Soil Dynamics and Earthquake Engineering 89 (2016) 75–84

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Soil-pile-structure kinematic and inertial interaction observed

in geotechnical centrifuge experiments
Mahmoud N. Hussien a,b,n, Tetsuo Tobita c, Susumu Iai c, Mourad Karray a
a
Department of Civil Engineering, Faculty of Engineering, Sherbrooke University, Sherbrooke, QC, Canada
b
Department of Civil Engineering, Faculty of Engineering, Assiut University, Assiut, Egypt
c
Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan

art ic l e i nf o a b s t r a c t

Article history: This paper reports the results of a series of centrifuge tests examining dynamic responses of single and
Received 29 August 2014 3
3 grouped piles embedded in sand and supporting SDOF and 2DOF structures. A total of 7 model tests
Received in revised form were conducted with the centrifugal acceleration of 40 g. Each model was subjected to 12 sinusoidal
11 February 2016
waves with constant acceleration amplitude and varying frequencies. The results of the tests indicate
Accepted 2 August 2016
that pile-head motion is dominated by two sequential frequencies: a lower frequency (fSSI) where pile-
head motion is substantially maximized and a higher one (fpSSI) where the response is minimized with
Keywords: respect to free field surface motion. These results confirm recent published numerical results on single
Centrifuge piles supporting SDOF structures and generalize their findings to grouped piles supporting SDOF and
Soil-pile-structure interaction
2DOF structures. The results show also strong mobilized kinematic interaction effect generating sig-
Pseudo-natural SSI frequency
nificant pile bending when the ground is excited at its resonant frequency. On the other hand, structural
Dynamic pile bending
Pile group vibrations tend to impose large bending moments as the excitation frequency approaches the natural
frequency of the coupled soil-pile-structure system. Distribution of pile bending moments in the group is
found to be a function of the pile position and the excitation frequency. In contrast to inner piles having
the greatest kinematic bending moments, outer piles have a more pronounced inertial ones.
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Single and grouped piles are widely used in the foundations of
many civil structures such as bridges and buildings. During
earthquakes, piles undergo stresses stemming from both the vi-
bration of the surrounding soil (kinematic interaction) and that of
the superstructure (inertial interaction). Evaluating these interac-
tions is an important issue in the seismic design of piled buildings,
highway-bridges, overcrossings and ramps, several of which ex-
perienced considerable damage in the 1994 Northridge earth-
quake, the 1995 Hyogo-Ken Nanbu earthquake, and the 2011 To-
hoku Pacific Earthquake [1,2].
Post-earthquake field investigations [3–5], well documented
centrifuge and shaking table tests [6–9], and numerical simula-
tions of piled structures [10–13] have clearly highlighted the im-
pact of soil-pile-structure interaction on the dynamic character-
istics of the structure and the seismic motion imposed at its base
[10]. Such effects may be present in some degree for every
n
Corresponding author.
E-mail addresses: Mahmoud.Nasser.Ahmed@USherbrooke.ca (M.N. Hussien),
tobita.tetsuo.8e@kyoto-u.ac.jp (T. Tobita), iai.susumu.6x@kyoto-u.ac.jp (S. Iai),
Mourad.Karray.Benhassen@USherbrooke.ca (M. Karray).
structure and is what has been recognized by the modern seismic
codes, such as Eurocode8 [14] that specify conditions where soil-
pile-structure interaction should be accounted for in the seismic
design of foundations and superstructures. These specifications
are set based on a variety of numerical and analytical methods that
utilize either simplified multi-step method that uncouple the
structure and foundation portions, commonly referred to in lit-
erature as kinematic-inertial decomposition method (substructure
approach) [15–17], or a direct analysis of the coupled soil-pile-
structure system in a single step [18–20]. Several comprehensive
reviews on the subject have been published, among others, by
Novak [21], Pender [22], and Gazetas and Mylonakis [23]. The
single-step approach gives a direct and often more convenient
estimation of the seismic response of soil-pile-structure systems
since inertial and kinematic effects are simultaneously modeled.
However, it is very complex and requires huge computational load
and resources, and consequently it is rarely performed in en-
gineering practice [10]. On the contrary, the multi-step approach
has been extensively implemented in professional engineering and
research practices in lieu of the direct approach. A key limitation of
the multi-step approach is that it refers to linear or moderately
nonlinear response of the coupled soil-pile-structure system. In
reality, the soil, the soil-pile interface and often the structure

http://dx.doi.org/10.1016/j.soildyn.2016.08.002
0267-7261/& 2016 Elsevier Ltd. All rights reserved.
76 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84
respond nonlinearly during an earthquake. Accordingly, either
high-quality laboratory tests or case history data are required to
validate the results from the substructure approach. The ad-
vantage of laboratory testing compared to field testing lies in its
ability to define the site conditions and structural systems that can
be in turn tested under controlled-intensity earthquake motions
[24]. In addition, it provides relatively rapid, precise and detailed
analyses of factors affecting the soil-pile-structure behavior under
earthquake events. A powerful laboratory device for soil-pile-
structure interaction testing is the geotechnical centrifuge since it
allows the researcher to monitor the simultaneous seismic re-
sponse of the soil, pile foundation, and structure [6,7].
At present, it seems that certain aspects of soil-pile-structure
interaction require further evidence based on well-documented
experimental data such as: (i) the identification of characteristic
frequencies dominating pile group response in coupled soil-pile
group-structure systems; (ii) the relative contributions of kine-
matic and inertial interaction to the dynamic bending of a pile as
part of a pile group; (iii) pile-soil-pile interaction under kinematic
and inertial loading as affected by the frequency content of input
motion. These three aspects of soil-pile-structure interaction are
discussed in this paper based on a comprehensive set of centrifuge
test data, which consist of end-bearing single and 3
3grouped
piles embedded in dry sand layer and supporting single (SDOF)
and two (2DOF) degree of freedom structures.
The study is at an initial phase where an overview of the be-
havior of group pile is obtained through model tests on dense
sand. The selected tested piles referred to typical hollow steel pipe
piles with length, Lp ¼ 10 m; diameter, Dp ¼ 0.4; Lp/Dp ¼ 25; and
pile spacing to diameter ratio, S/Dp ¼4. For the superstructures: we
used combination of SDOF and 2DOF structures with different
frequencies that are encountered in practice. In fact, pile founda-
tions are primarily used in soft soils with dominant frequencies
less than a few Hz and below resonant frequencies of structures on
rigid bases. In the current study, a dry dense sand deposit is used,
thereby aiming at the fundamental knowledge on the soil-pile
group-structure interactions under seismic loading as a basis for
further study including cases where soil is expected to liquefy. A
total of 7 model tests including single and grouped piles cases
were conducted with the centrifugal acceleration of 40 g. Each
model was subjected at least to 12 sinusoidal waves with constant
acceleration amplitude and varying frequencies. The results of the
study are compared and presented in the form of dimensionless
graphs, covering a wide range of excitation frequencies. The pri-
mary findings from this study are summarized as conclusions.
2. Centrifuge modeling
2.1. Centrifuge modeling principles
Geotechnical centrifuge modeling has been used over the past
three decades to simulate dynamic and seismic events. Model si-
militude is an important issue in centrifuge experimentation using
model scale models that are intended to capture the response of
prototype scale [25–27]. During a centrifuge test, according to the
scaling rules for n g centrifugal field, the gravity, frequency, and
acceleration are to be increased by n times while the length and
time are reduced by n. The stress and strain measured at model
scale is the same as the stress and strain experienced by the
prototype. Soil has stress-dependent strength, deformation, and
volume change properties; therefore, the centrifuge is a useful tool
for examining geotechnical earthquake engineering problems.
Additional information on geotechnical centrifuge scaling laws can
be found in Schofield [25,26]. All results presented in this study
have been converted to prototype scale unless otherwise noted.
2.2. Centrifuge facility
The centrifuge tests were performed at the Disaster Prevention
Research Institute, Kyoto University (DPRI-KU), Japan. The cen-
trifuge facility has an in-flight platform radius of 2.5 m and a ca-
pacity of 24 g-tons. It is equipped with a one dimensional shake
table (allowable displacement: 7 5 mm), which is operable under
the centrifugal accelerations of up to 50 g. It has a single servo
hydraulic actuator parallel to the rotation of the centrifuge arm,
and it is controlled through a laptop personal computer (PC) on
the centrifuge arm. The PC is fixed near the rotation axis of the
centrifuge to minimize the centrifugal force acting on it. It is
connected to a PC in the control room by a wireless LAN, and the
data loggers attached on the arm are accessible from a PC in the
control room through a wireless USB connection. A counter-
weight is loaded on other side of the arm to maintain balance
during rotation [28]. In this study, all the tests were performed at a
centrifugal acceleration of 40 g (n ¼ 40).
2.3. Experimental-setup
Ground model was constructed with Silica sands No. 7. The
physical properties of the sand are listed in Table 1, and the grain-
size distribution curve is shown in Fig. 1. The soil is classified into
“poorly graded sand (SP)”. The model ground was prepared in a
strong container, which has nominal inner dimensions of
0.45
0.15
0.30 m3 with a glass wall on one side. The prototype
scale depth of the model ground was 10 m. A total of 7 model tests,
listed in Table 2, were performed. Schematics of experimental
model tests No. 4 and 7 listed in Table 2 are provided in Fig. 2
(a) and 2(b), respectively. Fig. 2 includes important dimensions
and the locations of selected instrumentation. Fig. 3 shows a plane
view of the grouped pile arrangement and locations of in-
strumented piles. The pile group was lined up 3 by 3 with a spa-
cing of 4 pile diameters from center to center in both directions.
The tested pile had a 0.4 m outside diameter and wall thickness of
30 mm. The pile tips were set in motion fixed at the bottom of the
container using a 10 mm thickness steel plate. The soil and
structures were instrumented with accelerometers while piles
were instrumented with accelerometers and strain gauges. These
instruments recorded the seismic response of the soil, piles, and
structures during test program. Detailed descriptions of the in-
strumentations can be found in [29].
Model piles and columns shown in Fig. 2(a) and 2(b) were
made of steel tubes having the material properties listed in Ta-
ble 3. Material properties and dimensions of square steel plates
(Mass 1–4 shown in Fig. 2 and Table 2), used to model the struc-
tural masses, are listed in Table 4.
2.4. Model construction
The centrifuge model is normally prepared outside the cen-
trifuge pit, and then transferred to the swinging platform. Each
model was constructed very carefully so that all ground models
were as identical to each other as possible. Piles were fixed in the
model before the dry sand was placed, attempting to simulate a
pile installed with minimal disturbance to the surrounding soil, as
may be the case when a pile inserted into a pre-augered hole. The
piles in the group are kept vertical in their positions using a guided
plate as shown in Fig. 4. Dry silica sand was placed in the soil
container using the air pluviation method [29]. Gaffer tape was
used to supplementary fix the sensors in position as well as to fix
the cables of the sensors to the back wall of the box to prevent
unwanted influences on the sand deposit and movement of sen-
sors. Single (Tests No. 3 and 6) or two (Tests No. 4 and 7) DOF
structures were placed and fixed at pile-heads. The completed
 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84 77

Table 1 structures were then subjected to the same sinusoidal waves as

Physical properties of Silica sand No. 7. the coupled soil-pile-structure models. The experimentally de-
termined frequencies of the SDOF and 2DOF structures on fixed
Specific gravity (Gs) 2.66
D50 0.130 mm bases are listed in Table 2.
Coefficient of uniformity (Cu) 1.875
Maximum void ratio (emax) 1.19 2.6. Processing of records
Maximum void ratio (emin) 0.71

Displacement time histories were computed from the post-

100 processing acceleration records using the software SeismoSignal
[30]. The pre-processing acceleration time history data was base-
Percent finer by weight (%)

Silica sand No. 7

line corrected and filtered using the same software to remove the
80
undesirable frequency (noise) content and minimize the long
period drift after integration to velocity and displacement time
60
histories. The acceleration time history data was filtered using a
fourth-order bandpass filter allowing frequencies between 0.1 Hz
40 and 25 Hz to pass through.

20
0
0.01 0.1
Diameter (mm)
Fig. 1. Particle size distribution curve for Silica sand No. 7.
model was weighed and fixed on the swinging platform of the
centrifuge arm and the counterweight on the opposite arm was
adjusted.
2.5. Test procedures
After confirming that all equipment and sensors are well
functioning without any abnormality, centrifugal acceleration was
increased gradually up to 40 g. To properly consolidate the model
ground before shaking, the model is put under 40 g for 5 min. By
measuring the heights of the ground surface after the consolida-
tion, the actual relative density before shaking was obtained. The
centrifugal acceleration is again increased up to 40 g to apply the
dynamic motion to the model. Each model was subjected at least
to 12 sinusoidal waves as input base motions having constant
acceleration amplitude of about 1.5 m/s2 and frequencies ranging
from 0.5 to 12 Hz. As the input motion was imposed, the soil-
model was progressively densified. However, calculations based on
the total cumulative surface settlement shown in Table 2 show
that the maximum change in relative density between the start of
the testing (Dr ¼ 78%) (i.e., when the model was in a pristine
condition) and the end of the testing (Dr ¼ 82%) was on the order
of 4%. In summary, the soil model was very dense at the beginning
and end of testing, so densification of soil is expected to have little
influence on the results presented herein.
The fixed-base frequencies of the used SDOF and 2DOF struc-
tures are determined experimentally by connecting the structures
rigidly to the shaking table floor of the centrifuge. The fixed-base
3. Experimental results and discussions
1
3.1. Free field response
Test No. 1 was used as a benchmark test and performed to
monitor the free field motion of the dry sand deposit to act as a
baseline in subsequent discussions of both kinematic and inertial
interaction effects on pile behavior. In this study, recorded accel-
eration time histories of the free field were obtained at different
frequencies of excitations (1–12 Hz). From these time histories, the
amplitude of steady-state acceleration was noted and normalized
with respect to the amplitude of the base acceleration (E1.5 m/
s2). Thus amplification of free field response is derived at different
frequencies and plotted in Fig. 5. Fig. 5 shows that the funda-
mental period of the ground Tg equals to 0.143 s. In other words,
the predominant frequency of the ground fg E 7.0 Hz.
3.2. Kinematic soil-pile interaction
Even in the absence of a superstructure, piles tend to diffract
the upward propagating S-waves, thereby modifying soil de-
formations, so that the horizontal displacement of the pile-head,
Up may be different from the free field surface motion, Uff. This
type of interaction between piles and soils is called kinematic in-
teraction, which is usually portrayed in terms of the kinematic
amplification factor Au = Up/Ub [16] and the kinematic interaction
factor Iu = Up/Uff [1,16,31], where Ub is the base displacement. Tests
No. 2 and 5 were conducted to investigate kinematic soil-pile in-
teraction for single and grouped piles cases, respectively. Time
histories of pile-head displacements were obtained at different
frequencies of excitations (1–12 Hz). From these time histories, the
amplitude of steady-state pile-head displacement is noted and
normalized with respect to input displacement ( Au = Up/Ub ) and
with respect to free field surface motion ( Iu = Up/Uff ). Kinematic

Table 2
Summary of centrifuge model tests.

Test No. Test setup Soil deposit Structural mass fstr.fixed (Hz) Ground relative density (%)

f1 f2 Before the test After the test

1 Free field Dry sand 77 81

2 Free-head single pile 78 80
3 Single pile þ SDOF structure Mass 1 5 80 82
4 Single pile þ 2DOF structure Mass 1 Mass 2 2 11.5 80 82
5 Free-head pile group 78 82
6 Pile group þ SDOF structure Mass 3 8.5 82 83
7 Pile group þ 2DOF structure Mass 3 Mass 4 5 10 82 84
78 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84

Accelerometers
Mass 2 Mass 4
Columns
Column
Piles 2.5 m
2.5 m
Mass 1 Mass 3
Strain gauges
1.8 m 1.8 m 2.0 m

1.8 m 1.8 m
12 m 1.8 m 1.8 m Sand
2.0 m 9.6 m 10 m 2.0 m
9.6 m
2.2 m 2.2 m
2.2 m 2.2 m

18 m 18 m

Fig. 2. Centrifuge models with important instrumentation demarcated for: (a) Test No. 4; and (b) Test No. 7.

4Dp 4Dp follow the movements of the free field and may thereby experi-
ence considerably reduced deformations especially in a frequency
Pile 4 band approaching the natural frequency of the ground.
In order to discern kinematic soil-pile interaction in the pile
Pile 3 Pile 2 Pile 1 4Dp group, amplitudes of pile-heads displacements in the free-head
pile group (Test No. 5) are normalized with respect to base dis-
4Dp
Pile 5 placement and plotted as a function of base excitation frequencies
in Fig. 6(c). The corresponding free field curve is also plotted for
comparison. Due to symmetry, only the results of piles No. 1, 2, and
Instrumented pile 4 are presented. The small plot at the bottom right of Fig. 6
(c) represents the nine piles of the group, where the shadowed
Fig. 3. Plan view of the pile group arrangement.
ones indicate the piles being analyzed. Fig. 6(c) shows that the
displacements of the piles in the free-head pile group are not the
Table 3
same, but a function of both pile position and frequency content of
Material properties of piles and columns.
base excitation. The spatial dependence of piles displacements is
Model Prototype more pronounced in a frequency band approaching the natural
frequency of the ground. Fig. 6(c) illustrates that for relatively low
Outer diameter (mm) 10 400
excitation frequency; displacements of piles in the group are al-
Wall thickness (mm) 0.75 30
Young's modulus (GPa) 206 206 most the same and approach that of the free field. At such lower
2nd moment of inertia (mm4) 2.35
102 6.00
108 frequencies of excitation and for a relatively close pile group em-
Bending stiffness (MN-mm2) 48.41 1.24
108 bedded in a dense sand soil layer, the pile group and the sand
enclosed between the piles may serve as a block or a rigid pier that
vibrates as an integral unite. With the increase of the base ex-
amplification factor (Au) of the single pile is compared to the free
citation frequencies, the piles and the enclosed sand no longer
field amplification ( Uff /Ub ) in Fig. 6(a) while the kinematic inter-
serve as a rigid pier and piles in the group start to move in-
action factor is portrayed in Fig. 6(b). Figs. 6(a) and 6(b) show that
dependently with a decrease in pile displacement from the center
pile-head motion is in general different from that of the free field of the group to its edge.
because of the scattered wave field generated from the difference
between pile and soil rigidities in agreement with real earthquake 3.3. Combined kinematic and inertial action
observations of pile foundation responses reported by Kawamura
et al. [32], Tajimi [33], and Ohta et al. [34]. Fig. 6(a) and 6(b) show Tests (3, 4, 6, and 7) were performed to investigate soil-pile-
that up to a relatively low frequency of about 4 Hz, the pile ap- structure interaction by fixing SDOF or 2DOF structures on pile-
pears to essentially follow the ground motion. On the other hand, heads. In particular Tests No. 3 and 4 correspond to single piles
at higher frequencies (i.e., 44 Hz), the pile may not be able to supporting SDOF and 2DOF structures, respectively while Tests No.

Table 4
Material properties and dimensions of structural masses.

Mass 1 Mass 2 Mass 3 Mass 4

Model Prototype Model Prototype Model Prototype Model Prototype

Side length (mm) 40 1600 50 2000 120 4800 105 4200

Thickness (mm) 30 1200 15 600 30 1200 13.5 540
Mass (kg) 0.3792 24,269 0.297 19,008 3.4128 218,419 1.176 75,252
Young's modulus (GPa) 206 206 206 206 206 206 206 206
2nd moment of inertia (mm4) 9.0
104 2.30
1011 1.4
104 3.61
1010 2.7
105 6.91
1011 2.1
104 5.50
1010
Bending stiffness (MN-mm2) 1.85
104 4.75
1010 2.9
103 7.42
109 5.5
104 1.42
1011 4.4
104 2.65
109
 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84
Fig. 4. Model setup of free-head group pile test (Test No. 5).
Fig. 5. The base-to-surface free field amplification ratio.
Fig. 6. Kinematic interaction effects: (a) single pile amplification factor, (b) single
pile interaction factor, and (c) group pile amplification factor.
79
6 and 7 correspond to group piles supporting SDOF and 2DOF
structures, respectively. The frequency at which the ratio of
structural mass to free field horizontal displacement (UMass/Uff) is
maximized was computed for each examined case, to determine
the effective natural frequency of the coupled soil-pile-structure
system, fSSI according to the definition by [10]. fSSI of the system
was then compared to the natural frequency of the structure under
fixed-base conditions, thus quantifying soil-structure interaction
(SSI) in terms of fundamental dynamics considerations. Amplifi-
cation ratios correspond to single piles supporting SDOF (Test No.
3) and 2DOF (Test No. 4) structures are shown in Fig. 7(a) and 7(b),
respectively. Fig. 7(a) shows strong SSI resulting in a significant
reduction of the effective natural frequency (fSSI ¼1 Hz) with re-
spect to the natural frequency of the SDOF structure under fixed-
base conditions (fstr.fixed ¼5 Hz). Fig. 7(b) illustrates the strong SSI
effect on the reduction of resonant frequencies (fSSI1 ¼ 0.5 Hz and
fSSI2 ¼5 Hz) of the coupled soil-pile-2DOF structure system with
respect to the fundamental (fst.fixed1 ¼2 Hz) and the second mode
(fst.fixed2 ¼ 11.5 Hz) frequencies of the 2DOF structure on rigid
ground. The ratios of the pile-head to free field horizontal dis-
placements (Up/Uff) are also examined in Fig. 7(a) and 7(b). These
ratio incorporates the combined effect of inertial and kinematic
interaction as opposed to the kinematic interaction factor (Iu ¼ Up/
Uff) presented in Fig. 6(b) which expresses solely kinematic effects.
Fig. 7(a) shows that pile-head displacement, Up is dominated by
two sequential frequencies: a lower frequency (the effective nat-
ural frequency (fSSI)) where the pile-head motion is amplified and
a higher one (the pseudo-natural frequency (fpSSI) of the system)
where the response is suddenly de-amplified with respect to the
free field motion, Uff. It is worth noting that this variation pattern
has been reported by Rovithis et al. [10] and observed in actual
earthquake observations of the response of a pile foundation [34],
and analytical studies of soil-pile-structure systems using sub-
structure approach [20]. Based on the comparative results in Fig. 7
(a), it is obvious that both fSSI and fpSSI have a double function. fSSI
defines the frequency where free field motion (Uff) is minimized
with respect to both structure (UMass1) and pile-head (Up) motions.
On the other hand, fpSSI defines the frequency where pile-head
motion (Up) is minimized with respect to both structure (UMass1)
and free field (Uff) motions.
Fig. 7(b) corresponding to a single pile supporting a 2DOF
structure shows that the variation pattern of pile-head motion
relative to that of the free field is similar to that shown in Fig. 7(a).
Evidently, pile-head motion is amplified with respect to the free
field surface motion at the fundamental frequency of the coupled
system, fSSI1 ¼ 0.5 Hz and de-amplification at the first pseudo-
natural frequency fpSSI1 ¼ 1.0 Hz. Pile-head motion is amplified
again at the second mode fundamental frequency of the coupled
80 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84

Fig. 7. Amplification ratios of single piles supporting: (a) a SDOF and (b) a 2DOF structure.

Fig. 8. Amplification ratios for piles in a group supporting: (a) SDOF and (b) 2DOF structures.
 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84 81

system fSSI2 ¼ 5.0 Hz and then de-amplification at the second
mode pseudo-natural frequency fpSSI2 ¼ 6.0 Hz. It can be con-
cluded from Fig. 7(b) that effective and pseudo-effective natural
frequencies are found to exist not only in soil-pile-SDOF structure
systems but also in soil-pile-2DOF structure systems. In all cases,
the pseudo-effective natural frequency is generally higher than the
corresponding effective natural frequency.
Amplification ratios of piles No. 1, 2, and 4 in the group sup-
porting SDOF (Test No. 6) and 2DOF (Test No. 7) structures are
shown in Fig. 8(a) and 8(b), respectively. Similar behavior is ob-
served for all piles. Similar to single pile cases shown in Fig. 7(a),
Fig. 8(a) shows strong SSI resulting in a significant reduction of the
effective natural frequency (fSSI ¼ 3 Hz) with respect to the natural
frequency of the structure under fixed-base conditions (fstr.fixed ¼
8.5 Hz, from Table 2). Fig. 8(b) illustrates the strong SSI effect on
the reduction of resonant frequency (2.0 Hz) of the soil-pile-2DOF
structure system with respect to fundamental frequency (5 Hz,
from Table 2) of 2DOF structure on rigid ground. The peak

Fig. 9. Distributions of normalized steady state bending moments of piles in a group: (a) kinematic, (b) supporting a SDOF structure, and (c) supporting a 2DOF structure.
82 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84
corresponding to the second mode of vibration of the soil-pile-
2DOF structure system is not pronounced as they may correspond
to frequencies exist in-between to integers. Fig. 8 indicates also
that pile-heads displacements are dominated by two discrete
frequencies: a lower one (fSSI) where the response amplified and a
higher one (fpSSI) where the response de-amplified with respect to
free field motion. Fig. 8 shows also that there is no difference in
the values of these frequencies among the piles in the group.
These results confirm recent published numerical results by Ro-
vithis et al. [10] on single piles supporting SDOF structures and
generalize their findings to grouped piles supporting SDOF and
2DOF structures. It is worth noting that the significant de-ampli-
fication of the pile-head motion relative to that of the free field
may be of importance in the design of pile foundation supporting
structures especially when the predominant frequency of the de-
sign excitation is close to fpSSI of the soil-pile-structure system. In
other words, estimating fpSSI may provide a possible design cri-
teria/frequency where pile-head motion is minimized with respect
to the free field ground response.
3.4. Dynamic pile bending
The seismic response of the coupled soil-pile-structure system
is furthered examined in terms of dynamic pile bending as a
function of excitation frequency. The pile bending moments were
normalized to the amplitude of input acceleration according to
Kavvadas and Gazetas [35]:
M trends at fSSI (2.0 Hz) and fpSSI (3.0 Hz) of the pile group supporting a
Mnor =
ρp Dp4 U¨g (1)
where M is the absolute value of pile bending moment, ρp is the
mass density of the pile material, Dp is the pile diameter, and Üg is
the amplitude of the harmonic input motion introduced at the
base of the soil profile. To investigate the role of the excitation
frequency as well as the relative contributions of kinematic and
inertial interactions on dynamic pile bending, pile bending mo-
ments profiles calculated at selected excitation frequencies in-
cluding the ground resonance as well as fSSI and fpSSI of coupled
soil-pile-structure systems are plotted and compared to the cor-
responding kinematic moments for free-head piles as shown in
Fig. 9. Fig. 9 presents normalized bending moments of pile No. 1, 2,
and 4 obtained at 1.5 m/s2 input acceleration and at excitation
frequencies of 2, 3, 4, and 7.0 Hz.
For a free-head pile group (kinematic case) shown in Fig. 9
(a) and at relatively low frequencies of the base excitation (2 and
3 Hz), the measured bending moments in all piles are almost the
same. At such lower frequencies of excitation and for a relatively
close pile group embedded in dense sand, the pile group and the
sand enclosed between the piles may serve as a block or a rigid
pier therefore the effect of pile-soil-pile interaction is not sig-
nificant. When the frequency content of base excitation increases
and approaches the fundamental frequency of the ground (7 Hz),
bending moments of all piles increase as expected and the dif-
ference between piles bending moments also increases. It is worth
noting that Waas and Hartmann [36], and Finn et al. [37] reported
strong kinematic interaction effect for a large group of piles with
spatial dependence of pile bending. The fact that the kinematic
bending moments are not the same for different piles in the group
implies that the seismic induced forces in the piles do not attain
their maximum values at the same time. The most interesting
result in Fig. 9(a) is that the center pile (pile No. 2) receives the
maximum kinematic bending value followed by the side pile (pile
No. 1) followed by the corner pile (pile No. 4) in agreement with
those reported by Kaynia and Kausel [38] based on boundary in-
tegral analyses of different pile group configurations. The spatial
variability of kinematic pile bending in the group may be attrib-
uted to different pile-soil-pile interaction phenomena similar to
those reported under monotonic loading such as shadowing and
edge effects, being this a phenomenon that cannot be explained by
the authors at the moment and which should be further studied.
In passing, Fig. 9(a) shows also that the depth of maximum mo-
ments is not affected by the pile position or the frequency content
of the base excitation.
Fig. 9(b) and 9(c) depict the dynamic bending moments of piles
No. 1, 2, and 4 in pile groups supporting SDOF and 2DOF structures,
respectively. Unlike the kinematic case (shown in Fig. 9(a)); a
general trend displayed by these figures (Fig. 9(b) and 9(c)) is an
increase in the dynamic bending moment from the center of the
group to its edge. In other words, the piles farther from the center
are subjected to larger bending moments as they receive the
maximum inertial forces (i.e., swaying and maximum rocking
components) transmitted from the superstructures. This observa-
tion is in agreement with the results reported earlier by Wolf et al.
[39], Waas and Hartmann [36], Kaynia and Mahzooni [18], and
Zarzalejos et al. [40]. As shown in Fig. 9(b), when the frequency
content is close to the effective natural frequency of the coupled
soil-pile-structure system (3 Hz), structural vibrations tends to im-
pose significant increases in the dynamic bending moments of all
piles in the group with great difference in piles bending moments
indicating strong effect of pile-soil-pile interaction. On the other
hand, a sudden decrease in pile bending moments is observed as
the frequency content of the base excitation approaches the pseu-
do-natural frequency of the system (4.0 Hz). Fig. 9(c) shows similar
2DOF structure. Similar to the kinematic case, the depth of max-
imum inertial moments does not appear to be affected by the pile
position or the frequency content of the base excitation.
To clearly investigate the relative contributions of kinematic
and inertial interactions on dynamic bending of piles as affected
by the frequency content of the input motion, bending moments
profiles calculated at selected frequencies content of input motions
(2, 3, 4, and 7 Hz) for piles No. 1, 2, and 4 in groups supporting
SDOF and 2DOF structures are plotted and compared to the cor-
responding kinematic moments in Fig. 10. The small plot at the
bottom right of Fig. 10(a–c) represents the nine piles of the group,
where the shadowed ones indicate the piles being analyzed. For all
piles shown in Fig. 10, when the frequency content of the base
excitation is close to 2 Hz, the maximum bending moment
corresponds to the case of a group supporting 2DOF structure. In
this case, SSI has a major effect on the dynamic characteristics of
the structure, resulting in an effective natural frequency fSSI of 2 Hz
significantly lower than the fundamental frequency of the 2DOF
structure on rigid ground (5 Hz as shown in Table 2). When the
frequency content of the base excitation is closed to 3 Hz, the
maximum pile bending moment corresponds to the case of a
group supporting SDOF structure. In this case, SSI has also a major
effect on the dynamic characteristics of the structure, resulting in
an effective natural frequency fSSI of 3 Hz significantly lower than
the fixed base frequency of the SDOF structure (8.5 Hz as shown in
Table 2). These two cases reveal the dominant role of inertial in-
teraction in the development of bending moments of piles when
the superstructure is close to resonance. In other word, the re-
sonance effect that is evident when the excitation frequency is
close to fSSI of the coupled system mobilizes strong inertial inter-
action that imposes larger bending moment onto the piles.
At frequency content of the base excitation closed to 3 Hz, the
increase of bending moments of piles in the pile group supporting
SDOF structure is associated with a significant decrease in piles
bending of the group supporting 2DOF structure as a result of fre-
quency shifting from the effective natural frequency of the soil-pile
group-2DOF structure system (2.0 Hz) to its pseudo-natural
 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84 83

Fig. 10. Distributions of normalized steady state bending moments of piles in the group: (a) pile No. 1, (b) pile No. 2, and (c) pile No. 4.

frequency (3.0 Hz). Focusing on the case of the pile group sup-
porting a SDOF structure, there is a substantial reduction in pile
bending under the combined action of kinematic and inertial in-
teraction when the frequency content of the input motion is closed
to pseudo-natural frequency of the system (4.0 Hz). It is worth to
note that, despite this significant reduction, the maximum pile
bending at 4.0 Hz still corresponds to the case of the group pile
supporting SDOF structure. As the excitation frequency approaches
the fundamental frequency of the ground (7.0 Hz), strong kinematic
effects are mobilized generating significant increases in bending
moment of all piles. These latter results suggest that one cannot
ignore, without justification, the kinematic interaction in the design
of piles under seismic loading especially when the predominant
frequency of the design excitation is close to the ground resonance.
84 M.N. Hussien et al. / Soil Dynamics and Earthquake Engineering 89 (2016) 75–84
4. Conclusions
A series of systematic centrifuge tests is conducted in order to
study the seismic response of end-bearing single and 3
3grouped
piles embedded in dry sand layer and supporting single and two
degree of freedom structures. All tests are conducted with the
centrifugal acceleration of 40 g. Air pluviation method was used to
place the dry sand in the soil container. A total of 7 tests including
free field, single and grouped piles cases were performed. Each
model was subjected to at least 12 sinusoidal waves as input base
accelerations. These waves have constant amplitude of 1.5 m/s2
and varying frequencies ranging from 0.5 to 12 Hz. Of the findings
of this study, the following conclusions can be drawn:
● For both single and grouped piles supporting single or two de-
gree of freedom structures: the pile-head motion is found to be
dominated by two discrete frequencies: a lower frequency (the
effective natural frequency (fSSI)) where the pile-head motion is
amplified and a higher one (the pseudo-natural frequency (fpSSI)
of the system) where the response is suddenly de-amplified
with respect to the free field motion. These results confirm re-
cent publish numerical results found in the literature and gen-
eralize the finding to grouped piles supporting higher degree of
freedom structures. Estimating fpSSI may provide a possible de-
sign criteria/frequency where pile-head motion is minimized
with respect to the free field ground response.
● For free-head grouped piles: as the frequency content of base
excitation approaches the fundamental frequency of the ground,
strong kinematic effects are mobilized generating significant
increases in bending moment of all piles with great differences
between piles bending moments indicating strong effect of pile-
soil-pile interaction. In this case, the center pile in the group has
the maximum value of bending moment followed by the side
pile followed by the corner pile. These results suggest that the
kinematic interaction should not be neglected in the seismic
design of piles especially when the predominant frequency of
the design excitation is close to the ground resonance.
● For group piles supporting a structure: as the frequency content of
base excitation approaches the effective natural frequency of the
coupled soil-pile-structure system, structural vibrations tend to
impose large bending moments at all piles with large difference
between piles bending moments indicating strong effect of pile-
soil-pile interaction. Unlike the free-head group piles case, the
center pile in the group has the minimum value of inertial bending
moment followed by the side pile followed by the corner pile.
It is fair to mention here that more experimental work is
needed to better interpret soil-pile-structure interaction in more
complex conditions including the cases where liquefaction occurs.
In these situations, the relative contributions of kinematic and
inertial interaction to the seismic forces induced in pile founda-
tions may be different to those observed in the current centrifuge
experiments.
References
[1] Dezi F, Carbonari S, Leoni G. Static equivalent method for the kinematic in-
teraction analysis of single piles. Soil Dyn Earthq Eng 2010;30(8):679–90.
[2] Makris N, Tazoh T, Yun X, Fill AC. Prediction of the measured response of a
scaled soil–pile–superstructure system. Soil Dyn Earthq Eng 1997;16(2):113–
24.
[3] Mizuno H. Pile damage during earthquakes in Japan. In: Nogami T, editor.
Dynamic response of pile foundations. New York: ASCE Special Publication;
1987. p. 53–78.
[4] Ross GA, Seed BH, Migliaccio R. Bridge foundations in the Alaska earthquakes. J
Soil Mech Found Eng ASCE 1989:95–101.
[5] Tazoh T, Dewa K, Shimizu K, Shimada M. Observations of earthquake response
behavior of foundation piles for road bridge. In: Proceedings of the 8th World
Conference on Earthquake Engineering, San Francisco; 1984, p. 577–84.
[6] Boulanger RW, Curras CJ, Kutter BL, Wilson DW, Abghari A. Seismic soil-pile-
structure interaction experiments and analyses. J Geotech Geoenviron Eng
1999;125(9):750–9.
[7] Curras CJ, Boulanger RW, Kutter BL, Wilson DW. Dynamic experiments and
analyses of a pile-group supported structure. J Geotech Geoenviron Eng
2001;127(7):585–96.
[8] Tokimatsu K, Suzuki H, Sato M. Effects of inertial and kinematic interaction on
seismic behavior of pile with embedded foundation. Soil Dyn Earthq Eng
2005;25(7–10):753–62.
[9] Chau KT, Shen CY, Guo X. Nonlinear seismic soil–pile–structure interactions:
Shaking table tests and FEM analyses. Soil Dyn Earthq Eng 2009;29(2):300–10.
[10] Rovithis EN, Pitilakis KD, Mylonakis GE. Seismic analysis of coupled soil-pile-
structure systems leading to the definition of a pseudo-natural SSI frequency.
Soil Dyn Earthq Eng 2009;29(6):1005–15.
[11] Lu XL, Li PZ, Chen B. Computer simulation of the dynamic layered soil-pile-
structure interaction system. Can Geotech J 2005;42(3):742–51.
[12] Kagawa T, Kraft L. Seismic P-Y responses of flexible piles. Journal of Geo-
technical and Geoenvironmental Engineering, ASCE 1980; 106(8): p. 899–18.
[13] Hussien MN, Karray M, Tobita T, Iai S. Kinematic and inertial forces in pile
foundations under seismic loading. Comput Geotech 2015;69:166–81.
[14] EN 1998-1. Eurocode 8 – Design of structure for earthquake resistance – Part 1:
General rules, seismic actions and rules for buildings; 2004..
[15] Roesset JM, Whitman RV, Dobry R. Modal analysis for structures with foun-
dation interaction. J Struct Div 1973;99(3):399–416.
[16] Gazetas G. Seismic response of end bearing single piles. Soil Dyn Earthq Eng
1984;3(2):82–93.
[17] Makris N, Gazetas G, Delis E. Dynamic soil-pile-foundation-structure interac-
tion: records and predictions. Géotechnique 1996;46(1):33–50.
[18] Kaynia AM, Mahzooni S. Forces in pile foundations under seismic loading. J
Eng Mech 1996;122(1):46–53.
[19] Guin J, Banerjee PK. Coupled soil-pile-structure interaction analysis under
seismic excitation. J Struct Eng 1998;124(4):434–44.
[20] Mylonakis G, Nikolaou A, Gazetas G. Soil-pile-bridge seismic interaction: ki-
nematic and inertial effects. Part I: soft Soil. Earthq Eng Struct Dyn
1997;26:337–59.
[21] Novak M. Piles under dynamic loads. In: State of the art paper. Second inter-
national conference on recent advances in geotechnical earthquake en-
gineering and soil dynamics, vol. III. Missouri: University of Missouri-Rolla;
1991: p. 250–73.
[22] Pender MJ. A seismic pile foundation design analysis. Bull N Z Natl Soc Earthq
Eng 1993;26(1):49–160.
[23] Gazetas G, Mylonakis G. Seismic soil-structure interaction: new evidence and
emerging issues. Geotech Earthq Eng Soil Dyn III 1998;2:1119–74.
[24] Mason HB, Trombetta NW, Chen Z, Bray JD, Hutchinson TC, Kutter BL. Seismic
soil-foundation-structure interaction observed in geotechnical centrifuge ex-
periments. Soil Dyn Earthq Eng 2013;48(5):162–74.
[25] Schofield AN. Cambridge geotechnical centrifuge operations. Géotechnique
1980;30:227–68.
[26] Schofield AN. Dynamic and earthquake geotechnical centrifuge modelling. In:
Proceedings of the international conference on recent advances in soil dy-
namics and earthquake engineering, St. Louis, Missouri; 1981.
[27] Kutter BL. Recent advances in centrifuge modeling of seismic shaking. In:
Proceedings of the third international conference on recent advances in
geotechnical earthquake engineering and soil dynamics, St. Louis, MO; 1995.
[28] Tobita T, Kang G-C, Iai S. Centrifuge modeling of manhole uplift in liquefied
trench. Soils Found 2011;51(6):1091–102.
[29] Hussien MN. Effects of nonlinear soil-structure interaction on lateral behavior
of pile foundations [PhD Thesis].Kyoto, Japan: Kyoto University; 2011.
[30] SeismoSignal v.4.3.0. Seismosoft Ltd.; 2011. 〈http://www.seismosoft.com〉.
[31] Flores-Berrones R, Whitman RV. Seismic responses of end-bearing piles. J
Geotech Eng 1982;108(4):554–69 [ASCE].
[32] Kawamura S, Umemura H, Osawa Y. Earthquake motion measurement of a
pile-supported building on reclaimed ground. In: Proceedings of the 6th
World conference on earthquake engineering, India; 1977: p.103–08.
[33] Tajimi H. Seismic effects on piles. State-of-the-art-report special session,
Proceedings of the 9th international conference on soil mechanics, Tokyo;
1977: p. 1–12.
[34] Ohta T, Uchiyama S, Niwa M, Ueno K. Earthquake response characteristics of
structure with pile foundation on soft subsoil layer and its simulation analysis.
In: Proceedings of the 7th World conference on earthquake engineering. Is-
tanbul, Turkey;1980: vol. 3.
[35] Kavvadas M, Gazetas G. Kinematic seismic response and bending of free-head
piles in layered soil. Géotechnique 1993;43(2):207–22.
[36] Waas G, Hartmann HG. Seismic analysis of pile foundations including pile-
soil-pile interaction. In: Proceedings of the 8th World conference on earth-
quake engineering. San Francisco, USA; 1984: vol. 5.
[37] Finn WDL, Dowling J, Taiebat M, Wu G. Distribution of seismic loads in large
pile groups. In: Oka F, Murakami A, Uzuoka R, Kimoto S, editors. Computer
methods and recent advances in geomechanics. London: Taylor and Francis
Group; 2015. p. 803–7.
[38] Kaynia AM, Kausel E. Dynamic stiffness and seismic response of pile groups.
Technical Report R83-03. Cambridge, MA: MIT; 1982.
[39] Wolf JP, von Arx GA, de Barros FCP, Kakubo M. Seismic analysis of the pile
foundation of the reactor building of the NPP Angra 2. Nucl Eng Des 1981;65
(3):329–41.
[40] Zarzalejos JM, Aznárez JJ, Padrón LA, Maeso O. Influences of type of wave and
angle of incidence on seismic bending moments in pile foundations. Earthq
Eng Struct Dyn 2014;43(1):41–59.