1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia,

24-26 June 2019

Model Factors for Foundation Design – A Comprehensive Review

Kok-Kwang Phoon1 and Chong Tang2
1
Professor, Department of Civil and Environmental Engineering, National University of Singapore, Singapore
2
Research fellow, Department of Civil and Environmental Engineering, National University of Singapore, Singapore
E-mail: [email protected]

ABSTRACT: The calculated response from a numerical model will deviate from the measured one given the presence of modeling
idealizations and real world construction effects. This deviation can be directly captured by a ratio between the measured and the calculated
quantity. The ratio is also called a model factor in many design guides and codes. The probabilistic distribution of the model factor is
arguably the most common and simplest complete representation of model uncertainty. The characterization of model uncertainty is
identified as one of the critical elements in a geotechnical reliability-based design process in Annex D of ISO 2394:2015 “General Principles
on Reliability of Structures”. This keynote paper presents a large generic database (PILE/2739) that contains 2739 field load tests conducted
on various piles and installed in different soils and countries. The bias (mean) and dispersion (COV) of the model factor for a range of
foundations, geomaterials, and calculation models at the ultimate limit state are summarized in a form suitable for adoption in design and
codes of practice. Based on this summary, it is proposed that a model factor for a design model can be classified as: (1) moderately
conservative (1≤mean<2), (2) highly conservative (2≤mean<3), or (3) very highly conservative (mean≥3). The model uncertainty can be
classified as: (1) low dispersion (COV<0.3), (2) medium dispersion (0.3≤COV<0.6), (3) high dispersion (0.6≤COV<0.9), and (4) very high
dispersion (COV≥0.9). These model factors are necessary for the calibration of resistance factors in the Load and Resistance Factor Design.

KEYWORDS: Model uncertainty, Model factor, Load test database, PILE/2739, Load and Resistance Factor Design

1. INTRODUCTION 2. DEFINITION OF MODEL FACTOR

In general, it is difficult to “predict” the behaviour of a geo-structure For foundation capacity, the model uncertainty can be characterized
due to complicated soil-structure interaction (including interface in a natural way by the ratio of the measured to the calculated value
behaviour), complex geological conditions, and construction effects (Eq. D.1 in Annex D of ISO 2394:2015):
(Phoon and Tang 2019a). This is well known in geotechnical
engineering that distinguishes between a Class A prediction made M = Qum Quc (1)
before construction and a Class C prediction made after the response
of interest has been measured. Lambe (1973) advocated that the
where M is the model factor, Qum is the measured capacity, and Quc
geotechnical engineering profession “is in great need of simple
is the calculated capacity. It is customary to model M as a random
techniques to make type A predictions.” The predicted value (e.g.
variable. The simplest method to characterize a random variable is
failure load, stress, or deformation) from simple techniques will
to calculate the first two moments: mean (bias) and COV
deviate from the measured value (typically on the safe side). This
(dispersion). A mean and COV of M close to 1 and 0, respectively,
deviation can be directly captured by a ratio of the measured value
would represent a near perfect calculation method that matches
to the predicted (or calculated) value. The ratio is also known as a
measured responses for all scenarios in the calibration database. It
model factor (ISO 2015). There is overwhelming empirical evidence
goes without saying that such methods do not exist in geotechnical
showing that assigning a single value to the model factor is grossly
engineering, regardless of their numerical sophistication. In fact, as
incomplete - different design scenarios will produce different values
mentioned above, it is unlikely for the COV to be less than 10%.
of the model factor, even for the same geo-structure and the same
The review conducted by Phoon and Tang (2019a) supports this
design (calculation) model. This model uncertainty is regarded as
lower bound.
epistemic in nature (Section 2.2.3, DNV 2017), because it
At the design stage, it is obviously useful to know the model
predominantly arises from the simplifications, assumptions and
factor associated with a calculation method given there is always a
approximations made in the design model. Some design guidelines
gap between the calculated value and the measured (actual) value.
(JCSS 2006; DNV 2017; ISO 2015) have referred to model
The mean of M would provide an engineer with a sense of the
uncertainty and it is apparently being considered in the current
hidden factor of safety that either adds or subtracts from the nominal
revision process of the Eurocodes (Lesny 2017).
global factor of safety, depending on whether the calculation method
Characterization of model uncertainties for various geo-
is conservative (mean>1) or unconservative (mean<1) in the average
structures is limited, despite its importance being highlighted in a
sense. It should not be inferred that a calculation method is
number of design guidelines. An extensive review was conducted
conservative or otherwise for a specific case because M takes a
recently to summarize the mean and coefficient of variation (COV)
range of values in actuality (hence it is random). This random nature
of the model factor for a range of geo-structures, geomaterials, limit
is practically significant, because it implies that a calculation
states (both ultimate and serviceability), and calculation models in a
method can be unconservative when applied to a specific case even
form suitable for adoption in these design guidelines (Phoon and
though the method is conservative on the average. One example can
Tang 2019a). In the context of a model factor, Lambe (1973)’s
be found in Tang and Phoon (2018a) for low displacement helical
remark can be more realistically interpreted as making Class A
piles in sand under compression. The empirical torque-capacity
predictions with minimum bias (mean close to 1) and dispersion
correlation was used to calculate axial pile capacity. Based on 115
(COV close to 10%). Phoon and Kulhawy (1999) recommended a
static load tests, the model factor ranges between 0.53 and 2.82 with
COV of less than 10% for the unit weight of soil and it is reasonable
mean=1.39 and COV=0.33. The torque-capacity correlation method
to say that this COV is a lower bound for all practical purposes. This
is conservative on the average, but it can over-predict and under-
keynote paper presents the most comprehensive review of model
predict the actual capacity by roughly 2 and 3 times, respectively.
statistics for foundation design to date. These statistics are necessary
Therefore, it is important to consider both the mean and COV of the
for the calibration of resistance factors in the Load and Resistance
model factor. The LRFD approach discussed below can do this
Factor Design (LRFD). They also provide a quantitative feel for
without sophisticated probabilistic calculations.
how a design method adds/subtracts to the nominal factor of safety.
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

3. PILE/2739 (Department of Transportation) (Roling et al. 2011) and Deep

Foundation Load Test Database (DFLTD) maintained by the Federal
Load test databases play a key role in the characterization of model Highway Administration (FHWA) (Abu-Hejleh et al. 2015). More
factors for foundations. For these databases to be useful, they should details can be found in Tang and Phoon (2018b). In accordance with
contain information beyond just the foundation geometry, soil the study of Phoon and Tang (2019b), three types of definition of
profile, and load displacement curves. Over the past few years, the “capacity” are considered – (1) plunging failure; (2) Davisson
authors compiled a generic database (PILE/2739) covering many (1972) that allows the engineer, when proof testing a pile for a
foundation types installed in different soil types and different certain allowable load, to determine in advance the maximum
countries (see Figure 1). It includes 2739 full-scale field load tests, allowable movement for this load with consideration of the length
which are compiled from several existing databases (e.g. AAU-NGI, and size of the pile (Fellenius 1980); and (3) Chin (1970) that
DSHAFT, FHWA DFLTD, PILOT, WBPLT, and ZJU-ICL), real assumes the load-movement curve follows a hyperbolic shape when
construction projects, published technical reports and papers. The the load approaches the failure and allows a continuous check on the
geographical regions cover Australia, Belgium, Brazil, Canada, test if a plot is made as the test proceeds (Fellenius 1980). The
China, Denmark, France, Germany, Indonesia, Iraq, Iran, Ireland, capacity is calculated using the static design methods in ISO 19901-
Italy, Kuwait, Mexico, Netherlands, Norway, Singapore, Sweden, 4:2016 (ISO 2016) for sand (β-method) or clay (α-method). For
Taiwan, Thailand, Turkey, United Kingdom and United States. The mixed soils, either the α-method or β-method is used as appropriate.
geomaterial covers soft to stiff clay, loose to dense sand, silt, and The model statistics are summarized in Table 1. When the
gravel. The pile types are also diverse, covering small to large “capacity” interpreted by the Davisson (1972) or Chin (1970) failure
displacement piles (e.g. steel H-piles, torque-driven helical piles, criterion is adopted as the measured capacity, the resulting COV is
driven cast-in-situ piles, and driven closed/open-end concrete/steel higher than the case where peak or plunging failure loads exist. This
piles) and non-displacement piles [e.g. drilled shafts and augered is largely attributed to the bias in the interpretation of “capacity”
cast-in-place (ACIP) piles]. such as the consistency of the failure criterion and its application to
the test results (Phoon and Tang 2019b). Compared to the Davisson
(1972) method, extrapolation using the hyperbolic model of Chin
(1970) does exhibit a tendency to over-predict the interpreted
capacity (largest mean>1) and does result in more uncertainty
(highest COV). Therefore, Kyfor et al. (1992) and NeSmith and
Siegel (2009) stated that any attempt to extrapolate load-movement
curves should be avoided and pile capacity should be established by
measurements only and not from any hypothetical models.

Table 1 Comparison of model statistics of steel H-piles for different

failure criteria. The capacity is calculated using static design
methods in ISO 19901-4:2016 (Source: Phoon and Tang 2019b).
Soil type Data Plunging Davisson Chin
Clay N 18 21 21
Mean 0.95 0.99 1.34
COV 0.37 0.46 0.54
Sand N 5 28 28
Mean 0.99 0.82 1.35
COV 0.27 0.39 0.52
Mixed N 18 27 27
Mean 0.87 0.86 1.21
COV 0.25 0.3 0.32
All N 41 76 76
Figure 1 Distribution of load test data by pile type Mean 0.92 0.88 1.3
(Source: Phoon and Tang 2019a). COV 0.31 0.4 0.47

3.1 Interpretation of “capacity” 3.2 Model statistics

For foundation design, the measured capacity is understood to be the Axial pile capacity can be calculated by static or dynamic analysis
load for which rapid movement occurs under sustained or slight methods as detailed in AASHTO (2014). Static analyses include
increase of applied load – the pile plunges (Fellenius 1980). This indirect and direct methods. Indirect methods, although involving
definition may not be applicable to some load tests, because large some degree of empiricism, are built on established equations of
movements are required to reach plunging failure. For case histories bearing capacity and skin friction used to describe the behaviour of
where a distinct plunging failure is not observed at the maximum piles and buried anchors in compression and uplift (e.g. α- and β-
applied load, other failure criteria have been proposed to interpret methods). Direct methods eliminate the need to estimate
the “capacity”. The section 1810.3.3.1.3 on “Load test evaluation geotechnical input parameters by transforming the in-situ test results
methods” in the 2018 International Building Code (ICC 2018) to the bearing capacity and skin friction (e.g. CPT- and SPT-based
explicitly recommends three methods: (1) Davisson (1972), (2) methods) without the mediation of a physical model. Dynamic
Hansen (1963) (90% criterion), and (3) Butler and Hoy (1977). Note methods are discussed elsewhere (Phoon and Tang 2019a).
that this interpreted capacity may not be the actual value (if it exists Because of the availability of a large number of load tests,
at all). To compute model statistics, it is clearly important to apply a static analysis methods have been extensively calibrated by many
consistent failure criterion to all the load-movement curves in a researchers (AbdelSalam et al. 2012, 2015; Abu-Farsakh et al. 2009,
database, be it Davisson or otherwise. 2013; Adhikari et al. 2018; Asem et al. 2018; Burlon et al. 2014;
An example of steel H-piles under axial compression is Dithinde et al. 2011; Eslami and Fellenius 1997; Gilbert and Tang
extracted from PILE/2793 to illustrate the effect of failure criterion 1995; Lacasse and Nadim 1996; Lacasse et al. 2013a, b; Lehane et
on the statistics of M. The static load tests for 300 steel H-piles are al. 2017; Long and Anderson 2014; McVay et al. 2016; Motamed et
compiled from PIle LOad Test (PILOT) developed by Iowa DOT al. 2016; Ng and Fazia 2012; Ng et al. 2014; Paikowsky et al. 2004;
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Reddy and Stuedlein 2017; Stark et al. 2013, 2017; Tang and Phoon 3. Driven concrete/steel piles with closed/open-end in sand
2018a-d, 2019a-b; Tang et al. 2019; Yang et al. 2017; Zhang and (pile diameter D=140-760 mm, slenderness ratio H/D=13-
Chu 2009). Ching et al. (2011) calibrated the resistance factors in 251, and internal friction angle ϕ=30-42º) (Tang and
LRFD of drilled shafts with incomplete load test results and hybrid Phoon 2018c) and clay (D=100-810 mm, H/D=10-213,
Gibbs-Metropolis algorithm. The model statistics are summarized in plasticity index Ip=11-160, overconsolidation ratio
Table 2 for driven piles and Table 3 for drilled shafts. OCR=1-432, and su=10-590 kPa) (Tang and Phoon
The evaluated model statistics are closely related to the 2018d).
calibration database including: (1) number of load tests (N); (2) 4. Drilled shafts in clay (D=350-1800 mm, H/D=1.6-56, and
geological conditions covered; and (3) pile type and dimensions. su=41-246 kPa), sand (D=300-2000 mm, H/D=1.4-69, and
Although different model statistics can be obtained for the same ϕ=30-45º), and gravelly soils (D=430-2260 mm, H/D=1.5-
design method, the ranges of model statistics are similar, 30, and ϕ=37-48º) (Tang et al. 2019).
particularly the COV. For driven piles in sand/clay, mean=0.39–
1.81 and COV=0.26–0.62 in Paikowsky et al. (2004), mean=0.86– 4. CLASSIFICATION OF MODEL UNCERTAINTY
1.66 and COV=0.24–0.56 in Lehane et al. (2017), and mean=0.74–
1.6 and COV=0.13–0.64 obtained by the authors (Tang and Phoon Table 3.7.5.1 of the JCSS Probabilistic Model Code (2006) only
2018a-d). For drilled shafts in sand/clay (compression), mean=0.81- provides first-order estimates of the expected means and standard
1.02 and COV=0.37-0.41 in AbdelSalam et al. (2015), while deviations for some commonly used geotechnical calculation
mean=0.99-1.18 and COV=0.38–0.53 in Tang et al. (2019). The models (Table 5). The procedure and data sources underlying these
range and the average of the mean and COV in Tables 2 and 3 for indicative model statistics are not provided. In Phoon and Tang
pile foundations are given in Table 4. The number of tests is also (2019a), the mean and the COV of the model factor for a range of
presented as an indicator of the degree of statistical uncertainty geo-structures, geomaterials, limit states, and design models are
associated with the mean and COV. A good rule-of-thumb for the summarized in tables such as Tables 2 and 3. References to the data
lower and upper bound of the COV would be 0.1 and 0.6, sources/analyses are provided in the last column.
respectively. Randolph (2003) opined that “we may never be able to
estimate axial pile capacity in many soil types more accurately than Table 2 Indicative computation model uncertainty factors (Source:
about ±30%. We therefore need to rely on pile tests conducted early Table 3.7.5.1 of the JCSS Probabilistic Model Code, 2006).
during the construction phase to refine the final design (generally in Type of problem Calculation model Mean S.D.
terms of varying the embedded pile length, but possibly also the Embankment: slope stability
diameter or number of piles).” This observation is largely based on Homogeneous Failure arc analysis 1.1 0.05
past experiences and judgment. A systematic and comprehensive Non homogeneous 2-D FEM 1.1 0.1
statistical review of the type shown in Table 2 and 3 in fact reveals Embankment: settlement prediction 1 0.2
that the model factor could vary in a much wider range. Retaining (sheet piled walls)
Furthermore, the data in Table 4 indicates a considerable degree of Stability Brinch Hansen 1.0 0.1
uncertainty, where many COV values are not smaller than 0.3 on the Shallow foundations: stability
average. Homogeneous Brinch Hansen 1 0.15
A recent body of work has emerged that is distinctive from past Non homogeneous 1 0.2
studies in 3 aspects (Tang and Phoon 2018a-d, 2019; Tang et al.
Shallow foundations: settlement prediction
2019): (1) substantial sample size and substantial coverage of
— Brinch Hansen 1 0.2 – 0.3
different design conditions (e.g. soil and pile types); (2) statistical
Driven piles
rigor, including the verification of the randomness of M and
Base capacity CPT based-methods 1 0.25
performing appropriate corrections when this is not true (corrections
Shaft resistance 1 0.15
are physics-based which means the resulting statistics are founded
Note: S.D. = standard deviation; FEM = finite element method; and
on best available knowledge), (3) consistent extension of M as a
CPT = cone penetration test.
random variable for ultimate limit state (ULS) to a bivariate random
vector (a, b) for serviceability limit state (SLS) (regarded as the
Their extensive review of foundation and other databases with
simplest probabilistic model that can address the salient nonlinear
20 or more tests is plotted in Figure 2. Note that the mean model
feature in all load-movement curves). The scope covered by this
factor (bias) is “unconservative” if it is less than one and
body of work is briefly outlined below:
“conservative” if it is larger than one. For “unconservative” models,
1. Single- or multi-helix low/high displacement helical piles
they can be classified as: (1) moderate (0.5≤mean<1) or (2) high
(shaft diameter d=48-506 mm, helix diameter D=203-
(mean<0.5). For “conservative” models, they can be classified as:
1016 mm, and the ratio of embedment pile length to
(1) moderate (1≤mean<2), (2) high (2≤mean<3), or (3) very high
diameter, namely, slenderness ratio H/D=5-110) in soft to
(mean≥3). The model COV (dispersion) is classified as (1) low
very stiff clay (undrained shear strength su=24-406 kPa) or
(COV<0.3), (2) medium (0.3≤COV<0.6), (3) high (0.6≤COV<0.9),
loose to dense sand (internal friction angle ϕ=30-45º)
and (4) very high (COV≥0.9). This broadly follows the three-tier
(Tang and Phoon 2018a, 2019). Recently, it has been
classification for soil property variability proposed by Phoon and
recognized that helical piles offer an efficient solution for
Kulhawy (2008). A model factor with an excessively large COV
offshore structures (Byrne and Houlsby 2015; Lutenegger
may be indicative of an overly simplified design model.
2017; Al-Baghdadi 2018) and can be used for earthquake
mitigation because of their slenderness, higher damping
5. LRFD CALIBRATION
ratios, ductility, and ability to resist uplift (Cerato et al.
2017; Sakr 2018; Elsawy et al. 2019). More studies are LRFD could be the most popular simplified reliability-based design
encouraged to investigate the use of helical piles in (RBD) format in North America that fits within the broader
offshore structures and earthquake mitigation. Several framework of limit state design (Kulhawy et al. 2012). The
manufacturers are working together to draft a design AASHTO and FHWA mandated the use of LRFD in the United
guideline for high displacement helical piles. States for all new federal-funded bridges (e.g. super- and sub-
2. Steel H-piles under axial compression in clay (averaged structures) after September 2007. Seo et al. (2015) conducted a
SPT blow count NSPT=5-40, pile embedment length review of research reports, bridge or geotechnical design manuals,
H=5.6-33 m), sand (NSPT=7-40 and H=6.6-46 m), and and standard specifications published by 49 state DOTs. In addition
mixed soils (NSPT=4-29 and H=4.9-47 m) (Tang and to AASHTO (2014), several state DOTs carried out research
Phoon 2018b)
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

projects to calibrate resistance factors to achieve a prescribed target where γQD and γQL are dead and live load factors, λQD and λQL are
reliability index using field load tests on driven piles, drilled shafts dead and live load bias factors, λR is resistance bias that is equal to
or both, such as Indiana (Salgado et al. 2011), Iowa (AbdelSalam et the mean of the model factor M in Eq. (1), COVR is the COV of the
al. 2012; Ng et al. 2014), Louisiana (Abu-Farsakh et al. 2009, 2013), model factor M, η=QD/QL is the dead to live load ratio, and βT is
Nevada (Motamed et al. 2016), and New Mexico (Ng and Fazia target reliability index.
2012). As an illustration, the capacity model statistics of steel H- According to Paikowsky et al. (2004), γQD=1.25 and γQL=1.75,
piles under axial compression are applied to calibrate the resistance λQD=1.05 and λQL=1.15, COVQD=0.1 and COVQL=0.2. The values of
factor in LRFD. For simplicity, the AASHTO first-order second- λR and COVR are given in Table 2 that are highlighted in bold font.
moment method is used. When only dead and live loads are In accordance with the International Building Code (ICC 2018), the
considered, the resistance factor ψ is estimated by (Paikowsky et al. Hansen (90% criterion) (1963), Davisson (1972), and Butler-Hoy
2004): (1977) methods are considered. The dead to live load ratio QD/QL is
chosen as 3 beyond which the resistance factor is almost constant
(Paikowsky et al. 2004; Tang and Phoon 2018b). The results of the
resistance factor ψ for βT=2.33 and 3 are given in Table 6 with the
efficiency factor ψ/λR that represents the effectiveness of the design
method (Paikowsky et al. 2004). In AASHTO (2014), ψ=0.35 (α-
method) and 0.25 (β-method) for clay and mixed soil and ψ=0.45
(Nordlund/Thurman method) for sand that are recommended for all
driven piles (e.g. precast concrete and steel pipe closed or open)
under axial compression. Note that the resistance factors
recommended by AASHTO (2014) (shown in parentheses in Table
6) are different from the ones computed in this paper, because (1)
the database used in this paper is larger and cover a larger range of
conditions and (2) load tests without sufficient movement required
by the Davisson (1972) method are not considered to avoid the
uncertainty arising from extrapolation.

Table 3 Calibrated resistance factors for steel H-piles under axial

compression.
Interpretation of βT=2.33 βT=3
Soil type “capacity” ψ ψ/λR ψ ψ/λR
Clay Hansen (90%) 0.4 0.34 0.28 0.24
Davisson 0.37 0.37 0.26 0.27
(0.35)
Butler-Hoy 0.46 0.45 0.35 0.34
Sand Hansen (90%) 0.31 0.3 0.21 0.2
Davisson 0.33 0.42 0.25 0.31
(0.45)
Butler-Hoy 0.33 0.33 0.23 0.23
Mixed Hansen (90%) 0.52 0.51 0.4 0.4
Davisson 0.49 0.56 0.39 0.45
(0.35)
Butler-Hoy 0.44 0.49 0.34 0.38
All Hansen (90%) 0.39 0.36 0.28 0.26
Davisson 0.37 0.42 0.27 0.31
Butler-Hoy 0.38 0.39 0.28 0.28
Note: the values in parentheses are those recommended in AASHTO
(2014).

6. CONCLUDING REMARKS
The ratio of the “capacity” interpreted from a measured load-
movement curve to the capacity calculated from a design model is
called a model factor. Capacity interpretation criteria include the
Davisson (1972) and Chin (1970) methods. In foundation design,
static or dynamic analysis methods are commonly used. The
following observations are noteworthy:
1. The concept of a model factor appears in many design
guidelines (JCSS 2006; DNV 2017) and codes (ISO 2015;
EN 1997). While load tests exist, a systematic
Figure 2 Classification of model uncertainty based on the mean and
characterization of this concept to provide numerical
COV of the model factor (Source: Phoon and Tang 2019a).
guidance is lacking until recently (Phoon and Tang 2019a).
2. The simplest model for a model factor is a random

R ( Q  +  Q
(1 + COV + +COV )
2 2 variable and the simplest characterization is to determine
D L
) QD

(1 + COV )
2
QL
its mean and COV.
R 3. A large generic database (PILE/2739) that contains 2739
=
(QD L 
 + Q ) exp T ln (1 + COV ) (1 + COV + +COV )  
2
R
2
QD
2
QL
field load tests conducted on various piles and installed in
different soils and countries has been compiled. The
foundation types covered include small to large
(2)
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

displacement piles (e.g. steel H-piles, torque-driven Butler, H. D., and Hoy, H. E. (1977) The Texas quick-load method
helical piles, driven cast-in-situ piles, and driven for foundation load testing – users manual, FHWA-IP-77-8,
closed/open-end concrete/steel piles) and non- Federal Highway Administration.
displacement piles (e.g. drilled shafts and ACIP piles). Byrne, B. W., and Houlsby, G. T. (2015) “Helical piles: an
4. The bias (mean) and dispersion (COV) of the model factor innovative foundation design option for offshore wind
for a range of foundations, geomaterials, and calculation turbines”. Philosophical Transactions of the Royal Society A:
models at the ultimate limit state have been summarized in Mathematical, Physical and Engineering Sciences, 373, Issue
tables suitable for adoption in design and codes of practice. 2035, pp1-11.
This effort constitutes the most significant update of Table Carter, J. P., and Kulhawy, F. H. (1988) Analysis and Design of
3.7.5.1 in the JCSS Probabilistic Model Code (2006) to Foundations Socketed into Rock, EPRI EL-5918, Electric
date. Power Research Institute.
5. These model statistics are expected to vary, depending on Cerato, A. B., Vargas, T. M., and Allred, S. M. (2017) “A critical
the failure criterion selected to interpret the “capacity” review: state of knowledge in seismic behaviour of helical
from measured load-movement curves, among a range of piles”. The Journal of the Deep Foundations Institute, 11, Issue
other factors. 1, pp39-87.
6. The random nature of the model factor can only be Chin, F. K. (1970) “Estimation of the ultimate load of piles not
addressed consistently using the reliability-based design carried to failure”, Proceedings of 2nd Southeast Asian
(RBD) method. The most popular implementation of RBD Conference on Soil Mechanics, Southeast Asian Society of
is the partial factor approach and the Load and Resistance Soil Engineering, Singapore, pp81-90.
Factor Design (LRFD). It is straightforward to include the Ching, J. Y., Lin, H-D., and Yen, M-T. (2011) “Calibrating
mean and COV of the model factor in LRFD. No resistance factors of single bored piles based on incomplete
probabilistic calculations are needed. load test results”. Journal of Engineering Mechanics, 137, Issue
5, pp309-323.
7. REFERENCES Davisson, M. T. (1972) “High capacity piles”, Proceedings of Soil
Mechanics Lecture Series on Innovations in Foundation
AbdelSalam, S., Sritharan, S., Suleiman, M., Ng, K., and Roling, M. Construction, pp81-112. Illinois section (ASCE), Chicago.
(2012) Development of LRFD design procedures for bridge Det Norske Veritas (DNV). (2017) Recommended Practice:
pile foundations in Iowa, Vol. 3: recommended resistant Statistical representation of soil data, DNVGL-RP-C207,
factors with consideration to construction control and setup, Oslo, Norway.
Report No. IHRB Projects TR-584, Iowa Department of Dithinde, M., Phoon, K. K., Wet, M., and Retief, J. (2011)
Transportation. “Characterization of model uncertainty in the static pile
AbdelSalam, S., Baligh, F., and El-Naggar, H. M. (2015) “A design formula”. Journal of Geotechnical Geoenvironmental
database to ensure reliability of bored pile design in Egypt.” Engineering, 137, Issue 1, pp70-85.
Proceedings of the Institution of Civil Engineers – Elsawy, M. K., El Naggar, M. H., Cerato, A., and Elgamal, A.
Geotechnical Engineering, 168, Issue 2, pp131-143. (2019) “Seismic performance of helical piles in dry sand from
Abu-Farsakh, M., Chen, Q. M., and Haque, M. N. (2013) large-scale shaking table tests”. Géotechnique, in press.
Calibration of resistance factors for drilled shafts for the new Eslami, A., and Fellenius, B. H. (1997) “Pile capacity by direct CPT
FHWA design method, Report No. FHWA/LA.12/495, and CPTu methods applied to 102 case histories”. Canadian
Louisiana Department of Transportation and Development. Geotechnical Journal, 34, Issue 6, pp886-904.
Abu-Farsakh, M., Yoon, S., and Tsai. C. (2009) Calibration of Fellenius, B. (1980) “The analysis of results from routine pile load
resistance factors needed in the LRFD design of driven piles, tests”. Ground Engineering, 13, Issue 6, pp19-31.
Report No. FHWA/LA.09/449, Louisiana Department of Gilbert, R. B., and Tang, W. H. (1995) “Model uncertainty in
Transportation and Development. offshore geotechnical reliability”. Offshore Technology
Abu-Hejleh, N., Abu-Farsakh, M., Suleiman, M., and Tsai, C. Conference, OTC-7757-MS, American Petroleum Institute.
(2015) “Development and use of high-quality databases of Hansen, J. B. (1963) “Discussion of ‘Hyperbolic stress-strain
deep foundation load tests”. Journal of the Transportation response: cohesive soils by Robert L. Kondner’.” Journal of
Research Board, No. 2511, pp27-36. Soil Mechanics and Foundation Division, 89, Issue 4, pp241-
Adhikari, P., Gebreslasie, Y., Ng, K., Sullivan, T., and Wulff, S. 242.
(2018) “Static and dynamic analysis of driven piles in soft International Code Council (ICC). (2018) International Building
rocks considering LRFD using a recently developed electronic Code. Washington DC.
database”, Proceedings of Installation, Testing, and Analysis of International Organization for Standardization (ISO). (2015)
Deep Foundations (GSP 294), pp83-92. Reston, VA: ASCE. General principles on reliability of structures, ISO 2394,
Al-Baghdadi, T. (2018) “Screw piles as offshore foundations: Geneva, Switzerland.
numerical and physical modelling”. Ph.D. thesis, University of International Organization for Standardization (ISO). (2016)
Dundee, UK. Petroleum and natural gas industries – specific requirements
American Association of State Highway and Transportation for offshore structures, part 4: geotechnical and foundation
Officials (AASHTO). (2014) LRFD Bridge Design design consideration, ISO 19901-4:2016, Geneva,
Specifications, 7th edition, p1704. Switzerland.
Asem, P., Long, J., and Gardoni, P. (2018) “Probabilistic model and Joint Committee on Structural Safety (JCSS). (2006) Probabilistic
LRFD resistance factors for the tip resistance of drilled shafts Model Code, Denmark.
in soft sedimentary rock based on axial load tests”, Proceedings Kulhawy, F. H., Phoon, K. K., and Wang, Y. (2012) “Reliability-
of Innovations in Geotechnical Engineering: Honoring Jean- based design of foundations – a modern view”, Geotechnical
Louis Briaud (GSP 299), pp1-46. Reston, VA: ASCE. Engineering State of the Art and Practice: Keynote Lectures
Brown, D. A., Turner, J. P., and Castelli, R. J. (2010) Drilled shafts: from GeoCongress 2012 (GSP 226), pp102-121. Reston, VA:
construction procedures and LRFD design methods, FHWA- ASCE.
NHI-10-016, Federal Highway Administration. Kyfor, Z. G., Schnore, A. R., Carlo, T. A., and Baily, P. F. (1992)
Burlon, S., Frank, R., Baguelin, F., Habert, J., and Legrand, S. Static testing of deep foundations, FHWA-SA-91-042. New
(2014) “Model factor for the bearing capacity of piles from York State Department of Transportation.
pressuremeter test results: Eurocode 7 approach”.
Géotechnique, 64, Issue 7, pp513-525.
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Lacasse, S., and Nadim, F. (1996) “Model uncertainty in pile axial Geotechnical Engineering: Computations and Applications,
capacity calculations”. Offshore Technology Conference, pp344–384. London, UK: Taylor & Francis group.
OTC-7996-MS, American Petroleum Institute. Phoon, K. K., and Tang, C. (2019a) “Characterisation of
Lacasse, S., Nadim, F., Andersen, K. H., Knudsen, S., Eidsvig, U. geotechnical model uncertainty”. Georisk: Assessment and
K., Yetginer, G., Guttormsen, T. R., and Eide, A. (2013a) Management of Risk for Engineered Systems and
“Reliability of API, NGI, ICP and Fugro axial pile capacity Geohazards, 13, Issue 2, pp101-130.
calculation methods”. Offshore Technology Conference, OTC Phoon, K. K., and Tang, C. (2019b) “Effect of extrapolation on
24063-MS, American Petroleum Institute. interpreted capacity and model statistics of steel H-piles”.
Lacasse, S., Nadim, F., Langford, T., Knudsen, S., Yetginer, G. L., Georisk: Assessment and Management of Risk for
Guttormsen, T. R., and Eide, A. (2013b) “Model uncertainty in Engineered Systems and Geohazards, under review.
axial pile capacity design methods”. Offshore Technology Randolph, M. F. 2003. “Science and empiricism in pile foundation
Conference, OTC-24066-MS, American Petroleum Institute. design”. Géotechnique, 53, Issue 10, pp847-875.
Lambe, T. W. (1973) “Predictions in soil engineering”. Reddy, S., and Stuedlein, A. (2017) “Ultimate limit state reliability-
Géotechnique, 23, Issue 2, pp151-202. based design of augered cast-in-place piles considering lower-
Lehane, B. M., Kim, J. K., Carotenuto, P., Nadim, F., Lacasse, S., bound capacities”. Canadian Geotechnical Journal, 54, Issue
Jardine, R. J., and Van Dijk B. F. J. (2017) “Characteristics of 12, pp1693-1703.
unified databases for driven piles”, Proceedings of 8th Reese, L. C., and Wright S. J. (1977) Construction Procedures and
International Conference of Offshore Site Investigation and Design for Axial Loading, Drilled Shaft Manual HDV-22.
Geomechanics, vol. 1, pp162-191. London, UK: Society for Roling, M., Sritharan, S., and Suleiman, M. (2011) Development of
Underwater Technology. LRFD procedures for bridge pile foundations in Iowa. Vol. 1:
Lesny, K. (2017) “Evaluation and consideration of model An electronic database for pile load tests (PILOT), IHRB
uncertainties in reliability based design”. Joint ISSMGE TC Project TR-573, Iowa Department of Transportation.
205/TC 304 Working Group on “Discussion of Sakr, M. (2018) “Performance of laterally loaded helical piles in
Statistical/Reliability Methods for Eurocodes.” London, UK: clayey soils established from field experience”. The Journal
International Society for Soil Mechanics and Geotechnical of the Deep Foundations Institute, 12, Issue 1, pp28-41.
Engineering. Salgado, R., Woo, S. I., and Kim, D. (2011) Development of load
Long, J., and Anderson, A. (2014) Improved design for driven piles and resistance factor design for ultimate and serviceability limit
based on a pile load test program in Illinois: phase 2, FHWA- states of transportation structure foundations, FHWA/IN/JTRP-
ICT-14-019, Illinois Department of Transportation. 2011/03, Indiana Department of Transportation.
Lutenegger, A. J. (2017) “Support of offshore structures using Seo, H., Moghaddam, R. B., Surles, J. G., and Lawson, W. D.
helical anchors”, Proceedings of 8th International Conference (2015) Implementation of LRFD geotechnical design for deep
on Offshore Site Investigation Geotechnics, pp995-1004. foundations using Texas penetrometer (TCP) test, FHWA/TX-
London, UK: Society for Underwater Technology. 16/5-6788-01-1. Texas Department of Transportation.
McVay, M., Wasman, S., Huang, L., and Crawford, S. (2016) “Load Stark, T., Long, J., and Asem, P. (2013) Improvement for
and resistance factor design (LRFD) resistance factors for determining the axial capacity of drilled shafts in shale in
auger cast in place piles”. Florida Department of Illinois, FHWA-ICT-13-017, Illinois Department of
Transportation. Transportation.
Motamed, R., S. Elfass, and K. Stanton. (2016) LRFD resistance Stark, T., Long, J., Baghdady, A., and Osouli, A. (2017) Modified
factor calibration for axially loaded drilled shafts in the Las standard penetration test-based drilled shaft design method for
Vegas Valley, Report No. 515-13-803, Nevada Department of weak rocks (phase 2 study), FHWA-ICT-17-018, Illinois
Transportation. Department of Transportation.
NeSmith, V. M., and Siegel, T. C. (2009) “Shortcomings of the Tang, C., and Phoon, K. K. (2018a) “Statistics of model factors and
Davisson offset limit applied to axial compressive load tests consideration in reliability-based design of axially loaded
on cast-in-place piles”, Proceedings of International helical piles”. Journal of Geotechnical Geoenvironmental
Foundation Congress and Equipment Expo (GSP 185), Engineering, 144, Issue 8, 04018050.
pp.568-574. Reston, VA: ASCE. Tang, C., and Phoon, K. K. (2018b) “Evaluation of model
Ng, K., Sritharan, S., and Ashlock, J. (2014) Development of uncertainties in reliability-based design of steel H-piles in
preliminary load and resistance factor design of drilled shafts in axial compression”. Canadian Geotechnical Journal, 55, Issue
Iowa, Report No. InTrans Project 11-410, Iowa Department of 11, pp1513-1532.
Transportation. Tang, C., and Phoon, K. K. (2018c) “Statistics of model factors in
Ng, T., and Fazia, S. (2012) “Development and validation of a reliability-based design of axially loaded driven piles in sand”.
unified equation for drilled shaft foundation design in New Canadian Geotechnical Journal, 55, Issue 11, pp1592-1610.
Mexico”. Report No. NM10MSC-01, New Mexico Department Tang, C., and Phoon, K. K. (2018d) “Characterization of model
of Transportation. uncertainty in predicting axial resistance of piles driven into
O’Neill, M., and Reese, L. (1999) Drilled Shafts: Construction clay”. Canadian Geotechnical Journal, in press.
Procedures and Design Methods, Federal Highway Tang, C., and Phoon, K. K. (2019) “Statistical evaluation of model
Administration. factors in reliability calibration of high displacement helical
O’Neill, M., Townsend, F., Hassan, K., Buller, A., and Chang, P. piles under axial loading”. Canadian Geotechnical Journal,
(1996) Load Transfer for Drilled Shafts in Intermediate under review.
Geomaterials, Federal Highway Administration. Tang, C., Phoon, K. K., and Chen, Y. J. (2019) “Evaluation and
Paikowsky, S., Birgisson, R., McVay, M., Nguyen, T., Kuo, C., consideration of model factors in reliability-based design of
Baecher, G., Ayyub, B., Stenersen, K., O’Malley, K., axially loaded drilled shafts”. Journal of Geotechnical and
Chernauskas, L., and O’Neill, M. (2004) Load and resistance Geoenvironmental Engineering, in press.
factor design (LRFD) for deep foundations, NCHRP Report Yang, Z. X., Guo, W. B., Jardine, R. J., and Chow, F. (2017)
507, Transportation Research Board. “Design method reliability assessment from an extended
Phoon, K. K., and Kulhawy, F. H. (1999) “Characterization of database of axial load tests on piles driven in sand”. Canadian
geotechnical variability”. Canadian Geotechnical Journal, 36, Geotechnical Journal, 54, Issue 1, pp59-74.
Issue 4, pp 612-624. Zhang, L. M., and Chu, L. F. (2009) “Calibration of methods for
Phoon, K. K., and Kulhawy, F. H. (2008) “Serviceability-Limit designing large-diameter bored piles: ultimate limit state”.
State Reliability-Based Design.” Reliability-based Design in Soils and Foundations, 49, Issue 6, pp883-895.
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Table 4 Statistics of the capacity model factor for static design methods of driven piles under axial loading (Source: Phoon and Tang 2019a).
Soil type Pile tip Material/shape Load type N Design method Mean COV Reference
Clay Closed/open-end Concrete Compression 18 λ-method 0.76 0.29 Paikowsky et al. (2004)
17 α-API 0.81 0.26
8 β-method 0.81 0.51
18 α-Tomlinson 0.87 0.48
Closed/open-end Steel pipe Compression 18 α-Tomlinson 0.64 0.5
19 α-API 0.79 0.54
12 β-method 0.45 0.6
19 λ-method 0.67 0.55
12 SPT-97 0.39 0.62
Closed/open-end Concrete/steel Compression/Tension 49 API-00 1.54 0.33 Lehane et al. (2017)
49 Fugro-96 1.21 0.24
49 ICP-05 0.86 0.45
49 NGI-05 1.17 0.33
49 UWA-13 1.14 0.25
49 Fugro-10 1.01 0.31
Closed-end Concrete/steel pipe Compression 115 ISO 19901-4:2016 1.08 0.34 Tang and Phoon (2018d)
115 NGI-05 1.05 0.31
115 SHANSEP 1.06 0.32
115 ICP-05 1.08 0.33
Tension 32 ISO 19901-4:2016 0.92 0.27
32 NGI-05 0.96 0.26
32 SHANSEP 1.11 0.29
32 ICP-05 1.08 0.29
Open-end Steel pipe Compression 60 ISO 19901-4:2016 0.97 0.3
60 NGI-05 1.03 0.25
60 SHANSEP 1.16 0.25
60 ICP-05 1 0.24
Tension 32 ISO 19901-4:2016 0.85 0.3
32 NGI-05 1.01 0.27
32 SHANSEP 0.96 0.3
32 ICP-05 0.97 0.39
Open-end Steel-H Compression 4 β-method 0.61 0.61 Paikowsky et al. (2004)
16 λ-method 0.74 0.39
17 α-Tomlinson 0.82 0.4
16 α-API 0.9 0.41
8 SPT-97 1.04 0.41
Open-end Steel-H Compression 20 α-API 1.15 0.52 AbdelSalam et al. (2012)
Open-end Steel-H Compression 26 α-API 1.1 0.4 Tang and Phoon (2018b)
Closed-end Helical pile (SS) (d≤44 mm, n=1) Compression 16 Torque-capacity correlation 0.88 0.15 Tang and Phoon (2018a)
Tension 14 Torque-capacity correlation 0.74 0.27
Helical pile (SS) (d≤44 mm, n>1) Compression 14 Torque-capacity correlation 1.04 0.19
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Clay Closed-end Helical pile (SS) (d≤44 mm, n>1) Tension 10 Torque-capacity correlation 0.93 0.26 Tang and Phoon (2018a)
Compression 49 Individual plate bearing 1.25 0.41
Open-end Helical pile (RS) (73≤d≤114 mm, n=1) Compression 75 Torque-capacity correlation 1.09 0.26
Tension 54 Torque-capacity correlation 0.92 0.23
Helical pile (RS) (73≤d≤114 mm, n>1) Compression 71 Torque-capacity correlation 1.16 0.18
Tension 69 Torque-capacity correlation 1.02 0.27
Open-end Helical pile (RS) (d≥219 mm, n=1) Compression 11 CFEM-2006 1.27 0.17 Tang and Phoon (2019)
ISHF-2015 0.98 0.13
ISO 19901-4:2016 1.14 0.13
Tension 4 CFEM-2006 1.43 0.15
ISHF-2015 1.05 0.13
ISO 19901-4:2016 1.25 0.16
Helical pile (RS) (d≥219 mm, n>1) Compression 23 CFEM-2006 1.16 0.26
ISHF-2015 0.97 0.24
ISO 19901-4:2016 1.03 0.23
Tension 13 CFEM-2006 1.08 0.32
ISHF-2015 0.93 0.31
ISO 19901-4:2016 0.99 0.31
Sand Closed/open-end Concrete Compression 36 Nordlund 1.02 0.48 Paikowsky et al. (2004)
35 β-method 1.1 0.44
36 Meyerhof 0.61 0.61
36 SPT-97 1.21 0.47
Steel pipe Compression 19 Nordlund 1.48 0.52
20 β-method 1.18 0.62
20 Meyerhof 0.94 0.59
19 SPT-97 1.58 0.52
Closed/open-end Concrete/steel Compression/Tension 71 API-00 1.66 0.56 Lehane et al. (2017)
71 Fugro-05 0.99 0.4
71 ICP-05 1.04 0.27
71 NGI-05 0.99 0.34
71 UWA-05 1.06 0.27
71 ICP-API 1.2 0.31
71 UWA-OS 1.28 0.29
Open-end Concrete/steel pipe Compression 16 ICP-05 1.07 0.24 Tang and Phoon (2018c)
16 UWA-05 1.07 0.21
Closed-end Concrete/steel pipe Compression 52 ICP-05 1.1 0.31
52 UWA-05 1 0.39
Open-end Steel pipe Tension 19 ICP-05 1.36 0.38
19 UWA-05 1.3 0.37
Closed-end Steel pipe Tension 9 ICP-05 1.02 0.35
9 UWA-05 1.02 0.37
Sand Open-end Steel-H Compression 19 Nordlund 0.94 0.4 Paikowsky et al. (2004)
18 Meyerhof 0.81 0.38
19 β-method 0.78 0.51
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Sand Open-end Steel-H Compression 18 SPT-97 1.35 0.43 Paikowsky et al. (2004)
Open-end Steel-H Compression 34 Nordlund 0.92 0.53 AbdelSalam et al. (2012)
Open-end Steel-H Compression 46 Nordlund 0.82 0.47 Tang and Phoon (2018b)
Closed-end Helical pile (SS) (d≤44 mm, n=1) Compression 6 Torque-capacity correlation 1.51 0.39 Tang and Phoon (2018a)
Tension 7 Torque-capacity correlation 1.2 0.56
Helical pile (SS) (d≤44 mm, n>1) Compression 10 Torque-capacity correlation 1.54 0.39
Tension 10 Torque-capacity correlation 1.06 0.22
Compression 55 Individual plate bearing 1.46 0.42
Open-end Helical pile (RS) (73≤d≤114 mm, n=1) Compression 50 Torque-capacity correlation 1.23 0.37
Tension 47 Torque-capacity correlation 0.98 0.3
Helical pile (RS) (73≤d≤114 mm, n>1) Compression 49 Torque-capacity correlation 1.51 0.26
Tension 51 Torque-capacity correlation 1.2 0.24
Open-end Helical pile (RS) (d≥219 mm, n=1) Compression 5 CFEM-2006 1.39 0.64 Tang and Phoon (2019)
ISHF-2015 1.57 0.62
ISO 19901-4:2016 1.25 0.55
Tension 5 CFEM-2006 1.41 0.42
ISHF-2015 1.6 0.39
ISO 19901-4:2016 1.41 0.33
Helical pile (RS) (d≥219 mm, n>1) Compression 8 CFEM-2006 1.2 0.37
ISHF-2015 1.4 0.36
ISO 19901-4:2016 1.11 0.45
Tension 9 CFEM-2006 1.31 0.38
ISHF-2015 1.52 0.37
ISO 19901-4:2016 1.21 0.33
Soft rock Open-end Steel-H Compression 23 Nordlund/α-method 1.2 1.6 Adhikari et al. (2018)
β/α-method 1.5 1.2
Meyerhof/α-method 3.1 1.2
Mixed Closed/open-end Concrete Compression 33 α-Tomlinson/Nordlund 0.96 0.49 Paikowsky et al. (2004)
80 α-API/Nordlund 0.87 0.48
80 β-method/Thurman 0.81 0.38
71 SPT-97 1.81 0.5
30 FHWA CPT 0.84 0.31
Steel pipe Compression 13 α-Tomlinson/Nordlund 0.74 0.59
32 α-API/Nordlund 0.8 0.45
29 β-method/Thurman 0.54 0.48
33 SPT-97 0.76 0.38
Open-end Steel-H Compression 20 α-Tomlinson/Nordlund 0.59 0.39
34 α-API/Nordlund 0.79 0.44
32 β-method/Thurman 0.48 0.48
40 SPT-97 1.23 0.45
Open-end Steel-H Compression 26 α-API/Nordlund 1.04 0.4 AbdelSalam et al. (2012)
Open-end Steel-H Compression 32 α-API/Nordlund 0.92 0.4 Tang and Phoon (2018b)
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Note: API = American Petroleum Institute; ICP = Imperial College Pile; UWA = University of Western Australia; NGI = Norwegian Geotechnical Institute; SHANSEP = Stress History and Normalized Soil
Engineering Properties; CFEM = Canadian Foundation Engineering Manual; ISHF = International Society for Helical Foundation; d = shaft width or diameter of helical piles; n = number of helices in
helical piles; SS = square shaft; RS = round shaft.

Table 5 Statistics of the capacity model factor for static design methods of drilled shafts under axial loading (Source: Phoon and Tang 2019a).
Load type Soil type N Design model Construction method Mean COV Reference
Compression Sand 30 Static formula ─ 0.98 0.24 Dithinde et al. (2011)
Clay 53 1.15 0.25
Compression Sand/silt 11 O’Neill and Reese (1999) Casing 0.6 0.58 Zhang and Chu (2009)
17 FHWA (Hong Kong data) 1.06 0.28
Rock 15 O’Neill and Reese (1999) RCD 0.48 0.52
15 COP (BD 2004) 2.57 0.31
Compression Sand 32 O’Neill and Reese (1999) Mixed 1.71 0.60 Paikowsky et al. (2004)
12 Casing 2.27 0.46
9 Slurry 1.62 0.74
32 Reese and Wright (1977) Mixed 1.22 0.67
12 Casing 1.45 0.5
9 Slurry 1.32 0.62
Clay 53 O’Neill and Reese (1999) Mixed 0.9 0.47
13 Casing 0.84 0.50
40 Dry 0.88 0.48
Clay+sand 44 O’Neill and Reese (1999) Mixed 1.19 0.30
21 Casing 1.04 0.29
12 Dry 1.32 0.28
10 Slurry 1.29 0.27
44 Reese and Wright (1977) Mixed 1.09 0.35
21 Casing 1.01 0.42
12 Dry 1.2 0.32
10 Slurry 1.16 0.25
Rock 46 Carter and Kulhawy (1988) Mixed 1.23 0.41
29 Dry 1.29 0.4
46 O’Neill et al. (1996) Mixed 1.3 0.34
29 Dry 1.35 0.31
Tension Sand 11 O’Neill and Reese (1999) ─ 1.09 0.51 Paikowsky et al. (2004)
11 Reese and Wright (1977) 0.83 0.54
Clay 13 O’Neill and Reese (1999) 0.87 0.37
Clay+sand 14 O’Neill and Reese (1999) 1.25 0.29
14 Reese and Wright (1977) 1.24 0.41
All soil 39 O’Neill and Reese (1999) 1.08 0.41
25 Reese and Wright (1977) 1.07 0.48
Rock 16 Carter and Kulhawy (1988) 1.18 0.46
16 O’Neill et al. (1996) 1.25 0.37
Compression Mixed 34 O’Neill and Reese (1999) ─ 1.27 0.3 Abu-Farsakh et al. (2013)
 1st Malaysian Geotechnical Society (MGS) and Geotechnical Society of Singapore (GeoSS) Conference 2019, Petaling Jaya, Malaysia, 24-26 June 2019

Compression Mixed 34 Brown et al. (2010) ─ 0.99 0.3 Abu-Farsakh et al. (2013)
Compression Sand 24 O’Neill and Reese (1999) ─ 1.14 0.58 Ng and Fazia (2012)
24 Brown et al. (2010) 1.21 0.6
Compression Mixed 11 O’Neill and Reese (1999) ─ 1.18 0.16 Ng et al. (2014)
Compression Clay 22 Brown et al. (2010) ─ 1.02 0.41 AbdelSalam et al. (2015)
Sand 45 0.91 0.4
Mixed 90 0.81 0.37
Tension Clay 32 Brown et al. (2010) ─ 1 0.34 Tang et al. (2019)
Sand 30 1.14 0.32
Gravelly soil 49 0.84 0.5
Compression Clay 64 Brown et al. (2010) ─ 1.15 0.5
Sand 44 0.99 0.38
Gravelly soil 41 1.18 0.53
Note: FHWA = Federal Highway Administration; COP = code of practice for foundation; and BD = Building Department of Hong Kong.

Table 6 Summary of capacity model statistics for various foundation types (Source: Phoon and Tang 2019a).
No. of tests per group Mean COV
Pile type Load type Geomaterial No. data groups Range Mean Design method Range Mean Range Mean
Driven piles Compression Clay 26 4 – 115 28 α-method 0.39 – 1.54 0.97 0.13 – 0.62 0.34
Sand 24 5 – 71 29 β-method 0.61 – 1.66 1.18 0.21 – 0.64 0.43
Mixed 17 13 – 80 40 α/β-method 0.48 – 1.81 0.91 0.31 – 0.59 0.45
Tension Clay 8 4 – 69 28 α-method 0.74 – 1.43 1.01 0.13 – 0.39 0.27
Sand 8 5 – 51 20 β-method 0.98 – 1.6 1.26 0.22 – 0.56 0.36
Drilled shafts Compression Clay 6 13 – 64 41 α-method 0.84 – 1.15 0.99 0.25 – 0.5 0.44
Sand 11 9 – 46 30 β-method 0.48 – 2.57 1.35 0.24 – 0.74 0.48
Mixed 9 10 – 90 28 α/β-method 0.6 – 1.32 1.09 0.16 – 0.58 0.32
Tension Clay 2 13 – 32 22 α-method 0.87 – 1 0.94 0.34 – 0.37 0.36
Sand 4 11 – 49 26 β-method 0.83 – 1.25 1.06 0.32 – 0.54 0.45
Mixed 3 14 - 39 26 α/β-method 1.07 – 1.25 1.16 0.29 – 0.48 0.4