Transportation Geotechnics 1 (2014) 63–73

Contents lists available at ScienceDirect

Transportation Geotechnics
journal homepage: www.elsevier.com/locate/trgeo

Load transfer of hollow Fiber-Reinforced Polymer (FRP) piles

in soft clay
J. Giraldo a, M.T. Rayhani b,⇑
a
Carleton University, Civil and Environmental Engineering, Ottawa, Ontario, Canada
b
Dept. of Civil and Environmental Engineering, Carleton University, Ottawa K1S 5B6, Canada

a r t i c l e i n f o a b s t r a c t

Article history: This paper describes results of a series of small scale static axial and lateral pile load tests
Received 3 February 2014 carried out on model Fiber Reinforced Polymer (FRP) piles and steel piles. The goal of this
Revised 6 March 2014 study was to establish the geotechnical frictional performance of FRP piles in clayey soils
Accepted 13 March 2014
compared to hollow steel piles. Hollow FRP piles were manufactured using both carbon
Available online 20 March 2014
and glass fibers with different fiber orientations. Pile load testing was carried out in large
undisturbed clay samples using small scale piles of 55 mm in diameter by 700 mm in
Keywords:
length. The axial bearing capacity of FRP piles were shown to be 5-40% higher compared
Pile
Interface
to the steel pile. Under lateral loading, the FRP piles presented increased deflections com-
Load test pared to the steel counterpart. Based on the results presented here, FRP piles present suit-
Shear strength able characteristics to act as a load bearing member based on axial geotechnical capacity.
FRP The structural integrity and excessive deflections need further investigation.
Composite Ó 2014 Elsevier Ltd. All rights reserved.

Introduction technology. In recent decades researchers and industry

partners have strived to develop and introduce the use of
Pile foundations are typically constructed using three FRP piles for waterfront and harbour fender pile applica-
types of materials or combinations thereof: steel, concrete tions. Iskander and Hassan (1988) highlighted the predom-
and wood. However, the use of these materials faces sev- inant use of Fiber Reinforced Polymer (FRP) piles for
eral challenges when installed in harsh environments such fendering and light load bearing applications in harbours
as waterfront areas or contaminated and harsh soil condi- and piers. They cited the lack of an established track re-
tions. Steel and concrete are subject to corrosion and deg- cord, the lack of design guidelines and high initial material
radation, leading to significant reduction of pile cross- costs as reasons for the limited use in other piling applica-
sectional area and therefore impacting the pile’s perfor- tions. Typical FRP piles are fabricated using either a high
mance and structural integrity, and drastically reducing density polyethylene (HDPE) matrix reinforced with steel
service life. In order to address these adverse effects, or glass fiber, or fabricated from carbon or glass fiber in
researchers have been exploring the use of alternate mate- the form of hollow shells (Guades et al., 2012). Carbon or
rials such as glass or carbon fibers and plastic composites glass FRP hollow shells are also in-filled with concrete, in
as piling alternatives. Development and use in other indus- which case the outside fiber shell acts as the pile reinforce-
tries has driven the price of production down to an attrac- ment and protect the concrete. In addition the confinement
tive price point and produced a commercially viable enhances the structural strength and stiffness of the pile
(Fam and RizKalla, 2001a,b).
FRP pile performance and soil–pile interface behaviour
⇑ Corresponding author. Tel.: +1 61352026008890.
has been poorly characterized due to the lack of
E-mail address: mohammad.rayhani@carleton.ca (M.T. Rayhani).

http://dx.doi.org/10.1016/j.trgeo.2014.03.002
2214-3912/Ó 2014 Elsevier Ltd. All rights reserved.
64 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73

widespread use of the material. Researchers such as O’Rou- eral deflections. In addition, FRP material influence, com-
rke et al. (1990) carried out direct shear box interface char- paring both carbon and glass fiber piles, fiber orientation
acterization tests using sands and HDPE geotextiles where and soil interaction are explored to determine optimal con-
it was found that the interface shear strength increased ditions for increased pile performance under field condi-
with sand density and decreased with surface roughness. tions. This was achieved by conducting a series of small
Frost and Han (1999) conducted a series of interface char- scale pile load tests where model piles were driven in
acterization tests involving FRP interfaces and sand in undisturbed clay samples collected in industrial sized steel
which they identified various parameters that affected drums. Following driving, axial and lateral static load tests
interface shear strength including normal stress levels, soil were carried out. A description of the experimental setup
particle angularity and particularly interface surface and analysis of the results is presented in the sections
roughness. Further studies carried out by Pando et al. below.
(2002) confirmed that interface surface roughness, particle
shape and angularity play significant roles in the shear
Soil sampling and properties
strength behaviour of the soil–pile interface.
Pile drivability and installation is crucial when assess-
Undisturbed clay sampling
ing the use of FRP piles as deep foundation. Due to inherent
material properties such as lower stiffness, impedance and
Pile load tests were carried out in undisturbed clay
density leading to higher damping, hollow FRP piles have a
samples contained in industrial sized steel drums
lower driving performance compared to traditional piles.
(876 mm in height and 597 mm in diameter). Soil sampling
Iskander et al. (2001) carried out a numerical analysis
was conducted at a clay rich site in Navan, Ontario. The
focusing on the drivability of FRP composite piles in which
sampling was conducted as follows: (1) a thick section of
they identified that material properties such as specific
clayey material was prepared by removing the top soil
weight and elastic modulus played critical roles in driving
and leveling the surface ensuring that a clean profile of clay
performance, whilst for stiffer piles, driving behaviour de-
of at least 1.5 meters was present, (2) steel drums were
pended mostly on the soil properties. Ashford and
placed upside down on the leveled soil surface, (3) using
Jakrapiyanun (2001) analysed the dynamic pile driving re-
an excavator each steel drum was carefully embedded in
sponse of a series of steel, concrete and FRP piles. Their re-
the clay layer ensuring no cracking and minimal distur-
sults indicate that FRP piles perform comparatively well
bance of the surrounding soil, (4) following embedment,
compared to traditional piles, however lower material
the samples were carefully excavated and extracted, (5)
impedance limited the ultimate capacity achieved at driv-
the filled drums were sealed to prevent soil desiccation
ing refusal. Mirmiran et al. (2000) analysed the dynamic
and to maintain the in situ conditions, (6) the soil contain-
response to driving for a series of concrete, concrete in-
ers were then shipped to the testing facility where the pile
filled and hollow FRP piles, and concluded that concrete
load tests were conducted. Fig. 1 illustrates the soil sam-
in-filled FRP piles performed on par with conventional con-
pling procedure during the encasing of the steel drums.
crete piles, however hollow FRP shells were at risk of com-
pression failure and could support driving stresses of up to
50% at driving refusal compared to the concrete pile. Soil properties
Pando et al. (2006) carried out a large scale pile load test
investigating the performance of FRP piles as the support- The soil used for this study is a type of marine soft clay
ing structure for a highway overpass in Virginia. They com- formed during the most recent ice age (cira 10,000 years
pared driven pre-cast concrete piles to concrete in-filled ago) in the Ottawa River lowlands. This material, typically
FRP piles. Axial pile load tests showed that the FRP piles referred as Leda clay or Champlain Sea clay, is the result of
performed comparably to the concrete pile. Han and Frost glacial abrasion of the Canadian Shield resulting in a fine
(2000) highlighted the importance of non-linearity effects rock flour which was deposited at the bottom of the pre-
of the flexural stiffness of FRP piles and increased deflec- historic saltwater Champlain Sea that flooded the region.
tions caused by shear deformations due to high elastic
modulus to shear modulus ratios. Sakr et al. (2004) com-
pared the performance of driven FRP piles to steel piles
in sand and found that FRP piles performed favorably with
respect to axial compression compared to steel piles; how-
ever under lateral loading, the FRP piles exhibited larger
deflections due to the inherent lower pile stiffness.
Most of the pile load tests and interface characteriza-
tion studies using FRP materials have been conducted in
sandy soils. There is a gap in knowledge regarding the
behaviour of these materials, particularly frictional resis-
tance under undrained conditions in clayey soils. This pa-
per provides baseline information to compare the
performance of FRP and steel piles in soft clays. This in-
cludes driving response, ultimate pile capacity under axial
and lateral pile load tests and pile response under large lat- Fig. 1. Soil sample preparation for pile load tests.
 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73 65

available geotechnical numerical software (Geostudio

2007). The goal of this analysis was to identify a pile length
and diameter where boundary conditions of the soil con-
finement did not have a significant impact on the test re-
sults. The numerical analysis simulated small scale pile
load testing using the geometry of the clay sample con-
fined by the steel drum. Boundary conditions for the soil
sample were simulated by applying displacement con-
strains along the outside model nodes representing the
steel drum confinement of the soil sample. The pile was
simulated as a steel member embedded in the soil with
the appropriate geometric and elastic properties.
The model used linear-elastic properties for the soil and
Fig. 2. Particle size distribution of Leda Clay.
pile materials and Mohr–Coulomb material properties for
the soil–pile interface. The analysis was carried out using
The corresponding ASTM Soil characterization tests were axisymmetric behaviour under total and effective stress
carried out in order to determine the soil properties. Soil conditions utilizing interface parameters for steel under
unit weight was measured at 15.3 kN/m3 and it was classi- both drained and undrained conditions. The pile load test
fied as highly plastic clay (CH) according to the Unified Soil was simulated by applying 10 displacement increments
Classification System. Undisturbed undrained in situ shear of 1 mm in either the axial or lateral dimension at the pile
strength was measured with a vane shear apparatus head location for a total pile displacement of 10 mm. It was
according to ASTM D2573 (2008) and was measured as found through various iterations that a 700 mm pile length
55 kPa throughout the steel drum depth profile. Atterberg and a diameter of 55 mm would induce less than 10% of
limit tests were performed according to ASTM D4318 the applied stresses at the soil boundary, and accordingly
(2010) with measurements of a liquid limit and plasticity these were the dimensions used for the model piles. The
index of 51 and 23, respectively. Particle size distribution selected dimensions allow for 5 d (pile diameter) distance
was determined as per ASTM D422 (2007) using a hydrom- between the steel drum wall and the pile, providing ade-
eter test, and the results are illustrated in Fig. 2. The soil quate spacing to prevent influence from the boundary.
internal friction angle was determined using a direct shear Similar spacing of 3–5 d is typically used for preventing
box apparatus according to ASTM D6528 (2007) and was group pile effects for frictional piles (Tomlinson, 1994;
measured as 23.3°, under drained conditions at a rate of Franke, 1984).
0.05 mm/min. Table 1 summarizes the complete list of Five open-ended model piles were manufactured: 2
measured soil properties. glass FRP piles (GFRP) and 3 carbon fiber (CFRP) piles. Pile
wall thicknesses and diameter to thickness (d/t) ratios
were maintained as constant as manufacturing permitted,
Model piles
and ranged from 3.6 to 4.15 mm in thickness and from 13.3
to 15.1 in d/t ratios (Table 2). These dimensions also al-
A series of model FRP piles were manufactured with
lowed for a comparison to the control steel pile with a wall
commercially available carbon and glass fabrics typically
thickness of 3.45 and d/t of 14.2. For all piles, a total of 3
used for the reinforcement and retrofit of structural
layers of fiber material was used which was the controlling
members. The raw material is in the form of fiber bundles
parameter that dictated the wall thickness. Piles were
woven into a coherent fabric in a uniaxial direction. Satu-
manufactured with uniaxial alternating fiber direction
rating the fabric in epoxy forms the FRP structure which
which allowed for isolating the pile response to fiber orien-
can be molded into a number of different shapes and
tation. In addition, the different fiber orientation and mate-
arrangements. The structural integrity of the cured fiber
rial fabrics cause changes in pile texture due to the
is provided primarily along the fiber direction. Manufac-
differing wave patterns of the fiber fabric used. The GFRP
turing of the model FRP piles was carried out by using a
piles were manufactured with three layers in either 0° or
steel mandrel around which the fiber fabric was fitted
90° orientations. Similarly the CFRP piles had the same
and allowed to cure following epoxy impregnation. Follow-
arrangement, but with the addition of a third pile with
ing a 48 h curing time, as per the manufacture’s guidelines,
two inner layers having fibers arranged along the pile axis
the pile was extracted from the steel mandrel and cut to
(0°) and one outer layer with fibers perpendicular to the
the required length. Pile dimensions were selected by car-
pile axis (90°). A control steel pile was prepared in order
rying out a stress distribution analysis using commercially

Table 1
Soil properties for Leda clay.

q (Mg/m3) w (%) LL (%) PI (%) wopt (%) qd(max) (Mg/m3) Cv (cm2/s) cu (kPa) / (deg) c (kPa)
1.53 49 51 23 30 1.41 1.40E04 55 23.3 42.2

Notes: q, density; w, moisture content; wopt, optimum moisture content; qd(max), maximum dry density; cv, coefficient of consolidation; cu, undrained shear
strength; /, internal friction angle; c, apparent cohesion.
66 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73

Table 2
Model pile geometric properties.

Pile Diameter Toe area Length Thickness d/t Fiber direction Epoxy to fiber
d (mm) (mm2) L (mm) t (mm) (°) content (%)
C-0 54.7 629.8 700 4.12 13.3 0 52.5
C-90 55.1 639.2 700 3.98 13.8 90 53.5
C-0/90 57.1 649.4 700 4.15 13.8 90/0 55
G-0 55.4 608.5 700 3.79 14.6 0 53.5
G-90 54.5 590.5 700 3.62 15.1 90 55.5
SP 49 493.7 700 3.45 14.2 N/A N/A

C/G – denotes fiber material, number denotes fiber orientation.

SP – steel pile.

Table 3
Model pile material properties.

Pile Ultimate tensile Tensile modulus Moment of Initial flexural Ultimate flexural Initial stiffness Stiffness at failure
strength (MPa) (GPa) Inertia (mm4) modulus (GPa) modulus (GPa) (EI) (kN-m2) (EI) (kN-m2)
C-0 986 98.5 2.04E+05 30 56.2 6.12 11.46
C-90 986 98.5 2.10E+05 – – – –
C-0/90 986 98.5 2.29E+05 – – – –
G-0 575 26.1 2.04E+05 25 32.6 5.10 6.8
G-90 575 26.1 1.91E+05 – – – –
SP 400 200 1.29E+05 200 200 25.75 25.75

C/G – # denotes fiber material, number denotes fiber orientation.

SP – steel pile.

to compare the pile performance. Table 2 presents a sum- was placed across the channel section which served as a
mary of the geometric properties of the model piles tested. reaction platform allowing for the application of compres-
Additionally, a steel encasing mechanism at the pile head sive loads. Axial load was applied using a hydraulic jack
was used to apply axial and lateral loads to the piles and coupled to a steel rod connected to the pile head which
to prevent pile damage during driving. runs vertically aligned with the pile axis. Force measure-
Mechanical properties for the FRP composites, as re- ment was carried out using a load cell placed at the jack
ported by the manufacturer, are ultimate tensile strengths – pile head connection ensuring proper alignment. Lateral
of 986 and 575 MPa and tensile modulus of 98.5 and loading was applied using a lateral bracing system con-
26.1 GPa for CFRP and GFRP, respectively. Additionally, test nected to the vertical reaction columns. A steel rod was
piles manufactured with fiber direction along the pile lon- used to attach the pile head to the hydraulic jack which
gitudinal were prepared and tested under three point was resting horizontally against the reaction lateral brace.
bending according to ASTM D790 (2007) in order to char- The steel rod was attached to the pile head by means of a
acterize the pile flexural strength. Ultimate values for elas- U-shaped brace around the pile head which was connected
tic modulus E, pile stiffness and other mechanical to the steel rod by an end-bearing joint allowing for hori-
properties are listed in Table 3. It was observed however zontal self-alignment as load was applied. A schematic
that pile stiffness EI varied non-linearly with the loading top view for the lateral loading system is illustrated in
level. This non-linear behaviour is typical of FRP piles (Pan- Fig. 3b.
do et al., 2006) and its implications regarding pile response All model piles were open-ended and driven to an
will be discussed in subsequent sections. Finally, the embedment depth of 700 mm. Driving records and rates
epoxy/fiber ratio ranged between 50:50 and 45:55 as fiber of penetration for the different piles are shown and dis-
saturation was observed in this range. This is typically the cussed in the following sections. The load tests were car-
epoxy saturation ratio used for FRP piles and tubes (Fam ried out immediately following driving to prevent time-
and Rizkalla, 2002). dependent pile set-up effects. Axial and lateral resistance
was measured using a load cell connected to a data logging
Experimental program station, while axial and lateral displacement at the pile
head was recorded with an LVDT device connected to the
Test setup reference beams. All the data was collected by an auto-
mated data logging station with calibrated recording soft-
A testing frame was assembled at Carleton University to ware. Fig. 4a shows the experimental assembly before
apply monotonic axial and lateral loads for pile load tests. driving of the pile. Fig. 4b illustrates the assembly of the
Schematics of the test assembly are shown in Fig. 3a. The pile head system and the vertical steel rod undergoing ax-
horizontal reaction beam is assembled using two channel ial loading and Fig. 4c shows the setup undergoing lateral
sections attached to the vertical columns. A steel plate loading.
 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73 67

Fig. 3. (a) Experimental setup and (b) lateral loading schematics.

(a) Test frame (b) Axial loading (c) Lateral loading

Fig. 4. Experimental assembly for axial and lateral load tests.

Pile driving resistance at approximately 4 diameters of penetration

possibly following the penetration of the top few inches
Pile driving was carried out using a pulley and drop of disturbed soil material due to sampling and pile driving.
hammer setup. The drop hammer weight and drop dis- In addition, all FRP piles presented similar penetration ra-
tance were kept constant throughout the test at 8.8 kg tios, while the steel pile penetration ratio was up to 25%
and 150 mm, respectively. The impact energy was approx- lower at lower driving depths.
imately 15 joules per blow. Driving was carried out at
approximately 10–15 blows per minute with a rate of pen- Test procedure
etration ranging from 10 mm/blow at shallow depths to as
low as 2 mm/blow when reaching the target depth. Fig. 5a Static compressive pile load tests
shows the cumulative blow count vs. penetration depth ra- Compressive pile load tests were carried out following a
tio d/z (diameter over depth) and Fig. 5b shows the rate of modified method based on ASTM D1143 (2007). The pile
penetration vs. the depth ratio. It is evident there are two head mechanism was used to connect a steel rod to the
main regions where there is a marked increase in driving hydraulic jack and the reaction frame allowing for the
68 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73

Fig. 5. (a) Cumulative # of blows vs depth ratio; (b) rate of penetration vs depth ratio.

application of compressive loads. The loading procedure

was designed to bring the pile to failure by applying incre-
mental loading at 100 s intervals, allowing for interface
forces to reach equilibrium. The failure criteria used to
determine ultimate pile capacity was the pile resistance
at a pile head displacement of 10% pile diameter as sug-
gested by De Nicola and Randolph (1999). For all tests,
the number of loading cycles required to reach the men-
tioned failure criteria ranged from 14 to 17. Pile unloading
was carried out in 4 steps and pile rebound was measured.

Static lateral load tests

Following axial compression testing, a lateral pile load
test was carried out according to a modified procedure Fig. 6. Pile resistance vs pile head movement.
based on ASTM D3966 (2007). The test was carried out
by applying load increments at 100 s intervals. Unlike com-
pressive tests, lateral load tests did not reach the defined for the FRP model piles was 25–40% higher than the steel
failure behaviour. Typically, the design criteria for lateral pile with the exception of the C0 pile which was on par
pile capacity are based on limiting pile head deflection. with the steel pile ultimate capacity. The loading response
The failure criteria used for determining ultimate pile of all piles was similar, with a rapid linear increase in pile
capacity was the capacity measured 6.25 mm of pile head resistance at very low pile head displacements, attributed
displacement as suggested by Prakash and Sharma to ultimate shaft frictional capacity being mobilized fol-
(1990), however the loading test was continued until lowed by a hardening region until ultimate capacity is
approximately 0.20 was reached for the displacement ratio reached at approximately 5.0–5.5 mm of displacement.
(pile head movement/pile diameter) to observe pile behav- Pile frictional resistance was observed to engage at very
iour at large lateral deflections. The average number of cy- low displacements ranging from 0.08 mm to 0.16 mm cor-
cles required to reach the failure criteria ranged from 20 to responding to 0.15–0.30% of pile diameter. Pile G90 pre-
25, at which unloading was carried out in 4–5 cycles and sented the fastest rate of capacity increase while the
pile rebound was measured. steel and C0 piles presented the most gradual increases.
The rapid change in slope of the load–displacement curve
Results and discussion after the early portion of loading can be attributed to fric-
tional failure of the soil surrounding the pile shaft. Further
Pile axial compression capacity pile movement engages soil at the pile toe providing a very
gradual increase of capacity due to the open-ended nature
Pile load test results are summarized in Fig. 6 where ax- of the pile geometry.
ial load capacity is plotted against vertical pile head move- Based on the pile load tests results shown above, it is
ment. Ultimate pile capacity was defined as the load evident that the FRP piles tested present a significant
corresponding to a pile head displacement equal to 10% improvement over traditional steel piles in terms of their
of the pile diameter (De Nicola and Randolph, 1999). At axial capacity performance in soft clays. Full scale pile load
this level of displacement (5.5 mm), pile capacity was ob- tests are required to verify pile driving performance, struc-
served to remain approximately constant with each addi- tural resistance, and long term effects such as creep and
tional loading cycle. Table 4 summarizes the pile capacity weather resistance. As FRPs gain a better established track
for all FRP and steel piles. In all cases, ultimate capacity record and increased industry use, economies of scale will
 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73 69

Table 4
Axial compressive capacities of FRP and steel piles.

Pile Compressive Shaft resistance Estimated shaft Toe resistance Shaft resistance
capacity (N) (N) resistance (N) (N) ratio (%)
C-0 1670 1330 1453 340 79.6
C-90 2567 2220 2041 345 86.4
C-0/90 2786 2436 N/A 350 89.8
G-0 2711 2383 2255 328 89.9
G-90 2722 2404 2194 318 88.3
SP 1710 1440 1546 266 84.2

bring prices down to match that of steel piling. Further- where Q is the measured pile capacity, Qb is the toe resis-
more, GFRP piles exhibited some of the largest pile resis- tance, C is pile circumference, and L is pile embedment.
tances tested while material costs are half of the CFRP Fig. 7 illustrates the comparison of average unit shaft resis-
piles. tance plotted against displacement ratio (pile head move-
ment/average pile diameter) for all piles. In all cases
Pile shaft resistance failure was estimated to occur at a displacement ratio of
Pile capacity of open-ended piles in cohesive soils is approximately 0.1. It is clear that both glass fiber G0, G90
developed primarily in the form of soil–pile adhesion, also and the carbon C90/0 model piles reached the highest shaft
referred to as pile shaft resistance, in undrained conditions. resistance at approximately 20 kPa compared to steel pile
Various authors (Poulos and Davis, 1980; Fleming, 1992) at 13.5 kPa for an overall improvement of up to a 30% in-
and design guidelines such as the CFEM (Canadian geo- crease. In contrast, pile C0 presented an overall decrease
technical society, 2007) estimate the pile bearing capacity in shaft resistance of up to 15% at pile failure. The higher
as a function of pile geometry (length and diameter) and shaft resistances of FRP piles are attributed to their surface
soil–pile interface resistance. The shaft resistance (Qs) is roughness compared to the steel pile. Pile surface texture
typically estimated from the following equation: plays a significant role in the increased shaft resistance be-
cause grooves and textures in the pile fabric create in-
Q s ¼ CLqs ð1Þ
creased contact area between the soil and pile. Similar
where C is pile circumference, L is pile embedment and qs behaviour was observed when different FRP fabrics were
is the average shear strength, also known as unit shaft sheared against soft clay in a direct shear box (Giraldo
resistance, along the soil–pile contact area. Furthermore and Rayhani, 2013). Large scale pile load tests are needed
the average shear strength qs can be expressed in terms to corroborate these results.
of the soil undrained shear strength (cu) and an empirical The corresponding a-coefficient for each pile load test
reduction factor, a, accounting for pile–soil adhesion which was estimated using Eq. (2). The a value for piles G0, G90
is typically correlated to cu. qs is estimated by the following and C90/0 was 0.36, for pile C90 it was 0.33 and for pile
equation: C0 it was the lowest value at 0.21. In addition, the control
steel pile was measured at a of 0.25. The calculated values
qs ¼ acu ð2Þ
are, in general, lower than the typical estimates reported in
Bearing capacity of open-ended driven piles in clay is the literature for clays of similar undrained shear strength
mainly provided by shaft resistance, with a minimal con- (Poulos and Davis, 1980; Fleming, 1992; Canadian geotech-
tribution from toe bearing resistance. Toe bearing resis- nical society, 2007). Additionally, shaft resistances esti-
tance has been estimated and removed from the total mated based on undrained interface parameters obtained
pile capacity in order to isolate the shaft resistance. Toe from interface shear strength values carried out by Giraldo
capacity (Qb) was calculated according to the following and Rayhani (2013) are presented in Table 4. The results
expression: indicate a good agreement with the measured shaft
Q b ¼ N c c u Ab ð3Þ
where Nc is a bearing capacity factor, typically 9 for pile
diameter under 0.5 m as suggested by CFEM (Canadian
geotechnical society, 2007). cu is the soil undrained shear
strength, and Ab is the pile toe contact area. On average,
the calculated toe bearing capacity was estimated to be
approximately 10–15% that of the measured pile resistance
for all model piles.
The average unit shaft resistance, qs, was estimated
from the measured pile capacities by dividing the load
transferred to the soil between the pile head and toe by
the surface area of the pile, as follows:
Q  Qb
qs ¼ ð4Þ
CL Fig. 7. Average unit shaft resistance of FRP and steel piles.
70 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73

resistance values from pile load testing with the largest optimal structural integrity of the pile in order to avoid
discrepancy of 8% between the estimated and measured structural failure. In both cases, (GFRP and CFRP) the opti-
values. mal outer surface fiber orientation was 90° with pile shaft
resistance. The CFRP piles showed a higher sensitivity to fi-
ber orientation than GFRP piles. Inner layers of FRP should
Effect of fiber orientation on pile capacity
be arranged in a fashion to increase pile structural integ-
Fiber orientation was investigated in order to character-
rity. Full scale pile load tests are needed in order to explore
ize its influence on pile capacity. Since pile surface texture
the pile behaviour under working loads.
was dictated by the weave pattern of the fiber fabric used
to manufacture the piles, changing the direction of the fi-
Effect of FRP material selection on pile capacity
ber significantly modified the pile texture in contact with
Selection of pile material has an important role in two
the soil. More importantly, GFRP piles had a distinct inter-
key aspects of pile performance: pile structural integrity
woven fiber bundling pattern which gave the pile surface
and geotechnical response. Structural integrity is directly
protruding ridges and valleys with approximately 0.5 mm
linked to the mechanical material properties such as elastic
peak to valley distances that changed direction depending
modulus and yield strength, whereas pile geotechnical per-
on pile manufacturing. In contrast, for CFRP piles the tex-
formance is strongly influenced by the pile surface texture
ture was streamlined with less pronounced ridges. How-
which is dictated by the weaving of the fabric. Based on the
ever CFRP oriented at 0° allowed for streamlined groves
results of this study, the GFRP piles tested in both 90° and
along the pile axis which is suspected to allow the unob-
0° orientations presented a significant improvement over
structed sliding of clay, leading to the lower pile capacity
steel, while fiber orientation strongly influenced pile resis-
measured. Fig. 8 illustrates the frictional performance of
tance for CFRP piles, leading to significantly lower pile
three CFRP model piles each with fibers parallel and per-
capacity in the 0° fiber direction. Fiber weaving and texture
pendicular to each other against pile displacement ratio.
had a significant effect and the GFRP pile specimens
The third pile C90/0 has the outer surface with fibers ori-
showed protruding ridges due to the fabric weaving man-
ented at 90° with the objective of providing better soil–pile
ufacturing process which is suspected to allow for in-
resistance, while the inner fibers provide the pile with
creased soil–pile surface area or a better soil–pile
structural integrity. This combination is shown to present
interlocking mechanism. From a geotechnical point of
the best results in terms of shaft resistance for the CFRP
view, the surface texture produced by the GFRP material
piles with over a 35% increase of the C90/0 pile at pile fail-
yields a better performance than the CFRP fabric. To date
ure. Further testing and stringent manufacturing quality
most studies regarding structural response of hollow and
control to ensure pile surface texture repeatability are
concrete filled FRP tubes have been conducted with GFRP
needed to characterize the influence of pile texture and fi-
fibers since the high cost of CFRP materials has not made
ber orientation in a consistent manner. With respect to
the production of CFRP piles feasible for large scale use.
GFRP piles, both piles reached similar frictional capacity
However, studies on the structural response of GFRP piles
at failure suggesting that the pronounced texture of the
have shown that hollow GFRP piles are susceptible to
pile surface interacting with the soil is independent of
buckling (Frost and Han, 1999). In addition, long term
groove orientation, suggesting that the increased surface
deformations such as creep can be significant issues for
area available at the soil–pile interface was responsible
FRP piles due to material anisotropy, lower moduli and
for the increased shaft resistance. However, it was ob-
higher shear to elastic moduli ratios compared to steel. Ap-
served that pile G90 experienced shaft resistance mobiliza-
proaches to increase structural performance of FRP piles
tion at lower displacement ratios than pile G0. No GFRP
include the partial or total infilling of the hollow void with
pile with combined fiber orientation was produced since
concrete in order to take advantage of the confining effect
pile G90 was observed to have the sufficient structural
of the concrete by the FRP shell (Fam and RizKalla,
integrity required to withstand testing at the low axial
2001a,b).
loads used in this small scale study. Further analysis of
optimal fiber orientation is needed in order to establish
Pile lateral capacity

Pile lateral capacity is typically determined by estimat-

ing the ultimate lateral failure load divided by a safety fac-
tor or by estimating the allowable lateral load
corresponding to acceptable pile deflections based on var-
ious design criteria. The failure load, based on the static lat-
eral loading test, was taken as the pile resistance measured
at 6.25 mm (0.1 displacement ratio) of lateral pile head
movement (Prakash and Sharma, 1990). However, the
loading test was carried out until a pile head movement
of 20% pile diameter (displacement ratio) occurred to
observe the behaviour of the model piles under large
deformations. Piles were assumed to behave as short free
Fig. 8. Comparison of CFRP pile performance on based on fiber – head piles as the L/d ratio for all the piles ranged from
orientation. 13.5 to 12.7. Typically, short pile behaviour is expected
 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73 71

Fig. 9. Lateral pile resistance vs pile displacement ratio. Fig. 10. Effect of FRP material selection on lateral pile capacity.

for pile with L/d ratios of 10–12 (Tomlinson, 1994). In addi-

tion, the maximum applied lateral load was lower than the
load at failure under 3-point bending tests for the tested
FRP piles illustrating that flexural material failure is not
the controlling parameter for the tested piles.
Fig. 9 illustrates the results for 5 pile load tests showing
lateral resistance and pile displacement ratio (lateral
movement divided by pile diameter). Pile C90 failed struc-
turally when subjected to lateral loads due to the lack of
longitudinal fiber reinforcement; hence the results are
not included in this analysis. In all cases the load–displace-
ment behaviour exhibits two approximately linear sec-
tions. At low lateral displacement ratios of 0.01–0.02,
loading response is characterized by a rapid increase in lat-
eral resistance, followed by an approximately linear trend Fig. 11. Comparison of pile flexural stiffness under 3-point bending.
with a less pronounced slope. The steel pile presented an
increase in lateral resistance up to 28% higher at failure
Effect of pile material on lateral capacity
compared to the FRP piles. Similar behaviour was reported
Pile response under lateral loading is typically controlled
by Sakr et al. (2004) in scale piles tested in a pressure
by soil response, pile stiffness and soil–pile interface inter-
chamber filled with sand. The lower pile response was
action. For this research program, lateral response is mainly
mainly attributed to the lower flexural stiffness of the
controlled by the model pile structural properties such as
FRP piles (EpIp). In addition Pando et al. (2006) discussed
the piles flexural stiffness EpIp since soil conditions and pile
the susceptibility of FRP piles, especially hollow FRP piles
geometry is approximately equal for all cases. Fig. 10 shows
to increased lateral deflections due to shear deformation
the lateral response of piles C0 and G0 illustrating no signif-
effects caused by high ratios of elastic to shear modulus.
icant difference in pile behaviour, with pile G0 showing only
Table 5 shows a summary of the lateral capacity for the dif-
a 5% increase in pile capacity. Fig. 11 illustrates the results
ferent piles and their corresponding flexural stiffness (EI)
for flexural stiffness vs loading level carried out under 3-
estimated from 3 point loading tests. It was observed based
point bending. It is shown that the pile behaviour is highly
on these flexure tests that the piles at peak lateral displace-
non-linear at low loading levels reaching a plateau after
ment ratios of 0.2 were only approximately under 20–30%
approximately 30% of failure load. The FRP model piles un-
of their ultimate flexural capacity, and well below struc-
der lateral load at 0.2 displacement ratio face a load of
tural failure. Full scale load tests are needed to explore
approximately 2000 N at the pile head corresponding to
FRP pile behaviour under working load conditions.
approximately 20% of the failure loading under 3-point

Table 5
Lateral capacity and properties of FRP and steel piles.

Pile Lateral capacity Pile stiffness (EpIp) Pile stiffness (EpIp) Lateral resistance
at 6.25 mm (N) at 2000 N load (kN-m2) at failure (kN-m2) ratio to steel
C-0 880 6.4 6.8 80.0
C-90 N/A N/Ma N/Ma 86.4
C-0/90 650 N/Ma N/Ma 59.0
G-0 950 9.7 11.4 86.6
G-90 580 N/Ma N/Ma 52.0
SP 1100 25.75a 25.75b
a
N/M – not measured under 3-point bending.
b
Calculated based on 200 GPa elastic modulus.
72 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73

Fig. 12. Measured lateral capacities vs LPILE analyses for FRP and steel piles (G0; glass FRP, SP: steel and C0: Carbone FRP).

bending. This indicates that the pile flexural stiffness for into account the steel drum sample confinement and as-
piles undergoing lateral loading is in the range of 6.7– sumes an infinitely long soil layer in the horizontal direc-
9.3 kN-m2 for G0 and C90, correspondingly. Although 3- tion. This effect can be one of the reasons for some of the
point bending does not accurately represent the bending discrepancies observed between the measured and esti-
moments applied to a pile under lateral loading, these val- mated lateral deflections of some of the tested piles. The
ues give a basic understanding of why these piles experience results comparing the behaviour of the steel pile, C0 and
higher deflections than their steel counterpart. Based on this G0 piles are show in Fig. 12.
study it is difficult to identify which material performs opti- Similar resistance–displacement behaviour was found
mally under lateral loading since the pile stiffness behaves between the steel pile and LPILE results with similar pile
non-linearly and the soil yielded before pile failure oc- head displacements throughout the loading procedure.
curred. It was obvious however, that both FRP piles present Although the GFRP pile presented a reasonable agreement
a significantly softer response than steel piles. with LPILE results, the CFRP pile presented a discrepancy of
up to 35% pile head deflection difference. This could be
Comparison to analytical lateral capacity estimates attributed to non-linear material deformation of the CFRP
Lateral pile capacity is usually carried out by estimating pile not accounted in the LPILE model as the element was
ultimate pile capacity through methods considering lateral modelled as a linear-elastic pipe with constant EI. A more
earth pressure theory such as the Brinch Hansen Method elaborate numerical analysis accounting for FRP non-linear
(Brinch Hansen, 1961) or Broms’ theory (Broms, 1964) or deformation is probably needed to accurately model the
by estimating acceptable deflections at a working load. FRP piles behaviour in LPILE. Pando et al. (2006) carried
Estimating pile deflections typically uses the elastic contin- out a similar analysis for concrete filled FRP piles in sand
uum solution and modulus of subgrade reaction ap- with a similar approach using LPILE software where they
proaches such as p–y curves to calculate pile deflections found good agreement between the modelled and mea-
at a given loading condition. A numerical analysis was car- sured pile deflections with the implementation of a non-
ried out using the p–y method to compare typical design linear EI in their modelling. Full scale tests using instru-
approaches applied to the response of the FRP piles. P–y mented piles are required to accurately measure pile
curves are typically developed for specific soil types and deformations along the shaft in order to characterize the
pile geometries (Prakash and Sharma, 1990). The family pile stiffness non-linearity, and better predict pile deflec-
of p–y curves used for this study was based on method pro- tions under working loads.
posed by Matlock (1970) which was developed for soft to
medium clays. The analysis was carried out using the com- Conclusions
mercially available software typically used for design esti-
mates LPILE (Ensoft Inc., 2013) which allows the input of A series of small scale static pile load tests in laboratory
pile properties, boundary and loading conditions and the was carried out to characterize the performance of open-
built in p–y curves. ended FRP piles in clayey soils. The major findings of this
The simulation used the same laboratory parameters research programme are summarized below:
with the same thickness for the uniform clay layer, pile
dimensions and geometric properties and stiffness re-  Compressive ultimate pile resistance of the model FRP
sponse based on results measured from the 3-point bend- piles was up to 40% higher than the control steel pile.
ing tests on piles at a similar loading level as the lateral Similarly skin frictional resistance of the FRP piles was
loading test. The soil was modeled as an elasto-plastic measured to be up to 30% higher with both CFRP and
medium with Mohr–Coulomb constitutive behaviour, GFRP.
while the pile was modeled as a linear elastic element.  GFRP piles presented a highly textured surface due to
Loading was applied in 10 increments until a total the raw fiber weaving pattern with both directions 0°
2000 N lateral loading was reached. Measurements of pile and 90° showing significant protrusions. In contrast
head deflection at the different loading intervals were then the CFRP pile presented a less pronounced surface tex-
compared to the lateral test results. It is important to note ture with the 90° orientation showing an increased
that the numerical simulation using LPILE does not take profile.
 J. Giraldo, M.T. Rayhani / Transportation Geotechnics 1 (2014) 63–73 73

 FRP pile texture and waviness appeared to have a signif- ASTM D422. Standard test method for particle-size analysis of soils. West
Conshohocken, PA: ASTM International; 2007. http://dx.doi.org/
icant influence on pile performance with CFRP piles
10.1520/D0422-63R07.
ranging from just 5% capacity increase to up to 40% by ASTM D4318. Standard test methods for liquid limit, plastic limit, and
changing pile fiber direction and surface texture. In con- plasticity index of soils. West Conshohocken, PA: ASTM International;
trast, the GFRP piles presented virtually the same ulti- 2010. http://dx.doi.org/10.1520/D4318-10.
ASTM D6528. Standard test method for consolidated undrained direct
mate capacity at a 40% increase over steel despite simple shear testing of cohesive soils. West Conshohocken, PA: ASTM
changes in fiber orientation. International; 2007.
 Lateral loading of the piles showed that the lower stiff- ASTM D790. Standard test methods for flexural properties of polymer
matrix composite. West Conshohocken, PA: ASTM International;
ness of the FRP piles leads to increased pile head deflec- 2007. http://dx.doi.org/10.1520/D1143_D1143M-07E01.
tions at similar loading levels compared to the baseline Brinch Hansen J. The ultimate resistance of rigid piles against transversal
steel pile. These results agree with similar tests carried forces. 1961, Geoteknisk Institut, Copenhagen, Bull No. 12.
Broms BB. Lateral resistance of piles in cohesive soils. J Soil Mech Found
out in sandy soils Div., ASCE 1964;90(2):27–67.
 The modelling of the steel, C0, and G0 piles under lateral Canadian geotechnical society. Canadian geotechnical engineering
loading was carried out using LPILE numerical software manual. 4th ed. Richmond: BC, BiTech Publishers Ltd.; 2007.
De Nicola A, Randolph MF. Centrifuge modelling of pipe piles in sand
to compare common design methods to the perfor- under axial loads. Géotechnique 1999;49(3):295–318.
mance of FRP piles. The steel pile presented comparable Ensoft Inc. LPILE v. 6. 2013, [Online] available at: http://
results while both FRP piles presented higher pile head www.ensoftinc.com/ [Cited 23 07 2013].
Fam AZ, RizKalla SH. Behavior of axially loaded concrete-filled circular
deflections compared to the numerical simulation. In
fiber-reinforced polymer tubes. ACI Struct J 2001a;98(3):280–9.
the analysis, linear elastic properties were provided as Fam AZ, Rizkall SH. Confinement model for axially loaded concrete
input as is common practice in pile design. FRP piles confined by FRP tubes. ACI Struct J 2001b;98(4):251–61.
however can present non-linear behaviour under load- Fam AZ, Rizkalla SH. Flexural behaviour of concrete-filled fiber-reinforced
polymer circular tubes. J Compos Constr 2002;6(2):123–32.
ing which can lead to the measured discrepancies. Fleming WG. Piling engineering. New York: Halsted Press; 1992.
Franke E. Group action between vertical piles under horizontal loads.
The results of this study highlight the viability of FRP Proceedings of the International Geotechnical Seminar on Deep
Foundations on Bored and Auger Piles, Ghent, Belgium, 1984, pp.
piles in soft clays as an alternative to traditional steel piles. 83–93.
Axial compressive capacity was measured to be up to 40% Frost JD, Han J. Behaviour of interfaces between fiber-reinforced polymers
higher than steel, where pile surface and texture appeared and sands. J Geotech Environ Eng 1999;125(8):633–40.
Geostudio 2007. A software package for geotechnical analysis.
to have a significant influence on pile performance. Lateral Giraldo J, Rayhani MT. Influence of fiber-reinforced polymers on pile–soil
response of FRP piles showed higher deflections at similar interface strength in clays. Adv Civil Eng Mater 2013;2(1). ID
loading levels compared to steel. These results are in line ACEM20120043.
Guades E, Aravinthan T, Islam M, Manalo A. A review on the driving
with similar research conducted in sands and are primarily performance of FRP composite piles. Compos Struct 2012;94:
due to the lower elastic modulus and stiffness compared to 932–1942.
steel and prestressed concrete. Han J, Frost JD. Load–deflection response of transversely isotropic piles
under lateral loads. Int J Numer Anal Meth Geomech 2000;24:509–29.
Iskander MG, Hassan M. State of the practice review in composite piling. J
Acknowledgements Compos Constr 1988;2:116–20.
Iskander MG, Hanna S, Stachula A. Driveability of FRP composite piling. J
Geotech Geoenviron Eng 2001;127:169–76.
This study was financially supported by the Natural Sci- Matlock H. Correlation for design of laterally loaded piles in soft clay.
ences and Engineering Research Council of Canada Proceedings Offshore Technology Conference, 1970, paper OTC 1204.
Mirmiran A, Shahawy M, El Khoury C, Naguib W. Large beam-column
(NSERC). The authors are also grateful to their industrial tests on concrete-filled composite tubes. ACI Struct J 2000;97:268–76.
partner, Fyfe Co., for providing FRP samples for testing; O’Rourke TD, Druschel SJ, Netravali AN. Shear strength characteristics of
however, the views expressed herein are those of the writ- sand–polymer interfaces. J Geotech Eng 1990;116(3):451–69.
Pando AM, Flitz MG, Dove JE, Hoppe EJ. Interface shear tests on FRP
ers and not necessarily those of our partner.
composite piles. International Deep Foundations Congress. Orlando,
FL: ASCE; 2002. pp. 1486–1500.
Pando AM, Ealy CD, Flitz MG, Lesko JJ, Hoppe EJ. A laboratory and field
References study of composite piles for bridge substructures. McLean,
VA: Federal Highway Administration; 2006. Report No. FHWA-HRT-
Ashford SA, Jakrapiyanun W. Drivability of glass FRP composite piling. J 04-043.
Compos Constr 2001;5:58–60. Poulos HG, Davis EH. Pile foundation analysis and design. New York: John
ASTM D1143. Standard test methods for deep foundations under static Wiley and Sons Inc.; 1980.
axial compressive load. West Conshohocken, PA: ASTM International; Prakash S, Sharma HD. Pile foundations in engineering practice. New
2007. http://dx.doi.org/10.1520/D1143 D1143M-07E01. York: John Wiley and Sons Inc.; 1990.
ASTM D2573. Standard test method for field vane shear test in cohesive Sakr M, Elnaggar H, Nehdi M. Novel toe driving for thin-walled piles and
soil, west. West Conshohocken, PA: ASTM International; 2008. http:// performance of fiber-reinforced polymer (FRP) pile segments. Can
dx.doi.org/10.1520/D2573-08. Geotech J 2004;41:313–25.
ASTM D3966. Standard test methods for deep foundations under lateral Tomlinson M. Pile design and construction practice. 4th Ed. New York: E &
load. West Conshohocken, PA: ASTM International; 2007. http:// FN Spon, an Imprint of Chapman & Hall; 1994.
dx.doi.org/10.1520/D3966-07.