Vol. 7, No.

2 December 2013

DFI JOURNAL
The Journal of the Deep Foundations Institute

PAPERS:
A Driveability Study of Precast Concrete Piles in
Dense Sand
– Paul Doherty, David Igoe [3]
Ant Colony Optimization Method for Design of
Piled-Raft Foundations (DFI 2013 Student Paper
Competition Winner)
– Hessam Yazdani, Kianoosh Hatami, Elahe Khosravi [17]
Piezocone Penetration Testing in Florida High Pile
Rebound Soils
– Fauzi Jarushi, Paul J. Cosentino, Edward H. Kalajian [28]
Factors Affecting the Reliability of Augered Cast-In-Place
Piles in Granular Soils at the Serviceability Limit State
(DFI 2013 Young Professor Paper Competition Winner)
– Armin W. Stuedlein, Seth C. Reddy [46]
A Review of the Design Formulations for Static Axial
Response of Deep Foundations from CPT Data
(DFI 2013 Student Paper Competition Runner-Up)
– Fawad S. Niazi, Dr. Paul W. Mayne [58]

Deep Foundations Institute is the Industry Association of

Individuals and Organizations Dedicated to Quality and
Economy in the Design and Construction of Deep Foundations.
From the Editors and Publisher 2013 DFI Board
To quote Heraclitus, “There is nothing permanent except change.” Change is inevitable of Trustees
and essential in a growing organization, being preferable over remaining static. Following
seven years of in-house production of this journal, there are going to be changes. Manny President:
Fine is retiring from the Publisher position and DFI has contracted with Maney Publishing, Robert B. Bittner
(similarity of name is coincidental), an independent publishing company specializing in Bittner-Shen Consulting
journals in materials science and engineering, the humanities, and health science. Maney Engineers, Inc.
is committed to publishing high quality journals in print and electronic formats that are Portland, OR USA
international in scope and peer-reviewed. The partnership allows DFI to retain editorial Vice President:
control and ownership of the DFI Journal so that we can continue to produce a journal Patrick Bermingham
containing practice-oriented papers that are useful to our readers. We look forward to a Bermingham Foundation
successful relationship with our new publisher and all the developments this will entail. Solutions
DFI members will continue to have online access to the journal as a perk of membership Hamilton, ON Canada
and previous issues, back to Vol. 1, will also be available online. Secretary:
Matthew Janes
Additionally there will be changes to the editorial board with two of the three current editors, Isherwood Associates
Dan Brown and Ali Porbaha, stepping down, along with associate editor Lance Roberts, while Burnaby, BC Canada
Zia Zafir, Ph.D., P.E., of Kleinfelder will remain on the editorial board to provide continuity Treasurer:
during the transition. Two new co-editors were appointed by DFI president Robert Bittner; John R. Wolosick
Anne Lemnitzer, Ph.D., Assistant Professor of Structural/Geotechnical Engineering at Hayward Baker Inc.
University of California and Timothy Siegel, P.E., G.E., D.GE, Principal Engineer at Dan Brown Alpharetta, GA USA
and Associates, LLC. Anne’s current research work focuses on soil-structure interaction Immediate Past President:
of various bridge foundation systems and lead to significant changes in California bridge James A. Morrison
design (Caltran’s Seismic Design Criteria). She serves as reviewer for multiple geotechnical Kiewit Infrastructure
journals and conferences and is an active committee member of DFI, EERI and ASCE. Tim Engineers
began his career as a Geotechnical Consultant with S&ME, Inc. from 1993-2005 and then at Omaha, NE USA
Berkel & Company Contractors for the following 5 years, joining DBA in 2010. He also has Other Trustees:
served as Adjunct Faculty at The University of Tennessee since 2003. Additional members David Borger
of the editorial board will be Antonio Marinucci, Ph.D., P.E., Director of Sales and Research Skyline Steel LLC
at American Equipment and Fabricating as well as Thomas Weaver, Ph.D., P.E. of the Nuclear Parsippany, NJ USA
Regulatory Commission, previously associate editor. The editorial board will be rounded Maurice Bottiau
out in 2014 to ensure coverage of all technical areas of expertise. Franki Foundations Belgium
Saintes, Belgium
This edition includes a variety of papers, both as to subject matter and as to geographic
Dan Brown
location of the authors. This diversity of papers illustrates both the technical and geographic Dan Brown and Associates,
success of the DFI Journal in attracting papers of interest to our membership. PLLC
Sequatchie, TN USA
Gianfranco Di Cicco
GDConsulting LLC
Lake Worth, FL USA
Rudolph P Frizzi
Langan Engineering &
Environmental Services
Journal Publisher Elmwood Park, NJ USA
DFI JOURNAL Manuel A. Fine, B.A.Sc, P.Eng

Journal Editors
Bernard H. Hertlein
GEI Consultants Inc.
Mission/Scope Ali Porbaha, Ph.D., P.E.
Libertyville, IL USA
The Journal of the Deep Foundations Institute Central Valley Flood Protection James O. Johnson
publishes practice-oriented, high quality papers Board Sacramento, CA, USA Condon-Johnson & Associates,
related to the broad area of “Deep Foundations Dan A. Brown, Ph.D. Dan Brown Inc.
Engineering”. Papers are welcome on topics of interest and Associates, Sequatchie, TN, USA Oakland, CA USA
to the geo-professional community related to, all Zia Zafir, Ph.D., P.E. Douglas Keller
Kleinfelder Sacramento, CA, USA
systems designed and constructed for the support Richard Goettle, Inc.
of heavy structures and excavations, but not limited Associate Editors Cincinnati, OH USA
to, different piling systems, drilled shafts, ground Lance A. Roberts, Ph.D., P.E. Samuel J. Kosa
improvement geosystems, soil nailing and anchors. RESPEC Consulting & Services Monotube Pile Corporation
Rapid City, SD USA Canton, OH USA
Authors are also encouraged to submit papers on new
Thomas Weaver, Ph.D., P.E.
and emerging topics related to innovative construction Nuclear Regulatory Commission
Kirk A. McIntosh
technologies, marine foundations, innovative retaining Rockville, MD USA AMEC Environment &
systems, cutoff wall systems, and seismic retrofit. Infrastructure, Inc.
Case histories, state of the practice reviews, and Jacksonville, FL USA
innovative applications are particularly welcomed and Raymond J. Poletto
encouraged. Mueser Rutledge Consulting
Engineers
Published By Deep Foundations Institute Written permission must be obtained Contact New York, NY USA
Copyright © 2013 Deep Foundations from DFI to reprint journal contents, in DFI, 326 Lafayette Avenue
whole or in part. Hawthorne, NJ 07506 Michael H. Wysockey
Institute. AII rights reserved.
[email protected], www.dfi.org Thatcher Engineering Corp.
DFI, its directors and officers, and journal editors assume no responsibility for the statements expressed by the journal’s authors. Chicago, IL USA
International Standard Serial Number (ISSN): 1937-5247

DFI JOURNAL Vol. 7 No. 2 December 2013 [1]

A Driveability Study of Precast Concrete Piles in
Dense Sand
Paul Doherty, Post-Doctoral Researcher, University College Dublin, Ireland, [email protected]
David Igoe, Post-Doctoral Researcher, University College Dublin, Ireland

ABSTRACT
A research study was recently completed by University College Dublin to examine the performance of
various pile types including open steel tubular piles, concrete precast piles, and helical piers. At the
outset of this project, one of the key risks identified was that the concrete piles could not be installed
to the target depth due to (i) insufficient energy from the available hammer and (ii) the onset of pile
material damage. In order to mitigate this risk a detailed pile driveability analysis was completed to
predict the installation performance during driving. Selecting an appropriate model for predicting the
Static Resistance to Driving (SRD) was seen as a critical component of the driveability process in order
to predict reasonable stresses and blow counts. This paper describes the procedures adopted for a
base case driveability analysis and the outcome of the pile installations. A comparison of the SRD using
other models (including the API and IC-05 methods) was conducted and the results were compared
to SRD profiles derived from dynamic pile monitoring conducted on one of the concrete piles. The
base case driveability analysis indicated that the piles could be installed with the available hammer
equipment, however it was noted that the driving stresses were relatively high and approached the
failure stress of the concrete as the pile approached the target penetration. While hard driving was
observed in the field, all of the piles reached their design depth of 7m (23 ft) with the exception of
one pile which refused due to structural failure near the pile head. The driveability analysis and the
measured stresses were interpreted to identify the cause of failure for the single pile, which was linked
to the material properties of that specific pile.

INTRODUCTION that Goble, et al. (1976) cautioned that wave

Pile driveability analysis is used to assess equation analysis is “meaningful only if the
whether piles can be installed to a target driving system and soil conditions, as modelled
depth in a reasonable timeframe without in the computer, realistically reflect the actual
overstressing the pile material or generating conditions”. To date, accurately capturing the
excessive fatigue damage. Most pile driving site specific soil-pile behaviour remains one of
analysis procedures adopt a wave equation the most challenging aspects of undertaking
approach that simulates the entire pile, soil and a driveability analysis. This paper describes a
hammer system. This methodology is routinely case study where precast concrete piles were
implemented for offshore projects where pile installed at a very dense sand test site, where
installation is deemed to be a significant risk the soil relative density approached 100%, and
due to the high cost associated with vessel driveability was a concern from the outset.
delays (Toolan & Fox, 1977). Pile driveability The initial driving analysis are discussed
analysis is also employed for onshore projects and compared to the measured installation
and is particularly important where there is a performance.
specific concern over the installation stresses,
soil resistance, or hammer capabilities. In BACKGROUND TO PILE
1976, Goble et al. presented an alternative DRIVEABILITY
procedure for predicting pile installation The procedure for undertaking pile driveability
performance, which was deemed superior to analysis is illustrated schematically in Fig. 1
the empirical hammer-pile ratios employed below. The analysis undertaken for this project
at the time. The recommendations included used one-dimensional wave equation software
undertaking wave equation predictions and also to model the energy transfer from the hammer
measuring the force and acceleration of the into the pile-soil system.
pile throughout installation. It is worth noting,

DFI JOURNAL Vol. 7 No. 2 December 2013 [3]

 The main inputs for the analysis include the main differences relate to ageing and partial
soil resistance, the pile properties and the consolidation effects. To overcome these
hammer properties. The pile properties include differences, specific procedures have been
the geometry, the material type and relevant developed to predict the static resistance to
mechanical parameters such as modulus of driving (SRD). The baseline SRD calculation
elasticity. The hammer properties are used method considered in this study was
to determine the energy imparted to the formulated by Alm & Hamre (2001).
pile head (enthru energy) and therefore the
hammer model, anvil mass, efficiency and SELECTING A CONSERVATIVE SRD
drop heights are all important parameters MODEL FOR PILE DRIVEABILITY
in the analysis. The ability of the hammer to ANALYSIS
advance the pile into the ground will depend
The objective of a pile drivability analysis is to
on both the hammer energy during impact and
ensure that a pile can be safely installed to a
also the contact time over which the energy
target penetration with the available equipment
is transferred from the hammer into the pile.
without overstressing the pile material. This
In this regard, the cushion properties are also
paper focuses on pile driveability analyses
a critical input parameter, as a piling cushion
and is not considering the long-term pile
can be used to decrease the peak stresses in
behaviour. The SRD model selected for input
the pile while simultaneously increasing the
into the pile drivability analysis should assume
contact duration and consequently the pile
an upper bound profile for the soil resistance,
penetration rate.
such that the hammer and pile are analysed
against the hardest driving conditions that may
occur at site. A conservative SRD model would
typically over predict the average in-situ soil
resistance to ensure that the hammer is capable
of delivering sufficient energy to overcome
the maximum possible ground resistance
that may be encountered. By contrast, a non-
conservative SRD model would under predict
the driving resistance and as a result the pile
may experience premature refusal due to
the hammer being undersized for the given
installation conditions. An upper bound
SRD model will also yield the highest driving
stresses and therefore increasing the SRD values
is a conservative approach with respect to
material damage. When specifically considering
pile driveability analysis, the concept of
conservatism is reversed in comparison to static
capacity predictions.
For this study, the A&H method (Alm & Hamre,
2001) was preferred because of (i) its calibration
against North Sea pile driving records, (ii) the
direct use of the CPT test and (iii) consideration
[FIG. 1] Driveability Process
of friction fatigue effects in assessing the shaft
The soil driving resistance consists of two friction acting on the pile.
parts, the static component (SRD) and a velocity The friction degradation (or friction fatigue)
dependent dynamic component, termed concept was first introduced for clay by
damping. The damping factors implemented in Heerema (1980), and a complete SRD model
the wave equation model account for inertial that considers friction fatigue effects was
and viscous rate effects. The static component described by Heerema (1981),. Friction fatigue
is similar to the static capacity determined has also been documented in static load tests
from traditional pile capacity formulas. The undertaken by Chow (1997), and others. Alm

[4] DFI JOURNAL Vol. 7 No. 2 December 2013

& Hamre (1998) highlighted the limitations of For application in driveability analysis, Alm &
the Heerema SRD approach and proposed an Hamre (2001) suggest using an upper bound
alternative method that explicitly considered prediction of the SRD by multiplying the
friction fatigue in dense sands. This method resistance by a factor of 1.25 to provide an
subsequently evolved into the A&H approach, upper bound resistance.
which considers that the model should define For this paper, soil profiles calculated from the
an initial friction value and a residual static models were incremented to give approximately
friction, along with a shape function describing 100 elements for input into the wave equation
the relative friction fatigue reduction. This software. Issues arose due to friction fatigue,
resulted in the general formula for shaft shear as the distribution of shaft friction varied
stress (fs) defined by equation 1, below: with pile tip penetration. Schneider & Harmon
[1] (2010) found that the shape of the shaft friction
݂‫ ݏ‬ൌ ݂௥௘௦ ൅ ሺ݂‫ݏ‬௜ െ ݂௥௘௦ ሻ݁ ௞ሺௗି௣ሻ
distribution had little effect on the resultant
bearing graph and as a result the incremental
Where fres is the residual shaft shear stress
change in shaft capacity as the pile penetrates
after complete degradation; fsi is the initial
a distance ∆L could be used to calculate the
shaft shear stress value; k is a shape factor
pseudo average shaft friction (∆τf,avg):
controlling the rate of degradation; d is the
depth and p is the pile tip penetration. The ο߬௙ǡ௔௩௚ ൌ
σ ொೄǡಽ ିσ ொೄǡಽషభ
[7]
residual shaft shear stress is defined by: గ஽Ǥο௅

‫்ݍ‬ [2] where ∑QS,L is the cumulative shaft resistance

݂௥௘௦ ൌ ͲǤͲͲͶ‫ ்ݍ‬ሺͳ െ ͲǤͲͲʹͷ ൗߪ ᇱ ሻ
௩଴ at tip depth; ∑QS,L-1 is the cumulative shaft
resistance at the previous depth increment;
Where qT is the CPT cone tip resistance and σ'v0 and D is the pile diameter. The pseudo average
is the vertical effective stress. values can then be used to generate a synthetic
The initial shaft friction is defined by the soil shear stress profile with depth that can
traditional static formula: be implemented within the wave equation
software.
ᇱ
݂‫ݏ‬௜ ൌ ‫ߪܭ‬௩଴ ‫ߜ݊ܽݐ‬ [3]
It is usual for an SRD model to be accompanied
by a set of damping factors and quake values.
ᇲ ଴Ǥଵଷ
ᇱ
‫ߪܭ‬௩଴ ൌ ͲǤͲͳ͵ʹ‫ ்ݍ‬ቀ
ఙೡబ
ቁ [4] Quake is defined as the displacement required
௣௔
to mobilise the soil-pile resistance at a yield
Where δ is the mobilised friction angle at condition. The parameters used in this paper
failure, and pa is atmospheric pressure. were those associated with the original A&H
model. Quake values were taken as 2.5 mm
Calibration of this approach against a database
(0.1 in) for both side and tip yield, and soil
of piling records in the North Sea yielded a
damping constants of 0.5 s/m and 0.25 s/m
shape factor defined below:
଴Ǥହ
(0.15 s/ft and 0.08 s/ft) were used for the tip
݇ൌ൬
௤೅
൰ ȀͺͲ [5] and the side respectively.
ᇲ
ఙೡబ
SITE CONDITIONS AND GROUND
This shape factor suggests that rapid
degradation will occur for dense deposits with MODEL
high cone tip resistances. The UCD Blessington test site is an over-
The A&H formula for unit base resistance acting consolidated, glacially deposited, very dense
on a pile in sand was derived as: fine sand bed with a CPT qc resistance in the
range of 15- 20 MPa (2,175 – 2,900 psi) and
௤೅ ଴Ǥଶ
‫ݍ‬௕ ൌ ͲǤͳͷ‫ ்ݍ‬ቀ ᇲ ቁ [6] small strain shear stiffness in the range of 100
ఙೡబ
to 150 MPa (14.5 to 21.8 ksi). The water table
Equation 6 yields base stress values that is approximately 13m (42.7 ft) below ground
increase with sand relative density from 0.35 to level. Details of the deposit have been provided
0.55 as the sand density changes from loose to in Gavin & O’Kelly (2007), Gavin, et al. (2009),
very dense. Doherty, et al., (2012) and others. Eight CPTs
were conducted in the vicinity of the piling

DFI JOURNAL Vol. 7 No. 2 December 2013 [5]

 location and are shown in Fig. 2. Laboratory from 52 – 85 MPa (7,540 psi – 12,300 psi)
testing was used to supplement the in-situ site depending on the concrete batch. Details of
investigation and establish basic soil properties. the concrete mix are provided in Table 1. The
Particle size distributions conducted on the frequency distribution of 28-day cube strengths
Blessington sand classified the material as fine from concrete batches taken over the course of
to medium grained, with a mean particle size a year are shown in Fig. 3. The average 28-day
varying from 0.1 - 0.15 mm (.004 - .006 in). The strength of the concrete cubes was 64 MPa
moisture content increased with depth from (9,300 psi) but significant variability in concrete
8% at ground surface to approximately 12% at strength is evident. Pile reinforcement consisted
4.5 m (15 ft) depth. of eight H12 (0.47 in) helical cold drawn
rebars, with H5 (0.20 in) shear links spaced
every 55 mm (2.2 in) in the bottom and top 1m
(3.3 ft) and spaced every 150mm (5.9 in) in the
middle section.

[TABLE 1] Typical Concrete Mix Design for

Precast Piles

Quantity
Material
(kg/m3)

Cement 42.5N 400

PFA 80
Water 160
Admixture (Sika) 3.3
14mm Aggregate 670
10mm Aggregate 300
Concrete sand 850
Water/Cement Ratio = 0.4

[FIG. 2] CPT Cone Resistances from Blessington

Experimental Piling Programme

The pile tests described in this paper were part
of an extensive investigation of displacement
pile behaviour in sand including; driveability,
ageing, cyclic loading, dynamic loading
and lateral loading. In total, five 275 mm [FIG. 3] Variability of Concrete Cube Strength from 84
(10.8 in) square precast concrete piles and Samples
seven 340 mm (13.4 in) diameter open-ended
The piling rig used to drive the piles at
steel piles were driven into a dense sand
Blessington was a 4 tonne (4.4 ton) Junttan
deposit in Blessington, Ireland. This paper will
hammer (see Fig. 4), with a maximum rated
focus on the driveability of the precast concrete
energy of 47kJ (34,700 ft lb) for a ram fall
piles, which were 8m (26.2 ft) long and had a
height 1.2 meters (3.94 ft). For the concrete
target penetration of 7m (23 ft) below ground
piles a 75mm (3 in) thick beech pile cushion was
level. The piles were made from grade C50/60
also used to manage the applied stresses to the
concrete with 28-day cube strengths varying
concrete pile.

[6] DFI JOURNAL Vol. 7 No. 2 December 2013

 reasonable driving time and to minimise fatigue
damage, practical refusal was defined as a
penetration rate of less than 250 mm (10 in)
per 100 blows. The blow counts predicted for
this site were well below the design limits and
therefore the hammer was deemed appropriate
for driving the piles. However, the compressive
stresses imparted to the pile material were
relatively constant with depth at approximately
42MPa (6,100 psi) and were an area of concern.

[FIG. 4] Driving Concrete Piles

BEST ESTIMATE PREDICTION OF

PILE DRIVEABILITY
The pile, soil and hammer model was [FIG. 5] SRD predictions using A&H method
implemented in a one dimensional wave
equation programme as described in the
previous sections. The A&H SRD values and the
synthetic shear stress profile implemented in
the one dimensional wave equation software are
plotted in Figs. 5 and 6 respectively. The SRD
value is seen to increase from 0.6MN (67 ton)
for initial penetration to approximately 1.3MN
(146 ton) when the pile tip reaches 7m (23.0 ft)
depth. The base resistance accounts for nearly
100% of the initial SRD with the proportion of
shaft resistance increasing to 44% of the total
SRD at the end of installation.
The resultant installation response predicted by
the base case driveability analysis is shown in
Fig. 7. The initial analysis adopted the maximum
hammer drop heights to assess whether
sufficient energy was available to overcome
the soil resistance. The analysis yielded blow
counts that ranged from 10 blows/250 mm
[FIG. 6] Synthetic shear stress profile for A&H method
(10 blows/10 in) at shallow depths to
approximately 20 blows/250 mm (20 blows
/10 in) at the target penetration. To ensure a

DFI JOURNAL Vol. 7 No. 2 December 2013 [7]

 [FIG. 7] Driving Response for Hammer Operating at
Maximum Energy

The analysis was repeated for various drop

heights, as shown in Fig. 8, where increasing
the drop height increases the compressive
stress but reduces the blow counts. Both the
standard and upper bound A&H approaches
were used to predict the driving behaviour. To
achieve the optimum installation performance
it was decided to maintain the hammer drop
height between 400mm and 600mm (16 in
and 24 in) (i.e. less than 50% of the potential [FIG. 8] Driveability analysis to determine hammer drop
hammer energy). By restricting the hammer heights
drop heights the analysis showed that it was
possible to maintain the compressive stresses blow counts is due to the varying height of the
at an acceptable level, allowing the piles to be hammer impacts. PC5 was consistently struck
installed without damage. using lower hammer drop heights of 400mm
(16 in), whereas several of the other piles
PILE INSTALLATION PERFORMANCE used drop heights up to 600mm (24 in), which
The five concrete piles were installed in the resulted in lower blow counts.
dense sand in June 2012. The applied drop One of the precast piles (PC3) installed at the
heights were limited to 600 mm (24 in), which Blessington test site, exhibited early flaking of
allowed four of the piles to be installed to the the concrete near the pile head, which during
target depth without any issues. This procedure subsequent driving resulted in vertical cracks
allowed the blow counts to be maintained below extending several meters down the pile shaft
100 blows/250 mm (100 blows/10 in) as shown and ultimately led to complete failure of the
in Fig. 9, with the cumulative blows typically concrete (as shown in Fig. 10). The material
ranging between 1000 and 1500 to reach the failure was sufficient to reduce the cross
7m (23.0 ft) penetration target. The range in section to 65% of the original value, exposing

[8] DFI JOURNAL Vol. 7 No. 2 December 2013

 [FIG. 10] Pile damage on PC3

[FIG. 11] Piles installed to target depth

some variations in the hammer operation

[FIG. 9] Blowcount Records
between piles led to a spread in the measured
blowcounts (see Fig. 9). As the enthru energy
the longitudinal reinforcement and the shear from the hammer to the piles was not recorded
links. The damage was sufficient to force for every precast pile installed, this paper
premature refusal, with the pile tip terminating focuses primarily on the single pile for which
approximately 1.3 meters (4.3 ft) above the detailed driving measurements are available.
target penetration as seen in Fig. 11. The lack of monitoring on all five piles was
It is worth noting that due to the uniformity purely due to budgetary and time constraints
of the underlying ground conditions, all of and the authors recommend that for further
the piles were expected to exhibit similar field studies every effort should be made to
driving behaviour. However despite this, dynamically monitor every pile installed.

DFI JOURNAL Vol. 7 No. 2 December 2013 [9]

 MEASURED RESPONSE FROM PILE The CAPWAP signal matching process can be
DRIVING ANALYZER RESULTS used to determine an estimate of the shear
stress distribution along the pile shaft. The
Due to practical and financial constraints, only
shaft stress profile derived for each of the tip
one of the concrete piles (PC4) was monitored
penetrations at 3, 5 and 7m (10.0 ft, 16.4 ft
throughout driving, using a pair of strain
and 23.0 ft) is shown in Fig. 13, where friction
gauges and accelerometers placed on opposite
fatigue is evident between each of the profiles.
faces of the pile. The strain and acceleration
For example, at 4m (13 ft) depth, the shaft
generated by every hammer blow was measured
shear stress reduces from 140kPa (20.3 psi)
throughout driving and simultaneously
when the pile tip is at 5m (16.4 ft) to less than
recorded on an on-site dynamic data acquisition
system. As the strains and accelerations were
recorded for every blow during installation, it
was possible to derive a continuous estimate of
the SRD during installation for comparison with
the original design assumptions. The measured
SRD values are shown in Fig. 12. The continuous
profiles represent CASE method estimates of
the pile resistance (RMX and RX5). The CASE
estimates are determined from the force and
velocity inputs at times corresponding to the
ground and toe wave arrivals from the hammer
impact. The RMX value is determined for the
maximum total resistance during driving,
whereas RX5 assumes a damping factor, J
of 0.5. Selected blows at 3m, 5m and 7m
(10.0 ft, 16.4 ft and 23.0 ft) penetration have
also been extracted for further study using a
signal matching process. CAPWAP (CAse Pile
Wave Analysis Program) analysis allowed more
accurate estimates of the pile resistance during [FIG. 12] Measured Static Resistance to Driving (SRD) on
driving to be derived. In this instance, the Pile PC4
CAPWAP estimates are seen to provide excellent
agreement with the more rudimentary CASE
method capacities. The CAPWAP resistance
estimates are seen to increase from 1000kN
(112 ton) at 3m (10 ft) depth to approximately
1450kN (163 ton) at 7m (23.0 ft) depth. It is
worth noting that these values are slightly
higher than the range predicted by the original
A&H SRD model, which estimated values
that would increase from 700kN (78.7 ton) to
1200kN over the same penetration interval.
Furthermore, the factor of 1.25 for the
A&H UB approaches the measured values,
indicating that the driving resistance are a
close approximation to the upper bound design
predictions. It is worth noting that the A&H
methodology was derived for the specific case
of coring pipe piles and the full-displacement
nature of the concrete piles and the increased
surface roughness may have yielded the higher
resistances recorded.
[FIG. 13] Shaft Shear Stresses Derived from CAPWAP

[10] DFI JOURNAL Vol. 7 No. 2 December 2013

30kPa (4.4 psi) when the pile tip advances to 7m industry and are summarised in the American
(23.0 ft) depth. In short, continuous driving can Petroleum Institute API-RP2A (2007) design
cause a serious reduction in the shear strength guidelines. The API method has traditionally
at any given depth as the pile penetrates further been based on the earth pressure approach
into the ground. This implies that a suitable which relates the local unit shaft friction, τf,
SRD model should capture this effect, standing to the vertical effective stress, σ′v0, using an
to the applicability of the A&H approach which empirical factor β:
has been developed to incorporate this effect. ᇱ
߬௙ ൌ ߚǤ ߪ௩଴ [8]
The CAPWAP analysis can also be used to derive
the normal stress acting on the base of the where β ranges from 0.36 for medium-dense
concrete pile, as shown in Fig. 14. This stress deposits to 0.7 for very dense sand (for full
increases slightly with depth from 10500 kPa displacement piles). The API method calculates
(1,523 psi) to 12000 kPa (1,740 psi) over the the unit base resistance using traditional
depth range from 3m to 7m (10.0 ft to 23.0 ft). bearing capacity theory where:
This stress represents between 0.6 and 0.7
ᇱ
times the CPT tip resistance, qc. ‫ݍ‬௕ ൌ ܰ௤ Ǥ ߪ௩଴ [9]

where values of Nq range from 12 – 50 for

medium-dense to very dense sands. Several
studies, Chow (1997), Lehane, et al. (2005)
and others have shown the API method has
poor reliability, typically underestimating the
shaft resistance of piles in dense sand and
overestimating the resistance in loose sand. In
recent API editions, the method is precluded
from use in loose sands as the pile lengths
obtained were noted to be un-conservative
(API-2007).
Dennis & Olson, (1983), Lehane & Jardine,
(1994) and others have shown that the local
radial effective stress at failure, σ′rf, which
controls τf can be described using the Coulomb
failure criteria:
ᇱ
߬௙ ൌ ߪ௥௙ Ǥ ‫ߜ݊ܽݐ‬ [10]

where δf is the interface friction angle at failure.

[FIG. 14] Measured Base Stress During Driving Pile PC4 Tests performed using the closed-ended
Imperial College Pile (ICP) by Lehane (1992)
PARAMETRIC SRD STUDY and Chow (1997) showed that σ′rf comprised
A parametric study was completed to determine two discrete components, namely σ′rc, the
the impact of various SRD models on the pile radial effective stress after pile installation and
installation response to determine the worst equalisation and ∆σ′rd the increase in stress due
case scenario and to identify the upper bound to dilation during loading.
to the installation risk profile. ᇱ ᇱ ᇱ [11]
ߪ௥௙ ൌ ߪ௥௖ ൅ 'ߪ௥ௗ
Three static capacity approaches were
considered, including recent CPT based Jardine, et al. (2005) propose a direct correlation
methods that incorporate friction fatigue, between σ′rc and qc known as the IC-05 method.
to assess which of the methods are most The effects of friction fatigue (discussed earlier)
appropriate for use as an SRD model for driven are considered through a geometric term, h/R
concrete piles. (where h is the distance from the pile tip to
Design methods for estimating pile capacity the point under consideration and R is the
in sands have been developed by the offshore pile radius).

DFI JOURNAL Vol. 7 No. 2 December 2013 [11]

 ି଴Ǥଷ଼ ᇲ ଴Ǥଵଷ
ᇱ
ߪ௥௖ ൌ ͲǤͲʹͻǤ ‫ݍ‬௖ ቂ݉ܽ‫ ݔ‬ቀ ǡ ͺቁቃ
௛
൬
ఙೡబ
൰ [12]
ோ ௉ೝ೐೑

where pref =100 kPa (14.5 psi). Lehane et al.

(2005) suggest an alternative expression for σ′rc
known as the UWA-05 method:
௛ ି଴Ǥହ
ᇱ
ߪ௥௖ ൌ ͲǤͲ͵Ǥ ‫ݍ‬௖ ቂ݉ܽ‫ ݔ‬ቀ ǡ ͺቁቃ  [13]
஽

where D is the pile diameter. Both IC-05 and

UWA-05 approaches include expressions to
predict the stress increase ∆σ′rd caused by
dilation. Both are based on the work of Lehane
& Jardine (1994), who demonstrated through
cavity expansion theory that ∆σ′rd is inversely
proportional to the pile diameter pile diameter.
ସீ
ᇱ
οߪ௥ௗ ൌ ቀ ቁ Ǥ ο‫ݕ‬ [14]
஽

where G is the operational shear modulus of

the soil (which can be correlated with CPT qc) [FIG. 15] Comparison of SRD Methods
and ∆y is the radial displacement during pile
loading. [TABLE 2] SRD Predictions at Final Depth
Both the IC and UWA design methods use
notably different base resistance calculations. QT Qs Qb
In the IC-05 method the base resistance is SRD Model
[MN] [MN] [MN]
considered as a direct function of the average
CPT qc and the pile diameter. The CPT cone A&H 1.19 0.5 0.69
resistance is averaged over ±1.5 diameters
A&H UB 1.49 0.63 0.86
above and below the pile tip (qc,avg) and the base
capacity is calculated as follows: API 0.82 0.3 0.52
௤್
ൌ ƒšሾͳ െ ͲǤͷ Ž‘‰ ቀ
஽
ቁ ǡ ͲǤ͵ሿ [15] UWA05 1.69 0.79 0.90
௤೎ǡೌೡ೒ ஽಴ು೅
IC-05 1.68 0.76 0.92
Where DCPT = 0.036m (0.118 ft). The UWA-05
design method uses a more complex Dutch CPT
this study is focussing on their applicability
averaging technique and assumes qb is only a
for use as SRD models within a drivability
function of qc,dutch as follows:
framework. The static capacity approaches
௤್
ൌ ͲǤ͸ [16] have been designed to predict a long-term
௤೎ǡ೏ೠ೟೎೓
static capacity that accounts for some degree
of pile aging /setup, where the resistance will
DISCUSSION – SRD MODELS increase over time. As a result the red dots that
depict the end of installation (EOI) resistance in
The three different static capacity methods
Fig. 15 are likely to migrate closer to the UWA
described above were used to determine
prediction over time, suggesting that the UWA
the ultimate pile resistance values, which
method may be the most accurate approach
are illustrated in Fig. 15 and documented in
for long term resistance, while simultaneously
Table 2. The API method is seen to predict the
providing a conservative SRD estimate for
lowest resistance (0.82MN or 92.2 ton), while
driving analysis. By sharp contrast, the API
the UWA approach predicts the highest value
approach is seen to grossly underestimate the
(1.69MN or 190 ton). It should be noted that
installation resistance which would yield a non-
in contrast to the A&H method, the UWA, API
conservative SRD for input into a driveability
and IC approaches have been developed to
analysis. For long-term capacity predictions
calculate static pile capacity; and therefore
the API values would provide a conservative

[12] DFI JOURNAL Vol. 7 No. 2 December 2013

estimate of the pile failure loads; however been cast 156 days before driving compared
this may not result in an efficient pile design. with 66 days for pile PC3. Concrete cubes taken
As the pile capacity increases over time, the during pile casting from the two different
measured pile resistance will migrate further concrete batches are compared in Fig.17.
from the API predictions, leading to an even Seven-day and 28-day cube tests from the
greater inaccuracy in the predicted API PC3 batch indicated strength values of 72 and
values. Direct use of simplified static capacity 76 MPa (10,440 and 11,020 psi) respectively.
methods such as the API model for driveability Further cube tests from the PC4 batch indicated
predictions was recognised by Stevens et al. higher strengths varying from 90 – 96 MPa
(1982) to provide a non-conservative driveability (13,050 – 13,920 psi) although these were
prediction, whereas the UWA method appears first tested 157 days after casting (the day
to provide a more conservative SRD prediction after driving).
for implanting in driveability analysis. The
IC-05 method shows reasonable correlation [TABLE 3] Concrete Casting Times
with the measured values, whereas the Alm
and Hamre method slightly under predicts the Casting Driving
driving resistance. Pile Age (days)
Date Date
The relative conservatism of the different
design methodologies is shown in Fig. 16, which PC4+other 23/11/10 29/04/11 156
plots the maximum predicted blow counts and PC3 21/02/11 29/04/11 66
compressive stresses for the different design
SRD models. The API method results in the
lowest values while the UWA method predicts
higher stresses and blow counts. Employing the
UWA approach within a driveability analysis
for piles driven in dense sand will yield more
conservative estimates of the driving behaviour
and should allow a more appropriate driving
system to be selected without damaging
the pile.

[FIG. 17] Concrete Cube Strengths from the two

concrete batches

A further investigation into the concrete

[FIG. 16] Stresses and Blow Counts for Different SRD properties of PC3 was conducted by cutting
Models a 0.5m (1.64 ft) vertical section from the top
of pile PC3. Three horizontal drilled cores,
DISCUSSION – PILE PC3 FAILURE 100mm (4 in) in diameter and 200mm (8 in) in
After failure of pile PC3 during driving, it length were taken from the cut off section and
was discovered that this pile had been cast compression strength tests were conducted in
from a different concrete batch than the a Controls Automax 5 concrete compression
remaining precast concrete piles. Table 3 shows machine. The cores from pile PC3 indicated
a summary of the concrete casting and pile peak strengths of between 42 – 46 MPa (6,090
driving dates. PC4 and the rest of the piles had – 6,670 psi), significantly below the cast cube

DFI JOURNAL Vol. 7 No. 2 December 2013 [13]

 strength. Bungey & Millard (1996) suggest that the concrete strength would be an issue
that concrete cubes may have compression and this was verified by the failure of pile PC3.
strengths between 1.25 – 1.65 times higher In cases where a driveability analysis indicates
than equivalent cut cores (with an L/D=2), hard driving and high concrete stresses are to
depending on the curing, which may explain
this discrepancy.
Cast cylinder strengths, which are more typical
in the United States, were suggested to be
≈15% lower than cut cores, due to the weaker
top surface zone of cast cylinders (Bungey &
Millard, 1996). Traditional US design practice
recommends that driving stress not exceed
85% of the concrete cylinder strength minus
any effective pre-stressing (Goble & Hussein,
2000). Using the measured core strength
values this design practice would limit the
allowable compressive stress during driving on
pile PC3 to less than 31 MPa (4,500 psi). The
measured stresses on Pile PC4 are shown in
Fig. 18, where CSX is the average compressive
stress at the pile head; CSB is the average
compression stress on the pile base; and CS1
is the maximum compressive stress recorded
from the individual strain gauges. The CS1
values are typically 15 to 20% higher than the {FIG. 18] Measured Static Resistance to Driving (SRD) on
CSX values indicating some stress eccentricity Pile PC4
and non-uniform load transfer. This stress
eccentricity was not considered within the wave
equation simulations of the driving system.
The applied compression stresses are seen to
exceed 31MPa (4,500 psi) and approach the
failure strength of the concrete in pile PC3.
The other pre-cast piles (including PC4), which
had measured cube strengths approximately
25% higher than pile PC3, did not fail during
driving. The predicted stress range from 25 to
27 MPa (3,635 to 3,915 psi) (as shown in Fig. 8)
are seen to provide a reasonable match to the
average CSX stresses but are slightly lower than
the maximum measured stresses due to the
slightly higher SRD encountered in the field and
the stress eccentricity recorded.
It is worth noting that typical construction
practice often uses piles that have cured for
significantly shorter durations than used for
this project, with driving often taking place
7 days after casting. For relatively fresh piles,
the concrete is unlikely to have reached full
compressive strength and concrete crushing can
be a significant risk. This case history highlights
the benefit of performing a pile driveability
analysis before pile installation. The driveability
study performed as part of this paper, indicated
[FIG. 19] Concrete Cores from pile PC3

[14] DFI JOURNAL Vol. 7 No. 2 December 2013

be expected, it may be preferable to specify REFERENCES
a higher grade of concrete, or longer curing 1. Alm, T. & Hamre, L. (1998), "Soil model
times to allow the concrete to achieve the for driveability calculations", OTC 8835,
required strength. Offshore Technology Conference, paper No.
OTC 8835. Houston, TX, 4-7 May, 1998.
CONCLUSIONS
2. Alm, T. & Hamre, L. (2001), "Soil model for
This paper provides an interesting case study pile driveability predictions based on CPT
into the driveability of concrete piles in dense interpretations", Proceedings, The XV-th
sands. Five precast concrete piles were driven International Conference. on Soil Mechanics
into the dense sand deposit in Blessington, and Geotechnical Engineering, Istanbul,
Ireland. An original driveability study was Vol.3, pp. 1297-1302.
undertaken to analyse the performance of
the hammer-soil-pile system. The measured 3. API 2007, "Recommended Practice for
response showed that the Alm and Hamre SRD Planning, Designing and Constructing
model provided reasonable estimates of the Fixed Offshore Platforms – Working Stress
installation resistance. By contrast, the UWA Design", API RP2A. Washington, D.C.,
and IC-05 pile capacity models were relatively American Petroleum Institute.
conservative yielding higher SRD values, which 4. Bungey, JH. & Millard, SG. (1996), "Testing of
were more in line with long-term static values. concrete in structures, 3rd Edition", Blackie
The traditional API model was the least accurate Academic & Professional, Cambridge.
and non-conservative of all models considered 5. Chow, F.C. (1997), "Investigations into the
in the analysis (when used for SRD calculations). behaviour of displacement piles for offshore
Friction fatigue was also measured in the field foundations", PhD thesis, University of
and therefore should be incorporated in any London (Imperial College).
driveability approach considered.
6. Dennis, N. D. & Olson, R. E. (1983),
One pile (PC3) experienced concrete cracking "Axial capacity of steel pipe piles in
and failed during driving. Analysis of the pile sand", Geotechnical Practice in Offshore
indicated that the pile that failed was cast from Engineering. pp. 389–402.
a different concrete batch and cores tested
7. Doherty, P., Kirwan, L., Gavin, K.; Igoe, D.;
after driving indicated the pile had a lower
Tyrrell, S.; Ward, D. & O'Kelly, B., (2012).
concrete strength. Using the measured core
"Soil properties at the UCD geotechnical
strength values, traditional design practice
research site at Blessington", Bridge and
would limit the compressive stress during
Concrete Research in Ireland 2012, Dublin,
driving to less than 31 MPa (4,500 psi). The
Irl, 6-7 September 2012, 499-504.
measured stresses on an adjacent pile showed
the stresses to exceed this value, pointing 8. Gavin, K, Adekunte, A & O’Kelly, B. (2009),
toward the most likely cause of failure. The "A field investigation of vertical footing
other precast piles, which had 25% higher cube response on sand", Proceedings of the ICE -
strengths, all survived driving and reached their Geotechnical Engineering, 162 (5) 2009-10,
target penetration. pp.257-267.
9. Gavin, K. & O'Kelly, B. (2007), "Effect
ACKNOWLEDGMENTS of friction fatigue on pile capacity in
The authors would like to acknowledge the dense sand", Journal of Geotechnical and
financial support received from Science Geoenvironmental Engineering, 133(1), pp.
Foundation Ireland, Enterprise Ireland and 63-71.
IRCSET to undertake field testing at Blessington. 10. Goble, G. G., Fricke, K. E. & Likins, G. E.
Furthermore, the first author would like to (1976), “Driving Stresses in Concrete Piles”,
thank our enterprise partner, Mainstream Prestressed Concrete Institute Journal,
Renewable Power, for supporting this research Vol.21(1) pp. 1-20.
initiative. Technical assistance received from Dr.
Kenneth Gavin is gratefully acknowledged.

DFI JOURNAL Vol. 7 No. 2 December 2013 [15]

 11. Goble, G. G., Hussein, M.H., (2000). 16. Lehane B.M. & Jardine R.J. (1994), "Shaft
“Potential for HPC in Driven Pile capacity of driven piles in sand: a new
Foundations“, Proceedings of the PCI/ design approach", Proceedings of Conference
FHWA/FIB International Symposium on High on the Behaviour of Offshore Structures,
Performance Concrete: Orlando, FL;pp. 608- Boston, 1 pp. 23- 36.
615. 17. Lehane, B.M., Schneider, J.A., and Xu, X.
12. Heerema, E.P, (1980), "Predicting pile (2005). "The UWA-05 method for prediction
driveability: heather as an illustration of the of axial capacity of driven piles in sand",
friction fatigue theory", Ground Engineering, Proceedings, International Symposium,
13(3), pp. 15-37. Frontiers Offshore Geomechanics. ISFOG,
13. Heerema, E., (1981), "Dynamic point Perth, 683-689.
resistance in sand and in clay, for pile 18. Schneider, J. A. & Harmon, I. A., (2010).
driveability analysis", Ground Engineering, 1 "Analyzing Drivability of Open Ended Piles
4(6), pp. 30-46. in Very Dense Sands", DFI Journal, IV(1), pp.
14. Jardine R, Chow, F, Overy, R & Standing, 3-15.
J (2005), "ICP design methods for driven 19. Stevens, R.F., Wiltsie, E.A., & Turton, T.H.
piles in sands and clays", London: Imperial (1982), "Evaluating Pile Driveability for
College London Hard Clay, Very Dense Sand and Rock",
15. Lehane, B. M. (1992), "Experimental Proceedings, 14 Offshore technology
investigations of pile behaviour using Conference, Houston, Vol 1, pp 465-479
instrumented field piles", PhD thesis, 20. Toolan, F.E. & Fox, D.A. (1977), "Geotechnical
Imperial College, University of London. planning of piled foundations for offshore
platforms", Proceedings Institution of Civil
Engineers, May 1977 62(1) pp. 22 -244

[16] DFI JOURNAL Vol. 7 No. 2 December 2013

Ant Colony Optimization Method for Design of
Piled-Raft Foundations (DFI 2013 Student Paper
Competition Winner)
Hessam Yazdani, PhD Candidate, School of Civil Engineering and Environmental Science,
University of Oklahoma, Norman, OK, USA; (405) 325-5218; [email protected]
Kianoosh Hatami, Associate Professor, School of Civil Engineering and Environmental Science,
University of Oklahoma, Norman, OK, USA
Elahe Khosravi, Graduate Student, School of Civil Engineering and Environmental Science,
University of Oklahoma, Norman, OK, USA

ABSTRACT
In comparison to conventional piled foundations, piled-raft foundations provide a more economical
solution to support high-rise buildings constructed on compressible soils. In this type of foundation,
the bearing capacity of the underlying soil is taken into account in supporting the superstructure
loads, and the piles are placed to control both the total and differential movements of the
superstructure. Currently, there are no universally accepted methods to design piled-raft foundations
including the selection of the piles locations and dimensions. Most piled-raft foundation designs
are based on empirical methods and the experience of designers. However, piled-raft foundations
are massive and expensive. Therefore, developing methodologies for their optimal design could
significantly help minimize their otherwise high construction costs and would make them more
feasible and common practice. This paper examines the capability of the ant colony optimization
(ACO) algorithm to optimize piled-raft foundations. The soil-pile interactions are taken into account
by modeling the side and tip capacities of the piles using the nonlinear p-y, t-z, and Q-z springs
in the OpenSees platform. The soil-raft interaction is taken into consideration using the Winkler
springs beneath the raft. The objective of the optimization problem is to minimize the volume of the
foundation by taking the number, configuration, and penetration depth of the piles, as well as the
thickness of the raft, as design variables. The side and tip forces of the piles, the pressure applied on
the underlying soil, and the total and differential movements of the foundation under the serviceability
limit state are the constraints adopted for the optimization problem. Results indicate that the ACO
algorithm is a suitable method for optimal design of piled-raft foundations. Findings of the study also
indicate that including soil nonlinearity in the analysis (as opposed to a linear elastic soil model) can
lead to a more economical design for these foundation systems.

INTRODUCTION required factors of safety within the framework

Piled foundations are best suited for sites of the allowable stress design could often
where a shallow foundation may incur excessive lead to overly conservative, and hence costly
movements (settlements) and may not provide solutions (Poulos and Davids, 2005).
adequate bearing capacity to carry structural Piled-raft foundations are an economical
loads. Current design guidelines primarily alternative to the conventional piled
require that the piles should carry the entire foundations when competent soil strata exist
structural load of a piled foundation and immediately beneath the raft. In contrast to
transfer it to deeper and more competent layers piled foundations, structural loads supported
(de Sanctis and Mandolini, 2006; Sales et al., by piled-raft foundations are mostly carried
2010). However, field monitoring of several by the raft (Burland et al., 1977). The piles,
piled foundations has revealed that the raft known as the settlement-reducing piles, are
could significantly increase the overall bearing therefore located strategically to enhance the
capacity of a piled-raft foundation system bearing capacity of the raft besides controlling
(Kakurai, 2003). Consequently, designing a piled or minimizing the total and differential
foundation merely as a pile group to meet the movements that may cause distortion and

DFI JOURNAL Vol. 7 No. 2 December 2013 [17]

 cracking of the superstructure (Randolph, 1994; combining basic heuristic methods in higher
Momeni et al., 2012; Yazdani et al., 2013). Such level frameworks (Blum et al., 2011). This class
a design approach can significantly reduce the of algorithms includes swarm intelligence
cost of the foundation without jeopardizing the (imitating the processes of decentralized,
safety and performance of the superstructure self-organized systems such as ant colony
(Sales et al., 2010). optimization - ACO and particle swarm
Optimal design of piled-raft foundations could optimization - PSO), evolutionary computation
significantly help minimize their construction approaches (which are non-gradient, population-
costs. The optimal design of piled-raft based algorithms such as genetic algorithms
foundations includes the selection of type, - GAs and evolutionary programming - EP),
number, configuration and penetration depth of simulated annealing (SA), and tabu search (TS),
the piles in addition to the thickness of the raft among others. The efficiency and robustness
in conformance with the existing design and of an optimization approach depend on the
construction standards (Gates and Scarpa, 1984; problem in hand, and therefore, no globally
Prakoso and Kulhawy, 2001). Several factors accepted approach has been proven to best fit
influence the design of piled-raft foundations all engineering optimization problems (Rajeev
including the structural loads, the properties of and Krishnamoorthy, 1992).
foundation soil and the negative skin friction on Geotechnical systems in general and
piles associated with long-term settlements of foundations in particular have not so far
a foundation underlain by soft ground (Poulos benefitted significantly from the existing
and Davis, 1980). metaheuristic algorithms for optimal design
Optimization of piled and piled-raft (Cheng et al., 2007). Chan et al. (2009) proposed
foundations has been the subject of a few a modified genetic algorithm, which showed
past studies. Chow and Thevendran (1987) promise in optimizing a series of pile groups.
carried out an optimization analysis on pile Khajehzadeh et al. (2011) proposed a modified
groups and concluded that the central and particle swarm optimization technique for
peripheral piles of a pile group need to be optimal design of spread footings and retaining
designed with different degrees of rigidity in walls. In this paper, an ACO algorithm is
order to minimize the differential movements developed and used to optimize piled-raft
of a group under a flexible pile cap, or the foundations. The ACO approach was selected
load differentials among piles under a rigid because it retains the memory of the entire
cap. Valliappan et al. (1999) used a theoretical colony from all generations to approach the
optimization approach together with the finite optimal solution, as opposed to GA-based
element method to analyze and optimize methods in which the search information is
piled-raft foundations. They investigated two contained only in the current generation of a GA
cases of uniform and non-uniform pile lengths (Camp and Bichon, 2004).
and concluded that using non-uniform pile
lengths increases the contribution of the raft in ANT COLONY OPTIMIZATION
supporting the structural load. Kim et al. (2001) TECHNIQUE
used recursive quadratic programming to find Inspired by the foraging behavior of blind
an optimum configuration for the piles which animals such as ants (see Deneubourg et
would minimize differential movements of al., 1990), Dorigo et al. (1996) proposed a
piled-raft foundations. However, their approach population-based metaheuristic approach,
did not fully account for the interactions among called ant colony optimization (ACO), in which
the elements of the foundations (i.e. the raft, a colony of artificial ants is used to construct
piles and the underlying soil). Wang et al. (2002) solutions guided by pheromone intensities and
introduced an analytical method for the optimal heuristic information. Pheromone is a substance
design of piled foundations in nonlinear soils. ants deposit on the ground when carrying
However, their procedure included only a food from sources to the nest. Ants use this
limited number of design parameters. medium to indirectly communicate information
Metaheuristic algorithms are approximate regarding the shortest paths between feeding
but efficient algorithms that can be used to sources and the nest and use the intensity of
explore a search space for optimal solutions by pheromone to evaluate the potential of marked

[18] DFI JOURNAL Vol. 7 No. 2 December 2013

paths. In other words, the probability of a
path to be chosen by an isolated ant moving
essentially at random, searching for food, is
proportional to the pheromone intensity of
the path. The greater the pheromone intensity
marking the path (i.e., the greater the number
of ants selecting the path), the stronger the
stimulus for an ant to follow that path, thus
reinforcing the trail with its own pheromone
and increasing the probability that subsequent
ants will follow this path (Blum et al., 2011).
In the ACO technique, several terms are adopted
from the graph theory. Graphs are mathematical
structures used to model pairwise relations
between objects. The objects in a graph are
represented by vertices and the pairwise
[FIG. 1] ACO Algorithm
relations are established using edges connecting
the vertices. In ACO, the optimization problem
1
is projected on a graph, where the shortest path τ0 = [1]
nLnn
determines the optimal solution of the problem.
Consider a combinatorial optimization problem where Lnn is the length of the tour between n
projected to a multigraph (a graph which is cities created by the nearest neighbor heuristic
permitted to have multiple edges, also known as (i.e., the minimum value of the optimization
pseudograph), defined over a set x = {xi | i = 1, problem which is obtained by assigning the
…, n} of design variables, where n is the number smallest options to n design variables - Camp
of design variables (vertices), and a set E = { eij and Bichon, 2004). It does not matter whether
| i = 1, …, n; j = 1, …, Ji} of options (potential this assignment satisfies the design constraints
values) for variables, where Ji is the number of or not (Rajasekaran and Chitra, 2009; Aydoğdu
options for the ith design variable (number of and Saka, 2012).
edges departing from the ith vertex when moving
forward). A subset ψ of edges represents a Construction:
solution of the problem. Since each edge has A set of concurrent and asynchronous agents
the possibility of being selected or rejected, (a colony of ants) travel between vertices and
there are 2Jn feasible solutions for each problem, construct solutions to the problem. Starting
where J = ¦ J i Let C = { c | i = 1, …, n; j = 1, …, from an arbitrary or pre-selected design
ij
i =1 variable, i, ant k applies a stochastic local
Ji} is the set of costs (weights) corresponding
decision policy to select one of the Ji available
to the potential values for variables. Given this,
options for the design variable. Known as
the major steps in an ACO algorithm, shown
random proportional transition rule (also
in Fig. 1, are as follows (Maniezzo et al., 2004;
known as pseudorandom proportional rule -
Uğur and Aydin, 2009; Moeini and Afshar,
Maniezzo et al., 2004; Dorigo et al., 2006), this
2012):
decision policy is governed by two parameters,
Initialization namely visibility (also known as attractiveness
of the move) and trail level (also known as
The size of the colony (number of ants), m, is pheromone intensity). Visibility indicates a
chosen and a proper initial intensity of priori desirability of a move and is an artificial
pheromone, τ 0, is assigned to all options, eij. sight for selecting the shortest path among
Increasing the size of the colony can improve the options without experience or observation.
the quality of the final results achieved by ACO. Trail level, in contrast, can be interpreted as
However, colony size should be chosen such an adaptive artificial memory, and indicates
that a good tradeoff between solution quality a posteriori, the desirability of the move and
and computational effort is guaranteed (Viana reflects the experience acquired by the ants at
et al., 2008). Initial intensity of pheromone is this stage. The transition rule used by ant k to
usually assumed to be: select one of the options is:

DFI JOURNAL Vol. 7 No. 2 December 2013 [19]

 α β positive feedback from the ants constructing
ªτ ij (t ) º¼ ª¬ηij º¼
pij (k , t ) = Ji¬ [2]
satisfactory solutions by reinforcing the
α β
trail level on the paths they have selected.
¦ ª¬τ ij (t )º¼ ª¬ηij º¼
j =1 Secondly, a negative feedback mechanism
reduces the trail level on the selected paths
where pij(k,t) is the probability that ant k selects
to promote exploration of the search space.
option j of the ith decision point, eij, at iteration
The rule corresponds to the evaporation of
t; τ ij is the trail level on option eij at iteration t;
ηij is the visibility of the ant representing the the substance in nature and helps prevent
local cost of choosing option j at the ith decision early stagnation, an undesirable situation in
point ( ηij =1/cij ); and α and β are two which all ants repeatedly construct the same
parameters regulating the relative importance solutions making any further exploration in the
of trail level versus visibility. An iteration comes search space impossible. The amount of this
to an end when m moves are carried out by m trail reduction is kept low to guarantee overall
ants, each making one move, in the time interval solution convergence.
(t, t + 1). ACO algorithms are different from each other
After each iteration, a local update rule is with respect to the techniques adopted to
applied to reduce the trail level relative to update the pheromone level and to implement
the options most recently chosen for design the random proportional transition rule.
variables (the paths chosen by ants) in order Interested readers are referred to a summary
to prevent premature convergence to a sub- provided by Uğur and Aydin (2009). A variation
optimal solution. When an ant travels from of the ACO approach, called the ranked-
city i to city j, an optimization rule is applied based ant system (RBAS - Bullnheimer et al.,
to adjust the intensity of trail on the path 1997) is used in this study. In the RBAS, only
connecting these two cities by: λ top ranked ants, having the best designs,
are selected in the global update scheme. As
τ ij (t ) = (1 − ij )τ ij (t ) + ijτ 0 [3] positive feedback, the RBAS increases the trail
level by Δτ ij , corresponding to the solution
+
+
found by the elite ant, ψ , as:
where ij (0 ≤ ij ≤ 1) is the coefficient of decay
(Uğur and Aydin, 2009), representing the 1 [4]
Δτ ij+ =
persistence of the trail. Ants incrementally f (ψ + )
construct m solutions for the problem by
choosing paths to travel between decision where f (ψ ) is the value of the objective
+
+
points, visiting each point once, until all points function associated with the solution ψ . The
have been visited and they arrive back at their change in the trail level of the path i-j, if chosen
starting points. When they return to their point by the ant ranked μ (1 ≤ μ ≤ λ), is given by:
of origin, the ants have completed a tour, and R
each has constructed its own trial solution, ψk. Δτ ijμ = [5]
f (ψ μ )
A cycle is complete when m ants complete their
tours and construct m trial solutions.
where f (ψ ) is the fitness value of the solution
μ

Evaluation: made by the ant ranked μ, and R is a quantity

regulating the contribution of the top ranked
The objective function is evaluated for the trial
ants called pheromone reward factor (Moeini
solutions and their fitness values (global scores)
and Afshar, 2012). Camp and Bichon (2004)
are calculated.
used R = λ - μ in their study, meaning the
Modification: contribution of a top ranked solution is linearly
proportional to its ranking. The path i-j may be
Once m ants have completed a tour and the
selected by more than one ant. Therefore, the
objective function has been evaluated for the
total increase in the trail level of the path is
solutions constructed, a global update rule
given by:
is applied to the options selected for design λ −1
variables. Typically, the global update rule Δτ ijr = ¦ Δτ ijμ [6]
has a dual function in communication among μ =1

the colony agents. First, it implements a

[20] DFI JOURNAL Vol. 7 No. 2 December 2013

Therefore, the updated trail levels at the end of ANALYSIS OF PILED-RAFT
a cycle (at time t + n) are a function of the trail FOUNDATIONS
levels at the beginning of the cycle, the tour
In this study, the OpenSees (Open Source
constructed by the elitist ant, and the tours
for Earthquake Engineering Simulation)
made by the top ranked ants, and they are
finite element platform was used for three-
calculated as:
dimensional analysis of piled-raft foundations.
τ ij (t + n) = (1 − ρ )τ ij (t ) + λΔτ ij+ + Δτ ijr [7] The piles were assumed to be drilled piles,
where the pile capacity is provided by a
where ρ is an adjustable parameter in the combination of soil-pile friction and end
range 0 ≤ ρ ≤ 1 so that (1 – ρ) represents the bearing resistance at the pile tip (Brown et
evaporation rate (Camp and Bichon, 2004). al., 2010). In practice, the geological and local
soil conditions govern the depth of the piles.
The trail level update rule is followed by a
However, the underlying soil was assumed to
feasibility analysis of the solutions constructed.
be a homogenous, dry medium dense sand
An analysis is carried out for each solution.
for simplicity. The analysis was carried out
If any constraints are violated, a penalty is
using two different assumptions of linear and
applied to the objective function corresponding
nonlinear material behavior. Piles were modeled
to the solution. The value of the penalty is
using nonlinear beam-column elements. The
proportional to the extent of the violation of
soil-pile interactions were modeled using the
constraints (Camp et al., 1998). Many constraint-
nonlinear p-y, t-z, and Q-z springs (Fig. 2).
handling techniques for evolutionary algorithms
Vertical nonlinear springs (t-z) were used to
have been proposed in the literature (e.g.,
describe the relationship between mobilized
static penalties, dynamic penalties, adaptive
soil-pile shear transfer and local pile deflection
penalties). Interested readers are referred to
at any depth. The relationship between the
a comprehensive survey by Coello (2002). In
lateral soil resistance and pile deflection
this study, objective functions that violate the
was modeled using p-y springs (Boulanger et
imposed constraints are penalized using a static
al., 1999; API, 2000). The load-displacement
penalty approach as follows:
response of the tip resistance of the piles
f p (ψ k ) = f (ψ k ) [1 + Φ ] [8]
ε
was modeled using vertical Q-z springs. The
beam on nonlinear Winkler foundation (BNWF)
where f p (ψ ) is the penalized objective function
k
framework (Raychowdhury and Hutchinson,
of the kth ant, Φ is the total penalty and the 2010) was adopted to model the raft resting
summation of the force and deflection penalties on the soil. BNWF includes a fine mesh of
(described in “Formulation of the Design independently distributed, nonlinear inelastic
Problem”), and ε is a positive penalty exponent, springs placed vertically beneath the raft to
which remains constant in the static penalty capture its rocking and settlement movements.
approach and is adjusted proportionally to The mechanical response of the springs was
the extent of the violation of the constraints assumed as linear elastic in the linear analysis.
in adaptive penalties (Camp and Bichon, 2004;
Rajasekaran and Chitra, 2009; Aydoğdu and
Saka, 2012).

Termination:
The ant decision mechanism, steps 2-4,
continues until either a maximum number of
cycles has been completed or all ants construct
the same solutions.

Production:
The outputs of the optimization process are
obtained.

[FIG. 2] Soil-Pile-Raft Interaction Model

DFI JOURNAL Vol. 7 No. 2 December 2013 [21]

 Table 1 summarizes the soil and other input design were not included in the objective
parameters used in the analysis. function because they can be determined using
pertinent design guidelines after the design
[TABLE 1] Soil Properties and other Input variables are determined numerically using the
Parameters optimization algorithm.

Design Constraints
Model Model Parameters Value
The structural and geotechnical capacities
of the piles, and the total and differential
Initial unit weight (kN/m3) 18.0 movements of the foundation under the
Peak friction angle, ϕp (°) 35 serviceability limit state were considered as the
Void ratio, e 0.55 constraints for the optimization problem in this
Initial modulus of study. Depending on their type, the ultimate
p-y
subgrade reaction, 43.0 structural capacities of piles in compression
(API, 2000) k, (MN/m3) c
and tension, Ps −u and Pst−u , are governed by the
The friction angle of 20 compressive strength of concrete, yield strength
t-z
soil-pile interface, δ (°) of steel, and the contributions of concrete,
(API, 2000)
(z/D)max(1) 0.01 steel reinforcing bars, and steel casing in the
Q-z cross-sectional area of the piles. The ultimate
c
capacity of piles in compression, Pg − u, is the
(pile tip) (z/D)max(2) 0.1
sum of the tip and side resistances mobilized at
(API, 2000)
failure. In contrast, the ultimate uplift capacity
Q-z (raft) t
of piles, Pg − u, is governed by the self-weight of
Elastic modulus of soil, 35.0
(Raychowdhury the piles and their side resistances, neglecting
and Es (MPa)
the weight of the wedge of the soil around the
Hutchinson, Stiffness ratio, Rk 1
2010) piles accompanying them in tension (Brown
et al., 2010). The allowable values of the piles
1
z: Local pile deflection, D: Pile diameter vertical movement, δ v − a , and pile-head lateral
2
z: Axial tip deflection, D: Pile diameter displacement, δ h − a , were set to 60 mm and 25
mm (2.4 in and 1.0 in), respectively (Budhu,
FORMULATION OF THE DESIGN 2007). The allowable differential movement
PROBLEM of the piles is expressed in terms of angular
distortion of the raft (defined as the ratio of
Objective Function and Design Variables differential movement between two adjacent
In this study, an ACO approach was used to columns/piles to the distance between them),
minimize the material cost of a piled-raft βa, which was set to 1/500 (Chan et al., 2009).
foundation and meet design requirements. The The piles are allowed to operate at 100% of
objective function is formulated as: their ultimate load capacity (Randolph, 1994).
np However, a margin of safety was applied on
π
Minimize V (x) = ¦ d p2 L p + BLt [9] the ultimate bearing capacity of the raft, σu,
p =1 4
using a safety factor of 3 (σa = σu/3, where σa
where x is a vector containing n design is the allowable bearing capacity of the raft).
variables; dp and Lp are the diameter and length Finally, the constraints of the problem and their
of the pth pile (np piles overall); and B, L and t associated penalties as well as the constraints-
are the breadth, length, and thickness of the handling technique used in this study are as
raft, respectively. For ease of construction as given below:
well as to reduce the pile-pile interaction, the • The structural force penalty for pile p,
piles were assumed to be located on the nodes Φ Pp − s , is calculated as (where compressive
of a regular latticework with a constant spacing forces are taken to be positive):
of three times the average piles diameter
­ Pi − Psc−,ut
(Brown et al., 2010). The total number of piles, t c p
° Pi < Ps −u or Pi > Ps −u Ÿ Φ P − s =
® Psc−,ut [10]
their configuration and penetration depth and
° t c p
the thickness of the raft were taken as design ¯ Ps −u ≤ Pi ≤ Ps −u Ÿ Φ P − s = 0
variables. Details of the steel reinforcement

[22] DFI JOURNAL Vol. 7 No. 2 December 2013

• The geotechnical force penalty for
p
RESULTS AND DISCUSSIONS
each pile p, Φ P − g , is calculated as:
The capability of the ACO algorithm to
­ Pi − Pgc−,tu optimize the piled-raft foundations is
°° Pi < Pgt −u or Pi > Pgc− u Ÿ Φ Pp − g = demonstrated using the following example.
® Pgc−,tu [11]
° t c p
The optimization task was carried out using
°̄ Pg − u ≤ Pi ≤ Pg − u Ÿ Φ P − g = 0 a mutual communication between MATLAB®
and OpenSees. The algorithm was executed
Hence, the total force penalty for the solution
in MATLAB environment and the foundations
generated by ant k is:
np
generated by ants were analyzed in OpenSees
Φ kP = ¦ ª¬ Φ Pp − s + Φ Pp − g º¼ [12] platform. OpenSees results were imported in
p =1 MATLAB to evaluate the objective function,
fitness value, and penalties corresponding to
• For pile p, the vertical movement penalty,
Φδpv , and the pile-head lateral displacement each solution. This information was used to
p
penalty, Φδ h, are stated as: direct the next ant colony towards the optimal
solution. The process continued until either a
­ p δ vp − δ v − a maximum number of cycles was met or all ants
°δ v > δ v − a Ÿ Φδ v =
p

® δ v−a [13] constructed the same solutions.

°δ p ≤ δ Ÿ Φ p = 0
¯ v v−a δv
Example Problem
­ p δ hp − δ h − a Consider a high-rise building with the structural
°δ h > δ h − a Ÿ Φδ h =
p

δ h−a [14] loads shown in Fig. 3. A 20 m × 20 m (66 ft

®
°δ p ≤ δ Ÿ Φ p = 0 × 66 ft) reinforced concrete raft has been
¯ h h−a δh proposed to safely transfer the building loads
• The angular distortion penalty between piles to the underlying sandy soil, with the properties
p,q
p and q, Φ β , is: as given in Table 1.
­ β p,q − β a Assuming that the groundwater level is well
°β p,q > β a Ÿ Φ β =
p,q

βa below the ground surface, and the embedment

°
® [15] depth of the raft is 1 m, the ultimate net
° δ vp − δ vq bearing capacity of the raft using Meyerhof’s
° β p , q ≤ β a Ÿ Φ p,q
β = 0; β p , q =
¯ Lp , q equation is calculated as (Buhdu, 2007):

where Lp,q is the distance between piles p and q.

Therefore, the total deflection penalty for the
solution generated by ant k is:
np np np

Φδk = ¦ ª¬Φδpv + Φδph º¼ + ¦¦ Φ βp , q [16]

p =1 p =1 q =1
q≠ p

• For the solution generated by ant k, if the

bearing pressure of the raft, σ, exceeds its
allowable bearing capacity, σmax, the follow-
ing penalty is applied:
­ σ − σ max
°σ > σ max Ÿ Φσ =
k

® σ max [17]
°
¯σ ≤ σ max Ÿ Φσ = 0
k

Therefore, the total penalty for ant k, Φ k , in

Equation 8 is given by:

Φ k = Φ kP + Φδk + Φσk [18]

[FIG. 3] Structural Loads on Raft
A static constraints-handling technique was
considered in this study in which a constant
positive penalty exponent of ε = 2 was used in
Equation 8.

DFI JOURNAL Vol. 7 No. 2 December 2013 [23]

 [TABLE 2] Input Parameters for the ACO
σ u = γ D f ( N q − 1) sq d q + 0.5γ B ' Nγ sγ dγ [19] Algorithm

where Nq and Nγ are bearing capacity factors

that are functions of the peak friction angle, Parameter Value
Φp; sq and sγ are shape factors; dq and dγ are
embedment depth factors; and B' is the Size of the colony, m 100
equivalent footing breadth. After the ultimate Evaporation rate, ρ 0.1
net bearing capacity is determined, the Coefficient of decay, φ 0.1
allowable bearing capacity can be calculated as: α 5
σu β 0.2
σ max = + γ Df [20]
FS λ 10

Therefore,
σ u = 18 ×1× 32.2 ×1.7 ×1 [TABLE 3] Potential Values for Design Variables

+ 0.5 ×18 × 20 × 37.1× 0.6 ×1 = 4,991kPa

16 × 20, 000 Parameter Value
σ= = 800 kPa
20 × 20
4991 Pile diameter, dp (m) 0.8, 1.0, 1.2
800 = + 18 ×1 Ÿ FS = 6.4
FS Pile length, Lp (m) 20, 25, 30
The vertical elastic settlement of the raft is Raft thickness, t (m) 1.0, 1.2, 1.4
given by: Grid spacing (m) 3

σ B(1 −ν 2 )
δv = Is Fig. 4 shows the convergence history of the
E [21] foundation design represented by the raft
L volume. The optimization process is initiated
I s = 0.62 ln( ) + 1.12
B with a design generated by randomly chosen
where σ is the surface stress and Is is a values for the design variables and evolves to
settlement influence factor. The settlement is an optimal design. Results in Fig. 4 indicate
calculated as: that the ACO algorithm yields an optimum
solution in approximately 25 cycles using linear
I s = 0.62ln(1) + 1.12 = 1.12 analysis and 31 cycles using nonlinear analysis.
800 × 20 × (1 − 0.32 ) However, including the soil nonlinearity in
δv = ×1.12 = 485mm the analysis results in a 5% more economical
30,000
design based on the volume of the foundation
Therefore, the raft provides a bearing capacity raft. The 5% reduction in the volume of the raft
adequate to carry the working loads with a large could prove significant, or even critical, in cases
safety factor. However, the raft will experience where physical obstacles are present for the
excessive settlements from the serviceability construction of the foundation.
point of view. Therefore, settlement-reducing
piles are required to control the building
settlements, and to reduce the bending
moments in the raft.
The ACO algorithm is used to determine
the optimal design of the foundation. The
values used for the algorithm parameters are
summarized in Table 2. A set of candidate
values for the design variables are given in
Table 3.

[FIG. 4] Foundation Design’s Convergence History

[24] DFI JOURNAL Vol. 7 No. 2 December 2013

The influence of soil nonlinearity on the CONCLUSIONS
optimum configuration of piles is depicted in The capability of the ACO algorithm for
Fig. 5. It can be observed that assuming a linear optimizing piled-raft foundations was
elastic behavior for the soil increases the load investigated. Soil-pile-raft interactions were
concentration of the peripheral piles and results included in the analysis using nonlinear p-y,
in an overdesigned foundation system. In t-z and Q-z Winkler springs in the OpenSees
contrast, taking soil nonlinearity in the analysis platform. The analysis of a piled-raft foundation
reduces the share of the load carried by the system indicated that including soil nonlinearity
peripheral piles and results in a redistribution in the analysis results in a more uniform
of loads among the load carrying elements. As a distribution of predicted loads among the piles
result, it helps achieve a more favorable design and hence a more economical design. Results
in which all piles carry comparable loads as shown in this paper illustrated that the ACO
the load applied on the foundation increases. algorithm developed in this study is capable
Results shown in Fig. 5 are in conformity with of finding an optimal solution for piled-raft
those reported in the literature (e.g., Basile, foundation systems. Further work is underway
2003). to compare the performance of the ACO
algorithm with other metaheuristic algorithms
in order to identify faster and more effective
methods for optimal design of foundations.

REFERENCES
1. API, (2000) “Recommended practice for
planning, designing and constructing fixed
offshore platforms—working stress design”,
API Recommended Practices, American
Petroleum Institute.
2. Aydoğdu, İ. and Saka, M. P., (2012) “Ant
colony optimization of irregular steel
frames including elemental warping effect”,
Advances in Engineering Software, Vol. 44,
No. 1, pp. 150–169.
3. Basile, F., (2003) “Analysis and design of
pile groups”, in Numerical Analysis and
Modelling in Geomechanics, J. W. Bull,
Editor, Spon Press (Taylor & Francis Group
Ltd), Oxford, pp. 278–315.
4. Blum, C., Puchinger, J., Raidl, G. R. and
Roli, A., (2011) “Hybrid metaheuristics in
combinatorial optimization: A survey”,
Applied Soft Computing, Vol. 11, No. 6, pp.
4135–4151.
5. Boulanger, R., Curras, C., Kutter, B., Wilson,
D. and Abghari, A., (1999) “Seismic soil-
pile-structure interaction experiments and
analyses”, ASCE Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 125, No.
9, pp. 750–759.
6. Brown, D. A., Turner, J. P. and Castelli,
R. J., (2010) “Drilled shafts: construction
procedures and LRFD design methods”,
National Highway Institute, U.S. Department
of Transportation, Federal Highway
[FIG. 5] Piles Configuration in Optimal Solutions; a)
linear analysis, b) nonlinear analysis Administration, Washington DC.

DFI JOURNAL Vol. 7 No. 2 December 2013 [25]

 7. Budhu, M., (2007) Soil Mechanics and 18. Deneubourg, J.-L., Aron, S., Goss, S. and
Foundations. Wiley, Hoboken, NJ. Pasteels, J. M., (1990) “The self-organizing
8. Bullnheimer, B., Hartl, R. F. and Strauß, C., exploratory pattern of the argentine ant”,
(1997) “A new rank based version of the ant Journal of Insect Behavior, Vol. 3, No. 2, pp.
system - a computational study”, Central 159–168.
European Journal for Operations Research 19. Dorigo, M., Birattari, M. and Stutzle, T.,
and Economics, Vol. 7, pp. 25–38. (2006) “Ant colony optimization”, IEEE
9. Burland, J.B., Broms,, B.B. and de Mello, Computational Intelligence Magazine, Vol. 1,
V.F.B. (1977) "Behaviour of foundations No. 4, pp. 28–39.
and structures, State of the Art Review, 20. Dorigo, M., Maniezzo, V. and Colorni,
Proceedings IX ICSMFE, Tokyo, 2: pp. 495- A., (1996) “Ant system: optimization by
546, Rotterdam, Balkema a colony of cooperating agents”, IEEE
10. Burland, J. B., (1990) “On the compressibility Transactions on Systems, Man, and
and shear strength of natural clays”, Cybernetics, Part B: Cybernetics, Vol. 26, No.
Géotechnique, Vol. 40, No. 3, pp. 329–378. 1, pp. 29–41.

11. Camp, C. V. and Bichon, B. J., (2004) 21. Gates, M. and Scarpa, A., (1984) “Optimum
“Design of space trusses using ant colony penetration of friction piles”. Journal of
optimization”, Journal of Structural Construction Engineering and Management,
Engineering, Vol. 130, No. 5, pp. 741–751. Vol. 110, No. 4, pp. 491–510.

12. Camp, C., Pezeshk, S. and Cao, G. (1998) 22. Kakurai, M., (2003) “Study on vertical load
“Optimized design of two-dimensional transfer of piles”, Ph.D. Thesis, Tokyo
structures using a genetic algorithm”, Institute of Technology, (in Japanese).
Journal of Structural Engineering, Vol. 124, 23. Khajehzadeh, M., Taha, M. R., El-Shafie, A.
No. 5, pp. 551–559. and Eslami, M., (2011) “Modified particle
13. Chan, C., Zhang, L. and Ng, J., (2009) swarm optimization for optimum design of
“Optimization of pile groups using hybrid spread footing and retaining wall”, Journal
genetic algorithms”, Journal of Geotechnical of Zhejiang University SCIENCE A, Vol. 12,
and Geoenvironmental Engineering, Vol. No. 6, pp. 415–427.
135, No. 4, pp. 497–505. 24. Kim, K. N., Lee, S.-H., Kim, K.-S., Chung, C.-K.,
14. Cheng, Y. M., Li, L. and Chi, S. C., (2007) Kim, M. M. and Lee, H. S., (2001) “Optimal
“Performance studies on six heuristic pile arrangement for minimizing differential
global optimization methods in the location settlements in piled raft foundations”,
of critical slip surface”, Computers and Computers and Geotechnics, Vol. 28, No. 4,
Geotechnics, Vol. 34, No. 6, pp. 462–484. pp. 235–253.

15. Chow, Y. K. and Thevendran, V., (1987) 25. Maniezzo, V., Gambardella, L. M. and Luigi,
“Optimisation of pile groups”, Computers F. de., (2004) “Ant colony optimization”,
and Geotechnics, Vol. 4, No. 1, pp. 43–58. New Optimization Techniques in
Engineering, Studies in Fuzziness and Soft
16. Coello, C. A., (2002) “Theoretical and
Computing, Springer, Heidelberg, Berlin, pp.
numerical constraint-handling techniques
101–121.
used with evolutionary algorithms: a survey
of the state of the art”, Computer Methods 26. Moeini, R. and Afshar, M. H., (2012) “Layout
in Applied Mechanics and Engineering, Vol. and size optimization of sanitary sewer
191, No. 11–12, pp. 1245–1287. network using intelligent ants”, Advances in
Engineering Software, Vol. 51, pp. 49–62.
17. De Sanctis, L. and Mandolini, A., (2006)
“Bearing capacity of piled rafts on soft
clay soils”, Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 132, No.
12, pp. 1600–1610.

[26] DFI JOURNAL Vol. 7 No. 2 December 2013

27. Momeni, M., Yazdani, H., Fakharian, K., 35. Sales, M. M., Small, J. C. and Poulos, H.
Shafiee, A., Salajegheh, J. and Salajegheh, G., (2010) “Compensated piled rafts in
E., (2012) “Case study of a micropiled raft clayey soils: behaviour, measurements,
foundation design in soft calcareous sandy and predictions”, Canadian Geotechnical
soil, Kerman–Iran.”, The 4th International Journal, Vol. 47, No. 3, pp. 327–345.
Conference on Geotechnical and Geophysical 36. Uğur, A. and Aydin, D., (2009) “An
Site Characterization, Porto de Galinhas, interactive simulation and analysis
Pernambuco, Brazil, pp. 1063-1068. software for solving TSP using ant colony
28. Poulos, H. G. and Davids, A. J., (2005) optimization algorithms”, Advances in
“Foundation design for the Emirates Twin Engineering Software, Vol. 40, No. 5, pp.
Towers, Dubai”, Canadian Geotechnical 341–349.
Journal, Vol. 42, No. 3, 716–730. 37. Valliappan, S., Tandjiria, V. and Khalili,
29. Poulos, H. G. and Davis, E. H., (1980) Pile N., (1999) “Design of raft–pile foundation
Foundation Analysis and Design. John Wiley using combined optimization and finite
& Sons, Australia. element approach”, International Journal
30. Prakoso, W. and Kulhawy, F., (2001) for Numerical and Analytical Methods in
“Contribution to piled raft foundation Geomechanics, Vol. 23, No. 10, pp. 1043–
design”, Journal of Geotechnical and 1065.
Geoenvironmental Engineering, Vol. 127, No. 38. Viana, F. A. C., Kotinda, G. I., Rade, D. A.
1, pp. 17–24. and Steffen Jr., V., (2008) “Tuning dynamic
31. Rajasekaran, S. and Chitra, J. S., (2009) vibration absorbers by using ant colony
“Ant colony optimisation of spatial steel optimization”, Computers & Structures, Vol.
structures under static and earthquake 86, No. 13–14, pp. 1539–1549.
loading”, Civil Engineering and 39. Wang, S.-T., Reese, L. C. and Farmer,
Environmental Systems, Vol. 26, No. 4, pp. G., (2002) “Optimum design of three-
339–354. dimensional pile groups in nonlinear soil.”
32. Rajeev, S. and Krishnamoorthy, C. S., (1992) Deep Foundations 2002, M. W. O’Neill and F.
“Discrete optimization of structures using C. Townsend, Eds., American Society of Civil
genetic algorithms.” Journal of Structural Engineers, pp. 245–261.
Engineering, Vol. 118, No. 5, pp. 1233–1250. 40. Yazdani, H., Momeni, M. and Hatami, K.,
33. Randolph, M. F., (1994) “Design methods for (2013) “Micropiled-raft foundations for
pile groups and piled rafts”, XIII fCSMFE, high-rise buildings on soft soils - a case
Rotterdam: Balkema, New Delhi, pp. 61–82. study: Kerman, Iran”, The 7th International
Conference on Case Histories in Geotechnical
34. Raychowdhury, P. and Hutchinson,
Engineering, Chicago, US, Paper No. 2.40
T. C., (2010) “Sensitivity of shallow
foundation response to model input
parameters.” Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 136, No.
3, pp. 538–541.

DFI JOURNAL Vol. 7 No. 2 December 2013 [27]

 Piezocone Penetration Testing in Florida High Pile
Rebound Soils
Fauzi Jarushi, Ph.D., Department of Civil Engineering, Tripoli University, Tripoli, Libya,
[email protected]
Paul J. Cosentino, Ph.D., P.E., Professor of Civil Engineering, Florida Institute of Technology,
Melbourne, FL, USA, [email protected]
Edward H. Kalajian, Ph.D., P.E., Professor of Civil Engineering and Associate Dean, College of
Engineering, Florida Institute of Technology, Melbourne, FL, USA, [email protected]

ABSTRACT
Contractors and engineers have experienced pile installation problems while driving high displacement
piles with single-acting diesel hammers at Florida Department of Transportation (FDOT) construction
sites located throughout the Central and Panhandle regions of Florida. At certain depths during pile
driving in saturated soils, rebound exceeding 1 inch (25 mm) was experienced, followed by a small
permanent-set during each hammer blow.
High pile rebound (HPR) may cause false refusal to occur, stopping the pile driving and resulting in a
limited pile capacity. In some cases, rebound leads to pile damage, delaying of the construction project
and foundation redesign. In this paper, the response of HPR is investigated using cone penetrometer
testing (CPT) and a pile driving analyzer (PDA). PDA data, which yielded the amount and the depth
where rebound occurred, produced the pile movement per blow, Nineteen Piezocone soundings were
performed at seven FDOT sites in Florida, of which five sites experienced a rebound greater than 0.6
inches (15 mm), one site yielded rebound of 0.35 inches (9 mm), and one site’s rebound was less than
the FDOT limit of 0.25 inches (6 mm).
In order to improve the knowledge about the soil types producing HPR, a traditional geotechnical
investigation on grain-size distribution and soil plasticity allowing for classification using Unified Soil
Classification System (USCS) was conducted. Piezocone data were interpreted using the CPT and CPTu
soil behavior type (SBT) charts proposed by Schmertmann (1978), Robertson (1990) (i.e., Q-Fr, Q-Rf, and
Q-Bq), Eslami and Fellenius (2004), and Schneider et al. (2008). Comparison with classification data from
laboratory tests was in excellent agreement with the CPT soil type, indicating that the CPT is a useful
tool in evaluation of HPR or “large quake” soils. Correlations between rebound and CPTu data were
developed showing that rebound is a direct function of both friction ratio Rf and pore pressure u2.

INTRODUCTION AND OBJECTIVE Pile rebound is defined as the upward elastic

At numerous sites throughout Florida, the pile displacement that occurs during a hammer
Florida Department of Transportation (FDOT) blow. FDOT considers that excessive rebound
contractors and engineers have experienced takes place when it is greater than 0.25 inch
serious pile installation problems while driving (6 mm) per hammer blow (FDOT Road and
high displacement 18, 24 and 30 inch (457, 610 Bridge Construction Specification 455, 2010) or
and 762 mm) square prestressed concrete piles when the blow count exceeds 20 blows per inch
(PCPs) using diesel and air hammers. During (25 mm). HPR is the condition where the set
these installations, excessive rebound per (i.e., plastic soil deformation) represents a small
hammer blow occurred, followed by either small portion of the maximum displacement and the
or no permanent-set. Excessive or high pile rebound (i.e., recovered elastic deformation)
rebound (HPR) may stop the pile driving and constitutes the majority of the displacement.
result in limited pile capacity. In some cases, The design-phase geotechnical investigations
rebound leads to pile damage, delaying the at HPR sites did not produce any unusual
construction project, and foundation redesign soil properties. The research objective was
(Hussein et al., 2006; Cosentino et al., 2010). to develop geotechnical-testing protocol that
would identify HPR.

[28] DFI JOURNAL Vol. 7 No. 2 December 2013

BACKGROUND causing rebound. However, no analytical proof
The background will show a series of case of this conclusion is available.
studies describing HPR followed by studies Likins (1983) analyzed three sites with HPR
about the soil and excess pore pressures between 0.4 and 1 in (10 and 25 mm). He
associated with HPR. The common variable determined that the only common geotechnical
associated with HPR is excess pore pressure; parameter observed at each site was the fully
while the common variable associated with soil saturated soils. Preliminary analysis using the
type is the percent fines, particularly silt. basic wave equation was conducted for each
Murrell et al. (2008) presented a case history of site. The author then modified the results to
HPR, which occurred during the construction of match field data acquired by CAPWAP (Case
a new ferry terminal in coastal North Carolina. Pile Wave Analysis Program). Likins proposed
Twenty inch (508 mm) square, 70 ft (21.3 m) that the only reasonable cause of HPR was the
long PCPs, were designed to support an over buildup of excess pore pressure beneath the
water structure. The authors describe the pile tip. It was also clear through testing, that
high rebound using the term “bounce” which pile capacities decreased when high quake/
was observed at an overburden depth of 53 ft rebound occurred. Findings from the work
(16.2 m) (elevation -43ft or -13 m) as the piles indicated that high quake lowers the pile
penetrated into saturated, firm to stiff, fine- capacity by a factor of 3. Field observations
grained marine soils along the southeastern often led to a false interpretation that the
coast of the United States. hammer is not large enough for the pile. In
these cases where the hammer size is increased,
During the geotechnical investigation at this
the pile could be damaged. Likins (1983)
site, excess pore water pressures greater than
concludes that alternative pile designs, such
20 tsf (1915 kPa) were measured during CPTu
as hollow piles, should be considered as an
testing, at the u2 position behind the cone tip,
effective way to avoid high soil quake.
at the depth where bouncing occurred. When
the blow counts during pile driving were at 303 Jarushi et al., (2013) studied the effect of fines
blows per foot (bpf) (248 blows per 250 mm) , content (material passing # 200 sieve) and
the pile displacement became zero. The driving uncorrected standard penetration test (SPT)
process was then stopped for two hours and blow count (NSPT) on HPR at a group of Central
restrike was conducted. However, an additional Florida sites. The study showed that permanent-
2.5 ft (0.76 m) of pile length was driven with set and rebound were a direct function of NSPT
blow counts of 73 bpf (60 blows per 250 mm), and fines content of the soil at the pile tip. The
112 bpf (92 blows per 250 mm), and finally 87 authors reported that when NSPT was less than
blows per 6 inches (150 mm). Again the driving 15 bpf (12 blows per 250 mm) and the fines
was halted when large rebound resulted in near content was less than 25 percent, the rebound
zero set. In order to achieve the pile capacity was less than 0.25 inches (6 mm), yielding an
and overcome pile rebound, the pile was driven acceptable permanent-set of up to 3 in (76 mm).
after four days using a hammer with a larger For NSPT values between 15 and 40 bpf (12
ram and a shorter stroke. and 33 blows per 250 mm) and a fines content
of 25 to 40 percent, the pile rebound varied
Hussein et al., (2006) presented a case study
between 0.25 and 0.6 in (6 and 15 mm), yielding
related to HPR during driving of PCP’s for the
acceptable permanent-set values. For cases
State Road 528 Bridge over the Indian River in
where the NSPT exceeded 40 bpf (33 blows
Florida. A group of 30-inch (762-mm) square,
per 250 mm) with a fines content greater than
hollow core PCP’s with a length of 115 ft (35 m)
40 percent, pile rebound was greater than
were used to support the bridge. The piles
0.6 in (15 mm) and was accompanied by an
rebounded when they penetrated into hard
unacceptable permanent-set.
soils consisting of saturated medium dense
sand with silt (SP-SM) to fine silty sand (SM) to Baziar et al., (2011) studied pore pressure
clayey sands and sandy clays (SC). The authors generation during the cyclic loading of silty
believe that the incompressible water in the soil sands. The results showed that the magnitude
near and below the pile tip produced excess of the excess pore water pressures during
pore pressure during the driving process which loading of silty sand with 15% to 30% silt was
created an upward force on the pile, thereby similar to the pore pressure behavior observed

DFI JOURNAL Vol. 7 No. 2 December 2013 [29]

 in clean sand. However, as the silt content displacement in inches on the vertical axis and
increased past 30%, the buildup of excess pore time in milliseconds on the horizontal axis,
pressure was much faster. shows a maximum displacement (DMX) of 1 in
Bingjian (2011) studied excess pore pressure (25 mm), an inspector permanent-set (iSet) of
generated during pile driving. The pore pressure 0.11 in (2.8 mm) and a digital set (DFN or dSet)
under the tip of the pile was equivalent to 1.25 of 0.27 in (7 mm). The dSet was not presented
of the effective stress, and the soil disturbance here because it is currently industry practice
was obvious. The excess pore pressure extended to use iSet values. The dSet is recorded in 200
a distance of 5 to 6 pile diameters around milliseconds, while the iSet is based on the total
the pile. time between hammer blows. Jarushi (2013)
used both dSet and iSet data and concluded that
Moayed (2006) reported that as the silt content
the iSet produced more reliable correlations.
exceeded 30%, the time for 50% of the excess
The number of hammer blows per foot is
pore water pressure dissipation increased.
used to produce an average inspector set per
Results from a laboratory study on pore
blow. The maximum displacement and iSet
pressure generation by Erten et al., (1995),
were subtracted to determine the rebound per
showed that pore pressures increased when
hammer blow (i.e., DMX-iSet =Rebound), yielding
the silt content exceeded 30% at a constant
a rebound of 0.89 in (23 mm).
void ratio.
Studies by Jefferies et al., (2006) of California’s CPTu Testing
1918 Calaveras Dam failure showed that as the As shown in Fig. 1, nineteen CPT and CPTu
fines content of the granular soils approached soundings were performed at the seven FDOT
30%, the fines appeared to fill the void space sites. CPTu testing pore water measurements
between the sand particles. The soils then were performed by hydraulically advancing the
acted similar to a fine-grained material with cone penetrometer while signals were digitally
the coarse grains being surrounded by the silt recorded using the Hogentogler® standard
(Baziar et al., 2006). recording system. The CPTu soundings were
conducted using 10-cm2 (1.55 sq in) piezocones;
TESTING PROGRAM with type 2 porous filter elements located at
shoulder (i.e., u2). CPTu testing followed ASTM
Dynamic Testing
D 5778 "Electronic Friction Cone and Piezocone
During FDOT construction at seven sites (see Penetration Testing." During testing, data was
Fig. 1), pile driving was monitored with Pile collected digitally producing the cone resistance
Driving Analyzer (PDA) accelerometer and qc, sleeve friction fs, inclination, and pore water
strain gage equipment. Only site number 5 was pressure every 2 in (51 mm) of penetration.
outside of the Central Florida region. Ramsey The rod insertion speed was 0.75 in/sec
Branch Bridge is located in Florida’s Panhandle. (19 mm/sec). Tests were conducted until refusal
These sites were accessible and allowed the of the CPTu or until the desired depth was met.
research team to further evaluate the soils using
field-testing equipment. HIGH PILE REBOUND SITES
HPR was categorized using two levels: The seven FDOT sites, summarized in both
Unacceptable rebound with minimal or near Fig. 1 and Table 1, are discussed in this paper.
zero set per blow, and acceptable rebound with Six sites were along I-4 near Orlando, and one
sets large enough that continual pile penetration site was located in Florida’s panhandle region.
occurred (Fig. 1). Five of the seven sites had The results from an additional HPR site located
unacceptable HPR (i.e., > 0.25 in (6 mm) with in North Carolina as presented by Murrell et
minimal set) while two of the seven produced al., (2008), were also analyzed and compared to
HPR but with acceptable set (i.e., > 0.25 in data from the FDOT sites.
(6 mm) with enough set to allow driving until A summary of pile driving information obtained
completion). As part of this research cone from the case histories is continued in Table 1.
penetration testing either with (CPTu) or without Although not shown in the table, all HPR piles
(CPT) pore pressure readings was completed. were set into predrilled holes about 25 ft
Figure 2 shows typical HPR PDA data from (7.6 m) deep. In all but one instance the piles
one pile blow at a FDOT site. The plot, with were square PCP’s (SPCP’s), however, due to

[30] DFI JOURNAL Vol. 7 No. 2 December 2013

 FDOT Site Rebound Number of
Type
Descriptions Level Soundings
CPTu 3
Anderson Street
1 Unacceptable
Overpass
CPT 2
2 SR 50 over SR 436 Unacceptable CPTu 2
3 I-4/US192 Unacceptable CPTu 3
4 I-4 / Osceola Parkway Unacceptable CPTu 1
5 Ramsey Branch Bridge Unacceptable CPTu 2
I-4/ SR 408
6 Acceptable CPTu 3
(Ramp B)
SR 417 /International
7 Acceptable CPTu 3
Parkway

[FIG. 1] HPR Site Description, Location according to FDOT District, Rebound Level, and CPT tests conducted

FIG. 2] Typical PDA Pile Top Displacement versus Time Diagram from One Hammer Blow for FDOT HPR Site

construction problems at FDOT’s Anderson • The majority of pile driving hammers were
Street Overpass H-Piles were used. This table single-acting diesel hammers.
includes site description, pile description, pile • Rebound occurred at depths from 50 to
spacing, hammer characteristics, driving blow 80 ft (15 to 24 m).
counts, rebound and elevations.
• Average pile driving blow counts in the
The common characteristics among the HPR rebound layers were greater than 105 bpf
sites are as follows: (86 blows per 250 mm).
• Piles were concrete displacement piles Data analysis of the hammer type, pile size, and
ranging from 18 to 24 in (457 to 610 mm) spacings did not produce any clear correlations
in diameter. to HPR Jarushi (2013).
• Both the PDA test piles and production piles
were longer than 70 ft (21.3 m). OVERVIEW OF CPT SOIL PROFILING
• Pile spacings were 6 to 11 ft (1.8 to 3.4 m) or LITERATURE
2.5 to 5.5 diameters. Begemann (1965) developed the first CPT soil-
• Piles were set into predrilled holes. profiling chart based on qc and fs. Soils with

DFI JOURNAL Vol. 7 No. 2 December 2013 [31]

 [TABLE 1] Pile driving information summary at selected FDOT sites

Pile Pile a
Hammer b
Average Rebound MAX
Site Pile size
Length Spacing Model BL Elevation Rebound
Description and type
(ft) (ft) Type (blows/ft) (ft) (in)
24-in Delmag
124 7 135 15 to -10 1.4
SPCP D62-22
Anderson Street
1
Overpass
HP
120 NA ICE I-30 NA No Rebound
(14 x 89)

24-in APE
2 SR 50/SR 436 overpass 105 8 143 26 to 17 1.1
SPCP D62-42

24-in ICE
Pier 6 106 7 220 35 to 25 0.6
SPCP 120 S

I-4/ 24-in ICE

3 Pier 7 112 9 140 35 to 20 0.6
US.192 SPCP 120 S

24-in ICE
Pier 8 100 6 111 30 to 15 1.25
SPCP 120 S

24-in ICE
4 I-4/ Osceola Parkway 95 6 105 15 to 8 0.9
SPCP 120 S

18-in Vulcan
5 Ramsey Branch Bridge 100 10 & 20 110 -28 to -70 1.2
SPCP 512 Air

I-4/ SR 408 18-in Delmag

6 100 NA 50 30 to 0 0.5
(Ramp B) SPCP D36-32

SR 417 /International 24-in APE

7 100 6 42 5 to 0 0.25
Parkway SPCP D46-42

SPCP=square prestressed concrete pile; asingle acting; baverage driving blow counts at HPR layer; NA=
not available

higher qc and fs were considered coarse-grained Where:

while soils with low qc and fs are fine-grained. Bq = pore pressure ratio
Begemann (1965) showed that the soil types are
Fr = normalized friction ratio
a function of the friction ratio (Rf).
Rf = friction ratio, percent
Robertson et al., (1986) developed the first Soil
Behavior Type (SBT) chart taking into account Qtn = normalized cone resistanceQ —
the effect of pore pressure u2, and presented u2 = pore pressure at shoulder
a refinement to a normalized chart, which uo = hydrostatic pore pressure
considered overburden stress. Robertson (1990)
developed SBT’s with cone tip resistance as Q qc = cone resistance
in terms of Q vs Rf, Qtn vs Fr and Qtnvs Bq. The σv = total overburden stress
Bq ratio, friction ratio, Rf, normalized friction σv ́ = effective stress
ratio Fr, and normalized cone resistance are be
fs = sleeve friction
defined as follows:
pa = atmospheric pressure
௨మ ି௨೚
‫ܤ‬௤ ൌ , n = stress exponent
௤೎ ିఙೡ
ܴ௙ ൌ ݂‫ݏ‬Ȁqc) 100%, Eslami and Fellenius (2004) presented a CPTu
௙ೞ approach for soil classification. This approach,
‫ܨ‬௥ ൌ ቂ ቃ ͳͲͲΨ,
௤೎ ିఙೡ where the soil type depends on u2, is based on
௤೎ష഑ೡ ௣௔
and ܳ௧௡ ൌ ቂ ቃ ሺ ƴ ሻ௡ the effective cone resistance (qE) where: qE=qc-u2.
௣௔ ఙ௩
In soils where u2 is very high, qE can be smaller

[32] DFI JOURNAL Vol. 7 No. 2 December 2013

than qc. The authors profile chart contains five representing over 50% of the soil. The fines
main soil type: (1) Sensitive and Collapsible Clay content at these sites increased at HPR layers to
and/or Silt, (2) Clay and/or Silt, (3) Silty Clay over 30%.
and/or Clayey Silt, (4) Sandy Silt and/or Silty Using the results from SPT borings, PDA, and
Sand, (5) Sand and/or Sandy Gravel. the CPT testing, plots, shown in Figs. 3 through
11, for each of the case studies were developed
RESULTS AND DISCUSSIONS relating soil descriptions to elevations of
iSet, rebound and CPT output. The elevations
Index Properties of HPR Soils
associated with the start of PDA collection,
Disturbed samples obtained during Standard corresponded to the depth at which pile driving
Penetration Testing were used to determine the commenced. This elevation was below the
USCS symbol using the procedure outlined in ground surface, since the piles were set into
ASTM D-2488. The soils at the HPR sites were predrilled holes.
mainly sand with varying percentages of silts
and clays. Table 2 is a summary of the soil GEOTECHNICAL INVESTIGATION
classification, Atterberg limits, natural moisture RESULTS
content and average fines content of the soils.
Most HPR layers had high fines content with The soils at site 1 (Anderson Street Overpass),
the natural moisture content less than the consisting of an 18 ft (5.5 m) thick HPR zone
liquid limit. The soils plotted near the A-line (Fig. 3a), included a greenish-gray silty clayey
on the Cassagrande plasticity chart. These soils fine-sand (SM-SC), and dark greenish-gray clayey
displayed an olive green to light green color fine-sand (SC). The bottom soil layer in this HPR
with visual descriptions ranging from clayey zone was greenish-gray clay (CH) with a trace of
and silty fine sands, to highly plastic clays. phosphate and shell.
Direct correlations between Atterberg limits, The soils at site 2 (SR50/SR436 Overpass),
natural moisture and rebound content were shown in Fig. 4a, consisted of mainly medium
developed and examined from the HPR sites. dense sands with varying percentages of
Consequently, the data analyses of the water silts and clays to elevation 30 ft (9 m); while
content did not provide any clear findings, classifying as SP, SP-SM, and SM. The soils
which could be used to predict HPR soils. At within the HPR zone from elevations 28 to 17 ft
all sites, the water table was located near the (8.5 to 5.2 m), generally were dense to very
surface and well above the HPR elevation. Sand dense or stiff green silty fine-sand (SM) to green
was the predominate material, consistently clay (CH) with a trace of phosphate.

[TABLE 2] Summary of Soil Properties at Elevations where HPR soils were Encountered

Rebound Maximum
Site Site Name Elevationa Rebound USCS FC (%) wn (%) LL (%) PI (%)
(ft) (in)
Anderson Street SM-SC, SC 40 13
1 15 to -10 1.4 >40 30-50
Overpass CL & CH 86 42
2 SR50/SR436 overpass 26 to 17 1.1 CH >40 63 155 110
Pier 6 35-25 0.6 SM >30 31 NP NP
3 I-4/ US.192 Pier 7 35-20 0.6 SM 30 NA NA NA
Pier 8 35-15 1.25 SM >40 47 NP NP
4 I-4/Osceola Parkway 15-8 0.9 SM 25 NA NA NA
5 Ramsey Branch Bridge -28 to -70 1.2 SP-SC& SC >40 38 45 25
6 I-4/ SR 408 (Ramp B) 30-0 0.5 SC 20 23 NA NA
SR417/International SP-SM & 31 NP NP
7 5 to 0 0.25 20
Parkway SM 41 48 17

NOTE: aNAVD elevations; USCS=Unified Soil Classification System; FC= fine content; wn= natural
moisture content; LL= Liquid Limit; PI =Plasticity index; NA=not available; NP=non plastic.

DFI JOURNAL Vol. 7 No. 2 December 2013 [33]

 (a) (b) (c) (d) (e)
[FIG. 3] (a) USCS Soil Profile, (b) PDA output (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 1:
Anderson Street Overpass

Inspector set
s and qc(tsf) Rf (%) U2 (tsf)
Elev. GSE 99ft USCS Type Rebound d (in)
(fft) 0 1 2 3 4 0 300 600 0 2 4 6 0 10 20
2 30 40
100 Silty Fine Sand (SM)
Fin
ne Sand (SP)
90 Silty Fine Sand (SM)
Fin
ne Sand (SP)
SP
P-SM) Predrille
led
80 Silty Fine Sand (SM)
Depth
th

70
Fin
ne Sand (SP)
60
Silty Fine Sand (SM)
Fin
ne Sand (SP)
50 Silty Fine Sand (SM)

40 Claay (CH)
Silty Fine Sand (SM)
Fin
ne Sand (SP)
30 Silty Fine Sand (SM)

20 Sandy Fat Clay (CH)

10 CPT-1 (U2)
EB 4 WB B Pile 5 Rebound
EB 4 WB B Pile 5 iSet CPT-5b (U2)
0.25 inch
h Rebound

(a)) (b) (c) (d) (e)

[FIG. 4] (a) USCS Soil Type, (b) PDA output, (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 2: SR50/
SR436

[34] DFI JOURNAL Vol. 7 No. 2 December 2013

The three soil profiles at I-4/US.192 (site 3), The soils encountered at site 4 (I-4/ Osceola
shown in Figs. 5a, 6a, and 7a, included loose to Parkway), shown in Fig. 8a, consisted of loose
medium dense brown fine-sand (SP), fine-sand to medium dense fine sand (SP) and silty sand
with silt (SP-SM), gray silty fine-sand (SM) with (SM) with a lense of gray silt with sand (ML) to
abundant shell and a trace of phosphate, plus elevation 40 ft. The rebound layer elevation
gray clayey fine-sand (SC). HPR occurred when from 15 to 0 ft (4.6 to 0 m) is medium dense
the piles encountered very dense silty sand (SM) gray silty fine sand (SM), some shell, and trace
between elevations 25 and 0 ft (7.6 and 0 m). phosphate, which is underlain by tan sandy
weathered limestone, with a trace phosphate to
elevation -30 ft (-9.1 m).

Inspector set
s and
Eleev. GSE 106ftt USCS Type Rebound d (in)
qc (tsf) Rf (%) U2 (tsf)
(ftt) 0 Inspector
0.25 0.5 et0.75
0 and 1 0 300 600 0 2 4 6 0 10 20 30
0
El v. GSE 106f USCS Type Reboun (in)
q (tsf) Rf (%) U2 (tsf)
Sand with
w silt to Silty 0 0.25 0.5 .75 1
(f100
1) Sand (SM)
0 300 600 0 2 4 6 0 10 20 3
Predrillled
90
00
Sand ith silt to Silty Depthh
Sand (SM)
Predril ed
80
90 Dept
70
80 Sand with
w Silt
(SP-SMM)
60
70 Sand ith Silt
(SP-S )
50
60 Fine Sand
S (SP)

50
40 Fine and (SP)

40
30 Silty Sand (SM)

30 Silty Sand (SM)

20
20
10 Pier 6 Pile 16 Rebound
Pier 6 Pile 16 iSet CPTu-4
10 Pier 6 Pile
0.25 inch16Rebound
RRebound
Pier 6 Pile 16 iSet
(a) (b)
0.25 inch ebound (c) (d) (e)
CPTu-4

[FIG. 5] (a) USCS Soil Type, (b) PDA output, (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 3: I-4/
US192 Pier 6

Inspectorr set and qc (tsf) Rf (%) U2(tsf)

Eleev. GSE 106ftt USCS Type Reboun nd (in)
0 0.25 0.50 0.75 0 200 400 0 2 4 6 0 20 40
(ftt)
(SP-SMM)
1
100 Silty Sand
S (SM) Preddrilled
90 (SP-SMM) Deepth
Silty Sand
S (SM)
80 Sand with
w Silt
70 (SP-SM M)
60 Fine Sand
S (SP)

50 (SP-SM
M)

40
30 Silty Sand
S (SM)
20
10
0
Pier 7 Pilee 10 Rebound
Pier 7 Pilee 10 iSet CPTu-3
0.25 inch Rebound

(a) (b) (c) (d) (e)

[FIG
G 6] ( ) USCS S il T (b) PDA t t( )C R i t (d) F i ti R ti d( )P P f Sit 3 I 4/US192 Pi
S 7
[FIG. 6] (a) USCS Soil Type, (b) PDA output (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 3: I-4/
US192 Pier 7

DFI JOURNAL Vol. 7 No. 2 December 2013 [35]

 Inspector set and qc (tsf) Rf (%) U2(tsf)
Eleev. GSE 90ft USSCS Type Rebouund(in)
Inspecto r set and U2(tsf)
q200
c (tsf) 400 Rf (%)
(ft))v. GSE 90ft U CS Type
El 0 0.25 0.5 0.75 nd(in)
Rebo 1 1.25 0 0 2 4 6 0 40
0 10 20 30
(ft 90 0 0.25 0.5 0.75 1 1.25 0 200 400 0 2 4 0 10 20 3 40
Fin
ne Sand (SP)
90
80
80 Sand with
70 Silt
Sat(SP-SM)
nd with
70
60 Sil (SP-SM)
Fine Sand (SP)
60 (SP-S
50 FineSM)
Sand (SP)
50 (SP- M)
40
40
30
30 Siltyy Sand (SM)
20
Silt Sand (SM)
20
10
10
0
-100 Sandyy Limestone
Pier 8 Pilee 11 Rebound
Sand Limestone CPTu-2
-10 Pier 8 Pilee 11 iSet
Pier 8 Pil 11 Rebound
0.25 inch Rebound
Pier 8 Pil 11 iSet CPTu-2
(a) (b)
0.25 inch Rebound
(c) (d) (e)
[FIG
G 7]
[FIG. 7] ((a)) USCS S ilType,
USCS Soil T (b)(b)
PDAPD
DA t (c)t Cone
output ( ) C Resistance, (d) F Ratio
R i t (d) Friction R ti
i ti and d (Pressure
(e) Pore )P P Sit 3
for Site 3: fI-4/S
US192 Pier 8

Inspector et and qc (tsf) Rf (%) U2 (tsf)

El v. GSE 92ft USCS Type Reboun1 (in) 1.5
0 0.5
Inspector sset and 0 300
qc (tsf)600 2R (%)
4 6 0 10U (tsf)
20 30
(ft f 2
Eleev. GSE 92ft USCS Type Rebound d (in)
0 (SP-S ) 0 0.5 1 1.5 0 300 600 0 2 4 6 0 10 20 30
(ft))
9
900 Silty and (SM)
(SP-SM
M)

80
80 Silty Sand
S (SM)
Fine and (SP)
70
70 Sand ith Silt
Fine SSand
(SP-S ) (SP)
60
60 Sand w
with Silt(ML)
Silt w th sand
0 Silty
(SP-SMS M)
nd (SM)
5
50
Claye Sand (SC)
Silt wiith sand (ML)
40
40 Silty Saand (SM)
Clayeyy Sand (SC)
20
3
30
Silty and (SM)
10
2
20
Silty Sand
S (SM)
0
10
Tan Sandy Limestone
-100
Pier 2 Pile Rebound
Tan Sandy Limestone
-10 Pier 2 Pile iSet CPTu 1
Pierinch
0.25 8 Rebound
2 Pileebound
Pier 2 Pile 8 iSet CPTu--1
0.25 inch Rebound
R
(a) b)
(b (c) (d) (e)
[FIG
G 8]
[FIG. 8] ((a)) USCS
USCSSoil
S Type,
il T (b)(b)
PDAPD
DA t (c)t Cone
output, ( ) CResistance, (d) FRatio
R i t(d) Friction ti PoredPressure
i ti andR(e) ( ) P for f Sit
P Site 4: I-4/ S
Ocela Parkway

[36] DFI JOURNAL Vol. 7 No. 2 December 2013

Fig. 9a shows the soil profile for site 5 (Ramsey underlain by very soft sandy clay and clayey
Branch Bridge). In the HPR zone below elevation fine-sand (SC) to elevation 45 ft (13.7 ft). This
-28 ft (-8.5 m), the soils include greenish- layer was underlain by the HPR soil, a medium
gray clayey sand (SC) and greenish-gray sand dense clayey fine-sand and silty clayey fine-sand
with clay (SP-SC). A trace of shell was found (SC) to elevation 9 ft (2.7 m).
throughout soils in the HPR zone. The soils at site 7 (I-4/SR417), shown in Fig. 11a,
The soil profile at site 6, along I-4/SR408 varied between silty fine-sand (SM) and clayey
(Ramp B), is shown in Fig. 10a and consisted fine-sand (SC), and included thin layers of silt
generally of three layers. The first was an upper (MH) and sandy fat clay (CH). Minor rebound less
stratum of loose to medium dense sand with than the FDOT limits of 0.25 in (6 mm) occurred
silt and clay (SP-SM & SP-SC) to silty sand (SM) between elevation 20 and -2 ft (6 m and -0.6 m),
to elevation 60 ft (18.3 ft). This stratum was where the soils were silty fine-sand (SM).

Inspector set and qc(tsf) Rf (%) U2 (ttsf)

bound (in)
Reb
Elev. GSE 7 ft USCS Type 0 0.5 1 1.5 0 150 00
30 0 2 4 6 8 0 10 20
0 30 40
(ft)
Sand with Silt
0 (SP-SM) nspector no
In
Reebound Soils
-10
Sand with Silt
-20 to Silty Sand (SP-SM)
Silty Sand (SM)
-30 (SP-SM)
SSand with Clay
(SP-SC)
-40
-50
-60 Clayey Sand (SC)

-70 Cemented Sand

C
with Limestone(SP)
-80
B 3 Pile 11 Rebound
EB
CPTu1
EB
B 3 Pile 11 iSet
CPTu2
25 inch Rebound
0.2

(a) (b) (c) (d) (e)

[FIG. 9] (a) USCS Soil Type, (b) PDA output, (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 5: Ramsey
Branch bridge

Inspector set and qc(tsf) Rf (%) 2 (tsf)

Elev. GSE 1 5ft USCS Type Reb und (in)
((ft) 0 0.5 1 1.5 0 150 30 0 2 6 0 5 10 15 20

100 ilty Fin Sand (SM) Pred illed

90 De t
F ne Sand(SP)
80
( P-SC)
70 ( P-SM)
60 S nd With Clay SP- C)
DFI JOURNAL Vol. 7 No. 2 December 2013 [37]
50 C ayey Fine Sand (S )
 Inspector set and qc(tsf) Rf (%) U2 (tsf)
Elev. GSE 10
05ft USCS Type Reboound (in)
((ft) 0 0.5 1 1.5 0 150 0
300 0 2 4 6 0 5 10 15 20

100 S Fin Sand (SM)

Silty Predrrilled
90 Deppth
Fiine Sand(SP)
80
(SSP-SC)
70 (SSP-SM)
60 Saand With Clay SP-SSC)

50 C)
Cllayey Fine Sand (SC
40 Cllay (CH)

30
C
Clayey Fine Sand (SSC)
20
10
0
EB 1 Pile 2 Rebound CPTu-B109
EB 1 Pile 2 iSet CPTu-B118
25 inch Rebound
0.2 CPTu -A1-105A

(
(a) (b)
( (c) (d) (e))
[FIG. 10] (a) USCS Soil Type, (b) PDA output, (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 6: I-4/
SR408 Ramp B

Inspecttor set and qc (tsf) Rf (%)) U2 (tsf)

Elev. GSE 75ftt USCS Type Reboound (in)
(ft) 0 0.5 1 1.5 2 0 200 4
400 0 2 4 6 0 2 4 6

Silty Sand (SM)

70 Fine Sand (SP) Preddrilled
(SSP- SM)
60 Deepth
Saandy Fat Clay (CH)
(SSP- SM)
50
Elastic silt (MH)
(SSP- SM)
40

30 Silty Sand (SM)

20
(SSC)
10

0 S
Silty Fine Sand (SM
M)

-10
EB 2 Pile 5 Rebound CPTu1
CPTu2
EB 2 Pile 5 iSet
CPTu3
0.25
5 inch Rebound
(
(a) (b) (c) (d) (e)
[FIG. 11] (a) USCS Soil Type, (b) PDA output, (c) Cone Resistance, (d) Friction Ratio and (e) Pore Pressure for Site 7: I-4/
SR417

[38] DFI JOURNAL Vol. 7 No. 2 December 2013

PDA AND CPT PROFILES SOIL BEHAVIOR TYPE (SBT) CHART
Typical piezocone profiles from the seven sites, CORRELATIONS
which include cone resistance qc, friction ratio Two CPT and four CPTu SBT charts were used
Rf, and pore pressure u2, are shown in Figs. 3 to study HPR and non-HPR soil characteristics.
to 11. The rebound, shown as a red line and
(1) Schmertmanns’ (1978) CPT chart.
iSet shown as a green line, were plotted versus
elevation. A vertical dashed line at 0.25-in (2) Eslami and Fellenius (2004) qE-fs CPTu Chart.
(6-mm) was included to differentiate between (3) Robertson (1990) Q-Rf CPT Chart.
high and acceptable rebound. (4) Robertson (1990), Q-Fr, and Q-Bq CPTu Chart.
At site one (Anderson Street Overpass) HPR (5) Schneider et al. (2008) Q-∆u2 ⁄ σvo CPTu Chart.
problems that occurred during installation of
In general, these SBT charts show coarse-grained
24-inch (610 mm) PCP, caused construction
soils (sand and gravel) to have a higher qc and
delays and required the redesign of
a lower Rf, while fine-grained soils produce a
the foundations and replaced with low
lower qc and a higher Rf (Lunne et al. 1997).
displacement steel H-piles (HP 14 x 89 or
HP 360 x 132). As shown in Fig. 3b, the PDA Fig. 12a illustrates the CPT data plotted on
rebound ranged between 0.25 and 1 in (6 and the Schmertmann (1978) chart, which relates
25 mm) beginning at elevation 15 ft (4.6 m) to cone tip to cone friction. The soils change
the end of driving at elevation -7 ft (-2.1 m). from cohesive at low qc and high fs values (i.e.
In this HPR lower zone, the CPT pore pressure lower right portion) to sands with higher qc and
increased to 30 tsf (2873 kPa) and friction ratio lower fs values (i.e., upper left portion). The
increased to 5 percent. HPR soils from this study are shown in red and
plot mostly in the dense sand region and also
Consistent behavior was observed for soils that
the stiff or very stiff zones near the cohesive-
produced HPR (Figs. 3 to 9); pore water pressure
sand boundary. The non-HPR soils from this
u2 is typically greater than 20 tsf (1915 kPa) and
study are shown in blue and generally plot in
Rf is greater than 2%. At several locations shown
the loose sand regions, with some data also in
in Figs. (5e, 6e, 7e, and 8e), u2 decreased in the
the soft to medium cohesive zones. Data from
layers below where rebound occurred. The pore
Murrell et al., (2008) were added to this chart
water pressure at the tip increased throughout
and generally plotted near the blue-red border
the driving process and as the pile tip passed
of loose to dense sands. NSPT in the HPR soils
through the intact soil and advances down,
were greater than 35 bpf (28 blows per 250 mm)
excess pore water pressure still building up
(i.e., dense to very dense or hard soils), while
due to large soil deformation, allowing rebound
those in non-HPR soils were less than 30 bpf
for continue.
(25 blows per 250 mm) (loose to medium dense
Lower qc observed at these soils is typically soils). Just below fs values of 1 tsf (96 kPa) there
indicative that these soils are more fine-grained is a clear delineation between HPR and non-HPR
soils than coarse-grained. This is in good soils in Schmertmanns’ (1978) CPT chart.
agreement with laboratory results that HPR
CPTu data from HPR and non-HPR soils were
soils had fines content of greater than 30%.
also plotted on the Eslami and Fellenius (2004)
For non-HPR soils at sites where minor rebound chart. As shown in Fig. 12b, HPR soils plotted
was observed followed by an acceptable set in silty sandy to silty clay zones with fs values
(Figs. 10 and 11), the pore pressure u2 exhibited greater than 1 tsf (96 kPa), while Non-HPR
values of less than 17 tsf (1628 kPa) and Rf soils plotted in the sandy silt and sensitive-
of less than 2%. The non-HPR soils behaved collapsible clay-silt or clay zone with a fs values
more like coarse soils than fine soils, as these less than 1 tsf (96 kPa). This trend generally
soils exhibited higher qc and lower u2. Sand matches the trend from Schmertmanns’ (1978)
is representing over 70% of the non-HPR chart.
soils, which confirmed that pore pressure in
The HPR and non-HPR data was plotted on
soils with low fines content may dissipate
Robertson’s (1990) Q-Rf charts (Fig. 13). Fig. 13a
very quickly.
shows the HPR soils as sand mixtures in zone
5 (Silt mixtures: silty sand to sandy silt). The
majority of these soils have Rf values greater

DFI JOURNAL Vol. 7 No. 2 December 2013 [39]

 than 1.5%. Although the trends are not as
clear as those shown in Fig. 12, the non-HPR
soils, shown in Fig. 13b, plot from zones 4 (silt
mixtures clayey silt & silty clay) to 6 (sands:
clean sands to silty sand) and a large percentage
have Rf values less than 1%. The data from the
North Carolina site presented by Murrell et al.,
(2008) is shown in Fig. 13a and generally plots
in Zone 6.
HPR soils were plotted as sand mixture zone
4 (clayey silt to silty clay) on the normalized
Robertson (1990) Q-Fr chart (Fig. 14a), where
atmospheric pressure was used to normalize
each axis. These plots (shown on the left side) (a)
are similar to those shown in Fig. 13. In this
case most HPR soils plotted with Fr values
greater than 1.0%, while most non-HPR soils
plot below 1% between zones 3 and 6 or clays:
clay to silty clay to sands: clean sand to silty
sand. On the Q-Bq charts HPR soils (Fig. 14a),
produced pore pressures of greater than 20 tsf
(1915 kPa), which led to a Bq ratio of 0.4, while
non-HPR soils (Fig. 14b) had a Bq ratio of near
zero when u2 was small or approached the
hydrostatic pressure.
HPR soils exhibited excess pore pressures (i.e.,
u2-u0) greater than 20 tsf (1915 kPa). The CPTu
data from the HPR sites was plotted on the two
SBT charts proposed by Schneider et al., (2008).
The excess pore pressures were denoted as ∆u2,
(b)
and were normalized by dividing them by the
effective overburden stress (σv). As shown in [FIG. 12] Location of HPR and Non-HPR soils on (a)
Schmertmann (1978) Chart, (b) Eslami and Fellenius
both parts of Fig. 15a, HPR soils behave as clays (2004) Chart
or silts (Zones 1a or 1b). These soils produced
a ∆u2 ⁄ σv ratio greater than 1 and a ratio of

!"
"%'
% 
"
+*
.422:/

3222

3222


!!"*,


!!"*,

322

322

32
32

3
3

2+3
3
32
2+3
3
32


 #
#*

.>/

 #
#*


.>/





(a) (b)
[FIG. 13] Location of (a) HPR soils and (b) Non-HPR soils on Q-Rf classification approaches proposed by Roberston (1990)

[40] DFI JOURNAL Vol. 7 No. 2 December 2013

 (a)

(b)

[FIG. 14] Location of (a) HPR soils and (b) Non-HPR soils on Q-Fr and Q-Bq classification approaches proposed by Roberston
(1990)

net cone point resistance to effective stress different zones than non-HPR soils. The charts
(qcnet ⁄ σv) greater than 10. Non-HPR soils, shown proposed by Schmertmann (1978) Eslami and
in Fig. 15b, plotted well to the left of the HPR Fellenius (2004) and Schneider et al., (2008)
soils in Zone 3 which is the transitional zone produced the clearest delineations between
between drained and undrained behavior. The the two. Both the CPT friction and CPTu excess
normalized pore pressures are consistently pore pressure highlighted these differences.
below 1. HPR soils classified as dense according to
In summary, the proposed CPT SBT charts from Schmertmann (1978), silty sand to silty clay
all four sources show that the HPR soils plot in according to Eslami and Fellenius (2004), sandy

DFI JOURNAL Vol. 7 No. 2 December 2013 [41]


!"
"%'
% 
"
+*
.422:/

3222
3222

322
322






 
32
32

3
3

-4
-3
2
3
4
5
6
7
8
9
:
;
32
2+3
3
32
322


 



(a)

(b)

[FIG. 15] (a) HPR soils and (b) Non-HPR soils on the Schneider et al. (2008) chart

silt to silty sand acording to Robertson (1990), SC, SM-SC. There was close agreement between
and silt or clay according to Schneider et al., the USCS method and the CPT soils type.
(2008). Non-HPR soils behave more like sandy
soils than fine-grained soils. CORRELATING REBOUND TO Rf, qc,
Based on SPT and laboratory tests, HPR soils AND u2
were classified as one of the following USCS Correlations between the pile rebound, Rf, qc,
groups: dense to very dense or hard: SM, SC, and u2 were developed and examined. These
SM-SC, and CH. Conversely, non-HPR soils were correlations, shown in Fig. 16a and 16b, were
classified as: loose to medium dense SP, SP-SM, developed based on averages obtained within

[42] DFI JOURNAL Vol. 7 No. 2 December 2013

 1.6
Excludes Murrell, R² = 0.80 Excludes Murrell, R² = 0.80
1.2

Rebound (in) 0.8

0.4

0
0 1 2 3 4 5 0 5 10 15 20 25 30 35
Friction Ratio, Rf (%) Pore Pressure, U2 (tsf)

(a) (b)
[FIG. 16] Correlation between Rebound and (a) Rf and (b) u2

HPR zones (rebound > 0.25 in or 6 mm) and R = 0.25Rf + 0.005u2 + 0.05 [1]
non-HPR zones (rebound < 0.25 inches or Where: R= rebound (inches);
6 mm). The rebound versus friction ratio (Rf) Rf= Friction Ratio (%);
(Fig. 16a) produced an increasing linear trend
u2= CPTu pore pressure at
with a regression coefficient R2 of 0.80 while
shoulder (tsf).
an Rf of 1% was associated with 0.25 in (6 mm)
of rebound. The rebound versus qc did not The applicability of Eqn. 1 was evaluated by
produce desirable correlations (i.e., R2=0.1) and plotting its predicted rebound versus actual
are not shown (Jarushi, 2013). Rebound versus rebound, using the data from the seven sites
pore pressure u2 (Fig. 16b) produced a linearly used in this study plus the one site presented
increasing trend. Soils with u2 of 5 tsf (479 kPa) by Murrell et al., (2008). As shown in Fig. 17, the
or less produced an acceptable rebound of equation produced R2 values of 0.80, showing
less than 0.25 in (6 mm). As u2 exceeded 20 tsf an ability to predict rebound using Rf and u2.
(1915 kPa), the rebound increased to over Two of the three data points from the Murrell
0.60 inches. The CPTu data Rf data from Murrell et al., (2008) study do not agree with data from
et al. (2008) did not agree with the results this study.produced R2 values of 0.80, showing
from this study. The pore pressure data from an ability to predict rebound using Rf and u2.
Murrell et al., (2008) did plot near the data from Two of the three data points from the Murrell
this study. et al., (2008) study do not agree with data from
this study.
Statistical analysis using SPSS (Statistical
Package for the Social Sciences) software
was carried out on the CPT data from the
1.5
seven sites. An analysis of variance (ANOVA)
was included to determine if there was any
Predicted Rebound (in)

significant relationship between rebound and

Rf, qc, and u2. The ANOVA result produced 1
R² = 0.75
0
very strong correlations from both Rf and u2.
A design equation to predict pile rebound was
developed. It included both Rf and u2, but as 0.5
did the previous correlations excluded qc due to
poor correlations. These results were further
validated via examination of the experimental 0
residuals, which were normally distributed and
0 0.5 1 1.5
showed no patterns that would cause concern.
Actual Rebound (in)
Equation 1 is presented to relate HPR rebound
to CPTu parameters Rf and u2: [FIG. 17] Predicted Rebound using Rf and u2 versus Actual
Rebound from Equation 1

DFI JOURNAL Vol. 7 No. 2 December 2013 [43]

 SUMMARY be checked with the predicted rebound from
Large displacement piles at numerous Florida Equation 1 to help validate its accuracy.
locations experienced HPR during driving
into saturated soils. The overburden depth at REFERENCES
which HPR occurred was typically greater than 1. ASTM D2488, Standard Practice for
50 ft (15 m). There was a large increase in the Description and Identification of Soils.
CPTu pore pressure u2 from negative or near 2. ASTM D5778, Standard Test Method for
zero to high positive values whenever HPR Performing Electronic Friction Cone and
was identified. Piezocone Penetration Testing of Soils.
3. Baziar, Mohammad H., and Reza Z. Moayed.
CONCLUSIONS
(2006), “Evaluation of cone penetration
Based on SPT blow counts, HPR soils were very resistance in loose silty sand using
dense or hard while non-HPR soils were loose calibration chamber”, International Journal
to medium dense. HPR soils classified as one of of Civil Engineering 4.2, pp106-119.
the following groups: SC, SM-SC, SM, CL, SP-SM,
4. Baziar, Mohammad H., Habib Shahnazari,
and CH; while non-HPR soils as SP, SP-SM, SC,
and Hassan Sharafi. (2011), “A laboratory
and SM-SC.
study on the pore pressure generation
An equation for predicting rebound, based on Rf model for firouzkooh silty sands using
and u2, was developed, and the results showed hollow torsional test”, International Journal
the ability to predict pile rebound during design of Civil Engineering vol 9(2), pp126-34.
if large displacement piles are to be driven.
5. Begemann, H. K. S. (1965), “The friction
Three of the four SBT charts evaluated (i.e., jacket cone as an aid in determining the soil
Schmertmann’(1990), Eslami and Fellenus profile”, Proceedings 6th ICSMFE 1, pp17-20.
(2004) and Schneider et al., (2008)) would enable
6. Bingjian, Zhu. (2011), “Study of the pore
geotechnical engineers to predict the behavior of
water pressure variation rule in saturated
HPR soils based on clear limits from either CPT
soft soil caused by prestressed concrete pile
friction or CPTu pore pressures. Engineers may
penetration”, Electric Technology and Civil
encounter HPR problems if soils plot on the:
Engineering Conference, 22-24 April 2011,
• Schmertmann (1978) SBT chart as dense or Taizhou, China, pp.756-59.
stiff soils silts with fs values greater than
7. Cosentino, P. Kalajian, E. Misilo, T, Chin
1 tsf (96 kPa)
Fong, Y. Davis, K., Jarushi F., Bleakley A.,
• Eslami and Fellenius (2004) SBT chart as Hussein M. H., and Bates, Z. (2010), “Design
silty sand or silty clay to clay silts with fs phase identification of high pile rebound
values greater than 1 tsf (96 kPa) soils”, Technical report, Contract BDK81
• Schneider et al., (2008) SBT chart as 1a, Work Order 977-01, Florida Department of
and 1b, (silt and clays) with ∆u2 ⁄ σvo values Transportation.
greater than 1 8. Erten, D. and Maher, M. H. (1995), “Cyclic
• Robertson (1990) (Q-Rf) SBT chart as silty undrained behavior of silty sand”, Journal of
sand to sandy silt where qc/pa values Soil Dynamics and Earthquake Engineering,
are greater than 20 and Rf greater than 14(2), pp 115-123.
0.8 percent. 9. Eslami, A. & Fellenius, B.H. (2004), “CPT and
• Robertson (1990) (Qtn-Fr) SBT chart where CPTu data for soil profile interpretations;
most HPR soils have Fr values greater than review of methods and a proposed new
0.80 percent. approach”, Iran Journal of Science and
Technology 28(B1), pp 69-86.
RECOMMENDATIONS 10. FDOT. (2010), Standard Specification for
It is recommended that engineers use the Road and Bridges Section 455.
three most promising SBT charts when large 11. GeoLogismiki. (Released 2007), CPeT-IT,
diameter, high displacement piles are to be Version 1.70. GeoLogismiki.
driven into saturated very-dense or hard silty
or clayey fine sands. Measured rebound should

[44] DFI JOURNAL Vol. 7 No. 2 December 2013

12. Hussein, M.H., Woerner, W. A., Sharp, M. 18. Moayed, R.Z. (2006), “Evaluation of the fine
and Hwang, C. (2006), “Pile driveability and contents of silty sands using CPTU results”,
bearing capacity in high-rebound soils”, Proceedings of the 10th IAEG International
American Society of Civil Engineers GEO Congress, Nottingham, United Kingda;
Congress CD-ROM, Atlanta, GA. publication #506.
13. Jarushi, F., (2013), “Evaluating geotechnical 19. Murrell, Kyle L., Canivan, Gregory J., Camp,
engineering properties associated with high William M. III. (2008), “High and low strain
pile rebound”, Ph.D. Dissertation, Florida testing of bouncing piles”, Proceedings of
Institute of Technology, Melbourne Florida. the 33rd Annual and 11th International
14. Jarushi, F , Cosentino, P.J, and Kalajian E.H., Conference on Deep Foundations.article
(2013), “Using fines content and uncorrected #1603; publication #85
SPT blow counts of soils to predict high pile 20. Robertson, P.K. (1990), “Soil Classification
rebound”, Journal of the Transportation using the cone penetration test”, Canadian
Research Board: Paper No. 13-2880. Geotechnical Journal, 27(1), pp151-8.
15. Jefferies, Mike, and Ken Been. (2006), “Soil 21. Robertson. P.K., Campanella, R.G., Gillespie,
Liquefaction: A critical state approach”, D., and Greig, J., (1986), “Use of Piezometer
London: Taylor & Francis. Cone data”, In-Situ’86 Use of In-situ testing
16. Likins, Garland E. (1983), “Pile installation in Geotechnical Engineering, GSP 6 , ASCE,
difficulties in soils with large quakes”, Reston, VA, Specialty Publication, pp 1263-
In G.G. Globe, editors, Proceedings of 1280.
Symposium 6 at the 1983 ASCE Convention, 22. Schmertmann, J.H. (1978), “Guidelines for
Philadelphia, May 18, 1983. ASCE cone penetration test, performance and
Geotechnical Engineering Division. design”, Report No. FHWA-TS-78-209, U.S.
17. Lunne, T., Robertson, P.K., and Powell, Department of Transportation, Washington,
J.J.M. (1997), “Cone penetration testing in D.C., pp. 145.
geotechnical practice”, Blakie Academic and 23. Schneider, J.A., Randolph, M.F., Mayne,
Professional, Melbourne, 312. P.W., and Ramsey, N. (2008), “Analysis of
factors influencing soil classification using
normalized piezocone tip resistance and
pore pressure parameters”, ASCE Journal
of Geotechnical and Geoenvironmental
Engineering, 134(11), pp1569-1586.

DFI JOURNAL Vol. 7 No. 2 December 2013 [45]

 Factors Affecting the Reliability of Augered
Cast-In-Place Piles in Granular Soils at the
Serviceability Limit State (DFI 2013 Young
Professor Paper Competition Winner)
Armin W. Stuedlein, Ph.D., P.E., Assistant Professor and Loosley Faculty Fellow, School of Civil
and Construction Engineering, Oregon State University, Corvallis, OR, USA; (541) 737-3111,
[email protected]
Seth C. Reddy, E.I., Graduate Research Asst., School of Civil and Construction Engineering,
Oregon State University, Corvallis, OR, USA, [email protected]

ABSTRACT
Owing to an increasing demand to manage risk and maximize cost-effectiveness, preference for
reliability-based design (RBD) over traditional deterministic design procedures has increased for deep
foundation elements. In this study, factors affecting the reliability of augered cast-in-place (ACIP) piles
under axial compression at the serviceability limit state (SLS) are addressed using a simple probabilistic
hyperbolic model and a database of static loading tests conducted on ACIP piles in cohesionless
soils. The aleatory and model uncertainty in a selected two-parameter load-displacement model is
statistically characterized for use in reliability simulations. Reliability simulations incorporating the
correlated bivariate model parameter distribution were generated using a statistical translational
model and various parametric and non-parametric correlation coefficients to assess the effect of
correlation coefficient type on the reliability simulations. The first-order reliability method (FORM)
was used to determine the effect of sample size on the stability and uncertainty of the serviceability
limit state reliability index. Sample sizes greater than about 40 provided relatively consistent estimates
of the reliability index; however, its uncertainty continued to decrease with increasing sample sizes.
A parametric study was conducted in order to determine the variables (i.e. allowable displacement,
predicted pile capacity, slenderness ratio) which govern reliability. In general, the uncertainty in the
model used to predict pile capacity had a more significant impact on foundation reliability compared
to the uncertainty in allowable displacement; this finding illustrates one advantage of having an
accurate capacity prediction model. The slenderness ratio had the largest effect on foundation
reliability at the SLS, and illustrates the importance for accounting for the pile geometry in reliability
assessments.

INTRODUCTION uncertainty is increasing. Consequently, RBD

It is widely recognized that many geotechnical methods are replacing traditional deterministic
design parameters exhibit uncertainties; these design procedures (e.g. allowable stress design
have been historically accounted for in a [ASD]) for many deep foundation alternatives.
deterministic framework using a global safety Augered cast-in-place (ACIP) piles have generally
factor based on experience or general rules received limited attention; however, Stuedlein et
of thumb. This approach, however, does not al. (2012) present one example of a probabilistic
recognize the uncertainty in the individual ACIP-specific design model intended for the
underlying variables nor their potential ultimate limit state (ULS).
correlation. Reliability-based design (RBD) Although the reliability of the serviceability
can overcome many of these limitations, and limit state (SLS) is not as well understood as
can provide a means for estimating the risk that of the ULS, it is often the governing design
or probability of exceeding a particular design criterion (Uzielli and Mayne 2011). Phoon
limit state. The demand for managing risk by et al. (2006) discussed key factors impeding
explicitly recognizing and mitigating sources of RBD for the SLS and indicated the lack of

[46] DFI JOURNAL Vol. 7 No. 2 December 2013

model statistics for pile displacement. Phoon MODELING OF THE SERVICEABILITY
(2006) recognized that the uncertainty in LIMIT STATE
foundation displacement will have a significant
Ideally, RBD principles should be used to
impact on the reliability of a foundation and
estimate the likelihood of failure at both the
proposed a simple probabilistic hyperbolic
ULS and SLS. Reliability of foundations for the
model that captures the uncertainty in the
ULS is associated with the probability that a
entire load-displacement relationship. The
single value of pile capacity (i.e. the ultimate
applied load was normalized by an interpreted
resistance) will be less than the applied
failure load and the remaining uncertainty
load. If pile capacity is determined using a
in the load-displacement curve was modeled
consistent criterion, a model factor (i.e., the
using a bivariate random vector consisting
bias), defined as the ratio of the measured to
of hyperbolic curve-fitting parameters that
predicted capacity can be used to assess the
were found to be correlated and non-normally
uncertainty in the design model. The SLS,
distributed. Phoon and Kulhawy (2008)
however, is defined by one or more pre-defined
describe two methods to incorporate the
allowable displacements, which are related
correlated random variables comprising the
to corresponding allowable loads. Due to the
load-displacement model, and point to their
nonlinearity of the load-displacement curve, the
performance in consideration of a database of
model factor distribution must be evaluated for
40 loading tests on ACIP piles.
each specified allowable displacement. Efficient
This paper examines certain aspects of RBD RBD analyses for the SLS can consider the
for ACIP piles at the SLS using an expanded entire load-displacement curve in which a range
load test database. Contrary to previous work, of allowable displacements may be specified
dependence between the load-displacement (Uzielli and Mayne 2011).
model parameters and pile slenderness ratio
Multiple sources of uncertainty influence the
was observed; this correlation was eliminated
behavior of the load-displacement relationship
through the use of two transformed load-
of a deep foundation element. Development
displacement model parameters, which were
of a load-displacement model from a database
then used to assess foundation reliability for
allows the aleatory uncertainty resulting from
various pile geometries using the translation
the uniqueness of each test, as a consequence
model described in Phoon and Kulhawy (2008)
of inherent soil variability, and the epistemic
and the first-order reliability method (FORM).
model uncertainty to be captured (albeit
Because the correlation coefficient used in the
indirectly) and statistically characterized.
translation model is not suitable for highly
Epistemic uncertainties resulting from errors
correlated model parameters, random pairs
in measurement and observation of a loading
of model parameters were sampled from the
test contribute to the variability in the load-
database in order to estimate the minimum
displacement behavior of a pile, and this effect
sample size needed to use the translation
is lumped into the model uncertainty.
model. The effect of sample size on the
distribution of computed reliability indices Although any appropriate nonlinear function
was assessed for a variety of correlation can be used to model load-displacement data,
coefficients and a range of slenderness ratios Phoon (2006) showed that a hyperbolic curve
in order to assess the effect of correlation provided an adequate fit to the ACIP pile load
modeling on the reliability of the SLS design test data compiled in Chen (1998) and Kulhawy
model. By varying the mean and uncertainty and Chen (2005). This model was adopted
of allowable displacement, uncertainty of the herein for the purpose of identifying factors
predicted resistance, and slenderness ratio, a controlling the determination of the reliability
parametric study was conducted to determine index, β. Hyperbolic load-displacement curves
the variables that govern ACIP pile foundation were fitted to the observed load-displacement
reliability. Because slenderness ratio was found using the applied load, Q, normalized by the
to be the most critical parameter, additional slope-tangent capacity, QSTC, given by:
investigations were undertaken to better Q ya
quantify the influence of slenderness ratio on = [1]
QSTC k1 + k2 ⋅ ya
foundation reliability at the SLS.

DFI JOURNAL Vol. 7 No. 2 December 2013 [47]

 where y is the pile head displacement, and k1 RANDOMNESS OF THE HYPERBOLIC
and k2 are the curve-fitting parameters that MODEL PARAMETERS
describe the shape of the load-displacement
In order to use traditional statistical methods
curve. The slope-tangent capacity, similar to
and be considered random, k1 and k2 must not
the Davisson offset load but uses the initial
be statistically correlated to soil strength or
load-displacement curve slope rather than the
geometrical parameters (e.g., SPT-N and D/B).
pile elastic compression line and described in
The non-parametric Kendall’s Tau test (Daniel
detail by Kulhawy (2004), is not necessarily the
1990) for correlation between k1 and average
ultimate pile resistance but a simple reference
SPT-N and k2 and average SPT-N yielded p-values
capacity that can be derived from most load-
= 0.81 and 0.93, respectively, indicating no
displacement curves and is used to reduce the
correlation at a 5 percent level of significance.
scatter in the observed test results through
Conversely, there was convincing evidence
normalization.
(p-values < 0.05) that suggested neither k1 nor
The hyperbolic model parameters are k2 were independent of D/B. Considering only
physically meaningful: the reciprocal of k1 the Phoon and Kulhawy (2008) dataset, the
and k2 equal the initial slope and asymptotic Kendall Tau test indicated k1 and D/B were
(or ultimate) resistance, respectively. The correlated (p-value = 0.02), whereas k2 and
curve fitting parameters from the Chen (1998) D/B were independent (p-value = 0.40). Fig. 1
and Kulhawy and Chen (2005) database were shows that the potential correlation between
obtained directly from Phoon and Kulhawy the model parameters and the slenderness
(2008); ordinary least squares regression was ratio is strengthened by the addition of the new
performed to determine the model parameters data. A positive correlation between k1 and
for the newly added load test data. D/B is physically intuitive; that is, a larger k1
represents a less stiff pile, corresponding to a
LOAD TEST DATABASE larger slenderness ratio. Similarly, a larger k2
The database compiled herein consisted of 87 indicates less ultimate resistance which likely
load tests performed on ACIP piles installed in corresponds to a smaller embedment depth
predominately cohesionless soils: 40 loading since the range of B in the database is relatively
tests were compiled by Chen (1998) and small. These findings indicate one pitfall
Kulhawy and Chen (2005), 23 were compiled by associated with the use of a limited dataset for
McCarthy (2008), ten were presented in local reliability calculations.
ASCE Chapter meetings and were reported by To assess the likelihood that k1 and k2 were
Stuedlein et al. (2012), ten were compiled by actually correlated with D/B, the new database
Park et al. (2012), three from load tests were was randomly sampled using sample sizes,
conducted by Mandolini et al. (2002), and one n, ranging from 5 to 85, and the p-values
load test was reported by O’Neill et al. (1999). associated with the Kendall’s Tau correlation
The diameter, B, depth, D, and slenderness test were calculated. One million subsets of k1,
ratio, D/B, ranged from 300 to 800 mm (11.8 to k2, and D/B were generated using Monte Carlo
31.5 in), 7.5 to 29.0 m (24.6 to 95.1 ft), and 20 simulations for each sample size evaluated.
to 68.5 m (65.6 to 224.7 ft), respectively. The Monte Carlos simulations, in this case repetitive
database compiled by Chen (1998) and Kulhawy deterministic calculations that incorporate
and Chen (2005) included load tests on short samples of independent random variables from
piles (D/B < 20), but these were not assessed their source distributions, provide a convenient
herein. Kulhawy and Chen (2005) observed that framework for estimated probabilistic outcomes
shaft resistance constituted a small portion based on limited and/or uncertain datasets.
of the total resistance for short piles, and Fig. 2 shows the relationship between the
consequently, the load-displacement behavior mean p-value and sample size, as well as the
is different from other piles in the database. probability that the p-value is less than the
In order to assess the effect of sample size selected significance level, α of 0.05, Pr(p-value
on reliability and make direct comparisons < 0.05). The coefficient of variation (COV),
to previously reported studies, the expanded defined as the standard deviation divided by
database is limited to slenderness ratios greater the mean, was used to assess the degree of
than 20. scatter of the simulated p-values, which ranged

[48] DFI JOURNAL Vol. 7 No. 2 December 2013

 20 1.5
New Data
Phoon and Kulhawy (2008)
15
1.0

k2
10
k1

0.5
5

(a) (b)
0 0.0
0 20 40 60 80 0 20 40 60 80
D/B D/B
[FIG. 1] Correlation between slenderness ratio D/B and hyperbolic parameters k1 (a) and k2 (b) from Phoon and Kulhawy
(2008) and new data

from 54 to 350 percent as a function of sample Considering the typical ACIP pile response, a
size. Because p-values cannot be used directly small initial slope of the load-displacement
to determine the strength of the likelihood of curve (i.e., a large k1 value) implies a slowly
correlation between two variables, and due to decaying curve, and is generally associated
the large amount of scatter associated with with a less well-defined and larger asymptote
the simulated p-values, a curve representing (i.e. smaller k2), and vice versa. Therefore,
the probability of rejecting the null hypothesis it is expected that k1 and k2 will be inversely
(i.e. the probability of obtaining a p-value correlated to some degree. The magnitude of
< 0.05), was constructed. Fig. 2 indicates the dependence between normally distributed
that the mean p-value is less than 0.05 for k1 and k2 can be characterized using the Pearson
sample sizes equal to 19 and 23 for k1 and k2, product-moment correlation coefficient, ρ, as
respectively. However, the probability that k1 shown in Fig. 3:
and k2 are correlated with D/B is approximately n
73 percent. At n = 40, Pr(p-value < 0.05) = 99.89 ¦(k 1,i − k1 )( k2,i − k2 )
and 99.04 percent for k1 and k2, respectively; ρ= i =1
[2]
n n
whereas Pr(p-value < 0.05) = 100 percent at n =
¦(k − k1 ) ⋅ ¦ ( k2,i − k2 )
2 2
55 and 60 for k1 and k2, respectively. Therefore, 1,i
i =1 i =1
direct use of k1 and k2 should not be permitted
for reliability simulations as they are very likely where k1- and k2- equal the mean value of the
to be correlated to slenderness ratio. corresponding hyperbolic model parameter.
0.60 100 Fig. 3a shows that the new model parameters
agree with those reported by Phoon and
0.50 Kulhawy (2008), but indicates that the
80
strong inverse correlation between k1 and
Pr(p-value < 0.05)

p-value (k1-D/B)
0.40 k2 is significantly stronger than previously
Mean p-value

p-value (k2-D/B)
60
Pr(p-value<0.05) (k1-D/B) characterized. To perform accurate and
0.30 Pr(p-value<0.05) (k2-D/B) unbiased reliability analyses at the SLS, the
40 correlation between model parameters and D/B
0.20
must be considered. The dependence of k1
Uncorrelated 20
0.10 and k2 on D/B was eliminated for the purposes
5% Significance Level
of simulation by transforming the model
Correlated
0.00 0 parameters:
0 10 20 30 40 50 60 70 80
Sample Size, n B
k1,t = k1 ⋅ [3]
[FIG. 2] Average p-values from the Kendall’s tau D
correlation test and the probability of obtaining a p-value
less than 0.05 as a function of sample size using Monte
Carlo simulations from the expanded database

DFI JOURNAL Vol. 7 No. 2 December 2013 [49]

 lognormality for k1,t and k2,t at α = 5 percent.
D
k2,t = k2 [4] Although the distribution of k2,t could also be
B considered normal based on the Anderson-
Darling test, the null hypothesis of normality
The Kendall’s Tau correlation test between k1,t
for k1,t was rejected (p-value < 0.05). In addition
and D/B, k2,t and D/B, k1,t and SPT-N, and k2,t
to the evidence provided by the Anderson-
and SPT-N yielded p-values = 0.78, 0.56, 0.37,
Darling test, a lognormal marginal distribution
and 0.96, respectively, indicating no correlation
was adopted for k1,t and k2,t because it
for α = 5 percent. Fig. 3b illustrates that the
appeared to fit the data better than the normal
hyperbolic model parameters remain correlated
distribution, and the lognormal distribution is
to one another once transformed using Eqns. 3
confined to positive real values.
and 4, respectively.
1.2 1.0
New Data Lognormal

Cumulative Probability Density

Phoon and Kulhawy (2008)
0.9
1.0 Normal
ȡ = -0.67 (Phoon and Kulhawy
0.8
2008 dataset)
0.8 0.7
ȡ = -0.80 (Expanded dataset)
0.6
0.6
k2

0.5
0.4 0.4 Mean = 0.16
St. Dev. = 0.08
0.2 0.3
COV = 48.7
(a) 0.2 n = 87
0.0
0.1
0 5 10 15 20 (a)
k1 0.0
7.0 0 0.1 0.2 0.3 0.4 0.5
k1,t
6.0 1.0
ȡ = -0.69 (Phoon and
Cumulative Probability Density

5.0 Kulhawy 2008 dataset) 0.9

ȡ = -0.73 (Expanded dataset)
4.0 0.8
k2,t

0.7
3.0
0.6
2.0 0.5
1.0 0.4 Mean = 3.40
(b) St. Dev. = 0.80
0.0 0.3 COV = 23.4
0.2 n = 87
0 0.2 0.4 0.6
k1,t 0.1
(b)
[FIG. 3] Correlation between model parameters (a) k1 0.0
and k2 and (b) k1,t and k2,t from Phoon and Kulhawy (2008)
and the new data 0 2 4 6
k2,t
[FIG. 4] Empirical, lognormal, and normal marginal
The statistical distributions of k1,t and k2,t
cumulative distributions for the hyperbolic model
must be determined in order to accurately parameters: (a) k1,t, and (b) k2,t.
simulate the uncertainty in the observed
load-displacement curves. The empirical, In order to adequately represent the
fitted normal, and fitted lognormal marginal uncertainty associated with the SLS model,
cumulative distribution functions (CDF) load-displacement curves were simulated using
for k1,t and k2,t are shown in Fig. 4a and 4b, randomly generated, lognormally distributed,
respectively. The Anderson-Darling goodness- correlated pairs of k1,t and k2,t. The simulated
of-fit test (Anderson and Darling 1952) provided curves were then compared to the normalized
no evidence to reject the null hypothesis of

[50] DFI JOURNAL Vol. 7 No. 2 December 2013

empirical curves from the database to assess herein (ρln, ρs, and ρτ) for k1 and k2, and k1,t
the appropriateness of the simulations. The and k2,t.
simulation of k1,t and k2,t using a translational To verify that the translational model described
model requires generating uncorrelated above can adequately reproduce the uncertainty
standard normal random variables Z1 and Z2 in the observed load-displacement curves,
(mean = 0, standard deviation = 1) which are k1,t and k2,t were first simulated using Eqns.
then transformed into correlated random 8a-b and then back-transformed into k1 and
variables X1 and X2 (Phoon and Kulhawy 2008): k2 using deterministic values of D/B because
X 1 = Z1 [5a] the uncertainty associated with the individual
pile geometry due to construction error could
[5b] not be evaluated and the distribution of
X 2 = Z1 ⋅ ρln + Z 2 1 − ρ ln2 D/B in the database was relatively uniform.
Fig. 5 compares the observed and simulated
where ρln = an equivalent-normal correlation
model parameters and corresponding load-
coefficient and is given by:
displacement curves using ρln and D/B = 25, 45,
ª
ln « ρ ⋅ (e ζ 1,2t
)(
−1 e
ζ 2,2 t
) º
− 1 + 1» [6]
and 65. The slenderness ratios in Fig. 5 were
selected in order to represent the distribution
ρln = ¬ ¼
of observed values in the database and illustrate
ζ 1,t ⋅ ζ 2,t the difference in the range of simulated model
parameters for different D/B.
Where λ1,t, ζ1,t and λ2,t, ζ2,t are the approximate
lognormal mean and standard deviation of k1,t In general, the simulated model parameters
and k2,t, respectively. These second moment in Fig. 5 represent the scatter of the observed
descriptors were calculated from the sample values well. Owing to the values of D/B used
mean, ki,t, and standard deviation, σi,t: to back-transform k1,t and k2,t, the simulated
model parameters in Fig. 5 are associated
ζ i ,t = ln(1 + σ i2,t / ki ,2t ) [7a] with different sections of the observed k1-k2
relationship. Smaller slenderness ratios are
associated with smaller and larger k1 and k2,
λi ,t = ln(ki ,t ) − 0.5 ⋅ ζ i2,t [7b]
respectively, whereas the opposite holds for
larger D/B. For each independent average
Lognormally distributed k1,t and k2,t simulations D/B in Fig. 5 (i.e. 25, 45, 65), the correlation
were then calculated by: coefficient of the back-transformed simulated
(ζ 1,t ⋅ X1 + λ1,t ) [8a] model parameters is equal to ρln(k1,t,k2,t) = -0.82
k1,t = e as shown in Table 1; the combined coefficient
of the back-transformed simulated model
(ζ 2,t ⋅ X 2 + λ2,t ) [8b]
k2,t = e parameters should be approximately equal to
ρln(k1,k2) = -0.92 in Table 1) but will depend on
Because ρln is unstable for highly correlated the number and values of D/B selected.
variables and small sample sizes, the
In order to validate the use of a non-parametric
appropriateness of two non-parametric
correlation coefficient in place of ρln in Eqn. 5b
correlation coefficients for use in Eqn. 5b,
and compare the degree of scatter produced
including Kendall’s Tau, ρτ, and Spearman’s
in the simulated load-displacement curves,
rank, ρs, for simulation of load-displacement
5,000 simulations of k1,t and k2,t were generated
curves was investigated. Substituting ρs for
using Eqns. 8a and 8b, respectively, and the
ρln in Eqn. 5b is equivalent to the rank model
correlation coefficients shown in Table 1. The
described by Phoon and Kulhawy (2008).
simulated, transformed model parameters
Spearman’s rank coefficient is calculated by
were then converted to k1 and k2 for D/B = 25,
applying Eqn. 2 to the ranked distributions of
35, 45, 55, and 65 in order to provide a higher
k1,t and k2,t, whereas ρτ is based purely on the
resolution of simulated load-displacement
number of concordant and discordant pairs
curves.
of k1,t and k2,t and thus provides an unbiased
estimate of the population parameter. Table 1
shows each correlation coefficient investigated

DFI JOURNAL Vol. 7 No. 2 December 2013 [51]

 1.2 2.5
Simulated (D/B=25)
1.0 Simulated (D/B=45)
Simulated (D/B=65) 2.0
0.8 Observed

STC
ρln(Observed) = -0.92 1.5
k2

0.6 ρln(Simulated) = -0.91

Q/Q
1.0
0.4

0.2 0.5
(a) (b)
0.0 0.0
0 5 10 15 20 0 10 20 30 40 50
k y
1
[FIG. 5] (a) Observed and simulated hyperbolic model parameters, k1 and k2, and (b) load displacement curves for pln and
D/B = 25, 45, and 65

[TABLE 1] Correlation coefficients for original applied load, Q, equals or exceeds an allowable
and transformed model parameters load, Qa, which can be related to ya through
the use of a selected load-displacement model.
Correlation The assessment of the SLS can be evaluated
k1, k2 k1,t, k2,t quantitatively through the use of a performance
Coefficient
function, P, defined here as (Phoon and Kulhawy
pln -0.919 -0.819 2008):
ps -0.850 -0.824 P = ya − y ( Q ) = Qa ( ya ) − Q [9]

pτ -0.702 -0.648
which is equal to or less than zero for the SLS.
If Q is assumed to be deterministic, y remains
Fig. 6 shows the simulated and observed a random variable because of the uncertainty
distribution of k1 and k2 and the corresponding present in the load-displacement model (Phoon
load-displacement curves using ρln, ρs, and 2006). Similarly, Qa is a random variable even
ρτ. In general, the observed and simulated if ya is assumed to be deterministic. The
correlation coefficients are in good agreement, probability of exceeding the limit state (i.e.,
where ρln(Simulated), ρs(Simulated), and failure), pf, can be defined as:
ρτ(Simulated) are calculated with Eqn. 6 in order
to be consistent with the methodology used p f = Pr ( P < 0 ) [10]
to simulate k1,t and k2,t. Ideally, the combined
correlation coefficient of the back-transformed Eqns. 1 and 9 were substituted into Eqn. 10 and
simulated model parameters should equal that rearranged in terms of a deterministic mean
of the observed model parameters. The degree factor of safety, FS, to determine the probability
of scatter in the simulated k1 and k2 values is of failure (Phoon and Kulhawy 2008):
inversely proportional to the magnitude of the
correlation coefficient, where the scatter in Figs.
§ ya 1 Q′ ·
p f = Pr ¨ < ⋅ [11]
6a and 6b is smaller compared to Figs. 6e and ¨ k1 + k2 ⋅ ya FS Q′p ¸¸
© ¹
6f since | ρln | > | ρτ |. Based on the simulated
load-displacement curves, it appears that the where Q’ and Q’p are unit mean random
use of ρτ provides the most appropriate k1 - k2 variables associated with the applied load
correlation. and predicted pile capacity, respectively, and
FS is associated with the ULS and equivalent
INVESTIGATION OF RBD FOR THE to a global value adopted in current practice
SERVICEABILITY LIMIT STATE (Phoon 2006). The inverse standard normal
The serviceability limit state (SLS) occurs when function, Φ-1, maps the probability of failure to
the foundation settlement, y, equals or exceeds the reliability index, β, in order to provide an
a predetermined allowable settlement, ya. In estimate of the foundation reliability at the SLS:
terms of load, the SLS takes place when the β = −Φ −1 ( p f ) [12]

[52] DFI JOURNAL Vol. 7 No. 2 December 2013

[FIG. 6] Observed and simulated hyperbolic curve-fitting parameters, k1 and k2, and load-displacement curves for: (a) and
(b) pln, (c) and (d) ps, and (e) and (f ) pτ .

defined as the number of standard deviations be known (or assumed) for all variables
between the mean of the multivariate resistance contributing to foundation reliability to perform
distribution and the limit state surface. To RBD simulations. Model statistics for k1,t and
accurately assess foundation reliability at k2,t were obtained from the database herein
the SLS, the uncertainty in y and ya must be and found to be correlated and lognormally
characterized (Zhang and Phoon 2006). In this distributed, as described earlier. Table 2
probabilistic framework, the uncertainty in y summarizes the statistics for the other random
is captured by the bivariate distribution of k1 variables in Eqn. 11, selected for comparison
and k2 that is associated with the hyperbolic to a previous study by Phoon and Kulhawy
model, whereas ya in Eqn. 11 is treated as a (2008). The model statistics for Q’p given in
random variable with a pre-defined magnitude Table 2 were associated with the Meyerhof
of uncertainty. method for shaft resistance that was modified
by Kulhawy and Chen (2005) for use with ACIP
Model Statistics piles. The model statistics for Q’ are consistent
Second-moment model statistics (i.e. mean with that recommended for live loads by
and COV) and the type of distribution must Paikowsky et al. (2004). Phoon and Kulhawy

DFI JOURNAL Vol. 7 No. 2 December 2013 [53]

 (2008) recommended a mean ya of 25 mm (1 in) Effect of Sample Size and Slenderness Ratio on
based on Skempton and McDonald (1956), and Reliability
a COV of 60 percent based on probabilistic In order to study the effect of sample size
distributions of limiting tolerable displacements on the distribution of the reliability index,
reported by Zhang and Ng (2005). Each random simulations that randomly sampled the new
variable in Table 2 was modeled using a k1,t and k2,t data set were performed. Fifty-
lognormal distribution. thousand subsets of k1,t and k2,t with sample
size n ranging from 5 to 85 were generated
[TABLE 2] Statistics for the transformed model using Monte Carlo simulations. The Spearman
parameters, k1,t and k2,t from the new database
rank, Kendall’s Tau, and equivalent normal
and model statistics from Phoon and Kulhawy
correlation coefficients (ρs, ρτ, and ρln) and the
(2008)
corresponding β were calculated for each subset
using the model statistics for ya, Q’, Q’p, and FS
Random in Table 2.
Mean COV %
Variable
Figs. 7a-c show the effect of sample size on
k1 0.16 48.7 the mean and COV of β for the SLS at y =
25 mm (1 in) assuming D/B = 65, 45, and 25,
k2 3.40 23.4 respectively. Although the mean β stabilizes
ya 25 60 rapidly, the uncertainty in β continues to
decrease with increasing sample size for each
Q'p 1 50
D/B and correlation coefficient. However, the
Q' 1 20 COV(β) remains relatively small compared to
FS 3 - typical geotechnical design parameters, and
falls below 2 and 1 percent between n = 25
and 58, and 55 and 78, respectively, for the
First-order Reliability Method range of D/B and the correlation coefficients
Although closed-form solutions exist for investigated. In general, ρs and ρln produced
estimating β at the SLS, the distribution of the similar β for larger sample sizes. Though the
model factor has to be evaluated each time a use of ρln and ρs consistently produces larger β
new allowable displacement is specified (Phoon, compared to ρτ, the difference is relatively small
2006). The first-order reliability method (FORM) (e.g., 0.024 at n = 85 and D/B = 25). Thus, β is
is therefore a more suitable approach as ya is relatively insensitive to the type of correlation
treated as a random variable. In FORM, each coefficient and the sample size for n ≥ 40.
random variable in the limit state function (k1,
Effect of Design Variables on Pile Foundation
k2, ya, Q’, Q’p) is transformed into a standard Reliability at the SLS
normal variable such that the differences
In order to determine which design variables
in magnitude of the random variables are
most strongly influence foundation reliability
eliminated (Hasofer and Lind 1974). The
at the SLS, reliability indices were calculated for
random variables in Eqn. 11 are represented
different values of the mean and uncertainty
by a multivariate probability distribution that
of allowable displacement, the uncertainty of
is traversed by the performance function. The
predicted resistance, and slenderness ratio.
portion of the multi-dimensional distribution
Each random variable in the performance
beneath the hyper-surface where P ≤ 0 is equal
function (Eqn. 11) was presumed to be
to the probability of failure. FORM assumes
lognormally distributed, while the statistics
the limit state function is linear at the most
for k1,t and k2,t (i.e. mean, COV) were obtained
probable failure point, so the probability of
directly from the database herein. For allowable
failure can be calculated by evaluating the
displacement, the mean and COV was varied
standard normal CDF as a function of β. This
from 10 to 50 mm (0.4 to 2.0 in) and 5 to
approach may not be applicable for strongly
85 percent, respectively. The uncertainty of
nonlinear limit state functions (Chan and Low
predicted resistance, which relates to different
2011) or when probabilities of failure are very
capacity prediction methodologies with an
large, though the latter case is typically avoided
associated amount of uncertainty, ranged
in geotechnical applications.

[54] DFI JOURNAL Vol. 7 No. 2 December 2013

 12 1.84
D/B = 65
10

Mean Reliability Index, ȕ

COV (Kendall Tau)
COV (Spearman rank)
1.82
COV (Eqv. Normal)
8
COV (ȕ), %

ȕ (Spearman rank)
ȕ (Kendall Tau) 1.80
6 ȕ (Eqv. Normal)

1.78
4

2 1.76
(a)
0 1.74
0 20 40 60 80

8 2.09
D/B = 45

Mean Reliability Index, ȕ

2.08
6
COV (ȕ), %

2.07
4
2.06

2
2.05
(b)
0 2.04
0 20 40 60 80

6 2.22
D/B = 25

5
Mean Reliability Index, ȕ

2.21
COV (ȕ), %

4
2.20
3
2.19
2

1 2.18
(c)
0 2.17
0 20 40 60 80
Sample Size, n
[FIG. 7] The relationship between the mean and COV of
reliability index and sample size using 50,000 realizations
for pln, ps, and pτ and D/B = (a) 65, (b) 45, and (c) 25

from 5 to 85 percent. In order to investigate

the effect of pile geometry on β, slenderness
ratios of 25 and 65 were chosen based on the [FIG. 8] The effect of COV(ya), COV(Q’p), and D/B on β for
approximate upper and lower bounds observed mean ya = 10 (a), 20 (b), 30 (c), 40(d), and 50 mm (e) using
in the database. The statistics for the other the first-order statistics of k1,t and k2,t (Table 2)
variables in the performance function (Q’ and
FS) are listed in Table 2.

DFI JOURNAL Vol. 7 No. 2 December 2013 [55]

 The effect of varying the mean and COV of ya, 2.3
COV of Q’p, and D/B on β is shown in Figs. 8a-e Correlation Coefficient
for allowable pile head displacements ranging 2.2 Spearman Rank

Mean Reliability Index, ȕ

from 10 to 50 mm (0.4 to 2.0 in). In general, Eqv. Normal
decreasing the uncertainty in Q’p and ya, and 2.1
Kendall Tau
increasing the mean ya results in a larger
2.0
reliability index and a smaller probability of
exceeding ya. When the uncertainties in Q’p and 1.9
ya are relatively small (5-45 percent), foundation
reliability decreases rapidly with increasing 1.8
uncertainty in Q’p; this trend is more prominent
for smaller slenderness ratios and larger mean 1.7
allowable displacements. As the uncertainties 20 30 40 50 60 70
Slenderness Ratio, D/B
for Q’p and ya become larger, β continues to
decrease though at a slower rate. For each [FIG. 9] The relationship between the mean reliability
mean ya and slenderness ratio investigated index and D/B using pln, ps, and pτ .

herein, the uncertainty in Q’p had a larger

effect on β as compared to the uncertainty in uncertainty in the computed reliability index,
ya. For larger allowable displacements and β, was assessed for an equivalent normal
smaller slenderness ratios, β was observed to and two different non-parametric correlation
be insensitive to the level of uncertainty in ya, coefficients (Spearman rank, Kendall Tau) by
compared to Q’p; this demonstrates the benefit generating random subsets of the transformed
of an accurate ACIP design methodology. As load-displacement model parameters from the
mean ya increases and its associated uncertainty ACIP pile database. The mean reliability index
decreases, β approaches an upper bound limit rapidly stabilized with sample size, n, and was
regardless of level of uncertainty in Q’p. Overall, shown to be relatively insensitive for n > 40.
the slenderness ratio had a considerable impact Although β is fairly stable with relatively low
on foundation reliability, especially at lower database sample sizes, this work shows that
mean allowable displacements. Fig. 9 shows the accurate characterization of the degree of
the dependence of β on slenderness ratio; for correlation among bivariate random variables
example β = 2.21 and 1.76 (pf = 1.34 and 3.90 requires as many samples as possible. A
percent) for D/B = 25 and 65, respectively. The parametric study was conducted in order to
contrast in the level of reliability is directly due identify the factors which govern foundation
to the difference in the statistical parameters reliability at the SLS. The uncertainty in the
of k1 and k2 that describe the characteristic capacity prediction model was found to have a
behavior in the load-displacement relationship larger effect on β compared to the uncertainty
for different D/B. As a result, taking into in allowable displacement; this finding
account the correlation between model illustrates one benefit of an accurate capacity
parameters and slenderness ratio is essential prediction model. Owing to the dependence
when assessing the reliability of ACIP piles. of the model parameters on pile stiffness and
geometry, β was found to be very sensitive to
SUMMARY AND CONCLUSIONS D/B, and this effect must be accounted for in
This paper investigated the effect of various the reliability-based serviceability limit state
factors in reliability-based serviceability design of deep foundations.
limit state (SLS) design using an expanded
database consisting of ACIP piles installed
REFERENCES
in cohesionless soils. The hyperbolic model 1. Anderson, T.W. and Darling, D.A. (1952)
parameters, k1 and k2, used to model the load- “Asymptotic Theory of Certain Goodness-of-
displacement data were shown to be more Fit Criteria Based on Stochastic Processes”,
strongly correlated than previously reported. The Annals of Mathematical Statistics, 23(2),
Additionally, these model parameters were pp. 193-212.
found to be correlated to the slenderness
ratio (D/B). After removing the dependence
from D/B, the effect of sample size on the

[56] DFI JOURNAL Vol. 7 No. 2 December 2013

2. Chan, C.L. and Low, B.K. (2011) “Practical 12. Paikowsky, S.G., with contributions from
Second-order Reliability Analysis Applied Birgisson, B., McVay, M., Nguyen, T., Kuo,
to Foundation Engineering”, International C., Baecher, G., Ayyab, B., Stenersen, K.,
Journal for Numerical & Analytical Methods O’Malley, K., Chernauskas, L., and O'Neill, M.
in Geomechanics, DOI: 10.1002/nag.1057. (2004) “Load and Resistance Factor Design
3. Chen, J.R. (1998) “Case History Evaluation of (LRFD) for Deep Foundations”, NCHRP
Axial Behavior of Augered Cast-in-Place Piles Report 507, Transportation Research Board,
and Pressure-Injected Footings”, MS Thesis, Washington, DC, 126 pp.
Cornell University, Ithaca, NY. 13. Phoon, K.K. (2006) “Serviceability Limit
4. Daniel, W.W. (1990) “Applied Nonparametric State Reliability-based Design”, International
Statistics”, 2nd Edition, PWS-Kent, Boston. Symposium on New Generation Design
Codes for Geotechnical Engineering Practice,
5. Hasofer, A.M. and Lind, N. (1974) “An Exact
November 2-3, Taipei, Taiwan, 18 pp.
and Invariant First-Order Reliability Format”,
Journal of Engineering Mechanics, ASCE, 14. Phoon, K.K., Chen, J.R., and Kulhawy,
100(1), pp. 111-121. F.H. (2006) “Characterization of Model
Uncertainties for Auger Cast-in-Place (ACIP)
6. Kulhawy, F.H. (2004) “On the Axial Behavior
Piles under Axial Compression”, Foundation
of Drilled Foundations”, Geo-Support 2004:
Analysis & Design: Inn. Meth. GSP No. 153,
Drilled Shafts, Micropiling, Deep Mixing,
ASCE, pp. 82–89.
Remedial Methods, and Specialty Foundation
Systems, GSP No. 124, ASCE, Reston, VA, 18 15. Phoon, K.K. and Kulhawy, F.H. (2008)
pp. “Serviceability Limit State Reliability-
based Design”, Reliability-Based Design in
7. Kulhawy, F.H. and Chen, J.R. (2005) “Axial
Geotechnical Engineering – Comp. & App.,
Compression Behavior of Augered Cast-in-
London, UK: pp. 344-384.
Place (ACIP) Piles in Cohesionless Soils”,
Advances in Designing & Testing Deep 16. Skempton, A.W. and McDonald, D.H. (1956)
Foundations, ASCE, pp. 275-289 “The Allowable Settlement of Buildings”,
Proceedings, Institute of Civil Engineering,
8. Mandolini, A., Ramondini, M., Russo, G., and
Part 3(5), pp. 727-768.
Viggiani, C. (2002) “Full Scale Loading Tests
on Instrumented CFA Piles”, Proceedings, 17. Stuedlein, A.W., Neely, W.J., and Gurtowski,
Deep Foundations 2002, GSP No. 116, pp. T.M. (2012) “Reliability-Based Design of
1088-1097. Augered Cast-in-Place Piles in Granular
Soils”, Journal of Geotechnical &
9. McCarthy, D.J. (2008) “Empirical
Geoenvironmental Engineering, 138(6), pp.
Relationships Between Load Test Data
709-717.
and Predicted Compression Capacity of
Augered Cast-In-Place Piles In Predominately 18. Uzielli, M. and Mayne, P.W. (2011) “Statistical
Cohesionless Soils”, M.S. Thesis, University Characterization and Stochastic Simulation
of Central Florida, Orlando, 209 pp. of Load-displacement Behavior of Shallow
Footings”, Geotechnical Risk Assessment &
10. O’Neill, M.W., Vipulanandan, C., Ata, A.,
Management, GSP 224, ASCE, pp. 672-679.
and Tan, F. (1999) “Axial Performance of
Continuous Flight Auger Piles for Bearing”, 19. Zhang, L.M. and Ng, A.M.Y. (2005)
Project Report No. 7-3940-2, Texas “Probabilistic Limiting Tolerable
Department of Transportation, 253 pp. Displacements for Serviceability Limit State
Design of Foundations”, Geotechnique, 55(2),
11. Park, S., Roberts, L.A., and Misra, A.
pp. 151-161.
(2012) “Design Methodology for Axially
Loaded Auger Cast-in-Place (ACIP) and 20. Zhang, L.M. and Phoon, K.K. (2006)
Drilled Displacement (DD) Piles”, Journal “Serviceability Considerations in Reliability-
of Geotechnical & Geoenvironmental based Design”, Foundation. Analysis &
Engineering, 138(12), pp. 1431-1441. Design: Inn. Meth. GSP No. 153, pp. 127–136.

DFI JOURNAL Vol. 7 No. 2 December 2013 [57]

 A Review of the Design Formulations for Static
Axial Response of Deep Foundations from CPT Data
(DFI 2013 Student Paper Competition Runner-Up)
Fawad S. Niazi, Graduate Research Assistant, School of Civil and Environmental Engineering,
Georgia Institute of Technology, Atlanta, Georgia, USA, [email protected]
Research Advisor: Dr. Paul W. Mayne, School of Civil and Environmental Engineering, Georgia
Institute of Technology, Atlanta, Georgia, USA, [email protected]

ABSTRACT
Axial capacity analysis of deep foundations has been a topic of great interest in the soil-structure
interaction problems. Soil behavior is governed by a series of complex stress-strain changes that
occur during pile installation and subsequent loading. Owing to the difficulties and uncertainties on
the basis of the soil strength-deformation characteristics, one of the most frequently followed design
practice is to refer to the formulae correlating directly the pile axial capacity components of unit base
resistance (qb) and unit shaft resistance (fp) to the data collected from cone penetration test (CPT). The
elementary basis for such formulations has been the idea of considering cone penetrometer as a mini-
pile foundation. This has resulted in plethora of correlative relationships in the past over 60 years.
Such correlations, although empirical, have been worked out on the basis of load test results from
both instrumented and un-instrumented full scale piles and are able to accommodate many important
variables. A quick review of the evolution process and development of such design formulations
is presented. An existing method is refined and modified to bring more convenience in extended
applications. Few recommendations are proposed for future research directions, where the latest
version of CPT i.e., seismic piezocone test (SCPTu) can be used to advance from capacity singularity to
the complete axial pile load – displacement (Q – w) response.

INTRODUCTION Thorough geotechnical site investigations have

With the advancement of engineering become far more vital prerequisite to making
knowledge, considerable improvements have learned decisions for selection of reliable and
been made in the piling industry in terms of economical designs and choice of construction
materials, installation methods, and the sizes methods. Conventional investigations
of these foundations. In the contemporary (e.g., boring and sampling) for subsurface
history of deep foundations technology, the characterization of soil layers affecting the
pace of change has been particularly rapid. performance of deep foundations are time
Until recently it was possible to categorize consuming, expensive, and tedious. Laboratory
deep foundations according to their methods tests are conducted on disturbed samples,
of installation as driven piles or drilled shafts. or costly undisturbed samples obtained
This, however, does not satisfactorily cope from selected depths. The results are highly
with the many different forms of piles now dependent on the tools employed in samples’
in use. The newer pile designs are dictated retrieval and skills and expertise of individuals
by such factors as economy, durability, time assigned to the task. In-situ geotechnical
schedules, and heavier loads due to the size/ investigation tools offer quick and economical
scale of the modern state-of-the-art projects alternatives. Their measurements can be
including multi-span bridges, high-rise towers, employed in a wide variety of interpretive
and offshore structures. Fig. 1 shows common schemes and simulation models to match with
applications of deep foundations in the modern the full-scale axial pile response (see Fig. 2).
times and the basic concept of pile-to-ground In the realm of in-situ geotechnical methods,
load transfer. cone penetration testing (CPT) stands out as

[58] DFI JOURNAL Vol. 7 No. 2 December 2013

[FIG. 1] Examples of modern applications of pile foundations and the load transfer concept (Niazi, 2014)

a modern, expedient, economical and reliable pressure (u1) and/or shoulder porewater
means of obtaining detailed subsurface profiles. pressure (u2)] at every 10 to 50 mm (0.4 to
In CPT, an electronic steel probe is hydraulically 2.0 in) depth interval from a single vertical
pushed into the ground to collect multiple sounding. Here, qt = qc + (1 – an)u2, where an =
continuous readings throughout the depth net area ratio of the particular penetrometer
of investigation in a much shorter period of determined through calibration in a triaxial
time. The data can be simultaneously logged chamber. Fig. 3 shows a schematic illustration
and post-processed in a field computer to of implementing SCPTu.
evaluate the geostratigraphy and engineering
parameters of the geomaterials on-site, thereby PILE CAPACITY FROM CPT DATA
offering quick and preliminary conclusions Ever since the first use of CPT in geotechnical
for final design parameters and analysis. The investigations, research efforts have advanced
seismic piezocone penetration test (SCPTu), a the very elementary idea of considering it
newer version of CPT, is a hybrid geotechnical- as mini-pile. This has resulted in plethora of
geophysical in-situ device. It provides downhole correlations between CPT readings and pile
measurements of shear wave velocity (Vs) at capacity components of unit shaft resistance
every 1-m (3.3 ft) depth interval in addition to (fp) and unit base resistance (qb) on the basis
the penetration test parameters [tip stress (qc) of load test results from instrumented and un-
or more proper corrected tip stress (qt), sleeve instrumented piles. These correlations are able
friction (fs), and tip or mid-face porewater to accommodate many important variables.

DFI JOURNAL Vol. 7 No. 2 December 2013 [59]

 [FIG. 2] Alternatives to interpret axial pile response from in-situ geotechnical investigations (Niazi, 2014)

There are two main approaches to accomplish su via adhesion factor (α), or effective stress
axial pile capacity analysis from CPT data: (a) analysis (β-method) for sands and clays that
indirect (or rational) methods, and (b) direct relates effective overburden (σvo') via empirical
methods. The rational methods require a two- parameter β. The fundamental formulations of
step approach. As a first step, CPT data are the two approaches for fp are shown below:
used to provide assessments of stress history
[overconsolidation ratio (OCR)], in-situ radial α-Method: fp ≈ α ∙su [1]
stress coefficient (Ko), undrained shear strength
β-Method: fp ≈ β ∙σvo' ≈ Ko∙σvo'∙tanδ [2]
(su), relative density (Dr), effective stress
strength (φ'), total unit weight (γt), fundamental For pile foundations, an important factor of
soil stiffness [intial shear modulus (Gmax), or relevance to qb is the likely strain compatibility
initial Young’s Modulus (Emax)], interface friction differences between the unmatched
between soil and pile material (δ), and bearing mobilization of fp and qb components during
capacity coefficients (Nc, Nq). Some of the pile loading. For undrained loading (in clays and
pertinent relationships have previously been cohesive silts) beneath the base, qb can fully
summarized by Niazi and Mayne (2010), and mobilize within tolerable limits of settlements,
Niazi et el. (2010). Utilizing these input values usually taken as w/d = 0.10, where w = pile
of geoparameters, the second step enables settlement, and d = base diameter. In the case
assessments of fp and qb components of pile of drained loading (in sands and granular
capacity within a selected analytical framework. materials) it is impractical to assume full
The pile fp can be evaluated using total stress mobilization of end bearing resistance for the
analysis (α-method) for clays that relates fp to range of tolerable settlements.

[60] DFI JOURNAL Vol. 7 No. 2 December 2013

 used besides the direct use
of CPT readings. Herein,
such semi-empirical direct
methods have been grouped
with the direct methods.

EVOLUTION OF CPT-
BASED DIRECT PILE
DESIGN METHODS
According to Begemann
(1969), the earliest research
on the use of CPT-pile
relationship was focused
on estimating the driving
depth of piles. Plantema
(1948) concluded simple
one-to-one correspondence
[FIG. 3] A conceptual scheme of acquiring continuous between qb and qc for
multiple SCPTu readings (Niazi, 2014) jacked square concrete
To achieve a settlement ratio corresponding to piles in dense sand. Huizinga (1951) introduced
w/d = 0.10, it is customary to use an operational the concept of computing qb from average qc
value of qb, reduced from the theoretical value in the soil zone influenced by pile base, and
[qb = 0.1qb(theory)]. In sands and slow loading noted that the total shaft friction (Qs) was
(long-term analysis) in clays and silts, drained about twice the value computed by assuming
conditions are assumed (use bearing capacity fp = fs. Meyerhof (1951) identified the scale
coefficient for overburden, Nq, corresponding to effect on qb vs. qc due to the difference in
effective stress analysis). In clays, silts, and soils pile and cone diameters. Van der Veen and
with low permeability (assume φ' = 0 for rapid Boersma (1957) further explored this scale
loading), undrained conditions are evaluated effect relationship on concrete piles and CPT
(use bearing capacity term for cohesion, Nc soundings, concluding that qb equals average qc
corresponding to total stress analysis). Here qb over the influence zone (L – 3.75d to L + 1.0d,
is evaluated using relevant coefficients to relate where L and d are pile length and diameter,
to su or σvo' (qb ≈ Nc∙su for undrained loading, respectively). Menzenbach (1961) identified the
and qb ≈ 0.1Nq∙σvo' for drained loading). Nc ≈ 9 type and strength of soil as additional factors
for deep foundations, while Nq is function of φ'. affecting the relationship between qb and qc.
Doherty and Gavin (2011), Jamiolkowski (2003), During the following six decades CPT-based
Patrizi and Burland (2001), and Karlsrud (2012) direct pile design methods have considerably
have effectively reviewed significant relevant evolved. Various researchers extended the
contributions. scope of investigations in their respective
In this paper, no detailed discussions are efforts to make improvements in the predictive
included on the indirect methods. Yet, a reliability of these methods. They also extended
summary list of the various factors considered such design applications to larger varieties of
by different researchers in their respective piles and soils by evaluating the influence of
studies for their CPT-based rational evaluations following factors, parameters and variables
has been included (see Tables 1 and 2). (also see Table 3, where applicable piles and
soils are also listed):
The direct CPT methods use measured
penetrometer readings by scaling algorithms to • Pile embedment through varying strata
directly evaluate fp and qb for full-size pilings. • Pile length and diameter (and slenderness
Fig. 4 presents various paths to evaluate the two ratio, L/d)
components of qb and fp from CPT readings. As • Pile load application procedure [e.g.,
shown in this tree chart, fp and qb can also be slow maintained load test (SMLT), quick
estimated using semi-empirical direct methods, maintained load test (QMLT), constant rate
in which few additional parameters are also of penetration test (CRPT)]

DFI JOURNAL Vol. 7 No. 2 December 2013 [61]

 • Penetrometer-to-pile scale effects • Diameter to wall thickness (d/t) for OE piles
• Pile-to-soil interface friction (δ) • Relative position from pile base (h/r*),
• Sand Dr, Ko and stress history where r* is the modified radius of OE pile

• Thin seams of weak or stiff soil layers, • Different combinations of soil types
leading to extreme values of CPT qc profiles including clays, silts, sands and mixed type
(being non-influential for large piles) • Difference of CPT readings in mechanical vs.
• Definition of influence zone around the pile electrical friction sleeve penetrometers
base for correlating qb with qc (or qt) • Piles in offshore environment
• qc averaging procedures in the influence • Use of excess porewater pressure (u2)
zone readings in pile-CPT correlations
• Uplift (tension) vs. compression capacity • Overburden stress (σvo ), and use of qt(net) = qt
• Pile fp to penetrometer fs correlations – σvo instead of qt

• Pile installation methods: non-displacement • Tapered piles

piles (bored piles, drilled shafts), driven and • Spatial correlation between qc (or qt) values
jacked displacement piles [closed-ended of different soil layers
(CE) steel piles, square or circular precast Of particular interest amongst the direct
concrete piles etc.] and piles with small methods are following design formulations
displacement [driven and jacked open-ended that utilize multiple readings from CPTu (i.e.,
(OE) steel piles, steel H piles etc.] maximum data from cone penetration tests):
• Pile end conditions (OE vs. CE) • Norwegian Geotechnical Institute – Building
• Base plugging for OE piles, separating qb Research Establishment (NGI–BRE) Method
into plug resistance (qplug) and annulus • UniCone Method
resistance (qann) and correlating both with qc
• Kajima Technical Research Institute, Japan
(or qt)
(KTRI) Method

CPT Based Pile Capacity Evaluations

Direct Indirect
Methods Methods

Total Stress Approach Effective Stress Approach

fp via α-Methods: fp via β-Methods:

(fine grained soils) (coarse and fine grained soils)
fctn(su, σvo', OCR, Ip, L, plugging, fctn(σr, δ, φ', OCR, K, σvo', L, d,
progressive failure) su, Dr, St, Ip)

qb for undrained loading qb for drained loading

(fine grained soils): fctn(su) (coarse grained soils and slow loading
in fine grained soils):
fctn(φ', σvo', L, d, Dr)

Pure Empirical Methods: Semi-Empirical Methods:

Evaluate fp and qb directly using Evaluate fp and qb using qc (or qt),
qc (or qt), and/or fs and/or u2 and/or fs with additional geoparameters
(σr, δ, φ', K, σvo', L, d, su, Dr, plugging)

[FIG. 4] Alternative paths for CPT-based evaluations of fp and qb components of pile capacity (adapted from Niazi and
Mayne, 2013)

[62] DFI JOURNAL Vol. 7 No. 2 December 2013

[TABLE 1] Factors considered in the total stress approach (α-methods) for estimating pile unit shaft
resistance (fp)

Method/Reference Length effect Stress history Ip su σvo' φ' Progressive failure Plugging effect
Tomlinson (1957) x x x √ x x x x
Peck (1958) x x x √ x x x x
Skempton (1959) x x x √ x x x x
Woodward et al. (1961) x x x √ x x x x
Kerisel (1965) x x x √ x x x x
API (1969) x √ x x √ x x x
McClelland (1974) x √ x √ x x x x
Vijayvergiya and Focht √ x x √ √ x x x
(1972)
Vesić (1977) x x x √ x x x x
Drewry et al. (1977) x x x √ x x x x
API (1975; 1976) x √ √ √ √ x x x
Kraft et al. (1981) √ x x √ √ x √ x
Dennis and Olson (1983) √ x x √ x x x x
Randolph (1983) √ x x √ x x √ x
Semple and Rigden (1984) √ √ x √ √ x x x
Randolph and Murphy x √ x √ √ x x x
(1985)
API (1987) x √ x √ √ x x x
API (1993) x √ x √ √ x x x
Karlsrud et al. (1993) x x √ √ √ x x x
Chen and Kulhawy (1994) x x x √ x x x x
Kolk and van der Velde √ √ x √ √ x x x
(1996)
Miller and Lutenegger x x x x x x x √
(1997)
O'Neill and Reese (1999) x x x √ x x x x
Jamiolkowski (2003) x √ x √ √ x x x
Goh et al. (2005) x √ x √ √ x x x
NGI-05 (Karlsrud et al., x √ √ √ √ x x x
2005)
Salgado (2006; 2008; x √ x √ x x x x
2010)
German Method x x x √ x x x
(Kempfert and Becker,
2010)
Karlsrud (2012) x √ √ √ √ x x x
Chakraborty et al. (2013) x x x √ √ √ x x

Notes: Ip = plasticity index; su = undrained shear strength; σvo' = effective overburden stress;
φ' = friction angle; √ = factor considered in the design formulation; x = factor not considered in the
design formulation.

DFI JOURNAL Vol. 7 No. 2 December 2013 [63]

 [TABLE 2] Factors considered in the effective stress approach (β-method) for estimating pile unit shaft
resistance (fp)

Method/Reference σr' δ φ' OCR K σvo' L d su Dr St Ip

Chandler (1968); Burland (1973); √ √ √ x √ √ x x x x x x
Pelletier and Doyle (1982)
Meyerhof (1976) x √ √ √ x √ x x x x x x
Flaate and Selnes (1977) x x x √ x √ √ x x x x x
Coyle and Castello (1981) x x √ √ x x √ √ x x x x
Kulhawy et al. (1983) √ √ √ x √ √ x x x x x x
Twine (1987); Patel (1989) x x x x x √ x x x x x x
Reese and O'Neill (1988) x x x x x √ √ x x x x x
Fleming et al. (1992) √ √ x x √ √ x x x x x x
Burland (1993) x x x √ x √ x x √ x x x
de Nicola and Randolph (1999) x x x x x √ x x x √ x x
Patrizi and Burland (2001) x x x √ x √ x x √ x x x
ICP-05 Method (Jardine et al., 2005) √ √ x √ √ x √ √ √ x √ √
Karlsrud (2012) x x x √ x √ x x x x x √

Notes: σr' = radial effective stress; δ = soil-pile interface friction angle; φ' = effective friction angle; OCR
(= σp/σvo') = overconsolidation ratio; σp = preconsolidation stress; K (= σr'/σv') = radial effective stress
coefficient; σvo' = effective vertical overburden stress; L = pile length; d = pile diameter; su = undrained
shear strength of clay; Dr = relative density of sand; St = sensitivity of clayey soils; Ip = plasticity index;
√ = factor considered in the design formulation; x = factor not considered in the design formulation.

• Fugro V-K Method A soil classification chart is used to select

• UWA'13 Method relevant values of Cse (see Fig. 5), making it
a two-step method. The classification chart
The respective design equations of these five
presents five soil zones delineated by sharp
methods are presented in Table 4. Some of
boundaries. The recommended Cse values
these apply to specific pile and soil types, while
show abrupt variations in any two adjacent
others were based on wider varieties of piles
zones, rather than gradual transitional
installed in different geomaterials. Selected
values for intermediate soil types.
observations specific to each method are
appended below: • KTRI method was derived by comparing CPT
fs with pile fp over a range of applicable ∆u2
• NGI-BRE method applies to driven and
of a small local database. Here, it may be
jacked piles in clays. Majority of piles
noted that fs is the weakest CPTu readings.
analyzed were tested in tension (fp is
The most consistent qt reading was not
the main component of pile capacity).
used.
Correlation for qb was based on fewer
compression tests. The fs readings of CPTu • Fugro V-K method was developed from a
were not utilized. database of driven offshore piles in clays,
with limited applicability. Like the NGI-
• UniCone method developed from a large pile
BRE method, fs readings of CPTu were not
and soil database, applies to a wide variety
explored. For over one-third of the database,
of situations. Eslami and Fellenius (1997)
u2 readings were not available, and were,
noted that the number of cases was limited,
thus, derived from correlations. For almost
and newer field experience should result
50% of the database, an was not available, in
in modifications of the situations. Eslami
which case a value of 0.75 was assumed.
and Fellenius (1997) noted that the number
of cases was limited, and newer field • UWA method, which was developed from 53
experience should result in modifications previous and 22 newer load tests, concerns
of the correlation coefficients (Cse and Cte). driven and jacked piles in clayey soils only,

[64] DFI JOURNAL Vol. 7 No. 2 December 2013

[TABLE 3] Summary of the CPT based direct pile design methods

CPT readings, pile information and soil parameters used in the design
equations
Method/reference Pile unit shaft resistance (fp) Pile unit end bearing (qb)
Bogdanovic (1961) fs, d, and dCPT qca(tip)
(driven and jacked concrete piles in
dense sand)
Begemann (1963; 1965; 1969) qc, fs and pile type qca(tip)
(driven piles in sand)
Meyerhof (1976; 1983) fs and pile type, or alternatively qc qca(tip), d, zd, and pile and soil types
(driven piles and drilled shafts in and pile type
sand)
Aoki and Velloso (1975) qca(side), and pile and soil types qca(tip), and pile type
(piles in all soils)
Nottingham (1975); fs, pile and soil types, penetrometer qca(tip), and soil and penetrometer
Schmertmann (1978) types, and relative depth (z/d) types
(driven concrete, steel, timber piles
and drilled shafts in all soils)
Penpile Method fs qca(tip), and soil type
(Clisby et al., 1978)
(piles in all soils)
Dutch Method qca(side), soil types, OCR, and loading qca(tip), and soil type
(de Ruiter and Beringen 1979) direction
(offshore piles in all soils)
Philipponnat (1980) qca(side), and pile and soil types qca(tip), and soil type
(for all pile types in all soil types)
LCPC or French Method qca(side), pile and soil types, and qca(tip), and pile and soil types
(Bustamante and Gianeselli, 1982; installation procedure
Bustamante and Frank, 1997)
(all piles in all soils)
Cone-m Method fsa qca(tip)
(Tumay and Fakhroo, 1982)
(piles in clay)
Price and Wardle (1982) fs, and pile type qca(tip), and pile type
(driven and jacked piles, and drilled
shafts in stiff clay)
Gwizdala (1984) qca(side), and soil type qca(tip), d, and pile type
(drilled shafts in sand)
Kulhawy and Phoon (1993); Kulhawy qca(net), and σatm qca(net)
(2004); Lunne et al. (1997)
(drilled shafts in clay)
Alsamman (1995) qca(side), and soil and penetrometer qca(tip), and soil and penetrometer
(drilled shafts in all soils) types types
NGI-BRE Method qt(net), and σvo' qt(net), clay type
(Almeida et al., 1996; Powell et al.,
2001; Powell and Quarterman, 1988)
(driven and jacked piles in clay)
Politecnico di Torino Method qca(side) qca(tip), wt, d, and soil type
(Fioravante, 1994; Fioravante et al.,
1995) (drilled shafts in sands)
Lee and Salgado (1999) This method does not indicate a qta(tip), wb, and d
(piles in sands) means for evaluating fp
UniCone Method qt(side), u2(side), and soil classification qt(tip), u2(tip), and d
(Eslami and Fellenius, 1997; based on qt(side), u2(side) and fs
Fellenius, 2002)
(all piles in all soils)

DFI JOURNAL Vol. 7 No. 2 December 2013 [65]

 CPT readings, pile information and soil parameters used in the design
equations
Method/reference Pile unit shaft resistance (fp) Pile unit end bearing (qb)
KTRI Method fs and ∆u2(side) This method does not indicate a
(Takesue et al., 1998) means for evaluating qb
(all piles in all soils)
UWA-99 Method This method utilizes effective stress qca(tip), d, wb, σvo', pile end conditions,
(De Nicola and Randolph, 1999) approach as a means for evaluating and sand plugging
(driven pipe piles in medium - dense fp
sand)
TCD-01 Method This method does not indicate a qca(tip), pile end conditions, and sand
(Lehane and Gavin, 2001) means for evaluating fp plugging
(jacked pipe piles in loose sand;
measured IFR)
TCD-03 Method qc(side), d, t, σrc', δ, σvo', qc(tip), loading qca(tip), d, t, σvo', and sand plugging
(Gavin and Lehane, 2003; de Nicola direction, sand plugging
and Randolph, 1999)
(OE jacked pipe piles in loose sand;
measured IFR)
Fugro-05 Method qc(side), d, t, σvo', loading direction, qca(tip), d, t, wb
(Kolk et al., 2005) and pile shape
(driven piles in sands, mostly
offshore piling)
UCD-05 Method This method does not indicate a qca(tip), t, pile end conditions, and
(Gavin and Lehane, 2005) means for evaluating fp sand plugging
(driven and jacked pipe piles in
dense sand)
ICP-05 Method qc(side), d, t, σr', δ, σvo', loading qca(tip), d, t, dCPT, Dr, pile end
(Jardine et al., 2005) direction, and pile end conditions conditions, and sand plugging
(driven piles loaded first time via
SML test around 10 days after
driving in sand)
ICP-05 Method This method utilizes effective stress qca(tip), d, t, dCPT, pile loading rate and
(Jardine et al., 2005) approach as a means for evaluating end conditions, and soil plugging
(driven piles in clay) fp
UWA-05 Method qc(side), d, σr', δ, σvo', loading direction, qca(tip), d, t, and sand plugging
(Lehane et al., 2005) and sand plugging
(driven piles in sand)
NGI-05 Method qc(side), Dr, σvo', loading direction, and qca(tip), Dr, σvo', L, d, t, pile end
(Clausen et al., 2005) pile material and end conditions conditions, and sand plugging
(driven piles in sand)
Cambridge-05 Method This method does not indicate a qca(tip), L, d, and zd
(White and Bolton, 2005) means for evaluating fp
(CE pipe piles, Franki piles, PCC
piles jacked or driven through soft
into hard sand layer)
Togliani (2008) qc(side), L, d, and pile type and shape qca(tip), L, dtip, and pile type
(cylindrical and tapered driven piles
and drilled shafts in all soils)
German Method qc(side) qca(tip)
(Kempfert and Becker, 2010) (piles
in sand)
UCD-11 Method qc(side), σr', δ, d, t, σvo', and sand This method does not indicate a
(Igoe et al., 2010; 2011) plugging and density means for evaluating qb
(OE piles in sand)
Fugro V-K Method qt(net), and depth qt(net)
(Van Dijk and Kolk, 2011)
(offshore piles in clay)

[66] DFI JOURNAL Vol. 7 No. 2 December 2013

 CPT readings, pile information and soil parameters used in the design
equations
Method/reference Pile unit shaft resistance (fp) Pile unit end bearing (qb)
SEU Method fs, and ∆u2 Similar to Unicone Method
(Cai et al., 2011; 2012)
(driven or jacked PCC thin-wall
high-strength caissons, cement fly-
ash grave pile in clay)
HKU Method This method does not indicate a qca(tip), soil plugging
(Yu and Yang, 2012) means for evaluating fp
(OE steel pipe piles in sand)
UWA-13 Method qt, d, t, and depth This method does not indicate a
(Lehane et al., 2013) means for evaluating qb
(for shaft capacity of driven and
jacked piles in clays)

Notes: d = pile diameter; dCPT = penetrometer diameter; qca and qta = averages of qc and qt values
(respectively) in the applicable zone/soil layers that depend on the method; zd = embedment depth
in dense sand layer (in m); z = depth below ground surface; OCR = overconsolidation ratio; fsa = Ft/L
= average layer friction, Ft = total sleeve friction determined for pile penetration length (L) in the
layer; qc(net) = qc – σvo; qt(net) = qt – σvo; σvo' = effective vertical overburden stress = σvo – uo; σvo = vertical
overburden stress; wt and wb are displacements at pile head and base, respectively; u2 = porewater
pressure measured at cone shoulder; ∆u2 = excess pore water pressure = u2 – uo; uo = hydrostatic
pore water pressure; σr' = radial effective stress; δ = pile-soil interface friction angle; t = pile wall
thickness; Dr = relative density; OE = open-ended; CE = close ended; SML = slow maintained load; IFR =
incremental filling ratio.

[TABLE 4] Design formulations of CPTu based direct pile design methods

Method/Reference Design Equations

Pile unit shaft resistance (fp) Pile unit end bearing (qb)
NGI-BRE Method fp = qt(net)/k1 qb = qt(net),avg/k2
(Almeida et al., 1996; Powell et al., k1 = 10.5 + 13.3log[qt(net)/σvo'] k2 = Nkt/9; Nkt = 15 (soft – firm intact
2001) clays), 25 to 35 (fissured to hard
(for driven/jacked piles in clays) clays)

UniCone Method fp = Cse∙qE qb = Cte∙qEg

(Eslami and Fellenius, 1997; qE = qt – u2 Cte is generally taken as 1; for pile
Fellenius, 2002) (see Fig. 5 for Cse) diameter d > 0.4 m, Cte = 1/(3d)
(for all pile and soil types)
KTRI Method For ∆u2 > 300 kPa: fp /fs = ∆u2/200 This method does not indicate a
(Takesue et al., 1998) – 0.5 means for evaluating qb

(for all pile and soil types) For ∆u2 < 300 kPa: fp /fs = ∆u2/1250
+ 0.76
Fugro V-K Method fp = ks(z)∙qt(net),z qb = 0.7qt(net),avg
(Van Dijk and Kolk, 2011) ks(z) = 0.16(h/uL)–0.3[Qt(z)]–0.4 < 0.08
(for offshore driven piles in clays) uL = 1.0 m (=3.3 feet)
Qt(z) = qt(net),z/σvo' at z

UWA Method fp = 0.055 qt [max(h/r*, 1)]–0.2 This method does not indicate a
(Lehane et al. 2013) means for evaluating qb

(for driven/jacked piles in clays)

Notes: qt(net) = qt – σvo; σvo = vertical overburden stress; σvo' = effective vertical overburden stress = σvo –
uo; uo = hydrostatic pore water pressure; qt(net),avg = qt(net) averaged + 1.5d over pile toe level; Nkt = cone
factor; Cse = shaft correlation coefficient from soil classification chart (from qt; fs and u2); Cte = toe

DFI JOURNAL Vol. 7 No. 2 December 2013 [67]

 correlation coefficient; qEg is the geometric average of qE values over the influence zone (from 4d below
pile toe to 8d above pile toe if pile is installed from weak soil into dense soil, and from 4d below pile
toe to 2d above pile toe when pile is installed from dense soil into weak soil) after correction for u2 and
adjustment to σvo'; d = pile diameter; ∆u2 = excess pore water pressure = u2 – uo; z = depth below the
surface; h = distance between pile toe level and z; uL = unit length to render expression dimensionless;
r* = modified radius term for OE piles = (r2 – ri2)0.5; r = external radius; ri = internal radius.

and therefore, it has limited applicability. besides simple reliance on CPT readings.
This method does not indicate a means for Certain methods falling in the category of semi-
evaluating qb. empirical methods take the basic concept from
either total or effective stress approach, while
A REFLECTION ON CPT-BASED also establishing direct correlations with CPT
DIRECT METHODS FOR AXIAL PILE readings.
DESIGN From the list of CPT-based direct methods, 26
In order to select and implement any suitable methods provide estimates for both fp and qb,
CPT-based direct method detailed above, while the remaining 9 methods account for
following factors warrant deliberations: either of the two. Most of the direct methods
draw purely on cone tip stress readings. Only
Measured vs. Corrected Cone Tip Resistance a few rely on multiple CPTu parameters. Fig.
The methods that utilize tip stress do not 6 provides an overview of the dependence of
commonly account for correction of u2 acting these methods on various combinations of CPT
on unequal tip area of the cone to obtain qt. In readings and additional measurements.
clean sands and dense granular soils, it may be
Soil Influence Zone at Pile Tip
reasonable to assume qt ≈ qc because u2 remains
nearly hydrostatic (uo). In soft to stiff intact Direct methods relate unit base resistance
clays and silts considerable Du2 are generated to the cone tip resistance (qc or qt). Here, the
during cone penetration, warranting significant penetrometer readings are averaged over the
corrections to the measured qc in order to depth interval of "influence zone" around the
obtain qt (Mayne, 2007). pile base (or toe). Different methods provide
different definitions of the influence zone to
100 account for the rupture path around to the pile
Zone No. Soil Type Cse
1 Soft sensitive clay 0.08 5 toe. Various considerations that have led to
2 Soft clay and silt 0.05
3 Stiff clay and silt 0.025 these recommendations include: (1) the trend
qE = qt - u2 (MPa)

4
10
4
5
Silty sandy mix
Sand
0.01
0.004
of cone tip resistance values around the pile
3
toe, (2) extent of soil variability around the pile
toe, (3) pile diameter, (4) pile embedment depth
into the bearing stratum, (5) existence of weak
1 2 layer beneath the bearing layer, and (6) soil
1
fp = Cse qE compressibility.
where Cse = shaft correlation coefficient
0.1 Averaging Procedure for qc
1 10 100 1000
Sleeve Friction, fs (kPa)
The CPT readings typically display squiggly
profiles of random peaks and troughs
[FIG. 5] UniCone chart for zone numbers and soil types representing thin seams of variable soils.
(after Eslami and Fellenius, 1997)
For large piles, inclusion of these readings in
Dependence on CPT Data and Additional averaging can result in non-representative qb
Measurements values. Readings may be filtered out applying
a "minimum path" rule (e.g., Begemann,
From within the two main groups of CPT-
1963), or by simply removing the peaks and
based direct methods, for pure-empirical
troughs from the records (e.g., Bustamante and
category, simple direct relationships have been
Gianeselli, 1982). Eslami and Fellenius (1997)
suggested between CPT readings and fp and/or
recommended use of geometric average, as it
qb components of pile capacity. Semi-empirical
results in filtered representation of readings.
methods require additional parameters,

[68] DFI JOURNAL Vol. 7 No. 2 December 2013

 provided detailed recommendations for Ab
calculations, based on soil type, estimates of
plugging and pile loading rate.

Capacity Criteria Definition

A single value of axial load on the entire
load-displacement (Q – w) response curves is
selected for design load purposes, commonly
termed as "pile capacity." There are at least 45
different criteria defining the axial capacity. The
capacities interpreted via different criteria span
over a wide range (e.g., Niazi, 2011). Mostly,
CPT-based methods do not explicitly refer to
[FIG. 6] Dependence of different CPT-based direct the specific capacity criterion on the basis of
methods on combinations of CPT readings and additional which the design equations were formulated,
parameters (adapted from Niazi and Mayne, 2013)
thus adding to the degree of uncertainty.
Without reference to any specific criterion, the
Shaft and Base Areas for Pile Capacity
Calculations engineer in-charge of piling project has to rely
on subjective judgment and apply Factors of
The pile Qs and Qb calculations involve pile shaft Safety to strike a balance between economical
area (As) and base area (Ab), respectively. As is and safe design.
simply given as the product of pile perimeter
and length. For a circular pile, As = π∙d∙L. Practicality of CPT-based Methods
For square and rectangular piles, equivalent
Most of the research on pile-CPT correlations
diameter can be adopted: d = (2b + 2w)/ π,
has addressed driven piles in sands. Drilled
where b and w are the depth and width of pile
shafts in clays and silts have attracted lesser
cross-section, respectively. For OE piles, soil
attention. Large amounts of latest database
plug forms through the base during driving. It
from the modern bi-directional Osterberg-cell
provides additional internal shaft resistance
(O-cell) type of proofing method have also
during loading, commonly considered as part
been included in the inventory of load tests.
of base resistance. For H piles, As contributing
Existing CPT-based methods focus purely on
to shaft capacity depends on the degree
"pile capacity," without recourse to the behavior
of soil that enters the flanges and remains
in terms of Q – w response. In particular, the Vs
attached during loading. Seo et al. (2009)
component of the newer SCPTu can provide
recommended two options for As calculations
Gmax = (γt/g)∙Vs2, where g = gravitational
for H piles: (1) use average reduction factor
acceleration. Gmax can be utilized within an
of 0.65 when assuming full soil-pile interface
analytical framework (e.g., Randolph and Wroth,
contact perimeter; (2) assume outer perimeter
1978) towards extension of pile design to
in shaft capacity calculations to get comparable
"complete axial Q – w response." An appropriate
estimates.
modulus reduction scheme [e.g., G/Gmax vs.
For circular solid or CE pipe piles Ab is obtained Q/Qmax, or G/Gmax vs. w/d (i.e., pseudo-strain)]
as π∙d2/4. For non-circular solid piles of square may be adopted to account for non-linear
or rectangular cross-sections, equivalent base response of settlements (w) corresponding
diameter is obtained from: d = (4 b∙w/π)0.5. For to increasing loads. A schematic illustration
OE pipe piles, following the recommendations of implementing such a model is shown in
by Gavin and Lehane (2005), and Yu and Yang Fig. 7. The results of one such application has
(2012), separate contributions from annulus previously been shown in the bottom right
and plug should be considered: portion of Fig. 2.

Qb = (π/4)∙[d2 qplug + (d2 – di2) qann] [3] AN UNDATED DESIGN METHOD

where qplug and qann are unit plug resistance In a most recent effort, a much larger database,
and unit annulus resistance of pile; d = pile including the latest cases of pile load tests and
outer diameter; di = pile inner diameter. For H CPTu soundings was collected (Niazi, 2014).
piles, Jardine et al. (2005) and Seo et al. (2009) The aim was to utilize all 3 CPTu readings to

DFI JOURNAL Vol. 7 No. 2 December 2013 [69]

 refine the formulation of an existing method
and to make it more convenient, and applicable
to a larger variety of situations. This database
includes 153 pile load tests from 52 sites at
worldwide locations. The piles include drilled
shafts, continuous flight auger (CFA) piles,
driven and jacked OE and CE steel pipe piles,
H-section steel piles, and square as well as
circular precast concrete piles installed in
variety of geomaterials. The pile L/d ratios of
the database range from 3.9 to 142.0, while the
load carrying capacities range from 10 kN to
75,500 kN (1.12 ton to 8487 ton).
From the penetrometer database at 52 sites,
46 sites provided CPTu data (i.e., penetrometer
readings including porewater pressure data,
u2 or u1, in addition to qt and fs), while for the
remaining 6 sites both qc and fs readings were
present. In all these 6 cases, u2 readings were
assumed hydrostatic (i.e., u2 ≈ uo) because of
the sandy soil profiles being encountered.
Additionally, at 6 of the 46 sites with CPTu [FIG. 7] Randolph analytical elastic model for evaluating
pile load-displacement response (adapted from Niazi,
soundings, u1 readings were measured instead 2011)
of u2, where following correlations were used
for appropriate conversions:
The pile capacity components of fp for
Chen and Kulhawy (1994): u2 = 0.742 (u1) [4] respective layers through the pile embedded
length and qb were calculated from the peak
Peuchen et al. (2010): u2 = K (u1 – uo) + uo [5a] applied load for each test pile. In addition,
from the Q – w curve of each case, the axial
where pile capacity was also calculated based on 3
K = 0.91 exp(– 0.09 Qtn0.47) {1/[1 + Fr (0.17 different definitions: (1) Davisson's offset line
+ 0.061 (Qtn – 21.6)1/3)] – exp(–2 Fr)} [5b] criterion (Davisson 1972), (2) French criterion:
w/d = 10% (Vesić 1977), and (3) Hyperbolic
In calculations of K from Eqn. [5b], the asymptote criterion by Chin-Kondner (Chin
parameters Qtn and Fr relate to the CPT-based 1970; Kondner 1963). The goal was to associate
soil behavior type (SBT) classification index, Ic. these capacity criteria to the measured peak
Robertson (2009) presented SBT boundaries on values of applied loads in the database, and so
Qtn–Fr chart from the contours of Ic, leading to to the correlations that are developed based on
the following: the measured peak values.
For non-circular solid piles, equivalent
Ic = [(3.47 – logQtn)2 + (logFr + 1.22)2]0.5 [6a]
pile diameters were adopted as per the
Qtn = [(qt – σvo)/σatm] (σatm/σ'vo)n [6b] recommendations given in the preceding section.

Fr = [fs/(qt – σvo)] 100% [6c] Shaft Correlation Coefficient

As a first step, the corresponding sets of mean
n = 0.381 (Ic) + 0.05 (σ'vo/σatm) – 0.15 [6d]
values were plotted on the log-log qE vs. fs type
where σatm is a reference stress = 100 kPa (14.5 of soil classification diagram. Accordingly,
psi). new sets of boundaries and sub-boundaries
were delineated on the UniCone type chart,
The UniCone method was selected for
thus proposing 11 soil sub-zones, in contrast
refinement because it accommodates larger
to the 5 originally presented by Eslami and
variety of piles and soils. The influence zone
Fellenius (1997) (see Fig. 8). With fp and qb
and averaging technique were adopted as
obtained from the load test results, and their
recommended in this method.

[70] DFI JOURNAL Vol. 7 No. 2 December 2013

corresponding averaged piezocone readings Log[Cse(mean)] = 0.732 (Ic) – 3.605 [7]
available for different soil layers of each site,
correlations were established between the Cse(c) = 1.11 [Cse(mean)] [8]
two sets of the data. This followed uniting the Cse(t) = 0.85 [Cse(mean)] [9]
correlation experience from the entire database
to draw conclusive results based on the soil Reverse-transformation of the Cse values from
typology. The shaft correlation coefficients Eqn. [7] gives reasonable estimates of Cse. In
(i.e., Cse = fp/qE) and their relevant statistics use of this correlation, attention must be paid
so obtained are shown in Fig. 8. These values, to the applicable range of Ic from the database
which present gradual transition, are the results shown in Fig. 9. The following two alternative
of calibration experience from the extended correlation functions were also developed:
database of this study. Hyperbolic Tangent Function:
For further simplification, the soil behavior type
(SBT) classification index, Ic presented in Eqn. [6] Cse = 0.044 + 0.0416 tanh[1.51 (Ic) – 4.53] [10]
was considered. Thus, log-transformed values
Modified Hyperbolic Function:
of Cse were plotted against Ic, giving a linear
relationship (see Fig. 9, where SBT boundaries Cse = 0.1 – 0.097/[1 + (Ic/3.10)8.2] [11]
are also shown). The data was separated on
the basis of installation methods, the effects of
load application procedure (MLT vs. CRPT), and
the loading modes (compression vs. tension) to
study their influence on the correlations trends.
The most clearly discernible trend pertains
to the loading mode, where it was found that
Cse(t)/Cse(c) ≈ fp(t)/fp(c) ≈ 0.763. Here subscripts
t and c represent tension and compression,
respectively. Following general expression
was generated from the overall correlation
experience:
[FIG. 9] Variation of Log[Cse (= fp/qE)] with CPT SBT Ic

Use of any of the above 3 correlation

expressions provides a direct approach to shaft
capacity estimation from a continuous function,
eliminating the need for use of Fig. 8.

Toe Correlation Coefficient

A similar methodology of back-analysis was
employed for the toe correlation coefficients
(Cte = qb/qE). The resulting trends plotted in
Fig. 10, with a generalized correlation given
in Eqn. [12], apply to limited range of Ic
(i.e., 1.69 > Ic > 3.77).

Log[Cte(mean)] = 0.325 (Ic) – 1.218 [12]

For base coefficient, the influence of additional

factors like pile loading procedures (SMLT vs.
QMLT vs. CRPT), and pile end conditions (CE
vs. OE) were also investigated. Visibly, the data
points plotted in Fig. 10 indicate much greater
scatter in the Cte vs. Ic correlation than that
observed for Cse vs. Ic. This is besides the fact
[FIG. 8] Modified UniCone chart for zone numbers, soil that no clear distinction could be made either
types, and shaft correlation coefficients

DFI JOURNAL Vol. 7 No. 2 December 2013 [71]

 evolved through years of experience. Some
methods are appropriate for wider variety
of situations, while others are relevant to
specific pile and soil types. In application
of these methods, situations similar to the
database used in their formulations may
present expected results, but uncertainty
becomes dominant when dealing with different
structures or new types of geomaterials.
The direct CPTu-pile correlations offer a
quick and convenient method for axial pile
[FIG. 10] Variation of Log[Cte (= qb/qE)] with CPT SBT Ic capacity from multiple penetrometer readings.
Specifically, five CPTu based methods are
due to the installation method, or the pile end discussed, along with respective observations
condition. This study, however, clearly points concerning each. The UniCone method is
to the fact that Cte cannot be simply assumed selected for refinement that uses a 5-part soil
equal to unity, or based solely on pile diameter, classification system to assign pile side friction
as proposed in the original UniCone method. coefficients to each elevation of CPTu data.
Although, Eqn. [12] may provide a reasonable A case is made for extending this application
estimate for preliminary analysis and design, based on new test piles and additional soil
more focused studies on the influence of pile deposits. A database of 153 pile load tests
end conditions and loading procedures, coupled from 52 sites is compiled, representing a
with further information from pile load tests, considerable increase over the original study.
are warranted to further refine the proposed The database includes diversity in terms of pile
formulations. and soil types. The data selection process aimed
at the availability of piezocone type of CPT
Capacity Criteria testing.
The correlations developed in this study were Results of the derived coefficients Cse, and
based on the measured peak loads [Qmax (meas.)]. Cte of the axial pile capacity from CPTu data
For cautious application of the updated design are presented. The UniCone type of log-log
formulations, the Qmax (meas.) from each load test qE vs. fs soil classification chart is refined by
was associated to the pile capacities calculated delineating 11 soil sub-zones along with their
from 3 different criteria in the arrangement of respective shaft coefficients, in contrast to the
ratios: (1) Qcap (Chin-Kondner)/Qmax (meas.), (2) Qcap (w/d = 5 zones originally proposed. Further, the CPT
10%)
/Qmax (meas.), and (3) Qcap (Davisson)/Qmax (meas.). The material index, Ic is used to establish direct
applicable statistics were calculated for each correlations between Cse and Ic, and Cte and Ic.
ratio, as shown in Table 5. Thus, in application The results offer a continuous function for
of the proposed design formulations to estimate estimating the coefficients over a wide value of
the response of a particular pile loading, these Ic, thus disregarding the need for use of the soil
statistical results must be considered. classification chart and improving the reliability
in the evaluations of fp and qb.
CONCLUSIONS The analysis also reveals that the statistical
A quick review of the CPT-based axial pile reliability of estimating fp is much superior to
design formulations was presented. These that for qb.
designs are based largely on empirical methods

[TABLE 5] Statistics of the ratios between capacity definitions and measured peak loads
Mean Standard Variance Skewness Range Minimum Maximum
Deviation
Qcap (Chin-Kondner) /Qmax (meas.) 1.101 0.112 0.012 3.222 0.770 1.000 1.770

Qcap (w/d = 10%) /Qmax (meas.) 0.986 0.084 0.007 -0.979 0.470 0.720 1.190

Qcap (Davisson) /Qmax (meas.) 0.852 0.104 0.011 -0.847 0.520 0.470 0.990

[72] DFI JOURNAL Vol. 7 No. 2 December 2013

RECOMMENDATIONS FOR FUTURE 3. American Petroleum Institute, (1969),
RESEARCH Recommended practice for planning,
designing, and constructing fixed
Different CPT-based methods were derived
offshore platforms. API RP2A, 1st edition,
from diverse datasets of piles installed in
Washington, D.C.
different types of geomaterials. The reliability
of no single method can be regarded as 4. American Petroleum Institute, (1975),
universal. The more recent methods that exploit Recommended practice for planning,
maximum CPT parameters, and which were designing, and constructing fixed
developed from larger and latest database of offshore platforms. API RP2A, 6th edition,
pile load tests may be considered for prediction Washington, D.C.
analysis. However, the results must be checked 5. American Petroleum Institute, (1976),
against the estimates of other recent direct Recommended practice for planning,
methods before finalizing the design. Following designing, and constructing fixed
recommendations are offered towards future offshore platforms. API RP2A, 7th edition,
application and research on the topic: Washington, D.C.
• In design practice, caution must be 6. American Petroleum Institute, (1987),
exercised when applying empirical Recommended practice for planning,
approaches to field situations. Engineering designing, and constructing fixed offshore
judgment must be exercised for platforms. API RP2A, 17th edition,
interpretation of the data. Washington, D.C.
• Latest database of newer piles, top-down 7. American Petroleum Institute, (1993),
compression, top-up tension as well as Recommended practice for planning,
the new O-cell load testing arrangements, designing and constructing fixed offshore
and the modern multi-channel hybrid platforms – working stress design. API
geophysical-geotechnical SCPTu testing RP2A, 20th ed., Washington, D.C.
records should form part of evolution
8. Aoki, N., and Velloso, D. A., (1975), “An
process of pile-CPT correlations.
approximate method to estimate the bearing
• Maximum use of SCPTu readings should be capacity of piles”, Proceedings of 5th Pan-
made for the complete axial pile response American Conference of soil Mechanics and
by utilizing three penetrometer readings Foundation Engineering, Buenos Aires, pp.
(qt, fs, u2) for axial capacity evaluations, and 367–376.
geophysical Vs component for axial load-
9. Begemann, H. K. S. Ph., (1963), “The use of
displacement response within an analytical
the static penetrometer in Holland”, New
framework.
Zealand Engineering, Vol. 18, No. 2, pp. 41.
• Similar to the analysis presented herein, any
10. Begemann, H. K. S. Ph., (1965), “The
future developments in this regard must
maximum pulling force on a single tension
explicitly indicate the measurement basis
pile calculated on the basis of results of
for fp and qb from their datasets, and the
the adhesion jacket cone”, Proceedings of
average relationships thereof with some of
the 6th International Conference on Soil
the more frequently used capacity criteria.
Mechanics and Foundation Engineering,
Montreal, Vol. 2, 229 p.
REFERENCES
11. Begemann, H. K. S. Ph., (1969), “The Dutch
1. Almeida, M.S.S., Danziger, F.A.B., and Lunne,
static penetration test with the adhesion
T., (1996), “Use of the piezocone test to
jacket cone”, Delft soil mechanics and delft
predict the axial capacity of driven and
hydraulics laboratory, XIII(1): pp. 1.
jacked piles in clay”, Canadian Geotechnical
Journal, Vol. 33, No. 1, pp. 33–41. 12. Bogdanovic, L., (1961), “The use of
penetration tests for determining the
2. Alsamman, O. M., (1995), “The use of CPT
bearing capacity of piles”, Proceedings of
for calculating axial capacity of drilled
the 5th International Conference on Soil
shafts”, PhD Thesis, UIUC, IL, 299 p.
Mechanics and Foundation Engineering,
Paris, Vol. 2, pp. 17.

DFI JOURNAL Vol. 7 No. 2 December 2013 [73]

 13. Burland, J. P., (1973), “Shaft friction of piles Vol. I, Final Report, Mississippi State Highway
in clay – a simple fundamental approach”, Department.
Ground Engineering, Vol. 6, No. 3, pp. 30–42. 24. Coyle, H. M., and Castello, R. R., (1981),
14. Burland, J. P., (1993), “Closing address. “New design correlations for piles in sand”,
Proceedings of Recent Large-scale Fully Journal of the Geotechnical Engineering
Instrumented Pile Tests in Clay”, Institute of Division, 107(GT7): 965–986.
Civil Engineers, London, pp. 590–595. 25. Davisson, M. T., (1972), “High capacity
15. Bustamante, M., and Frank, R., (1997), piles”, Proceedings of the Lecture Series on
“Design of axially loaded piles - French Innovations in Foundation Construction,
practice. Design of Axially Loaded Piles – ASCE, Reston, VA, 81–112.
European practice”, Proceedings of ERTC3 26. Dennis, N. D., and Olson, R. E., (1983),
seminar, Brussels, Balkema, Rotterdam, The “Axial capacity of steel pipe piles in clay”,
Netherlands, pp. 161–175. Proceedings of Geotechnical Practice in
16. Bustamante, M., and Gianeselli, L., (1982), Offshore Engineering, Austin, Texas. Edited
“Pile bearing capacity predictions by means by Stephen Wright. ASCE, New York: 370–
of static penetrometer CPT”, Proceedings 388.
of 2nd European symposium on penetration 27. De Nicola, A., and Randolph, M. F., (1999),
testing, Amsterdam, pp. 493–500. “Centrifuge modeling of pipe piles in sand
17. Cai, G., Liu, S., and Puppala, A. J., (2011), under axial loads”, Géotechnique, Vol. 49,
“Evaluation of pile bearing capacity from No. 3, pp. 295–318.
piezocone penetration test data in soft 28. De Ruiter, J., and Beringen, F. L., (1979), “Pile
Jiangsu Quaternary clay deposits”, Marine foundations for large North Sea structures”,
Georesources and Geotechnology, Vol. 29, Marine Geotechnology, Vol. 3, No. 3, pp.
pp. 177–201. 267–314.
18. Chakraborty, T. Salgado, R. Basu, P., and 29. Doherty, P., and Gavin, K., (2011), “The shaft
Prezzi, M., (2013), “Shaft capacity of drilled capacity of displacement piles in clay: a
shafts in clay”, Journal of Geotechnical state of the art review”, Geotechnical and
and Geoenvironmental Engineering, ASCE, Geological Engineering, Vol. 29, No. 4, pp.
139(4): 548–563. 389–410.
19. Chandler, R. J., (1968), “The shaft friction of 30. Drewry, J. M., Weidler, J. B., and Hwong, S.
piles in cohesive soils in terms of effective T., (1977), “Predicting axial pile capacities
stress”, Civil Engineering and Public Works for offshore platforms”, Petroleum
Review, Vol. 60(708), pp. 48–51. Engineering, May 1977, pp. 41–44.
20. Chen, Y. J., and Kulhawy, F. H., (1994), “Case 31. Eslami, A., and Fellenius, B. H., (1997), “Pile
history evaluation of the behavior of drilled capacity by direct CPT and CPTu methods
shafts under axial and lateral loading”, EPRI applied to 102 case histories”, Canadian
TR-104601, Research Project No. 1493–04. Geotechnical Journal, Vol. 34, No. 6, pp.
21. Chin, F. K., (1970), “Estimation of the 880–898.
ultimate load of piles not carried to failure”, 32. Fellenius, B. H., (2002), Excerpt from Chapter
Proceedings of the 2nd Southeast Asian 6 of the red book: “Direct methods for
Conference on Soil Engineering, Singapore, estimating pile capacity”, in Background to
81–90. UniCone: http://www.fellenius.net, accessed
22. Clausen, C. J. F., Aas, P. M., and Karlsrud, May 22, 2010.
K., (2005), “Bearing capacity of driven piles 33. Fioravante, V., (1994), “Centrifuge modeling
in sand, the NGI approach”, Proceedings of driven and bored piles in sand subjected
of International Symposium on Frontiers to axial load”, Workshop on Pile Foundations
in Offshore Geomechanics, Perth, Taylor & Experimental Investigation, Analysis and
Francis, London, pp. 677–681. Design, Napoli, pp. 125–163.
23. Clisby, M. B., Scholtes, R. M., Corey, M. W., 34. Fioravante, V., Ghionna, V. N., Jamiolkowski,
Cole, H. A., Teng, P., and Webb, J.D., (1978), M., and Pedroni, S., (1995), “Load carrying
“An evaluation of pile bearing capacities”, capacity of large diameter bored piles in

[74] DFI JOURNAL Vol. 7 No. 2 December 2013

 sand and gravel”, Proceedings of 10th Asian the 4th International Geotechnical Seminar:
Regional Conference on Soil Mechanics and Deep Foundations on Bored and Auger Piles,
Foundation Engineering, Beijing Vol. 2, pp. Ghent, pp. 83–100.
3–15. 45. Jardine, R. J., Chow, F. C., Overy, R., and
35. Flaate, K., and Selnes, P., (1977), “Side Standing, J., (2005), “ICP design methods for
friction of piles in clay”, Proceedings of driven piles in sands and clays”, Thomas
the 9th International Conference on Soil Telford Publishing, London, 105 p.
Mechanics and Foundation Engineering, 46. Karlsrud, K., Hansen, S. B., Dyvik, R., and
Tokyo, pp. 517–522. Kalsnes, B., (1993), “NGI's pile tests at
36. Fleming, W. G. K., Weltman, A. J., Randolph, Tilbrook and Pentre – review of testing
M. F., and Elson, W. K., (1992), Piling procedures and results. large-scale pile
Engineering. 2nd edition, Blackie/Halsted tests in clay”, Thomas Telford, London, pp.
Press, John Wiley & Sons, NY, 390 p. 549–583.
37. Gavin, K. G., and Lehane, B. M., (2003), 47. Karlsrud, K., Clausen, C. J. F., and Aas, P.
“The shaft capacity of pipe piles in sand”, M., (2005), “Bearing capacity of driven piles
Canadian Geotechnical Journal, Vol. 40, pp. in clay, the NGI approach”, Proceedings of
36–45. Frontiers in Offshore Geotechnics, Perth,
38. Gavin, K. G., and Lehane, B. M., (2005), Taylor and Francis Group, London, pp.
“Estimating the end bearing resistance 775–782.
of pipe piles in sand using the final 48. Karlsrud, K., (2012), “Prediction of load-
filling ratio”, Proceedings of International displacement behavior and capacity of
Symposium on Frontiers in Offshore axially-loaded piles in clay based on
Geomechanics, Perth, Taylor and Francis analyses and interpretation of pile load
Group, London, pp. 717–723. test results”, PhD Dissertation, Norwegian
39. Goh, A. T. C., Kulhawy, F. H., and Chua, University of Science and Technology,
C. G., (2005), “Bayesian neural network Trondheim, 320 p.
analysis of undrained side resistance of 49. Kempfert, H-G., and Becker, P., (2010),
drilled shafts”, Journal of Geotechnical and “Axial pile resistance of different pile types
Geoenvironmental Engineering, 131(1): based on empirical values”, Proceedings of
84–93. Geo-Shanghai 2010: Deep Foundations and
40. Gwizdala, K., (1984), “Large diameter Geotechnical In-Situ Testing (GSP 205), ASCE,
bored piles in non-cohesive soils”, Swedish Reston, VA, pp. 149–154.
Geotechnical Institute, Report No. 26, 50. Kerisel, J. L., (1965), “Vertical and horizontal
Linkoping, Sweden. bearing capacity of deep foundations in
41. Huizinga, T. K., (1951), “Application clay”, Bearing capacity and settlement of
of results of deep penetration tests to foundations, Duke University, Durham, pp.
foundation piles”, Proceedings of Building 45–51.
Research Congress, London, Div. 1, Part 3, 51. Kolk, H.J., Baaijens, A. E., and Senders. M.,
173 p. (2005), “Design criteria for pipe piles in
42. Igoe, D. J. P., Gavin, K. G., and O'Kelly, B. C.,( silica sands”, Proceedings of International
2010), “Field tests using an instrumented Symposium on Frontiers in Offshore
model pile in sand”, Proceedings of Physical Geomechanics, Perth, Taylor and Francis
Modelling in Geotechnics, Taylor and Francis Group, London, pp. 711–716.
Group, pp. 775–780. 52. Kolk, H. J., and Van der Velde, E., (1996),
43. Igoe, D. J. P., Gavin, K. G., and O'Kelly, B. “A reliable method to determine the
C., (2011), “Shaft capacity of open-ended friction capacity of piles driven into clays”,
piles in sand”, Journal of Geotechnical and Proceedings of 28th Annual Offshore
Geoenvironmental Engineering, Vol. 137, No. Technology Conference, Houston, pp. 337–
10, pp. 903–913. 346.

44. Jamiolkowski, M., (2003), “Deep foundations 53. Kondner, R. L., (1963), “Hyperbolic stress-
on bored and auger piles”, Proceedings of strain response: cohesive soils”, Journal of

DFI JOURNAL Vol. 7 No. 2 December 2013 [75]

 Soil Mechanics and Foundations Division, 63. Mayne, P. W., (2007), “Cone penetration
Vol. 89, No. 1, 115–143. testing – a synthesis of highway practice”,
54. Kraft, L. M., Focht, J. A., and Amerasinghe, NCHRP Synthesis 368, Transportation
S. F., (1981), “Friction capacity of piles Research Board, Washington, D.C., 117 p.
driven into clay”, Journal of Geotechnical 64. McClelland, B., (1974), “Design of deep
Engineering Division, Vol. 107, pp. 1521– penetration piles for ocean structures”,
1541. Journal of Geotechnical Engineering Division,
55. Kulhawy, F. H., (2004), “On the axial Vol. 100, pp. 709–747.
behavior of drilled foundations”, 65. Menzenbach, E., (1961), “The determination
GeoSupport 2004: Drilled shafts, of permissible point loads of piles by means
micropiling, deep mixing, remedial methods, of static penetration tests”, Proceedings
and specialty foundation systems (GSP 124), of the 5th International Conference on Soil
J. P. Turner and P. W. Mayne, editors., ASCE, Mechanics and Foundation Engineering,
Reston, VA, pp. 34–51. Paris, Vol. 2, 99 p.
56. Kulhawy, F. H., and Phoon, K., (1993), 66. Meyerhof, G. G., (1951), “The ultimate
“Drilled shaft side resistance in clay soil bearing capacity of foundations”,
to rock”, Design and Performance of Deep Géotechnique, Vol. 2, No. 4, pp. 301–332.
Foundations: Piles and Piers in Soil and Rock, 67. Meyerhof, G. G., (1976), “Bearing capacity
ASCE (GSP 38), pp. 172–183. and settlement of pile foundations”, Journal
57. Kulhawy, F. H., Trautmann, C. H., Beech, J. of Geotechnical Engineering Division, Vol.
F., O'Rourke, T. D., and McGuire, W., (1983), 102, pp. 195–228.
“Transmission line structure foundations 68. Meyerhof, G. G., (1983), “Scale effects
for uplift-compression loading”, Report EL- of ultimate pile capacity”, Journal of
2870, Electric Power Research Institute, Palo Geotechnical Engineering, Vol. 109, pp.
Alto, CA, 412 p. 797–806.
58. Lee, J. H., and Salgado, R., (1999), 69. Miller, G. A., Lutenegger, A. J., (1997),
“Determination of pile base resistance “Influence of pile plugging on skin
in sands”, Journal of Geotechnical and friction in overconsolidated clay”, Journal
Geoenvironmental Engineering, Vol. 125, No. of Geotechnical and Geoenvironmental
8, pp. 673–683. Engineering, Vol. 123, pp. 525–533.
59. Lehane, B. M., and Gavin, K. G., (2001), 70. Niazi, F. S., (2014), “Static axial pile
“Base resistance of jacked pipe piles foundation response using seismic
in sand”, Journal of Geotechnical and piezocone data”, PhD Thesis, School of Civil
Geoenvironmental Engineering, Vol 127, No. and Environmental Engineering, Georgia
6, pp. 473–480. Institute of Technology, Atlanta, Georgia
60. Lehane, B. M., Schneider, J. A., and Xu, X., (under approval and submission).
(2005), “The UWA-05 method for prediction 71. Niazi, F. S., and Mayne, P. W., (2013), “Cone
of axial capacity of driven piles in sand”, Penetration Test Based Direct Methods
Proceedings of International Symposium on for Evaluating Static Axial Capacity of
Frontiers in Offshore Geomechanics, Perth, Single Piles: A State-of-the-Art Review”,
Taylor and Francis, London, pp. 683–689. Geotechnical and Geological Engineering,
61. Lehane, B. M., Li, Y., and Williams, R., (2013), 31(4): 979–1009.
“Shaft capacity of displacement piles in clay 72. Niazi, F. S., (2011), “Axial pile displacement
using the cone penetration test”, Journal evaluations from seismic piezocone data
of Geotechnical and Geoenvironmental and backanalysis of load tests”, Proceedings
Engineering, 139(2): 253–266. of the Deep Foundations Institute 36th
62. Lunne, T., Robertson, P. K., and Powell, J. Annual Conference on Deep Foundations.
J. M., (1997), “Cone penetration testing in Boston, MA, pp. 191–200.
geotechnical practice”, Blackie Academic/ 73. Niazi, F. S., and Mayne, P. W., (2010),
London, Routledge, NY, 312 p. “Evaluation of EURIPIDES pile load tests
response from CPT results”, International

[76] DFI JOURNAL Vol. 7 No. 2 December 2013

 Journal of Geoengineering Case Histories, the 2nd International Conference on Soil
Vol. 1, No. 4, pp. 367–386: www.geoengineer. Mechanics and Foundation Engineering,
org. Rotterdam, Vol. 2, 112 p.
74. Niazi, F. S., Mayne, P. W., and Woeller, D. 84. Powell, J. J. M., Lunne. T., and Frank, R.,
J., (2010), “Review of CPT-based methods (2001), “Semi-empirical design for axial
for response evaluation of driven piles in pile capacity in clays”, Proceedings of 15th
dense sands”, Proceedings of International International Conference on Soil Mechanics
Conference on Geotechnical Engineering, and Geotechnical Engineering, Istanbul,
Pakistan, pp. 259–266. Balkema, Rotterdam, The Netherlands, pp.
75. Nottingham, L. C., (1975), “Use of quasi- 991–994.
static friction cone penetrometer data to 85. Powell, J. J. M., and Quarterman, R. S.
predict load capacity of displacement piles”, T., (1988), “The interpretation of cone
PhD Thesis, University of Florida. 553 p. penetration tests in clays with particular
76. O'Neill, M. W., and Reese, L. C., (1999), reference to rate effects”, Penetration
“Drilled shafts: Construction procedure Testing 1988, Orlando, FL, Balkema,
and design methods”, FHWA Report IF-99- Rotterdam, The Netherlands, Vol. 2, pp.
025, U.S. Department of Transportation, 903–909.
Washington, DC. 86. Price, G., and Wardle, I. F., (1982), “A
77. Patel, D. C., (1989), “A case study of the comparison between cone penetration
shaft friction of bored piles in London test results and the performance of small
clay in terms of effective stress”, Thesis, diameter instrumented piles in stiff clay”,
University of London, Imperial College, as Proceedings of the 2nd European Symposium
referenced in Patrizi and Burland (2001). on Penetration Testing, Amsterdam, Vol. 2,
pp. 775–780.
78. Patrizi, P., and Burland, J., (2001),
“Development in the design of driven piles 87. Randolph, M. F., (1983), “Design
in clay in terms of effective stresses”, Rivista considerations for offshore piles”,
ltaliana di Geotecnica, Vol. 35, No. 3, pp. Proceedings of Conference on Geotechnical
35–49. Practice in Offshore Engineering, Austin, pp.
422–439.
79. Peck, R. B., (1958), “A study of the
comparative behavior of friction piles”, 88. Randolph, M. F., and Murphy, B. S., (1985),
Special Report No. 36, Highway Research “Shaft capacity of driven piles in clay”,
Board. Proceedings of 17th Annual Offshore
Technology Conference, Houston, Vol. 1, pp.
80. Pelletier, J. H., and Doyle, E. H., (1982),
371–378.
“Tension capacity in silty clays – Beta pile
tests”, Proceedings of the 2nd International 89. Randolph, M. F., and Wroth, C. P., (1978),
Conference on Numerical Methods in “Analysis of deformation of vertically-
Offshore Piling, Austin, pp. 1–19. loaded piles”, Journal of Geotechnical
Engineering Division, Vol. 104, GT12, pp.
81. Peuchen, J., Vanden-Berghe, J. F., and
1465–1488.
Coulais, C., (2010), "Estimation of u1/
u2 conversion factor for piezocone." 90. Reese, L. C., and O'Neill, M. W., (1988),
Proceedings of the 2nd International “Drilled shafts: construction practices and
Symposium on Cone Penetration Testing design methods”, Publication No. FHWA-
(CPT’10), Huntington Beach (http://www. HI-88-042, Federal Highway Administration,
cpt10.com/). Office of Implementation, Washington, D. C.,
564 p.
82. Philipponnat, G., (1980), “Methode pratique
de calcul d'un pieu isole a l'aide du 91. Robertson, P. K., (2009), “Interpretation
penetrometre statique”, Revue Française de of cone penetration tests – a unified
Géotechnique, Vol. 10, pp. 55–64. approach”, Canadian Geotechnical Journal,
Vol. 46, No. 11, 1337–1355.
83. Plantema, G., (1948), “Results of a special
loading test on a reinforced concrete pile, 92. Salgado, R., (2006), “The role of analysis
a so-called pile sounding”, Proceedings of in non-displacement pile design”, Modern

DFI JOURNAL Vol. 7 No. 2 December 2013 [77]

 Trends in Geomechanics, Springer 102. Twine, D. (1987), “Shaft friction of bored,
Proceedings in Physics, Vol. 106, pp. 521– cast in-situ piles in stiff, overconsolidated
540. clays in terms of effective stress”, Thesis,
93. Schmertmann, J. H., (1978), “Guidelines for University of London, Imperial College, as
cone penetration test, performance and referenced in Patrizi and Burland (2001).
design”, U.S. Department of Transportation, 103. Van der Veen, C., and Boersma, L.,
Washington, DC, Report No. FHWA- (1957), “The bearing capacity of a pile
TS-78-209, 145 p. predetermined by a cone penetration
94. Semple, R. M., and Rigden, W. J., (1984), test”, Proceedings of the 4th International
“Shaft capacity of driven pipe piles in clay”, Conference on Soil Mechanics and
Proceedings of Symposium on Analysis and Foundation Engineering, London, Vol. 2,
Design of Deep Foundations, San Francisco, 72 p.
pp. 59–79. 104. Van Dijk, B. F. J., and Kolk, H. J., (2011),
95. Seo, H., Yildirim, I.Z., and Prezzi, M., “CPT-based design method for axial
(2009), “Assessment of the axial load capacity of offshore piles in clays”,
response of an H pile driven in multilayered Proceedings of Frontiers in Offshore
soil”, Journal of Geotechnical and Geotechnics II, Taylor and Francis Group,
Geoenvironmental Engineering, Vol. 135, No. London, pp. 555–560.
12, pp. 1789–1804. 105. Vesić, A. S., (1977), “Design of pile
96. Skempton, A. W., (1959), “Cast in-situ bored foundations”, NCHRP Synthesis of Highway
piles in London clay”, Géotechnique, Vol. 9, Practice 42, Transportation Research
No. 2, pp. 91–125. Board, National Research Council,
Washington, D.C., 68p.
97. Takesue, K., Sasao, H., and Matsumoto, T.,
(1998), “Correlation between ultimate pile 106. Vijayvergiya, V. N., and Focht, J. A., (1972),
skin friction and CPT data”, Proceedings “A new way to predict the capacity of
of Geotechnical Site Characterization piles in clay”, Proceedings of 4th Annual
Conference, Rotterdam, Vol. 2, pp. 1177– Offshore Techonology Conference,
1182. Houston, pp. 269–284.

98. Togliani, G., (2008), “Pile capacity 107. White, D.J., and Bolton, M. D., (2005),
prediction for in-situ tests”, Proceedings “Comparing CPT and pile base resistance
of Geotechnical and Geophysical Site in sand”, Proceedings of the Institute
Characterization, Taylor and Francis Group, of Civil Engineering: Geotechnical
London, pp. 1187–1192. Engineering, Vol. 158, No. GE1, pp. 3–14.

100. Tomlinson, M.J., (1957), “The adhesion of 108. Woodward, R. J., Lundgren, R., and
piles driven in clay soils”, Proceedings of Boitano, J. D., (1961), “Pile loading tests
the 4th International Conference on Soil in stiff clays”, Proceedings of the 6th
Mechanics and Foundation Engineering, International Conference on Soil Mechanics
London. and Foundation Engineering, Paris, pp.
177–184.
101. Tumay, M. T., and Fakhroo, M. (1982.),
“Friction pile capacity prediction in 109. Yu, F., and Yang, J., (2012), “Base
cohesive soils using electric quasi-static capacity of open-ended steel pipe piles
penetration tests”, Interim Research in sand”, Journal of Geotechnical and
Report No. 1, Louisiana Department of Geoenvironmental Engineering, Vol. 138,
Transportation and Development, Research No. 9, pp. 1116–1128.
and Development Section, Baton Rouge,
LA, 275 p.

[78] DFI JOURNAL Vol. 7 No. 2 December 2013

Underwriters: Gold

Your Solutions
Provider.
CANADA & WESTERN U.S.
CHAMPION EQUIPMENT SALES, LLC
562.634.8180

EASTERN & CENTRAL U.S.

AMERICAN EQUIPMENT & FABRICATING CORP
401.438.2626

www.soilmec.com

DFI JOURNAL Vol. 7 No. 2 December 2013 [79]

 DFI Journal Paper Review Process
The peer review process for documents considered for publication in the DFI Journal is still evolving.
The following is a description of the current process, however, the publication is still in its infancy and
the review process is still in a state of flux. DFI reserves the right to alter the procedures as necessary.

Paper Submittal
Papers may be submitted at any time. Authors wishing to submit their papers for consideration of
publication in the DFI Journal are invited to access www.edmgr.com/dfi. The website will ask for
a login or, for new submitters, will ask for creation of an account. Once logged in the author must
upload a full paper in MS Word format as well as any ancillary files such as figures, photos and other
graphics which are included in the paper. The paper is then converted to a PDF file which the author
must approve before the paper will be released to the publisher and journal editors for viewing. The
journal editors preliminarily review the paper for relevancy to the Journal mission.

Paper Review
The journal editors assign those papers deemed to be worthy of consideration for Journal publication
to the appropriate editorial board member, which currently consists of DFI technical committee
chairmen and other industry leaders, so that appropriate reviewers for the paper topic can be obtained.
Reviewers are chosen based on their knowledge, areas of expertise, and qualifications to act as a
reviewer on the particular subject matter of the paper in question. At least three reviewers will be
assigned to each paper.
After the reviewers are selected, they are provided with instructions and a password for entry into
the website where they can view the paper PDF and submit their evaluation. The criteria on which
they base their review fall under two areas: technical content and quality of paper presentation.
The criteria for technical content include relevancy, originality, appropriate references to support
statements, significance of results and exclusion of personal opinion and commercialism. The criteria
for paper presentation include quality of figures, quality of English language, paper organization and
completeness. The reviewers enter their evaluation by responding to a number of questions rating the
paper as well as entry of comments to authors. They are also required to make a recommendation to
the journal editors of: accept as is; accept with mandatory changes; or reject. The author is advised
by automatic email of the posting of reviews and he/she can access the reviews and respond and/or
modify the paper to satisfy comments by the reviewers. A second round review can then take place if
necessary, ultimately leading to second round reviewer recommendations. The publisher and editors,
acting as a final review committee, make the decision, based on the reviewers’ recommendations, as to
acceptance of the paper for publication in the next issue of the journal or in a subsequent issue.
Throughout the process, automatic emails are sent out to reviewers when papers are ready for their
review and to the authors to keep them aware of the progress of their paper.

Contact
For enquires relating to the submission and status of articles please email dfi.ed@maneypublishing.
com. Please quote the manuscript reference number (where possible) in all correspondence.

Paper Finalization
Upon acceptance, the final paper submission by the author and all graphic files are downloaded by
the publisher for processing and formatting for publication. The publisher is provided with proofs by
the production house and these are edited to ensure acceptable layout, the absence of typos, clarity of
figures, etc. In most cases the author(s) are provided with a final PDF for their review and approval.

[80] DFI JOURNAL Vol. 7 No. 2 December 2013

DFI Journal Call for Papers
The Deep Foundations Institute compiles and publishes a Journal of practical and technically rigorous
papers on a bi-annual schedule. The DFI Journal is distributed to ~3,000 DFI members plus non-
member subscribers.
The DFI Journal content is subject to quality technical review, and must meet a standard in quality on
practical subjects dealing with case studies, deep foundations history, design, construction, testing,
innovations and research in the field.
Each journal consists of at least five documents collected from technical papers that are invited
or selected from papers submitted by international industry members based on this call. Papers
presented at the DFI Annual Conference and Specialty Seminars may be included if expanded to the
Journal standard and review process.
The editors are herein sending out a call for original papers for consideration of inclusion
in the upcoming journals. Full draft papers up to 15 pages in length are to be submitted to:
www.edmgr.com/dfi for review. Authors will be required to create a login account and will be notified
via email on the status of their submission.
Papers are solicited on the following topics:
• Case studies involving foundation systems with technical data support
• Historical evolution of deep foundations
• Relationship between use of design, construction and equipment
• Quality control, quality assurance and non-destructive testing
• Innovation in all aspects of deep foundations and earth retention
• Practice-oriented research
The Publisher and the Journal Editorial Board will review submitted papers for acceptability for
publication in the current or future issues of the Journal, subject to full peer reviews as described
on the preceding page entitled "DFI Journal Paper Review Process". Authors of papers accepted for
publication will be required to sign a copyright licence agreement.
Further information about the journal is available at www.maneyonline.com/dfi.

DFI JOURNAL Vol. 7 No. 2 December 2013 [81]

NEW PUBLISHER
DFI Journal: The Journal of the Deep Foundations Institute

From 2014, the DFI Journal will be published as part of Maney Publishing’s growing materials
science and engineering collection. The Journal will continue to publish in print and online
and will publish 2 issues per year, with the intention to increase in frequency in the future.

What does this mean for the journal?

For members For subscribers For authors

Members will be able to access all Maney will be contacting existing Authors will be able to submit their
DFI Journal issues via the DFI subscribers directly concerning the papers online using Editorial
website from 2014. renewal of subscriptions. For help or ManagerTM, an online submission,
information, please email peer-review and tracking system.
Content alerts and RSS feeds can be [email protected]
set up to inform you of new content For more information on how to
as soon as an issue is published If you would like to subscribe to this submit, please see the instructions
online. journal, please visit the journal on the journal homepage:
homepage: www.maneyonline.com/dfi www.maneyonline.com/dfi

About Maney Publishing

Maney is an independent publishing company with a diverse and fast-expanding portfolio of materials science and engineering
journals, which the DFI Journal complements nicely. Maney’s publishing partners include the International Desalination
Association, ASM International, the MetSoc, the Institute of Materials, Minerals and Mining, AusIMM and many others. For more
information on the titles in the Materials Science & Engineering Collection, visit www.maneyonline.com/matscieng.

With offices in Leeds and London in the UK, and Philadelphia in North America, Maney focuses on the publication and
dissemination of high quality, peer-reviewed scientific research.

For more information, please visit the DFI Journal homepage: www.maneyonline.com/dfi

www.maneyonline.com/dfi www.dfi.org
 Maney
Online
Maney Publishing’s new
online journal platform

Have you tried reading the DFI Journal online?

All content published in the DFI Journal from 2014 will be available on
a new journal platform, Maney Online.

Maney Online, powered by Atypon’s What can you expect from Maney Online?
Literatum platform, brings greater n straightforward access to content
functionality and flexibility to the online
publication of the DFI Journal. n an intuitive user interface
n advanced search capabilities
Members will continue to access the DFI n popular research at your fingertips
Journal via their “myDFI” login and will be
n ability to save your favorite articles
notified shortly. Subscribers should register
on Maney Online now and be ready to n export citations easily
access the first issue of 2014. n recommended articles tailored to you

www.maneyonline.com/dfi
www.maneyonline.com/dfi www.dfi.org
Deep Foundations Institute was incorporated in 1976 in the State of New Jersey as a non-profit
educational activity. DFI is a technical association of firms and individuals in the field of designing and
constructing deep foundations and excavations. DFI covers the gamut of deep foundation construction
and earth retention systems.

Although the bulk of the membership is in North America, the Institute is worldwide.

DFI’s strengths are: DFI Sustaining Members

• Communication of information concerning AECOM USA INC.
the state-of-the-art and state of the practice AMEC - ENVIRONMENT & INFRASTRUCTURE
of deep foundation technologies AMERICAN EQUIPMENT & FABRICATING CORP.
• Offering networking opportunities for ANDERSON DRILLING
our members APE/J&M
BAUER FOUNDATION CORP.
• Offering opportunities for members to
BAUER - PILECO INC.
improve the industry through publications
produced by volunteer committees BEN C. GERWICK INC.
BERKEL & COMPANY CONTRACTORS INC.
• Offering educational conferences, seminars
BRASFOND FUNDAÇÕES ESPECIAIS S/A
and workshops in the industry
BRAYMAN CONSTRUCTION CORPORATION
The core strength of DFI is the broad spectrum
CAJUN DEEP FOUNDATIONS LLC
of its membership. All disciplines participate
on an equal footing, be they contractors, CASE FOUNDATION COMPANY
engineers, owners, academicians, equipment CIPORT S.A.
manufacturers and distributors or materials DEAN CONSTRUCTION CO. LTD.
manufacturers and suppliers. All types of DEWITT CONSTRUCTION INC.
foundation systems are represented, whether
FOUNDATION CONSTRUCTORS INC.
installed by driving, drilling or other means.
This diversity and openness without bias FOUNDATION SUPPORTWORKS INC.
provides a forum for the free exchange of FOUNDATION TECHNOLOGIES INC.
knowledge and a platform for the development GEOKON INC.
of new technology and opportunity. GOETTLE
HAYWARD BAKER INC.
DFI is: HJ FOUNDATION COMPANY
• An international network of heavy KELLER FOUNDATIONS LTD.
construction professionals dedicated to KIEWIT INFRASTRUCTURE ENGINEERS
quality and economy in foundation design KLEINFELDER
and construction L.G. BARCUS & SONS INC.
• A forum open to all construction LANGAN ENGINEERING AND ENVIRONMENTAL
professionals across disciplines SERVICES
and borders. MCKINNEY DRILLING COMPANY
• A technological association devoted to MENARD
gathering, storing and disseminating MORETRENCH
practical information MUESER RUTLEDGE CONSULTING ENGINEERS
• A resource for identifying and locating the NICHOLSON CONSTRUCTION COMPANY
specialists and sources of expertise. O.C.I. DIVISION / GLOBAL DRILLING SUPPLIERS INC.
PND ENGINEERS INC.
• An initiator and participant in research
SAS STRESSTEEL INC.
SCHNABEL FOUNDATION COMPANY
Deep Foundations Institute Sustaining Members
are Corporate Members of DFI who have TEI ROCK DRILLS INC.
voluntarily granted funding to the Institute for THATCHER ENGINEERING CORPORATION
expanded support of the Industry. The fund is URBAN FOUNDATION/ENGINEERING LLC
managed by the DFI Educational Trust. WILLIAM F. LOFTUS ASSOCIATES FOUNDATION
ENGINEERS
WURSTER ENGINEERING & CONSTRUCTION INC.
DFI JOURNAL
The Journal of the Deep Foundations Institute

International Standard Serial Number (ISSN): 1937-5247

Deep Foundations Institute

326 Lafayette Avenue
Hawthorne, New Jersey 07506 USA
Tel: 973-423-4030
Fax: 973-423-4031
www.dfi.org