1] Beskos DE. Boundary element methods in dynamic analysis: Part II
compression–extension modes become evident for the (1986–1996). Appl Mech Rev 1997;50:149–97.
horizontal and rocking cases as the soil stiffness increases. [2] Kaynia AM. Dynamic stiffness and seismic response of pile groups.
At the same time, the group effect becomes more evident as Research Report R83-03, Massachusetts Institute of Technology.
Cambridge, MA; 1982.
the ratio E p =E s turns higher. [3] Sen R, Davies TG, Banerjee PK. Dynamic analysis of piles and pile
Secondly, the seismic response of a square 3  3 hinged groups embedded in homogeneous soils. Earthquake Eng Struct Dyn
pile group embedded in different soil profiles and under 1985;13:53–65.
vertically incident plane time-harmonic shear waves and [4] Mamoon SM, Kaynia AM, Banerjee PK. Frequency domain
dynamic analysis of piles and pile groups. J Eng Mech ASCE 1990;
two different strong ground motions, have been studied.
116(10):2237–57.
General conclusions cannot be drawn without more and [5] Mamoon SM, Banerjee PK. Response of piles and pile groups to
deeper analysis in the subject but, as for the presented case, travelling SH-waves. Earthquake Eng Struct Dyn 1990;19:597–610.
it has been shown that: [6] Kaynia AM, Kausel E. Dynamics of piles and pile groups in layered
soil media. Soil Dyn Earthquake Eng 1991;10:386–401.
 the presence of a soft layer atop the soil yields to much [7] Gazetas G, Fan K, Kaynia AM, Kausel E. Dynamic interaction
factors for floating pile groups. J Geotech Eng ASCE 1991;117:
more rapidly decreasing displacement transfer functions 1531–48.
than those corresponding to foundations in the reference [8] Miura K, Kaynia AM, Masuda K, Kitamura E, Seto Y. Dynamic
homogeneous soil. In fact, two different trends are behaviour of pile foundations in homogeneous and non-homoge-
clearly defined, depending on whether the soft layer is neous media. Earthquake Eng Struct Dyn 1994;23:183–92.
[9] Davies TG, Sen R, Banerjee PK. Dynamic behavior of pile groups in
considered or not.
inhomogeneous soil. J Geotech Eng 1985;111(12):1365–79.
 the addition of further intermediate shear wave velocity [10] Kattis SE, Polyzos D, Beskos DE. Vibration isolation by a row of
layers in between is of minor importance. piles using a 3-D Frequency domain BEM. Int J Numer Meth Eng
 the incident field at the free surface is perturbed by the 1999;46:713–28.
presence of the piled foundation, being the magnitude of [11] Kattis SE, Polyzos D, Beskos DE. Modelling of pile wave barriers by
this perturbation larger when a soft layer exists atop. effective trenches and their screening effectiveness. Soil Dyn Earthq
Eng 1999;18:1–10.
The maximum value of this perturbation is of similar [12] Vinciprova F, Aznárez JJ, Maeso O, Oliveto G. Interaction of BEM
order in both the direction of motions induced by the S- analysis and experimental testing on pile–soil systems. In: Davini C,
waves and its perpendicular. Viola E, editors. Problems in structural identification and diagnostic:
 pile-to-pile interaction under seismic excitation is almost general aspects and applications. Berlin: Springer; 2003. p. 195–227.
non-existent. [13] Maeso O, Aznárez JJ, Garcı́a F. Dynamic impedances of piles and
groups of piles in saturated soils. Comput Struct 2005;83:769–82.
 piled foundations embedded in the non-homogeneous soil [14] Mamoon SM, Ahmad S. Seismic response of piles to obliquely
profiles studied here filter out a great part of the harmonic incident SH, SV, and P waves. J Geotech Eng ASCE 1990;116(2):
components of the seismic input, in such a way that the 186–204.
 ARTICLE IN PRESS
346 L.A. Padrón et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 333–346

[15] Masayuki H, Shoichi N. A study on pile forces of a pile group in [22] Guin J, Banerjee PK. Coupled soil-pile-structure interaction analysis
layered soil under seismic loadings. In: Proceedings of the II under seismic excitation. J Struct Eng 1998;124(4):434–44.
International Conference on Recent Advances in Geotech Earth- [23] Koo KK, Chau KT, Yang X, Lam SS, Wong YL. Soil-pile-structure
quake Engineering and Soil Dynamics, St Louis, Missouri, 1991. interaction under SH wave excitation. Earthquake Eng Struct Dyn
[16] Fan K, Gazetas G, Kaynia A, Kausel E, Ahmad S. Kinematic seismic 2003;32:395–415.
response of single piles and pile groups. J Geotech Eng ASCE 1991; [24] Padrón LA, Aznárez JJ, Maeso O. BEM–FEM coupling model for
117(12):1860–79. the dynamic analysis of piles and pile groups. Eng Anal Boundary
[17] Kaynia A, Novak M. Response of pile foundations to Rayleigh Elem 2007;31:473–84.
waves and obliquely incident body waves. Earthquake Eng Struct [25] Mendonc- a AV, de Paiva JB. A boundary element method for the
Dyn 1992;21:303–18. static analysis of raft foundations on piles. Eng Anal Boundary Elem
[18] Nogami T. Dynamic group effect of multiple piles under vertical 2000;24:237–47.
vibration. In: Proceedings of the ASCE Engineering Mechanics [26] Mendonc- a AV, Paiva JB. An elastostatic FEM/BEM analysis of
Division Specialty Conference, Austin, Texas; 1979. p. 750–4. vertically loaded raft and piled raft foundation. Eng Anal Boundary
[19] Velez A, Gazetas G, Krishnan R. Lateral dynamic response of Elem 2003;27:919–33.
constrained-head piles. J Geotech Eng ASCE 1983;109(8):1063–81. [27] Matos Filho R, Mendonc- a AV, Paiva JB. Static boundary element
[20] Gazetas G, Fan K, Tazoh T, Shimizu K, Kavvadas M, Makris N. analysis of piles submitted to horizontal and vertical loads. Eng Anal
Seismic pile-group-structure interaction. Geotechnical Special Pub- Boundary Elem 2005;29:195–203.
lication No. 34, ASCE, 1992. p. 56–93. [28] Domı́nguez J. Boundary elements in dynamics. Southampton, New
[21] Mylonakis G, Nikolaou A. Soil-pile-bridge seismic interaction: York: Computational Mechanics Publications & Elsevier Applied
kinematic and inertial effects. Part I: soft soil. Earthquake Eng Science; 1993.
Struct Dyn 1997;26:337–59.