Flexible Pile Group Interaction Factors Under

Journal of
Marine Science
and Engineering
Article
Flexible Pile Group Interaction Factors under
Arbitrary Lateral Loading in Sand
Miloš Marjanović 1, * , Mirjana Vukićević 1 and Diethard König 2
1 Faculty of Civil Engineering, University of Belgrade, Bulevar kralja Aleksandra 73, 11000 Belgrade, Serbia;
mirav@grf.bg.ac.rs
2 Chair of Soil Mechanics, Foundation Engineering and Environmental Geotechnics,
Ruhr-Universität Bochum, Universitätsstraße 150, 44780 Bochum, Germany; diethard.koenig@rub.de
* Correspondence: mimarjanovic@grf.bg.ac.rs

Received: 6 August 2020; Accepted: 1 October 2020; Published: 15 October 2020
Abstract: Marine and harbor structures, wind turbines, bridges, offshore platforms, industrial
chimneys, retaining structures etc. can be subjected to significant lateral loads from various sources.
Appropriate assessment of the foundations capacity of these structures is thus necessary, especially
when these structures are supported by pile groups. The pile group interaction effects under lateral
loading have been investigated intensively in past decades, and the most of the conducted studies
have considered lateral loading that acts along one of the two orthogonal directions, parallel to the
edge of pile group. However, because of the stochastic nature of its source, the horizontal loading
on the pile group may have arbitrary direction. The number of studies dealing with the pile groups
under arbitrary loading is very limited. The aim of this paper is to investigate the influence of the
arbitrary lateral loading on the pile group response, in order to improve (extend) the current design
approach for laterally loaded pile groups. Free head, flexible bored piles in sand were analyzed
through the extensive numerical study. The main hypothesis of the research is that some critical
pile group configurations, loading directions, and soil conditions exist, which can lead to the unsafe
structural design. Critical pile positions inside the commonly used pile group configurations are
identified with respect to loading directions. The influence of different soil conditions was discussed.
Keywords: pile group interaction; lateral loading; PLAXIS 3D; Hardening Soil model; python
1. Introduction
In situations where surface soil layers have low bearing capacity, pile foundations are the common
foundation solution. Pile foundations are usually designed as closely spaced, square, or rectangular pile
groups. Beside the primary function to transfer the vertical loads to the stiff soil layers, pile foundations
can be significantly loaded with horizontal (lateral) loads. These loads may originate from different
sources, such as: wave, current and ice action, ship impacts, wind pressure, earthquake, earth pressures,
traffic acceleration, braking forces, soil displacements etc. The examples of laterally loaded engineering
structures are: marine and harbor structures, wind turbines, bridges, offshore platforms, industrial
chimneys, retaining structures, etc. The magnitude of lateral load is usually 10–15% of vertical load
(and up to 30% in offshore structures) [1]. This fact makes the problem of the laterally loaded pile
groups very challenging in marine engineering applications.
When the pile group is laterally loaded, stress–strain fields of adjacent piles overlap, followed
with the separation (“gapping”) between the piles and the soil at the back of the piles. The influence
of leading (front) pile row on lateral response of trailing (rear) pile rows is called “shadowing”
because the rear rows are in the shadow of the front pile row (Figure 1a). The problem of interaction
between the piles inside the closely spaced laterally loaded pile group has been recognized by the
J. Mar. Sci. Eng. 2020, 8, 800; doi:10.3390/jmse8100800 www.mdpi.com/journal/jmse

J. Mar. Sci.Sci.
J. Mar. Eng. 2020,8,8,800
2020,
Eng. x FOR PEER REVIEW 2 of 21
2 of 21
between the piles inside the closely spaced laterally loaded pile group has been recognized by the
scientific
scientificcommunity.
community. Ideally, thebearing
Ideally, the bearingcapacity
capacity of the
of the pilepile
group group
shallshall be equal
be equal to the to
sumtheofsum
the of
thebearing
bearing capacities of the individual piles. However, due to the soil–structure interaction
capacities of the individual piles. However, due to the soil–structure interaction effects, load- effects,
load-displacement behavior
displacement behavior of ainside
of a pile pile inside the group
the group is different
is different from thefrom the behavior
behavior of the equivalent
of the equivalent single
single pile (Figure 1b). In addition, the maximum pile bending moment in the pile
pile (Figure 1b). In addition, the maximum pile bending moment in the pile group will be larger than group will be
larger than for an equivalent single pile. When the pile spacing is increased, the effects
for an equivalent single pile. When the pile spacing is increased, the effects of overlapping between of overlapping
between the resisting
the resisting zones become
zones become less significant.
less significant.
(a) (b)
Figure1. 1.(a)(a)Front
Frontand
andtrailing
trailingrows
rows in
in laterally
laterally loaded
loadedpile
pilegroup
group(H(H
G—total lateral force acting on
Figure G —total lateral force acting on
pile group, sx and sy—center-to-center spacing between the piles in X and Y directions, respectively);
pile group, sx and sy —center-to-center spacing between the piles in X and Y directions, respectively);
(b) Influence of the “shadowing” on the load distribution inside the pile group (Hi—lateral force
(b) Influence of the “shadowing” on the load distribution inside the pile group (Hi —lateral force acting
acting on pile i).
on pile i).
Engineering problems that can arise due to the inappropriate assessment of the pile lateral load
Engineering problems that can arise due to the inappropriate assessment of the pile lateral load
can be very serious. Failure in such cases can lead to significant damage or even collapse of the entire
can be very serious. Failure in such cases can lead to significant damage or even collapse of the entire
superstructure. In the cases of bridges or other structures supported by pile foundations, only a few
superstructure.
centimeters of In the cases
lateral of bridgescan
displacements or cause
other the
structures supported
significant by pile foundations,
stress development [2]. Accordingonly toa few
centimeters
Eurocode 7 provisions [3], lateral pile displacements are usually limited to 3% of pile diameter, or to to
of lateral displacements can cause the significant stress development [2]. According
Eurocode
max. 2 cm.7 provisions
Despite the [3],fact
lateral
thatpile
the displacements
current design are usually
practice for limited to 3% of pile
pile foundations diameter,
is based on theor to
max. 2 cm. Despite
capacity-based design,the many
fact that the current
authors [2,4–9] design
pointedpractice
out thatfor thepile foundations
opposite, is based on the
displacement-based
capacity-based
design is moredesign, many authors [2,4–9] pointed out that the opposite, displacement-based design
appropriate.
is moreTheappropriate.
pile group interaction effects under lateral loading have been investigated intensively in
past decades,
The both numerically
pile group interaction and experimentally.
effects under lateral Many numerical
loading have solutions with different
been investigated level of in
intensively
complexity
past decades, have
both been proposed:
numerically andclosed form and empirical
experimentally. solutions solutions
Many numerical [10–12], limit
withequilibrium
different level
of methods
complexity [13,14],
havestrain
beenwedge
proposed:method [15–19],
closed form andandp-yempirical
curve method [20–26].
solutions The development
[10–12], limit equilibriumof
fast computers led to rapid development of continuum-based methods,
methods [13,14], strain wedge method [15–19], and p-y curve method [20–26]. The development which allow for the full
of discretization
fast computers of both
led to soil and the
rapid structure, combined
development with the use of
of continuum-based advancedwhich
methods, constitutive
allow models.
for the full
Continuum-based representation of both the soil and the piles can provide
discretization of both soil and the structure, combined with the use of advanced constitutive a more realistic analysis,
models.
with the use of material properties with a clear physical meaning. On the contrary, continuum-based
Continuum-based representation of both the soil and the piles can provide a more realistic analysis,
methods require more engineering effort for the numerical model preparation, as well as the higher
with the use of material properties with a clear physical meaning. On the contrary, continuum-based
computational hardware requirements. Continuum-based studies on the laterally loaded pile groups
methods require more engineering effort for the numerical model preparation, as well as the higher
are mainly based on the following numerical methods: finite element method (FEM) [27–31], finite
computational hardware
difference method (FDM) requirements.
[2,8,32–34] and Continuum-based
boundary element studies
method on the
(BEM) laterally loaded pile groups are
[35–37].
mainlyBeside
based on the following numerical methods: finite element method (FEM)
numerical studies, the pile group behavior under lateral loading has been investigated [27–31], finite difference
method (FDM) [2,8,32–34] and boundary element method (BEM) [35–37].
using different experimental methods. Full scale tests can eliminate most of the uncertainties of the
Beside numerical
considered studies, the
problem. However, duepile group
to the high behavior
costs, onlyunder
a limitedlateral
numberloading
of fullhas been
scale pileinvestigated
load test
using different
results experimental
are available methods.
in the literature. AsFull scale tests experiments
an alternative, can eliminate most
with of scale
small the uncertainties
models are also of the
considered
reported. The conducted experimental studies of the laterally loaded pile groups are summarized in test
problem. However, due to the high costs, only a limited number of full scale pile load
several
results arestudies
available(e.g.,
inFayyazi et al. [38],
the literature. As Lemnitzer et al. experiments
an alternative, [39], Ashour and withArdalan [40]).models
small scale The results
are also
are mainly
reported. Thepresented
conductedasexperimental
the p-multipliers for of
studies various pile groups.
the laterally loaded Recently,
pile groups improvement
are summarized of CT in
(computer
several tomography)
studies (e.g., Fayyazi and etPIV
al.(particle image velocimetry)
[38], Lemnitzer technologies
et al. [39], Ashour led to the[40]).
and Ardalan experimental
The results
studies [41–45] of soil deformation patterns under lateral loading,
are mainly presented as the p-multipliers for various pile groups. Recently, improvement with the main conclusions of CT
(computer tomography) and PIV (particle image velocimetry) technologies led to the experimental
studies [41–45] of soil deformation patterns under lateral loading, with the main conclusions regarding
the three-dimensional (almost conical) shape of the soil deformation zone, as well as the progressive
soil failure.
regarding the three-dimensional (almost conical) shape of the soil deformation zone, as well as the
J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 3 of 21
progressive soil failure.
The magnitude
regarding of the pile group
the three-dimensional interaction
(almost can be
conical) shape described
of the by comparing
soil deformation zone, asthewell
pileasgroup
the
response with the response
progressive soil failure. of equivalent single pile for the same mean loading. One of the possible
representations
J. Mar.The
Sci. Eng. is the
2020,
magnitude concept
8, 800 of pile
of the pile interaction
group factors,
interaction can which is still in
be described byuse in everyday
comparing engineering
the pile 3 of 21
group
practice
response in Germany, throughofthe
with the response recommendations
equivalent single pile forof the
DIN/EN
same mean codesloading.
[46,47] One
and ofworking group
the possible
“piles” (EAP)
representations
The [48].isThe
magnitude the ofpile
theinteraction
concept of pile
pile factor αi factors,
interaction
group for can
interaction the bepiledescribed
which i inside the
is still in pile
byuse in group
everyday
comparing is
thedefined using
engineering
pile group
theresponse
following
practice in expression:
Germany, through the recommendations of DIN/EN codes [46,47]
with the response of equivalent single pile for the same mean loading. One of the possibleand working group
representations[48].
“piles” (EAP) The
is the pile interaction
concept factor αi factors,
of pile interaction for the which
Hi
pile i inside
is still the pileingroup
in use is defined
everyday using
engineering
the following expression:
practice in Germany, through the recommendations α = of ,DIN/EN codes [46,47] and working group(1)
i
“piles” (EAP) [48]. The pile interaction factor αi for the
H
HSPpile i inside the pile group is defined using the
following expression: α i = i
, (1)
where Hi is the lateral force on pile i and HSP is the H
lateral
H SP
force for on equivalent single pile, for the
same displacement level. According to the studies αi = byi Klüber
, (1)
[49] and Kotthaus [50], pile interaction
H
where Hi is the lateral force on pile i and HSP is the lateral
SP force for on equivalent single pile, for the
factors are mainly dependent on both the pile spacing and pile position inside the pile group.
same displacement
where level.
H is the lateral According
force to the
on pile i and HSPstudies
is theby Klüber
lateral [49]
force forand
on Kotthaus
equivalent [50], pilepile,
single interaction
for the
Interactioni factors αi are calculated using the partial interaction factors αL (in the loading direction),
factorsdisplacement
same are mainly level.
dependent on both
According thestudies
to the pile spacing
by Klüberand[49]
pile
andposition
Kotthausinside
[50], the
pilepile group.
interaction
αQA (perpendicular to loading direction, outer piles) and αQZ (perpendicular to loading direction,
Interaction
factors are factors
mainlyαdependent
i are calculated usingthe
on both the pile
partial interaction
spacing factors
and pile αL (in inside
position the loading direction),
the pile group.
inner piles), using following equations:
Interaction factors αto
αQA (perpendicular i loading
are direction,
calculated using outer
the piles)
partial and α
interaction
QZ factors α
(perpendicular
L (in to
the loading
loading direction,
direction),
α
inner piles), using following
QA (perpendicular equations:
to loading direction, outer αi =piles)
αL · αQA , αQZ (perpendicular to loading direction,(2)
and
inner piles), using following equations:
αi = αL · αQA, (2)
αi α=i =αLα·Lα· αQZ,.
QA (2)(3)
αi =
αi α=Lα· Lα· αQZ. .
QZ (3)
The determination of pile interaction factors is illustrated in Figure 2.
The
The determination
determination of
of pile
pile interaction
interaction factors
factors is
is illustrated
illustrated in
in Figure
Figure 2.
2.
Figure 2. Determination of pile interaction factors based on the pile position inside the pile group
Figure
(after 2. Determination
Kotthaus [50]). sL andof pile interaction
the pilefactors based on the pile position
in the inside the of
pile group
Figure 2. Determination ofsQpile
denote
interaction center-to-center
factors based on the spacings
pile position direction
inside thegroup
the pile lateral
(after
loading Kotthaus [50]). sL and sQ denote the pile center-to-center spacings in the direction of the lateral
(after H G, and perpendicular
Kotthaus [50]). sL and sQtodenote
the direction
the pile of the lateral loading,
center-to-center spacingsrespectively. Fourofdifferent
in the direction pile
the lateral
loading
types, basedHG, and perpendicular
pile positionto the direction
the pileof the lateral
are loading, respectively. Four different pile
loading HGon the
, and perpendicular toinside
the direction ofgroup,
the lateral distinguished
loading, and associated
respectively. withpile
Four different four
types, based on the pile position inside the pile group, are distinguished and associated with four
different colors.on the pile position inside the pile group, are distinguished and associated with four
types, based
different colors.
different colors.
Proposed values for the partial interaction factors αL, αQA, and αQZ are given in Figure 3. Based
Proposed values for the partial interaction factors αL, αQA, and αQZ are given in Figure 3. Based
Proposed values for
on the recommendation ofthe
EAPpartial
[48],interaction factors αL , α
the pile interaction QA , and
factors αiαare
QZ are given
used for in
theFigure 3. Basedofonthe
assessment
on the recommendation of EAP [48], the pile interaction factors αi are used for the assessment of the
the recommendation
modulus of of
subgrade of EAP
reaction [48], the pile interaction factors α are used for the assessment
group. of the
reactionand
andpile
pilebending
bending moments for each
eachpile
pileinside
insidethe
thepile
pilegroup.
i
modulus subgrade moments for
modulus of subgrade reaction and pile bending moments for each pile inside the pile group.
(a)(a)
pile spacings
pile spacings (b) pile(b)
spacing perpendicular
pile spacing perpendicular
Figure 3. Values of partial interaction factors αL , αQA , and αQZ based on the pile spacings in the loading
direction sL (a) and the pile spacing perpendicular (b) to the loading direction sQ (after Kotthaus [50]).
Note that at the pile spacings sL ≥ 6D and sQ ≥ 3D, no interaction occurs (D—pile diameter).
J. Mar. Sci. Eng. 2020, 8, 800 4 of 21
Total pile group efficiency under lateral loading can be described using the following
non-dimensional parameter (n—total number of piles) GW:
n
P
Hi
HG i=1
GW = = . (4)
n · HSP n · HSP
The pile group efficiency GW is equal to the mean value of all interaction factors αi . Lateral force
distribution between the piles can also be described using the load proportions LPi :
Hi
LPi = . (5)
HG
So far, most of the studies on pile group interaction effects have considered lateral loading acting
along one of the two orthogonal directions, parallel to the edge of pile group (namely X and Y directions,
Figure 1a). However, because of the stochastic nature of its source, the horizontal loading on the
pile group may have arbitrary direction. The number of studies dealing with the pile groups under
arbitrary loading is very limited. Randolph [51] defined the influence of the loading direction via simple
algebraic expressions for two laterally loaded flexible piles, both for elastic homogeneous and Gibson
soil. Ochoa and O’Neill [52], based on the experimental results in submerged sand, have developed the
interaction factors dependent on the angle between the load vector and the line that connects the pile
heads. Fan and Long [53] analyzed the interaction between the piles under various loading directions
and derived the modulus reduction factors to account for the interaction effects at the ultimate limit
state. Su and Yan [54] formulated the multidirectional p-y model for sands that was incorporated into
FEM and validated through the simulations of piles under unidirectional and multidirectional lateral
loading. Mayoral et al. [55] formulated the p-y curves for piles under multidirectional loading in soft
clays. The independent application of the common p-y curves in two orthogonal directions was found
to be impossible. Georgiadis et al. [56] analyzed the influence of the arbitrary lateral loading on the
ultimate resistance of the two rigid piles using FEM.
In their recent paper, Su and Zhou [57] presented the experimental study of 2 × 2 square and
rectangular pile groups under static lateral loading in different directions. The results of this study
have shown that the loading direction has great influence on the response of the pile group, mainly
on the redistribution of the lateral load between the piles, and on the total group bearing capacity.
The pile groups at medium pile spacing were found to be more sensitive to the variation in loading
direction. In addition, some pile load distributions have increased when the loading direction was
different than the usual 0/90◦ case, which leads to the hypothesis that some critical (worst-case) loading
cases exist, which can lead to the minimum foundation capacity. Such findings of this study, as well
as the fact that, up to the authors’ knowledge, arbitrary loading case was studied in a very limited
number of research papers, mainly motivated further research presented in this paper.
Therefore, the main aim of this paper is to investigate the influence of arbitrary lateral loading
on the pile group response, in order to improve (extend) the current design approach for laterally
loaded pile groups. The main hypothesis of the research is that some critical pile group configurations,
loading directions, and soil conditions exist, which can lead to unsafe design. This hypothesis is
analyzed through the broad numerical study, and critical pile positions inside the commonly used pile
group configurations are identified with respect to loading directions. The influence of different soil
conditions, mainly sandy soil, is discussed.
2. Research Methodology and Scope

The methodology used in this study follows the concept of “numerical experiment”, where a
numerical simulation is used to mimic the real experiment. Compared to the real experiments, numerical
simulations allow for the large number of analyses to be done faster and cheaper. Full scale experiments
are expensive and therefore not feasible solution for the larger studies. First, the numerical model of
laterally loaded pile group was generated and validated, and then the broad parametric study was
conducted in order to assess the influence of various parameters on the pile group interaction.
J. Mar. Sci. Eng. 2020, 8, 800 5 of 21
Interaction inside the pile group is influenced by various factors, such as: pile spacing and
arrangement inside the pile group, total number of piles, pile stiffness, length and diameter,
pile installation method, soil conditions, loading level, pile head conditions, pile-soil slippage and
separation, and presence of axial loading. A significant number of research papers [58–65] points out
and ranks the most important factors that influence this problem. “Apriori” sensitivity analysis of
different problem parameters, through the broad literature survey, was done in order to select the most
important parameters for further analysis.
Regarding the scope of the present study, bored piles are analyzed, as more expected in common
engineering practice. These piles can also sustain larger vertical loading, so the horizontal loading
is expected to be high as well. According to Van Impe [66], “bored and CFA piles account for 50%
of the world pile market, while the remaining is mainly covered by driven (42%) and screw piles”.
Regarding the pile length, long, flexible piles are considered, as the more common case [51]. Pile spacing
and configuration are identified as the governing factors for the pile group interaction by different
authors [2,67], while pile length and diameter are identified as the less important [32,51].
In this study, only free head piles were considered. According to Mokwa and Duncan [68],
boundary conditions at the pile head are somewhere in between the fixed and free head pile. The same
authors pointed out that the pure fixed head conditions are hard to achieve in reality, even when the
pile cap is very stiff. Despite the different deflection response of fixed and free head piles, pile load
redistribution can be considered independent on pile head conditions, especially within the working
load conditions. This statement is supported by various authors [21,32,69].
3. Numerical Model
The FEM is today considered as the most reliable and widely used numerical method for engineering
analysis of complex foundation systems. Regarding the laterally loaded pile group, FEM allows for
detailed modeling of all important model components: pile geometry, soil continuity and nonlinear
behavior, and especially the pile–soil interaction through the slippage and gapping. According to Dodds
and Martin [23], “modeling lateral behavior in any way other than with three-dimensional models
using nonlinear soil models and interface elements must constitute a compromise.” Eurocode 7 [3]
strongly suggests the use of soil–structure interaction in the numerical analysis, especially in the case of
complex foundations, such as laterally loaded pile foundations or piled-raft foundations.
Within this study, the response of the pile groups under arbitrary static lateral loading was
computed using FEM code PLAXIS 3D (Anniversary Edition) [70]. PLAXIS 3D is a world-wide
used code for the stress–strain, stability, and groundwater flow analysis in geotechnical engineering,
that supports an easy graphical input of the models with complex geometry, as well as the illustrative
presentation of the results. It also features various constitutive models for soil and rock, with different
levels of complexity.
3.1. Model Components

Numerical prediction of the laterally loaded pile group behavior is a challenging task that requires
realistic modeling of the deformation behavior of the soil, pile, and the contact between the soil and
the pile. As follows, model details for different model components are briefly presented.
3.1.1. Pile
Two pile modeling techniques are mainly used in PLAXIS 3D: full 3D (solid) pile model,
and recently implemented embedded beam model. The comparison of two modeling techniques
was studied by Marjanović et al. [71] on an idealized example of 2 × 2 pile group in loose and
dense sand, with conclusions that, in the case of the weaker pile–soil contact, pile group behavior
cannot be resembled using the embedded beam model in PLAXIS 3D. Therefore, the full 3D pile
model was adopted for this study. Pile volumes are modeled using 10-node tetrahedral elements
(nodes in the corners and the middles of the edges), with three translational degrees of freedom
J. Mar. Sci. Eng. 2020, 8, 800 6 of 21
(ux , uy , uz ) per node. This type of element provides a second-order (quadratic) interpolation of nodal
displacement components.
Pile behavior is assumed as linear elastic. Bowles [72] pointed out the possibility of concrete
cracking as the problem in realistic modeling of the laterally loaded pile. Comodromos et al. [73]
analyzed the influence of concrete cracking on the response of laterally loaded piles, and concluded
that, at the working load levels, fully elastic assumption of the pile behavior will not lead to the
significant change in the interaction effects and stress redistribution. Therefore, the pile constitutive
behavior is defined using the linear elastic model, based on the generalized Hooke’s law.
The extraction of the pile group results from the 3D FEM simulation is not a simple task. This is
mainly related to the determination of pile section forces (shear forces and bending moments).
Theoretically, this task is done by integration of component stresses over the pile cross section.
Despite the fact that the calculation of the section forces from volume elements is implemented in
PLAXIS 3D, this step cannot be done automatically, but requires that the user performs the whole
procedure manually for each pile volume. In order to simplify the calculation of pile section forces
and to automatize the post-processing as well, the “dummy” beams are added into the numerical
model. Such a modeling concept is common when the laterally loaded piles are modeled using the 3D
elements [74,75]. The elastic beam finite elements with very small bending stiffness (106 times smaller
than the pile bending stiffness EI) are inserted along the pile axes. Because the “dummy” beam stiffness
is very small, system stiffness matrix remains almost unchanged. “Dummy” beams are enforced to
deform together with the pile axis, and the pile section forces are then easily computed by simple
multiplication of “dummy” beam section forces with 106 .
3.1.2. Pile–Soil Interface

In reality, the contact between the piles and the surrounding soil is not fully rigid, but weaker,
so there are small relative displacements along the pile–soil interface. In general, the pile–soil interface
strength depends on the type of pile material (wood, steel, concrete, etc.) and pile installation effects
(bored, driven, jacked-in). Because this study considers only the bored piles, installation effects and
loading history were neglected.
The pile–soil contact was modeled using thin 2D interface elements. These elements are different
from the regular finite elements: they have pairs of nodes instead of single nodes, with the distance
between the two nodes of a node pair equal to zero. Each node has three translational degrees of
freedom (ux , uy , uz ). As the result, these elements allow for differential displacements between the
node pairs to simulate both slipping and gapping on the pile–soil contact [70].
The constitutive behavior of the pile–soil interface is defined by the elastic-perfectly plastic
Mohr–Coulomb (MC) model with a non-associated flow rule and zero tension cut-off criterion
(when tension develops, a gap between the pile and the surrounding soil is generated). The shear
strength parameters of the pile–soil interface (cinter , ϕinter ) are related to the strength reduction factor
Rinter , which reduces the shear strength parameters (c0 and ϕ0 ) of the MC model, according to the
following expressions:
cinter = Rinter c0 , (6)
tan ϕinter = Rinter tan ϕ0 . (7)
In general, there is lack of experimental data for real pile–soil interface parameters. Suitable
values of Rinter are recommended in the literature [70] for different soil types, and the usual value is
around 0.5.
3.1.3. Soil
Soil constitutive behavior is modeled using the Hardening Soil (HS) model (Schanz and
Vermeer [76]). As in the case of the MC model, the limit stress states of the HS model are defined
using the parameters for the MC failure criterion (cohesion c0 , angle of internal friction ϕ0 , and dilation
J. Mar. Sci. Eng. 2020, 8, 800 7 of 21
angle ψ). However, the soil stiffness is described more accurately, using advanced stiffness parameters:
the triaxial loading stiffness E50 , the triaxial unloading stiffness Eur , and the oedometer loading stiffness
Eoed . These stiffnesses are not constants, but are dependent on the loading level (principal stress state).
The magnitude of the stress level dependency is governed using the parameter m. The values of these
three stiffness parameters are not mutually independent—usually the following relationships are used:
E50 = Eoed , (8)
Eur = 3E50 . (9)
Opposite to the MC model, the yield surface of the HS model is not fixed in the principal stress
space, but it can expand due to plastic straining. Two types of hardening are included in the model:
shear hardening (due to primary deviatoric loading) and compression hardening (due to oedometer
and isotropic loading).
Due to the fact that the pile group response under lateral loading is mainly governed by the
properties of the top soil layer (near-surface soil conditions) [2], the soil was modeled as homogeneous
and isotropic. Because the static loading is assumed, drained soil conditions are assumed in the model.
The soil volumes are also modeled using 10-node tetrahedral elements.
3.2. Model Geometry, Discretization, and Boundary Conditions

Model side domain limits were chosen far enough from the center of the pile group, in order
to avoid the boundary effects on the model response. Sensitivity analyses were done to determine
the optimum size of the model. Because the model geometry will change for different pile group
configurations, it was adopted that the distance between the model domain sides and the edges of
the pile group remain the same. The spacing between the pile bottom (tip) and the bottom model
boundary was chosen to be relatively small, because the lateral pile group response is mainly governed
by the top soil layer in the case of flexible piles, as well as with the active pile length.
PLAXIS 3D does not allow for very precise control of the model discretization. However, it allows
the user to define “refinement zones” inside the model volume, where the mesh discretization can be
finer. Around the pile group, the rectangular prismatic refinement zone was adopted. Trial analyses
were executed on models with different discretization setups, in order to optimize the accuracy
and computational cost. The most of the model geometry parameters have been designed as the
multiplicators of pile diameter D, so the whole geometry can be parameterized as the function of pile
diameter D.
The displacements on the model boundaries were fully fixed in all directions at the bottom of the
model, while the displacements on the side planes were limited to vertical direction. The final model
geometry and the mesh refinement zone are presented in Figures 4 and 5.
3.3. Calculation Stages

Numerical simulations within the study were done as the staged (phase) analysis, and consisted
of the following stages:
1. Initial (K0 ) stage—initial stress field in the soil is established. Horizontal stress state is calculated
using the Jaky’s formula [77]: K0 = 1 − sin ϕ0 .
2. Construction stage—soil volume is replaced with 3D finite elements that represent the piles
(wished-in-place concept). Interface elements are also activated in this stage
3. Prescribed displacements (4 increments)—the prescribed displacements up to 0.1D are applied at
the pile top, in desired loading direction. The prescribed displacements simulate the displacement
control test under static loading conditions.
3. Prescribed displacements (4 increments)—the prescribed displacements up to 0.1D are applied
3. Prescribed displacements (4 increments)—the prescribed displacements up to 0.1D are applied
J. Mar.atSci.the
Eng.pile 8, 800 in desired loading direction. The prescribed displacements simulate8 of
2020, top, the21
at the pile top, in desired loading direction. The prescribed displacements simulate the
displacement control test under static loading conditions.
displacement control test under static loading conditions.
Figure 4. Sketch of the numerical model with parameterized geometry. N, M denote the number of
Figure
Figure4.4.Sketch
Sketchofofthe
thenumerical
numericalmodel
modelwith
withparameterized
parameterizedgeometry.
geometry.N,N,MMdenote
denotethethenumber
numberofof
piles in X and Y direction, respectively. SX and SY are center-to-center pile spacings, in diameters D.
piles
pilesininXXand
andYYdirection,
direction,respectively.
respectively.SX
SXand
andSY
SYare
arecenter-to-center
center-to-centerpile
pilespacings,
spacings,in
indiameters
diametersD.
D.
Figure 5. Numerical model components in the mesh refinement zone.

Figure
Figure5.5.Numerical
Numericalmodel
modelcomponents
componentsininthe
themesh
meshrefinement
refinementzone.
zone.
3.4. Post-Processing of the Results

Beside the complexity of described numerical model and numerical simulation itself,
data management after calculations was an equally challenging task, because all raw data had
3.4. Post-Processing of the Results

3.4. Post-Processing
J. Mar. Sci. Eng. 2020, 8,of the Results
800 9 of 21
Beside the complexity of described numerical model and numerical simulation itself, data
Beside the complexity of described numerical model and numerical simulation itself, data
management after calculations was an equally challenging task, because all raw data had to be
management
to be reordered after calculations
to interpret was group
the pile an equally
results.challenging
PLAXIS 3D task, because
allows all raw data
the extraction of allhad to be
results in
reordered to interpret the pile group results. PLAXIS 3D allows the extraction of all results in
reordered
Euclidean to XYinterpret the pile for
space. However, group results. PLAXIS
the arbitrary 3D allows
loading case, the extraction
the resultant maximum of displacement,
all results in
Euclidean XY space. However, for the arbitrary loading case, the resultant maximum displacement,
Euclidean
shear forces, XYandspace. However,
bending for the
moments arbitrary
must loading case,
be recalculated, thesimple
using resultant maximum
vector algebra displacement,
(Figure 6):
shear forces, and bending moments must be recalculated, using simple vector algebra (Figure 6):
shear forces, and bending moments must be recalculated, using simple vector algebra (Figure 6):
Figure6.6.Directions
Figure Directionsofofthe
themaximum
maximumdisplacements
displacementsy ymax , ,shear
shearforces
forcesHH , ,and
max andbending
bendingmoments
moments
Figure 6. Directions of the maximum displacements ymax max
max, shear forces Hmax, and bending moments
Mmax . Loading direction is governed by the loading direction angle β. X, Y, Q12, Q13, M2, and
M max . Loading direction is governed by the loading direction angle β. X, Y, Q12, Q13, M2, andM3
M3
Mmax . Loading direction is governed by the loading direction angle β. X, Y, Q12, Q13, M2, and M3
denotes
denotesthe the directions
thedirections
directionsof of displacements,
of displacements, shear
displacements,shear forces,
shearforces,
forces,andand
andbendingbending moments
bendingmoments
momentsin in
inXYXY space
XY space
space inin PLAXIS
in PLAXIS
PLAXIS
denotes
3D,respectively.
3D, respectively.
3D, respectively.
Themost
The mostimportant
importantstep
stepinin the
the post-processing
post-processing of of
thethe pile
pile group
group results
results is the
is the precise
precise evaluation
evaluation of
The most important step in the post-processing of the pile group results is the precise evaluation
of pile
the the pile
shearshear forces.
forces. However,
However, the forces
the shear shear extracted
forces extracted
directlydirectly
from thefrom the “dummy”
“dummy” beams in beams
PLAXISin
of the pile shear forces. However, the shear forces extracted directly from the “dummy” beams in
PLAXIS
3D 3D are
are slightly slightly unrealistic,
unrealistic, which is the which
issueisrecognized
the issue recognized
and analyzed andbyanalyzed
Tedesco by Tedesco
[74], [74], who
who concluded
PLAXIS 3D are slightly unrealistic, which is the issue recognized and analyzed by Tedesco [74], who
concluded
that that such
such behavior behavior
could could bewith
be associated associated with the
the PLAXIS 3DPLAXIS 3D beamthat
beam elements, elements,
compute thatthe
compute
shear
concluded that such behavior could be associated with the PLAXIS 3D beam elements, that compute
the shear
forces usingforces using moment
the bending the bending moment
derivative alongderivative along the
the pile length. pile length.
In order In order
to properly to properly
evaluate the pile
the shear forces using the bending moment derivative along the pile length. In order to properly
evaluate
shear thefirst,
forces, pile shear forces,
the pile bendingfirst,moments
the pile bending moments were
were approximated approximated
using the B-splineusing the B-spline
approximations,
evaluate the pile shear forces, first, the pile bending moments were approximated using the B-spline
approximations,
with with
10 interior knots 10 5th
and interior
orderknots
splineand 5th order spline
interpolation. interpolation.
Other fitting techniques,Othersuchfitting techniques,
as polynomials
approximations, with 10 interior knots and 5th order spline interpolation. Other fitting techniques,
such
and as polynomials
cubic splines, were and cubic splines,
also considered. were
Then, thealso
shearconsidered.
forces profileThen,
was the shear forces
computed profile was
by differentiation
such as polynomials and cubic splines, were also considered. Then, the shear forces profile was
ofcomputed
the fittedby differentiation
pile bending moments of the fitted pile bending
line along the pile,moments
and thenlinethe along the pile,
pile shear force and
at then the pile
the ground
computed by differentiation of the fitted pile bending moments line along the pile, and then the pile
shearwas
level force
usedat for
the the
ground level was
calculation used
of the for the calculation
interaction factors. The ofshape
the interaction factors. The
of the characteristic shape of
diagrams
shear force at the ground level was used for the calculation of the interaction factors. The shape of
the characteristic diagrams obtained in the
obtained in the numerical analysis is given in Figure 7. numerical analysis is given in Figure 7.
the characteristic diagrams obtained in the numerical analysis is given in Figure 7.
Figure 7. Shape of the characteristic diagrams for a free head pile. Q12, Q13, M2, and M3 denotes the
Figure 7.
7. Shape
Shapeof the
thecharacteristic
ofbending
characteristic diagrams for aafree head pile. Q12,
Q12,Q13,
Q13,M2,
M2,and
andM3
M3denotes
denotesthe
Figure
shear forces and momentsdiagrams forin
in XY space free head3D,
PLAXIS pile.
respectively. the
shear forces and bending moments in XY space in PLAXIS 3D, respectively.
shear forces and bending moments in XY space in PLAXIS 3D, respectively.
3.5. Automatization of the Calculation Procedures
3.5.
3.5. Automatization
Automatizationofofthe
theCalculation
CalculationProcedures
Procedures
Parametric study done within this research includes multiple FEM simulations, with different
Parametric
Parametricstudy
study done
done within
within this
this research
research includes
includes multiple
multiple FEM
FEM simulations,
simulations, with
with different
different
combinations of input data. In such case, one of the critical “bottlenecks” are the repetitive pre-/post-
combinations of input data. In such case, one of the critical “bottlenecks” are the
combinations of input data. In such case, one of the critical “bottlenecks” are the repetitive repetitive pre-/post-
processing activities, and this part of the analysis is usually prone to errors. Despite the fact that the
processing activities, and
pre-/post-processing this part
activities, ofthis
and the part
analysis is usually
of the analysisprone to errors.
is usually proneDespite the fact
to errors. that the
Despite the
PLAXIS 3D is very user friendly, the preparation and (especially) the post-processing of multiple
PLAXIS
fact that 3D
the is very user
PLAXIS 3D isfriendly,
very userthefriendly,
preparation and (especially)
the preparation the post-processing
and (especially) of multiple
the post-processing of
numerical models of the pile group is time-consuming. Scripting is a popular approach for the
numerical models ofmodels
multiple numerical the pile group
of the pileisgroup
time-consuming. Scripting
is time-consuming. is a popular
Scripting approach
is a popular for the
approach for
automatization of numerical analysis, and such features are nowadays implemented in PLAXIS 3D
automatization
the automatizationof numerical
of numericalanalysis, andand
analysis, such
suchfeatures
featuresareare
nowadays
nowadaysimplemented
implementedininPLAXIS
PLAXIS3D 3D
and many other FEM packages, mainly using the Python programming language. PLAXIS 3D input
and
and many
many other
other FEM
FEM packages,
packages, mainly
mainly using
using the
the Python
Python programming
programming language.
language. PLAXIS
PLAXIS 3D3D input
input
J. Mar. Sci. Eng. 2020, 8, 800 10 of 21
and output modules can be fully controlled through the user defined Python scripts, instead the GUI
(graphical user interface).
In order to perform the most of the calculation procedure in this study automatically, the broad
development and implementation of original Python scripts was done throughout every numerical
analysis step. In this manner, the PLAXIS 3D becomes a “problem calculator”, while the simulation
parameters and data analysis are handled by the separate computer programs. Such approach cannot
increase the single simulation calculation time, neither prevent the conceptual errors in the numerical
model, but data handling in all simulation stages becomes substantially faster.
Beside scripting, PLAXIS 3D allows for the use of multiple computers through the local area
network. Due to availability of multiple licenses of PLAXIS 3D, the set of computer programs for
multiple simulation running on different computers was developed. The parametric study was
then done on (in average) 4–7 computers at the same time, depending on the current availability.
The program was tested on up to 30 computers, and the only limit is the number of available PLAXIS
3D licenses. It is important to note that single computer calculates only one simulation at a time, but as
soon the simulation is finished, the next simulation is started automatically. This process lasts as long
as the simulation queue contains at least one simulation, and the total computation time is decreased
with the number of used computers, assuming the same simulation complexity and the same computer
hardware. This original solution has provided the optimum use of relatively expensive resources (code
licenses and computer power).
3.6. Model Validation

Numerical model was validated based on the back-calculation of the small-scale pile group test
from the literature. Trial and error analysis was used to match the experimental results with the
model response.
3.6.1. Experimental Results

A well-documented centrifuge test at 50 g, done by Kotthaus [50], was used for model validation.
Both single pile and the three pile rows at 3D spacing were back-calculated. The experimental setup
was prepared by the sand rainfall method, to achieve the desired relative sand density. Dry sand was
used for the centrifuge test. Experimental piles were fabricated as hollow aluminum tubes. The outside
pile diameter was 30 mm, and the pipe wall thickness was 2 mm. Model dimensions were chosen to
mimic the real structure, made of solid concrete piles. The model and prototype dimensions are given
in Table 1. Prototype model dimensions are determined according to scaling laws [78].
Table 1. Centrifuge model dimensions (after Kotthaus [50]). Prototype model mimic the real concrete
pile, with the concrete class between C20/25 and C25/30.
Centrifuge Model Prototype Real Structure

Pile diameter D (m) 0.030 1.5 1.5
Pile embedment length L (m) 0.600 30 30
Pile length above ground level L2 (m) 0.085 4.25 4.25
Pipe wall thickness t (m) 0.002 0.10
Pipe inner diameter d (m) 0.026 1.30
Young’s modulus E (MN/m2 ) 70 000 70,000 30,508
Moment of inertia I (m4 ) 1.73 × 10−8 0.108 0.2485
Bending stiffness EI (kNm2 ) 1.2130 7,581,448 7,581,448
Free head pile top conditions were provided using the screws that allowed for the rotation of
the pile head, but enforced the same pile top displacements (displacements control test). In order
to increase the pile soil friction inside the dense sand, soil particles were glued onto the pile shaft.
The point of load application was above the ground level, because of the experimental setup.
J. Mar. Sci. Eng. 2020, 8, 800 11 of 21
3.6.2. Validation Results

3.6.2.For
Validation Results
the accurate numerical simulation of geotechnical problem, the first step is selection and
proper calibration of the
For the accurate numericalsoil constitutive
simulationmodel. The available
of geotechnical experimental
problem, the firstsoil testing
step results and
is selection for
the dense sand (C U = 2.08, D 10 = 0.12 mm, D 50 = 0.23 mm, e max = 0.914, e min = 0.583) used
proper calibration of the soil constitutive model. The available experimental soil testing results for in the centrifuge
experiment by Kotthaus
the dense sand (CU = 2.08, [50] Dwere used for the calibration of the hardening soil model in PLAXIS 3D.
10 = 0.12 mm, D50 = 0.23 mm, emax = 0.914, emin = 0.583) used in the
The
centrifuge experiment by Kotthaus [50]for
effective angle of internal friction were theused
densefor sand was initially
the calibration assumed
of the as ϕ’
hardening = model
soil 41°, and in
oedometer
PLAXIS 3D.modulus as Eoed
The effective = 28 of
angle MPa, basedfriction
internal on thefor
results of thesand
the dense sandwas laboratory
initiallytests givenasinφ’
assumed [50,79].
= 41◦ ,
However, slight modulus
and oedometer trial and as error adjustments
Eoed = 28 MPa, of initially
based on theassumed
results values had to
of the sand be done in
laboratory order
tests to
given
better match the measured pile group response. This alteration is justified by the
in [50,79]. However, slight trial and error adjustments of initially assumed values had to be done in fact that the stress
states around
order to betterthe
match laterally loaded are
the measured pilenot
group fully matched
response. by alteration
This conventional laboratory
is justified by thetests. Finally
fact that the
adopted constitutive model parameters are given in Table 2.
stress states around the laterally loaded are not fully matched by conventional laboratory tests. Finally
adopted constitutive model parameters are given in Table 2.
Table 2. Hardening soil model parameters after back-calculation of experimental results.
Table 2. Hardening soil model
Modelparameters
Parameter after back-calculationValue
of experimental results.
Volumetric
Model weight
Parameterγ (kN/m3) 16.6
Value
Cohesion c’ (kN/m2) 0
Volumetric weight γ (kN/m3 ) 16.6
Angle Cohesion
of internal friction
c’ (kN/m 2 ) ϕ’ (°) 35
0
AngleDilation angle
of internal ψ (°)φ’ (◦ )
friction 5
35
Triaxial loading angle ψ (E
Dilationstiffness ◦)
50 (kN/m2) 28,800
5
Triaxial loading stiffness E 2) 2 28,800
Oedometer loading stiffnessE 50 (kN/m
oed (kN/m ) 28,800
Oedometer loading stiffnessEoed (kN/m2 )2 28,800
Triaxial unloading stiffness Eur (kN/m ) 115,200 (=4E50)
Triaxial unloading stiffness Eur (kN/m2 ) 115,200 (=4E50 )
Poisson’s
Poisson’s ratio
ratio νν(-)(-) 0.25
0.25
m
m (-)
(-) 0.47
0.47
Rinter
inter (-)
(-) 0.654
0.654
The
The measured
measured vs.vs. computed
computed load-displacement
load-displacementresponse
responseisis given
given in
in Figure
Figure 8.
8. Despite
Despite slight
slight
discrepancies, presented results of the model validation show a satisfactory match with
discrepancies, presented results of the model validation show a satisfactory match with the experimentalthe
experimental
results and theresults
overalland the overall
performance performance
of the of theisnumerical
numerical model consideredmodel is considered
to be acceptably to be
accurate for
acceptably accurate for the parametric study
the parametric study of the considered problem. of the considered problem.
(a) (b)
(c) (d)
Figure
Figure8.8.Measured
Measured(triangles) vs.vs.
(triangles) computed
computed(red(red
line)line)
load-displacement response
load-displacement for (a)
response single
for pile;
(a) single
(b) pile row—front pile; (c) pile row—middle pile; (d) pile row—rear pile. Horizontal
pile; (b) pile row—front pile; (c) pile row—middle pile; (d) pile row—rear pile. Horizontal (lateral) (lateral)
displacements
displacements areare given
given inin prototype
prototype scale.
scale.
J. Mar. Sci. Eng. 2020, 8, 800 12 of 21
4.4.Parametric
ParametricStudy
Study
The scope
The scope of
of this
this study
study was
was designed
designed with
with thethe aim
aim to
to assess
assess the
the most
most anticipated
anticipated practical
practical
situations with the reasonable complexity. The following major problem parameters,
situations with the reasonable complexity. The following major problem parameters, according to according to
“apriori” sensitivity
“apriori” sensitivity analysis,
analysis,were
wereselected
selectedfor
forthis study:
this soil
study: type,
soil pile
type, group
pile configuration,
group andand
configuration, pile
spacing, as well as the loading direction, as the main parameter under investigation
pile spacing, as well as the loading direction, as the main parameter under investigation (Figure 9).(Figure 9).
The selection of study parameters is briefly elaborated in the following subsections.
The selection of study parameters is briefly elaborated in the following subsections.
Figure
Figure 9.
9. Sketch
Sketchof
of the
the parametric
parametric study
study parameters:
parameters: left—side
left—sideview;
view;right—top
right—topview; β—angleof
view;β—angle of
direction of lateral loading.
direction of lateral loading.
In
In all simulations,
simulations,a apile pile Young’s
Young’s modulus
modulus of 30.5
of 30.5 GPa wasGPaadopted
was adopted to simulate
to simulate the mostthe most
common
common
concrete concrete
class used class used in everyday
in everyday engineeringengineering
practice. practice.
Pile lengthPileL length L and diameter
and diameter were kept were kept
constant:
D = 1.0 m, and L = 20.0 m. These values provide flexible pile behavior (L/D = 20). The pile head pile
constant: D = 1.0 m, and L = 20.0 m. These values provide flexible pile behavior (L/D = 20). The was
head was extended
extended above theaboveground thelevel (L2 =
ground level
0.25(L2
m),= to
0.25 m), to
avoid avoid numerical
numerical instabilities
instabilities caused by caused by
the high
the high concentrated
concentrated forces atforces at the
the place place of observation
of observation the horizontal
the horizontal displacement.
displacement.
Two
Two “synthetic”
“synthetic” soilsoil types,
types, loose
loose and
and dense
dense drydry sand,
sand, were
were simulated.
simulated. Brown
Brownet etal.
al.[26]
[26] noted
noted
that
thatthe
theshadowing
shadowingeffecteffectwas
wasmoremorepronounced
pronouncedin insand
sandcompared
comparedwith withstiff
stiffclay.
clay. Constitutive
Constitutivemodel
model
parameters densesand
parameters for dense sandwas wasadopted
adopted to to
be be close
close to parameters
to the the parameters of theofvalidated
the validated numerical
numerical model,
model, andsand
and loose loose sand parameters
parameters were as
were chosen chosen as significantly
significantly different,different,
in order toin compare
order to compare the soil
the soil stiffening
stiffening
effects on effects
the pileon the pile
group group response.
response. The parameters
The parameters of the investigated
of the investigated soils are summarized
soils are summarized in Table 3.
in Table 3.
Table 3. Hardening soil model parameters for the “synthetic soils”, compared with the parameters of
validated numerical model.
Table 3. Hardening soil model parameters for the “synthetic soils”, compared with the parameters of
validated numerical model. Model Validation Parametric Study
Model Parameter
Dense
Model Sand
Validation Loose Sand
Parametric Dense Sand
Study
Model Parameter
Volumetric weight γ (kN/m3 ) Dense
16.6Sand Loose16Sand Dense Sand18
Volumetric
Cohesionweightc’ (kN/m 2)
γ (kN/m 3) 16.6
0 16
0 18 0
Cohesion
Angle of internalc’friction
(kN/mφ’ 2) (◦ ) 350 0
30 0 42
AngleDilation
of internal ψ ( ) ϕ’ (°)
anglefriction◦ 535 0
30 42 12
Triaxial loading stiffness 2 28,800 15,000 30,000
Dilation angle Eψ50(°)(kN/m ) 5 0 12
Oedometer
Triaxial loading
loading stiffnessE
stiffness E (kN/m22))
50 (kN/m
oed 28,800
28,800 15,000
15,000 30,000
30,000
Triaxial unloading stiffness Eur oed(kN/m 2) 115,200 60,000 120,000
Oedometer loading stiffnessE (kN/m 2) 28,800 15,000 30,000
Poisson’s ratio
Triaxial unloading ν (-) Eur (kN/m2)
stiffness 0.25
115,200 0.25
60,000 0.25
120,000
m (-) 0.47 0.50 0.50
Poisson’s ratio ν (-) 0.25 0.25 0.25
Rinter (-) 0.654 0.65 0.65
m (-) 0.47 0.50 0.50
Rinter (-) 0.654 0.65 0.65
J. Mar. Sci. Eng. 2020, 8, 800 13 of 21
Pile groups with different pile arrangements were analyzed, as summarized in Table 4.
The numerical model symmetry could not be applied. In order to effectively analyze and discuss the
interaction effects, as well as to limit the calculation time, it was decided to limit the total number of
piles inside the pile group up to 3 × 3 piles.
Table 4. Summary of analyzed pile group configuration. The loading direction angle was varied in
15◦ increments.
Pile Group Shape N×M Sx (D) Sy (D) Loading Direction β (◦ )

2×2 2 2
Square 0/15/30/45
3×3 3 3
2 3
2 4
2 5
2×2
3 4
3 5
4 5
2 2
2 3
2 4
3 2
3 3
3 4
3 5
Rectangle 2×3 0/15/30/45/60/75/90
4 2
4 3
4 4
4 5
5 3
5 4
5 5
2 3
2 4
2 5
3×3
3 4
3 5
4 5
From the practical standpoint, it is a more common case that the configuration of the pile group
(NxM) varies (square, rectangular, circular, etc.), while the spacing between the piles is kept on the
lowest possible level. If the pile spacing is higher than 5D, the costs of the pile cap will increase to a
very high level [60]. In this study, pile spacing between 2D–5D were investigated.
5. Results and Discussion

The results of the parametric study are presented in the form of pile interaction factors and
maximum bending moments for each pile inside the pile group, with respect to loading direction.
The representative plots are given to illustrate the main results trends. All results are presented for the
displacement levels 0.03D (y/D = 0.03) and 0.10D (y/D = 0.10).
All original computer codes, PLAXIS 3D AE numerical models, raw simulation results and results
databases are available from the first author via e-mail upon request.
5.1. Pile Interaction Factors

Pile interaction factors for different pile group configurations are presented on Figures 10–13.
Full lines denote the results for loose sand, while dashed lines denote the results for dense sand.
5.1. Pile
5.1. Pile Interaction
Interaction Factors
Factors
5.1. Pile
J. Mar. Sci.Interaction
Eng. 2020, 8,Factors
Pile
Pile interaction800
interaction factors for
factors for different
different pile
pile group
group configurations
configurations are
are presented
presented onon Figures 14 of 21
Figures 10–13.
10–13.
Full Pile interaction
Full lines
lines denote the
denote factors for
the results
results forloose
for different
loose pile
sand,
sand, group
while
while configurations
dashed
dashed are
lines denote
lines denote presented
the
the on
results for
results for Figures
dense
dense 10–13.
sand.
sand.
Full lines denote the results for loose sand, while dashed lines denote the results for dense sand.
Figure 10. Pile interaction factors for square

square 2 × 2 pile group (s
(sx == 2D,
2D,sssyyy===2D).
2D). Different colors denote
Figure10.
Figure Pileinteraction
10.Pile interaction factors
factors for
for square 2 ×× 22 pile
pile group
group xx = 2D, 2D).Different
Differentcolors
colorsdenote
denote
piles
Figureinside
10. theinteraction
Pile pile group.factors
Full lines—loose
for square sand;
2 × 2 dashed
pile lines—dense
group (s x = 2D, sy = sand.
2D). Different colors denote
piles
pilesinside
inside the
the pile
pile group.
group. Full
Fulllines—loose
lines—loosesand;
sand;dashed
dashed lines—dense
lines—dense sand. sand.
piles inside the pile group. Full lines—loose sand; dashed lines—dense sand.
Figure 11. Pile interaction factors for rectangular 2 × 2 pile group (sx = 2D, sy = 3D). Different colors
Figure 11.
denote
Figure
denote
pilesPile
11. Pile
piles
interaction
inside the pilefactors
interaction
inside
group. for
the pile factors
Fullrectangular
for
group. Full rectangular 2×
lines—loose × 22 pile
sand;
lines—loose2 sand;
group
dashed
pile (sxx =
(s 2D, ssyy ==
= 2D,
lines—dense
grouplines—dense
dashed
3D). Different
sand.
3D).
sand.
Different colors
colors
denote piles
denote piles inside
inside the
the pile
pile group. Full lines—loose
group. Full lines—loose sand;
sand; dashed
lines—dense sand. sand.
denote
Figure piles inside
12. Pile the pile factors
interaction group. for
Fullrectangular
lines—loose ×sand; dashed lines—dense sand.
Pile
Figure piles
denote 12. interaction
inside the pile factors
group. for lines—loose22 sand;
Fullrectangular × 33 pile
pile grouplines—dense
group
dashed (sxx =
(s = 2D,
2D, ssyy == 2D).
2D). Different
sand. Different colors
colors
denote piles inside the pile group. Full lines—loose sand; dashed lines—dense
denote piles inside the pile group. Full lines—loose sand; dashed lines—dense sand. sand.
Pile interaction factors for the closely-spaced square 2 × 2 pile group (sx = 2D, sy = 2D) are
presented in Figure 10. The significant influence of the displacement level is observed—the interaction
factors are changing in a broader range at the higher displacement level. The influence of the soil
J. Mar. Sci. Eng. 2020, 8, 800 15 of 21
type is also pronounced, but at the smaller level, compared to the pile position inside the pile group.
Similar trends can be seen for the square 3 × 3 pile group interaction factors, given in Figure 13.
The effects of symmetry are clearly observed: the black and green pile have the same interaction factors
for the loading direction angle equal to zero. For the direction angle of 45 degrees, the green and blue
pile have the same interaction factors, due to symmetry. In addition, the well-known behavior of front
and rear piles can be observed: for the loading direction angle equal to zero, black and green (front)
piles carry higher loading than red and blue (rear) piles.
Figure
Figure13.
13.Pile
Pileinteraction
interaction factors
factors for
for square
square 33 ×
× 33 pile
pile group
group (s
(sxx ==3D,
3D,ssyy==3D).
3D).Different
Differentcolors
colorsdenote
denote
piles inside the pile group. Full lines—loose sand; dashed lines—dense
piles inside the pile group. Full lines—loose sand; dashed lines—dense sand. sand.
Pile
Pile interaction
interaction factors
factors for
for the closely-spaced rectangular
square 2 × 22 pile group
× 2 pile x = s2D,
(sx =(s2D,
group y = s y = are
2D) 3D)
presented
are presented in Figure
in Figure 10.11.The
Thesignificant
similar shape influence of the
of the plot displacement
is also observed for level is observed—the
different pile spacing,
interaction
with decreased factors are changing
interaction in a broader
with increased pile range
spacing.at As
theinhigher displacement
the previous level.10),
case (Figure Thetheinfluence of
significant
the soil type
influence of isthe
also pronounced,level
displacement but atand themedium
smaller level, compared
influence of soilto the on
type pilepile
position
groupinside the pile
interaction is
group. Similar trends can be seen for the square 3 × 3 pile group interaction
observed. The behavior of green and blue piles is significantly different, compared to the other two factors, given in Figure
13. TheThe
piles. effects of symmetry
changing are clearly
of interaction observed:
factors thepiles
for these blackisand more green pile have with
pronounced the same interaction
the variation in
factors
loadingfor the loading
direction. direction
Therefore, angle
these piles equal
can beto zero. For the
considered asdirection
sensitive angle of 45indegrees,
to change loadingthe green
direction,
and
which blue
canpile
leadhaveto thethe samedesign
unsafe interaction factors,
in the case due to symmetry.
of unpredictable changeInofaddition, the well-known
loading direction.
behaviorPile of front andfactors
interaction rear piles can closely-spaced
for the be observed: for the loading
rectangular 2× direction x = 2D,
angle(sequal
3 pile group y = 2D)
to szero, black
are
and green (front) piles carry higher loading than red and blue (rear) piles.
presented in Figure 12. The interaction factors in this case follows the same trends as in the previous
case.Pile
Theinteraction
green, purple,factors
andfor the pile
black closely-spaced rectangular
can be considered as more2 × 2sensitive.
pile group (sx = 2D, sy = 3D) are
presented in Figure
Finally, 11. The similar
pile interaction factorsshape of the 3plot
for square × 3ispilealsogroup
observed(sx =for sy = 3D)pile
3D,different arespacing,
presented with
in
decreased interaction with increased pile spacing. As in the previous
Figure 13. The trend lines, in general, follows the same pattern as in the case of the 2 × 2 squarecase (Figure 10), the significant
influence
pile group. of the displacement level and medium influence of soil type on pile group interaction is
observed. The behavior of green and blue piles is significantly different, compared to the other two piles.
5.2. changing
The Maximumof Bending Moments
interaction factors for these piles is more pronounced with the variation in loading
direction. Therefore,
Maximum pilethese piles can
bending be considered
moments as sensitive
for different pile togroupchange in loading direction,
configurations which can
are presented in
lead to the unsafe design in the case of unpredictable change of loading direction.
Figures 14–16. Full lines denote the results for loose sand, while dashed lines denote the results for
dense Pile interaction factors for the closely-spaced rectangular 2 × 3 pile group (sx = 2D, sy = 2D) are
sand.
presented in Figure
As shown 12. The
in Figure 14,interaction
maximum pile factors in thismoments
bending case followsare notthesignificantly
same trendsinfluenced
as in the previous
with the
case. The green, purple, and black pile can be considered as more sensitive.
loading direction. As expected, higher bending moments occur in dense sand (due to the displacement
Finally,
control test pile
andinteraction
larger soilfactors for square
resistance in dense3 × 3sand).
pile group
Beside (sx =the
3D,higher
sy = 3D)absolute
are presented
bending in Figure
moment,13.
The trend lines, in general, follows the same pattern as in the case of the 2 × 2
the displacement level does not affect the distribution of the bending moments between the piles. It can square pile group.
be concluded that the pile group configuration and pile position are the governing factors in this case.
5.2. Maximum Bending Moments
Maximum pile bending moments for different pile group configurations are presented in
Figures 14–16. Full lines denote the results for loose sand, while dashed lines denote the results for
dense sand.
J. Mar. Sci. Eng. 2020, 8, 800 16 of 21
J.J.Mar.
J. Mar.Sci.
Mar. Sci.Eng.
Sci. Eng.2020,
Eng. 2020,8,
2020, 8,8,xxxFOR
FORPEER
FOR PEERREVIEW
PEER REVIEW
REVIEW 16of
16
16 of21
of 21
21
Figure14.
Figure
Figure
Figure 14.Maximum
14.
14. Maximumbending
Maximum
Maximum bendingmoments
bending
bending momentsfor
moments
moments forsquare
for square222×
square ×××222pile
pilegroup
pile group(s
group (s(sxxxx==
==2D,
2D,sssysyyy===
2D,
2D, =2D).
2D).
2D). Different
Different
2D).Different colors
colors
Differentcolors
colors
denote
denote
denote piles
piles
piles inside
inside
inside the
the
the pile
pile
pile group.
group.
group. Full
Full
Full lines—loose
lines—loose
lines—loose sand;
sand;
sand; dashed
dashed
lines—dense
lines—dense
denote piles inside the pile group. Full lines—loose sand; dashed lines—dense sand. sand.
sand.
sand.
Figure 15.
Figure
Figure 15. Maximum
Maximum bending
bending moments
moments forfor rectangular
rectangular 22 ×× 33 pile
pile group
group(s
(s(sxxx===2D,
2D,ssysyy===2D).
2D, 2D).Different
2D). Different
Different
Figure 15.
15. Maximum
Maximum bending
bending moments
moments for
for rectangular
rectangular 22 ×× 33 pile
pile group
group (s x = 2D, sy = 2D). Different
colorsdenote
colors
colors denotepiles
denote pilesinside
piles insidethe
inside thepile
the pilegroup.
pile group.Full
group. Fulllines—loose
Full lines—loosesand;
lines—loose sand;dashed
sand; dashedlines—dense
dashed lines—densesand.
lines—dense sand.
sand.
colors denote piles inside the pile group. Full lines—loose sand; dashed lines—dense sand.
Figure16.
Figure 16.Maximum
Maximumbending
bendingmoments
momentsfor forsquare
square 3×××333pile
pilegroup
group(s(s(sxxx==
==3D,
3D, sy =3D). 3D).Different
Differentcolors
colors
Figure
Figure 16.
16. Maximum
Maximum bending
bending moments
moments for square 333 ×
for square pile
3 pile group
group (s x
3D,
3D, sssyyy===3D).
3D).Different
Different colors
colors
denotepiles
denote
denote pilesinside
piles insidethe
inside thepile
the pilegroup.
pile group.Full
group. Fulllines—loose
Full lines—loosesand;
lines—loose sand;dashed
sand; dashedlines—dense
dashed lines—densesand.
lines—dense sand.
sand.
denote piles inside the pile group. Full lines—loose sand; dashed lines—dense sand.
Asthe
As
As
In shown
shown
shown
casein inFigure
in
of Figure
Figure
the 14,maximum
14,
14, maximum
maximum
rectangular pilegroup
pile
pile
2 × 3 pile bending
bending
bending(sx = moments
moments arenot
are
2D, sy = 2D),
moments are not
not significantly
significantly
significantly
maximum influenced
influenced
influenced
pile bending with
with
with
moments
the loading
the
the
(Figureloading
loading direction.
direction.
direction.
15) are distributedAsmore
As
As expected,
expected,
expected, higher
higher
higher
unequally bending
bending
bending
between moments
moments
the moments
piles, occurwith
occur
occur
compared inin dense
in dense
dense sandpile
sand
sand
the square (due
(due
(due to the
to
to
group the
the
at
displacement
displacement
displacement control
control
control test
test
test and
and
and larger
larger
larger soil
soil
soil resistance
resistance
resistance inin
in dense
dense
dense sand).
sand).
sand). Beside
Beside
Beside the
the
the higher
higher
higher
the same pile spacing. Again, higher bending moments occur in dense sand. In addition, the influence absolute
absolute
absolute bending
bending
bending
moment,
moment,
moment,
of thedisplacement
the
the
the lateral displacement
displacement
loading level
level
level
direction does
does
ondoes notaffect
not
not affectthe
affect
the maximum thedistribution
the distribution
distribution
bending moments ofthe
of
of thebending
the
can bending
bending
be moments
moments
moments
considered between
between
between
as less the
the
the
important,
piles.It
piles.
piles. ItItcan
canbe
can beconcluded
be concludedthat
concluded thatthe
that thepile
the pilegroup
pile groupconfiguration
group configurationand
configuration andpile
and pileposition
pile positionare
position arethe
are thegoverning
the governingfactors
governing factors
factors
in
ininthis
thiscase.
this case.
case.
J. Mar. Sci. Eng. 2020, 8, 800 17 of 21
compared to the pile position inside the pile group. The same pile groups (2 × 3) with different spacing
show similar trends, with less moment differences between the piles at larger pile spacing (as expected).
Maximum pile bending moments for the square 3 × 3 pile group (sx = 3D, sy = 3D) follows the
similar trend as the smaller square pile group, but with significantly higher differences of the maximum
moment between the piles. In general, the maximum bending moment between the piles shows larger
discrepancies with the increased number of the piles in the pile group configuration.
6. Conclusions
A broad numerical experiment has been conducted to examine the influence of the arbitrary lateral
loading direction on the interaction inside the pile group. Free head, flexible bored piles in dry sands
were considered. The FEM 3D numerical model was designed and validated by back-calculation of the
existing experimental results. The capabilities of PLAXIS 3D software to perform multiple numerical
simulations have been tested in an automated scripting environment, using original computer programs.
The common approach to the analysis of the laterally loaded pile group is expanded with the pile
interaction factors for the arbitrary loading direction. The results of the study provide more clear insight
into the pile group behavior under lateral loading. Based on the presented analyses, the following
conclusions can be made:
1. Modeling of the laterally loaded pile group using the full 3D finite element model provides
reasonable results for the case of the bored piles.
2. As expected, the level of interaction between the piles is higher at higher displacement levels.
The influence of the displacement level on the pile group interaction effects should be investigated
in more detail, especially at small displacement levels.
3. The interaction level is decreased with increased pile spacing. By means of quantification,
interaction factors are between 0.6–0.9 for working load levels, and 0.4–0.9 for high loading levels.
4. The pile interaction factors are dependent on soil type, but soil conditions are less important
factor, compared to pile group configuration, displacement level, and pile position inside the
pile group.
5. The influence of the loading direction on the maximum bending moments is relatively small, which
is expected, given the central symmetry of the circular piles. The soil conditions significantly affect
the bending moments. The displacement level does not affect the maximum bending moment
distribution between the piles. Bending moment discrepancies are more pronounced with the
increased number of piles.
6. The force in some individual piles inside the pile group significantly changes with the change of
loading direction. These piles can be considered as sensitive, because the change of the loading
direction can be unpredictable. The critical positions of these piles inside the pile group have
been identified for the considered pile group configurations.
7. The concept of multiple numerical simulations, followed with original computer programs,
could be easily extended to other numerical models and more used in everyday practice. Bearing
in mind that the simulation sets can be run during the night, this approach leads to the optimum
use of hardware and software resources.
The presented conclusions emphasize the importance of the considered problem and can lead to
further research topics in various fields. Due to availability of modern experimental techniques, such as
CT and PIV, the experimental analysis of soil deformation patterns under arbitrary loading would
provide better insight in the problem and lead to more accurate numerical models. The identified
concept of sensitive piles should be investigated in more detail, with the aim to eliminate the sensitive
elements from the foundations design. The influence of the different water level in the soil could also be
investigated in more detail, since this paper considered only dry sand. Finally, the speed of numerical
simulations could be reduced with the alternative modeling approaches—through development of
J. Mar. Sci. Eng. 2020, 8, 800 18 of 21
improved embedded beam model formulations (with advanced interface) in FEM, as well as the more
advanced strain wedge model (with arbitrary wedge orientation).
Author Contributions: Conceptualization, D.K.; methodology, M.M.; software, M.M.; validation, M.M., D.K.,
M.V.; formal analysis, M.M.; funding acquisition, M.V., D.K.; investigation, M.M.; resources, D.K.; data curation,
M.M.; writing—original draft preparation, M.M., M.V., D.K.; writing—review and editing, M.M., M.V., D.K.;
visualization, M.M.; supervision, M.V., D.K.; project administration, M.M., M.V., D.K. All authors have read and
agreed to the published version of the manuscript.
Funding: This research was funded partially by Serbian Government, Ministry of Education, Science and
Technological Development, via Project TR 36046, and partially by German Academic Exchange Service (DAAD),
via project SEEFORM (South Eastern European Graduate School for Master and Ph.D. Formation in Engineering).
First author was supported with a short-term scholarship for international PhD guest researcher by German
Research Foundation (DFG) within the Collaborative Research Center SFB 837 “Interaction Modeling in Mechanized
Tunneling”, subprojects A5 and C2. Network license of the PLAXIS 3D software and significant financial
support were provided by the Chair of Soil Mechanics, Foundation Engineering and Environmental Geotechnics,
Ruhr-Universität Bochum.
Acknowledgments: The authors express their sincerest reverence to Tom Schanz, who passed away during this
research, for his scientific contributions, dedication and leadership. Valuable initial topic discussions with René
Schäfer and technical (IT) assistance by Reinhard Mosinski are acknowledged.
Conflicts of Interest: The authors declare no conflict of interests.
References
1. Rao, S.N.; Ramakrishna, V.; Rao, M.B. Influence of Rigidity on Laterally Loaded Pile Groups in Marine Clay.
J. Geotech. Geoenviron. 1998, 124, 542–549. [CrossRef]
2. Comodromos, E.M.; Pitilakis, K.D. Response evaluation for horizontally loaded fixed-head pile groups using
3-D non-linear analysis. Int. J. Numer Anal. Methods Geomech. 2005, 29, 597–625. [CrossRef]
3. European Committee for Standardization. Eurocode 7: Geotechnical Design—Part. 1: General Rules; European
Committee for Standardization: Bruxelles, Belgium, 2004.
4. Randolph, M. Design methods for pile group and piled rafts. In Proceedings of the 13th International
Conference on Soil Mechanics and Geotechnical Engineering, New Delhi, India, 5–10 January 1994; pp. 61–82.
5. Burland, J.; Broms, B.; De Mello, V.F.B. Behaviour of foundations and structures. In Proceedings of the 9th
International Conference on Soil Mechanics and Geotechnical Engineering, Tokyo, Japan, 10–15 July 1977;
pp. 495–546.
6. Mandolini, A.; Viggiani, C. Settlement of piled foundations. Géotechnique 1997, 47, 791–816. [CrossRef]
7. Poulos, H.G.; Davis, E.H. Pile Foundation Analysis and Design; Wiley: New York, NY, USA, 1980.
8. Papadopoulou, M.C.; Comodromos, E.M. On the response prediction of horizontally loaded fixed-head pile
groups in sands. Comput. Geotech. 2010, 37, 930–941. [CrossRef]
9. Lesny, K. Design of laterally loaded piles-limits of limit state design. In Proceedings of the Geo-Risk 2017:
Reliability-Based Design and Code Developments, Denver, CO, USA, 4–7 June 2017; pp. 267–276.
10. Winkler, E. Die Lehre von der Elasticität und Festigkeit: Mit Besonderer Rücksicht auf ihre Anwendung in der Technik,
für Polytechnische Schulen, Bauakademien, Ingenieure, Maschinenbauer, Architekten, etc.; Dominicius: Prague,
Czech Republic, 1867.
11. Hetenyi, M. Beams on Elastic Foundation: Theory with Applications in the Fields of Civil and Mechanical Engineering;
The University of Michigan Press: Ann Arbor, MI, USA, 1946.
12. Terzaghi, K. Evaluation of coefficients of subgrade reaction. Geotechnique 1955, 5, 297–326. [CrossRef]
13. Broms, B.B. Lateral resistance of piles in cohesive soils. J. Soil Mech. Found. Div. 1964, 90, 27–64.
14. Broms, B.B. Lateral resistance of piles in cohesionless soils. J. Soil Mech. Found. Div. 1964, 90, 123–158.
15. Norris, G. Theoretically based BEF laterally loaded pile analysis. In Proceedings of the 3rd International
Conference on Numerical Methods in Offshore Piling, Nantes, France, 21–22 May 1986; pp. 361–386.
16. Ashour, M.; Pilling, P.; Norris, G. Documentation of the strain wedge model program for analyzing laterally
loaded piles and pile groups. In Proceedings of the 32nd Engineering Geology and Geotechnical Engineering
Symposium, Boise, ID, USA, 26–28 March 1997; pp. 26–28.
17. Ashour, M.; Norris, G.; Pilling, P. Lateral loading of a pile in layered soil using the strain wedge model.
J. Mar. Sci. Eng. 2020, 8, 800 19 of 21
18. Ashour, M.; Pilling, P.; Norris, G. Assessment of pile group response under lateral load. In Proceedings of
the 4th International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil
Dynamics, San Diego, CA, USA, 25–28 June 2001.
19. Ashour, M.; Pilling, P.; Norris, G. Lateral behavior of pile groups in layered soils. J. Geotech. Geoenviron. 2004,
130, 508–592. [CrossRef]
20. McClelland, B.; Focht, J. Soil modulus for laterally loaded piles. J. Soil Mech Found. Div 1958, 82, 1–22.
21. Matlock, H. Correlations for design of laterally loaded piles in soft clay. In Proceedings of the Offshore
Technology Conference, Houston, TX, USA, 22–24 April 1970; pp. 77–94.
22. Reese, L.C.; Cox, W.R.; Koop, F.D. Analysis of laterally loaded piles in sand. In Offshore Technology in Civil
Engineering Hall of Fame Papers from the Early Years; American Society of Civil Engineers: Reston, VA, USA,
1974; pp. 95–105.
23. Dodds, A.M.; Martin, G.R. Modeling Pile Behavior in Large Pile Groups under Lateral Loading. Technical Report
mceer-07-0004; University of Southern California: Los Angeles, CA, USA, 2007.
24. Robertson, P.K.; Davies, M.P.; Campanella, R.G. Design of laterally loaded driven piles using the flat
dilatometer. Geotech. Test. J. 1989, 12, 30–38.
25. Briaud, J.L.; Smith, T.; Meyer, B. Laterally loaded piles and the pressuremeter: Comparison of existing methods.
In Laterally Loaded Deep Foundations: Analysis and Performance; ASTM International: West Conshohocken, PA,
USA, 1984.
26. Brown, D.A.; Morrison, C.; Reese, L.C. Lateral load behavior of pile group in sand. J. Geotech. Eng. ASCE
1988, 114, 1261–1276. [CrossRef]
27. Brown, D.A.; Shie, C.F. Three dimensional finite element model of laterally loaded piles. Comput. Geotech.
1990, 10, 59–79. [CrossRef]
28. Brown, D.A.; Shie, C.F. Some numerical experiments with a three dimensional finite element model of a
laterally loaded pile. Comput. Geotech. 1991, 12, 14–162. [CrossRef]
29. Brown, D.A.; Shie, C.F. Numerical experiments into group effects on the response of piles to lateral loading.
Comput. Geotech. 1990, 10, 211–230. [CrossRef]
30. Wakai, A.; Gose, S.; Ugai, K. 3-D elasto-plastic finite element analyses of pile foundations subjected to lateral
loading. Soils Found. 1999, 39, 97–111. [CrossRef]
31. Yang, Z.; Jeremic, B. Numerical study of group effects for pile groups in sands. Int. J. Numer. Anal. Met. 2003,
27, 1255–1276. [CrossRef]
32. Comodromos, E.; Papadopoulou, M. Response evaluation of horizontally loaded pile groups in clayey soils.
Geotechnique 2012, 62, 329–339. [CrossRef]
33. Papadopoulou, M.C.; Comodromos, E.M. Explicit extension of the p-y method to pile groups in sandy soils.
Acta Geotech 2014, 9, 485–497. [CrossRef]
34. Comodromos, E.M.; Papadopoulou, M.C. Explicit extension of the p-y method to pile groups in cohesive
soils. Comput Geotech 2013, 47, 28–41. [CrossRef]
35. Poulos, H.G. Behavior of laterally loaded piles: I—single piles. J. Soil Mech. Found. Div. 1971, 97, 711–731.
36. Poulos, H.G. Behavior of laterally loaded piles: II—pile groups. J. Soil Mech. Found. Div. 1971, 97, 733–751.
37. Banerjee, P.K.; Driscoll, R.M. Three-dimensional analysis of vertical pile groups. In Proceedings of the
Second International Conference on Numerical Methods in Geomechanics, Blacksburg, VA, USA, 1 June 1976;
pp. 438–450.
38. Fayyazi, M.S.; Taiebat, M.; Finn, W.L. Group reduction factors for analysis of laterally loaded pile groups.
Can. Geotech. J. 2014, 51, 758–769. [CrossRef]
39. Lemnitzer, A.; Khalili-Tehrani, P.; Ahlberg, E.R.; Rha, C.; Taciroglu, E.; Wallace, J.W.; Stewart, J.P. Nonlinear
Efficiency of Bored Pile Group under Lateral Loading. J. Geotech. Geoenviron 2010, 136, 1673–1685. [CrossRef]
40. Ashour, M.; Ardalan, H. Employment of the p-multiplier in pile-group analysis. J. Bridg. Eng 2011, 16,
612–623. [CrossRef]
41. Otani, J.; Pham, K.D.; Sano, J. Investigation of failure patterns in sand due to laterally loaded pile using X-ray
CT. Soils Found. 2006, 46, 529–535. [CrossRef]
42. Hajialilue-Bonab, M.; Azarnya-Shahgoli, H.; Sojoudi, Y. Soil deformation pattern around laterally loaded
piles. Int. J. Phys. Model. Geo. 2011, 11, 116–125. [CrossRef]
43. Hajialilue-Bonab, M.; Sojoudi, Y.; Puppala, A.J. Study of strain wedge parameters for laterally loaded piles.
Int. J. Geomech. 2013, 13, 143–152. [CrossRef]
J. Mar. Sci. Eng. 2020, 8, 800 20 of 21
44. Iai, S.; Tobita, T.; Hussien, M.N.; Ozutsumui, N.; Rollins, K. Soil-pile interaction under lateral load. In
Soil-Foundation-Structure Interaction; CRC Press: Boca Raton, FL, USA, 2010; pp. 117–124.
45. Morita, K.; Otani, J.; Mukunoki, T.; Hironaka, J.; Pham, K.D. Evaluation of vertical and lateral bearing
capacity mechanisms of pile foundations using X-ray CT. In Proceedings of the International Workshop on
Recent Advances of Deep Foundations, Yokosuka, Japan, 1–2 February 2007; pp. 217–223.
46. German Institute for Standardization. Subsoil—Verification of the Safety of Earthworks and Foundations—Supplementary
Rules to DIN EN 1997–1; German Institute for Standardization: Berlin, Germany, 2010.
47. German Institute for Standardization. DIN 4014: Bored Cast-in-Place Piles: Formation, Design and Bearing
Capacity; German Institute for Standardization: Berlin, Germany, 1990.
48. Deutschen Gesellschaft für Geotechnik, e.V. Deutschen Gesellschaft für Geotechnik e.V.: EA-Pfähle: Empfehlungen
des Arbeitskreises “Pfähle”; Wilhelm Ernst & Sohn: Berlin, Germany, 2012.
49. Klüber, E. Tragverhalten von Pfahlgruppen unter Horizontal Belastung. Ph.D. Thesis, Technische Universität
Darmstadt, Darmstadt, Germany, 1988.
50. Kotthaus, M. Zum Tragverhalten von horizontal belasteten Pfahlreihen aus langen Pfählen in Sand.
Ph.D. Thesis, Ruhr-Universität Bochum, Bochum, Germany, 1992.
51. Randolph, M.F. The response of flexible piles to lateral loading. Geotechnique 1981, 31, 247–259. [CrossRef]
52. Ochoa, M.W.; O’Neill, M. Lateral pile interaction factors in submerged sand. J. Geotech. Eng. ASCE 1989, 115,
359–378. [CrossRef]
53. Fan, C.C.; Long, J.H. A modulus-multiplier approach for non-linear analysis of laterally loaded pile groups.
Int. J. Numer Anal. Met. 2007, 31, 1117–1145. [CrossRef]
54. Su, D.; Yan, W. A multidirectional p-y model for lateral sand-pile interactions. Soils Found. 2013, 53, 199–214.
[CrossRef]
55. Mayoral, J.M.; Pestana, J.M.; Seed, R.B. Multi-directional cyclic p-y curves for soft clays. Ocean. Eng. 2016,
115, 1–18. [CrossRef]
56. Georgiadis, K.; Sloan, S.; Lyamin, A. Ultimate lateral pressure of two side-by-side piles in clay. Geotechnique
2013, 63, 733–745. [CrossRef]
57. Su, D.; Zhou, Y.G. Effect of Loading Direction on the Response of Laterally Loaded Pile Groups in Sand.
Int. J. Geomech. 2015, 16, 04015051. [CrossRef]
58. Brown, D.A.; O’Neill, M.W.; Hoit, M.; McVay, M.; El Nagger, M.H.; Chakraborty, S. Static and Dynamic
Lateral Loading of Pile Groups. Technical Report NCHRP 24; Transportation Research Board: Washington, DC,
USA, 2001.
59. Huang, A.B.; Hsueh, C.K.; O’Neill, M.W.; Chern, S.; Chen, C. Effects of construction on laterally loaded pile
groups. J. Geotech. Geoenviron. 2001, 127, 385–397. [CrossRef]
60. McVay, M.; Casper, R.; Shang, T.I. Lateral response of three-row groups in loose to dense sands at 3D and 5D
pile spacing. J. Geotech. Eng. 1995, 121, 436–441. [CrossRef]
61. Rollins, K.M.; Peterson, K.T.; Weaver, T.J. Lateral load behavior of full-scale pile group in clay.
62. Cox, W.R.; Dixon, D.A.; Murphy, B.S. Lateral-load tests on 25.4-mm (1-in.) diameter piles in very soft clay
in side-by-side and in-line groups. In Laterally Loaded Deep Foundations: Analysis and Performance; ASTM
International: West Conshohocken, PA, USA, 1984.
63. Reese, L.C.; Van Impe, W.F. Single Piles and Pile Groups under Lateral Loading; CRC Press: Boca Raton, FL, USA,
2011. Available online: https://books.google.com.hk/books/about/Single_Piles_and_Pile_Groups_Under_
Later.html?id=7qg0SwAACAAJ&redir_esc=y (accessed on 14 October 2020).
64. Chandrasekaran, S.; Boominathan, A.; Dodagoudar, G. Group interaction effects on laterally loaded piles in
clay. J. Geotech. Geoenviron. 2010, 136, 573–582. [CrossRef]
65. McVay, M.; Zhang, L.; Molnit, T.; Lai, P. Centrifuge testing of large laterally loaded pile groups in sands.
66. Van Impe, W. Foreword: Belgian geotechnics’ experts research on screw piles. In Belgian Screw Pile Technology
Design and Recent Developments; A.A. Balkema: Rotterdam, The Netherlands; Momenta: London, UK, 2003.
67. Basile, F. Analysis and design of pile groups. In Numerical Analysis and Modelling in Geomechanics; CRC Press:
Boca Raton, FL, USA, 2003; pp. 278–315.
68. Mokwa, R.L.; Duncan, J.M. Rotational Restraint of Pile Caps during Lateral Loading. J. Geotech. Geoenviron.
2003, 129, 829–837. [CrossRef]
J. Mar. Sci. Eng. 2020, 8, 800 21 of 21
69. Reese, L.C.; Van Impe, W.F. Single Piles and Pile Groups under Lateral Loading; CRC Press: Boca Raton, FL, USA,
2000. Available online: https://books.google.com.hk/books/about/Single_Piles_and_Pile_Groups_Under_
Later.html?id=jEs5z6-2xG4C&redir_esc=y (accessed on 14 October 2020).
70. Brinkgreve, R.B.J.; Kumarswamy, S.; Swolfs, W.M.; Waterman, D.; Chesary, A.; Bonnier, P.G.; Haxaire, A.
PLAXIS 3D AE—User Manual; Plaxis BV: Delft, The Netherlands, 2015.
71. Marjanovic, M.; Vukicevic, M.; König, D.; Schanz, T.; Schäfer, R. Modeling of laterally loaded piles using
embedded beam elements. In Proceedings of the International Conference “Contemporary achievements in
Civil Engineering 2016”, Subotica, Serbia, 22 April 2016; pp. 349–358.
72. Bowles, J. Foundation Analysis and Design; McGraw-Hill: New York, NY, USA, 1996.
73. Comodromos, E.M.; Papadopoulou, M.C.; Rentzeperis, I.K. Effect of cracking on the response of pile test
under horizontal loading. J. Geotech. Geoenviron. 2009, 135, 1275–1284. [CrossRef]
74. Tedesco, G. Offshore Tower or Platform Foundations: Numerical Analysis of a Laterally Loaded Single Pile
or Pile Group in Soft Clay and Analysis of Actions on a Jacket Structure. Master’s Thesis, University of
Bologna, Bologna, Italy, 2013.
75. Zhang, Y.; Andersen, K.H.; Tedesco, G. Ultimate bearing capacity of laterally loaded piles in clay - Some
practical considerations. Marine Struct. 2016, 50, 260–275. [CrossRef]
76. Schanz, T.; Vermeer, P.A. Formulation and verification of the Hardening Soil Model. In Beyond 2000 in
Computation Geotechnics—10 Years of PLAXIS; Balkema: Rotterdam, The Netherlands, 1999; pp. 1–16.
77. Jaky, J. The coefficient of earth pressure at rest. J. Soc. Hung. Arch. Eng. 1944, 78, 355–358.
78. Garnier, J.; Gaudin, C.; Springman, S.M.; Culligan, P.J.; Goodings, D.; König, D.; Kutter, B.; Phillips, R.;
Randolph, M.F.; Thorel, L. Catalogue of scaling laws and similitude questions in geotechnical centrifuge
modelling. Int. J. Phys. Model. Geo. 2007, 7, 1–23. [CrossRef]
79. Kayalar, A. Bettung von Horizontal Belasteten Pfahlreihen und Bohrpfahlwänden. Bachelar‘s Thesis,
Ruhr-Universität Bochum, Bochum, Germany, 2012.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Flexible Pile Group Interaction Factors Under

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flexible Pile Group Interaction Factors Under

Uploaded by

Copyright:

Available Formats

Journal of

J. Mar. Sci. Eng. 2020, 8, 800; doi:10.3390/jmse8100800 www.mdpi.com/journal/jmse

2. Research Methodology and Scope

3.1. Model Components

3.1.2. Pile–Soil Interface

tan ϕinter = Rinter tan ϕ0 . (7)

E50 = Eoed , (8)

Eur = 3E50 . (9)

3.2. Model Geometry, Discretization, and Boundary Conditions

3.3. Calculation Stages

Figure 5. Numerical model components in the mesh refinement zone.

3.4. Post-Processing of the Results

3.4. Post-Processing of the Results

3.6. Model Validation

3.6.1. Experimental Results

Centrifuge Model Prototype Real Structure

3.6.2. Validation Results

Pile Group Shape N×M Sx (D) Sy (D) Loading Direction β (◦ )

5. Results and Discussion

5.1. Pile Interaction Factors

Figure 10. Pile interaction factors for square

You might also like