Some aspects of the design of pile foundations

under seismic motion

Raffaele Di Laora,* Alessandro Mandolini*

Summary
The design of piles under seismic loads is conventionally carried out with reference to the inertial forces arising from struc-
tural vibrations and, hence, neglecting kinematic interaction between pile and soil. Even if in the last years several research
efforts gave particular attention to kinematic effects, they focused on bending moments at the interface between two soil layers
with different stiffness, while few contributions investigated kinematic effects at pile head.
The paper presents the results (both in the frequency and time domain) of extensive parametric studies, leading to a simplified
formula for predicting kinematic bending moments at pile head. The paper gives some indications about their relative impor-
tance with respect to the inertial ones and provides simply rules to combine inertial and kinematic maximum effects taking into
account their phase lag.

1. Introduction Some comments about Codes statements will be ex-

pressed later on.
During earthquakes, even in absence of super- As can be noticed, Codes suggest to design piles
structure, a pile foundation is forced to deform (and with reference to inertial and kinematic effects,
then to experiment internal forces) by ground implicitly splitting the complex soil-pile-structure
shaken due to seismic motion. Moreover, because of interaction in the light of substructure method.
its different stiffness, pile opposes to soil displace- Indeed, under the hypothesis of applicability of
ment, resulting in an interaction phenomenon re- superposition principle (and hence linear elastic be-
ferred to as kinematic. haviour of materials), it would be straightforward to
The problem has been studied by several re- show that the complete system may be analysed per-
searchers [FLORES-BERRONES et al., 1982; DOBRY et al., forming three consecutive steps [KRAMER, 1996]:
1983; KAVVADAS et al., 1993; NIKOLAOU et al., 1995;
1) a kinematic interaction analysis, in which super-
NIKOLAOU et al., 2001; MYLONAKIS, 2001; MAIORANO et
structure has no mass, aimed to provide (a) the
al., 2009], and its destructive effects have been con-
first part of stresses and strains acting along
firmed by some post-earthquake investigations [MI-
piles and (b) the foundation motion;
ZUNO, 1987].
2) an inertial interaction analysis in which the
The importance of kinematic interaction has
been recently recognized by several recent seismic structure, shaken by foundation input motion,
Codes such as Eurocode 8 [EN 1998-5, 2003] and induce forces and displacements at its founda-
the Italian Code [D.M. 14.1.08]. For example, Eu- tion, eventually replaced by a set of springs and
rocode 8 states that piles shall be designed to resist the dashpots representing the dynamic impedance
following two types of action effects: (a) inertia forces from of the soil-foundation system;
the superstructure; (b) kinematic forces arising from the 3) derivation of the second part of stresses and
deformation of the surrounding soil due to the passage of strains induced along piles by inertial forces.
seismic waves, and that bending moments developing Moreover, a further simplification may be in-
due to kinematic interaction shall be computed only when troduced in the procedure. DI LAORA (2009) showed
all of the following conditions occur simultaneously: (1) that the importance of the filtering effect exerted
the ground profile is of type D, S1 or S2, and contains con- by the piles on seismic motion depends on various
secutive layers of sharply differing stiffness; (2) the zone is parameters; with reference to fixed-head piles em-
of moderate or high seismicity, i.e. the product agS exceeds bedded in two-layers soil deposits, the discrepancy
0.10g; (3) the supported structure is of class III or IV. between free-field and foundation motions gener-
The Italian Code provides quite similar indications. ally increases with increasing the stiffness contrast,
the stiffness ratio between pile and soil, the fre-
quency of excitation and with decreasing the inter-
* Department of Civil Engineering Second University of Naples. face depth.

RIVISTA ITALIANA DI GEOTECNICA 1/2011

64 DI LAORA - MANDOLINI

The results of finite element analyses of 612 program) and, hence, the keypoint lies in evaluat-
cases in linear elasticity hypothesis showed that in ing the pile-soil curvature ratio, that clearly express
rare cases, in which the following conditions: the interaction.
shallow interface (i.e., located inside the so Using Beam on Dynamic Winkler Foundation
called active lenght of the pile) models, it is straightforward to show that, for a long
large stiffness contrast between layers (Vs2/Vs1 > fixed-head embedded in a homogeneous halfspace
3, with Vs1 and Vs2 shear wave velocity of first and the ratio of pile and soil curvature at the top is ob-
second soil layer) tained as:
large pile-to-soil stiffness ratio (Ep/E110000,
with Ep and E1 Youngs moduli of pile and first
soil layer) (1)
are contemporary present, it is opportune to
study the interaction case by case; it is possible to ne- in which accounts for pile-soil interaction. Under
glect the filtering effect (without a too large conserv- harmonic oscillations, it can be determined analyti-
atism) only for large structural periods Ts > 0.5 s. cally as [FLORO-BERRONES et al., 1982; NIKOLAOU et al.,
For the other (frequent) cases it is always possi- 1995]:
ble to refer to free-field motion with reference to
low-frequency bedrock signals. In the case of high-
frequency signals it is possible to neglect the filter-
ing effect exerted by the piles only when the struc- (2)
tural period is larger than 0.3 s. Details are available
in the mentioned reference. where Ep and Ip are the pile Young modulus and
Despite the important research effort in last cross-sectional moment of inertia, Vs the velocity of
decades, engineers agree to design piles for resist- shear waves in the soil, the cyclic vibrational fre-
ing only inertial forces. Simplified formulas are quency of the excitation and k the modulus of the
available in scientific literature, but mainly referred Winkler springs (modulus of subgrade reaction). It
to kinematic moment generating at the interface be- may be shown that is always smaller than 1 and de-
tween two layers having different stiffness. Less at- creases monotonically with increasing frequency. DI
tention has been devoted in regards of kinematic LAORA [2009] highlighted that even for large pile-
bending at the top of a pile whose head is fixed soil stiffness ratios (up to 10000) curvature ratio in
against rotation. homogeneous soils starts to decrease in correspond-
This topic is of critical importance, as pile is ence of frequencies larger than those relevant in re-
head subjected to both kinematic and inertial al earthquakes and, hence, it may be assumed in de-
forces; the latter, as well-known, vanish within a sign equal to 1.
depth of about ten diameters [RANDOLPH, 1981; MY-
MYLONAKIS [1999], by using a BDWF model, de-
LONAKIS, 1995].
rived a closed-form expression of the curvature ra-
In the ensuing it is tried to clarify some aspects
tio for a fixed-head pile of finite length in a homo-
regulating the kinematic moment at pile head, pro-
geneous layer of thickness h over a rigid base, show-
viding a simplified formula for its estimation. Some
ing that it assumes values greater than unity at low
considerations about kinematic vs. inertial relative
importance are expressed, providing also a rule to frequencies for certain slenderness ratios h/d, where
combine their maximum effects in the light of their d is the pile diameter.
phase lag. Further contributions on kinematic bending as-
sessment at pile head in layered soils are provided
by DEZI et al., 2009 and DE SANCTIS et al., 2010; de-
2. Kinematic interaction spite these efforts, the role played by certain param-
eters, such as the interplay between interface and
head bending, have not been adequately addresses.
2.1. Available methods for assessing kinematic bending at
pile head
2.2. Numerical analyses for two-layer soils
Kinematic forces acting along piles may be
viewed as the result of two counteracting phenom- In order to explore pile-soil interaction at pile
ena: (1) the deflected shape that soil tries to impose head, some numerical analyses regarding a single
on pile and (2) the resistance that pile opposes fixed-head pile embedded in a two-layers soil de-
through its flexural stiffness. posit have been carried out by FE program ANSYS
The free-field deformations can be readily cal- v. 10.0, under the assumption of linear elastic be-
culated (for instance using an established computer haviour of both soil and pile.

RIVISTA ITALIANA DI GEOTECNICA

 SOME ASPECTS OF THE DESIGN OF PILE FOUNDATIONS UNDER SEISMIC MOTION 65

d and 55 d (respectively for the direction of shaking

and the perpendicular direction) away from the
pile. Vertical displacement have been neglected on
lateral boundaries to simulate 1-D conditions for
ground response, nodes at the base of the model
(whose height is set to 30 m) have been fully re-
strained to represent a rigid bedrock, properly re-
straints have been used on symmetry and antisym-
metry planes. 8-noded isoparametric brick elements
have been utilised to mesh soil and pile. A sensitivity
analysis has been performed to establish the height
of the elements, which has been set to 1 d except
within a region extending 1.5 d above and below the
interface, where element height is 0.1 d. No inter-
face elements has been considered between pile and
soil.
Reference geometrical scheme is reported in
Figure 1, while FE model is shown in Figure 2.

2.3. Static and frequency domain response

Fig. 1 Reference geometrical scheme. In Figure 3 pile-to-soil curvature ratios against

Fig. 1 Schema geometrico di riferimento. dimensionless frequency /1 (with 1 cyclic vibrational
frequency of the ground) are represented by varying
depth of interface, pile-soil stiffness ratio and stiff-
The analyses have been performed in the fre- ness contrast between the two layers.
quency domain, superimposing the effects of a sig- As evident from the Figure 3, curvature ratio de-
creases with increasing frequency and can be larger
nificant number of modes of vibration (whose fre-
than unity (in accord with MYLONAKIS, 1999). The
quency is lower than 12-15 times the fundamental
latter result for static loads (i.e., at 0) may be inter-
frequency of the ground, and in any case less than
preted as a consequence of an interplay between in-
25 Hz). The input load consists of a unit accelera- terface and head bending moment, so that a further
tion having different frequencies. FFT and IFFT al- bending moment is transmitted to pile head.
gorithms have been used to switch from time to fre- In the light of this interpretation, the stiffness
quency domain and vice versa. contrast between soil layers represents the degree of
The plan dimensions of the model have been restraint and is the main responsible of the amount
selected ensuring free-field response at boundaries of moment this restraint generates at interface. This
not affected by waves spreading away from pile-soil means that the larger is its value the larger is the
interface; the boundaries have been then located 80 bending induced at cap. In Figure 3a is evident that

Fig. 2 Finite Element model.

Fig. 2 Modello agli Elementi Finiti.

GENNAIO - MARZO 2011

66 DI LAORA - MANDOLINI

the additional contribution increases with increas-

ing Vs2/Vs1.
Nevertheless, by the analysis of the other graphs
it can be noticed that the additional bending has no
monotonous trend by varying depth of interface
and stiffness ratio. These results, apparently not in-
tuitive, may find a physical explanation by consider-
ing a beam resting on a Winkler medium loaded by
a moment at one end (representing the interface)
and fixed against rotation at the other end (repre-
senting pile head). On the basis of this simple
scheme, Figure 4 depicts the fraction of bending
transmitted at restraint in function of dimensionless
beam length L/with

(3)

It is worthy of note that the trend is very similar

to the distribution of bending moments along a
flexible Winkler beam loaded by a moment at one
end, thereby assuming a zero value at L/ (3/4)
and a local minimum at L/ .
In the light of this scheme, the interface acts as
a source and transmits bending to pile head pro-
viding a further contribution, that is, with reference
to Figure 3b subtractive up to h1/d = 6 and addi-
tional until h1/d = 10, while for larger values of in-
terface depth the bending contribution tends to
vanish.
This interpretation helps to justify the trends of
curvature ratio but should not be applied from a
quantitative point of view, as the mechanism is more
complicated.

Fig. 3 Influence of excitation frequency, by varying stiff-

ness contrast, pile-soil stiffness ratio, interface depth
(d=0.5 m, E1 =50 MPa).
Fig. 3 Influenza della frequenza delleccitazione al variare del Fig. 4 Fraction of bending moment transmitted at
contrasto di rigidezza, rapporto di rigidezza palo-terreno, restrained end.
profondit dellinterfaccia (d = 0.5 m, E1 = 50 MPa). Fig. 4 Frazione di momento trasmessa allestremit vincolata.

RIVISTA ITALIANA DI GEOTECNICA

SOME ASPECTS OF THE DESIGN OF PILE FOUNDATIONS UNDER SEISMIC MOTION 67

Tab. I Main features of the signals used in the analyses.

Tab. I Caratteristiche principali dei segnali usati nelle analisi.
Arias Predom. Mean
Ep. distance PGA
Station name Earthquake MI Component Intensity period period
[km] [g]
[m/s] [s] [s]
Tolmezzo -Diga
Friuli, 06/05/1976 6.4 20.23 NS 0.357 0.7865 0.26 0.39
Ambiesta
Irpinia 1st, 23/11/
Sturno 6.5 30.37 WE 0.321 1.3935 0.20 0.85
1980
Umbria-Marche
Borgo - Cerreto Torre 5.1 10.25 WE 0.162 0.0648 0.06 0.19
(AS), 12/10/1997
Friuli (AS), 11/09/
San Rocco 5.8 23.96 NS 0.090 0.0322 0.28 0.54
1976
Friuli (AS), 11/09/
Tarcento 5.5 8.38 NS 0.211 0.1079 0.10 0.22
1976
Nocera Umbra - Umbria-Marche
5.0 6.87 NS 0.186 0.1582 0.12 0.17
Biscontini (AS), 03/10/1997

Similar considerations can be drawn in regards of soil layers s1 = s2 = 0.1; pile Poisson coefficient
of the influence of stiffness ratio (Fig. 3c), reversing p = 0.2; soil Poisson coefficient s1 = s2 = 0.3.
the trends. For large pile-to-soil stiffness ratios (and while its variable parameters are:
then large pile wavelengths) bending moment at Ep/E1 = 300, 1000, 10000
pile head decreases, while for moderate stiffness ra- Vs2/Vs1 = 1.5, 2, 3, 6
tios the interface increases bending; as expected, for h1/d = 2, 4, 8, 16 with h1 interface depth.
very soft piles curvature ratio at low frequencies
In the second parametric study the fixed parame-
tends to unity.
ters are the following:
Hence, while the stiffness contrast mainly rules
H = 30 m; d = 0.5 m; Ep = 30 GPa; p = 2.5 Mg/m3;
the amount of bending generated at interface, the
other two parameters rule the fraction of bending s1 = 1.6 Mg/m3; s2 = 1.8 Mg/m3; s1 = 0.1; s2 =
transmitted from interface to pile head. 0.1; p = 0.2; s1 = 0.3; s2 = 0.3.
Its variable parameters are:
Ep/E1= 150, 666, 1500
2.4. Parametric investigation for time domain response Vs2/Vs1 = 1.5, 2, 3
h1/d = 4, 8, 16
As the curvature ratio is strongly dependent on
the exitation frequency, the maximum bending mo- L/d = 24, 40, with L pile length.
ment at pile head generating during earthquakes is In the first parametric study, the density of the
affected by the distribution of relevant frequencies second layer is equal to 1.8 Mg/m3 when the stiffness
into the input motion. contrast is 1.5 or 2, 2.0 Mg/m3 when the latter is
To find a simple criterion for estimating design equal to 3 or 6.
kinematic moment at pile head, several analyses Six signals taken from the Italian Database
have been carried out by varying key parameters in- [SCASSERA et al., 2006], selected in order to cover a
fluencing the interaction phenomenon. wide range of relevant frequencies, have been
By applying dimensional analysis and selecting adopted as input motion. Their main features are
relevant non-dimensional products that may con- reported in Table I.
cern, a large number of parameters should be taken In total 612 cases have been analyzed.
as variables to explore their influence. To limit the
number of cases, two different parametric studies 2.4.1. RESULTS OF THE ANALYSES
have been performed, by varying their fixed param-
eters. As evident from the previous graphs in Figure 3,
The latters, in terms of physical quantities, for curvature ratio is sensitive to the moment generated
the first one are: at the interface. If the latter is located beyond a cer-
Total height H = 30 m; pile diameter d = 1 m; tain distance, dependent on pile-soil stiffness ratio,
its effects are negligible at pile head.
E1 = 50 MPa; pile density p = 2.5 Mg/m3; first soil
Figure 5 shows the correlation between the
layer density s1 = 1.6 Mg/m3; second soil layer den- maximum pile and soil curvatures for cases in which
sity s2 = 1.8 Mg/m3 or 2.0 Mg/m3; damping ratios interface depth is smaller (empty dots) and larger

GENNAIO - MARZO 2011

68 DI LAORA - MANDOLINI

For these cases, although at present time no cor-

relation seems to exist in order to obtain a good es-
timation of pile curvature, it is clear that a unit cur-
vature ratio represents a conservative assumption.

3. Kinematic vs. inertial interaction

Several recent seismic Codes (among which Ital-
ian NTC and Eurocode 8) impose to evaluate the ef-
fects of kinematic interaction only under certain
conditions, but regardless of the amount and the
distribution of structural masses.
In the authors opinion, it should be allowable to
neglect kinematic interaction when its effects are
negligible with respect to inertial ones; hence, the
Codes indications should be improved, as the iner-
tial effects, contrary to kinematic, strongly depend
on the structural mass.
Fig. 5 Correlation between pile and soil curvature. To investigate the relative importance of iner-
Fig. 5. Correlazione tra curvatura del palo e del terreno. tial vs. kinematic interaction DI LAORA [2009] built a
simple analytical tool for evaluating the inertial re-
sponse at the top of a group of piles restrained to a
(full dots) than an active length defined as [GAZETAS, rigid raft, by using the static expressions provided
1991]: by RANDOLPH et al., [1979] and RANDOLPH, [1981]
both for axial/lateral stiffness of a single pile and in-
teraction coefficients. Soil is considered to be homo-
(4)
geneous; despite this, the presence of a deep inter-
face (located beyond the pile active length) doesnt
The graph clearly explains that for deep inter- affect the general trends. Further details are availa-
faces curvature ratio for kinematic loading may be ble in aforementioned reference.
still assumed equal to 1. The inertial horizontal action at pile cap is
This appears as a very useful result. First, no in- taken equal to the product of structure mass and
teraction analysis has to be performed to obtain a horizontal surface acceleration (i.e. the height of the
good estimation of kinematic bending moment at structure is zero).
pile head; second, neither a ground response anal- On the other hand, kinematic moment at pile
ysis is required, as it is straightforward to show that head can be well estimated, as shown above,
in one-dimensional case soil curvature may be writ- through the formula:
ten as:
(6)
(5)
In this manner kinematic vs. inertial relative im-
where as is the acceleration of soil at surface. portance may be investigated by varying key param-
The design value of surface acceleration may be eters, deriving general trends about the behaviour
assumed on the basis of the statements of the refer- of piles under seismic shaking.
ence Code, allowing to get a simple estimation of The most relevant issue in seismic performance
kinematic moment without performing complex dy- of piles is that large diameter piles tend to suffer
namic analyses. kinematic loads rather than inertial ones.
If the interface is close to pile head, a complex Indeed, as the piles are generally designed to
interaction between pile head and interface arises, carry vertical load, the mass pertaining to each pile
depending on all the aforementioned nondimen- (say Mp) is dependent on its radius r. In particular,
sional factors. it appears useful to refer to two extreme cases, un-
As can be noticed some cases referring to shal- der the hypothesis of constant ratio between pile
low interfaces still have a curvature ratio that is very bearing capacity and Mp:
close to unity. This reinforces the concept previ- considering only the shaft contribution in bear-
ously expressed about the influence of interface on ing capacity, structural mass per pile is propor-
head bending. tional to the radius by a coefficient 1 depend-

RIVISTA ITALIANA DI GEOTECNICA

SOME ASPECTS OF THE DESIGN OF PILE FOUNDATIONS UNDER SEISMIC MOTION 69

Fig. 7 Influence of soil Youngs modulus on inertial vs.

kinematic moment at pile head (r=0.4 m, Mp =80 Mg,
L=20 m, s=6r, n=9, Ep =30 GPa, s =1.6 Mg/m3, aff =3
m/s2, M/V=0).
Fig. 7 Influenza del modulo di Young del terreno sul rapporto
relativo momento inerziale/cinematico alla testa.

imposed to satisfy structural requirements. Further-

more, while group effects are negligible for kine-
matic bending, interaction among piles implies that
corner piles carry a larger inertial moment with re-
spect to central ones.
It is worthy of note that the abovementioned fig-
ures indicate just trends of kinematic vs. inertial
forces by varying pile diameter, and not their actual
values, the latter being strongly dependent on soil
Fig. 6 Influence of pile radius on inertial vs. kinematic
stiffness that is to say, even a large pile, if embed-
moment at pile head (a) Mp = 1 r, 1 = 200 Mg/m, (b)
ded in a stiff soil, is stressed by inertial rather than
Mp = 2 r2, 2 =500 Mg/m2 (L=20 m, s=6r, n=9, Ep =30
kinematic moments.
GPa, Es =30 MPa, s =1.6 Mg/m3, aff =3 m/s2, M/V=0). In Figure 7 the trend of seismic moments are
Fig. 6 Influenza del raggio del palo sul rapporto relativo
depicted by varying the soil Youngs modulus. As
momento inerziale/cinematico alla testa.
expected, kinematic moment at pile head rapidly
vanish with increasing soil stiffness.
In order to provide a quantitative example, a
ing on installation technology and soil proper- pile having a diameter equal to 1000 mm, embed-
ties; ded in a class C subsoil with Vs1 = 80 m/s, Vs2 = 350
if the bearing capacity of the pile is due only to m/s, h1 = 8m carries a kinematic bending moment
its tip, Mp is proportional to r2 by a coefficient 2 equal to 700 kNm. This value clearly proves that
depending on the same parameters mentioned kinematic effects may be of relevant importance
even in some cases not highlighted by Codes.
for 1.
Furthermore, its important to remind that
It follows that, as mass per pile is proportional
these kinematic forces are referred only to pile top;
to inertial moment at pile head and kinematic mo- in presence of sharp stiffness contrasts between two
ment is proportional to r4, the latter tend to be pre- consecutive layers bending at interface may be even
dominant with increasing diameter. more severe.
Figure 6 clearly shows these trends; it can be no- The influence of the pile number and the ratio
ticed that kinematic moment reaches large values between moment and lateral force M/V acting on
even for moderately large diameters; this means pile cap is depicted in Figures 8 and 9. To justify the
that in very soft soils a limit (maximum) diameter is trends some remarks are needed.

GENNAIO - MARZO 2011

70 DI LAORA - MANDOLINI

Hence, the sign and the amount of the resulting

moment at pile head (connected to a rigid cap) is
function of the generalised forces acting upon the
raft (lateral force and moment) and geometrical-
mechanical features of the soil-foundation system.
On the basis of these remarks, it can be seen
from Figure 8 that, even fixing the mass per pile, in-
ertial moments tend to increase with increasing pile
number.
Indeed, increasing the latter with a spacing pro-
portional to diameter, the subtractive moment con-
tribution induced by raft rotation decreases, taking
as positive the moments due to raft translation.
Figure 9 shows the effect of the ratio between
moment M and lateral force V loading the raft; this
ratio, in the case of a SDOF system, is clearly equal
to the height of the structure.
As loading moment increases, bending at pile
top decreases approaching zero, then changing in
Fig. 8 Influence of pile number on inertial vs. kinematic sign and growing up, in absolute value, as lateral
moment at pile head (r= 0.4 m, Mp =80 Mg, L=20 m, force is fixed and hence an increase in M/V means
s=6r, Ep =30 GPa, Es =30 MPa, s =1.6 Mg/m3, aff =3 m/ an increase in M.
s2, M/V=0).
Fig. 8 Influenza del numero di pali sul rapporto relativo
momento inerziale/cinematico alla testa.
4. Phase lag between kinematic and inertial
interaction
The ideal separation of total strains and stresses
acting on piles during earthquakes in the light of
substructure method appears advantageous, as a
simplified estimation of kinematic and inertial con-
tribution is allowed. Nevertheless, once one has es-
timated the moments at pile cap due to both inter-
action phenomena, a simple sum of them may lead
to a large overestimation of the actual forces acting
along the piles, as the superstructure can generate a
bending time history at pile top in opposition to the
kinematic one.
It appears useful, hence, to find a simple crite-
rion to predict the phase lag between kinematic and
inertial effects and then combine their maximum ef-
fects.
The distinction of inertial effects induced by raft
translation and rotation above introduced is here
essential, as it appears evident that the two different
contributions have a phase lag each other equal to
Fig. 9 Influence of M/V on inertial vs. kinematic moment
180.
at pile head (r=0.4 m, Mp =80 Mg, L=20 m, s=6r, n=9,
Ep =30 GPa, Es =30 MPa, s =1.6 Mg/m3, aff =3 m/s2). D I L AORA [2010] proposed a useful criterion
Fig. 9 Influenza di M/V sul rapporto relativo momento aimed to a quick estimation of kinematic vs. inertial
inerziale/cinematico alla testa. phase lag in a dynamic problem.
Starting from theoretical considerations he
found that, with reference to a SDOF structure and
With reference to a pile group surmounted by a in the hypotheses of:
rigid raft, inertial moments arising at pile heads are linear elastic behaviour of materials (that is also
the result of two separate contributions related to the fundamental assumption of the substructure
the translation and the rotation of the raft. This sep- method)
aration is fundamental, as the two mechanisms pro- inertial moments due to translation of raft pre-
duce opposite-signed bending. vailing over those induced by its rotation the

RIVISTA ITALIANA DI GEOTECNICA

SOME ASPECTS OF THE DESIGN OF PILE FOUNDATIONS UNDER SEISMIC MOTION 71

Tab. II Presented cases. fs and Hs are the natural frequency and the height of the structure.
Tab. II Casi presentati. fs e Hs sono la frequenza naturale e laltezza della struttura.
Case E1 [MPa] f1 [Hz] fS [Hz] Hs [m]
A 30 0.71 1.45 10
B 120 1.42 1.45 10
C 120 1.42 1.02 10
D 30 0.71 3.41 5

Fig. 10 Phase angle between force and displacement in Fig. 11 Surface signal Fourier spectra and structure fre-
a SDOF system. quencies of analysed cases.
Fig. 10 Angolo di fase tra forza e spostamento in un sistema Fig. 11 Spettri di Fourier de segnali in superficie e frequenze
SDOF. della struttura dei casi analizzati.

phase lag at a given frequency can be ex- Although the kinematic vs. inertial phase lag de-
pressed as: pends only, frequency by frequency, on inertial in-
teraction through the structural period, the overall
(7) response under seismic motion is strongly depend-
where KM and IM are the kinematic and inertial ent on kinematic interaction, as the latter rules the
bending moment at pile top, the well-known relative importance of the in-phase and out-of-
phase angle: phase harmonics within the foundation input mo-
tion.
(8) Moreover, the author highlighted that for prac-
tical problems the filtering effect exerted by piles
where IF is the force acting on a SDOF system and doesnt shift the relevant frequencies.
SD its displacement. In the light of these statements, it is possible to
This phase angle, as it can be noticed in Figure predict the phase lag, first performing a seismic
10, for zero damping is 0 [or 180] for frequencies ground response analysis, then comparing the rele-
of kinematic acceleration that are lower [or larger]
vant frequencies of the surface signal with the fun-
than the structural frequency, the latter evaluated
damental frequency of the structure (evaluated tak-
considering the period elongation due to soil-struc-
ing into account SSI interaction).
ture interaction.
The reliability of this simple method was also
If the bending moments due to the rotation of
the raft prevail the phase lag between inertial and checked by FE numerical analyses; some results (re-
kinematic effects is in opposition to the previous sumed in Tab. II) are shown herein. Numerical de-
case. tails are available in the original work.
The figure also shows that the presence of Cases A-C represent cases with inertial moments
damping attenuates the sudden response shift at pile head due to the rotation of the raft prevailing
around the structural frequency; nevertheless, as- over those generated by its translation; an opposite
suming that low and high frequencies produce re- behaviour is represented in the case D.
sponses in opposition appears a reasonable approx- Some relevant results are shown below; in Fig-
imation. ure 11 frequencies of structures are depicted (verti-

GENNAIO - MARZO 2011

72 DI LAORA - MANDOLINI

Fig. 12 Time histories of total, kinematic and inertial stress at outer fiber of central pile.
Fig. 12 Storie temporali della tensione totale, cinematica e inerziale alla fifra esterna del palo centrale.

cal lines) against the frequency content of the sur- The case C TOLMEZZO represents structural
face free-field acceleration. frequencies lower than those of the surface free-
In the case A STURNO signal the structural fre- field acceleration: no phase lag is expected.
quency is larger than the relevant frequencies of sur- The difference between case A and case D is in
face acceleration. The proposed method, then, pre- the structure height so that, in the latter, bending
dicts out-of-phase kinematic vs. inertial response. moments due to the translation of the raft prevail;
In the case B TOLMEZZO the frequencies of consequently, the response should be in-phase.
acceleration and structure are very similar, so that a The results are shown in Figure 12. As it can be
phase lag of 90 is expected. Its worthy of note that noticed, the results fully confirm the predictions.
in this case there is resonance between input signal Trying to derive general conclusions, if bend-
(at foundation) and structure. ing moments due to the rotation of the raft prevail,

RIVISTA ITALIANA DI GEOTECNICA

SOME ASPECTS OF THE DESIGN OF PILE FOUNDATIONS UNDER SEISMIC MOTION 73

the phase lag decreases (fixing the other condi- Acknowledgements

tions):
decreasing the frequency of the structure; This research has been developed under the
increasing the frequency of the ground; auspices of the research project ReLUIS Innova-
shifting the frequency content of input signal to tive methods for the design of geotechnical systems
higher frequencies. under earthquake excitation, funded by Diparti-
On the other hand, if bending moments due to mento della Protezione Civile (National Emergency
raft translation prevail, opposite conclusions have to Management Agency).
be derived (i.e., the abovementioned three condi-
tions increase the phase lag).
For practical purposes, it is suggested to com- References
bine maximum kinematic and inertial effects in the
following manner: DE SANCTIS L., MAIORANO R.M.S., AVERSA S. (2010)
A method for assessing bending moments at the pile
in the case they are in-phase, summing the
head. Earthquake Engng Struct. Dyn., vol.
maxima;
XXXIX, pp. 375-397.
if the phase lag is approximately 180, taking
DEZI F., CARBONARI S., LEONI G. (2009) Kinematic
the maximum between the kinematic and the
bending moments in pile foundations. Soil Dynamics
inertial contribution;
and Earthquake Engineering, 30, n. 3, pp. 119-
in the rare case in which the phase lag is about 132.
90, it appears still reasonable to sum maximum
DI LAORA R. (2009) Seismic soil-structure interaction
effects as this doesnt produce a significant over- for pile supported systems. Ph.D. Thesis, University
estimation; it is worth noting that in this case in- of Napoli Federico II.
ertial interaction may induce particularly severe DI LAORA R. (2010) Phase lag between kinematic and
effects because of the resonance between foun- inertial interaction. Eygec 2010, Brno.
dation input motion and structure.
D.M. 14.1.2008 del Ministero delle Infrastrutture -
Nuove norme tecniche per le costruzioni. S.O. n. 30
alla G.U. del 4.2.2008, n. 29 [In Italian].
5. Concluding remarks DOBRY R., O'ROURKE M. J. (1983) Discussion on
Seismic response of end-bearing piles. by Flores-Ber-
Some issues relative to seismic design of piled rones, R. & Whitman, R. V. J. Geotech. Engng
foundation have been discussed in the light of nu- Div., ASCE p. 109.
merical and analytical results regarding piles em- EN 1998-5 - Eurocode 8: Design of structures for earth-
bedded in two layers soil deposits. quake resistance Part 5: Foundations, retaining
It has been shown that pile-soil curvature ratio structures and geotechnical aspects. CEN European
at pile head, contrary to the homogeneous soil case, Committee for Standardization, Bruxelles, Bel-
may assume values much different from unity. De- gium.
spite this, as earthquakes are characterised by a FLORES-BERRONES R., WHITMAN, R. V. (1982) Seismic
rather large range of dominant frequencies, the response of end-bearing piles. J. Geotech. Engng
time domain results show that the assumption of a Div., ASCE, CVIII, n. 4, pp. 554-569.
pile head curvature equal to that of soil provides a GAZETAS G. (1991) Foundation Vibrations in Foundat-
bending moment rather accurate for deep inter- ion Engineering Handbook. H.Y. Fang (Ed.), Klu-
faces; in presence of a shallow interface it represents wer Academic Pub., Boston.
a conservative assumption. It follows that, for a KAVVADAS M., GAZETAS G. (1993) Kinematic seismic re-
given case (fixed pile and soil properties), only the sponse and bending of free-head piles in layered soil.
maximum value of surface free field acceleration is Gotechnique, vol. XXXXIII, n. 2, pp. 207-222.
required as design input (i.e, taken from Codes). KRAMER S.L. (1996) Geotechnical Earthquake Engin-
Some important aspects regulating kinematic eering. Publ. Prentice Hall Upper Saddle River,
vs. inertial effects have been highlighted, pointing N.J.
out also that kinematic bending moments may as- MAIORANO R.M.S., DE SANCTIS L., AVERSA S., MANDO-
sume relevant values also in those cases not consid- LINI A. (2009) Kinematic response analysis of piled
ered as relevant by the Codes. foundations under seismic excitations. Canadian Geo-
Finally, a simple method has been proposed to technical Journal, 46, n.5.
estimate the phase lag existing between kinematic MIZUNO H. (1987) Pile damage during earthquakes in
and inertial interaction, proposing a rule to com- Japan, In Dynamic response of pile foundations. T.
bine their maximum effects in order to avoid useless Nogami (Ed.), ASCE Special Publication, pp. 53-
overconservatism. 78.

GENNAIO - MARZO 2011

74 DI LAORA - MANDOLINI

MYLONAKIS G. (1995) Contributions to Static and Dy- U.S. National Conference on Earthquake Engi-
namic Analysis of Piles and Pile-Supported Bridge neering, San Francisco, CD ROM, Paper n. 1824.
Piers. Ph.D. Thesis, State University of New York
at Buffalo.
MYLONAKIS G. (1999) Analytical solutions for seismic
pile bending. Unpublished research report, City
University of New York.
MYLONAKIS G. (2001) Simplified model for seismic pile Alcuni aspetti del progetto di pali di
bending at soil layer interfaces. Soils and Founda-
tions, 41, n. 4, pp. 47-58. fondazione sotto moto sismico
N IKOLAOU A., M YLONAKIS G., GAZETAS G. (1995)
Kinematic Bending Moments in Seismically Stressed
Piles. Report. NCEER-95-0022, National Center Sommario
Il progetto dei pali sotto azioni sismiche convenzionalmente
for Earthquake Engineering Research, State
eseguito con riferimento alle forze di inerzia che provengono
University of New York, Buffalo.
dalle vibrazioni della struttura e, quindi, trascurando
NIKOLAOU A., MYLONAKIS G., GAZETAS G., TAZOH T.
linterazione cinematica tra palo e terreno. Anche se negli ultimi
(2001) Kinematic pile bending during earthquakes
anni molti contributi di ricerca hanno rivolto particolare
analysis and field measurements. Gotechnique, 51, attenzione agli effetti cinematici, essi si sono focalizzati sui
n. 5, pp. 425-440. momenti flettenti allinterfaccia tra due strati di terreno di
RANDOLPH M.F. (1981) Response of flexible piles to lat- diversa rigidezza, mentre pochi contributi hanno studiato gli
eral loading. Gotechnique, 31, n. 2, pp. 247-259. effetti cinematici alla testa del palo.
RANDOLPH M.F., WROTH C.P. (1979) An analysis of Larticolo presenta i risultati (sia nel dominio delle frequenze
the vertical deformation of pile groups. Gotech- sia del tempo) di estesi studi parametrici che portano a una
nique, 29, n. 4, pp. 423-439. formula semplificata per stimare i momenti cinematici alla testa
SCASSERRA G., LANZO G., MOLLAIOLI F., STEWART J.P., del palo.
BAZZURRO P., DECANINI L.D. (2006) Preliminary Inoltre, larticolo d alcune indicazioni circa la loro
comparison of ground motions from earthquakes in importanza relativa rispetto a quelli inerziali e fornisce semplici
Italy with ground motion prediction equations for regole per combinare i massimi momenti inerziali e cinematici
active tectonic regions. in Proceedings of the 8th tenendo conto del loro sfasamento.

RIVISTA ITALIANA DI GEOTECNICA