You are on page 1of 19

Statistical Analyses of Model Factors in Reliability-Based

Limit-State Design of Drilled Shafts under Axial Loading


Chong Tang 1; Kok-Kwang Phoon, F.ASCE 2; and Yit-Jin Chen 3
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This study compiles 320 static-load tests to quantify the model factors in reliability-based limit-state design of drilled shafts under
axial loading. At ultimate limit-state, the model factor is defined as the ratio of the measured capacity to the calculated capacity. It character-
izes the bias in capacity calculation. The measured capacity is interpreted from load test data by the modified Davisson offset limit, and
current design methods are utilized to compute the ultimate axial capacity. The load-displacement data are simulated by two-parameter
hyperbolic curves. Based on the database, the statistics and probability distributions of the capacity and load-displacement model factors
are established. Several copulas are selected for goodness-of-fit tests on the observed correlation of the hyperbolic model factors. The model
factor at serviceability the limit-state is then described by two correlated hyperbolic parameters. Finally, the model statistics are applied to
implement the LRFD of drilled shafts under axial loading by Monte Carlo simulations. DOI: 10.1061/(ASCE)GT.1943-5606.0002087.
© 2019 American Society of Civil Engineers.

Introduction respecting the diversity of practical design procedures (Phoon et al.


2016). The intended use of ISO 2394 is to serve as a unified frame-
Various sources of uncertainties are encountered within geotechni- work for drafting structural and geotechnical design codes where
cal design. All uncertainties are considered by assuming a factor of reliability theory is employed. Phoon and Retief (2016) provided
safety. This design philosophy corresponds to the allowable stress detailed guidance on how to draft geotechnical design codes in
design (ASD) method. The limitations of ASD were discussed by accordance with the reliability principles in ISO 2394. Phoon
Kulhawy and Phoon (2006, 2009) and Kulhawy (2010), which can (2017) further discussed how reliability principles can be integrated
be resolved with reliability-based design (RBD). Compared with within the prevailing norms of geotechnical practice.
ASD, RBD can ensure a compatible level of reliability between In the opinion of the authors, the first and foremost element is to
structure and foundation. In this context, probabilistic methods and characterize the variability of geotechnical properties, and the sec-
reliability theory have been increasingly applied in geotechnical ond one is to quantify the model uncertainty (Phoon 2016). The key
engineering (e.g., Casagrande 1965; Whitman 1984; Duncan 2000; features of the first element are (1) determination of the coefficient
Baecher and Christian 2003; Fenton and Griffiths 2008; Phoon of variation (COV) of a geotechnical parameter that does not take
2008; Phoon and Ching 2014; Phoon and Retief 2016). a unique value because of spatial variability (Phoon and Kulhawy
During the last decades, geotechnical design codes have been 1999) and (2) multivariate correlations of soil/rock parameters that
gradually migrating from ASD to RBD (Fenton et al. 2016). can be used to reduce the variability in estimating a geotechnical
Through significant efforts, a number of geotechnical RBD codes parameter (e.g., Ching et al. 2010; Ching and Phoon 2012). The
were developed worldwide, such as Geo-Code 21 (Honjo and second element is frequently quantified as the ratio of the measured
Kusakabe 2002) in Japan, Eurocode 7 (Orr and Farrell 1999) in response to the calculated response (i.e., model factor) as suggested
Europe, the load and resistance factor design (LRFD) specifications by the Joint Committee on Structural Safety (JCSS 2001) and ISO
adopted by the American Association of State Highway and Trans-
2394. The need for model uncertainty assessment has been empha-
portation Officials (AASHTO 2014) in the United states, and the
sized by the Eurocode 7 drafters, who are involved in the current
Canadian Highway Bridge Design Code (Canadian Standards
revision process of the Eurocodes (Lesny 2017). Reliability-based
Association 2014). The fourth edition of ISO 2394 (ISO 2015)
sensitivity analyses of vertically loaded piles were conducted by
contains an informative Annex D on “Reliability of Geotechnical
Teixeira et al. (2012). The results demonstrated that (1) model un-
Structures.” It emphasizes the identification and characterization
certainty has a significant effect on the reliability of piles in spite of
of the critical elements in a geotechnical RBD procedure while
the high soil variability, and (2) model uncertainty is more influ-
1 ential than soil variability in two cases studied.
Research Fellow, Dept. of Civil and Environmental Engineering,
National Univ. of Singapore, Block E1A, #07-03, 1 Engineering Dr. 2, With the guideline of Annex D in ISO 2394, the primary aim of
Singapore 117576 (corresponding author). ORCID: https://orcid.org/0000 this paper is to statistically quantify the model factors in reliability-
-0002-8415-2487. Email: ceetc@nus.edu.sg based limit-state designs of drilled shafts under axial loading in
2
Professor, Dept. of Civil and Environmental Engineering, National clay, sand, and gravel. First, RBD models for ultimate limit-state
Univ. of Singapore, Block E1A, #07-03, 1 Engineering Dr. 2, Singapore (ULS), related to capacity, and serviceability limit-state (SLS),
117576. Email: kkphoon@nus.edu.sg related to displacement, are introduced to show the consideration
3
Professor, Dept. of Civil Engineering, Chung Yuan Christian Univ., of model factors. Second, an integrated and comprehensive drilled-
Chung-Li 32023, Taiwan. Email: yjc@cycu.edu.tw
shaft load test database is developed. The database contains 320
Note. This manuscript was submitted on May 8, 2018; approved on
February 5, 2019; published online on June 27, 2019. Discussion period static-load tests compiled from several existing databases and
open until November 27, 2019; separate discussions must be submitted the literature. An interpretation method is used to define the
for individual papers. This paper is part of the Journal of Geotechnical measured capacity. A hyperbolic model with two parameters is
and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. employed to simulate the measured load-displacement curves that

© ASCE 04019042-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


are normalized by the measured capacity. Third, the design manual Serviceability Limit-State
recommended by the Federal Highway Administration (FHWA)
SLS occurs when foundation settlement s exceeds a predefined set-
(Brown et al. 2010) is used to calculate the ultimate axial capacity.
tlement sa . It requires the characterization of the ratio of the mea-
Fourth, rigorous analyses are carried out to perform statistical
sured displacement to the calculated displacement (e.g., Zhang
evaluation of the model factors. The results are discussed and com-
et al. 2008; Zhang and Chu 2009a). The randomness in the permis-
pared with those available in literature. Finally, the model statistics
sible settlement sa may also be required. Zhang and Ng (2005)
are applied to calculate the ULS and SLS resistance factors in
showed that the sa value is associated with the type and size of
LRFD of drilled shafts under axial loading by reliability analysis
structures, structural materials, and the underlying soils. Reliability
methods, where the load factors and statistics in the AASHTO
analyses with this approach were implemented by Zhang and Chu
LRFD Bridge Design Specifications (AASHTO 2014) are used.
(2009a, c). On the other hand, SLS can be equivalently specified as
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

the case when the applied load exceeds the permissible value Qa
(Phoon and Kulhawy 2008)
Reliability-Based Limit-State Design Models
pf ¼ PrðQa − Q ≤ 0Þ ≤ pfT ð3Þ
In foundation design, two typical limit-states (i.e., ULS and SLS)
are considered (Paikowsky et al. 2004). ULS involves the total col- where Qa depends on sa . Eq. (3) corresponds to the load method,
lapse of a structure and provides the proper structural safety. With which is applied similar to ULS. It gives a permissible capacity that
SLS, a structure should remain functional for its intended use. The is familiar to engineers (e.g., Phoon and Kulhawy 2008; Misra and
following sections summarize the RBD models for ULS and SLS. Roberts 2009; Moghaddam 2016). The factored capacity approach
is employed in this work, which necessitates the evaluation of the
Ultimate Limit-State SLS capacity statistics.
Phoon and Kulhawy (2008) demonstrated an alternate approach
ULS can be typically defined as the case when the applied load Q is to correlate Qa with sa . It is based on parameterization of the mea-
greater than the ultimate capacity Ru. The probability of failure pf sured load-displacement curve. This involves fitting the curve to a
at the ULS can be concisely expressed as follows (Phoon and simple model such as hyperbolic and power law, thus replacing
Kulhawy 2008): the continuous curve by two model parameters. Previous studies
(e.g., Phoon et al. 2006, 2007; Phoon and Kulhawy 2008; Dithinde
pf ¼ PrðRu − Q ≤ 0Þ ≤ pfT ð1Þ et al. 2011; Stuedlein and Reddy 2013; Reddy and Stuedlein 2017a;
Tang and Phoon 2018a, b, d) showed that the following hyperbolic
model can reasonably simulate the load-displacement curves of
where Prð·Þ = probability of an event; and pfT = target value for the piles:
probability of failure. The reliability analysis in Eq. (1) requires the
load and capacity statistics. The focus of the current work is to char- Q s
¼ ð4Þ
acterize the capacity statistics that are calculated according to Rum a þ b · s

M ¼ Rum =Ruc ð2Þ where ða; bÞ = hyperbolic parameters, which are calculated by
least-squares regression of the measured load-displacement curve.
Taking the limit of Eq. (4) as s goes to infinitely large, it has
where Rum = measured capacity; Ruc = calculated capacity; and
M u = model factor that quantifies the deviation of the calculated Q s 1
lim ¼ lim ¼ ð5Þ
capacity from the measured capacity. s→∞ Rum s→∞ a þb·s b
The model factor M u is not constant. The simplest method to
evaluate M u is to compute the mean (bias) and COV (dispersion). Differentiating Eq. (4) with respect to s and taking the limit as s
A mean and COV of M u close to 1 and 0 would represent a near approaches to zero, the initial stiffness K si is determined
perfect calculation method. Unfortunately, such a method does not
exist in geotechnical engineering, regardless of the method’s so- a 1
phistication. The random nature of Mu is practically significant, K si ¼ lim ¼ ð6Þ
because it indicates that a calculation method could be unconser-
s→0 ða þ b · sÞ2 a
vative even though the method is conservative on average. A better
method is to consider the degree of scatter in M u and ensure that the Eqs. (5) and (6) indicate that the reciprocals of a and b corre-
probability of a measured capacity being lower than the calculated spond to the initial slope and the asymptote of the normalized load-
capacity is capped at a known value. This is conceptually similar to displacement curve, respectively. For a given sa , Qa is calculated
Eurocode 7 (Lesny 2017).
The model factor M u has been extensively characterized and Qa ¼ M s · Rum ð7Þ
applied to reliability-based ultimate limit-state design of (1) spread
footings (Paikowsky et al. 2010); (2) replacement piles (e.g., Kuo where M s = SLS model factor defined as follows:
et al. 2002; Paikowsky et al. 2004; Zhang and Chu 2009b;
sa
Abu-Farsakh et al. 2010, 2013; Yang et al. 2010; Ng and Fazia Ms ¼ ð8Þ
2012; Stuedlein et al. 2012; Ng et al. 2014; AbdelSalam et al. 2015; a þ b · sa
McVay et al. 2016; Motamed et al. 2016; Reddy and Stuedlein
2017b); (3) small displacement piles (e.g., Paikowsky et al. 2004; When site-specific load tests are unavailable, Rum is usually
AbdelSalam et al. 2012; Tang and Phoon 2018b, c); and (4) large unknown at the design stage. It is convenient to replace Rum with
displacement piles (e.g., Paikowsky et al. 2004; Abu-Farsakh et al. Ruc , and the deviation M u in capacity calculation should be incor-
2009; Tang and Phoon 2018a, c). porated into Eq. (7)

© ASCE 04019042-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Qa ¼ M s · M u · Ruc ð9Þ They are based on the classification of pile type and soil profile:
(1) model factors are evaluated for the main relevant soil conditions
The load-displacement model factors have been statistically with considering the differences in pile installation and construc-
characterized for (1) spread footings (e.g., Uzielli and Mayne 2011; tion procedure, pile tip and material; (2) model factors are quanti-
Huffman and Stuedlein 2014; Huffman et al. 2015; Najjar et al. fied with similar soil and pile types grouped together; and (3) model
2014, 2017); (2) replacement piles (e.g., Stuedlein and Reddy factors are characterized for all piles in all soils, which leads to a
2013; Reddy and Stuedlein 2017a); (3) small displacement piles single model statistics. The first approach is ideal, but it needs a
(Tang and Phoon 2018b, d); and (4) large displacement piles (Tang significant number of load tests on each pile type for each geoma-
and Phoon 2018a). A very comprehensive summary of model sta- terial type. The second method requires a fairly comprehensive
tistics has been given by Phoon and Tang (2019). database, which is employed here. It does not differentiate the con-
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

struction procedure of drilled shafts (e.g., dry casing and wet). The
soil category for each site is determined based on the predominant
Drilled-Shaft Load Test Database soil type (i.e., clay, sand, and gravel) that exists along more than
70% of the profile. A site with a predominant soil type less than
The importance of geotechnical data was early emphasized by
70% is then considered as layered (AbdelSalam et al. 2012).
Lambe (1973), who opined that the quality of geotechnical calcu-
lation can be optimal for a particular combination of method and
data. Increasing the quality of data or method alone beyond this Interpretation of Drilled-Shaft Load Tests
“best” combination may not result in an improvement of the quality
of prediction (Lambe 1973). A more recent development is the The load-displacement curves for drilled shafts are illustrated in the
gradual adoption of RBD, which is more sensitive to information/ left-hand plots of Figs. 1(a–c). Pile plunge, where rapid movement
data (Fenton et al. 2016). Phoon and Kulhawy (2005) stated that a occurs under a small increment in applied load, is observed in many
comprehensive database with well documented field and laboratory load tests. The ultimate capacity can be taken as the load at deflec-
test data will be a good tool for model uncertainty evaluation. The tions near plunge, which is similar to the L1 –L2 method of Hirany
whole geotechnical community should actively work to collect and and Kulhawy (2002). This definition in general requires a large dis-
analyze a large number of full-scale load test data to reveal the placement. In the other cases, where the plunging mode is not ob-
underlying pattern, trend, and association. served, the ultimate capacity must be determined by a definition
based on the measured load-displacement data. For a standardized
evaluation of model uncertainty, the measured capacity in Eq. (2) is
Collection of Static-Load Test Data taken as an interpreted value from the measured load-displacement
Abu-Hejleh et al. (2015) suggested that the following information curve.
should be contained within a load test database for evaluation of Various methods have been proposed to define the measured
design methods and reliability calibration: (1) subsurface data at or capacity. Section 1810.3.3.1.3, “Load Test Evaluation Methods,”
around a test foundation to classify geomaterial type and determine in the International Building Code (ICC 2018) explicitly lists three
geomaterial parameters; (2) test foundation data such as embed- methods, where the Davisson offset limit (Davisson 1972) is widely
ment depth, width or diameter, material, and installation method; used in North America. This method was proposed by comparing
and (3) load-displacement data. Several load test databases for the results of wave equation analyses of driven steel pipes tested in
drilled shafts are available from Wysockey (1999), Garder et al. accordance with quick methods. The load at the intersection of the
(2012), Lin et al. (2012), AbdelSalam et al. (2015), and Abu-Hejleh load-displacement curve with an elastic line offset by 3.8 mm plus
et al. (2015). An integrated database is developed for this study, the soil quake (pile diameter divided by 120) is identified as the
with load test data compiled from the following sources: measured capacity (NeSmith and Siegel 2009). Because the quake
• FHWA deep foundation load test database (DFLTD) (Abu- offset is much larger in drilled shafts, the Davisson offset limit
Hejleh et al. 2015). This is the most comprehensive database produces smaller axial capacities compared with other methods
in the United States, containing 293 compression and 12 uplift (e.g., Ng et al. 2001; Hirany and Kulhawy 2002; Chen et al. 2008;
tests on drilled shafts. Chen and Fang 2009; Chen and Chu 2012; Stuedlein et al. 2014).
• Web-based pile-load test database (WBPLT) (Chen et al. 2014). More detailed fallacies in applying the Davisson offset limit to
This was developed by Chen et al. (2008), Chen and Fang replacement piles were discussed by Kulhawy and Chen (2005) and
(2009), and Chen and Chu (2012), with 272 compression and NeSmith and Siegel (2009).
169 uplift tests on drilled shafts worldwide. For drilled shafts, Davisson (1993) suggested the offset dis-
• Niazi (2014) summarized 330 pile load tests at 70 sites from 5 placement should be increased by a factor varying from 2 to 6. This
continents and 19 different countries of the world to evaluate the is because the tip deflection of 2%–5% of the base width is required
response of axially loaded piles using seismic piezocone data. to reach the ultimate capacity (Brown et al. 2010). The modified
Thirty-four static-load tests on drilled shafts are selected. Davisson offset limit is thus adopted in this paper, where the offset
• Qian et al. (2014, 2015) performed 60 uplift load tests to in- is taken as the shaft diameter divided by 30 (Kyfor et al. 1992).
vestigate the uplift behavior of drilled piers in northwest China Previous studies showed that the modified Davisson offset limit
that are used as foundations for transmission-line structures. will give a larger drilled-shaft capacity (e.g., Ng et al. 2001;
The dominant soil type is gobi gravel, which is classified as Stuedlein et al. 2014). Extrapolation is not recommended for model
cohesionless. calibration when a load test is terminated at insufficient settlement
level to determine the measured capacity (Lesny 2017). This is be-
cause (1) extrapolation could overpredict the pile capacity as high
Classification of Load Test Data and Geomaterial
as 50% (e.g., Paikowsky and Tolosko 1999; Hasan et al. 2018),
Categorization
and (2) an established guideline for extrapolation is unavailable
Bauduin (2002) discussed three philosophies for using a load (NeSmith and Siegel 2009).
test database to evaluate the model uncertainty in pile design, A filtered database is presented in Tables S1–S5, which contains
which lead to different quality levels in the derived model statistics. 320 load tests. Only tests that were loaded to failure in accordance

© ASCE 04019042-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Original Normalized
15000 2.5

2
10000
1.5

Q (kN)

Q/Rum
1
5000
0.5

0 0
0 50 100 150 200 250 0 50 100 150 200 250
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

s (mm) s (mm)
(a)

18000 2.5

2
12000
1.5
Q (kN)

Q/Rum
1
6000
0.5

0 0
0 50 100 150 200 250 0 50 100 150 200 250
s (mm) s (mm)
(b)
10000 1.5

8000
1
6000
Q (kN)

Q/Rum

4000
0.5
2000

0 0
0 20 40 60 80 100 0 20 40 60 80 100
s (mm) s (mm)
(c)

Fig. 1. Measured load-displacement curves of drilled shafts: (a) clay; (b) sand; and (c) gravel.

Table 1. Classification and ranges of the collected load tests


Shaft geometries
Soil Number of
profile Load type load tests (N) B (m) D (m) D=B su (kPa) ϕ (degrees)
Clay Compression 64 0.35–1.52 1.8–62 1.6–56 41–246 —
Uplift 32 0.36–1.8 3.1–60 3.4–55 21–250 —
Sand Compression 44 0.35–2 4.7–69 5.1–59 — 30–41
Uplift 30 0.3–1.31 1.4–26 2.5–43 — 30–45
Gravel Compression 41 0.59–1.5 4.7–30 6.2–30 — 37–47
Uplift 109 0.43–2.26 1.52–26 1.77–17.3 — 42–48

with the modified Davisson offset limit are compiled from the shaft geometries and soil properties are given in Table 1. Applica-
FHWA DFLTD, WBPLT, Qian et al. (2014, 2015), and Niazi tion of the derived model statistics to circumstances not covered by
(2014). This is why the number of selected drilled-shaft load tests the calibrate database should be verified.
is smaller than the number of load tests in these databases. The The original load-displacement curves in the left-hand plots
database is categorized into different groups, where each group re- of Fig. 1 are located within a wide range because of different
fers to a specific soil type and load type (32 uplift and 64 compres- drilled-shaft geometries and site profiles. When these curves are
sion tests in clay, 30 uplift and 44 compression tests in sand, and normalized by the interpreted capacities, they are located within a
109 uplift and 41 compression tests in gravel). The ranges of the narrower range as shown in the right-hand plots of Fig. 1. This is

© ASCE 04019042-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


because the interpreted capacities take into account many factors β ¼ ð1 − sin ϕÞ · ðσp0 =σv0 Þsin ϕ · tan ϕ ð13Þ
that could affect the load-displacement curves. Reduction in varia-
tion is essential to evaluate the load-displacement model factors where ϕ = friction angle of a cohesionless soil; σp0 = effective ver-
in Eq. (4). Similar observations were presented by Phoon and tical preconsolidation stress ¼ 0.47pa ðN 60 Þm for sandy soil with
Kulhawy (2008), Ching and Chen (2010), Dithinde et al. (2011), m ¼ 0.6 for clean quartzitic sand, m ¼ 0.8 for silty sand to sandy
Huffman et al. (2016), and Tang and Phoon (2018a, b, d). silt (Mayne 2007), and σp0 ¼ 0.15pa N 60 for gravelly soil (Kulhawy
and Chen 2007); and N 60 = blow count of standard penetration test
(SPT) corrected to 60% efficiency. The upper-bound value of β
FHWA Method for Drilled-Shaft Capacity is K p · tan ϕ, where K p is the coefficient of Rankine passive earth
pressure.
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

The ultimate axial capacity of a drilled shaft can be computed as the


summation of the limit side resistance Rs and ultimate base capac-
ity Rb (Salgado 2008) Base Capacity
The calculation of base capacity is given in Eq. (14) for cohesive
X
n soil and Eq. (15) for cohesionless soil (Brown et al. 2010)
Ruc ¼ Rb þ Rs ¼ Ab · rb þ π · B · Δzi · rsi ð10Þ
i¼1 rb ¼ N c · sub ð14Þ

where Ab = area of the shaft base; rb = unit base capacity; n = rb ¼ 0.06 · N 60 ≤ 2.9 MPa ð15Þ
number of distinct soil layers encountered in the foundation loca-
where sub = mean undrained shear strength over a depth of 2B
tion; B = shaft width or diameter; Δzi = thickness of the soil layer;
below the base; and N c = bearing capacity factor in Table 13-2 in
and rsi = unit side resistance within the soil layer. The underlying
Brown et al. (2010).
assumption of Eq. (10) is that the side and base resistances are
For axial uplift, suction could be developed between the base
mobilized simultaneously. Nevertheless, the displacement required
and underlying soils. However, it is very difficult to calculate
to develop the ultimate base capacity is much higher than that to
the tip suction accurately. Thus, the base capacity will not be
mobilize the full side resistance (Brown et al. 2010).
considered in the design procedure (Brown et al. 2010). The uplift
capacity is calculated in the same way for side resistance under
Side Resistance compression (AASHTO 2014).
The side resistance is often calculated using a total stress analysis in
Eq. (11) for cohesive soil and an effective stress method in Eq. (12)
Uncertainty in Calculation of Drilled-Shaft Capacities
for cohesionless soil (Brown et al. 2010)
The calculated capacities using the 2010 FHWA design method
rs ¼ α · sus ð11Þ (Brown et al. 2010) are reported in Appendix S1. Comparison
between the calculated and interpreted capacities is presented in
the plots on the left side of Fig. 2. A clear deviation is observed.
rs ¼ β · σv0 ¼ σv0 · K · tan δ ð12Þ The model factors appear to be independent of the interpreted
capacities, as shown in the plots on the right side of Fig. 2. Uncer-
where sus = average undrained shear strength; α = adhesion factor tainty in capacity calculation could be explained by the following
(e.g., Skempton 1959; Chen and Kulhawy 1994; O’Neill and Reese factors:
1999; Goh et al. 2005; Chakraborty et al. 2013); σv0 = average • A site profile is spatially variable even for a single and homo-
vertical effective; K = coefficient of lateral earth pressure (which geneous layer. This is because geomaterials are not manufac-
is a function of the in situ or at-rest value and changes in horizontal tured to achieve a desired level of quality, unlike concrete or
stress that occur during the construction of drilled shafts); δ = steel. A site profile that contains different strata is approximately
friction angle of the soil–shaft interface; and β = coefficient of the classified as cohesive or cohesionless by the 70% rule. Soil
side resistance (e.g., O’Neill and Reese 1999; Jamiolkowski 2003; parameters are determined by averaging along the shaft or
Rollins et al. 2005). around the base.
In the 2010 FHWA design manual (Brown et al. 2010), α ¼ • Empiricisms existing within the 2010 FHWA design method to
0.55 for sus =pa ≤ 1.5 and α ¼ 0.55 − 0.1ðsus =pa − 1.5Þ for 1.5 ≤ estimate the unit base and side resistances. The adhesion factor
sus ≤ 2.5, where pa is the atmospheric pressure (≈100 kPa). α and the coefficient K of lateral earth pressure are calibrated
Another commonly used α–su relation is α ¼ 0.21 þ 0.25pa = against load test data (typically limited). When the design sce-
sus ≤ 1 which was proposed by Kulhawy and Jackson (1989) based nario (e.g., shaft geometries and/or soil properties) is beyond
on 65 uplift and 41 compression field load tests on drilled shafts. the boundaries of the calibration database, additional deviation
This relation was adopted in the Fourth Canadian Foundation could be introduced.
Engineering Manual (Canadian Geotechnical Society 2006). The • Simplifications of the 2010 FHWA design method. For exam-
β coefficient in O’Neill and Reese (1999) is recommended in the ple, the side resistance is solely related to the undrained shear
AASHTO LRFD Bridge Design Specifications (AASHTO 2014), strength in the α-method, whereas the load transfer between a
which is expressed as a function of depth below the ground surface shaft and the soil is controlled by effective stress behavior. With
only, without accounting for the soil strength or in situ stress state. the β-method in Eq. (12), the side resistance is proportional
A more rational method was suggested by Chen and Kulhawy to the effective stress, disregarding the rotation of principle
(2002), where the values of K and δ are evaluated separately and stresses, relative movement, and many other aspects (Fellenius
then combined to calculate the coefficient β. This approach was 2018).
adopted in the 2010 FHWA design manual, where β is approxi- • Transformation uncertainties of empirical correlations such as
mated by Eq. 13–13 in Brown et al. (2010) Eqs. 3–8 to 3–10 in Brown et al. (2010) that are utilized to

© ASCE 04019042-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


100000 10

10000

Mu=Rum/Ruc
Ruc (kN)
1000 1

100 Uplift Uplift


Compression Compression
Equality line Mu=1
10 0.1
10 100 1000 10000 100000 10 100 1000 10000 100000
(a) R um (kN) (b) R um (kN)
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

100000 10

10000

Mu=Rum/Ruc
Ruc (kN)

1000 1

100 Uplift Uplift


Compression Compression
Equality line Mu=1
10 0.1
10 100 1000 10000 100000 10 100 1000 10000 100000
(c) R um (kN) (d) R um (kN)

100000 10

10000
Mu=Rum/Ruc
Ruc (kN)

1000 1

100 Uplift Uplift


Compression Compression
Equality line Mu=1
10 0.1
10 100 1000 10000 100000 10 100 1000 10000 100000
(e) R um (kN) (f) R um (kN)

Fig. 2. Comparison between calculated and measured resistances: (a and b) clay; (c and d) sand; and (e and f) gravel.

estimate su and ϕ based on in situ test results (e.g., SPT N-value Phoon 2017; Tang et al. 2017a), and (2) the load-displacement
and cone-tip resistance). model factors for pile foundations (e.g., Stuedlein and Reddy 2013;
• Bias in the definition of measured capacity and measurement Reddy and Stuedlein 2017a; Tang and Phoon 2018b). In this case,
errors in drilled-shaft load tests and site investigation. it is inappropriate to treat the model factor as a random variable;
In summary, the model statistics characterized by load test data therefore, the randomness of the model factor should be verified
will contain many sources of uncertainties (Lesny 2017). (Phoon et al. 2016).
The M u samples are plotted against the shaft slenderness D=B
and soil engineering parameters (su for cohesive soil and ϕ for co-
Statistical Evaluation of Model Factors hesionless soil) in Fig. 3. Visually, Mu may not be dependent on the
pile geometries and soil parameters. The randomness is further
The model factors calculated from the database take a range of verified with Spearman correlation analysis, which is conducted
values. Three steps are followed to evaluate the model factors by the MATLAB version 7.10.0 function “corr.” It gives the cor-
(Dithinde et al. 2011). Data outliers that deviate significantly relation quantified by the r-value with a probability p. If the
from the main trend of a data set may lead to a biased estimation p-value is less than 0.05, the correlation r is statistically significant.
of model statistics and should be identified first. Three methods The Spearman correlation analysis results are given in Tables S6
were suggested by Dithinde et al. (2011): (1) normalized load- and S7 in Appendix S2. Although a few p-values are 0, the range
displacement curves in Fig. 1, and (2) comparison between the of r is 0.2 to 0.4, indicating a low degree of correlation. On this
measured and calculated capacities in Fig. 2. Visual inspection basis, all model factors can be viewed as random variables.
of these figures indicates no data outliers.

Model Statistics and Probability Distributions


Verification of Randomness
The model factors may not be random in the sense that they are Capacity Model Factor
systematically affected by soil properties and foundation geome- The statistics for the model factor M u for drilled shafts under axial
tries. Examples can be found in (1) the model factor for uplift loading are given in Table 2. For the current FHWA design manual
capacity of helical anchors in clay (Tang and Phoon 2016) and (Brown et al. 2010), mean ¼ 1.19–1.69 and COV ¼ 0.39–0.64 for
bearing capacity of shallow foundations on sand (e.g., Tang and compression and mean ¼ 1.11–1.28 and COV ¼ 0.28–0.43 for

© ASCE 04019042-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


10 1 10 1

Mu

Mu
10 0 10 0

Uplift Uplift
Compression Compression
Mu=1 Mu=1
10 -1 10 -1
10 0 10 1 10 2 10 1 10 2
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

D/B su (kPa)
(a)
10 1 10 1
Mu

Mu
10 0 10 0

Uplift Uplift
Compression Compression
Mu=1 Mu=1
10 -1 10 -1
10 0 10 1 10 2 25 30 35 40 45 50
D/B (o )
(b)
1
10 Mu 10 1
Mu

10 0 10 0

Uplift Uplift
Compression Compression
Mu=1 Mu=1
10 -1 10 -1
10 0 10 1 10 2 35 40 45 50
D/B (o )
(c)

Fig. 3. Scatter plot of model factors with shaft slenderness ratio and soil parameters: (a) clay; (b) sand; and (c) gravel.

uplift that are obtained in this study. The probability plots of the measured capacity in AbdelSalam et al. (2015) was interpreted by
observed Mu values with the fitted normal and lognormal distribu- the Davisson offset limit, which generally leads to the lowest es-
tions are presented in Fig. 4. It can be seen that the lognormal dis- timation of the measured capacity (Chen and Fang 2009). The COV
tribution matches the samples better than the normal distribution. value of AbdelSalam et al. (2015) is about 0.4, which is also lower
For all cases studied, the capacity model factor M u follows a log- than the present result (COV ¼ 0.4–0.63). This is primarily attrib-
normal distribution. uted to more diverse site conditions in this work, which can be seen
Because different calibration databases (e.g., number of test in Tables S1, S2, and S3 in the Supplemental Data. This is reason-
sites, site conditions, and number of load tests) and interpretation able because the uncertainty in the capacity calculation is more
methods for measured capacity were adopted in the studies in closely related to the imperfection of the adopted method and pos-
Table 2, model statistics could be largely different even for the same sibly some site variabilities.
design method, soil type, and/or construction method. Despite this In addition, some previous design methods have been evalu-
fact, summary and comparison of the ranges for mean and COV ated by (1) Dithinde et al. (2011) using 83 static-load tests in
could provide useful information (i.e., bias and dispersion) on the South Africa; (2) Paikowsky et al. (2004) using 256 field load
respective design method that is discussed subsequently. For drilled tests mainly in the United States; (3) Abu-Farsakh et al. (2013);
shafts under compression, this design manual has also been evalu- and (4) Ng and Fazia (2012). For drilled shafts under compression,
ated by (1) AbdelSalam et al. (2015) using 157 static-load tests on the mean ¼ 0.84–2.27 and COV ¼ 0.27–0.74 for Reese and
large-diameter bored piles in Egypt; (2) Abu-Farsakh et al. (2013) O’Neill (1988), mean ¼ 1.04–1.45 and COV ¼ 0.25–0.67 for
using 30 Osterberg cell tests and four top-down static-load tests in Reese and Wright (1977), and mean ¼ 1.14–1.35 and COV ¼
Louisiana and Mississippi; and (3) Ng and Fazia (2012) using 24 0.3–0.58. For drilled shafts under uplift, mean ¼ 0.87–1.25 and
load tests from New Mexico Department of Transportation (DOT) COV ¼ 0.29–0.51 for Reese and O’Neill (1988), and mean ¼
and other DOTs (e.g., Iowa, Georgia, Texas, Florida, and Arizona). 0.83–1.24 and COV ¼ 0.41 and 0.54 for Reese and Wright
The mean ¼ 0.81–1.02 found by AbdelSalam et al. (2015) is lower (1977). For all 49 data groups in Table 2, there are 36 groups with
than the present result (mean ¼ 1.19–1.69). This is because that the mean ¼ 1–2 and COV ¼ 0.3–0.6. The other cases outside this

© ASCE 04019042-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Table 2. Summary of the model statistics for the static analysis methods of drilled-shaft capacity
Construction
References Load type Soil type N Design method method Mean COV
This study Uplift Clay 32 Brown et al. (2010) — 1.11 0.28
Sand 30 1.28 0.33
Gravel 109 1.14 0.43
Compression Clay 64 1.41 0.63
Sand 44 1.19 0.39
Gravel 41 1.69 0.47
Dithinde et al. (2011) Compression Sand 30 Static formula — 0.98 0.24
Clay 53 1.15 0.25
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

Zhang and Chu (2009b) Compression Sand/silt 11 O’Neill and Reese (1999) Casing 0.6 0.58
Rocks 15 RCD 0.48 0.52
15 Code of practice for foundations 2.57 0.31
Paikowsky et al. (2004) Compression Sand 32 Reese and O’Neill (1988) Mixed 1.71 0.60
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 Reese and O’Neill (1988) Mixed 0.9 0.47
13 Casing 0.84 0.50
40 Dry 0.88 0.48
Clay+sand 44 Mixed 1.19 0.30
21 Casing 1.04 0.29
12 Dry 1.32 0.28
10 Slurry 1.29 0.27
Compression Clay+sand 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 and Reese (1999) Mixed 1.3 0.34
29 Dry 1.35 0.31
Paikowsky et al. (2004) Uplift Sand 11 Reese and O’Neill (1988) Mixed 1.09 0.51
11 Reese and Wright (1977) 0.83 0.54
Clay 13 Reese and O’Neill (1988) 0.87 0.37
Mixed 14 1.25 0.29
14 Reese and Wright (1977) 1.24 0.41
All soil 39 Reese and O’Neill (1988) 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 and Reese (1999) 1.25 0.37
Abu-Farsakh et al. (2013) Compression Mixed 34 O’Neill and Reese (1999) — 1.27 0.3
34 Brown et al. (2010) 0.99 0.3
Ng and Fazia (2012) Compression Sand 24 O’Neill and Reese (1999) — 1.14 0.58
24 Brown et al. (2010) 1.21 0.6
Ng et al. (2014) Compression Mixed 11 O’Neill and Reese (1999) — 1.18 0.16
AbdelSalam et al. (2015) Compression Clay 22 Brown et al. (2010) — 1.02 0.41
Sand 45 0.91 0.4
Layered 90 0.81 0.37
Note: COP is approved by Buildings Department (2004). RCD = reverse circulation drilling.

range are largely due to the statistical uncertainty, namely very few and 2.97 for drilled shafts in clay, sand, and gravel, respectively.
load tests from different sites (typically fewer than 15). According A smaller mean of a for drilled shafts in gravel could be explained
to the classification of model uncertainty proposed by Phoon and by the difference in the slenderness D=B, where D=B ¼ 1.6–56 in
Tang (2019), the design methods for drilled-shaft capacity (com- clay, 5.1–59 in sand, and 1.77–17.3 in gravel. This can be easily
pression or uplift) are moderately conservative (mean ¼ 1–2) and understood with the physical implication of a. A higher a value
medium dispersion (COV ¼ 0.3–0.6). indicates a more flexible shaft with a larger slenderness D=B. The
COV values of a are 0.75 (clay), 0.74 (sand), and 0.6 (gravel),
Correlated Hyperbolic Parameters indicating a high level of variability. Phoon and Kulhawy (1999)
The statistics of the hyperbolic model factors a and b are summa- indicated that the uncertainties in stiffness are larger than those in
rized in Table 3. For the parameter b, the results are close to those of strength. The hyperbolic parameters with the fitted distributions are
Dithinde et al. (2011), where mean ≈ 0.8 and COV ≈ 0.15 (low illustrated in Fig. 5, where a follows a lognormal distribution.
level of variability). For parameter a, the mean values are 4.79, 4.3, However, b, which is usually bounded within 0 and 1 according to

© ASCE 04019042-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


0.999
0.99 0.99

0.9 0.9

Probability

Probability
0.75 0.75
0.5
0.25 0.5
0.1 0.25
0.01 Normal 0.1 Normal
0.001 Samples Samples
0.0001 Lognormal 0.01 Lognormal

0 0.5 1 1.5 2 0 1 2 3 4 5
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

(a) Mu (b) Mu

0.99 0.99

0.9 0.9
Probability

Probability
0.75 0.75
0.5 0.5
0.25 0.25
0.1 Normal 0.1 Normal
Samples Samples
0.01 Lognormal 0.01 Lognormal

0.5 1 1.5 2 2.5 3 0 1 2 3


(c) Mu (d) Mu

0.999
0.99
0.99
0.9 0.9
Probability

Probability
0.75 0.75
0.5 0.5
0.25
0.1 0.25
Normal 0.1 Normal
0.01 Samples Samples
0.001 Lognormal 0.01 Lognormal
0.0001
0 1 2 3 1 2 3
(e) Mu (f) Mu

Fig. 4. Probability plots for observed resistance model factors.

Eq. (5), can be modeled by a beta distribution. The mathematical the strength of dependence. The local optimization method is prone
equations of the probability distribution function (PDF) and statis- to getting trapped in local minima and lead to a biased result. On the
tics of a beta distribution are given in the Appendix. contrary, the Bayesian method can provide a robust estimation of
the global optima and also characterize the underlying uncertainty
in estimating θ. For drilled shafts in clay, the posterior distribu-
Simulation of Hyperbolic Parameters tions of θ obtained by Bayesian analyses are plotted in Fig. 7 for
Scatter plots of the observed load-displacement model factors five selected copulas (i.e., Gaussian, Clayton, Frank, Gumbel,
(solid squares) are presented in Fig. 6. A negative correlation is and Plackett). The asterisk on the top of each plot is the θ value
observed, which is quantified by the Kendall’s tau coefficient ρτ ¼ estimated by local optimization. The gray bins are the Bayesian-
−0.5 for clay and −0.4 for sand and gravel. Phoon and Kulhawy derived parameters and cross at the bottom of each plot is the maxi-
(2008) and Li et al. (2013) showed that uncorrelated hyperbolic mum likelihood parameter. It is found that the posterior parameters
parameters cannot reasonably capture the scatter and shape of mea- of Gaussian [Fig. 7(a)], Frank [Fig. 7(c)], and Plackett [Fig. 7(e)]
sured load-displacement curves and overestimate the probability of copula are well constrained. The corresponding θ values from local
failure at SLS. Correlated load-displacement model factors can be optimization are −0.69, −5.46, and 0.11. They coincide with the
simulated with (1) the translation probability model with Pearson mode of the distribution from Bayesian analyses, where θ ¼ −0.68,
correlation (e.g., Phoon and Kulhawy 2008; Dithinde et al. 2011; −5, and 0.11. For Clayton [Fig. 7(b)] and Gumbel [Fig. 7(d)] cop-
Stuedlein and Reddy 2013), and (2) copula theory (e.g., Li et al. ulas, the maximum likelihood parameters are placed on the boun-
2013; Huffman and Stuedlein 2014; Huffman et al. 2015; Reddy dary. It implies that local optimization fails to obtain the best-fit
and Stuedlein 2017a; Tang and Phoon 2018a, b, d). copula within the parameter bounds, and the choice of such a cop-
Copula theory is used in this section. Sadegh et al. (2017) ula for these data is inappropriate (Sadegh et al. 2017).
developed a Multivariate Copula Analysis toolbox (MvCAT) in Similar results are obtained for drilled shafts in sand and gravel.
MATLAB, which includes 26 copula families with different levels The copula families are evaluated by the Bayesian information cri-
of complexity, to perform rigorous and comprehensive bivariate terion (BIC). Table S8 in Appendix S3 summarizes the BIC and θ
dependence analyses. MvCAT uses both local optimization and values of the selected copulas. The lowest BIC values are obtained
Bayesian approaches to infer the copula parameter θ, which reflects for Gaussian copula. The simulated a and b values (solid cross)

© ASCE 04019042-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Table 3. Summary of statistics for the load-settlement model factors of foundations
a b
Load-displacement
References Foundation type Load type Soil type N model Mean COV Mean COV ρ
This study Drilled shaft Compression Clay 110 Hyperbolic 4.79 0.75 0.83 0.13 −0.5
Sand 76 4.3 0.74 0.82 0.14 −0.4
Uplift Gravel 60 2.97 0.6 0.82 0.13 −0.4
Phoon et al. (2006) ACIP pile (D=B > 20) Compression Sand 40 5.15 0.6 0.62 0.26 −0.67
Phoon et al. (2007) Spread footing Uplift Clay þ sand 85 7.13 0.65 0.75 0.18 −0.24
Drilled shaft Clay þ sand 48 1.34 0.54 0.89 0.07 −0.59
Pressure-injected footing Sand 25 1.38 0.68 0.77 0.27 −0.73
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

Dithinde et al. (2011) Driven pile Compression Clay 59 3.58 0.57 0.78 0.11 −0.74
Sand 28 5.55 0.54 0.71 0.14 −0.6
Drilled shaft Clay 53 2.79 0.73 0.82 0.11 −0.75
Sand 30 4.1 0.78 0.77 0.21 −0.75
Tang and Phoon (2017a) Multihelix steel Compression Clay 53 5.54 0.36 0.78 0.14 −0.46
square shaft pile Sand 49 5.84 0.27 0.76 0.14 −0.36
Tang and Phoon (2017b) Driven closed-end pile Compression Sand 111 6.26 0.75 0.8 0.15 −0.56
Stuedlein and Reddy (2013) ACIP pile Compression Sand 87 0.16 0.49 3.4 0.23 −0.73
Reddy and Stuedlein (2017a) ACIP pile Compression Sand 95 0.16 0.47 3.38 0.23 −0.67
Tang and Phoon (2018d) Steel H-pile Compression Clay 47 1.07 0.37 1.01 0.09 −0.51
Sand 52 1.17 0.6 0.69 0.18 −0.6
Layered 50 1.17 0.59 0.75 0.13 −0.53
Uzielli and Mayne (2011) Shallow foundation Compression Sand 30 0.1 0.45 6.9 0.58 −0.38
Huffman and Stuedlein (2014) Spread footing Compression Clay 30 0.02 0.79 1.15 0.25 —
Huffman et al. (2015) Spread footing Compression Clay 30 0.013 0.53 0.7 0.16 −0.85
Stuedlein and Uzielli (2014) Helical anchor Uplift Clay 37 0.03 0.81 0.65 0.26 −0.94
Tang et al. (2018b) Square foundation Compression Sand 63 0.015 0.4 0.72 0.17 −0.67
Uzielli and Mayne (2011) Shallow foundation Compression Sand 30 Power law 0.59 0.29 0.49 0.17 0.64
Huffman and Stuedlein (2014) Spread footing Compression Clay 30 3.09 0.41 0.45 0.23 0.43
Huffman et al. (2015) Spread footing Compression Clay 30 4.94 0.45 0.47 0.21 0.78
Stuedlein and Uzielli (2014) Helical anchor Uplift Clay 37 2.73 0.39 0.41 0.45 —
Note: Correlation ρ is quantified by Pearson correlation coefficient in Phoon et al. (2006, 2007) and Stuedlein and Reddy (2013), whereas it corresponds to the
Kendall’s tau correlation in the other studies; in Uzielli and Mayne (2011), the applied load is normalized by net cone-tip resistance of piezocone penetration
test; statistics in Stuedlein and Reddy (2013), Reddy and Stuedlein (2017b), and Tang and Phoon (2018b) were calibrated for the transformed hyperbolic
parameters after removing the dependency on pile slenderness; foundation displacement s in the hyperbolic model evaluated by Uzielli and Mayne (2011),
Huffman and Stuedlein (2014), Huffman et al. (2015), Stuedlein and Uzielli (2014), and Tang et al. (2017b) was normalized by foundation diameter or
width B, namely, s=B. ACIP = augered cast-in-place.

are given in Fig. 6. It shows that the Gaussian copula reasonably LRFD in 24 states, the resistance factors were simply derived by
captures the scatter within the regressed a and b values from the fitting to the ASD factor of safety based on previous experience.
database (solid circle). The simulated a and b values are then used Seo et al. (2015) presented a latest review for the pile design prac-
to generate the load-displacement curves. Fig. 8 illustrates that for tice of each state DOT in the United States. Some DOTs were de-
drilled shafts in clay and sand, the simulated load-displacement voted to calibrating region-specific resistance factors by reliability
curves (gray lines) satisfactorily characterize the scatter and shape analyses, such as Florida (Kuo et al. 2002), Indiana (Salgado et al.
of the measured curves (dark gray lines). 2011), Iowa (e.g., AbdelSalam et al. 2012; Ng et al. 2014), Kansas
The current practice in model uncertainty evaluation does not (Yang et al. 2010), Louisiana (e.g., Abu-Farsakh et al. 2009, 2010,
take full advantage of the data yet. There is a great potential for 2013), Nevada (Motamed et al. 2016), and New Mexico (Ng and
advanced data analytics tools to draw more insights from databases Fazia 2012).
(Machairas and Iskander 2018). Najjar et al. (2014, 2017) applied Ideally, ULS and SLS should be checked by the same reliability
the Bayesian methods to update the load-displacement curves sum- theory. Extensive studies have been implemented to calibrate ULS.
marized by Akbas (2007) with point measurements for larger foot- Recently, there has been an increasing trend in applying RBD for
ings. Other discussions and applications of Bayesian methods in SLS (e.g., Phoon et al. 1995; Paikowsky and Lu 2006; Phoon and
geotechnical engineering have been given by Zhang et al. (2009), Kulhawy 2008; Wang and Kulhawy 2008; Misra and Roberts 2009;
Baecher (2017), Juang and Zhang (2017), and Li et al. (2018). Zhang and Chu 2009a, c; Uzielli and Mayne 2011; Stuedlein and
Reddy 2013; Huffman and Stuedlein 2014; Huffman et al. 2015,
2016; Moghaddam 2016; Vu and Loehr 2017). In the following
Application of Model Statistics in RBD section, the statistics of the capacity and load-displacement model
factors will be applied to calibrate the ULS and SLS resistance fac-
tors in LRFD of drilled shafts under axial loading.
Overview
The FHWA mandated the use of LRFD for all new bridges after
September 2007. The nationwide survey conducted by AbdelSalam LRFD Calibration
et al. (2012) indicated that although 24 states implemented LRFD The basic idea of LRFD is that the factored loads effects should not
of bridge foundations, 5 states were still using the ASD, and 21 exceed the factored capacity. The performance function for LRFD
states were migrating toward LRFD. For the implementation of design is (AASHTO 2014)

© ASCE 04019042-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


0.999 0.999
0.99 Samples 0.99
Lognormal
0.9 Beta 0.9

Probability

Probability
0.75 0.75
0.5 0.5
0.25 0.25
0.1 0.1 Samples
Normal
0.01 0.01 Lognormal
0.001 0.001
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

0.4 0.6 0.8 1 0 5 10 15 20


b a
(a)
0.999
0.99 Samples 0.9999
Lognormal 0.999
0.9 0.99
Beta
Probability

Probability
0.75 0.9
0.5 0.75
0.25 0.5
0.1 0.25
0.1
0.01 0.01 Samples
0.001 0.001 Normal
0.0001 0.0001 Lognormal

0.4 0.6 0.8 1 0 5 10 15 20


b a
(b)
0.9999 0.9999
0.999 Samples 0.999
Lognormal
0.99
0.95 Beta
0.95
Probability

Probability

0.75
0.5 0.75
0.25 0.5
Samples
0.1 0.25
Normal
0.01 0.1
Lognormal
0.001 0.01
0.0001
0.4 0.6 0.8 1 0 2 4 6 8 10
b a
(c)

Fig. 5. Probability plots for observed hyperbolic parameters: (a) clay; (b) sand; and (c) gravel.

X
ψ · Rn ≥ γ i · Qni ð16Þ ULS Resistance Factor
For Strength limit I, the ULS performance function has the follow-
ing form:
where Rn = calculated nominal capacity; ψ (≤1) = resistance factor
applicable to Rn ; Qni = nominal value of load component i; g ¼ M u · Ruc − ðλDL · QDL þ λLL · QLL Þ ð18Þ
and γ i (≥1) = load factor applicable to Qni . LRFD aims to cali-
brate ψ to ensure that Eq. (16) is satisfied for a target reliability Based on Eq. (16), the calculated nominal capacity (i.e., Rn ¼
index β T . Ruc ) is expressed
By considering the AASHTO Strength limit I (AASHTO 2014)
(i.e., the combination of dead and live loads), the applied load Q is γ DL · QDL þ γ LL · QLL
Ruc ¼ ð19Þ
written ψULS

Q ¼ λDL · QDL þ λLL · QLL ð17Þ where γ DL = dead load factor ¼ 1.25; γ LL = live load factor ¼
1.75; and ψULS = ULS resistance factor. Substitution of Eq. (19)
into Eq. (18) leads to
where QDL = nominal dead load; λDL = bias factor of QDL ; QLL =
nominal live load; and λLL = bias factor of QLL . The load biases  
γ DL · QDL þ γ LL · QLL
λDL and λLL are lognormal random variables. The mean and COV g¼ · M u − ðλDL · QDL þ λLL · QLL Þ
ψULS
values of λDL are 1.05 and 0.1, and the mean and COV values of
λLL are 1.15 and 0.2. ð20Þ

© ASCE 04019042-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


40 40

35 Simulated (N=5000)
35
Simulated (N=5000)
30 Observed (N=110) 30 Observed (N=76)
= -0.5 = -0.4
25 25

20 20

a
15 15

10 10
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

5 5

0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
(a) b (b) b

25

Simulated (N=5000)
20
Observed (N=60)

= -0.4
15
a

10

0
0.2 0.4 0.6 0.8 1
(c) b

Fig. 6. Scatter plot of hyperbolic parameters: (a) clay; (b) sand; and (c) gravel.

0.4 0.4 0.4

0.3 0.3 0.3

0.2 0.2 0.2

0.1 0.1 0.1

0 0 0
-0.72 -0.67 -0.62 0 0.05 0.1 -5.7 -5.2 -4.7 -4.2
(a) (b) (c)

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
1 1.01 1.02 1.03 0.08 0.1 0.12 0.14
(d) (e)

Fig. 7. Posterior distribution of the selected copula parameter for drilled shafts in clay: (a) Gaussian; (b) Clayton; (c) Frank; (d) Gumbel; and
(e) Plackett.

© ASCE 04019042-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Comparison of measured and simulated load-displacement curves: (a) clay; (b) sand; and (c) gravel.

Introducing η ¼ QDL =QLL , Eq. (20) is rearranged as follows uplift resistance factors for a target reliability index. For a target
(e.g., Abu-Farsakh et al. 2009, 2010, 2013): reliability index, the compression resistance factors are lower than
  the uplift resistance factors for clay and sand. These resistance fac-
γ DL þ γ LL =η tors are calibrated for total capacity. In the AASHTO LRFD Bridge
g¼ · M u − ðλDL þ λLL =ηÞ ð21Þ
ψULS Design Specifications (AASHTO 2014), the resistance factors are,
respectively, recommended for side and base capacity components
Monte Carlo simulations are utilized to calibrate ψULS with the (Table 10.5.5.2.4-1 in AASHTO 2014). Because the equations for
following steps (e.g., Abu-Farsakh et al. 2009, 2010, 2013; Haque side resistance are the same for uplift and compression, the uplift
and Abu-Farsakh 2018): resistance factors are assumed to be 0.1 less than the compression
1. Select an initial value of the resistance factor (ψULS ). resistance factors. This assumption is also adopted in the current
2. Produce random numbers for three variables (i.e., Mu , λDL , and FHWA design manual (Brown et al. 2010) (Table 10-5 in Brown
λLL ) in Eq. (21), which are sampled independently. et al. 2010). The ratio of the resistance factor to the mean model
3. Count the number N f of cases for Eq. (21) is smaller or equal factor (ψULS =M u ) is also reported in Table 4, which represents the
to zero and calculate pf ¼ N f =N s (N s ¼ 1 million) and β 0 ¼ effectiveness of the design model (Paikowsky et al. 2004).
−Φ−1 ðpf Þ, where Φ−1 is the inverse standard normal cumula- Another rational framework was presented by Basu and Salgado
tive function. (2012), where the uncertainties of all basic soil parameters appear-
4. Repeat Steps 1–3 until jβ 0 − β T j < tolerance of 0.01, where ing in the design models are quantified individually to calculate
β T ¼ 2.33 for redundant piles or 3 for nonredundant piles the resistance factors. However, calibration against pile load test
(Paikowsky et al. 2004). data cannot identify different sources of uncertainties within the
The variation of ψULS with the load ratio η is presented in Fig. 9 problem, which are considered by a single model factor. When dif-
for β T ¼ 2.33, where η ¼ 1–10. It shows ψULS decreases nonli- ferent techniques are used for site investigation, such as cone pen-
nearly as η increases. When η is larger than 3, the reduction in etration tests, the model factor M u needs to be reevaluated. This is
ψULS is very minor for all cases. This has also been observed by because different transformation model errors are involved within
Paikowsky et al. (2004), AbdelSalam et al. (2012), Abu-Farsakh the estimation of soil parameters, which could result in different
et al. (2009, 2010, 2013), and Tang and Phoon (2018a, b, c, d). capacity model statistics.
The resistance factors for η ¼ 3 with β T ¼ 2.33 and 3 are summa-
rized in Table 4. It is observed that β T has a more pronounced
SLS Resistance Factor
impact than η on ψULS . For example, for drilled shafts in
sand under compression, ψULS reduces from 0.53 to 0.39 (26% re- For analysis purposes, the load factors γ DL and γ LL are considered
duction) when β T increases from 2.33 to 3. For drilled shafts in clay as unity, the mean values of λDL and λLL are 1.05 and 1.15, and the
and sand, the compression resistance factors are lower than the corresponding COV values are 0.1 and 0.2, in accordance with

© ASCE 04019042-13 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


0.8 0.8

0.7 0.7

0.6 0.6

ULS

ULS
0.5 0.5

0.4 0.4
Clay Clay
0.3 Sand 0.3
Sand
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

Gravel Gravel
0.2 0.2
0 2 4 6 8 10 0 2 4 6 8 10
(a) =QDL /QLL (b) =QDL /QLL

Fig. 9. Variation of ULS resistance factor with dead-load to live-load ratio: (a) uplift; and (b) compression.

Table 4. ULS resistance factors and efficiency factors for the 2010 FHWA Table 5. Variation of SLS resistance factors for different values of allow-
design method able settlement and reliability index
β T ¼ 2.33 βT ¼ 3 ψSLS
Soil Soil sa
Load type type N ψULS ψULS =μ ψULS ψULS =μ type (mm) β T ¼ 1.5 βT ¼ 2 β T ¼ 2.33 β T ¼ 2.66 βT ¼ 3
Uplift Clay 32 0.6 0.54 0.48 0.43 Clay 5 0.25 0.19 0.15 0.13 0.11
Sand 30 0.63 0.49 0.48 0.38 10 0.34 0.26 0.22 0.18 0.15
Gravel 109 0.45 0.4 0.33 0.29 15 0.4 0.3 0.25 0.21 0.17
Compression Clay 64 0.36 0.26 0.24 0.17 20 0.43 0.33 0.27 0.23 0.19
Sand 44 0.51 0.43 0.38 0.32 25 0.45 0.34 0.29 0.24 0.2
Gravel 41 0.61 0.36 0.44 0.26 Sand 5 0.3 0.24 0.2 0.17 0.15
10 0.41 0.33 0.28 0.24 0.2
15 0.47 0.38 0.32 0.28 0.24
20 0.51 0.41 0.35 0.3 0.26
the AASHTO LRFD specifications (AASHTO 2014). The SLS 25 0.53 0.43 0.37 0.32 0.27
performance function is Gravel 5 0.33 0.26 0.22 0.19 0.16
10 0.37 0.29 0.25 0.21 0.18
g ¼ Qa − ðλDL · QDL þ λLL · QLL Þ ð22Þ 15 0.43 0.34 0.28 0.24 0.21
20 0.46 0.36 0.31 0.26 0.22
From Eq. (9), Eq. (22) is reformulated as follows: 25 0.48 0.38 0.32 0.28 0.23
   
γ DL þ γ LL =η sa
g¼ · · Mu − ðλDL þ λLL =ηÞ
ψSLS a þ b · sa purposes. The present SLS resistance factors range from 0.11 to
ð23Þ 0.53, which are close to the results of Misra and Roberts (2009)
(ψSLS ¼ 0.23–0.55), Zhang and Chu (2009c) (ψSLS ¼ 0.18–0.49),
where ψSLS = SLS resistance factor; and load ratio η ¼ 3. and Vu and Loehr (2017) (ψSLS ¼ 0.1–0.4), but substantially
For a given sa value, Monte Carlo simulations are applied to de- smaller than the values of Phoon et al. (1995) (ψSLS ¼ 0.48–0.65)
termine ψSLS , with Steps 1–4 used to calculate ψULS . The difference and Moghaddam (2016) (ψSLS ¼ 0.18–0.86). Differences in the
in Step 2 is that the correlated load-displacement model factors a magnitude of the resistance factors could be attributed to (1) the
and b in Eq. (23) are randomly sampled by copula theory with the target reliability index, (2) the method utilized to calibrate ψSLS ,
source distributions. Similar studies were carried out by Huffman (3) model statistics [e.g., capacity model factor for sa ¼ 25.4 mm
and Stuedlein (2014) and Huffman et al. (2015, 2016) for spread with mean of 2.11 and COV of 0.33 in Moghaddam (2016)],
footings, Reddy and Stuedlein (2017a) for ACIP piles, and Tang (4) variability in the soil/rock strength, and (5) the load level or
and Phoon (2018a) for closed-end piles in sand. A wide range of permissible settlement considered.
the permissible settlement sa ¼ 5–25 mm and target reliability in-
dex β T ¼ 1.5–3 have been adopted to examine their effects on ψSLS .
It is also useful to compare the range of ψSLS with the results ob- Conclusions
tained by other researchers (e.g., Phoon et al. 1995; Misra and
Roberts 2009; Zhang and Chu 2009c; Moghaddam 2016; Vu and This paper developed an integrated database with 171 uplift and
Loehr 2017). 149 compression load tests on drilled shaft in clay, sand, and gravel.
The calculated ψSLS values are given in Table 5 and plotted in This database covers a wide range of shaft geometries and soil
Fig. 10. The results demonstrate that ψSLS increases as sa increases properties from a wide range of geographical regions. The modified
and β T decreases. This is intuitive in the sense that more robust Davisson offset limit was used to interpret the measured capacity.
foundation design is needed when the structure can tolerate only A normalized hyperbolic model with two correlated parameters
very small settlement (Moghaddam 2016). The results for ψSLS was employed to capture the uncertainties within the entire load-
available in literature are summarized in Table 6 for comparison displacement curves. Database studies showed a medium degree

© ASCE 04019042-14 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


1 1
=1.5 =1.5
T T
=2 =2
0.8 T 0.8 T
=2.33 T
=2.33
T
=2.66 T
=2.66
T
0.6 0.6 =3
=3

SLS

SLS
T T

0.4 0.4

0.2 0.2
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

0 0
5 10 15 20 25 5 10 15 20 25
(a) sa (mm) (b) sa (mm)

1
=1.5
T
=2
0.8 T

T
=2.33

T
=2.66
0.6
SLS

=3
T

0.4

0.2

0
5 10 15 20 25
(c) sa (mm)

Fig. 10. Variation of SLS resistance factor with allowable settlement and reliability index: (a) clay; (b) sand; and (c) gravel.

Table 6. Comparison of SLS resistance factors from different studies


References Calibration method Random variables COV Load level sa (mm) βT ψSLS
This study Factored capacity Dead load 0.1 — 5–25 1.5–3 0.11–0.53
Live load 0.2
Bias for resistance at ULS 0.28–0.52
Hyperbolic parameter a 0.6–0.75
Hyperbolic parameter b 0.13
Phoon et al. (1995) Factored capacity Strength and modulus 0.1–0.7 0.5Rn to Rn — 2.6 0.48–0.65
Curve-fitting parameters in 0.6 and 0.19
normalized hyperbolic
load-settlement model
Zhang and Chu (2009c) Settlement Bias for settlement prediction 0.22–0.42 0.5Rn — 1–2.5 0.18–0.49
Misra and Roberts (2009) Factored capacity Shaft–soil interface parameters 0.2–0.7 — 10, 20 2.6 0.23–0.55
Load-transfer model parameters
Moghaddam (2016) Factored capacity Dead load 0.1 — 2.54–25.4 2.33, 3 0.18–0.86
Live load 0.2
Bias for resistance at SLS 0.33–0.51
Vu and Loehr (2017) Factored capacity Dead load 0.1 0.1Rn to 0.4Rn — 1.8–2.3 0.1–0.4
Live load 0.12
Uniaxial compressive strength 0–1
Concrete modulus 0.15
Unit side resistance 0.66
Unit base resistance 0.25

of variability in capacity calculation by the 2010 FHWA design LRFD of drilled shafts in axial loading were determined by Monte
manual (COV ¼ 0.3–0.5), a low degree of variability in the asymp- Carlo simulations.
tote (COV ≈ 0.15), and a high degree of variability in the initial In the future, it may be very useful to apply new deep learning
slope (COV ≈ 0.75) of the normalized load-displacement curves. methods to identify similar load test data from a global database to
The model statistics in Table 2 could be an improvement of supplement limited site-specific load test data. An example was
Table 3.7.5.1 on “Indicative Computation Model Uncertainty presented by Ching and Phoon (forthcoming) for similarity be-
Factors” in JCSS (2001). The ULS and SLS resistance factors in tween site-specific data and records in a soil database. Site-specific

© ASCE 04019042-15 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


model factors may also be derived in a similar manner. This effort Akbas, S. O. 2007. “Deterministic and probabilistic assessment of settle-
will resolve the current ambiguity on whether model factors char- ments of shallow foundations in cohesionless soils.” Ph.D. thesis,
acterized from a global database can be applied to a new design Dept. of Civil Engineering, Cornell Univ.
situation that partially lies outside the database (e.g., pile and soil Baecher, G. B. 2017. “Bayesian thinking in geotechnics.” In Proc.,
types are comparable, but pile length is longer than those in the Geo-Risk 2017: Keynote Lectures, Geotechnical Special Publication
282, edited by D. Griffiths, G. Fenton, J. S. Huang, and L. M. Zhang,
database).
1–18. Reston, VA: ASCE.
Baecher, G. B., and J. T. Christian. 2003. Reliability and statistics in geo-
technical engineering. New York: Wiley.
Appendix. Mathematical Equations of PDF and Basu, D., and R. Salgado. 2012. “Load and resistance factor design
Statistics of a Beta Distribution of drilled shafts in sand.” J. Geotech. Geoenviron. Eng. 138 (12):
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

1455–1469. https://doi.org/10.1061/(ASCE)GT.1943-5606.0000714.
The probability distribution function of a beta distribution is Bauduin, C. 2002. “Design of axially loaded compression piles according
given by to Eurocode 7.” In Proc., Int. Conf. on Piling and Deep Foundations.
Paris: Presses de l’ENPC.
xξ−1 · ð1 − xÞζ−1
fðx; ξ; ζÞ ¼ ; x ∈ ½0; 1 ð24Þ Brown, D. A., J. P. Turner, and R. J. Castelli. 2010. Drilled shafts: Con-
Bðξ; ζÞ struction procedures and LRFD design methods. Publication FHWA-
NHI-10-016. Washington, DC: Federal Highway Administration.
where ξ and ζ = shape parameters; and Bðξ; ζÞ = beta function Buildings Department. 2004. Code of practice for foundations. Hong
defined as the Euler integral of the first kind Kong: Buildings Dept.
Z 1 Canadian Geotechnical Society. 2006. Canadian foundation engineering
Bðξ; ζÞ ¼ tξ−1 · ð1 − tÞζ−1 dt ð25Þ manual. 4th ed. Québec: Canadian Geotechnical Society.
0 Canadian Standards Association. 2014. Canadian highway bridge design
code. CAN/CSA-S6-14. Mississauga, ON, Canada: Canadian Geotech-
In this work, for the parameter b, ξ ¼ 6.83 and ζ ¼ 1.58 for nical Society.
clay; ξ ¼ 7.56 and ζ ¼ 1.7 for sand; and ξ ¼ 11.9 and ζ ¼ 2.72 for Carter, J. P., and F. H. Kulhawy. 1988. Analysis and design of drilled shaft
gravel. The mean μ and standard deviation σ are calculated by foundations socketed into rock. Rep. No. EL-5918. Palo Alto, CA:
Electric Power Research Institute.
ξ Casagrande, A. 1965. “Role of the calculated risk in earth work and foun-
μ¼ ð26Þ
ξþζ dation engineering.” J. Soil Mech. Found. Div. 91 (4): 1–40.
Chakraborty, T., R. Salgado, P. Basu, and M. Prezzi. 2013. “Shaft capac-
ξ·ζ ity of drilled shafts in clay.” J. Geotech. Geoenviron. Eng. 139 (4):
σ2 ¼ ð27Þ
ðξ þ ζÞ2 ðξ þ ζ þ 1Þ 548–563. https://doi.org/10.1061/(ASCE)GT.1943-5606.0000803.
Chen, Y. J., H. W. Chang, and F. H. Kulhawy. 2008. “Evaluation of uplift
interpretation criteria for drilled shaft capacity.” J. Geotech. Geoen-
viron. Eng. 134 (10): 1459–1468. https://doi.org/10.1061/(ASCE)1090
Supplemental Data
-0241(2008)134:10(1459).
Chen, Y. J., and T. H. Chu. 2012. “Evaluation of uplift interpretation criteria
Appendixes S1–S3 and Tables S1–S8 are available online in the
for drilled shafts in gravelly soils.” Can. Geotech. J. 49 (1): 70–77.
ASCE Library (www.ascelibrary.org).
https://doi.org/10.1139/t11-080.
Chen, Y. J., and Y. C. Fang. 2009. “Critical evaluation of compression in-
terpretation criteria for drilled shafts.” J. Geotech. Geoenviron. Eng.
References 135 (8): 1056–1069. https://doi.org/10.1061/(ASCE)GT.1943-5606
.0000027.
AASHTO. 2014. LRFD bridge design specifications. 7th ed. Washington, Chen, Y. J., and F. H. Kulhawy. 1994. Case history evaluation of behavior
DC: AASHTO. of drilled shafts under axial and lateral loading. Rep. No. TR-104601.
AbdelSalam, S. S., F. A. Baligh, and H. M. El-Naggar. 2015. “A database Palo Alto, CA: Electric Power Research Institute.
to ensure reliability of bored piles in Egypt.” Proc. Inst. Civ. Eng. Geo-
Chen, Y. J., and F. H. Kulhawy. 2002. “Evaluation of drained axial capacity
tech. Eng. 168 (2): 131–143. https://doi.org/10.1680/geng.14.00051.
for drilled shafts.” In Proc., Deep Foundations 2002, Geotechnical Spe-
AbdelSalam, S. S., S. Sritharan, M. T. Suleiman, K. W. Ng, and M. J.
cial Publication 116, edited by M. W. O’Neill, and F. C. Townsend,
Roling. 2012. “Development of LRFD design procedures for bridge pile
1200–1214. Reston, VA: ASCE.
foundations in Iowa.” Vol. 3 of Recommended resistant factors with
Chen, Y. J., M. R. Liao, S. S. Lin, J. K. Huang, and M. C. Marcos. 2014.
consideration to construction control and setup. Ames, IA: Institute
for Transportation, Iowa State Univ. “Development of an integrated web-based system with a pile load test
Abu-Farsakh, M. Y., Q. M. Chen, and M. N. Haque. 2013. Calibration of database and pre-analyzed data.” Geomech. Eng. 7 (1): 37–53. https://
resistance factors for drilled shafts for the new FHWA design method. doi.org/10.12989/gae.2014.7.1.037.
Rep. No. FHWA/LA.12/495. Baton Rouge, LA: Louisiana Transporta- Ching, J. Y., and J. R. Chen. 2010. “Predicting displacement of augered
tion Research Center. cast-in-place piles based on load test database.” Struct. Saf. 32 (6):
Abu-Farsakh, M. Y., S. M. Yoon, and C. Tsai. 2009. Calibration of 372–383. https://doi.org/10.1016/j.strusafe.2010.04.007.
resistance factors needed in the LRFD design of driven piles. Rep. Ching, J. Y., and K. K. Phoon. Forthcomina. “Measuring similarity between
No. FHWA/LA.09.449. Baton Rouge, LA: Louisiana Transportation site-specific data and records in a geotechnical database.” ASCE-ASME
Research Center. J. Risk Uncertainty Eng. Syst. Part A Civ. Eng.
Abu-Farsakh, M. Y., X. B. Yu, S. M. Yoon, and C. Tsai. 2010. Calibration Ching, J. Y., and K. K. Phoon. 2012. “Modeling parameters of structured
of resistance factors needed in the LRFD design of drilled shafts. clays as a multivariate normal distribution.” Can. Geotech. J. 49 (5):
Rep. No. FHWA/LA.10/470. Baton Rouge, LA: Louisiana Transporta- 522–545. https://doi.org/10.1139/t2012-015.
tion Research Center. Ching, J. Y., K. K. Phoon, and Y. C. Chen. 2010. “Reducing shear strength
Abu-Hejleh, N. M., M. Y. Abu-Farsakh, M. T. Suleiman, and C. Tsai. 2015. uncertainties in clays by multivariate correlations.” Can. Geotech. J.
“Development and use of high-quality databases of deep foundation 47 (1): 16–33. https://doi.org/10.1139/T09-074.
load tests.” Transp. Res. Rec. 2511: 27–36. https://doi.org/10.3141 Davisson, M. T. 1972. “High capacity piles.” In Proc., Lecture Series on
/2511-04. Innovations in Foundation Construction. Reston, VA: ASCE.

© ASCE 04019042-16 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Davisson, M. T. 1993. “Negative skin friction in piles and design deci- Safety and Reliability: Honoring Wilson H. Tang, Geotechnical Special
sions.” In Proc., 3rd Int. Conf. on Case Histories in Geotechnical Publication 286, edited by C. H. Juang, R. B. Gilbert, L. M. Zhang,
Engineering, 1793–1801. St. Louis: Univ. of Missouri-Rolla. J. Zhang, and L. L. Zhang, 215–246. Reston, VA: ASCE.
Dithinde, M., K. K. Phoon, M. D. Wet, and J. V. Retief. 2011. “Charac- Kulhawy, F. H. 2010. “Uncertainty, reliability, and foundation engineering:
terization of model uncertainty in the static pile design formula.” J. Geo- The 5th Peter Lumb lecture.” HKIE Trans. 17 (3): 19–24. https://doi.org
tech. Geoenviron. Eng. 137 (1): 70–85. https://doi.org/10.1061/(ASCE) /10.1080/1023697X.2010.10668200.
GT.1943-5606.0000401. Kulhawy, F. H., and J. R. Chen. 2005. “Axial compression behavior of
Duncan, J. M. 2000. “Factors of safety and reliability in geotechnical en- augered cast-in-place piles in cohesionless soils.” In Proc., Geo-
gineering.” J. Geotech. Geoenviron. Eng. 126 (4): 307–316. https://doi Frontiers 2005: Advances in Deep Foundations, Geotechnical Special
.org/10.1061/(ASCE)1090-0241(2000)126:4(307). Publication 132, edited by C. Vipulanandan, and F. C. Townsend, 1–15.
Fellenius, B. H. 2018. “Basics of foundation design.” Revised edition. Reston, VA: ASCE.
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

Accessed October 3, 2017. www.fellenius.net. Kulhawy, F. H., and J. R. Chen. 2007. “Discussion of ‘Drilled Shaft Side
Fenton, G. A., and D. V. Griffiths. 2008. Risk assessment in geotechnical Resistance in Gravelly Soils’ by Kyle M. Rollins, Robert J. Clayton,
engineering. New York: Wiley. Rodney C. Mikesell, and Bradford C. Blaise.” J. Geotech. Geoenviron.
Fenton, G. A., F. Naghibi, D. Dundas, R. J. Bathurst, and D. V. Griffiths. Eng. 133 (10): 1325–1328. https://doi.org/10.1061/(ASCE)1090-0241
2016. “Reliability-based geotechnical design in 2014 Canadian high- (2007)133:10(1325).
way bridge design code.” Can. Geotech. J. 53 (2): 236–251. https://doi Kulhawy, F. H., and C. S. Jackson. 1989. “Some observations on undrained
.org/10.1139/cgj-2015-0158. side resistance of CIDH piles.” In Vol. 2 of Proc., Foundation Engineer-
Garder, J. A., K. W. Ng, S. Sritharan, and M. J. Roling. 2012. Development ing: Current Principles and Practices, 1011–1025. Reston, VA: ASCE.
of a database for drilled shaft foundation testing (DSHAFT). Rep. No. Kulhawy, F. H., and K. K. Phoon. 2006. “Some critical issues in Geo-RBD
InTrans Project 10–366. Ames, IA: Institute for Transportation, Iowa calibrations for foundations.” In Proc., GeoCongress 2006: Geotechni-
State Univ. cal Engineering in the Information Technology Age, edited by D. J.
Goh, A. T. C., F. H. Kulhawy, and C. G. Chua. 2005. “Bayesian neural DeGroot, J. T. DeJong, D. Frost, and L. G. Baise, 1–6. Reston, VA:
network analysis of undrained side resistance of drilled shafts.” J. Geo- ASCE.
tech. Geoenviron. Eng. 131 (1): 84–93. https://doi.org/10.1061/(ASCE) Kulhawy, F. H., and K. K. Phoon. 2009. “Geo-RBD for foundations– let’s
1090-0241(2005)131:1(84). do it right!” In Int. Foundation Congress and Equipment Expo: Con-
Haque, M. N., and M. Y. Abu-Farsakh. 2018. “Estimation of pile setup and temporary Topics in in-situ Testing, Analysis, and Reliability of Foun-
incorporation of resistance factor in load and resistance factor design dations, Geotechnical Special Publication 186, edited by M. Iskander,
framework.” J. Geotech. Geoenviron. Eng. 144 (11): 04018077. D. F. Laefer, and M. H. Hussein, 442–449. Reston, VA: ASCE.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0001959. Kuo, C. L., M. McVay, and B. Birgisson. 2002. “Calibration of load and
Hasan, M. R., X. Yu, and M. Abu-Farsakh. 2018. “Extrapolation error resistance factor design: Resistance factors for drilled shaft design.”
analysis of bi-directional load-settlement curves for LRFD calibration Transp. Res. Rec. 1808: 108–111. https://doi.org/10.3141/1808-12.
of drilled shafts.” In Installation, Testing, and Analysis of Deep Foun-
Kyfor, Z. G., A. R. Schnore, T. A. Carlo, and P. F. Baily. 1992. Static testing
dations, Geotechnical Special Publication 294, edited by M. T.
of deep foundations. Rep. No. FHWA-SA-91-042. New York:
Suleiman, A. Lemnitzer, and A. W. Stuedlein, 331–340. Reston, VA:
New York State Dept. of Transportation.
ASCE.
Lambe, T. W. 1973. “Predictions in soil engineering.” Géotechnique 23 (2):
Hirany, A., and F. H. Kulhawy. 2002. “On the interpretation of drilled
149–202. https://doi.org/10.1680/geot.1973.23.2.151.
foundation load test results.” In Deep Foundations 2002, Geotechnical
Lesny, K. 2017. “Evaluation and consideration of model uncertainties in
Special Publication 116, edited by M. O’Neill, and F. Townsend, 1018–
1028. Reston, VA: ASCE. reliability based design.” Chap. 2 in Proc., Joint ISSMGE TC 205/
TC 304 Working Group on Discussion of Statistical/Reliability Methods
Honjo, Y., and O. Kusakabe. 2002. “Proposal of a comprehensive founda-
for Eurocodes. London: International Society for Soil Mechanics and
tion design code: Geo-code 21 ver. 2.” In Proc., Int. Workshop on
Foundation Design Codes and Soil Investigation in View of Int. Geotechnical Engineering.
Harmonization and Performance Based Design, 95–103. Lisse, Li, D. Q., X. S. Tang, K. K. Phoon, Y. F. Chen, and C. B. Zhou. 2013.
Netherlands: A.A. Balkema. “Bivariate simulation using copula and its application to probabilistic
Huffman, J. C., J. P. Martin, and A. W. Stuedlein. 2016. “Calibration and pile settlement analysis.” Int. J. Numer. Anal. Meth. Geomech. 37 (6):
assessment of reliability-based serviceability limit state procedures 597–617. https://doi.org/10.1002/nag.1112.
for foundation engineering.” Georisk Assess. Manage. Risk Eng. Syst. Li, J., P. Hu, M. Uzielli, and M. J. Cassidy. 2018. “Bayesian prediction of
Geohazards 10 (4): 280–293. https://doi.org/10.1080/17499518.2016 peak resistance of a spudcan penetrating sand-over-clay.” Géotechnique
.1183797. 68 (10): 905–917. https://doi.org/10.1680/jgeot.17.P.154.
Huffman, J. C., A. W. Strahler, and A. W. Stuedlein. 2015. “Reliability- Lin, S. S., M. C. M. Marcos, H. W. Chang, and Y. J. Chen. 2012. “Design
based serviceability limit state design for immediate settlement of and implementation of a drilled shaft load test database.” Comput. Geo-
spread footings on clay.” Soils Found. 55 (4): 798–812. https://doi tech. 41: 106–113. https://doi.org/10.1016/j.compgeo.2011.12.001.
.org/10.1016/j.sandf.2015.06.012. Machairas, N. P., and M. G. Iskander. 2018. “An investigation of pile design
Huffman, J. C., and A. W. Stuedlein. 2014. “Reliability-based serviceability utilizing advanced data analytics.” In Proc., Installation, Testing, and
limit state design of spread footings on aggregate pier reinforced clay.” Analysis of Deep Foundations, Geotechnical Special Publication 294,
J. Geotech. Geoenviron. Eng. 140 (10): 04014055. https://doi.org/10 edited by M. T. Suleiman, A. Lemnitzer, and A. W. Stuedlein, 132–141.
.1061/(ASCE)GT.1943-5606.0001156. Reston, VA: ASCE.
ICC (International Code Council). 2018. International building code. Mayne, P. W. 2007. Cone penetration testing. NCHRP Synthesis 368.
Washington, DC: ICC. Washington, DC: Transportation Research Board, National Research
ISO. 2015. General principles on reliability of structures. ISO 2394. Council.
Geneva: ISO. McVay, M. C., S. J. Wasman, L. Huang, and S. Crawford. 2016. Load
Jamiolkowski, M. 2003. “Soil parameters relevant to bored pile design from and resistance factor design resistance factors for augercast in place
laboratory and in-situ tests.” In Proc., 4th Int. Geotechnical Seminar on piles. Rep. No. BDV31-977-12. Tallahassee, FL: Florida Dept. of
Deep Foundations on Bored and Auger Piles, 83–100. Rotterdam, Transportation.
Netherlands: Millpress. Misra, A., and L. A. Roberts. 2009. “Service limit state resistance factors
JCSS (Joint Committee on Structural Safety). 2001. Probabilistic model for drilled shafts.” Géotechnique 59 (1): 53–61. https://doi.org/10.1680
code. Shanghai, China: JCSS. /geot.2008.3605.
Juang, C. H., and J. Zhang. 2017. “Bayesian methods for geotechnical Moghaddam, R. B. 2016. “Evaluation of the TxDOT Texas cone penetra-
applications-a practical guide.” In Proc., Geo-Risk 2017: Geotechnical tion test and foundation design method including correlation factors,

© ASCE 04019042-17 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


allowable total capacity, and resistance factors at serviceability limit Phoon, K. K., J. R. Chen, and F. H. Kulhawy. 2007. “Probabilistic hyper-
state.” Ph.D. thesis, Dept. of Civil Engineering, Texas Tech Univ. bolic models for foundation uplift movements.” In Probabilistic Appli-
Motamed, R., S. Elfass, and K. Stanton. 2016. LRFD resistance factor cations in Geotechnical Engineering, Geotechnical Special Publication
calibration for axially loaded drilled shafts in the Las Vegas 170. Reston, VA: ASCE.
valley. Rep. No. 515-13-803. Carson City, NV: Nevada Dept. of Phoon, K. K., and J. Y. Ching. 2014. Risk and reliability in geotechnical
Transportation. engineering. Boca Raton, FL: CRC Press.
Najjar, S. S., E. Shammas, and M. Saad. 2014. “Updated normalized load- Phoon, K. K., and F. H. Kulhawy. 1999. “Characterization of geotechnical
settlement model for full-scale footings on granular soils.” Georisk: variability.” Can. Geotech. J. 36 (4): 612–624. https://doi.org/10.1139
Asses. Manage. Risk Eng. Syst. Geohazards 8 (1): 63–80. https://doi /cgj-36-4-612.
.org/10.1080/17499518.2013.836419. Phoon, K. K., and F. H. Kulhawy. 2005. “Characterization of model un-
Najjar, S. S., E. Shammas, and M. Saad. 2017. “A reliability-based ap- certainties for laterally loaded rigid drilled shafts.” Géotechnique
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

proach to the serviceability limit state design of spread footings on 55 (1): 45–54. https://doi.org/10.1680/geot.2005.55.1.45.
granular soil.” In Proc., Geotechnical Safety and Reliability: Honoring Phoon, K. K., and F. H. Kulhawy. 2008. “Serviceability limit state
Wilson H. Tang, Geotechnical Special Publication 286, edited by C. H. reliability-based design.” Chap. 9 in Reliability-based design in geo-
Juang, R. B. Gilbert, L. M. Zhang, J. Zhang, and L. L. Zhang, 185–202. technical engineering: Computations and applications, edited by
Reston, VA: ASCE. K. K. Phoon, 344–384. London: Taylor & Francis.
NeSmith, V. M., and T. C. Siegel. 2009. “Shortcomings of the Davisson Phoon, K. K., F. H. Kulhawy, and M. D. Grigoriu. 1995. Reliability-based
offset limit applied to axial compressive load tests on cast-in-place design of foundations for transmission line structures. Rep. No.
piles.” In Proc., 2009 International Foundation Congress and Equip- TR-105000. Palo Alto, CA: Electric Power Research Institute.
ment Expo, Geotechnical Special Publication 185, 568–574. Reston, Phoon, K. K., and J. V. Retief. 2016. Reliability of Geotechnical Structures
VA: ASCE. in ISO 2394. Leiden, Netherlands: A.A. Balkema.
Ng, C. W. W., T. L. Y. Yau, J. H. M. Li, and W. H. Tang. 2001. “New failure Phoon, K. K., J. V. Retief, J. Y. Ching, M. Dithinde, T. Schweckendiek, Y.
load criterion for large diameter bored piles in weathered geomaterials.” Wang, and L. M. Zhang. 2016. “Some observations on ISO 2394:2015
J. Geotech. Geoenviron. Eng. 127 (6): 488–498. https://doi.org/10.1061 Annex D (Reliability of Geotechnical Structures).” Struct. Saf. 62:
/(ASCE)1090-0241(2001)127:6(488). 24–33. https://doi.org/10.1016/j.strusafe.2016.05.003.
Ng, K. W., S. Sritharan, and J. C. Ashlock. 2014. Development of prelimi- Phoon, K. K., and C. Tang. 2019. “Characterization of geotechnical
nary load and resistance factor design of drilled shafts in Iowa. Rep. model uncertainty.” Georisk Assess. Manage. Risk Eng. Syst.
No. InTrans Project 11–410. Ames, IA: Institute for Transportation, Geohazards 13 (2): 101–130. https://doi.org/10.1080/17499518.2019
Iowa State Univ. .1585545.
Ng, T. T., and S. Fazia. 2012. Development and validation of a unified Qian, Z. Z., X. Lu, X. Han, and R. Tong. 2015. “Interpretation of uplift load
equation for drilled shaft foundation design in New Mexico. Rep. tests on belled piers in gobi gravel.” Can. Geotech. J. 52 (7): 992–998.
No. NM10MSC-01. Albuquerque, NM: New Mexico Dept. of https://doi.org/10.1139/cgj-2014-0075.
Transportation. Qian, Z. Z., X. Lu, and W. Yang. 2014. “Axial uplift behavior of drilled
Niazi, F. S. 2014. “Static axial pile foundation response using seismic shafts in gobi gravel.” Geotech. Test. J. 37 (2): 205–217. https://doi.org
piezocone data.” Ph.D. thesis, School of Civil and Environmental /10.1520/GTJ20130083.
Engineering, Georgia Institute of Technology. Reddy, S. C., and A. W. Stuedlein. 2017a. “Serviceability limit state
O’Neill, M. W., and L. C. Reese. 1999. Drilled shafts: Construction pro- reliability-based design of augered cast-in-place piles in granular soils.”
cedure and design methods. FHWA Rep. No. IF-99-025. Washington, Can. Geotech. J. 54 (12): 1704–1715. https://doi.org/10.1139/cgj-2016
DC: Federal Highway Administration. -0146.
Orr, T. L. L., and E. R. Farrell. 1999. Geotechnical design to Eurocode 7. Reddy, S. C., and A. W. Stuedlein. 2017b. “Ultimate limit state reliability-
London: Springer. based design of augered cast-in-place piles considering lower-bound
Paikowsky, S., and Y. Lu. 2006. “Establishing serviceability limit state in capacities.” Can. Geotech. J. 54 (12): 1693–1703. https://doi.org/10
design of bridge foundations.” In Proc., Foundation Analysis and De- .1139/cgj-2016-0145.
sign: Innovative Methods, Geotechnical Special Publication 153, edited Reese, L. C., and M. W. O’Neill. 1988. Drilled shaft: Construction pro-
by R. L. Parsons, L. M. Zhang, W. D. Guo, K. K. Phoon, and M. Yang, cedures and design methods. Washington, DC: Federal Highway
49–58. Reston, VA: ASCE. Administration.
Paikowsky, S. G., et al. 2004. Load and resistance factor design (LRFD) for Reese, L. C., and S. J. Wright. 1977. Drilled shaft manual-volume I: Con-
deep foundations. NCHRP Rep. No. 507. Washington, DC: Transpor- struction procedures and design for axial loading. Austin, TX: Univ. of
tation Research Board. Texas.
Paikowsky, S. G., M. C. Canniff, K. Lesny, A. Kisse, S. Amatya, and R. Rollins, K. M., R. J. Clayton, R. C. Mikesell, and B. C. Blaise. 2005.
Muganga. 2010. LRFD design and construction of shallow foundations “Drilled shaft side friction in gravelly soils.” J. Geotech. Geoenviron.
for highway bridge structures. NCHRP Rep. No. 651. Washington, DC: Eng. 131 (8): 987–1003. https://doi.org/10.1061/(ASCE)1090-0241
Transportation Research Board. (2005)131:8(987).
Paikowsky, S. G., and T. A. Tolosko. 1999. Extrapolation of pile capacity Sadegh, M., E. Ragno, and A. Aghakouchak. 2017. “Multivariate copula
from non-failed load tests. Rep. No. FHWA-RD-99-170. Washington, analysis toolbox (MvCAT): Describing dependence and underlying un-
DC: Federal Highway Administration. certainty using a Bayesian framework.” Water Resour. Res. 53 (6):
Phoon, K. K. 2008. Reliability-based design in geotechnical engineering: 5166–5183. https://doi.org/10.1002/2016WR020242.
Computations and applications. London: Taylor & Francis. Salgado, R. 2008. The engineering of foundations. New York:
Phoon, K. K. 2016. “Reliability of geotechnical structures.” In Proc., 15th McGraw-Hill.
Asian Regional Conf. Soil Mech. Geotech. Eng., 1–9. Tokyo: Japanese Salgado, R., S. I. Woo, and D. Kim. 2011. Development of load and
Geotechnical Society. resistance factor design for ultimate and serviceability limit states of
Phoon, K. K. 2017. “Role of reliability calculations in geotechnical de- transportation structure foundations. Rep. No. FHWA/IN/JTRP-
sign.” Georisk Assess. Manage. Risk Eng. Syst. Geohazards 11 (1): 2011/03. Indianapolis: Indiana Dept. of Transportation.
4–21. https://doi.org/10.1080/17499518.2016.1265653. Seo, H. Y., R. B. Moghaddam, J. G. Surles, and W. D. Lawson. 2015. Im-
Phoon, K. K., J. R. Chen, and F. H. Kulhawy. 2006. “Characterization of plementation of LRFD geotechnical design for deep foundations using
model uncertainties for augered cast-in-place (ACIP) piles under axial Texas cone penetrometer (TCP) test. Rep. No. FHWA/TX-16/5-6788-
compression.” In Proc., Foundation Analysis and Design: Innovative 01-1. Austin, TX: Texas Dept. of Transportation.
Methods, Geotechnical Special Publication 153, 82–89. Reston, VA: Skempton, A. W. 1959. “Cast in-situ bored piles in London clay.” Géotech-
ASCE. nique 9 (2): 91–125. https://doi.org/10.1680/geot.1959.9.4.153.

© ASCE 04019042-18 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042


Stuedlein, A. W., W. J. Neely, and T. M. Gurtowski. 2012. “Reliability- Geomech. 17 (7): 04017015. https://doi.org/10.1061/(ASCE)GM
based design of augered cast-in-place piles in granular soils.” J. Geo- .1943-5622.0000898.
tech. Geoenviron. Eng. 138 (6): 709–717. https://doi.org/10.1061 Teixeira, A., Y. Honjo, A. Correia, and A. Henriques. 2012. “Sensitivity
/(ASCE)GT.1943-5606.0000635. analysis of vertically loaded pile reliability.” Soils Found. 52 (6):
Stuedlein, A. W., and S. C. Reddy. 2013. “Factors affecting the reliability of 1118–1129. https://doi.org/10.1016/j.sandf.2012.11.025.
augered cast-in-place piles in granular soils at the serviceability limit Uzielli, M., and P. Mayne. 2011. “Serviceability limit state CPT-based de-
state.” J. Deep Found. Inst. 7 (2): 46–57. https://doi.org/10.1179/dfi sign for vertically loaded shallow footings on sand.” Geomech. Geoeng.
.2013.7.2.004. 6 (2): 91–107. https://doi.org/10.1080/17486025.2010.531146.
Stuedlein, A. W., S. C. Reddy, and T. M. Evans. 2014. “Interpretation of Vu, T., and E. Loehr. 2017. “Service limit state design for individual drilled
augered cast in place pile capacity using static load tests.” J. Deep shafts in shale.” J. Geotech. Geoenviron. Eng. 143 (12): 04017091.
Found. Inst. 8 (1): 39–47. https://doi.org/10.1179/1937525514Y https://doi.org/10.1061/(ASCE)GT.1943-5606.0001791.
Wang, Y., and F. H. Kulhawy. 2008. “Reliability index for serviceability
Downloaded from ascelibrary.org by Indian Institute of Technology Bhubaneswar on 06/11/20. Copyright ASCE. For personal use only; all rights reserved.

.0000000003.
Stuedlein, A. W., and M. Uzielli. 2014. “Serviceability limit state design for limit state of building foundations.” J. Geotech. Geoenviron. Eng.
uplift of helical anchors in clay.” Geomech. Geoeng. 9 (3): 173–186. 134 (11): 1587–1594. https://doi.org/10.1061/(ASCE)1090-0241
https://doi.org/10.1080/17486025.2013.857049. (2008)134:11(1587).
Tang, C., and K. K. Phoon. 2016. “Model uncertainty of cylindrical shear Whitman, R. V. 1984. “Evaluating calculated risk in geotechnical engineer-
method for calculating the uplift capacity of helical anchors in clay.” ing.” J. Geotech. Eng. 110 (2): 145–188. https://doi.org/10.1061
Eng. Geol. 207: 14–23. https://doi.org/10.1016/j.enggeo.2016.04.009. /(ASCE)0733-9410(1984)110:2(143).
Wysockey, M. H. 1999. “Axial capacity of drilled shafts.” Ph.D. thesis,
Tang, C., and K. K. Phoon. 2017. “Model uncertainty of Eurocode 7 ap-
Dept. of Civil Engineering, Univ. of Illinois at Urbana–Champaign.
proach for bearing capacity of circular footings on dense sand.” Int. J.
Yang, X., J. Han, and R. L. Parson. 2010. Development of recommended
Geomech. 17 (3): 04016069. https://doi.org/10.1061/(ASCE)GM.1943
resistance factors for drilled shafts in weak rocks based on o-cell
-5622.0000737.
tests. Rep. No. K-Tran: KU-07-4. Topeka, CA: Kansas Dept. of
Tang, C., and K. K. Phoon. 2018a. “Characterization of model uncertainty
Transportation.
in predicting axial resistance of piles driven into clay.” Can. Geotech. J.
Zhang, J., L. M. Zhang, and W. H. Tang. 2009. “Bayesian framework for
https://doi.org/10.1139/cgj-2018-0386.
characterizing geotechnical model uncertainty.” J. Geotech. Geoen-
Tang, C., and K. K. Phoon. 2018b. “Evaluation of model uncertainties in viron. Eng. 135 (7): 932–940. https://doi.org/10.1061/(ASCE)GT
reliability-based design of steel H-piles in axial compression.” Can. .1943-5606.0000018.
Geotech. J. 55 (11): 1513–1532. https://doi.org/10.1139/cgj-2017 Zhang, L. M., and L. F. Chu. 2009a. “Calibration of methods for designing
-0170. large-diameter bored piles: Serviceability limit state.” Soils Found.
Tang, C., and K. K. Phoon. 2018c. “Statistics of model factors and con- 49 (6): 897–908. https://doi.org/10.3208/sandf.49.897.
sideration in reliability-based design of axially loaded helical piles.” Zhang, L. M., and L. F. Chu. 2009b. “Calibration of methods for designing
J. Geotech. Geoenviron. Eng. 144 (8): 04018050. https://doi.org/10 large-diameter bored piles: Ultimate limit state.” Soils Found. 49 (6):
.1061/(ASCE)GT.1943-5606.0001894. 883–895. https://doi.org/10.3208/sandf.49.883.
Tang, C., and K. K. Phoon. 2018d. “Statistics of model factors in reliability- Zhang, L. M., and L. F. Chu. 2009c. “Development of partial factors
based design of axially loaded driven piles in sand.” Can. Geotech. J. for serviceability limit state design of large-diameter bored piles.”
55 (11): 1592–1610. https://doi.org/10.1139/cgj-2017-0542. HKIE Trans. 16 (4): 28–35. https://doi.org/10.1080/1023697X.2009
Tang, C., K. K. Phoon, and S. O. Akbas. 2017b. “Model uncertainties for .10668172.
the static design of square foundations on sand under axial compres- Zhang, L. M., and A. M. Y. Ng. 2005. “Probabilistic limiting tolerable
sion.” In Proc., Geo-Risk 2017: Reliability-Based Design and Code displacements for serviceability limit state design of foundations.”
Developments, Geotechnical Special Publication 283, edited by J. S. Géotechnique 55 (2): 151–161. https://doi.org/10.1680/geot.2005.55
Huang, G. A. Fenton, L. M. Zhang, and D. V. Griffiths, 141–150. .2.151.
Reston, VA: ASCE. Zhang, L. M., Y. Xu, and W. H. Tang. 2008. “Calibration of models for pile
Tang, C., K. K. Phoon, L. Zhang, and D. Q. Li. 2017a. “Model uncertainty settlement analysis using 64 field load tests.” Can. Geotech. J. 45 (1):
for predicting the bearing capacity of sand overlying clay.” Int. J. 59–73. https://doi.org/10.1139/T07-077.

© ASCE 04019042-19 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2019, 145(9): 04019042

You might also like