materials

Article
Analysis Method for Laterally Loaded Pile Groups
Using an Advanced Modeling of Reinforced
Concrete Sections
Stefano Stacul * ID
and Nunziante Squeglia ID

Department of Civil and Industrial Engineering, University of Pisa, Largo Lucio Lazzarino, 56122 (PI) Pisa, Italy;
[email protected]
* Correspondence: [email protected]; Tel.: +39-050-221-7733

Received: 27 December 2017; Accepted: 13 February 2018; Published: 15 February 2018

Abstract: A Boundary Element Method (BEM) approach was developed for the analysis of pile
groups. The proposed method includes: the non-linear behavior of the soil by a hyperbolic modulus
reduction curve; the non-linear response of reinforced concrete pile sections, also taking into account
the influence of tension stiffening; the influence of suction by increasing the stiffness of shallow
portions of soil and modeled using the Modified Kovacs model; pile group shadowing effect, modeled
using an approach similar to that proposed in the Strain Wedge Model for pile groups analyses.
The proposed BEM method saves computational effort compared to more sophisticated codes such
as VERSAT-P3D, PLAXIS 3D and FLAC-3D, and provides reliable results using input data from a
standard site investigation. The reliability of this method was verified by comparing results from data
from full scale and centrifuge tests on single piles and pile groups. A comparison is presented between
measured and computed data on a laterally loaded fixed-head pile group composed by reinforced
concrete bored piles. The results of the proposed method are shown to be in good agreement with
those obtained in situ.

Keywords: laterally loaded piles; pile group; shadowing effect; boundary element method; tension
stiffening; reinforced concrete; suction

1. Introduction
The response to the horizontal loading of pile foundations, starting with the single pile, passing
through the pile groups and finally to the combined piled-rafts has been the focus of many studies.
However, as noted by many authors including Mokwa and Duncan [1] and Katzenbach and
Turek [2], additional tests are needed to better understand the interactions between the soil, piles
and superstructures. For the single pile case, it is well known that the key factors that influence the
response include the restraint condition at the pile-head and the pile-soil relative stiffness.
For laterally loaded pile groups, full scale tests conducted by O’Neill [3] and Huang et al. [4] failed
to provide definitive information about the influence of the execution technique while it is certain that
the additional factors to be considered, compare to the single pile case, are: pile spacing, pile-soil-pile
interactions, the stiffness of the connecting structure.
Generally, it is assumed that in pile-groups, under horizontal loads, the displacement of the
pile-heads is the same for all the piles and, so it is heterogeneous the load distribution between these.
In general, because of group effects and in particular the shadowing effect [5] the efficiency of a
pile-group is less than unity with a decreasing trend as the load increase. The efficiency achieves an
asymptotic value for displacements larger than 0.06 pile diameter D [6]. Group effects such as the
shadowing, which leads to a different load distribution between the piles in a group, tend to disappear
for spacing values larger than 5–7 diameters [7,8].

Materials 2018, 11, 300; doi:10.3390/ma11020300 www.mdpi.com/journal/materials

Materials 2018, 11, 300 2 of 21

The most common analysis methods are continuum-based or Winkler-based approaches

(non-linear transfer curves or p-y curves). p-y curves methods are limited to the use of a subgrade soil
reaction modulus which does not represent an actual property of the soil, and the soil is schematized
with a series of independent springs that require the introduction of specific multipliers less than
unity, [5], by which the transfer curves are scaled to consider the presence of group effects.
Since the introduction of the multipliers-concept a lot of works have been published [6,9,10] with
the aim to determine a so+rt of project curves, for these coefficients, as a function of the spacing and
the pile position in the group, although as evidenced by [11] would be more appropriate to evaluate
the trend of these multipliers as a function of the load level (or the displacement level).
Some of the most common p-y curves (and implemented in software such as LPILE [12]
GROUP [13], FLPIER [14]) include those recommended by the American Petroleum Institute.
These were obtained from experimental tests on steel pipe piles with an outer diameter of about
30 cm, and which are not affected by the non-linearity of the pile material. Even recently new
analytical methods based on the transfer curves have been developed [15,16] and new p-y curves were
experimentally determined [17].
Continuum-based approaches are usually solved with boundary element methods (BEM) and
finite element methods (FEM). Despite their potential in geotechnical engineering applications,
FEM suffer from the complexity of the domain discretization, difficulties in choosing the input
parameters and as evidenced in [18] FEM results are affected by the pile modeling. The high
computational costs also prevent their use in parametric studies.
Often, therefore, they are used as a benchmark to validate other simplified approaches or as a tool
to determine p-y curves for comparison with those obtained experimentally in situ [19,20].
BEM approaches, however, describe the soil as a homogeneous elastic half-space, characterized
by a Young modulus and Poisson ratio, and enable pile-soil-pile interactions to be directly evaluated,
and group effects can be considered.
These methods provide a complete solution at the interfaces of the problem domain but
entail numerical approximations when the analysis involves heterogeneous soils. To evaluate the
displacement induced at one point of the subsoil by a load acting in another point, the elastic Mindlin
solution is generally used [21].
The most important works and parametric studies conducted using BEM approaches have been
carried out by Spillers and Stoll [22], Poulos and Davis [23], Davies and Banerjee [24], Sharnouby and
Novak [25]; and Budhu and Davies [26,27].
The computer codes that make use of this approach include DEFPIG [28], PIGLET [29],
and PGROUPN [30]. Recently some other approaches, which permit analysis of pile groups and
piled-rafts [31–33], have been proposed.
Another interesting method is the Strain Wedge Model proposed by Ashour et al. [34,35] for the
analysis of single piles and pile-groups. This method links the response of the one-dimensional beam
on an elastic foundation with a three-dimensional representation of the pile-soil interaction, and thus
with the development and mobilization of a passive wedge of soil in front of the pile.

2. Proposed ‘BEM-Based’ Method

2.1. Key Features of the Proposed Method

The analysis of pile-groups requires non-linear methods that have the capability to reproduce all
relevant interactions between the foundation elements and the soil.
The originality of the proposed approach lies in its ability to provide a complete BEM analysis of
the soil continuum in which all the interactions are modeled. Compared to FEM or FDM analyses,
BEM provides a complete problem solution in terms of boundary values only, specifically at the
pile-soil interfaces.
Materials 2018, 11, 300 3 of 21

Materials 2018, 8, 300 3 of 23

This leads to a drastic reduction in unknowns to be solved for, thereby resulting in substantial
savings in computing time and data preparation effort. This feature is particularly significant for
The non-linear soil response is modeled by adopting a modified version of the quasi-hyperbolic
three-dimensional problems such as pile-groups.
elastic modulus reduction curve formulation proposed by Fahey and Carter [36]. The proposed
The non-linear soil response is modeled by adopting a modified version of the quasi-hyperbolic
method, relies upon the following assumptions:
elastic modulus reduction curve formulation proposed by Fahey and Carter [36]. The proposed
a) pile-soil, pile-pile interactions are considered using the Mindlin’s solution;
method, relies upon the following assumptions:
b) horizontally layered elastic soil;
c) non-linear
a) behavior
pile-soil, pile-pile for the reinforced
interactions concreteusing
are considered pile section;
the Mindlin’s solution;
d) non-linear soil behavior (incremental
b) horizontally layered elastic soil; analysis);
e) the so-called shadowing effect, has been implemented in the code using an approach similar to
c) non-linear behavior for the reinforced concrete pile section;
that described in [35];
d) non-linear soil behavior (incremental analysis);
f) the ultimate soil pressure profile is evaluated according to the relationship suggested by [37–40].
e) the so-called shadowing effect, has been implemented in the code using an approach similar to
To validate the proposed method with its main assumptions, some well-documented case
that described in [35];
histories have been collected and a prediction exercise has been carried out. The number of piles in
f) the ultimate
the group studiedsoil pressurerather
is generally profilesmall
is evaluated according
(the largest to thestudied
pile-group relationship suggestedbyby15[37–40].
was composed piles).
In allTothe examined
validate cases, a load
the proposed test on
method a single
with pile
its main was also available.
assumptions, some well-documented case histories
have been collected and a prediction exercise has been carried out. The number of piles in the group
2.2. Pile Modelling
studied is generally rather small (the largest pile-group studied was composed by 15 piles). In all the
The proposed
examined method
cases, a load test onwas developed
a single pile wasto also
capture the response of a pile group subjected to
available.
horizontal load. It consists of a BEM-based approach. The analysis is performed using a non-linear
2.2. Pile Modelling
incremental tangent method.
The pile is modeled
The proposed methodas a was
vertical strip, geometrically
developed to capture the defined by theof
response outer diameter
a pile group D and length
subjected to
L of the actual pile, discretized in 60 blocks of variable length with depth. With
horizontal load. It consists of a BEM-based approach. The analysis is performed using a non-linear this discretization, it
is possible totangent
incremental minimize the calculation-time.
method.
The discretization
The pile is modeled as is asafollows
vertical(Figure 1):
strip, geometrically defined by the outer diameter D and length
L of the
20 blocks with a thickness  = D/8, starting
actual pile, discretized in 60 blocks of variable from the ground
length with level upWith
depth. to a this
depth of 2.5D;
discretization, it is

possible10 blocks with a the
to minimize thickness  = D/4, starting from a depth of 2.5D up to a depth of 5D;
calculation-time.
 10 blocks
The with a thickness
discretization  = D/2,
is as follows starting
(Figure 1): from a depth of 5D up to a depth of 10D;
 10 blocks with a thickness  = D, starting from a depth of 10D up to a depth of 20D;
• 20 thickness ∆==(LD/8,
10 blocks with aa thickness starting
− 20D)/10, from the
starting fromground
a depthlevel up toup
of 20D a to
depth of 2.5D;
the pile base depth.
• 10
Theblocks with aofthickness
definition ∆ = D/4, starting
the discretization criterionfrom
wasa depth of 2.5D
suggested by up to a [41]
Landi depth asofresult
5D; of some
•
parametric
10 blocksanalyses.
with a The
thickness ∆ = D/2,
same pile modeling described
starting from a herein
depth of has5Dbeen recently
up to a depthused by the authors
of 10D;
in
• [42]
10to studywith
blocks the lateral
a thickness ∆ = D,
response of starting
the single pile.a To
from facilitate
depth of 10Dtheupreader‘s overall
to a depth understanding
of 20D;
of
• the10 manuscript,
blocks with athe authors
thickness ∆ =reported here some
(L − 20D)/10, details
starting from of the pile
a depth modeling
of 20D up to thethat
pilecan
basebedepth.
also
retrieved from [42].

Figure 1. Pile
Figure 1. Pile discretization.
discretization.

The pile flexibility matrix, in case of linear elastic behavior, is obtained using the elastic beam
theory, and each coefficient of this matrix can be expressed using Equation (1) (Figure 2).
Materials 2018, 11, 300 4 of 21

The definition of the discretization criterion was suggested by Landi [41] as result of some
parametric analyses. The same pile modeling described herein has been recently used by the authors
in [42] to study the lateral response of the single pile. To facilitate the reader‘s overall understanding of
Materials
the 2018, 8, 300the authors reported here some details of the pile modeling that can be also retrieved
manuscript, 4 of 23

from [42].
The pile flexibility matrix, in case
j  zi
z i2 z elastic
z i3 of linear  
behavior, is obtained using the elastic beam
theory, and each coefficient of  matrix 
aij this can be expressed z i  Equation
if using zj (1) (Figure 2).
3E p I p 2E p I p
z3i z2i (z j −zi ) (1)
z z  z 
aij 3= +
3E p I p 2 2E p I p if zi < z j
zj z3j j zi 2j (zi −zj j )
(1)
aij  aij = 3E I + 2E p I p i 
ififzi z≥ zj z j
3E p I p p p 2 E p I p

Figure
Figure 2.
2. Pile
Pile flexibility
flexibility matrix
matrix using
using the
the elastic
elastic beam theory (auxiliary
beam theory (auxiliary restraint
restraint approach).
approach).

In this way, the horizontal displacement of each pile-block assumes the expression as in
In this way, the horizontal displacement of each pile-block assumes the expression as in Equation (2).
Equation (2).
n
yi = n− ∑ aij Pj + y0 + θ0 zi
yi   aj=ij1Pj  y0   0 z i
(2)
(2)
j 1

in which Pj represents the load applied at the generic pile-block j (located at depth zj ), and y0 and θ 0
in which Pj represents the load applied at the generic pile-block j (located at depth zj), and y0 and 0
are the unknown displacement and rotation at the pile-head. Obviously if the pile-head is fixed, the
are the unknown displacement and rotation at the pile-head. Obviously if the pile-head is fixed, the
rotation becomes a known term. Each pile-point displacement is a function of n + 2 (or n + 1, for fixed
rotation becomes a known term. Each pile-point displacement is a function of n + 2 (or n + 1, for fixed
condition) unknowns, n pile-soil interface pressures, y0 and θ 0 .
condition) unknowns, n pile-soil interface pressures, y0 and 0.
The proposed method analyses both steel-pipe and reinforced concrete piles. For the analysis
The proposed method analyses both steel-pipe and reinforced concrete piles. For the analysis of
of steel piles, the flexural rigidity Ep Ip is assumed to be constant (which means hypothesizing a
steel piles, the flexural rigidity EpIp is assumed to be constant (which means hypothesizing a linear-
linear-elastic behavior of the section until the ultimate bending moment occurs). For reinforced
elastic behavior of the section until the ultimate bending moment occurs). For reinforced concrete
concrete sections, the development of cracks, even at low values of the bending moment, requires a
sections, the development of cracks, even at low values of the bending moment, requires a different
different modeling for the pile response. In this case, the “moment-curvature-axial load” relationship
modeling for the pile response. In this case, the “moment-curvature-axial load” relationship is
is obtained by a model that has the additional feature of taking the influence of tension stiffening into
obtained by a model that has the additional feature of taking the influence of tension stiffening into
account [43].
account [43].
Details on this model are presented in [43], however this model represents an extension to
Details on this model are presented in [43], however this model represents an extension to the
the circular section of another model that considers the tension stiffening influence for rectangular
circular section of another model that considers the tension stiffening influence for rectangular
reinforced concrete sections [44].
reinforced concrete sections [44].
Once the moment-curvature relationship has been obtained, the coefficients of the flexibility
Once the moment-curvature relationship has been obtained, the coefficients of the flexibility
matrix need to be defined using Equation (3) for the reinforced concrete pile, which is modeled in this
matrix need to be defined using Equation (3) for the reinforced concrete pile, which is modeled in
case as a step-tapered beam with a variable flexural rigidity, Ep Ip , along the pile shaft. In Equation (3),
this case as a step-tapered beam with a variable flexural rigidity, EpIp, along the pile shaft. In Equation
(3), the variation of both Ep and Ip along the shaft is fully considered by changing Ip of the section,
while Ep is kept constant. Consequently, in an incremental analysis, the pile flexibility matrix needs
to be updated at each load increment.
i 1 
l  l k 1 3 z j  l k   l k  l k 1 2   l k  l k 1 2 z j  l k   l k  l k 1   
a ij    k     z i  l k  (3)
Materials 2018, 11, 300 5 of 21

Materials 2018, 8, 300 5 of 23

the variation of both Ep and Ip along the shaft is fully considered by changing Ip of the section, while
Ep is kept constant. Consequently, in
 z i  li 1 3  
an incremental analysis,
z j  z i  z i  l i 1  
2 the pile flexibility matrix needs to be
 if z i  z j
updated at each load increment.  
 3E p I i 2 E I 
 p i 
i −1
 2
   
3
( l k − l k −1 ) (z j −lk )·(lk −lk−1 ) ( l k − l k −1 )2 (z j −lk )·(lk −lk−1 )
aij = ∑ + + + · ( z − l )
   
3E p Ik 2E p Ik 2 2E p Ik E p Ik i k
j 1 
l 
k =1
l 3
z j  l k l k  l k31    l k  l k 1 22 z j  l k  l k  l k 1   
a ij    k k 1
 ( z i − li −1 ) (z j−zi )·(zi −li−1 )    z i  l k 
k 1 
 3E I k + 2 E p3E I kp Ii +   2Ep2IE if z i < z jE I  
 j−1p    i p I k p k
    (3)
3 2 2
( l k − l k −1 ) ( j k ) k k −1
z − l ·( l − l ) ( l k − l k −1 ) ( j k ) k k −1
z − l ·( l − l )
aij = ∑ + + + · ( z − l )
    
3E p Ik 2E p Ik 2E2p Ik E p Ik i k
k =1  z j l j 1 3 z3i  z j  z j  l j 1 2 
  (z −l ) ( z − z ) · ( z j − l j −1 ) if z  z
 3+E p I j j3Epj−I 1 + 2iE pjI2E  if zi ≥ z j
i j
 j j pIj 
In Equation
In Equation (3),
(3),zzi iand
andzjzrepresent respectively the distance between the fixed node in Figure 2 and
j represent respectively the distance between the fixed node in Figure 2
and the point along the beam in whichdisplacement
the point along the beam in which the is considered
the displacement and the
is considered anddistance between
the distance the same
between the
fixed fixed
same node node
and the
andpoint where
the point the the
where load is applied.
load is applied.OnOnthetheother
otherhand,
hand,lkl,k ,represents
represents the distance
the distance
between the
between the fixed
fixed node
node and and the
thelower
lowerpart
partofofblock
blockk,k,and
andEEpIpkIis the flexural rigidity of block k.
k is the flexural rigidity of block k.

2.3. Soil Modelling

The soil is modeled as a multi-layered elastic half-space. BEM analysis requires an appropriate
elementary singular
elementary singularsolution solutiontotobebeintegratedintegrated onon thethe surface
surface of the
of the problem problem domain.
domain. In the Incase
the of case of
piles
piles subjected
subjected to horizontal
to horizontal loading, loading, the elastic
the elastic Mindlin Mindlin
solution solution is generally
is generally used [21]. usedThis [21]. This solution,
solution, which
which evaluates
evaluates the pile-soil the pile-soil interactions,
interactions, is validisand valid and rigorous
rigorous only inonly the casein the of case of a homogeneous
a homogeneous elastic
elastic half-space,
half-space, however however
it can still it can still be considered
be considered valid in the valid in of
case thea case of a multi-layered
multi-layered elastic half-spaceelastic half-[23].
spaceIn[23].
this work, the approximation suggested by Poulos and Davis [23] is used, so the soil elastic
modulusIn this work, thein
introduced approximation
the Mindlin equation suggested by Poulos
is the mean value and Davis between [23]theis used,
elasticsomodulusthe soil elastic
at the
modulus
point where introduced
the displacement in the Mindlin equation
is evaluated andis the
the mean
elasticvaluemodulus between at the the elastic
point wheremodulus
the force at the
is
point where the
applied: E = (Ei + Ej )/2. displacement is evaluated and the elastic modulus at the point where the force is
applied:
The Ehorizontal
= (Ei + Ej)/2. displacement sij at a point i belonging to the half space by a horizontal load Pj
appliedTheathorizontal
point j can displacement
be expressedsij as at in
a point
Equation i belonging
(4) (Figure to the3). half
Where space
the by term a horizontal
bij represents loadthePj
applied at point j can be expressed
general expression for each soil flexibility matrix coefficient. as in Equation (4) (Figure 3). Where the term bij represents the
general expression for  each soil flexibility matrixcoefficient.
Pj s1ij= 8πEj((31−)ν4) (3−R4ν1) + Rx1 2 + 3x 3+43−34ν
 
P 1+ ν
    bij Pj

  
4(1−ν)·(1−2ν)
 
2 2 3x2 x2 2
x 2 + 2cz
2 cz 13−x 4+1   1  2 1 − x = (4)
sij  s  1  23  R1 3 Rx2  3 R21  2 R2
2 3 2 R 2 + z + c 1  2 2
R ( R + z + c )   bij Pj (4)
8Es (1   )  R1 R2 R1 R2 R2  R2  R2  z  c  R2 R2  z  c  

Figure 3.
Figure 3. Mindlin
Mindlin solution
solution scheme.
scheme.

2.3.1. Soil Non-Linear Behavior

In [45–47] the shear stress-strain curves were approximated with hyperbolae, with the tangent
equal to Gmax at zero strain and where the tangent is asymptotic to max at infinite strain.
Materials 2018, 11, 300 6 of 21

Soil Non-Linear Behavior

In [45–47] the shear stress-strain curves were approximated with hyperbolae, with the tangent
equal to Gmax at zero strain and where the tangent is asymptotic to τ max at infinite strain.
By defining a reference strain (γref = τ max /Gmax ) it was possible to rewrite the equation of a
hyperbola as a normalized secant shear modulus (Gsec /Gmax ) that is reduced with a normalized shear
strain (γ/γref ), Equation (5).
Gsec 1
=   (5)
Gmax 1+ γ γre f

Fahey and Carter [36], instead, used the formulations in Equation (6) and Equation (7) for the
secant and the tangent elastic modulus reduction respectively.
 g
Gsec τ
= 1 − Rf (6)
Gmax τmax
 2
Gsec
Gtan Gmax
= h g i (7)
Gmax

1 − R f (1 − g) · τmax
τ

These represent a quasi-hyperbolic relation written in terms of shear stress rather than shear
strain, and employing the coefficient g to adjust the curve shape.
To model the non-linear behavior of the soil, therefore, a modified version of the formulation
proposed by Fahey and Carter [36] was adopted.
The vertical stresses (at the pile-soil interface points) are assumed not to vary during the horizontal
load analysis, so only the horizontal stresses change. An analogy can thus be assumed between the
“interface pressure-ultimate soil resistance” ratio and the “shear stress–maximum shear stress” ratio
(p/pult ≈ τ/τmax ).
With this assumption, at each step of the analysis the value of the tangent elastic modulus is
updated at each pile-soil interface point using Equation (8).
 2
Gsec
Gtan Gmax
= h g i (8)
Gmax

p
1 − R f (1 − g ) · p
ult

In the proposed method Rf is equal to 1, while the parameter g ranges between 0.25 and 1.
The appropriate value for g, to perform the analysis, can be easily estimated by trying to obtain the
best fit with the load-deflection curve of a lateral load test on a single pile or with the load-deflection
curve obtained with other available codes [12,14,28,30,34,42].
The input data required to define the soil flexibility matrix are: the Young’s modulus at small
strain level, Emax and the Poisson ratio. While the input data to define the soil resistance are: the angle
of friction or the undrained shear strength for cohesionless or cohesive, respectively.
The solving scheme, is typical of BEM methods, and requires the imposition of: a) compatibility
equations between the soil and pile displacements and b) equilibrium equations to translation and
rotation (using appropriate boundary conditions defined at the pile-heads).
In the following, the solution system is fully described for the free-head or fixed-head pile group
cases, however a different restraint condition can be considered.
The analyses are performed incrementally, using an adaptive step-size control.

2.4. Influence of Suction on Pile Group Response to Horizontal Loading

Suction is an important aspect in pile foundation subjected to lateral loads because the response
of this foundation system is mainly affected by the shallower soil layers. The proposed method uses
Materials 2018, 11, 300 7 of 21

the “MK-Model” (Modified-Kovacs Model) proposed by Aubertin et al. [48]. This model makes use
of a parameter defined as the equivalent capillary rise hc0 in the porous medium. The role of this
parameter is the same as the average capillary rise in the original model developed by Kovacs and is
calculated using the expression for the rise of water in a capillary tube (hc ) with a diameter d.
For the sake of convenience, the expression to estimate the equivalent capillary rise in granular
soils Equation (9) and in cohesive/plastic soils is reported below Equation (10).

0.75
hc0 (cm) = (9)
e · D10 [1 + 1.17 · log CU ]

where D10 (in cm) is the diameter corresponding to 10% passing on the grain-size distribution curve,
CU is the coefficient of uniformity (=D60 /D10 ), and e is the void ratio.

0.15 · ρs 1.45
hc0 = wL (10)
e

where wL is the liquid limit, and ρs is the solid grain density (kg/m3 ).
The MK-Model uses the equivalent capillary rise as a reference parameter to define the relationship
between the degree of saturation Sr (or volumetric water content θ) and the matric suction ψ. The model
considers that water is held by capillary forces responsible for capillary saturation Sc , and by adhesive
forces, causing saturation by adhesion Sa . The Sc component is more important at relatively low
suction values, while the Sa component becomes dominant at a higher suction when most capillary
water has been withdrawn. The relationship proposed in the MK-Model is written as in Equation (11)
for the degree of saturation:
θ
Sr = = Sc + S∗a · (1 − Sc ) (11)
n
In this equation, to ensure that this component does not exceed unity at low suction a truncated
value of the adhesion component Sa * is introduced in place of Sa used in the original model.
The contribution of the capillary and the adhesion components to the total degree of saturation
is defined as a function of hc0 and ψ using the equations reported in Aubertin et al. [48].
Implementing the “MK-Model” in the BEM-based method takes suction into account and increases
the effective stress state of the upper soil layers. This thus increases both the stiffness and the resistance
of the soil located close to the ground surface, which is expressed as a function of the soil stress state.

2.5. Group Effects Modelling

The experimental data revealed that for small spacing values the interaction between piles
belonging to different rows cannot be studied only considering a non-linear reduction of the soil
elastic modulus. This is because the movements of the front piles instantaneously cause an active
state condition in the soil behind the shaft. This causes not only a reduction in the stiffness of the soil
responsible for the back piles response, but also a reduction in resistance.
Therefore, the proposed BEM method required of an approach to better capture the behavior seen
in experimental data. The approach chosen is similar to that proposed by Ashour et al. [35].
In the latter work [35], the interaction among the piles in a group is determined based on the
envisioned geometry of the developing passive wedge of soil in front of the pile in addition to the pile
spacing. As shown in Figure 4, the soil passive wedge in front of a specific pile in the group overlaps
with those of adjacent piles by an area that changes with depth.
The overlap of these wedges of neighboring piles at a generic depth z is characterized as shown
in Figure 5.
According to the pile classification shown in Figure 5, the load carried by inner piles is less than
the load carried by the outer piles in each row. This fact was observed in several field tests as described
in [35].
Materials 2018, 11, 300 8 of 21
Materials 2018, 8, 300 8 of 23

Figure 4. Mobilized soil passive wedges and pile-group interaction scheme (similarly as described in
Figure 4. Mobilized soil passive wedges and pile-group interaction scheme (similarly as described in
the context of the so-called Strain Wedge Model [34,35]).
the context of the so-called Strain Wedge Model [35]).
The overlap of these wedges of neighboring piles at a generic depth z is characterized as shown
in Figure 5.
As stated in [35], at a given depth (Figure 5), overlapping areas exhibit larger values of soil strains
According to the pile classification shown in Figure 5, the load carried by inner piles is less than
and stresses compared to the isolated pile. The increase in the average soil stress attributable to the
the load carried by the outer piles in each row. This fact was observed in several field tests as
passive described
wedge ofina[35].
given pile depends on the number and area of interfering wedges overlapping the
wedge of the Aspile in question
stated in [35], at (Figure 5). (Figure 5), overlapping areas exhibit larger values of soil
a given depth
strains and stresses compared to the isolated pile. The increase in the average soil stress attributable
to the passive wedge of a given pile depends on the number and area of interfering wedges
overlapping the wedge of the pile in question (Figure 5).

Figure 5. Lateral
Figure interaction
5. Lateral for for
interaction a specific
a specificpile
pilein
in the group(similarly
the group (similarly as described
as described in thein the context
context of the of the
so-called
so-called Strain Strain
Wedge Wedge
ModelModel [34,35]).
[35]).

This overlap depends on the position of the pile in the group.

The average stress level in a soil layer (SLg ) due to passive wedge interference is evaluated based
on an empirical relationship Equation (12), which provides good agreement with field test results [35].

1.5
SL g = SLi 1 + ∑ R j ≤1 (12)
Materials 2018, 11, 300 9 of 21

where j = number of neighboring passive wedges in soil layer i that overlap the wedge of the pile in
question; R = ratio between the length of the overlapped portion of the face of the passive wedge (L)
and the total length of the face of the passive wedge (AB); and Rj is determined from all the neighboring
piles (sides and front piles) of the pile in question (Figure 5).
The SLi value on the right-hand side of Equation (12), which represents the SL of the single
isolated pile, for cohesionless soils in the Strain Wedge model, is defined in Equation (13).
◦
∆σh tan2 45 + ϕm /2 − 1


SL = = ◦ (13)
∆σh f tan2 45 + ϕ/2 − 1

where the horizontal stress change at failure ∆σhf (or the deviatoric stress at failure in the triaxial test) is
◦
∆σh f = σv0 tan2 45 + ϕ/2 − 1 . However, in the proposed method it is assumed that SLi ∼


= p/pult ,
and thus: ◦
p ∼ tan2 45 + ϕm /2 − 1


= ◦ (14)
tan2 45 + ϕ/2 − 1


pult
The mobilized friction angle, φm , can be easily obtained if SLi is known, which is assumed to be
approximately equal to the ratio p/pult .
The values of SLg vary with depth and level of loading. They can be used to evaluate the increased
value of the pressure at each pile-soil interface (p g ) (where this increase is caused by the interferences
of the passive wedges) Equations (15) and (16).
p
1.5
SL g = 1 + ∑ Rj (15)
pult
pg
SL g = (16)
pult
The value assumed by p g at each pile-soil interface is then used to update the value of the tangent
elastic modulus of the soil at each depth using Equation (8) in which p = pg .
For cohesive soils, on the other hand, is assumed to be in an undrained-condition (total stress).
Consequently, the value of φ is equal to 0◦ and also the value of φm is always 0◦ . This means that the
base angle of the passive wedge, for cohesive soils, is constantly equal to 45◦ and only the dimension
in depth (and thus on the plain) of the passive wedge changes when the load increases. However,
in this way only the interaction between the wedge of a pile positioned in a row different from the
front row with the wedge of the pile located in front of it can be considered, and thus the interactions
between the wedges of piles belonging to the same row are neglected (Figure 6).
To overcome this limit, and thus to consider the interactions between the wedges of piles located
side by side, we consider the extreme case in which a row of piles has a relative spacing s/D equal to 1.
In this condition, theoretically, the ultimate soil resistance profile should be coincident to the one in a
retaining wall, given by the difference of the passive earth pressure and the active earth pressure in an
undrained condition. In this case, the active and passive soil pressure profiles (in terms of force per
unit length) acting along the pile shaft are expressed by Equations (17) and (18) respectively.

p a (z) = [σv0 (z) − 2cu (z)] · D = [γ · z − 2cu (z)] · D (17)

p p (z) = [σv0 (z) + 2cu (z)] · D = [γ · z + 2cu (z)] · D (18)

Note that the value of pa , for shallow depths, could be negative and this means that the soil is in
tension. Assuming reasonably that the soil cannot support tension all the values of pa < 0 are corrected
considering directly pa = 0. The difference between pp and pa thus represents the ultimate soil pressure
profile, pr , (in terms of force per unit length) acting along a pile shaft in a row of piles with a spacing of
1D (or on a retaining wall). For example, Figure 7 shows all these steps to define the resulting lateral
Materials 2018, 11, 300 10 of 21

pressure profile, pr , considering a homogenous cohesive soil with a constant cu equal to 50 kPa, a soil
3
unit weight γ equal
Materials 30020 kN/m and a pile diameter D = 1 m.
2018, 8,to 10 of 23

Figure 6. Pile group effect: shadowing modeling.

To overcome this limit, and thus to consider the interactions between the wedges of piles located
side by side, we consider the extreme case in which a row of piles has a relative spacing s/D equal to
1. In this condition, theoretically, the ultimate soil resistance profile should be coincident to the one
in a retaining wall, given by the difference of the passive earth pressure and the active earth pressure
in an undrained condition. In this case, the active and passive soil pressure profiles (in terms of force
per unit length) acting along the pile shaft are expressed by Equation (17) and (18) respectively.

pa z    v 0 z   2cu z  D    z  2cu z  D (17)

p p z    v 0 z   2cu z   D    z  2cu z   D (18)

Note that the value of pa, for shallow depths, could be negative and this means that the soil is in
tension. Assuming reasonably that the soil cannot support tension all the values of pa < 0 are corrected
considering directly pa = 0. The difference between pp and pa thus represents the ultimate soil pressure
profile, pr, (in terms of force per unit length) acting along a pile shaft in a row of piles with a spacing
of 1D (or on a retaining wall). For example, Figure 7 shows all these steps to define the resulting
lateral pressure profile, pr, considering
Figure 6. Pilea group
homogenous cohesive soil
effect: shadowing with a constant cu equal to 50 kPa,
modeling.
Figure 6. Pile group effect: shadowing modeling.
a soil unit weight  equal to 20 kN/m3 and a pile diameter D = 1 m.
To overcome this limit, and thus to consider the interactions between the wedges of piles located
side by side, we consider the extreme case in which a row of piles has a relative spacing s/D equal to
1. In this condition, theoretically, the ultimate soil resistance profile should be coincident to the one
in a retaining wall, given by the difference of the passive earth pressure and the active earth pressure
in an undrained condition. In this case, the active and passive soil pressure profiles (in terms of force
per unit length) acting along the pile shaft are expressed by Equation (17) and (18) respectively.

pa z    v 0 z   2cu z  D    z  2cu z  D (17)

p p z    v 0 z   2cu z   D    z  2cu z   D (18)

Note that the value of pa, for shallow depths, could be negative and this means that the soil is in
tension. Assuming reasonably that the soil cannot support tension all the values of pa < 0 are corrected
considering directly pa = 0. The difference between pp and pa thus represents the ultimate soil pressure
profile, pr, (in terms of force per unit length)
Figure 7.acting alongprofile.
Soil pressure a pile shaft in a row of piles with a spacing
Figure 7. Soil pressure profile.
of 1D (or on a retaining wall). For example, Figure 7 shows all these steps to define the resulting
lateral pressure profile, pr, considering a homogenous cohesive soil with a constant cu equal to 50 kPa,
The a soil
soil resistance  equal is
unit weightprofile to now
20 kN/m 3 and a pile diameter D = 1 m.
considered, for the same soil condition shown before, as
defined
by Matlock [37] for a single isolated pile in soft clay, and thus expressed by the minimum values of
Equation (19).
γ0
  
J
min 3 + z + z cu ; 9cu D (19)
cu D
Figure 8 compares the values of pr (for a spacing of 1D) and pult (for a single isolated pile).
Based on all the experimental data studied, it is assumed that the ultimate soil resistance profile
for the single isolated is valid only for pile spacing s/D ≥ 6. For a pile spacing ratio of less than 6 it is
considered that the ultimate soil resistance profile is intermediate to the profile pr (assumed for spacing
ratio s/D = 1) and the profile pult (assumed for spacing ratio s/D ≥ 6).
To evaluate the definitive soil resistance profile (pult,def ), for spacing ratio between 1 and 6, it is
assumed that pult,def can be expressed as a function of the actual spacing ratio s/D and the depth, z,
using this relationship in Equation (20).Figure 7. Soil pressure profile.

pult,de f (z) = pr (z) + F (s/D ) · ( pult (z) − pr (z)) (20)

where, F(s/D) is a factor defined as a function of the relative spacing ratio according to Equation (21).

s/D − 1
F (s/D ) = (21)
5
Materials 2018, 11, 300 11 of 21

Materials 2018, 8, 300 11 of 23

With this procedure, even for cohesive soil in undrained condition, it is possible to consider the
interactions between The soil resistance
piles locatedprofile
sideis nowby considered,
side, simply for theby same soil condition shown
substituting before, as soil resistance
the ultimate
defined by Matlock [37] for a single isolated pile in soft clay, and thus expressed by the minimum
evaluated withvalues of Equation (20).defined by Matlock [37] with the pult,def values.
the expressions
In the proposed approach, the mobilized soilJ passive wedge starts to develop at a depth, along
 '  
the pile shaft, which is different frommin pile  pile,
3to z  where u D  pressure at the interface
z  cu ;9cthe (20)changes from a
 cu D  
positive to a negative value, passing from a passive to an active state.
Figure 8 compares the values of pr (for a spacing of 1D) and pult (for a single isolated pile).

Figure 8. Resulting soil pressure profile.

Figure 8. Resulting soil pressure profile.
Based on all the experimental data studied, it is assumed that the ultimate soil resistance profile
for the single isolated is valid only for pile spacing s/D  6. For a pile spacing ratio of less than 6 it is
2.6. Solution System
considered that the ultimate soil resistance profile is intermediate to the profile pr (assumed for
spacing ratio s/D = 1) and the profile pult (assumed for spacing ratio s/D  6).
The solution system
To evaluateis the
defined as:soil[F][X]
definitive = [P]
resistance Equation
profile (pult,def), for(22). [X]
spacing is between
ratio the unknowns
1 and 6, it isvector made up
of km + 2m + 1 terms
assumedor km
that +m
pult,def can+be1expressed
terms for as afree or fixed
function headspacing
of the actual conditions, the depth, z, where m is the
respectively;
ratio s/D and
using this relationship in Equation (20).
number of piles, k the number of the pile blocks for each pile, p are the km unknown pressures acting
at the pile-soil interfaces, y0 is the , def z   pr z  
pultpile-group F s / D   pult z  θpr zare
displacement,  the m pile-heads(20) rotations, Hm are
m
the m horizontal loads
where, at isthe
F(s/D) pile-heads
a factor defined as aand [P]ofisthethe
function known-terms
relative vectorto(with
spacing ratio according the
Equation same dimension
(21).
as for the vector [X]). [F] is a (km + 2m + F1)s×D (km s D  1
 + 2m + 1) or (km + m + 1) × (km + m + 1) matrix,
(21)
5
obtained by summing:
With this procedure, even for cohesive soil in undrained condition, it is possible to consider the
interactions between piles located side by side, simply by substituting the ultimate soil resistance
• the km × km pile flexibility
evaluated matrix
with the expressions [FP ],bycomposed
defined Matlock [37] with the pa
of the ij coefficients;
ult,def values.
In the proposed approach, the mobilized soil passive wedge starts to develop at a depth, along
• the km × km flexibility matrix [FS ], composed of the bij coefficients that represent the displacements
the pile shaft, which is different from pile to pile, where the pressure at the interface changes from a
induced bypositive
a load to aacting
negativeat thepassing
value, pile-soil
from ainterface j to
passive to an pile-soil interface i.
thestate.
active

The last 2m2.6. Solution

+ 1 or System
m + 1 rows and columns, of the [F] matrix, are necessary to impose
The solution
the equilibrium and to complete system is defined
the as: [F][X] = [P] Equation
compatibility (22). [X] is the
equations atunknowns vector madeinterface node.
each pile-soil
up of km + 2m + 1 terms or km + m + 1 terms for free or fixed head conditions, respectively; where m
In Equation (22), H number
is the is the ofapplied
piles, k thehorizontal
number of theload andfor
pile blocks f is the
each load
pile, p are eccentricity.
the km unknown pressures
The flexibility matrix [F] is updated at each step of the procedure.
acting at the pile-soil interfaces, y 0 is the pile-group displacement, m are the m The pilerotations,
pile-heads flexibility sub-matrix,
[FP ], is updated only in the case of a non-linear “moment-curvature” relationship for the pile section.
[FP ] is updated using the tangent flexural rigidities, according to the bending-moments reached at
each pile-node in the previous load increment.

p1 p · ∆1 D
    
a1,1 + b1,1 ··· a1,km + b1,km −1 −z1,1 ··· ··· 0 ··· 0 0
 .. .. .. .. .. .. .. .. .. ..  ..   .. 

 . . . . . . . . . . 
 .  
  . 


 akm,1 + bkm,11 · · · akm,km + bkm,km −1 · · · · · · −zkm,m 0 ··· 0 
 pkm p · ∆km D  
  0 

1 ··· 1 0 0 ··· 0 0 ··· 0 y0 H
    
    
    
 z1,1 ··· ··· 0 0 ··· 0 0 f f  θ1   fH 

 .. .. .. .. .. .. ..

 .. =
 
..

 (22)

 . . . . . . . f 0 f 

 .  
  . 


 ··· ··· zm,km 0 0 ··· 0 f f 0 

 θm  
  fH 

11,1 ··· ··· 0 0 ··· 0 −1 0 0  H1 0
    
   
.. .. .. .. .. .. .. .. ..
    
. .
    
 . . . . . 0 −1 0  .   . 
··· ··· 1m,km 0 0 ··· 0 0 0 −1 Hm 0
Materials 2018, 11, 300 12 of 21

Once the initial flexibility matrix, [F], has been computed, the total horizontal load is applied
in the first step of the solution procedure. At each generic load increment hk , an iterative process
is performed where two solutions are obtained, the first using hk as the load increment, the second
using two load steps equal to hk /2. The iterative scheme is described in Figure 9, which, for the sake
of simplicity refers to the explicit Euler method with step-doubling and adaptive step-size control.
However, a fourth order Runge-Kutta method can also be used to obtain some improvement in the
accuracy of the solution. The adaptive step-size control numerical technique is fully described in
Materials
Press et al.2018, 8, 300
[49]. 13 of 23

Figure9.9.Flow
Figure Flowchart
chartof
ofthe
theproposed
proposed non-linear adaptivestep-size
non-linear adaptive step-sizemethod.
method.

Once
Once these
these twotwo solutions
solutions havebeen
have beencomputed,
computed,the
theincremental ratioε,,isiscomputed
incrementalratio computed according
according to
to Equation
Equation (23). (23).
∆u − ∆u
u 2 2 u1 1
ε= (23)
 ∆u1 (23)
u1
where u1 and u2 are the incremental displacement at the pile-head evaluated using one and two
steps, respectively. The  value is compared with a predefined tolerance taken as equal to 0.001
(Figure 10).
Materials 2018, 11, 300 13 of 21

where ∆u1 and ∆u2 are the incremental displacement at the pile-head evaluated using one and two
steps, respectively. The ε value is compared with a predefined tolerance taken as equal to 0.001
(Figure 10).
Materials 2018, 8, 300 14 of 23

Figure 10.
Figure 10. Adaptive
Adaptive step-size
step-sizecontrol.
control.

When this convergence criterion is exceeded (  > tol), the iterative process starts again with an
When this convergence criterion is exceeded (ε > tol), the iterative process starts again with
updated load increment hknewnew
which should be able to achieve the desired accuracy and can be
anestimated
updated using
load increment hk [49].
Equation (24)
which should be able to achieve the desired accuracy and can be
estimated using Equation (24) [49].
1
 tol  p
1
hknew  SF  hk    tol
 p+1 1 (24)
hk = SF · hk·  
new
(24)
ε
where p is the order of the method used (in the Euler method p = 1, in the Runge-Kutta method p = 4),
where p isisthe
and SF order of
a “safety the method
factor” (taken asused
equal(intothe Euler
0.90) method pthe
to guarantee = 1,success
in the in
Runge-Kutta method
the next attempt. p = 4),
When
and SFconvergence
this is a “safety factor” (taken
criterion is passed (  to
as equal 0.90)
tol), to guarantee
Equation the success
(24) is used again toinestimate
the nextthe
attempt. When this
next step-size.
The procedure
convergence stops when
criterion is passed (ε ≤lateral
the final load H is reached.
tol), Equation Finally,
(24) is used againtheto
entire load-deflection
estimate curve
the next step-size.
Theand the deflection,
procedure shear, the
stops when bending moment
final lateral and
load H pile-soil interface
is reached. Finally,pressure profiles
the entire along the pile
load-deflection curve
andshaft
the at each load-step
deflection, shear,can be evaluated.
bending moment and pile-soil interface pressure profiles along the pile shaft
at each load-step can be evaluated.
3. Validation of the Proposed Method
3. Validation
In this of the Proposed
section, Method
are presented the results of the pile group analyses using the BEM-based
approach proposed
In this section, arewithin this paper.
presented Theseofresults
the results aregroup
the pile compared withusing
analyses thosethe
obtained in horizontal
BEM-based approach
proposed within this paper. These results are compared with those obtained in horizontal loadout
load tests on pile groups in sandy and cohesive/plastic soils. The lateral load tests were carried tests
onboth
pile on steel and
groups r.c. piles.
in sandy andThe experimental data
cohesive/plastic were
soils. Theretrieved from tests
lateral load well-documented
were carriedtestsout in the on
both
available
steel and r.c.literature
piles. Theforexperimental
a total of 15 pile groups
data werecase histories,
retrieved fromreported in Table 1. Further
well-documented tests ininformation
the available
about the validation procedure can be found in [50].
literature for a total of 15 pile groups case histories, reported in Table 1. Further information about the
validation procedure can be found in [50].
Materials 2018, 11, 300 14 of 21

Table 1. Case histories studied.

Case Reference Pile Material Pile Diameter (m) Pile Length (m) Soil Type Hmax (kN)
[51] 3 × 3; s = 3D Steel with Grout-fill 0.273 13.11 OC Clay 695
[5] 3 × 3; s = 3D Steel with Grout-fill 0.273 13.11 Sand 808.5
[4] 3 × 2; s = 3D Bored RC 1.5 34.9 Silty Sand 11043
[8] φ’ = 34◦ ; 3 × 3; s = 3D Aluminum 0.43 13.3 Sand 761.2
[8] φ’ = 39◦ ; 3 × 3; s = 3D Aluminum 0.43 13.3 Sand 1508.2
[8] φ’ = 34◦ ; 3 × 3; s = 5D Aluminum 0.43 13.3 Sand 1110.5
[8] φ’ = 39◦ ; 3 × 3; s = 5D Aluminum 0.43 13.3 Sand 1424
[52] 2 × 1; s = 2D Aluminum 0.72 12 Sand 1183
[52] 2 × 1; s = 4D Aluminum 0.72 12 Sand 1220.1
[52] 2 × 1; s = 6D Aluminum 0.72 12 Sand 1030.72
[53] 3 × 3; s = 3D Steel with Grout-fill 0.305 8.7 Clay 927.05
[11] 3 × 3; s = 3D Steel pipe 0.324 11.5 Sand 488.6
[54] 3 × 3; s = 5.65D Steel pipe 0.324 11.9 Clay 1407
[54] 3 × 4; s = 4.4D Steel pipe 0.324 11.9 Clay 1353.8
[54] 3 × 5; s = 3.3D Steel pipe 0.324 11.9 Clay 1942.5

In the case histories studied the maximum number of piles in the group is 15 and the lateral load
test data on a single pile were always included.
The purpose of the analyses is to validate the BEM approach developed in this work. The analyses
were realized as class A predictions by using the actual pile properties and the soil geotechnical
parameters obtained based on the in situ and laboratory tests data. In particular, the soil elastic
modulus to be considered is a Young’s Modulus at small strain level inferred from in situ tests.
It needs to be remembered that in the analyses of pile-groups the value of the exponent g of
the elastic modulus reduction curve must be defined. The appropriate g value to be inputted can be
easily estimated trying to obtain the best fit with the load-deflection curve of the horizontal test on
the single pile. In Table 2 are reported the input data used to perform the analyses with the proposed
method. These data refer to the properties of the soil layers at least in between the ground surface and
10-diameter depth. Additional details are presented in [50].

Table 2. Input data used to perform the analyses with the proposed method.

Emax (Linear
Ep Ip γ DR W.T. Head
Case φ (◦ ) cu (kPa) Increasing with G (-) F (m)
(MNm2 ) (kN/m3 ) (%) (m) B.C.
Depth) (MPa)
[51] 16.0 19.0 - - 58–145 (0–5.5 m) 70–200 (0–5.5 m) 0.25 0.305 0.0 Free
[5] 16.0 19.5 47 90 - 35–100 (0–2.0 m) 1.0 0.305 0.0 Free
[4] variable 18.5 34 50 - 40–400 (0–34.9 m) 0.5 1.0 1.0 Fixed
[8] 72.1 14.51 34 33 - 60–300 (0–13.3 m) 0.25 1.68 - Free
[8] 72.1 15.18 39 55 - 50–260 (0–13.3 m) 0.5 1.68 - Free
[52] 514.0 16.3 40 89 - 40–200 (0–12.0 m) 1.0 1.6 - Free
[53] 26.91 19.0 - - 50–75 (0–2.9 m) 60–170 (0–8.7 m) 0.25 0.4 0.0 Free
[11] 30.03 19.5 40 44 - 20–150 (0–11.5 m) 0.25 0.86 0.0 Free
60 (0–1 m)120
[54] 30.03 19.0 - - 50–60 (0–4.0 m) 0.25 0.48 1.0 Free
(1.0–4.0 m)
Table notes: Ep Ip = pile flexural rigidity; γ = soil unit weight; φ = peak friction angle; DR = relative density; cu =
undrained shear strength; Emax = soil elastic modulus at small strain level; g = parameter of the modulus reduction
curve; f = load eccentricity; W.T. = water table depth below the ground surface; Head B.C. = pile-head boundary
conditions (free-head or fixed head).

The comparison between computed and measured results (Figure 11) clearly show the capability
of the proposed BEM method to provide good predictions of the pile groups lateral response.
Materials 2018, 11, 300 15 of 21
Materials 2018, 8, 300 16 of 23

Figure11.
Figure 11.Measured
Measuredvs.vs.computed
computedhorizontal
horizontalloads
loads(H)
(H)for
foraagiven
givennormalized
normalizeddisplacement
displacement(y/D):
(y/D):
(a) y/D = 0–0.5%; (b) y/D = 1.0–2.0%; (c) y/D = 4.0–5.0%; (d) y/D = 8.0–10.0%.
(a) y/D = 0–0.5%; (b) y/D = 1.0–2.0%; (c) y/D = 4.0–5.0%; (d) y/D = 8.0–10.0%.

In Figure 11, the ratio among the experimental lateral load for a given normalized displacement
In Figure 11, the ratio among the experimental lateral load for a given normalized displacement
(y/D) and the experimental maximum horizontal load is on the x-axis, while the ratio among the
(y/D) and the experimental maximum horizontal load is on the x-axis, while the ratio among the
calculated and the measured load (for a given y/D) is on the y-axis.
calculated and the measured load (for a given y/D) is on the y-axis.
The error in the load forecasting at each y/D reached during the tests is included in the range of
The error in the load forecasting at each y/D reached during the tests is included in the range
±30%.
of ±30%.

3.1.Analysis
3.1. AnalysisResults
Resultswith
withthe
theProposed
ProposedBEM
BEMMethod
Methodfor
foraaSpecific
SpecificLateral
LateralLoad
LoadTest
Testonona aBored
BoredPile
PileGroup
Group

AA full-scale
full-scale lateral
lateral load
load test
test program
program[4] [4]was
wasrealized
realizedininTaiwan
Taiwan in in
2001. Two
2001. pilepile
Two groups, one
groups,
consisting
one in bored
consisting piles
in bored andand
piles the the
other in driven
other piles,
in driven were
piles, subjected
were to horizontal
subjected loading
to horizontal tests.tests.
loading The
tests were also conducted on single piles installed using the same two techniques. In
The tests were also conducted on single piles installed using the same two techniques. In this section, this section, all
thethe
all presented
presented results
resultsrefer
refertotothe
thefree-head
free-headbored
boredsingle
singlepile
pileand
andto
to the
the fixed-head
fixed-head bored pile group.
bored pile group.
The latter is a 3 × 2 (3 rows and 2 columns) pile group at
The latter is a 3 × 2 (3 rows and 2 columns) pile group at 3D spacing. 3D spacing.
Materials 2018, 11, 300 16 of 21

3.1.1. Soil and Pile Properties Description

The soil in the site was classified as silt or silty sand, with occasional layers of silty clay. The water
table was at 1.0 m below the ground surface and did not change considerably during the entire test
program period. In [4]
Materials 2018, 8, 300are fully presented the cone penetration tests and Gmax data profiles. 17 of 23

In the site were placed 13 cast in situ bored piles and 13 precast driven piles. Eleven of the 13 bored
3.1.1. Soil and Pile Properties Description
piles (D = 1500 mm, L = 34.9 m; EI = 6.86 GNm2 ) were realized using bentonite-mud with reverse
The soil in the site was classified as silt or silty sand, with occasional layers of silty clay. The
circulation.water
Two of the 13 bored piles were realized by means of a drilling device with hydraulic
table was at 1.0 m below the ground surface and did not change considerably during the entire
oscillator attest
full length.
program period. The In
measurement instruments
[4] are fully presented the cone(strain gauges
penetration testsand
and inclinometers)
Gmax data profiles. were attached
to the longitudinalIn the reinforcement
site were placed 13bars, inserted
cast in situ boredinto
pilesthe
andhole before
13 precast casting
driven piles.the concrete.
Eleven of the 13Bored pile
bored piles (D = 1500
properties are summarized in Table 3. mm, L = 34.9 m; EI = 6.86 GNm 2) were realized using bentonite-mud with reverse

circulation. Two of the 13 bored piles were realized by means of a drilling device with hydraulic
oscillator at full length. The measurement instruments (strain gauges and inclinometers) were
Table 3. Structural properties of bored pile.
attached to the longitudinal reinforcement bars, inserted into the hole before casting the concrete.
Bored pile properties are summarized in Table 3.
Pile Cross Concrete Reinforcement Intact Flexural
Pile Length Steel Ratio
Diameter D Sectional Compressive Yield Stress f Rigidity EI
(m) Table
2 3. Structural properties of bored pile. y ρs
(mm) Area (cm ) Strength f ’c (MPa) (MPa) (GNm2 )
1500 Pile 34.9 Cross
17672 Concrete
27.5 471 Steel
0.025 Intact flexural
6.86
Pile Length Reinforcement
Diameter Sectional Compressive Ratio Rigidity EI
(m) Yield Stress fy (MPa)
D (mm) Area (cm2) Strength f’c (MPa) s (GNm ) 2

1500 34.9 17672 471 27.5 0.025 6.86

3.1.2. Single Bored Pile B7 (Free-Head) and Pile Group (Fixed-Head): Analysis Results
3.1.2.
The soil Single
unit Bored Pile
weight (γ) B7
was(Free-Head)
evaluated andbased
Pile Group (Fixed-Head):
on the Analysis Results
cone penetration tests data fully reported
in [4]. Along theThefirst
soil unit meters()in
15 weight depth
was the based
evaluated tip resistance
on the conein the CPTtests
penetration tests was
data on
fully average
reported in equal to
[4]. Along the first 15 meters in depth the tip resistance in the CPT tests was on average equal to 5
5 MPa. The pile properties used to perform the BEM analyses are those reported in Table 3.
MPa. The pile properties used to perform the BEM analyses are those reported in Table 3.
The bending moment-curvature relationship of the pile section (Figure 12) was computed with
The bending moment-curvature relationship of the pile section (Figure 12) was computed with
the model the
thatmodel
can that
consider the influence
can consider of of
the influence tension
tension stiffening [43].
stiffening [43].

Figure 12. Pile section ‘bending moment-curvature’ relationship for B7 pile section.
Figure 12. Pile section ‘bending moment-curvature’ relationship for B7 pile section.
Based on the cone penetration tests data in [4], an angle of internal friction of 34°was obtained
using the correlation suggested by Mayne [55] Equation (25). ◦
Based on the cone penetration tests data in [4], an angle of internal friction of 34 was obtained
using the correlation suggested by Mayne [55] Equation  qt (25).
 v 0 
 '  17.6  11  log10   (25)
 ' p  
 0t −
vq a σv0
ϕ0 = 17.6 + 11 · log10 √ (25)
σ 0v0 p a
The Gmax profile was that provided in [4]. This profile was simplified and assumed linearly
increasing from 15 to 150 MPa. The Poisson ratio was set equal to 0.35.
The Gmax profile was that provided in [4]. This profile was simplified and assumed linearly
increasing from 15 to 150 MPa. The Poisson ratio was set equal to 0.35.
The ultimate soil pressure profile was evaluated according to relationship proposed by Reese et al. [38].
Since the water table was located 1 m below the ground surface, approximate suction effects were
considered, because of the lack of information to use the Modified Kovacs Model rigorously, thus
 Materials 2018, 8, 300 18 of 23

Materials 2018, 11, 300 17 of 21

The ultimate soil pressure profile was evaluated according to relationship proposed by Reese
et al. [38]. Since the water table was located 1 m below the ground surface, approximate suction effects
were considered, because of the lack of information to use the Modified Kovacs Model rigorously,
increasingthustheincreasing
vertical effective
the verticalsoil stresses
effective soil at the first
stresses meter
at the in depth.
first meter In fact,
in depth. a linearly
In fact, a linearlyincreasing
suction value
increasing suction value was assumed from 0 kPa to 10 kPa starting from 1.0 meter depth up ground
was assumed from 0 kPa to 10 kPa starting from 1.0 meter depth up to the to the surface.
ground surface.
As a consequence, the soil resistance profile, obtained with the relationships suggested in [38],
is increased inAs a consequence, the soil resistance profile, obtained with the relationships suggested in [38],
the first meter depth.
is increased in the first meter depth.
The comparison between
The comparison measured
between measured and
andcalculated results
calculated results forfree-head
for the the free-head single
single pile pile case are
case are
shown in Figure
shown in13 (considering
Figure andand
13 (considering notnot
considering suction).
considering suction).

Figure 13. Measured vs. computed results: Lateral Load versus Head Deflection curve.
Figure 13. Measured vs. computed results: Lateral Load versus Head Deflection curve.
The capability of the BEM method to forecast the laterally loaded pile response is good in both
cases, nevertheless it can be observed an improvement in the prediction considering suction. The
The capability of the BEM method to forecast the laterally loaded pile response is good in both
results for the 3 × 2 fixed-head pile group are presented in Figures. 14–16 in terms of the load
cases, nevertheless it canfor
deflection curve beanobserved aninimprovement
average pile the group, group inefficiency
the prediction
and the considering
pile-deflection suction.
profile at aThe results
for the 3 ×specific
2 fixed-head pile
value of the group
lateral loadare
forpresented
all the piles in Figures
in the group.14–16 in terms
The group of the
efficiency load
in the deflection curve
pile-group
analysis
for an average results
pile is defined
in the group, as:group
Hgroup/(nefficiency
Hsingle); where, Hgroup
and the= the total horizontal load
pile-deflection in the
profile atpile group, value of
a specific
Hsingle = the horizontal load in the isolated single pile (at the same displacement-level) and n = the
the lateral load for all the piles in the group. The group efficiency in the pile-group analysis results is
number of piles in the group.
defined as: Hgroup /(n Hefficiency
The group single ); where,
in FigureHgroup = the than
15 is higher totalonehorizontal
because theload
singleinpile
thewaspile group,
tested Hsingle = the
in free-
horizontalhead
loadrestraint
in the condition,
isolated while
singlethepile
pile(at
grouptheissame displacement-level) and n = the number of piles
fixed-head.
in the group.
Materials 2018, 8, 300 19 of 23

Figure 14. Computed vs. measured load-deflection curves.

Figure 14. Computed vs. measured load-deflection curves.
Materials 2018, 11, 300 18 of 21
Figure 14. Computed vs. measured load-deflection curves.

Figure 15. Computed vs. measured group efficiency.

Materials 2018, 8, 300 Figure 15. Computed vs. measured group efficiency. 20 of 23

Figure 16. Computed vs. measured deflection profiles at H = 10948 kN of piles B2, B5, B6, B8, B9, B10.
Figure 16. Computed vs. measured deflection profiles at H = 10948 kN of piles B2, B5, B6, B8, B9, B10.
The reader should note that in [4] and in [56] the load of 1462 kN was considered as the lateral
load
The groupcorresponding
efficiencytointhe first cracking
Figure bending
15 is higher moment
than based onthe
one because back-analysis
single pileresults.
was tested in free-head
For this reason, in [4] and more recently in [56] a reduced bending stiffness was set to the relevant
restraint condition, while the pile group is fixed-head.
section of the piles to consider cracking.
The reader should note
The computer codesthat
usedinin [4]
[4] and in [56]
and [56] werethe load
LPILE ofand
[57] 1462 kN was considered
VERSAT-P3D as the lateral
[58], respectively.
load corresponding to the first cracking bending moment based on back-analysis
The proposed BEM approach, instead, automatically updates each pile-block bending stiffness results.
Foraccording
this reason, in “average
to the [4] and more recently in [56]
moment-curvature” a reducedobtained
relationship bendingfrom
stiffness was pile
the actual set to the relevant
section
section properties.
of the piles to consider cracking.
With the proposed method, the analysis to obtain the bored pile group results requires less than
The computer codes used in [4] and [56] were LPILE [57] and VERSAT-P3D [58], respectively.
10 min to compute the entire lateral load-deflection curve on a laptop with an Intel Core i7 CPU
The proposed BEM approach, instead, automatically updates each pile-block bending stiffness according
processor (2.20 GHz). Analyses of similar problems by VERSAT-P3D [58] using a coarse finite
to the “average moment-curvature”
element mesh requires about 20relationship obtained
min to calculate from the actual
the displacement for a pile
singlesection properties.
point on the load-
With the proposed method, the analysis to
deflection curve, while FLAC-3D takes about 6 h [59]. obtain the bored pile group results requires less than
10 min to compute the entire lateral load-deflection curve on a laptop with an Intel Core i7 CPU
4. Conclusions
processor (2.20 GHz). Analyses of similar problems by VERSAT-P3D [58] using a coarse finite element
A pile group subjected to horizontal load is a complex soil-structure interaction (SSI) problem
affected by pile and soil non-linear behavior. Even nowadays, this specific SSI problem cannot be
easily solved, especially because most of the computer codes are specialized to solve/study either
structural or geotechnical issues.
One of the key aspects of the pile group lateral behavior is the continuous pile-soil relative
stiffness variation, while a horizontal load is applied.
Materials 2018, 11, 300 19 of 21

mesh requires about 20 min to calculate the displacement for a single point on the load-deflection
curve, while FLAC-3D takes about 6 h [59].

4. Conclusions
A pile group subjected to horizontal load is a complex soil-structure interaction (SSI) problem
affected by pile and soil non-linear behavior. Even nowadays, this specific SSI problem cannot be
easily solved, especially because most of the computer codes are specialized to solve/study either
structural or geotechnical issues.
One of the key aspects of the pile group lateral behavior is the continuous pile-soil relative stiffness
variation, while a horizontal load is applied.
To capture the latter, a new BEM-based approach for the analysis of laterally loaded pile group
has been developed and validated. Herein, the solution system of the proposed method is fully
presented for the free-head and fixed-head pile group cases, however a different restraint condition
can be considered.
The proposed BEM approach is innovative because can take into account for the highly non-linear
behavior of reinforced concrete piles, considering also the tension stiffening effect. Moreover, the influence
of suction in the upper soil layers, is also considered, by means of the Modified-Kovacs Model.
The method developed herein presents two significant merits compared to some FDM, FEM
and quasi-3D FEM codes: the reduction of computation (or running) time, and the easiness in the
selection/definition of the input parameters to perform the analyses.
The reliability of the proposed BEM method was tested by comparing computed and experimental
data from full scale and centrifuge tests on 15 pile groups, retrieved from the available technical
literature about this topic. The results presented herein have shown the capability of the BEM method to
provide a good prediction both qualitatively and quantitatively of the relevant aspects of the pile group
horizontal behavior. The prediction errors are lower than 30% for most of the case histories studied.
Finally, for comparison purpose, the proposed method was used to analyze a specific pile group
horizontally loaded in a full-scale test program realized in Taiwan in 2001.

Acknowledgments: This research did not receive any specific grant from funding agencies in the public,
commercial, or not-for-profit sectors.
Author Contributions: Stefano Stacul and Nunziante Squeglia developed the proposed model for the laterally
loaded pile groups, Stefano Stacul analyzed the data and performed the comparison with experimental tests;
Stefano Stacul and Nunziante Squeglia wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Mokwa, R.L.; Duncan, J.M. Experimental evaluation of lateral-load resistance of pile caps. J. Geotech.
Geoenviron. Eng. 2001, 127, 185–192. [CrossRef]
2. Katzenbach, R.; Turek, J. Combined pile-raft foundation subjected to lateral loads. In Proceedings of the
International Conference on Soil Mechanics and Geotechnical Engineering, Osaka, Japan, 12–16 Spetember
2005; p. 2001.
3. O’Neill, M.; Dunnavant, T.W. A Study of the Effects of Scale, Velocity, and Cyclic Degradability on Laterally Loaded
Single Piles in Overconsolidated Clay; University of Houston: Houston, TX, USA, 1984.
4. 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. Eng. 2001, 127, 385–397. [CrossRef]
5. Brown, D.; Morrison, C.; Reese, L. Lateral load behavior of pile group in sand. J. Geotech. Eng. 1988, 114,
1261–1276. [CrossRef]
6. McVay, M.; Zhang, L.; Molnit, T.; Lai, P. Centrifuge Testing of Large laterally loaded Pile Groups in Sand.
J. Geotech. Geoenviron. Eng. 1998, 124, 1016–1026. [CrossRef]
7. Lateral-Load Tests on 25.4 mm (1-in.) Diameter Piles in Very Soft Clay in Side-by-Side and in-Line Groups; ASTM:
West Conshohocken, PA, USA, 1984; STP 835; pp. 122–139.
Materials 2018, 11, 300 20 of 21

8. 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]
9. Rollins, K.M.; Gerber, T.M.; Lane, J.D.; Ashford, S.A. Lateral resistance of a full-scale pile group in liquefied
sand. J. Geotech. Geoenviron. Eng. 2005, 131, 115–125. [CrossRef]
10. Mokwa, R. Investigation of the Resistance of Pile Caps to Lateral Loading. PhD Thesis, Virginia Polytechnic
and State University, Blacksburg, VA, USA, February 2000.
11. Rollins, K.; Lane, J.; Gerber, T. Measured and computed lateral response of a pile group in sand. J. Geotech.
Geoenviron. Eng. 2006, 132, 103–114. [CrossRef]
12. Reese, L.C.; Wang, S.T.; Isenhower, W.M.; Arrellaga, J.A. Computer Program Lpile Plus Version 5.0 Technical
Manual; Ensoft: Austin, TX, USA, 2004.
13. Reese, L.C.; Wang, S.T.; Arrellaga, J.A.; Hendrix, J. Computer Program GROUP for Windows, User’s Manual,
Version 4.0; Ensoft: Austin, TX, USA, 1996.
14. Hoit, M.I.; McVay, M.; Hays, C.; Andrade, P.W. Nonlinear pile foundation analysis using Florida-Pier.
J. Bridg. Eng. 1996, 1, 135–142. [CrossRef]
15. Abdrabbo, F.M.; Gaaver, K.E. Simplified analysis of laterally loaded pile groups. Alexandria Eng. J. 2012, 51,
121–127. [CrossRef]
16. Hirai, H. A Winkler model approach for vertically and laterally loaded piles in nonhomogeneous soil. Int. J.
Numer. Anal. Methods Geomech. 2012, 36, 1869–1897. [CrossRef]
17. Khari, M.; Kassim, K.A.; Adnan, A. Development of Curves of Laterally Loaded Piles in Cohesionless Soil.
Sci. World J. 2014, 2014, 917174. [CrossRef] [PubMed]
18. Mardfekri, M.; Gardoni, P.; Roesset, J.M. Modeling laterally loaded single piles accounting for nonlinear
soil-pile interactions. J. Eng. 2013, 2013, 243719. [CrossRef]
19. Yang, Z.; Jeremić, B. Numerical analysis of pile behavior under lateral loads in layered elastic–plastic soils.
Int. J. Numer. Anal. Methods Geomech. 2002, 26, 1385–1406. [CrossRef]
20. Yang, Z.; Jeremić, B. Numerical study of group effects for pile groups in sands. Int. J. Numer. Anal.
Methods Geomech. 2003, 27, 1255–1276. [CrossRef]
21. Mindlin, R.D. Force at a point in the interior of a semi-infinite solid. Physics 1936, 7, 195–202. [CrossRef]
22. Spillers, W.R.; Stoll, R.D. Lateral response of piles. J. Soil Mech. Found. Div. 1964, 90, 1–10.
23. Poulos, H.G.; Davis, E.H. Pile Foundation Analysis and Design; Wiley & Sons: New York, NY, USA, 1980.
24. Davies, T.G.; Banerjee, P.K. The displacement field due to a point load at the interface of a two layer elastic
half-space. Geotechnique 1978, 28, 43–56. [CrossRef]
25. Sharnouby, B.E.; Novak, M. Flexibility coefficients and interaction factors for pile group analysis.
Can. Geotech. J. 1986, 23, 441–450. [CrossRef]
26. Budhu, M.; Davies, T.G. Nonlinear analysis of laterality loaded piles in cohesionless soils. Can. Geotech. J.
1987, 24, 289–296. [CrossRef]
27. Budhu, M.; Davies, T.G. Analysis of Laterally Loaded Piles in Soft Clays. J. Geotech. Eng. 1988, 114, 21–39.
[CrossRef]
28. Poulos, H.G. User’s Guide to Program DEFPIG 3/4 Deformation Analysis of Pile Groups; University of Sydney:
Sydney, Australia, 1990.
29. Randolph, M.F. PIGLET: A Computer Program for the Analysis and Design of Pile Groups; Report Geo 86033;
University of Western Australia: Perth, Australia, 1987.
30. Basile, F. A practical method for the non-linear analysis of piled rafts. In Proceedings of the 18th International
Conference on Soil Mechanics and Geotechnical Engineering, Paris, France, 2–5 September 2013; p. 2503.
31. Matos Filho, R.; Mendonça, A.V.; Paiva, J.B. Static boundary element analysis of piles submitted to horizontal
and vertical loads. Eng. Anal. Bound. Elem. 2005, 29, 195–203. [CrossRef]
32. Kitiyodom, P.; Matsumoto, T.; Horikoshi, K. Analyses of vertical and horizontal load tests on piled raft
models in dry sand. In Proceedings of the International Conference on Soil Mechanics and Geotechnical
Engineering, Osaka, Japan; 2005; pp. 2005–2008.
33. Small, J.C.; Zhang, H.H. Piled raft foundations subjected to general loadings. In Proceedings of the
Developments in Theoretical Geomechanics, Sydney, Australia, 1 January 2000; pp. 57–72.
34. Ashour, M.; Norris, G.; Pilling, P. Lateral Loading of a Pile in Layered Soil Using the Strain Wedge Model.
J. Geotech. Geoenviron. Eng. 1998, 124, 303–315. [CrossRef]
Materials 2018, 11, 300 21 of 21

35. Ashour, M.; Pilling, P.; Norris, G. Lateral Behavior of Pile Groups in Layered Soils. J. Geotech. Geoenviron. Eng.
2004, 130, 580–592. [CrossRef]
36. Fahey, M.; Carter, J. A finite element study of the pressuremeter test in sand using a nonlinear elastic plastic
model. Can. Geotech. J. 1993, 30, 348–362. [CrossRef]
37. Matlock, H. Correlations for design of laterally loaded piles in soft clay. In Proceedings of the 2nd Annual
Offshore Technology Conference, Houston, TX, USA, 22–24 April 1970; pp. 577–588.
38. Reese, L.C.; Cox, W.R.; Koop, F.D. Analysis of laterally loaded piles in sand. In Proceedings of the VI Annual
Offshore Technology Conference, Houston, TX, USA, 1974; pp. 473–485.
39. Welch, R.C.; Reese, L.C. Lateral Load Behavior of Drilled Shafts; University of Texas: Austin, TX, USA, 1972.
40. Reese, L.; Cox, W.; Koop, F. Field testing and analysis of laterally loaded piles in stiff clay. Proceedings of
Offshore Technology Conference, Houston, TX, USA, 5–8 May 1975; pp. 671–690.
41. Landi, G. Pali soggetti a carichi orizzontali: Indagini sperimentali ed analisi. PhD Thesis, University of
Naples Federico II, Pisa, Italia, 26–28 June 2006.
42. Stacul, S.; Squeglia, N.; Morelli, F. Laterally Loaded Single Pile Response Considering the Influence of
Suction and Non-Linear Behaviour of Reinforced Concrete Sections. Appl. Sci. 2017, 7, 1310. [CrossRef]
43. Morelli, F.; Amico, C.; Salvatore, W.; Squeglia, N.; Stacul, S. Influence of Tension Stiffening on the Flexural
Stiffness of Reinforced Concrete Circular Sections. Materials 2017, 10, 201. [CrossRef]
44. Salvatore, W.; Buratti, G.; Maffei, B.; Valentini, R. Dual-phase steel re-bars for high-ductile r.c. structures,
Part 2: Rotational capacity of beams. Eng. Struct. 2007, 29, 3333–3341. [CrossRef]
45. Kondner, R.L. Hyperbolic Stress-Strain Response: Cohesive Soils. J. Soil Mech. Found. Div. 1963, 89, 115–144.
46. Duncan, J.M.; Chang, C.Y. Nonlinear Analysis of Stress and Strain in Soil. J. Soil Mech. Found. Div. 1970, 96,
1629–1653.
47. Hardin, B.O.; Drnevich, V.P. Shear Modulus and Damping in Soils: Design Equations and Curves. Soil Mech.
Found. Div. 1972, SM7, 667–692.
48. Aubertin, M.; Mbonimpa, M.; Bussière, B.; Chapuis, R.P. A model to predict the water retention curve from
basic geotechnical properties. Can. Geotech. J. 2003, 40, 1104–1122. [CrossRef]
49. Press, W.H.; Teukolsky, S.A.; Vetterling, W.T.; Flannery, B.P. Numerical Recipes in C++. The Art of Scientific
Computing, 2nd ed.; Cambridge University Press: Cambridge, UK, 2002; pp. 714–718.
50. Stacul, S. Analysis of piles and piled raft foundation under horizontal load. PhD Thesis, University of
Florence, Florence, Italy, 2018.
51. Brown, D.; Reese, L.; O’Neill, M. Cyclic lateral loading of a large-scale pile group. J. Geotech. Eng. 1987, 113,
1326–1343. [CrossRef]
52. Remaud, D.; Garnier, J.; Frank, R. Laterally loaded piles in dense sand: Group effects. In Proceedings of the
International Conference Centrifuge, Tokyo, Japan, 23–25 September 1998; pp. 533–538.
53. Rollins, K.M.; Peterson, K.T.; Weaver, T.J. Lateral load behavior of full-scale pile group in clay. J. Geotech.
Geoenviron. Eng. 1998, 124, 468–478. [CrossRef]
54. Rollins, K.M.; Olsen, K.G.; Jensen, D.H.; Garrett, B.H.; Olsen, R.J.; Egbert, J.J. Pile spacing effects on lateral
pile group behavior: Analysis. J. Geotech. Geoenviron. Eng. 2006, 132, 1272–1283. [CrossRef]
55. Mayne, P.W. In situ test calibrations for evaluating soil parameters. In Proceedings of the Second
International Workshop on Characterisation and Engineering Properties of Natural Soils, Singapore,
Singapore, 29 November–1 December 2006; Tan, T.S., Phoon, K.K., Hight, D.W., Leroueil, S., Eds.; CRC Press:
Boca Raton, FL, USA; pp. 1601–1652.
56. Wu, G.; Finn, W.L.; Dowling, J. Quasi-3D analysis: Validation by full 3D analysis and field tests on single
piles and pile groups. Soil Dyn. Earthq. Eng. 2015, 78, 61–70. [CrossRef]
57. Reese, L.C.; Wang, S.T. LPILE 4.0.; Ensoft Inc.: Austin, TX, USA, 1993.
58. Wu, G. VERSAT-P3D: A Computer Program for Dynamic 3-Dimensional Finite Element Analysis of Single Piles
and Pile Groups; Wutec Geotechnical International: Vancouver, BC, Canada, 2006.
59. 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]

© 2018 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/).