Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

Contents lists available at SciVerse ScienceDirect

Soil Dynamics and Earthquake Engineering

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

Kinematic response of single piles for different boundary conditions:

Analytical solutions and normalization schemes
George Anoyatis a, Raffaele Di Laora b, Alessandro Mandolini b, George Mylonakis a,n
a
Department of Civil Engineering, University of Patras, Rio GR-26500, Patras, Greece
b
Department of Civil Engineering, Second University of Naples, Aversa (CE), Italy

a r t i c l e i n f o abstract

Article history: Kinematic pile–soil interaction is investigated analytically through a Beam-on-Dynamic-Winkler-

Received 10 July 2012 Foundation model. A cylindrical vertical pile in a homogeneous stratum, excited by vertically-
Received in revised form propagating harmonic shear waves, is examined in the realm of linear viscoelastic material behaviour.
10 September 2012
New closed-form solutions for bending, displacements and rotations atop the pile, are derived for
Accepted 20 September 2012
different boundary conditions at the head (free, fixed) and tip (free, hinged, fixed). Contrary to classical
Available online 3 November 2012
elastodynamic theory where pile response is governed by six dimensionless ratios, in the realm of the
proposed Winkler analysis three dimensionless parameters suffice for describing pile–soil interaction:
(1) a mechanical slenderness accounting for geometry and pile–soil stiffness contrast, (2) a dimension-
less frequency (which is different from the classical elastodynamic parameter a0 ¼ o d/Vs), and (3) soil
material damping. With reference to kinematic pile bending, insight into the physics of the problem is
gained through a rigorous superposition scheme involving an infinitely-long pile excited kinematically,
and a pile of finite length excited by a concentrated force and a moment at the tip. It is shown that for
long piles kinematic response is governed by a single dimensionless frequency parameter, leading to a
unique master curve pertaining to all pile lengths and pile–soil stiffness ratios.
& 2012 Elsevier Ltd. All rights reserved.

1. Introduction the accumulated evidence and enforced evaluation of kinematic

effects in design of deep foundations, even though only in the
It is well known that the passage of seismic waves through soft presence of a layered soil profile. Note in this regard that a wealth of
soil induces deformations in the soil mass that excite dynamically research results have demonstrated that significant kinematic
embedded bodies such as piles. As a result, a pile foundation will, bending can develop at the pile head even in perfectly homogeneous
even in the absence of a superstructure, be subjected to a soil [12, 20–23].
spatially-variable displacement field imposed by the surrounding The simplest approach for computing kinematic bending along
soil which gives rise to a dynamic interplay known as ‘‘kinematic a pile is to neglect pile–soil interaction and assume that pile and
interaction’’ [1–3]. The ensuing deformations naturally coexist soil movement coincides at all times. This procedure has been
with motions transmitted onto the pile through the pile cap due suggested by Margason [9] and yields the following predictive
to superstructure dynamics, an effect commonly referred to as equation for pile bending moment:
‘‘inertial interaction’’ [1–3]. Note that inertial interaction is as
affected by kinematic interaction as the input motion to the M ¼ Ep Ip ð1=RÞp ¼ Ep Ip ð1=RÞs ¼ Ep Ip ð1Þ
V 2s
former problem is the output motion of the latter [4–7].
Starting with the pioneering work by Blaney et al. [8], a large where Ep and Ip are Young’s modulus and the cross-sectional
number of analytical studies have demonstrated the importance moment of inertia of the pile, (1/R)p and (1/R)s are the pile and soil
of kinematic effects on piles [9–14]. In addition to the theoretical curvature, respectively, as ¼ as(z,t) or as ¼ as(z,o) is the depth-
work, post-earthquake investigations [15–17] have highlighted varying horizontal ground acceleration and Vs is the shear wave
the vulnerability of pile foundations (even in non-liquefied soil) propagation velocity in the soil. A drawback of this approach
by revealing damage at the pile head and/or depths where inertial clearly lies in the inability of Eq. (1) to handle layered soil (as soil
forces are negligible. Seismic regulations [18,19] have acknowledged curvature is infinite at interfaces separating soil layers of different
stiffness) and boundary conditions at the pile head and tip. For
instance, Eq. (1) would always predict maximum bending at the
n
Corresponding author. Tel.: þ30 2610 996542; fax: þ30 2610 996576. pile head even in the absence of a restraining cap (free head
E-mail address: [email protected] (G. Mylonakis). conditions).

0267-7261/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soildyn.2012.09.011
184 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

Nomenclature uff free-field displacement

uf f 0 free-field displacement amplitude
Latin symbols
ug base displacement
ug 0 base displacement amplitude
acutoff cutoff frequency V sn complex-valued soil shear wave propagation velocity
a0 dimensionless frequency w pile displacement
as soil acceleration z vertical coordinate
Ap pile cross-sectional area (1/R)c complementary curvature at depth z
A, B, C, D integration constants (1/R)p pile curvature at depth z
c Winkler dashpot coefficient ð1=RÞp0 pile curvature at pile head (level z ¼0)
d pile diameter ð1=RÞp0 ,static static pile curvature at head (level z¼0)
Ep pile Young’s modulus (1/R)s soil curvature at depth z
Es, Gs soil Young’s modulus, soil shear modulus
H thickness of soil layer
Greek symbols
Ip pile cross-sectional moment of inertia
Iu translational kinematic response coefficient
b Winkler damping coefficient
Ij, Iy rotational kinematic response coefficients
bs soil material damping coefficient
k* complex-valued Winkler modulus
G dimensionless response coefficient
k dynamic Winkler stiffness
d Winkler stiffness coefficient ( ¼k/Es)
L(¼ H) pile length, soil thickness
l Winkler wavenumber
M bending moment
~p lstatic static Winkler ‘‘wavenumber’’
m mass per unit pile length
Q shear force
ns soil Poisson’s ratio
q soil wavenumber
rs, rp soil, pile mass density
CRz pile–soil curvature ratio at depth z
t soil shear stress
CR0 pile head curvature over soil surface curvature (z¼0)
o cyclic excitation frequency
CRL pile tip curvature over soil surface curvature (z¼L)

To account for pile–soil interaction and, thereby, stiffness 2. Problem definition

mismatch between pile and soil, as well as different boundary
conditions at the ends of the pile, various analytical techniques The problem considered is depicted in Fig. 1: a single vertical
have been developed over the past decades. A particularly cylindrical pile of length L, diameter d, mass density rp and
attractive family of methods are the Winkler models which Young’s modulus Ep is embedded in a homogeneous soil layer of
consider the pile as a flexural beam connected to a bed of thickness H( ¼L) resting on a rigid base. Soil is modelled as a
independent springs and dashpots distributed along its axis, to linear elastic material of Poisson’s ratio ns, mass density rs and
simulate the restraining and dissipative action of soil. On the basis frequency-independent material damping bs, expressed through a
of this approach, Flores-Berrones and Whitman [10] derived complex-valued shear modulus Gns ¼ Gs ð1 þ 2ibs Þ. The pile is
(implicitly) the ratio of pile and soil curvature for a fixed-head loaded by vertically propagating shear waves expressed in the
pile embedded in a homogeneous halfspace under harmonic form of a harmonic horizontal displacement ug ðtÞ ¼ ug 0 exp½iot
excitation consisting of vertically-propagating S waves. In this applied at rock level. Considering different boundary conditions at
case, pile-to-soil curvature ratio was found to be always smaller the pile head (fixed, free to rotate) and pile tip (fixed, hinged, free
than unity and decreasing with frequency, thus reflecting the to displace and rotate), provides six distinct cases to be examined
inability of the pile to follow short wavelengths in the soil.
Further studies by Dobry and O’Rourke [11], Mylonakis [13],
Nikolaou et al. [16] and de Sanctis et al. [20], resulted in a number
of analytical solutions and empirical formulas for bending of piles
embedded in homogeneous or two-layer soil, showing that pile
curvature may exceed soil curvature under certain conditions.
Other contributions [17,24–30] have investigated the behaviour
of piles in two- and multi-layer soil deposits under both harmonic
and transient excitation.
Despite these efforts, certain fundamental mechanisms
governing the development of bending along kinematically-
loaded piles remain poorly understood, even for idealized
conditions such as homogeneous soil and low-frequency seis-
mic excitation. Of particular interest are counterintuitive cases
where pile curvature is larger than soil curvature and the role
of boundary conditions at pile head and tip. The work at hand
aims at offering insight into these aspects by: (1) presenting a
new set of analytical solutions pertaining to different boundary
conditions; (2) introducing new dimensionless parameters governing
static and dynamic pile response to vertically-propagating Fig. 1. Problem considered and associated boundary conditions at pile head
SH waves. and tip.
 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195 185

response and pile response. Each sub-problem is addressed

separately below.

3.1. Free-field response

In one-dimensional analysis, the linear stress–strain law is,

according to the notation of Fig. 2
@uf f
t ¼ Gns ð7Þ
@z
where Gns is the complex soil shear modulus, t is the shear stress
and uff ¼ uff(z, t) is the time- and depth-dependent displacement
Fig. 2. Positive notation for forces and stresses on pile and soil, respectively.
of the free-field soil.
Considering forced harmonic oscillations of the type uff
in the ensuing (Fig. 1). Positive notation for stresses and displace- (z, t)¼uff(z) eiot, the equilibrium of forces in the horizontal direction
ments is provided in Fig. 2. acting upon an arbitrary soil element yields the familiar second-
The problem at hand is governed by seven dimensional para- order differential equation
meters (L, d, Ep, Es, o, rp, rs), in addition to the inherently
du2f f
dimensionless ratios bs and ns. Given that three fundamental þ q2 uf f ¼ 0 ð8Þ
dimensions are involved (mass, time, length), Buckingham’s dz2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
theorem [31] suggests that the interaction problem can be fully where V ns ¼ V s 1 þ 2ibs and q ¼ o=V ns is the complex shear wave
described by six (¼7þ23) dimensionless ratios (e.g., L/d, Ep/Es, ns, propagation velocity in the soil and the corresponding wavenumber,
rp/rs, bs and o d/Vs). As will be shown in the ensuing, the adopted respectively.
Winkler model leads to a drastic reduction in the number of Solving Eq. (8) and imposing the boundary condition of a
governing independent variables. In particular, response in the static traction-free soil surface, the following well-known solution is
regime is found to be controlled by a unique dimensionless variable, obtained [32]:
whereas in the dynamic regime two parameters are generally
sufficient for describing the interaction problem. It will also be uf f ðz,tÞ ¼ uf f 0 cosðqzÞeiot ð9Þ
shown that in the realm of the Winkler model, pile–soil interaction
which describes a standing wave of amplitude uf f 0 at soil
for long piles can be described through a single backbone curve,
surface (z ¼0). Assuming that the amplitude of motion at base
depending solely on a novel frequency parameter.
level (z ¼H) is known, the familiar amplification function is
Solutions from such analyses can be conveniently expressed
recovered [32,33]
through the so-called kinematic response factors Iu and Ij. These
are defined, respectively, as the translation and rotation ampli- uf f 0 1
¼ ð10Þ
tudes at the pile head normalized by the corresponding displace- ug 0 cosðoH=V ns Þ
ment amplitude at the surface of the free-field soil i.e. [8],
wð0, oÞ
Iu  ð2Þ
uf f ð0, oÞ
3.2. Pile response
w0 ð0, oÞ d
Ij  ð3Þ
uf f ð0, oÞ In the realm of the approach at hand, free-field displacements
w(z, o) and uff ¼ uff(z, o) being the frequency- and dependent are applied at the base of the Winkler supports, which constitute
displacement of the pile and the free-field soil, respectively, and the dynamic excitation that forces the pile to deflect. The
d ¼pile diameter. equilibrium of horizontal forces acting on an arbitrary pile
In the same spirit, the following curvature ratios between pile segment yields the governing equation (Fig. 2):
and soil can be defined as @Q @2 w
n
~p
k ðwuf f Þm ¼0 ð11Þ
ð1=RÞp 9z ¼ 0 w ð0, oÞ
00
V 2s @z @t 2
CR0 ¼ ¼ ð4Þ
ð1=RÞs 9z ¼ 0 as ð0, oÞ ~ p ¼ rp Ap mass density per unit
where Q¼Q(z, t) is shear force, m
pile length and w¼w(z, t) pile displacement. k* ¼kþioc is the
ð1=RÞp 9z ¼ L w00 ðL, oÞ V 2s complex-valued Winkler modulus, k being the stiffness of the
CRL ¼ ¼ ð5Þ
ð1=RÞs 9z ¼ 0 as ð0, oÞ Winkler springs and c the corresponding dashpot coefficient
[3,34,35].
ð1=RÞp w00 ðz, oÞ V 2s Considering forced harmonic oscillations of the type
CRz ¼ ¼ ð6Þ
ð1=RÞs 9z ¼ 0 as ð0, oÞ wðz,tÞ ¼ wðzÞ exp½iot and given that shear force is related to
displacement through the strength-of-materials expression [36]
corresponding to the pile head (CR0), pile tip (CRL) and an
arbitrary elevation (CRz). In the above equations ( )’’ denotes @3 w
Q ðz,tÞ ¼ Ep Ip ð12Þ
double differentiation with respect to depth. @z3
the equation governing pile motion can be rewritten in the
3. Model development Navier form [10,16,35]
4 n
Following earlier studies, the problem is treated in the context d w 4 k
þ4l w ¼ u ð13Þ
of two modular problems, namely the analysis of free-field soil dz4 Ep Ip f f
186 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

where l ¼ l(o) is the characteristic wavenumber governing the (ratio of pile head absolute displacement to free-field surface
attenuations of pile displacement with depth absolute displacement) under free-tip, hinged-tip and fixed-tip
  conditions are, respectively
k þ ioco2 m~ p 1=4
l¼ ð14Þ 
2   
4Ep Ip q cosðqLÞ½coshðlLÞsinðlLÞcosðlLÞsinhðlLÞ q=l cosðlLÞcoshðlLÞsinðqLÞ
Iu ¼ G 1
l sinð2lLÞ þsinhð2lLÞ
Note that even though l is complex-valued, for the sake of ð22Þ
simplicity no superscript ()* is used to distinguish it from its real
2 h   i3
counterpart. q
2 cosðqLÞ 2ð11=GÞ l=q 2 cosðlLÞcoshðlLÞ þ sinðlLÞsinhðlLÞ
The general solution to the above equation is [10] Iu ¼ G41 5
l cosð2lLÞ þcoshð2lLÞ
wðz, oÞ ¼ ðA coslz þ B sinlzÞelz þ ðC coslz þ D sinlzÞelz þ Guf f ðz, oÞ
ð23Þ
ð15Þ
2   3
where A, B, C, D are integration constants dependent of the boundary cosðqLÞ cosðlLÞsinhðlLÞ þ coshðlLÞsinðlLÞ ð11=GÞ
6   7
conditions; G is a dimensionless response coefficient given by 6 þ q=l sinðlLÞsinhðlLÞsinðqLÞ 7
Iu ¼ G6
612
7
7
k þioc 4 sinð2lLÞ þ sinhð2lLÞ 5
G¼ 4
ð16Þ
Ep Ip ðq4 þ 4l Þ
ð24Þ

Note that for the limit case of an infinitely-long pile, all the
3.3. Pile–soil curvature ratio above relations converge to the simple solution of Flores-Berrones
and Whitman [10]
Enforcing the boundary conditions at the pile tip, the ratios of
Iu ¼ G ð25Þ
pile curvature at the pile head and the corresponding soil
curvature at the same elevation, CR0, for a fixed-head pile are a result which also holds for curvature ratios. Note that Eq. (22)
obtained as (Eq. (4)) has been presented (inadvertently referred to as curvature ratio
2 3 CR0) in Ref. [16].
cosðqLÞ½cosðlLÞsinhðlLÞ þcoshðlLÞsinðlLÞ
6 q 7 With reference to free-head piles, kinematic response coeffi-
6 þ sinðlLÞsinhðlLÞsinðqLÞ 7
6 l 7 cients for free-tip, hinged-tip and fixed-tip conditions are, respec-
CR0 ¼ G6
6 12 7
7 ð17Þ tively
6 sinð2lLÞ þsinhð2lLÞ 7
4 5
Iu ¼
 q
3 sinðqLÞ½cosðlLÞsinhðlLÞcoshðlLÞsinðlLÞ2l=qsinðlLÞsinhðlLÞcosðqLÞ
2 h  2 i3 G 1þ
cosðqLÞ cosðlLÞcoshðlLÞ2ð11=GÞ l=q sinðlLÞsinhðlLÞ l cosð2lLÞ þcoshð2lLÞ2
CR0 ¼ G412 5 ð26Þ
cosð2lLÞ þ coshð2lLÞ

ð18Þ 2
2   3
2ð11=GÞ lq cosðlLÞsinhðlLÞcoshðlLÞsinðlLÞ cosðqLÞ
2  2   3 6 7
6 þ sinðlLÞ½coshðlLÞcosðqLÞcosðlLÞ 7
2ð11=GÞ l=q cosðlLÞsinhðlLÞcoshðlLÞsinðlLÞ cosðqLÞ 6 7
6   7 6 7
6 þ2 l=q cosðlLÞcoshðlLÞsinðqLÞ 7 6 q
2 þ sinhðlLÞ½cosðlLÞcosðqLÞcoshðlLÞ 7
6 7 6 7
CR0 ¼ G612 7 Iu ¼ G61þ 7
6 sinð2lLÞ þsinhð2lLÞ 7 6 l sinð2lLÞsinhð2lLÞ 7
4 5 6 7
6 7
6 7
4 5
ð19Þ
corresponding to free-tip, hinged-tip and fixed-tip conditions, ð27Þ
respectively. Eq. (17) and (18) have first been reported in Ref. [16]. 2  2 3
For a fixed-tip pile, pile curvature at pile tip over soil curvature 4ð11=GÞ l=q cosðlLÞcoshðlLÞcosðqLÞ þ½coshð2lLÞcosð2lLÞ=2
6   7
6 q
2 2 l=q sinðqLÞ½cosðlLÞsinhðlLÞ þcoshðlLÞsinðlLÞ 7
at surface, CRL, for fixed- and free-head conditions is respectively 6
Iu ¼ G61þ
7
7
(Eq. (5)): 6 l cosð2lLÞ þ coshð2lLÞ þ 2 7
4 5
2 " #3
sinð2lLÞ þ sinhð2lLÞ
6 cosðqLÞ   2   7 ð28Þ
6 þ2 11=G l=q sinð2lLÞsinhð2lLÞ 7
4    5
2 l=q cosð2lLÞ þcoshð2lLÞ sinðqLÞ
CRL ¼ G
sinð2lLÞ þsinhð2lLÞ
ð20Þ 3.5. Rotational kinematic response factor
"  2 #
cos ðqLÞ½2 þ cos ð2lLÞ þ cosh ð2lLÞþ 2ð11=GÞ l=q ½cos ð2lLÞcosh ð2lLÞ To quantify pile head rotation, a second interaction coefficient
þ 2ðl=qÞ½sin ð2lLÞsinh ð2lLÞsin ðqLÞ4cosðlLÞ coshðlLÞ can be defined as
CRL ¼ G
2 þ cos ð2lLÞþ cosh ð2lLÞ
ð21Þ w0 ð0, oÞ
Iy ¼ ð29Þ
l uf f 0
where G is given by Eq. (16).
It should be noticed that the above definition is different from
3.4. Translational kinematic response factor the ordinary coefficient Ij (note the different subscript) in Eq. (3)
invariably employed in the literature [4,6,16], as in this way pile
For the aforementioned case of a fixed-head pile, correspond- head rotation depends solely on dimensionless frequency and
ing expressions for the kinematic response coefficient Iu in Eq. (2) Winkler parameter (o/lVs) and lL, respectively.
 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195 187

Expressions for Iy pertaining, respectively, to free-tip, hinged- Blaney et al. [8] and Syngros [39] for related problems. For the
tip and fixed-tip conditions are: problem at hand, only the first term of the series is relevant and,
" # thereby, solving a single FE configuration is sufficient for the
2cosðqLÞ½coshðlLÞsinðlLÞ þcosðlLÞsinhðlLÞsinð2lLÞsinhð2lLÞ
  lateral response mode. Owing to this procedure, the original
q
2 þ2 q=l sinðlLÞsinhðlLÞsinðqLÞ
Iy ¼ G three-dimensional problem is conveniently reduced to a two
l cosð2lLÞ þcoshð2lLÞ2
dimensional one. The domain was discretized using 4-noded
ð30Þ
axisymmetric elements; following a sensitivity analysis, the lateral
2 h  2 i3 dimension of the model was set equal to 200d, to ensure that soil
2cosðqLÞ 2ð11=GÞ l=q sinðlLÞsinhðlLÞcosðlLÞcoshðlLÞ
4 5 response close to the boundaries is not affected by outward-
q
2 þcosð2lLÞ þ coshð2lLÞ spreading waves emitted from the pile–soil interface. Likewise,
Iy ¼ G
l sinð2lLÞsinhð2lLÞ vertical displacements were restrained along the lateral boundary of
ð31Þ the mesh to simulate 1-dimensional conditions for S waves at large
distances from the pile. In addition, nodes at the base of the model
"  2 #
4ð11=GÞ l=q ½cosðlLÞsinhðlLÞcoshðlLÞsinðlLÞcosðqLÞ were fully restrained to represent the rigid bedrock. Vertical size of
 
q
2 sinð2lLÞsinhð2lLÞ þ 4 l=q cosðlLÞcoshðlLÞsinðqLÞ the elements was kept constant, equal to d/4, which was found to be
Iy ¼ G sufficiently accurate and economical. The analyses were carried out in
l cosð2lLÞ þ coshð2lLÞ þ 2
the frequency domain [8,21,37], the load being applied in the form of
ð32Þ
a harmonic lateral body force in each element.
Discussion of the above analytical developments is provided in
the remainder of the article. 4.1. Static response

It is well-known that in Winkler models pile–soil interaction is

4. Interpretation of results and comparison with other controlled by the key dimensionless parameter [10,35]
solutions  1=4  1=4   1=4
k 16d L Ep
lL ¼ L ¼ ð33Þ
4Ep Ip p d Es
For comparison purposes, rigorous Finite Element (FE) analyses
were performed by means of the commercial computer platform being the product of parameter l (evaluated for o ¼0) in Eq. (14)
ANSYS [37]. Given that the geometry is axisymmetric and the load and pile length L. In such models lL is a unique parameter
anti-symmetric, stresses and displacements were expanded in controlling static response, which can be interpreted as a
Fourier series along the circumferential direction, following the ‘‘mechanical slenderness’’ (as opposed to the familiar geometrical
technique introduced by Wilson [38] and later employed by slenderness L/d) for it encompasses both geometry (L/d) and

Fig. 3. Variation of pile–soil curvature ratio at pile head under static conditions (o ¼ 0), as function of pile slenderness for selected values of Winkler stiffness parameter d.
(a) Ep/Es ¼ 1000, (b) Ep/Es ¼10000, (c) Ep/Es ¼ 1000 and (d) Ep/Es ¼ 10000.
188 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

Fig. 4. Variation of pile–soil curvature ratio at pile tip under static conditions (o ¼ 0), as function of pile slenderness for selected values of Winkler stiffness parameter d.
(a) Ep/Es ¼ 1000, (b) Ep/Es ¼10000, (c) Ep/Es ¼1000 and (d) Ep/Es ¼ 10000.

pile–soil relative stiffness (Ep/Es). In addition, lL is function of

Winkler stiffness coefficient d ¼k/Es, the value of which lies in the
core of the Winkler representation of soil reaction. The effect of d
on kinematic response is examined in Figs. 3 and 4, where pile–
soil curvature ratio is plotted against pile slenderness L/d for two
values of pile–soil stiffness contrast. It is evident by inspecting
Figs. 3 and 4, that the predictions of Winkler models employing
the commonly used value d ¼1.2 pertaining to inertial interaction
analyses [40,41], are not in good agreement with the FE results for
certain cases examined. In Fig. 3a and 3b the value of d that
matches the FE results seems to decrease with increasing pile
slenderness and with decreasing pile–soil stiffness ratio. More-
over, different values for ‘‘optimum’’ d are obtained depending on
the boundary conditions at the tips as shown in Fig. 3c and 3d. In
addition, optimum d clearly depends on the parameter to be
matched. For instance, it is evident from Fig. 4a and 4b that values
of d matching CRL (ratio of pile curvature at tip over soil curvature
at soil surface) differ from those in Fig. 3, referring to CR0, and are
independent of pile slenderness L/d. In light of this observation,
results for static pile–soil curvature ratio presented in Figs. 5–14
are plotted in terms of lL.
In Fig. 5, static pile–soil curvature ratio at the pile head is
plotted against lL for different boundary conditions at the pile tip.
Naturally, all curves start from zero since for lL-0 the pile
degenerates into a rigid disk, thus essentially experiencing zero
moment regardless of restraints at the tip. With increasing lL
Fig. 5. Variation of static pile–soil curvature ratio at pile head with lL for a fixed- curvature ratio gradually increases attaining unity at points A1,2
head pile and various boundary conditions at the tip. and A3, and reaching a maximum, above unity, at points M1, M2,
 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195 189

Fig. 6. Kinematic bending along an infinitely-long pile as a superposition of a kinematic and an external load on two piles of finite length (hatched areas denote bending
moment).

Fig. 7. Variation of static pile–soil curvature ratio with depth for free-tip piles, Fig. 8. Variation of static pile–soil curvature ratio with depth for hinged-tip piles,
for different geometric and material configurations. (a) fixed-head free-tip and for different geometric and material configurations. (a) fixed-head hinged-tip and
(b) free-head free-tip. (b) free-head hinged-tip.
190 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

substituting the expression for (1/R)c (see [42]), Eq. (34) duly
reduces to Eq. (17) for z¼0.
The above procedure can be extended to account for the more
general case of a restrained pile tip. This can be achieved by
introducing in Fig. 6b the pertinent restraining actions due to full
or partial fixity at the tip. To ensure equilibrium, the opposite
actions must be applied at pile tip in Fig. 6c. For hinged-tip
condition a horizontal force must be applied at the tip, whereas
for a fixed-tip both a force and a moment are required. Note that
because of the statically indeterminate nature of the problem in
Fig. 6b, the values of these actions are not known a priori.
An alternative interpretation of the trends observed in Fig. 5 is
possible by means of the aforementioned superposition approach.
Indeed, the complementary curvature at the pile head may
possess different signs depending on the value of lL. Specifically:
for a very short pile the complementary moment at the pile head
will be equal to soil curvature, thus leading to a zero overall
moment at the top. On the other hand, for an infinitely-long pile
curvature at the head will be equal to soil curvature, as the
external moment in Fig. 6c will not be transmitted to the pile top.
For short piles the moment transmitted to the head has the same
sign as the applied moment. This results in pile–soil curvature
ratios lower than unity. For longer piles, the complementary
moment becomes negative leading to a curvature ratio higher
than unity.
The profile of pile curvature with depth over soil curvature at
the surface, CRz, is presented in Figs. 7–9 for various head and tip
conditions and the values of lL shown at the inset of Fig. 5. For
fixed-head, free-tip conditions (Fig. 7a) the bending moment
along a short pile (lLo5.49, Fig. 5) attains its maximum value
at the top, decreasing monotonically with depth. For long piles
(lL45.49, Fig. 5) the maximum curvature ratio develops at depth
Fig. 9. Variation of static pile–soil curvature ratio with depth for fixed-tip piles,
and attains the value of 1.04. The depth zmax corresponding to
for different geometric and material configurations. (a) fixed-head fixed-tip and maximum curvature ratio may be related to lL by means of the
(b) free-head fixed-tip. aforementioned superposition scheme through the easy-to-derive
expressions valid for free-tip conditions

M3 depending on the conditions at the tip. A further increase in lL lzmax ¼ lL3:14, lL4 5:49 ð35Þ
causes pile curvature to drop and gradually converge to soil For free-head, free-tip piles (Fig. 7b) all curves are symmetrical.
curvature (B1,2, B3, T), as the pile becomes sufficiently flexible to Specifically, for 0o lLo2p (short piles) pile bending attains its peak
follow soil deformation, regardless of tip conditions. value at mid-depth. For lL42p (long piles) a maximum is observed
A better understanding of the above trends, which may explain at two symmetric distances lzmax ¼ p from the head and tip.
the counterintuitive values of curvature ratios larger than one, Pile–soil curvature ratio for hinged-tip piles is presented in
may be achieved through a simple mechanistic approach, which Fig. 8. The behaviour of fixed-head piles depicted in Fig. 8a is
interprets the curvature of a fixed-head pile of length L as the similar to the one shown in Fig. 7a: bending moment attains its
superposition of: (1) the curvature of an infinitely-long pile maximum at the top for short piles (lLo4.71, Fig. 5), whereas for
(CR ¼1 at any depth) and (2) a ‘‘complementary’’ curvature profile long piles (lL4 4.71, Fig. 5) maximum curvature ratio develops at
accounting for finite pile length and the specific boundary condi- depth and attains a constant value of 1.07. In the same fashion,
tion at the tip. As an example, for a fixed-head pile the expression depth zmax corresponding to the maximum curvature ratio is
for static pile–soil curvature ratio at any depth can be cast in the related to lL through the expression
form:
ð1=RÞp ð1=RÞc lzmax ¼ lL2:4, lL4 4:71 ð36Þ
¼ 1 ð34Þ
ð1=RÞs ð1=RÞs
where 1 in the right side is the curvature ratio for an infinitely- The hinged-tip pile in Fig. 8b experiences a curvature pattern
long fixed-head pile in static regime [for which (1/R)p ¼(1/R)s], analogous to that in Fig. 7b. The maximum value of curvature
and (1/R)c is a ‘‘complementary’’ curvature at depth z as defined ratio is observed at z¼0.58L for 0 o lLo5.49 (short piles),
above. This interpretation is schematically shown in Fig. 6 for the whereas for higher values of lL a maximum is observed at the
case of a floating pile (free-tip condition): an infinitely-long pile depth given by Eq. (36).
embedded in homogeneous soil is conceptually separated from For fixed-tip piles, the maximum curvature is always observed at
the underlain material at depth z¼L (Fig. 6a), the curvature the tip and has the opposite sign to that at the top (Fig. 9). It is noted
pattern of the upper part of the pile in Fig. 6a is tantamount in passing that a quick estimate of the curvature ratio at the pile
to the superposition of the curvature along a free-tip pile of base can be obtained using the expression of Dobry and O’Rourke
length L (Fig. 6b) and the complementary curvature of the same [11] derived for an infinitely-long pile in two-layer soil, considering
pile subjected at its tip to an action Ep Ip(1/R)s (Fig. 6c), due an infinitely-stiff bottom layer. This leads to an overestimation of
to the ‘‘detached’’ lower part in Fig. 6a. For this particular case, curvature ratio at the pile tip by 2lL.
 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195 191

4.2. Dynamic response frequency a0 ¼ od/Vs for selected pile–soil configurations. Predic-
tions using the static value of l in Eq. (33) are compared to those
Employing the approximate relations for the distributed dashpot obtained from the complete formulation in Eq. (37) and to FE
1=4
coefficient along the pile c ¼ 6a0 rs V s d þ 2bs dEs =o derived using results. Different values for d are used for each case, based on an
planar wave-propagation analysis [34], the complex-valued wave- optimal selection according to Fig. 3. A convergence of all curves
number l can be related to its static counterpart through below cutoff frequency is observed. Beyond cutoff, however, the
8 1=4 static assumption leads to a better agreement with the more
>
>
> 1 p a20 rp rigorous FE results. This is probably due to the approximate
>
> þ i2 bs , a0 r acutof f
l < 8ð1 þ ns Þ d rs description of radiation damping employed in Eq. (37), which
¼ " !# 1=4
lstatic > 3=4 was based on inertial interaction considerations [34]. Nevertheless,
>
> p a20 rp 3a0 1
> 1
> þi þ 2bs , a0 4 acutof f even under a more realistic representation of geometric energy
: 8ð1þ ns Þ d rs 1 þ ns d
dissipation, the benefit stemming from the simplified approach
ð37Þ cannot be overstated. It is also worth noting that optimum d
exhibits only a weak dependence on frequency. Accordingly,
where l and lstatic are obtained from Eqs. (14) and (33), respectively,
optimum d for static analysis (Fig. 3) can be employed in the
and acutoff stands for a characteristic frequency (termed ‘‘cutoff
dynamic regime as well (Fig. 10).
frequency’’) below which no stress waves can be emitted from the
Additional comparisons of the proposed model against FE results
pile–soil interface to propagate horizontally in the soil medium and,
obtained as part of this study [37] and from the literature [6] are
thereby, no radiation damping is generated. The cutoff frequency is,
presented in Figs. 11 and 12 in terms of translational and rotational
accordingly, associated with a sudden increase in damping and
kinematic response factors Iu and If. It is evident that the predictions
coincides with the fundamental frequency of the soil layer in
of the model are in satisfactory agreement with the results of the
shearing and is expressed in dimensionless form as
more rigorous solutions for all configurations examined.
acutof f ¼ ðp=2ÞðH=dÞ1 ð38Þ In the remainder of the article, dynamic effects are discussed
in terms of pile curvature and kinematic response factors Iu and Iy.
Note that for the range of frequencies relevant to earthquake In light of the analytical developments in Eqs. (22)–(32), it can be
engineering, the term related to pile density in Eq. (37) may be readily recognized that the adoption of a0 (which is independent
neglected without significant error. of mechanical slenderness) as an independent frequency variable
In the same spirit as in static analysis, dynamic pile response would not allow (lL) to be the main parameter controlling the
can be described by a unique dimensionless parameter. This is response. It is observed that the excitation frequency appears in
achieved by using the static value of l in Eq. (33) in the dynamic the solutions only in dimensionless terms qH ( ¼ oH/Vs) and
regime. The validity of this approximation is explored in Fig. 10, q/l ( ¼ o/lVs), thereby, these frequency parameters can be used
in which pile–soil curvature ratio at the head is plotted against for expressing results in the dynamic regime.

Fig. 10. Variation of dynamic pile–soil curvature ratio at pile head with frequency for fixed-head free-tip piles, for different geometric and material configurations:
comparisons of rigorous elastodynamic FE results with Winkler solutions obtained using the optimum static d value in Fig. 3. bs ¼ 0.10. (a) L/d¼ 5, Ep/Es ¼1000, (b) L/d¼ 5,
Ep/Es ¼10000, (c) L/d ¼10, Ep/Es ¼ 1000 and (d) L/d¼ 10, Ep/Es ¼10000.
192 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

Fig. 11. Variation of kinematic response factor Iu for free-head free-tip piles:
comparisons of rigorous elastodynamic FE results with Winkler solutions obtained
using the optimum static d value in Fig. 3. bs ¼ 0.05 (a) Ep/Es=1000, (b) Ep/Es=10000. Fig. 12. Variation of kinematic response factor If for free-head free-tip piles:
comparisons of rigorous elastodynamic FE results with Winkler solutions obtained
using the optimum static d value in Fig. 3. bs ¼0.05 (a) Ep/Es=1000, (b) Ep/Es=10000.

For an infinitely-long fixed-head pile, pile–soil curvature ratio at

all depths, CR, and kinematic response factor Iu can be cast in the form mechanical slenderness are lower than in the previous case. This
" pattern may reflect that strains are more sensitive to boundary
 4 #1
1 o conditions than displacements.
CR ¼ Iu ¼ 1 þ ð39Þ
4 lV ns For free-head piles kinematic response factors Iu and Iy are plotted
in Figs. 15 and 16. A common trend is observed: both factors increase
Which coincides with factor G in Eq. (16) and clearly indicates with increasing frequency up to a certain value of (o/lVs). Beyond
that the response of long piles depends solely on the unique this value the trend is reversed with Iu and Iy decreasing with
frequency parameter (o/lVs) and soil material damping bs, not on frequency. This behaviour can be explained in light of the wave-
dimensionless frequency a0. lengths developing in the soil at different frequencies. With increasing
In Fig. 13, the dynamic de-amplification of curvature ratio at the frequency, wavelengths become shorter forcing the pile to experience
pile head is plotted against (o/lVs) for fixed-head piles under stronger rotations along its length. This also leads to higher displace-
different boundary conditions at the tip. Dynamic pile curvature ments atop free-head piles. Note that the maximum rotation at the
decreases with frequency for all piles of finite length, as the pile is pile head is equal to the ratio of free-field displacement at the soil
unable to follow short wavelengths in the soil. For the trivial case surface, uf f 0 , and the characteristic wavelength of the pile, (1/l).
lL¼ 0 dynamic de-amplification is equal to one, as dynamic pile
curvature is always zero. With increasing lL all curves approach that
corresponding to the infinitely-long pile regardless of tip conditions. 5. Conclusions
The threshold value of lL beyond which a pile behaves as an
infinitely-long beam depends on end condition and is strictly related A Beam-on-Dynamic-Winkler-Foundation model was employed
to the static behaviour depicted in Fig. 5. Indeed, if a pile behaves for investigating the behaviour of kinematically stressed piles of finite
as an infinitely-long beam under static conditions (lL44.71, length embedded in a homogeneous soil layer, for different boundary
5.49—Fig. 5), it behaves the same way in the dynamic regime as conditions at the head and tip. Analytical solutions for pile response
well. Note that with increasing lL the curves do not evolve in a were derived in closed form.
monotonic manner. Interestingly, the lower curve corresponds to The main conclusions of the study may be summarized as
the value of lL for which the curvature ratio is maximum in static follows:
conditions (lL corresponding to 2.57, 2.31, p, see Fig. 5).
Similar observations can be made for the kinematic response 1) Owing to its simplicity, the adopted analytical model can shed
factor Iu (Fig. 14), except for that the threshold values of light on certain fundamental mechanisms controlling pile–soil
 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195 193

Fig. 13. Variation of dynamic pile curvature ratio at pile head with frequency for
Fig. 14. Variation of kinematic response factor Iu with frequency for fixed-head
fixed-head piles under different tip conditions.
piles under different tip conditions.

interaction. Its performance, however, is related to the proper 3) Pile curvature may be decomposed into the sum of soil
selection of stiffness coefficient d which depends on a number of curvature and a complementary curvature that develops along
parameters such as pile slenderness, pile–soil stiffness ratio, the pile subjected to pertinent forces and moments at the two
boundary conditions, as well as on the parameter to be matched. ends. These forces depend on the specific boundary conditions
Nevertheless, it is observed (Fig. 3) that d attains higher values for and are responsible for the counterintuitive phenomenon of
small pile slenderness and large pile–soil stiffness ratios, and pile curvature higher than soil curvature for certain values of
appears to be independent of frequency (Figs. 3 and 10). pile slenderness.
2) In Winkler models, pile–soil kinematic interaction is governed 4) A new dimensionless frequency factor (o/lVs) was introduced
by a unique dimensionless parameter, lL (Eq. (33)), which can for normalizing response in the dynamic regime. It was shown
be interpreted as a ‘‘mechanical slenderness’’, encompassing that this allows long piles to exhibit the same response
key problem parameters namely pile slenderness, pile–soil regardless of actual length and pile–soil stiffness ratio. This
stiffness ratio and Winkler coefficient d. A unique parameter can be understood, since the dimensionless frequency is
(lL) governs the response at static conditions. The same expressed as ratio of characteristic pile wavelength (1/l) and
parameter controls the behaviour in the dynamic regime if soil wavelength (Vs/o) at a given frequency. As a follow up, a
pile inertia and radiation damping are neglected. This simpli- new kinematic response factor, Iy, was introduced to describe
fication allows a better understanding of the interaction pile head rotation (Eq. (29)). In this way, the interaction is
phenomenon and leads to a better agreement of the closed- function only of the aforementioned frequency factor and
form solutions with rigorous numerical results. mechanical slenderness.
194 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195

Fig. 15. Variation of kinematic response factor Iu with frequency for free-head Fig. 16. Variation of kinematic response factor Iy with frequency for free-head
piles under different tip conditions. piles under different tip conditions.

Acknowledgements [2] Wolf JP. Dynamic soil–structure interaction. Englewood Cliffs, NJ: Prentice-
Hall; 1985.
[3] Gazetas G, Mylonakis G. Seismic soil–structure interaction: new evidence and
The research reported in this paper was conducted under the emerging issues. In: Dakoulas P, Yegian EvlK, Holtz RD, editors. Geotechnical
auspices of the ReLUIS project ‘‘Methods for risk evaluation and earthquake engineering and soil dynamics III ASCE; 1998.
management of existing buildings’’, funded by the Italian National [4] Kaynia AM, Kausel E. Dynamic stiffness and seismic response of pile groups.
Research Report R82-03. Cambridge, MA: Massachusetts Institute of Tech-
Emergency Management Agency and was partially supported by
nology; 1982.
the University of Patras through a Caratheodory Grant (no. C.580). [5] Mamoon SM, Banerjee PK. Response of piles and pile groups to travelling SH
The authors are grateful for this support. The authors also would waves. Earthquake Engineering and Structural Dynamics 1990;19:597–610.
like to thank the anonymous Reviewers whose comments improved [6] Fan K, Gazetas G, Kaynia A, Kausel E, Ahmad S. Kinematic seismic response of
single piles and piles groups. Journal of Geotechnical Engineering Division
the quality of the manuscript. ASCE 1991;117(12):1860–79.
[7] Rovithis E, Mylonakis G, Pitilakis K. Inertial and kinematic response of piles in
layered inhomogeneous soil: Winkler analysis. In: Second international
References conference on performance-based design in earthquake geotechnical engi-
neering, ID:11.21, 2012.
[1] Roesset JM, Whitman RV, Dobry R. Modal analysis for structures with [8] Blaney GW, Kausel E, Roesset JM. Dynamic stiffness of piles. In: Proceedings
foundation interaction. Journal of the Structural Division 1973;99?(3): in 2nd international conference on numerical methods in geomechanics
399–416. ASCE; 1976.
 G. Anoyatis et al. / Soil Dynamics and Earthquake Engineering 44 (2013) 183–195 195

[9] Margason E. Pile bending during earthquakes. Lecture ASCE/UC-Berkeley [25] Kaynia AM, Mahzooni S. Forces in pile foundations under seismic loading.
seminar on design construction & performance of deep foundations; 1975. Journal of Engineering Mechanics ASCE 1996;122(1):46–53.
[10] Flores-Berrones R, Whitman RV. Seismic response of end-bearing piles. [26] Castelli F, Maugeri M. Simplified approach for the seismic response of a pile
Journal of Geotechnical Engineering Division ASCE 1982;108(4):554–69. foundation. Journal of Geotechnical and Geoenvironmental Engineering
[11] Dobry R, O’Rourke MJ. Discussion on ‘‘Seismic response of end-bearing piles’’ 2009;135(10):1440–51.
by Flores-Berrones R & Whitman RV’’. Journal of the Geotechnical Engineer- [27] Maiorano RMS, de Sanctis L, Aversa S, Mandolini A. Kinematic response
ing Division 1983; 109. analysis of piled foundations under seismic excitations. Canadian Geotech-
[12] Kavvadas M, Gazetas G. Kinematic seismic response and bending of free-head nical Journal 2009;46(5):571–84.
piles in layered soil. Géotechnique 1993;43(2):207–22. [28] Dezi F, Carbonari S, Leoni G. Kinematic bending moments in pile foundations.
[13] Mylonakis G. Analytical solutions for seismic pile bending. Unpublished
Soil Dynamics and Earthquake Engineering 2009;30(3):119–32.
research report. City University of New York; 1999.
[29] Sica S, Mylonakis G, Simonelli AL. Transient kinematic pile bending in two-
[14] Mylonakis G. Simplified model for seismic pile bending at soil layer inter-
layer soil. Soil Dynamics and Earthquake Engineering 2011;31(7):891–905.
faces. Soils and Foundations 2001;41(4):47–58.
[30] Cairo R, Chidichimo A. Nonlinear analysis for pile kinematic response. In:
[15] Tazoh T, Shimizu K, Wakahara T. Seismic observations and analysis of
grouped piles. Dynamic response of pile foundations: experiment, analysis Fifth international conference on earthquake geotechnical engineering; 2011.
and observation. Geotechnical Special Publication 1987;11:1–20. [31] Buckingham E. On physically similar systems; illustrations of the use of
[16] Nikolaou AS, Mylonakis G, Gazetas G, Tazoh T. Kinematic pile bending during dimensional equations. Physical Review 1914;4(4):345–76.
earthquakes analysis and field measurements. Géotechnique 2011;51(5): [32] Kramer SL. Geotechnical earthquake engineering. New York: Prentice-Hall;
425–40. 1996.
[17] Di Laora R, Mandolini A, Mylonakis G. Insight on kinematic bending of [33] Roesset JM. Soil amplification of earthquakes. In: Desai CS, Christian JT,
flexible piles in layered soil. Soil Dynamics and Earthquake Engineering editors. Numerical methods in geotechnical engineering. New York:
2012;43:309–22. McGraw-Hill; 1977.
[18] CEN/TC 250. Eurocode 8. Design of structures for earthquake resistance Part [34] Gazetas G, Dobry R. Horizontal response of piles in layered soils. Journal of
5: Foundations, retaining structures and geotechnical aspects. European Geotechnical Engineering ASCE 1984;110(1):20–40.
Committee for Standardization Technical Committee 250, Standard EN [35] Mylonakis G. Contributions to static and dynamic analysis of piles and pile-
1998–5, Brussels, Belgium; 2003. supported bridge piers. PhD thesis. State University of New York at Buffalo;
[19] National Earthquake Hazard Reduction Program. Recommended provisions 1995.
for the development of seismic regulations for new buildings. FEMA [36] Den Hartog JP. Advanced strength of materials. New York: McGraw-Hill; 1952.
publication 274. Washington, DC: Building Seismic Safety Council; 1997. [37] ANSYS Inc. ANSYS theory reference 10.0. Canonsburg, PA, USA: ANSYS Inc.;
[20] de Sanctis L, Maiorano RMS, Aversa S. A method for assessing kinematic 2005.
bending moments at the pile head. Earthquake Engineering and Structural [38] Wilson EL. Structural analysis of axisymmetric solids. American Institute of
Dynamics 2010;39(10):1133–54. Aeronautics and Astronautics Journal 1965;3:2269–74.
[21] Di Laora R. Seismic soil–structure interaction for pile supported systems. PhD
[39] Syngros C. Seismic response of piles and pile-supported bridge piers
thesis. Napoli: University of Napoli ‘‘Federico II’’; 2009.
evaluated through case histories. PhD thesis. City University of New York;
[22] Di Laora R, Mylonakis G, Mandolini A. Pile-head kinematic bending in layered
2004.
soil. Earthquake Engineering and Structural Dynamics 2012 http://dx.doi.or
[40] Novak M, Nogami T, Aboul-Ella F. Dynamic soil reactions for plane strain case.
g/10.1002/eqe.2201.
[23] Di Laora R, Mandolini A, Mylonakis G. Selection criteria for pile diameter in Journal of the Engineering Mechanics Division ASCE 1978;104(4):953–9.
seismic areas. In: Second international conference on performance-based [41] Roesset JM. The use of simple models in soil–structure interaction. In: ASCE
design in earthquake geotechnical engineering. ID:10.15, 2012. specialty conference, civil engineering and nuclear power; 1980.
[24] Pender M. Seismic pile foundation design analysis. Bulletin of the New [42] Hetenyi M. Beams on elastic foundations. Ann Arbor, MI: University of
Zealand National Society for Earthquake Engineering 1993;26(1):49–160. Michigan Press; 1946.