Professional Documents
Culture Documents
Souri 2015
Souri 2015
1
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Study of Static Lateral Behavior of Battered Pile Group Foundation at I-10 Twin Span Bridge
Using 3D Finite Element Modeling
Ahmad Souri1, Murad Abu-Farsakh2, M.ASCE, PhD, P.E., and George Voyiadjis3, PhD, P.E.
1
Ph.D. Graduate Student, Department of Civil and Environmental Engineering, Louisiana State
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Corresponding Author:
Murad Abu-Farsakh
4101 Gourrier Ave. Baton Rouge, LA 70808
Phone: (225) 767-9147 / Fax: (225) 767-9108
Email: cefars@lsu.edu
Page 2 of 41
2
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Abstract
In this study, the static lateral behavior of battered pile group foundation was investigated using
three-dimensional finite element (FE) analysis. The FE model was used to simulate the static lateral
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
load test that was performed during the construction of the I-10 Twin Span Bridge over Lake
Pontchartrain LA, in which two adjacent bridge piers were pulled against each other. The pier in
interest was supported by 24, 1:6 batter, 34 m-long piles in a 6x4-rows configuration. The FE model
of the battered pile group was developed in Abaqus and verified using the results from the field test.
The model utilized an advanced constitutive model for concrete, which allowed distinct behavior in
tension and compression, and introduced damage to the concrete stiffness. The soil domain comprised
of several layers in which the constitutive behavior of clay layers were modeled using the Anisotropic
Modified Cam-clay (AMCC) model, and for sands using the elastic-perfectly plastic Drucker-Prager
model. FE results showed good agreement with the results of the lateral load test in terms of lateral
deformations and bending moments. The results showed that the middle rows carried a larger share of
lateral load than the first and the last rows. The pile group resisted a maximum lateral load of 2494
metric tons at which the piles were damaged within a 6 m-zone from the bottom of the pile cap. The
edge piles carried larger internal forces and exhibited more damage compared to the inner piles. The
soil resistance profiles showed that soil layering influenced the distribution of resistance between the
soil layers. A series of p-y curves were extracted from the FE model, and then used to study the
influence of the group effect on the soil resistance. The p-y curves showed that the group effect
reduced the soil resistance in all rows, with the lowest resistance in the third row. Finally, the p-
multipliers were calculated using the extracted p-y curves, and compared to the reported p-multipliers
analysis
3
Keywords: Lateral behavior, battered pile, pile group, p-y curve, Soil-pile interaction, numerical
Page 4 of 41
4
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Introduction
The lateral behavior of piles has been well studied in the past decades, with the single pile case
mostly investigated. The lateral capacity of a single pile can be predicted by several methods
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
including the well-known p-y curve method (Matlock 1970; Reese et al. 1974), elastic solution
(Poulos and Davis 1980), the strain wedge model (Ashour et al. 2004), and the finite element (FE)
method (e.g., Muqtadir and Desai 1986; Brown and Shie 1990; Trochanis et al. 1991; Yang and
Jeremic 2002; Ahmadi and Ahmari 2009; Allahverdizadeh 2015). The p-y curve represents the
variation of soil resistance at certain depth with pile deflection. Experimental investigations showed
that the soil resistance is reduced for the pile-in-group compared to isolated single pile (e.g., Brown et
al. 1987; McVay et al. 1995; Rowesta and Townsend 1997). The phenomena is commonly called
“shadowing” or the group effect. The overlapping of stress zones from the surrounding piles
apparently weakens the soil in front of the pile. The group effect is usually quantified using the p-
The lateral behavior of vertical pile groups is experimentally studied by full-scale testing (e.g., Brown
et al. 1987; Rollins et al. 1998; Illyas et al. 2004; Rollins et al. 2005) or in centrifuge (e.g., McVay et
al. 1995; McVay et al. 1998). Single battered piles were also studied experimentally but at lesser
degree compared to single vertical piles (e.g., Tschebotarioff 1953; Murthy 1964; Meyerhof and
Yalcin 1992). Murthy (1964) reported that negative battered piles have better lateral resistance than
vertical piles and positive battered piles. Meyerhof and Yalcin (1992) studied the effect of load
inclination on the lateral resistance of single piles with different batter angles, and reached to similar
conclusion by Murthy (1964). Poulos and Davis (1980) suggested that single battered piles can be
analyzed as an equivalent case of vertical pile under inclined load. Contrary to vertical pile groups,
experimental studies on battered pile groups are rather limited in number (Ranjan et al. 1980; McVay
Page 5 of 41
5
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
et al. 1996; Zhang et al. 1999; Zhang et al. 2002). Ranjan et al. (1980) showed that a group of one
vertical pile and one battered (negative or positive) pile has greater lateral resistance than a group of
only vertical piles. McVay et al. (1996) studied the lateral behavior of battered pile groups in-
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
centrifuge at different pile spacing and pile-head fixity conditions, and found that the lateral capacity
of the battered pile group was greater than the vertical pile group at 3D and 5D spacings (D is pile
width). Zhang et al. (1999) studied the lateral behavior of single battered piles at different
inclinations. Their results showed that positive-battered piles had better resistance than vertical piles,
while negative-battered piles were weaker. They also proposed a modification for the original p-y
curves by Reese et al. (1974) to consider the effect of pile inclination. Zhang et al. (2002) investigated
the influence of dead load on the lateral capacity of battered pile groups. They found that the lateral
resistance of the battered pile groups increases with larger vertical deal loads. Meyerhof and Ranjan
(1973) studied the lateral resistance of single rigid battered piles in sand using the bearing capacity
theory. Battered piles were also studied using FE modeling. Rajashree and Sitharam (2001)
performed static and cyclic load analysis for single battered piles using hyperbolic model for the soil
reaction. Mroueh and Shahrour (2009) investigated the influence of load inclination with respect to
the pile’s axis on the axial and lateral capacity of a single battered pile in Abaqus. They showed that
the axial capacity of battered piles decreases when load inclination increase.
A full scale lateral load test was conducted on the M19 pier foundation in the I-10 Twin Span Bridge
over Lake Pontchartrain, LA (Abu-Farsakh et al. 2011a,b). The test was conducted by pulling two
adjacent piers toward each other using high strength steel strands. The maximum lateral load
achieved was 848 tons (1 ton = 9.81 kN = 1000 kg) at which one of the two strand jacks reached its
maximum stroke. One of the concerns in the test was to maintain the integrity of the structure so that
no damages or distresses occur. This was a limitation in addition to the high cost of full-scale testing.
Page 6 of 41
6
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
FE analysis provides an economical and robust tool to overcome the limitations found in experiments.
In the present study, a three dimensional FE model for the M19 pier foundation was developed in
Abaqus. The nonlinear behavior of the concrete and soil materials was considered by using advanced
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
constitutive models. Deflected shapes and bending moment results from the field test were used to
verify the model. The lateral load in the FE model was further increased beyond test loads to induce
greater deformation and fully mobilize the soil resistance. The lateral behavior of the pile group is
investigated through load-displacement curves, lateral deformation profiles in piles, damage in piles,
The lateral load test was conducted on the M19 eastbound pier of the I-10 Twin Span Bridge over
Lake Pontchartrain, LA. The pier comprised of 24 prestressed concrete battered piles arranged in 6x4-
row configuration. The adjacent rows of piles were battered at slope of 1:6, so that two rows were
battered to the east and the other two to the west, see Figure 1. Each pile was 33.5 m long and had 0.9
× 0.9 m square section. The pile spacing at the cap level was 4.3D between rows and 2.5D between
columns. The average embedded length of the piles was 26 m, and the mudline was located 3.3 m
from the bottom of the pile cap. The pile cap dimensions were 13.4 m long × 13 m wide × 2.1 m
deep. Figure 1 also shows the layout and numbering of the piles for the eastbound pier foundation.
Numbers were used for columns, and L/ML/MT/T refers to leading, middle-leading, middle-trailing,
trailing rows, respectively. A column refers to the piles arranged in the line parallel to the lateral load
direction. A row refers to the piles arranged in the transverse direction to the lateral load. Figure 1
also shows the piles instrumented with strain gages (SG) and/or in-place inclinometers (IPI). The SGs
Page 7 of 41
7
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
were initially installed in pairs at 4.9 and 6.4 m from the pile top at the time of casting. Considering
an average cutoff length of 2.0 m in the piles, the final location of the SGs was estimated at 3.0 and
4.5 m from the bottom of the pile cap. Six inclinometers were installed over the pile length at the
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
following levels: 1.5, 4.6, 7.6, 10.6, 13.7, and 19.8 m from the bottom of the pile cap.
The lateral load test was conducted by pulling the eastbound and westbound piers toward each other
using high strength steel tendons run through two 4-inch PVC pipes installed in both pile caps. The
steel strands were first anchored at the dead-end side, and then were threaded one-by-one through the
two 4-inch PVC pipes from the dead-end at the M19 eastbound pier toward the hydraulic jacks of the
live-end at the M19 westbound pier. The total lateral load was applied incrementally and designed to
reach a maximum of 907 tons. However, the maximum load achieved was 848 tons because one of
the strand jacks reached its maximum stroke. More details about the lateral load test can be found in
The subsurface soil conditions for the M19 pier were characterized by means of in-situ and laboratory
tests; this includes soil boring, laboratory testing, standard penetration tests (SPT), and cone
penetration tests (CPT). The site investigation revealed that the subsurface soil stratigraphy consists
mainly of medium-to-stiff silty clay to clay soil down to about 30.5 m deep with a layer of medium
dense sand between 11.6 and 14.9 m below the mudline. A dense sand layer was found at depths
greater than 30.5 m below the mudline. Figure 2 shows the assumed soil stratigraphy and the CPT
profile. The soil-type probability plot in Figure 2 was produced using the soil classification for CPT
sounding software “soil CPT 4.0” developed by Abu-Farsakh et al. (2008), which is based on the
probabilistic region estimation CPT classification method (Zhang and Tumay, 1999).
8
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
A three-dimensional geometry of the M19 pier and the soil domain was created in Abaqus v6.12, see
Figure 3. Owing to the symmetry in both load and geometry, only half of the M19 pier foundation
was modeled. The pile group and the soil domain models were created from two separate
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
parts/meshes, and allowed to interact with each other using the contact feature. The holes of piles
were cut out of the soil domain, and then the piles were placed into the holes (i.e. wish-in-place).
Roller-type boundary conditions were applied at the soil domain boundaries and the symmetry plane,
as shown in Figure 3. The dimensions of the soil domain are 67 m long x 24 m wide x 36.7 m deep.
The size of the soil domain was checked to be large enough to eliminate the boundary effects. The
mesh was designed to be finer near the piles, and gradually become coarser near the boundaries.
Initially, the geostatic (or in-situ) stresses were established by applying gravity load on both soil and
pile models, and then performing a static equilibrium step. Once equilibrium was achieved, the lateral
load was applied incrementally on the side of the pile cap as: 258, 349, 440, 535, 716, 791, 848, 1134,
1587, 2041, 2494 tons. The lateral load increments up to 848 tons followed the field test loading
sequence. The soil-pile interaction was introduced using Abaqus’s contact feature, which allows soil-
pile separation and models frictional behavior at the interface. The frictional behavior was governed
by the classical Coulomb friction law with coefficient of friction µ = 0.42 (or δ = 23 deg.). Different
values for µ were also considered in FE trial tests; however, the influence was negligible on the
lateral behavior of the pile group. The latter was similarly observed in the FE study by Mroueh and
Shahrour (2007).
9
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
The pile group model comprised of 10206 linear continuum brick elements (C3D8R). Two
constitutive laws for the concrete were used at the same time: linear elastic, and concrete damaged
plasticity (CDP) (Abaqus 2011). The main advantage in CDP is it allows distinct behavior in tension
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
and compression, which is a major aspect for concrete material. The CDP model is an elastoplastic
model, which features: linear elasticity, the yield surface by Lubliner et al. (1989) for plasticity, and
material damage. The yield surface by Lubliner et al. (1989) is based on the Drucker-Prager yield
=
3 + + 〈 〉 − 〈− 〉 − Eq. 1
where α and β are dimensionless constants related to the ratio of initial equibiaxial compressive yield
stress to the uniaxial compressive yield stress, γ is a dimensionless constant that controls the shape of
yield surface in the deviatoric plane, is the second devitoric invariant of stress, is the first
invariant of stress, σmax is the maximum principal stress, and σc is the compressive cohesion stress.
The flow rule in the CDP model is non-associative, and the plastic potential is based on the classical
Mohr-Coulomb yield function with the angle of dilation (ψ) replacing the angle of internal friction
(φ). Damage (i.e. degradation of concrete stiffness) is applied in the CDP model once the failure/yield
strength is exceeded. It is controlled by the scalar damage variable (d) which reduces the elastic
stiffness as follows:
Where , are the damaged and undamaged elasticity matrices.
Page 10 of 41
10
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
The scalar damage variable is estimated from the distinct damage variables in tension (dt) and
compression (dc). The damage variables (dc, dt) versus strain and stress-strain curves needed for use
in the CDP model were obtained using the analytical concrete models and available modeling data
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
from the literature (Mander et al. 1988; Gupta and Maesterini 1990; Jankowiak and Lodygowski
The linear elastic law was chosen for the pile cap and the upper 1.5 m of the piles, and the CDP
model was used for the remaining portion of the piles. This assumption was made to avoid the issue
of premature damage in the concrete elements at the pile cap-piles connection at small lateral load
magnitude (< 500 kips), which, in consequence, resulted in the early termination of the numerical
solution by the FE solver. The 1.5 m distance came after the choice of three elements (the element
size was 0.5 m) below the pile cap-piles connection, which were assigned with linear elastic model to
overcome the issue. The elastic modulus for the concrete was estimated using the well-known
Where, ′ is the concrete compressive strength after 28 days in MPa (the estimated concrete strength
was 55 MPa).
The value of Poisson’s ratio for both linear elastic and CDP models was 0.2.
The piles were built with prestressed steel strands as the main reinforcement. Each steel strand was
made of seven-wire, low relaxation Grade 270 bundled steel strands, and had a diameter of 1.5 cm.
The total number of strands per pile was 36 spaced at 8.4 cm in the pile’s square section, and were 7.6
Page 11 of 41
11
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
cm away from pile’s face. The estimated prestress force per strand after considering the prestress
In the current study, the steel reinforcement was modeled using embedded shell elements (S4R). The
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
embedded element technique is used to specify that an element or group of elements is/are embedded
in “host” elements. For a node of an embedded element, the translational degrees of freedom of the
embedded nodes are constrained to the interpolated values of the corresponding degrees of freedom
of the host element (Abaqus 2011). For each pile, four layers of shell elements were used in a box
shape, see Figure 4. For each shell layer, the configuration (spacing, orientation, number of bars),
properties (diameter, constitutive law), and the prestress force for steel bars was specified following
the pile’s section design. The von Mises elastic-perfectly plastic model was used to describe the
behavior of the steel material. The elastic properties for were Young’s modulus "( = 200 GPa,
Poisson’s ratio * = 0.2, and the yield stress was 1.6 GPa (ASTM A416).
Soil models
The soil domain was composed of eight layers: six clay and two sand layers, following the in-situ soil
stratigraphy. The soil domain was modeled using linear continuum brick elements (C3D8R) with a
total of 72160 elements. The constitutive laws for the soil layers were the Anisotropic Modified Cam-
clay (AMCC) model for clays (Dafalias et al. 2006), and the Drucker-Prager elastic-perfectly plastic
model for sands. The AMCC model is based on the modified Cam-clay (MCC) model, which is
derived from the critical state soil mechanics (Roscoe and Burland 1968). The AMCC model
incorporates a rotational hardening rule for the yield surface, which allows simulating the anisotropy
of naturally deposited soils (e.g.,Voyiadjis and Song 2000, Wei 2004, Elias 2008, Abu-Farsakh et al.
2015). The elastic bulk stiffness of the soil in AMCC model is proportional to the mean pressure,
which makes it increase linearly with depth. For plastic behavior, an associative flow rule was
Page 12 of 41
12
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
adopted in AMCC model, which means the direction of the plastic strain rate follows the direction of
the normal to the yield surface f, see Figure 5 . The yield function for general stress state in AMCC
, ,
= -./0 − 1 /0 2-./0 −1 /0 2 − 3 − /0 /0 11 − 1 Eq. 4
where p is the mean stress, ./0 is the deviatoric stress tensor, M is the slope of the critical state line in
triaxial stress space, 1 is the preconsolidation pressure, and /0 is the non-dimensional deviatoric
anisotropy tensor. More details for the AMCC model formulation and features can be found in
Dafalias et al. (2006). The AMCC model was written in FORTRAN code and incorporated in Abaqus
For the sand layers, the elastoplastic Drucker-Prager (DP) model was used. Similar to the AMCC
model, the elastic stiffness is proportional to the mean pressure, and the plastic behavior is triggered
by the hyperbolic DP yield surface (Abaqus 2011). The parameters for both AMCC and DP models
and the soil layering are summarized in Table 1. The parameters for AMCC model are: 4 is the slope
of unload-reload curve in e-log p plot, 5 is the slope of normal consolidation line in e-log p plot, Ko is
:,;<
the coefficient of lateral earth pressure (6 /8 ), /9/ is the degree of initial anisotropy ( /9/ = ),
;<
x and C are constants which control the rate of evolution of anisotropy, is the DP model friction
angle, = is the cohesion in DP model. The parameters (4, 5, Ko, ) were estimated with the aid of the
CPT results and correlations found in the reference for CPT (Robertson and Cabal 2015). For the DP
model, = represents the cohesion in the sand material and was assumed 0.5 kPa to avoid numerical
problems. The constants x and C were reasonably assumed following Dafalias et al. (2006).
13
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
The FE model was verified using the results from the lateral load test conducted on the M19 pier
(deformation profiles from IPIs, and bending moments from SGs). Additional lateral load was applied
in the FE model to induce more deformation and study the aspects of the lateral behavior of the
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
battered pile group under extreme load conditions. In addition to deflected shapes and bending
moments, internal forces such the axial force in the piles, soil resistance, and p-y curves were
extracted.
Load-displacement curves
Figure 6 shows the lateral load versus displacement for the pile cap and the average for each row in
the pile group. Recall that the maximum lateral load from the field test was 848 tons. Good agreement
between the field test results and the FE model results is observed. The cap displacement from the
experiment at 848 tons was 1.7 cm compared to 1.9 cm from the FE model. In the FE simulation, the
lateral load was further increased until the piles were severely damaged and unable to sustain
additional load. The maximum lateral load achieved in the FE model was 2494 tons, which resulted
in pile cap displacement of 7.3 cm. The load-displacement curve for the pile cap exhibited slight non-
linearity, which is attributed to the nonlinear behavior of the piles and the soil as well. Figure 6 also
depicts the average lateral load carried by the piles in each row, which shows that the middle rows
(ML, MT) carried larger lateral load compared to the leading (L) and trailing (T) rows. It is noticed
that rows ML and MT carried approximately similar load percentage, which also applies to rows L
and T. At 2494 tons, the average lateral load percentage carried by each row was 22%, 31%, 28%,
and 20% in rows L, ML, MT, and T, respectively. The slight difference between rows ML and MT
was due to the tension-damage in row MT, which initiated earlier at a total lateral load of 1587 tons.
The distribution of lateral load per row observed here is different from the case of vertical piles
groups. For vertical pile groups, the largest share of lateral load is usually carried by the leading row
Page 14 of 41
14
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
followed by the 2nd, 3rd, and 4th rows, respectively (e.g. Brown et al. 1988; Ruesta and Townsend
1997; McVay et al. 1998). Such difference is attributed to the effect of the batter angle, which
modifies the mechanism of lateral resistance in battered piles. In the vertical case, the pile resists the
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
lateral load by carrying shear load only, while in the battered case the pile carries both axial and shear
loads. As will be shown later, the axial load transferred to the middle rows ML and MT was
approximately three times larger than the axial loads in rows L and T.
The CDP constitutive law introduces damage by reducing the concrete modulus. This occurs in
tension and/or compression when the strength limit is exceeded. When damage is introduced, the
concrete material becomes weaker and, consequently, stresses are transferred to the embedded
reinforcing steel. In the current FE study, only damage due to tension was observed following the fact
that the tensile strength is significantly smaller (10 times less) than the compressive strength. The first
elements to get damage were in row MT piles at 1587 tons, and located 1.5 m below the pile cap.
This was expected since material damage was not a property of the elements within the top 1.5 m of
the piles.
The damage parameter represents the reduction to the elastic stiffness of the material, and assumes
values within the range 0 < d < 1.0. A zero value for means the material is damage free, while a
value of 1.0 means the material stiffness vanished. In numerical modeling, a value of 1.0 is not
applicable and introduces numerical problems; therefore, the maximum value of is typically below
0.9. Figure 7 shows the damage progression in the piles with lateral load. Damage first appeared in
row MT at 1587 tons, followed by rows T and L at 2041 tons, and finally in row ML at 2494 tons. At
2494 tons, the elements located at 1.5 m below the pile cap in the trailing rows (MT and T) were
Page 15 of 41
15
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
severely damaged with > 0.7 (notice the dark colored areas in the figure), and the damaged area
expanded to the elements located further below 1.5 m. Piles in the leading row (L) exhibited lesser
degree of damage with < 0.5 compared to the trailing rows. The lowest damage level was observed
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
in row ML with < 0.2. It should be noted that the FE solver aborted the solution at a total lateral
load less than 2721 tons (the load increment in the FE solution was 226 tons). Therefore, it was
assumed that the last load increment completed in the FE solution (2494 tons) was the maximum
The damage sequence in the rows can be explained by investigating the axial load carried in each
pile. Figure 8a summarizes the variation of the axial load in each pile with increasing group lateral
load. In Figure 8a, positive axial load means compression while negative is tension. The axial loads in
the leading rows (L and ML) were in compression and increased with increasing lateral load, while in
the trailing rows (T and MT) the axial loads increased in tension. Additionally, it is noticed that the
edge piles (column 1) were subjected to the largest magnitude of axial loads among other piles in the
same row. For the damage sequence, piles in row MT were the first to get tension damage due to the
relatively large tensile axial load, which in combination with bending moment initiated tension-
damage earlier than other rows. The second to experience tension-damage was the trailing row (T),
which was subjected to lesser tensile axial load. Lastly, the leading rows (L and ML) exhibited
damage at later stages due to the compressive axial load, which counteracts the action of the bending
moment.
Moreover, within the same row, the edge piles (column 1) received more damage compared to the
inner piles (columns 2 and 3) due to the larger bending moment. For example, Figure 8b shows the
bending moments developed at the top of piles in the leading row (L). The edge pile (L1) carried
Page 16 of 41
16
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
larger bending moment (8% more) compared to the inner piles (L2 and L3). Similar behavior was
The lateral deformation profiles for piles L6, ML2, MT2, T1, T2 are shown in Figure 9. Note that the
results of pile L6 from the full-scale test are compared to the mirror or symmetry pile L1 from the FE
model. The deflected shapes are shown for two load levels: 848 tons from the field test and FE, and
2494 tons from FE analyses only. Good agreement can be observed between the field test and FE
results at 848 tons. It can be noticed that the majority of the deformation occurred within the clay
layers (3.3-14.9 m), and diminished at the (medium dense) sand layer located at 14.9-18.3 m below
the mudline.
The lateral deformation for column 2 and row ML piles at 2494 tons are depicted in Figure 10a and
Figure 10b, respectively. For column 2 piles, it is noticed that the deflections for piles ML2 and MT2
(middle rows) was slightly greater than piles L2 and T2. Similarly, for row ML, piles ML2 and ML3
had slightly greater deformation compared to the edge pile ML1, Figure 10b. The latter observation is
also valid for the other columns / rows but not shown here for brevity.
The bending moments from the field were deduced from the strain gage readings at 3.0 and 4.5 m
from the bottom of the pile cap in each of the instrumented piles. For 3D solid continuum elements,
Abaqus provides output for cross sections with bending moment, axial force, and shear force results.
It performs the integration of stresses for cross sections defined in the mesh, and append the results to
the output file. To obtain the bending moment profiles, multiple cross sections were defined in the
pile group mesh at the pile cap-piles connection and at 1.5, 4.6, 7.6, 10.7, 13.7, 16.8, 19.8, 22.9, and
Page 17 of 41
17
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
25.9 m from the bottom of the pile cap. A comparison between the field results and the FE bending
moments at lateral loads of 258 and 791 tons is shown in Figure 11. Good agreement is observed
between the field measurements and FE results. The bending moment profiles from FE show that the
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
inflection point (i.e. zero moment) was located within 6.4-7.6 m from the bottom of the pile cap. The
location of the maximum positive moment was at the pile cap connection, while the maximum
negative moment was located at 12-13.7 m from the bottom of the pile cap. Within the same column,
the leading and trailing rows (L and T) had shallower inflection and maximum negative moment
points compared to the middle rows (ML and MT), as shown in Figure 12a.Within the same row, the
inflection and maximum negative moment points were slightly shallower for the edge piles (column
1), see for example the bending moment profiles for row L piles in Figure 12b. It is also noticed that
both points (inflection and maximum negative moment) shift deeper with increasing lateral load, as
Soil resistance
The soil resistance is investigated with the aid of soil resistance profiles and p-y curves. The soil
resistance profiles were obtained using the traditional method by double differentiating the bending
moment curve, which was fitted using high-order polynomial fit (e.g. Wilson 1998, Illyas et al. 2004,
Nip and Ng 2005, Abu-Farsakh et al. 2011a). Figure 13 presents the soil resistance profiles for
column 2 piles. It should be noted that the reference depth in the following discussion and in the
figures is taken from the mudline unless otherwise specified. The profiles show that the soil
resistance changed direction at an approximate depth of 13.7 m in all piles. The soil resistance
increased with increasing lateral load in all piles, but at different rates. The largest increase in soil
resistance was in the leading row pile L2, which is mainly due to the group effect. The resistance was
also influenced by soil layering; this can be noticed in the larger resistance in layer 2 (stiff clay, 4.5-
Page 18 of 41
18
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
7.6 m) compared to layer 3 (medium stiff clay, 7.6-11.6 m). In addition, a sharp increase in the
resistance at 2494 tons in the sand layer is noticed, which can be attributed to the larger stiffness of
The soil resistance is also investigated using the p-y curves. A p-y curve represents the soil resistance
per unit pile-length versus displacement at certain depth. For brevity, only column 2 and row ML p-y
curves are presented here, and the following discussion applies for other columns/rows as well.
Figure 14 shows the p-y curves for column 2 piles at different depths. The figure shows that the p-y
curves for pile L2 in the leading row were stiffer than those in other rows (ML2, MT2, T2). In
addition, the p-y curves for piles ML2 and T2 were approximately similar and stiffer than pile MT2,
Figure 15 compares the p-y curves for column 2 piles and row ML piles at depth of 3 m. The
comparison also includes the p-y curve for an isolated single vertical pile, single positive battered
pile, and single negative battered pile. Those p-y curves were obtained from a separate FE models for
each single pile case. These FE models maintained the same pile dimensions, pile head condition (i.e.
fixity), batter angle for battered piles, material properties (soil, steel, and concrete), constitutive
models, soil layering, and embedded shell elements for steel reinforcement. The p-y curves for single
pile cases show that the single positive battered pile was subjected to slightly higher soil resistance
compared to the vertical and negative battered cases, which can be attributed to the effect of the batter
angle.
Within the same column (Figure 15a), the soil resistance was notably reduced for all piles in the
group compared to the single pile case, again, due to the group effect. The largest reduction in the soil
resistance was in row MT followed by rows T then ML, and finally row L. Within the same row
Page 19 of 41
19
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
(Figure 15b), the soil resistance was largest in the edge pile ML1 followed by ML2 and ML3, which
The influence of the group effect is quantitatively evaluated using the p-multiplier concept. Brown et
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
al. (1988) defined the p-multiplier as the factor that reduces the p-y curve from a single pile case to
give the p-y curve for a pile-in-group configuration. The typical value of the p-multiplier is less than
unity. The p-multiplier is affected by pile group arrangement (i.e. number of rows and columns), piles
spacing (row and column spacing), and soil properties (e.g., McVay et al. 1995; Chandrasekaran et al.
2010; Ashour and Ardalan 2011). Table 2 summarizes the p-multipliers for all piles in the current
study obtained using the p-y curves at depth of 3 m, in which the p-y curve for the single vertical pile
was taken as the reference case. The p-multipliers in Table 2 for each pile are the average values of p-
multipliers calculated at displacements of 3, 4, and 5 cm, following the suggestions from previous
studies. McVay et al. (1998) reported that 2-3 cm displacement is needed to develop the average p-
multipliers. Ashour and Ardalan (2004) showed that the p-multiplier value stabilizes at
The p-multipliers were highest in the edge piles (column 1) followed by the inner columns 2 and 3 for
which the multipliers were almost the same. The average p-multiplier was largest in the leading row
L followed by rows ML, T, and MT, respectively. The p-multipliers for the current study are
comparable to those from experimental studies on vertical pile groups with 4-rows (e.g., Ilyas et al.
2004; Chandrasekaran et al. 2010). The p-multipliers from the study by Ilyas et al. (2004) were 0.65,
0.49, 0.42, 0.46, and from the study by Chandrasekaran et al. (2010) were 0.76, 0.56, 0.46, 0.54 for
the 1st, 2nd, 3rd, and 4th row, respectively. The most notable similarity is in the value of the p-
multiplier for the 3rd row, which was the lowest multiplier among other rows. The slightly higher p-
Page 20 of 41
20
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
multipliers for the battered pile group for the current study can be attributed to the larger row spacing,
which was 4.3D in the current study versus 3D in the referenced studies.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Conclusions
The lateral behavior of 6x4-row battered pile group foundation was studied using three-dimensional
finite element modeling in Abaqus. The FE model utilized advanced elastoplastic constitutive laws
for the concrete and soil materials. The constitutive model for the concrete allowed distinct behavior
in tension and compression, and introduced stiffness degradation once the failure limit stress is
exceeded. The constitutive model for the clay layers was the Anisotropic Modified Cam clay, and the
classical Drucker-Prager adopted for the sand layers. The FE model was verified using the results of
full-scale static lateral load test that was conducted on the M19 eastbound pier of the I-10 Twin Span
Bridge over Lake Pontchartrain, LA. Based on the findings of this study, the following conclusions
can be drawn:
• Good agreement was obtained between the field results and FE predictions in terms of lateral
• The lateral load distribution indicated that the middle rows carried 8-10% more lateral load
than the leading and trailing rows. The latter was different from the load distribution in
vertical pile groups in which the leading row carries the largest share of applied lateral load.
• The comparison between the piles in the same row showed that the edge piles carried larger
axial loads and bending moments than the other piles in the same row. Also, the edge piles
21
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
• Bending moment profiles showed that the maximum positive moment was located at the pile
cap connection, and the maximum negative moment within 10D from the mudline. The
inflection point (or zero bending moment) was found within 5D from the mudline.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
• At large lateral loads, only tensile failure in the concrete piles was observed, and was initiated
near the connection with the pile cap in the trailing rows with negative batter.
• The soil resistance profiles showed that the soil resistance was influenced by soil layering.
This was observed in the higher resistance in the stiff clay and sand layers compared to the
soft and medium stiff clay layers. In addition, the profiles showed that the influence depth for
• When compared to isolated single pile case, the soil resistance in the pile group was notably
reduced due to the group effect. The group effect was quantitatively represented by the p-
multipliers. The average p-multipliers per row for the battered pile group were 0.80, 0.64,
0.49, 0.59 for the 1st, 2nd, 3rd, and 4th row, respectively. The p-multipliers values were
comparable to the p-multipliers reported in previous studies for vertical pile groups.
Page 22 of 41
22
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
References
Abu-Farsakh, M., Zhang, Z., Tumay, T., and Morvant, M. 2008. Computerized Cone Penetration Test
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Abu-Farsakh, M., Yu, X., Pathak, B., Alshibli, K., and Zhang, Z. 2011a. Field Testing and Analyses
of a Batter Pile Group Foundation under Lateral Loading. Transportation Research Record:
Abu-Farsakh, M., Yoon, S., Ha, D., Marr, W. A., Zhang, Z., and Yu, X. 2011b. Development of a
Substructure Instrumentation System at the New I-10 Twin Span Bridge and its Use to
Investigate the Lateral Behavior of Batter Piles. Geotechnical Testing Journal, 34(4), 332-343.
Abu-Farsakh, M., Rosti, F., and Souri, A. 2015. Evaluating Pile Installation and The Following
Thixotropic And Consolidation Setup By Numerical Simulation For Full Scale Pile Load Tests.
Ahmadi, M., and Ahmari, S. 2009. Finite-Element Modelling of Laterally Loaded Piles in Clay.
Allahverdizadeh, P. 2015. Group Effects on Piles Behavior under Lateral Loading. Proceedings of
American Concrete Institute. 2011. Building Code for Structural Concrete and Commentary (ACI
American Society for Testing and Materials (ASTM). 2012. Specification for Steel Strand, Uncoated
Seven-Wire for Prestressed Concrete. ASTM International. Retrieved December, 2014 from
http://www.astm.org/doiLink.cgi?A416A416M
Page 23 of 41
23
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Ashour, M., and Ardalan, H. 2011. Employment of the p-Multiplier in Pile-Group Analysis. Journal
Ashour, M., Pilling, P., and Norris, G. 2004. Lateral Behavior of Pile Groups in Layered Soils.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Brown, D., Reese, L., and O’Neill, M. 1987. Cyclic Lateral Loading of a Large‐Scale Pile Group.
Brown, D., Reese, L., and O’Neill, M. 1988. Lateral Load Behavior of Pile Group in Sand. Journal of
Brown, D., and Shie, C. 1990. Three Dimensional Finite Element Model of Laterally Loaded Piles.
Chandrasekaran, S., Boominathan, A., and Dodagoudar, G. 2009. Group Interaction Effects on
136(4), 573–582.
Cicekli, U., Voyiadjis, G., and Abu Al-Rub, R. 2007. A Plasticity and Anisotropic Damage Model for
Dafalias, Y. 1986. An Anisotropic Critical State Soil Plasticity Model. Mechanics Research
Dafalias, Y., Manzari, M., and Papadimitriou, A. 2006. SANICLAY: Simple Anisotropic Clay
Plasticity Model. International Journal for Numerical and Analytical Methods in Geomechanics,
30(12), 1231–1257.
Elias, M. 2008. Numerical Simulation of Pile Installation and Setup. Ph.D. dissertation, The
Gupta, A., and Maestrini, S. 1990. Tension‐Stiffness Model for Reinforced Concrete Bars. Journal of
24
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Ilyas, T., Leung, C., Chow, Y., and Budi, S. 2004. Centrifuge Model Study of Laterally Loaded Pile
Jankowiak, T., and Lodygowski, T. 2005. Identification of Parameters of Concrete Damage Plasticity
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Lubliner, J., Oliver, J., Oller, S., and Oñate, E. 1989. A Plastic-Damage Model for Concrete.
Mander, J., Priestley, M., and Park, R. 1988. Theoretical Stress‐Strain Model for Confined Concrete.
Matlock, H. 1970. Correlations for Design of Laterally Loaded Piles in Soft Clay. Proc., 2nd Annual
Meyerhof, G., and Ranjan, G. 1973. The Bearing Capacity of Rigid Piles under Inclined Loads in
Meyerhof, G., and Yalcin, A. 1993. Behavior of Flexible Batter Piles Under Inclined Loads In
McVay, M., Casper, R., and Shang, T. 1995. Lateral Response of Three-Row Groups in Loose To
Dense Sands at 3D And 5D Pile Spacing. Journal of Geotechnical Engineering, 121(5), 436–
441.
McVay, M., Shang, T., and Casper, R. 1996. Centrifuge Testing of Fixed-Head Laterally Loaded
Battered and Plumb Pile Groups in Sand. Geotechnical Testing Journal, 19(1), 41–50.
McVay, M., Zhang, L., Molnit, T., and Lai, P. 1998. Centrifuge Testing of Large Laterally Loaded
1016–1026.
Page 25 of 41
25
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Mroueh, H., and Shahrour, I. 2009. Numerical Analysis of the Response of Battered Piles to Inclined
Pullout Loads. International Journal for Numerical and Analytical Methods in Geomechanics,
33(10), 1277–1288.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Murthy, V. 1964. Behavior of Batter Piles Subjected to Lateral Loads. Ph.D. dissertation, 1ndian
International Journal for Numerical and Analytical Methods in Geomechanics, 10(1), 41–58.
Nip, D., and Ng, C. 2005. Back-analysis of laterally loaded bored piles. Proceedings of the Institution
Poulos, H., and Davis, E. 1980. Pile Foundation Analysis and Design. John Wiley & Sons, Inc., USA.
Rajashree, S., and Sitharam, T. 2001. Nonlinear Finite-Element Modeling of Batter Piles under
Ranjan, G., Ramasamy, G., and Tyagi, R. 1980. Lateral Response of Batter Piles and Pile Bents in
Reese, L., Cox, W., and Koop, F. 1974. Analysis of Laterally Loaded Piles in Sand. Proc. 6th Annual
Robertson, P., and Cabal, K. 2015. Guide to Cone Penetration Testing for Geotechnical Engineering.
Rollins, K., Peterson, K., and Weaver, T. 1998. Lateral Load Behavior of Full-Scale Pile Group in
Rollins, K., Lane, J., and Gerber, T. 2005. Measured and Computed Lateral Response of a Pile Group
26
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Roscoe, K., and Burland, J. 1968. On The Generalized Stress-Strain Behaviour of ‘Wet’ Clay.
Engineering plasticity (eds. J. Heyman and F.A. Leckie), Cambridge University Press, London,
England, 535-609.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
Ruesta, P., and Townsend, F. 1997. Evaluation of Laterally Loaded Pile Group at Roosevelt Bridge.
Trochanis, A., Bielak, J., and Christiano, P. 1991. Three‐Dimensional Nonlinear Study of Piles.
Tschebotarioff, G. 1953. The Resistance to Lateral Loading of Single Piles and of Pile Groups.
Voyiadjis, G., and Song, C. 2000. Finite Strain, Anisotropic Modified Cam Clay Model with Plastic
Wei, L. 2004. Numerical Simulation and Field Verification of Inclined Piezocone Penetration Test in
Cohesive Soils. Ph.D. dissertation, Louisiana state university, Baton Rouge, LA.
Wilson, D. 1998. Soil-Pile-Superstructure Interaction in Liquefying Sand and Soft Clay. Center for
Yang, Z., and Jeremić, B. 2002. Numerical Analysis of Pile Behaviour under Lateral Loads in
Layered Elastic–Plastic Soils. International Journal for Numerical and Analytical Methods in
Zhang, L., McVay, M., and Lai, P. 1999. Centrifuge Modelling of Laterally Loaded Single Battered
Zhang, L., McVay, M., Han, S., Lai, P., and Gardner, R. 2002. Effects of Dead Loads on the Lateral
28
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Stiff Clay 4.5-7.6 18.7 0.20 0.85 0.111 1.1 0.01 0.12 1.33 4 - - -
Medium Stiff Clay 7.6-11.6 17.0 0.25 0.73 0.219 1.0 0.02 0.13 1.33 4 - - -
Medium Sand 11.6-14.9 18.8 0.38 0.70 - - - - - - 0.003 53 0.5
Stiff Clay 14.9-21.3 17.7 0.20 0.65 0.271 1.2 0.01 0.12 1.33 4 - - -
Stiff Clay 21.3-24.5 19.1 0.20 0.65 0.271 1.2 0.01 0.12 1.33 4 - - -
Stiff Clay 24.5-30.1 20.1 0.20 0.60 0.363 1.2 0.01 0.12 1.33 4 - - -
Dense Sand > 30.1 19.4 0.40 0.60 - - - - - - 0.003 58 0.5
Row
L ML MT T
Column 1 0.96 0.78 0.57 0.72
Column 2 0.74 0.59 0.46 0.55
Column 3 0.70 0.57 0.43 0.51
Average 0.80 0.64 0.49 0.59
Page 29 of 41
29
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
List of Figures
Figure 1: Schematic of (a) lateral load test, (b) piles numbering and instrumentation (after Abu-
Farsakh et al. 2011a)
Figure 2: Soil stratigraphy and CPT profile (1 ft = 0.3 m)
Figure 3: Geometry and mesh of the FE model for the battered pile group
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
20.8 m
(b)
0
500
1000
1500
2000
2500
3000
0
FE Pile cap
Expr. Pile cap
FE Avg Row T
FE Avg Row L
FE Avg Row MT
FE Avg Row ML
At 2041 tons
At 1587 tons
At 2494 tons
34
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
400 400
ML2
ML3
100 250
L
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by CAMBRIDGE UNIVERSITY LIBRARY on 12/27/15
0 200
T
-100 MT1 150
MT2
-200 MT3 100
T1 MT
-300 T2 50
T3
-400 0
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000
Group lateral load (tons) Group lateral load (tons)
Figure 8: Variation of piles’ internal forces with lateral load (a) axial load, (b) bending moment in
row L piles
Page 35 of 41
35
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
8
Pile L6 Pile ML2 Pile MT2
10
Expr. 848 tons Expr. 848 tons Expr. 848 tons
12 FE 848 tons FE 848 tons FE 848 tons
14 FE 2494 tons FE 2494 tons FE 2494 tons
2
Mudline Mudline
4
6
8
Pile T1 Pile T2
10
Expr. 848 tons Expr. 848 tons
12 FE 848 tons FE 848 tons
14 FE 2494 tons FE 2494 tons
16 Sand Sand
18
20
22
22
20
18
16
14
12
10
8
6
4
2
0
0
1
(a)
2
3
4
5
Mudline
6
Displacement (cm)
T2
L2
7
MT2
ML2
Sand
FE 2494 tons
8
0
1
(b)
2
3
4
5
Mudline
6
Displacement (cm)
7
ML3
ML2
ML1
FE 2494 tons
Sand
8
Figure 10: Lateral deformation profiles for (a) column 2 piles, (b) row ML piles
36
Page 36 of 41
Page 37 of 41
37
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
10
12
14
16 Sand Sand
18
20 Pile L3 Pile M L 3
22 E xpr. 258 tons Expr. 258 tons
24 E xpr. 791 tons Expr. 791 tons
FE 258 tons FE 258 tons
26 FE 791 tons FE 791 tons
28
Bending moment (ton-m) Bending moment (ton-m)
-50 -25 0 25 50 75 100 -50 -25 0 25 50 75 100
0
2 Mudline Mudline
Depth from pile cap bottom (m)
4
6
8
10
12
14
16 Sand Sand
18
20 Pile M T 2 Pile T 1
22 Expr. 258 tons Expr. 258 tons
Expr. 791 tons Expr. 791 tons
24
FE 258 tons FE 258 tons
26 FE 791 tons FE 791 tons
28
Figure 11: Bending moments from strain gages and FE bending moment profiles for instrumented
piles
Page 38 of 41
38
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
10
12
14
16 Sand Sand Sand
18
20 FE 848 tons Pile L2
FE 848 tons
L2 FE 258 tons
22 L1
ML2 FE 848 tons
24 L2
MT2 FE 1587 tons
26 L3
T2 FE 2494 tons
28
Figure 12: Bending moment profiles for (a) column 2 piles, (b) row L piles, (c) pile L2
Page 39 of 41
39
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
8
10
12
Sand Sand
14
16
18 Pile L2 Pile ML2
20 FE 848 tons FE 848 tons
22 FE 1587 tons FE 1587 tons
24 FE 2494 tons FE 2494 tons
26
Soil resistance (kN/m) Soil resistance (kN/m)
-120 -60 0 60 120 180 240 -120 -60 0 60 120 180 240
0
2
4
Depth from mudline (m)
6
8
10
12
14 Sand Sand
16
18 Pile MT2 Pile T2
20 FE 848 tons FE 848 tons
22 FE 1587 tons FE 1587 tons
24 FE 2494 tons FE 2494 tons
26
40
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
80
Pile L2 Pile ML2
70 Depth 1.8 m Depth 1.8 m
Depth 3.0 m Depth 3.0 m
40
30
20
10
0
80
Pile MT2 Pile T2
70 Depth 1.8 m Depth 1.8 m
Soil resistance - p (kN/m)
40
30
20
10
0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Displacement - y (cm) Displacement - y (cm)
0
10
20
30
40
50
60
0
1
2
Single vertical
p-y curves at 3.0 m
3
4
Displacement - y (cm)
5
T2
L2
(a)
MT2
ML2
6
0
1
ML3
ML2
ML1
2
Single vertical
p-y curves at 3.0 m
3
4
Figure 15: p-y curves comparison for (a) column 2 piles, (b) row ML
Displacement - y (cm)
5
(b)
6
41