Geotechnical Engineering Research Laboratory Samuel G. Paikowsky, Sc.D.

One University Avenue Professor

Lowell, Massachusetts 01854
Tel: (978) 934-2277 Fax: (978) 934-3046
e-mail: Samuel_Paikowsky@uml.edu
web site: http://geores.caeds.eng.uml.edu
DEPARTMENT OF CIVIL AND
ENVIRONMENTAL ENGINEERING

Load and Resistance Factor Design (LRFD) for

Deep Foundations

by
Samuel G. Paikowsky

Published in:
Keynote lecture in the proceedings of IWS Kamakura 2002 Foundation Design Codes
and Soil Investigation in view of International Harmonization and Performance, Honjo et
al. eds., Tokyo Japan, April 10-12, 2002, pp. 59-94.
Load and resistance factor design (LRFD) for deep foundations
Samuel G. Paikowsky
Geotechnical Engineering Research Laboratory
Department of Civil and Environmental Engineering
University of Massachusetts, Lowell, Massachusetts, 01854, USA

ABSTRACT: An ongoing project supported by the USA National Cooperative Highway Research Program
(NCHRP) under the Transportation Research Board (TRB) of the National Academy of Science (NAS), is
aimed at rewriting AASHTO Deep Foundation Specifications. The AASHTO specifications are traditionally
observed on all federally aided projects and generally viewed as a National code of the US Highway practice,
hence influences the construction of all the deep foundations of highway bridges throughout the USA.
Since 1993 (20th edition) the existing AASHTO code, similarly to others worldwide, is based on Load and
Resistance Factor Design (LRFD) principles. The current attempt is the first, however, to use reliability
based calibration utilizing databases. In the present phase, the developed databases relate to axial capacity of
single driven piles and drilled shafts only. Databases containing 527 and 129 case histories were compiled at
the university of Florida for static evaluation of driven piles and drilled shafts respectively. For dynamic
evaluation of driven piles, a database (PD/LT2000) containing information related to 210 piles and 403 dy-
namic measurements was compiled at the University of Massachusetts at Lowell.
A review of design methodologies, LRFD principles, and deep foundation codes is briefly discussed. A
summary of the large databases is presented, detailing the performance of the various static and dynamic
methods when compared to static load testing to failure. The parameters that control the accuracy of the pre-
dictions are analyzed.
Statistical analyses are then utilized for the development of resistance factors to be used in the new specifi-
cations (correct for the present stage). Recommended initial parameters are presented and discussed. Included
are discussions regarding target reliability and the need for knowledge based parameters accounting for sub-
surface variability, quality of soil parameters estimation, and previous experience as well as amount and type
of testing during construction.

1 BACKGROUND the design - construction - quality control sequence

(i.e. independence in resistance factors according to
National Cooperative Highway Research Program, the chronological stage and the evaluation proce-
project, NCHRP 24-17, “LRFD Deep Foundations dure) and (ii) Overcome the generic difficulties of
Design” was initiated to: (i) Provide recommended applying the LRFD methodology to geotechnical
revisions to the driven pile and drilled shaft portions applications, i.e. incorporation of indirect variability,
of section 10 of AASHTO Specifications and (ii) (e.g. site or parameters interpretation), judgment
Provide a detailed procedure for calibrating deep (e.g. previous experience), and other similar factors.
foundation resistance factors. The current AASHTO The project team, headed by the author, is divided
specifications as well as other existing codes based into three major groups dealing with static analyses
on Load and Resistance Factor Design (LRFD) prin- (University of Florida), probabilistic approaches and
ciples were calibrated using a combination of reli- structural analyses (University of Maryland), and
ability theory, fitting to ASD (Allowable Stress De- dynamic analyses (University of Massachusetts at
sign) and engineering judgment. The main Lowell). The present paper provides a background
challenges of the project are therefore: (a) for design methodologies and the LRFD. Databases
Compilation of large, high quality databases and are presented and analyzed. Selected design meth-
(b) Framework for a procedure and data manage- ods are described, followed by an in depth evalua-
ment to enable: (i) LRFD parameter evaluation and tion of the dynamic methods for driven piles evalua-
(ii) Future updates. These challenges include two re- tion, and examination of the controlling parameters.
quirements: (i) Organization of the factors following The performance of different prediction methods is
then provided, categorized according to the exam- 2.2 Limit states design
ined methods of analysis and the controlling parame- In the 1950’s the demand for a more economical de-
ters. The obtained results are used for the develop- sign of piles brought about the use of Limit States or
ment of resistance factors to be recommended for the Limit States Design (LSD). The two types of limit
new specifications. states are the Ultimate Limit State (ULS) and the
Serviceability Limit State (SLS) both of which must
be satisfied when using the LSD method. Regardless
2 STATE OF STRESS DESIGN of the complexity of the analysis and calculation, all
METHODOLOGIES
limit states designs are carried out to satisfy the fol-
lowing criteria for ULS and SLS, respectively,
2.1 Working stress design
Working Stress Design (WSD) method, also called Factored resistance ≥ Factored load effects (2)
the Allowable Stress Design (ASD) method, has
been used traditionally in Civil Engineering since it Deformation ≤ Tolerable deformation to remain ser-
was introduced in the early 1800s. The design loads, viceable (3)
Q, consist of the actual forces estimated to be ap-
plied directly to the structure. ULS pertains to structural safety and involves the
collapsing of the structure, or in relation to piles,
R n Qult when the ultimate bearing capacity of the soil is ex-
Q ≤ Qall = = (1) ceeded. SLS represents the conditions affecting the
FS FS function or service requirements (performance) of
the structure under expected service/working loads.
where Q = Design load (F); Qall = Allowable design These conditions can include excessive deformations
load (F); Rn= Qult = Ultimate geotechnical pile force and settlement or deterioration of the structure -
resistance; and FS = Factor of safety. pile(s).
The factor of safety is commonly defined as the
ratio of the resistance of the structure (Rn) to the load
effects (Q) acting on the structure. Table 1, last used 3 LOAD AND RESISTANCE FACTOR DESIGN
in AASHTO, (1997), presents common practice in (LRFD)
which different factors of safety are used in conjunc-
tion with the level of control in the analysis of the
pile’s ultimate geotechnical resistance. Presumably, 3.1 Principles
when a more reliable and consistent level of con- The design of a pile depends upon predicted loads
struction control is used, a smaller factor of safety and the pile’s capacity to resist them. Both loads
can be implemented, which leads to a more eco- and capacity have various sources and levels of un-
nomical design. Practically, however, the proposed certainty that engineering design methods and proc-
factors of safety in Table 1 (excluding the static load esses have historically compensated for by experi-
test to failure) do not necessarily consider the bias, ence and subjective judgment. These uncertainties
in particular, the conservatism (i.e. under-prediction) can be quantified using probability-based design, or
of the examined method and, hence, their economi- safety check expressions, aimed at achieving engi-
cal validity is questionable (to be further discussed neered designs with consistent levels of reliability.
in section 9.3). The intent of the Load and Resistance Factor Design
(LRFD) method is to separate uncertainties in load-
Table 1. Factor of safety on ultimate axial geotechnical capac-
ity based on level of construction control (AASHTO, 1997). ing from uncertainties in resistance and to assure a
prescribed margin of safety.
Basis for Design and Type Increasing Design/Construction Figure 1 shows the probability density functions
of Construction Control Control
for the load effect, Q, and the resistance, R. “Load
Subsurface Exploration X X X X X Effect” refers to the load calculated to act on a
Static Calculation X X X X X particular element, (e.g. a specific pile). As the
Dynamic Formula X loads are more deterministic than the resistances, the
Wave Equation X X X X
load effect has smaller variability (smaller
coefficient of variation, translating to a narrower
CAPWAP Analysis X X
probability density function) than the resistance.
Static Load Test X X Where the two curves overlap, the load effect is
Factor of Safety (FS) 3.50 2.75 2.25 2.00* 1.90 greater than the resistance, indicating a high
*For any combination of construction control that includes a probability of failure.
static load test, FS =2.0.
 __
Q 3.2 Background information
The concept of using the probability of failure as a
Qn
criterion for structural design was presented in the
late 1920s by the Russians N. F. Khotsialov and N.
Load Ef f ect (Q) S. Streletskii, and was first introduced in the US by
Freudenthal (1947). Recent development of LRFD
in civil engineering was initiated in structural engi-
f R(R), f Q(Q)

__
Rn R
neering (e.g. Ellingwood et al., 1980). Reliability-
Based Design codes utilizing the LRFD format are
Resistance (R) used by the American Institute of Steel Construction
(AISC 1994, Galambos and Ravindra 1978) and by
the American Concrete Institute (ACI, 1995). An ef-
fort was made by the National Standards Institute
(ANSI) to develop probability-based load criteria for
buildings (Ellingwood et al. 1982a,b) and ASCE 7-
93 (ASCE 1993). The American Petroleum Institute
(API) extrapolated LRFD technology for use in
fixed offshore platforms API (1989), and Moses
(1985, 1986). Additional comprehensive summaries
R, Q for the implementation of modern probabilistic de-
Figure 1. An illustration of probability density functions for
load effect and resistance. sign theory into design codes include Siu, et al.
(1975) for the National Building Code of Canada
(1977), Ellingwood et al. (1980) for the National Bu-
In LRFD, partial safety factors are separately ap- reau of Standards, and the CIRIA 63 (1977). Ayyub
plied to the resistance and load components. A fac- et al. (1998a) provide details on LRFD rules for ship
tored (reduced) strength (capacity) of a pile is larger structures that were developed for the US Navy. The
than a linear combination of factored (magnified) AASHTO LRFD Bridge Design and Construction
load effects. In this format, the strength is reduced Specifications (1994) resulted from work in NCHRP
and the load effects are increased, by multiplying the Project 12-33 (Nowak 1993, 1999) provide design
corresponding characteristic (nominal) values with guidance for girders.
factors, which are called strength (resistance) and
load factors, respectively. The nominal values (e.g. 3.3 LRFD performance and advantages
the nominal strength, Rn) are those calculated by the
specific calibrated design method and are not neces- Experience has shown that adopting a probability-
sarily the mean strength or resistance, Q and R (see based design code results in significant savings
Figure 1). For example, while R and its distribution and/or efficient use of materials. Reliability im-
represent the mean and performance of dynamic sig- provements are still under evaluation even though
nal matching analyses predictions from many case the new codes are designed for reliability to be equal
histories, Rn is the predicted value for the specific to or better than the older codes. Experiences are not
analyzed pile. yet well documented, but designers have commented
Based on considerations ranging from analyzed that, relative to the conventional working stress
case histories to existing design practices, a pre- code, the new AISC-LRFD requirements are saving
scribed value is chosen for the probability of failure. from 5% to 30% steel weight, with a 10% typical
The probability for failure of the designed pile fol- savings (Ayyub, 1999). This may or may not be the
lowing the application of the resistance and load fac- case for other industries. Specific benefits in pile de-
tors should be smaller than the prescribed value (the sign include the following:
overlapping shaded area in Figure 1). Practically, the 1. A more efficiently balanced design results in cost
resistance factors are calculated to provide the pre- savings and/or improved reliability.
scribed probability of failure. Referring to the factor 2. Uncertainties in the design are treated more ra-
of safety principle, the mean factor of safety is R Q tionally and rigorously.
where as the nominal factor of safety is Rn/Qn. 3. An improved perspective of the overall design
and construction processes (sub- and super-
structures), and the development of probability-
based design procedures can stimulate advances
in pile analysis and design.
4. The codes become a living document that can be Today, the situation has changed somewhat, but
easily revised to include new information reflect- not entirely. The research team of NCHRP 24-17
ing statistical data on design factors. gathered robust data sources on pile capacity from
5. The partial safety factor format used herein also which a more objective calibration of resistance fac-
provides a framework for extrapolating existing tors can be made. Nonetheless, there remain uncer-
design practice to new foundation concepts and tainties associated with (1) site conditions, (2) soil
materials where experience is limited. behavior and interpretation of soil parameters, and
(3) construction methods and their quality. These
factors are difficult or impossible to understand from
3.4 LRFD in geotechnical engineering the pile databases alone. Such knowledge factors
Early use of limit state design for geotechnical ap- must be combined with the reliability-theory-based
plications was examined by the Danish geotechnical calibration of the database records to achieve a
institute (Hansen 1953, 1956) and was later formu- meaningful LRFD approach, hence, require a long
lated into code (Hansen, 1966). Independent load process of a major effort.
and resistance factors were used, with the resistance
factors applied directly to the soil properties rather
than to the nominal resistance. 3.5 LRFD for deep foundations
Considerable effort has been directed over the
past decade to the application of LRFD in geotech- 3.5.1 2001 AASHTO LRFD bridge design specifi-
nical engineering. LRFD approaches have been de- cations for driven piles
veloped in offshore engineering (e.g. Tang, 1993; The ultimate resistance (Rn) multiplied by a resis-
Hamilton and Murff, 1992), in general foundation tance factor (φ), which is equal to the factored resis-
design (e.g. Kulhawy et al. 1988, 1994, 1996), and tance (Rr), must be greater than or equal to the sum-
in pile design for transportation structures (Barker et mation of the loads (Qi) multiplied by a modifier (ηi)
al. 1991, O’Neill, 1995). and load factor (γi) for the strength limits states:
In geotechnical practice, uncertainties in resis-
tance principally manifest in site characterization, R r = φR n ≥ ∑η γ i i Qi (4)
soil behavior, and construction quality. The uncer-
tainties have to do with interpreting site conditions, where:
understanding soil behavior (e.g. its representation in
property values), and accounting for construction ef- η i = η Dη Rη I > 0.95 (5)
fects. Uncertainties in external loads are small com-
pared with uncertainties in soil and water loads and ηi = factors to account for effects of ductility (ηD),
the strength-deformation behaviors of soils. The ap- redundancy (ηR), and operational importance (ηI).
proach for selecting load and resistance factors de- The 2001 AASHTO LRFD Bridge Design Speci-
veloped in structural practice, though a useful start- fications for Foundations provides the following
ing point for geotechnical applications, is not equations for the determination of the factored bear-
sufficient. Work is needed to incorporate in the ing resistance of piles, QR:
LRFD formulation factors that are unique to geo-
technical design. Q R = φ Q n = φ q Q ult = φ qp Q p + φ qs Q s (6)
Philosophically, the selection of load and resis-
tance factors need not be made probabilistically, al- for which:
though in current structural practice a reliability-
theory-based calibration is commonly used. This Q p = q p Ap (7)
approach focuses more on load uncertainties than re-
sistance uncertainties, and does not include many Q s = q s As (8)
subjective factors unique to geotechnical practice.
An expanded approach is needed if the full benefits where φq = resistance factor for the bearing resis-
of LRFD are to be achieved for foundation design. tance of a single pile specified for methods that do
The National Research Council reported, “this… not distinguish between total resistance and the indi-
subjective approach reflects the general lack of ro- vidual contributions of tip resistance and shaft resis-
bust data sources from which a more objective set of tance; Qult = bearing resistance of a single pile (F);
factors can be derived” (NRC, 1995). The report Qp = pile tip resistance (F); Qs = pile shaft resistance
continues, “…realistically, because of the tremen- (F); qp = unit tip resistance of pile (F/L2); qs = unit
dous range of property values and site conditions shaft resistance of pile (F/L2); As = surface area of
that one may encounter, it is unlikely that com- pile shaft (L2); Ap = area of pile tip (L2); and φqp, φqs
pletely objective factors can be developed in the = resistance factor for tip and shaft resistance, re-
foreseeable future.” spectively, for those methods that separate the resis-
tance of a pile into contributions from tip resistance therefore unique by providing a guide for choosing
and shaft resistance. the appropriate resistance factor. Interestingly, no
The resistance factors for use in the above equa- distinction is made regarding either soil type or time
tions are presented in Table 10.5.5-2 of the of driving (i.e. EOD, BOR) when referring to the
specifications. Details are provided for different signal matching based on dynamic measurements.
design methods based on soil type and area of The method by which the resistance factors were
resistance (tip and side). The resistance factors for generated is not provided in the code.
compression vary between 0.45 to 0.70, but the table The AUSTROADS Bridge Design Code (1992)
incorporates a factor λv, to account for the level of provides resistance factors for the construction stage
field capacity verification. As an example, if α alone including static load test (to failure φ=0.9,
method analysis is used to determine the pile’s proof test φ=0.8), and four categories of dynamic
friction resistance in clay, a resistance factor of 0.70 methods. The range of resistance factors is quite
is recommended. If a pile driving formula, e.g. ENR large and there is no explanation as to how the resis-
equation, is used to verify the pile capacity without tance factors were obtained. Goble (1999) postu-
stress wave measurements during driving, a λv factor lates that the resistance factors were calibrated via
of 0.80 is recommended. This results in an actual the working stress design method.
resistance factor of 0.56 to be used in the above The Ontario Bridge Code (1992) recommends
analysis/verification sequence (0.70 × 0.80). relatively low resistance factors with no differentia-
tion between the individual static or dynamic analy-
3.5.2 2001 AASHTO LRFD bridge design specifi- ses. For example, the resistance factors for static
cations for drilled shafts analyses and static load tests in compression and
Detailed resistance factors for a large number of dif- tension are 0.4, 0.3, 0.6 and 0.4 respectively. No in-
ferent design methods are provided. Differentiation formation is provided on how the resistance factors
is made between base and side, similarly to driven were obtained.
piles, with resistance factors varying overall between The Canadian Bridge Code (1992) is brief in its
0.45 to 0.65. Static testing is included with the same design requirements for deep foundations. Resis-
resistance factor as used for driven piles (0.8). Resis- tance factors are based on pile type, φ=0.4 for all
tance factors are not provided for drilled shafts in timber and concrete piles (precast, filled pipe, and
sand. The factor λv used for field verification for cast in place) and 0.5 for steel piles. For dynamic
driven piles, is not used for drilled shafts, and practi- load testing, resistance factors of 0.4 and 0.5 are rec-
cally no distinction is made on the basis of construc- ommended for routine testing and analyses based on
tion method. dynamic measurements, respectively.
Eurocode 7 (1997) deals with driven piles and
3.5.3 Worldwide LRFD for deep foundations drilled shafts at a single section. Factors for static
A review of foundation design standards in the load testing depends on the number of tested piles
world was conducted by the Japanese Geotechnical (irrelevant to the number of piles at the specific site).
Society (1998). A Review of the development of Range of values from 0.67 to 0.91 is provided for
LRFD applications for Geotechnical Engineering is one to three tests, related to the mean or lowest value
presented by Goble (1999). A short description of of the test results. The code is quite complex with
LRFD codes for deep foundations follow. quantitative descriptions and limiting conditions.
The Australian Standard for Piling-Design and The code is presented with multiple component fac-
Installation (1995) provides ranges of resistance fac- tors, and for comparison with the form used by USA
tors for static load tests (0.7 to 0.9) and static pile codes, Goble (1999) inverted and combined the fac-
analyses (0.40-0.65) related to the source of soil pa- tors resulting with values ranging from 0.63 to 0.77
rameters and soil type (e.g. SPT in cohesionless for base, skin, and total resistance of driven, bored,
soils). Detailed recommendations are provided for and CFA piles.
resistance factors to be used with the dynamic meth- Substantially less details are provided by the
ods ranging between 0.45 to 0.65 for methods with- codes for LRFD design of drilled shafts. The two ex-
out dynamic measurements (including WEAP), and tremes being the aforementioned Canadian Bridge
between 0.50 to 0.85 when utilizing dynamic meas- Code (1992), in which drilled shafts are included
urements with signal matching analysis. Selection of under a single category of cast in place piles (φ=0.4
the appropriate resistance factor depends on driving like all other concrete piles), and the AASHTO de-
conditions, geotechnical factors (e.g. extent of site tailed provisions described in section 3.5.2.
investigation), and extent of testing (e.g. low range
for <3% of the pile tested and high range for >15%). 3.5.4 Difficulties of the existing LRFD codes
In traditional structural design specifications, a All existing codes suffer from two major difficulties.
nominal value is given and the value used is based One is the application of LRFD to geotechnical
primarily on engineering judgment and cannot ex- problems as described in section 3.4 (e.g. site vari-
ceed the nominal value. The Australian Standard is ability, construction effects, etc.). The other problem
is lack of data. None of the reviewed codes and as- Table 3. Summary of the driven piles database:
sociated resistance factors were developed based on Soil type Number of Cases
databases enabling the calculation of resistance fac- Tip Side H-PILES PPC PIPE
Clay 3 0 0
tors from case histories. Sand 12 0 0
The current AASHTO specifications of driven Rock
Mix 6 15 3
piles reviewed in section 3.5.1 encounter additional Total 21 15 3
difficulty due to the multiplication of the resistance Clay 0 0 0
factor by the modifier λv. This procedure requires Sand
Sand 17 37 20
the interaction of two independent evaluations (e.g. Mix 13 50 19
static analysis and dynamic methods) and results in Total 30 87 39
unnecessary and confusing conservatism. A clear Clay 8 19 20
Sand 1 1 0
separation of the resistance factors on the basis of Clay
Mix 36 34 15
design and construction is required and is the aim of Total 44 54 35
the presented code development. Insufficient data 0 7 1
All cases 97 163 78

4 DATABASES
4.3 Driven pile database – dynamic analysis
Three major databases were developed for the pri- The database PD/LT2000 contains information re-
mary statistical evaluation of resistance factors for lated to 210 driven piles that have been statically
the design and construction of driven piles and load tested to failure and dynamically monitored
drilled shafts. during driving and/or restrike (403 analyzed meas-
urements). PD/LT2000 is comprised of the integra-
4.1 Drilled shaft database – static analysis tion of databases PD/LT (Paikowsky et al. 1994) and
PD/LT2 (Paikowsky and LaBelle, 1994) with expan-
The drilled shaft database summarized in table 2 was sion by an additional 57 pile cases as described by
developed at the University of Florida mostly as an Paikowsky and Stenerson (2000). The data of
integration of databases gathered by the Florida De- PD/LT2000 were carefully examined and analyzed
partment of Transportation (FDOT), the Federal following procedures described by Paikowsky et al.
Highway Administration (FHWA) and by O’Neill et 1994, resulting in detailed static and dynamic pile
al. (1996). The general database (prior to the final capacity evaluations. Table 4 presents a summary of
selection process) is presented by Kuo et al. (2002). the data contained in PD/LT2000 broken down ac-
cording to site location and soil type, pile type and
Table 2. Summary and breakdown - drilled shafts database.
capacity, driving behavior and time of driving.
Soil/Rock
Method of Construction
Type
Total=skin + Casing Slurry Dry
tip 5 PRINCIPLES AND FRAMEWORK FOR THE
Skin=side Total Skin Total Skin Total Skin CALIBRATION
alone
Sand 13 6 15 4 6 1
Clay 14 3 0 0 40 10
5.1 Existing AASHTO specifications
Mixed Soils 23 4 12 5 13 7
Rock 0 0 0 0 8 0 The existing AASHTO specifications are based on
Sand & Rock 4 4 7 5 20 0 the First-Order, Second-Moment (FOSM) principles,
Clay & Rock 2 0 2 0 19 7 and when using η=1 in equation 4, and assuming
Mixed Soils
& Rock
2 1 0 0 2 0 lognormal distribution for the resistance and bias
Total (256) 58 32 36 14 91 25 factors, the following relations can be established
(Barker et al., 1991).
2
1 + COVQ
4.2 Driven pile database – static analysis λR (∑ γ iQi ) 2
1 + COVR (9)
The driven pile database summarized in table 3 was φ=
Q exp{βT ln[(1 + COVR )(1 + COVQ )]}
2 2
developed at the University of Florida mostly as an
integration of databases gathered by the University
of Florida (UOF), FHWA database (see e.g. DiMil- where:
lio, 1999), the University of Massachusetts Lowell λR = resistance bias factor
(UML) (see e.g. Paikowsky et al., 1994), and the COVQ = coefficient of variation of the load
Louisiana Transportation Research Center (LTRC). COVR = coefficient of variation of the resis-
Some of the data is presented by Nguygen et al. tance
(2002). βT = target reliability index
Table 4. Summary of the data in the PD/LT2000 database.

Geographical
Pile Types Soil Types Soil Inertia Type of Data Pile Capacities
Location
Soil Blow Range
Pile Type No. Location No. Side Tip Criteria AR Time No. No.
Type Ct. (kN)
Northeast
H –Pile 37 44 EOD 0-445 2
USA ≥ 16 & 92
Southeast BOR
OEP 10 69 blows 272 ---- 445-890 6
USA Clay
67 61 /10cm
North /Till
CEP 61 24 EOD 890-1334 17
USA
& 30
Voided South
35 10 BOR’s 1334-1779 44
Concrete USA
< 16
Northwest
254 9 3 blows 112 ---- 1779-2224 27
USA
/10cm EOD 135
Southwest
305 5 14 2224-2669 25
USA Rock 0 11
356 8 Australia 2 2669-3114 15
BOR 239
406 1 Brunswick 3 ≥ 350 ----- 134 3114-3559 10
Sq.
Conc
457 8 Holland 4 3559-4003 13
EOR 11
Hong
508 8 4 4003-4448 13
Kong Sand
140 137
/Silt
610 16 Israel 4 < 350 ----- 255 4448-4893 11
DD 2
762 5 Ontario 22 4893-5338 6

Octagonal Sweden 1 5338-5783 5

3 DR 1
Concrete NA 3 1 NA 5 ---- 5783-6228 4
NA 6
Timber 2 ALT 1 6228-6672 6
Monotube 2 >6672 6
Total 210 210 210 210 389 389 389 210
Notes:
1. OEP - Open Ended Pipe Pile 11. No. - Number of Piles / Cases
2. CEP - Close Ended Pipe Pile 12. Sq. Conc. - Square Concrete
3. EOD - End of Driving 13. NA - Non Applicable / Unknown
4. BOR - Beginning of Restrike 14. USA - United States of America
5. EOR - End of Restrike 15. Northeast USA - Federal Highway Regions 1, 2 & 3
6. DD - During Driving 16. Southeast USA - Federal Highway Region 4
7. DR - During Restrike 17. North USA - Federal Highway Regions 5, 7, & 8
8. ALT - Alternate measurement 18. South USA - Federal Highway Region 6
9. Blow Ct. - Blow Count 19. Northwest USA - Federal Highway Region 10
10. AR - Area Ratio 20. Southwest USA - Federal Highway Region 9

When just the dead and live loads are considered 5.2 Present project calibration
equation 9 can be written as: The LRFD for the design of Structures was evolved
around modern methodology in the probability area
γ D QD (1 + COVQD 2 + COVQL 2 )
λR ( +γ L ) [ ] (10) (e.g. Ellingwood et al., 1980, Ravindra and Galam-
QL (1 + COVR 2 )
φ=
λQD QD
bas, 1978), while the geotechnical applications of re-
(
QL
+ λQL ) exp{βT ln[(1 + COVR 2 )(1 + COVQD 2 + COVQL 2 )]} liability in the US lag behind, (Meyerhof, 1994). It
is important that the calibrations of resistance factors
for deep foundations make use of methodology con-
where: γD, γL dead and live load factors sistent with procedures used to calibrate load factors.
QD/QL dead to live load ratio Otherwise, the factors for the substructure and su-
λQD, λQL dead and live load bias factors perstructure can be internally inconsistent with re-
spect to the statistical assumptions of distributional
form, uncertainty in structure, and correlation among
variables on which they are based (Nowak, 1999).
 Following Ayyub and Assakkaf (1999) and Ay- Definition Define Limit States at Single Pile
yub et al. (2000) the present project calibrates LRFD of Failure Level: Ultimate & Serviceability
partial safety factors using the First-Order Reliability
Method (FORM). FORM can be used to assess the
reliability of a pile with respect to specified limit Define Statistical Characteristics
states, and provides a means for calculating partial of Basic Random Variables
safety factors φ and γi for resistance and loads, re- Resistance Load
spectively, against target reliability levels, βO.
FORM requires only the first and second moment in-
formation on resistances and loads (i.e. means and Determine Model Determine Load
standard deviations), and an assumption of distribu- Uncertainty for Uncertainties from
tion type (e.g. normal, lognormal, etc.). The entire Strength (from Superstructure to
calibration process is presented in Figure 2. database) Foundation (from
In design practice, there are usually two types of ST code)
limit states: ultimate limit states and serviceability
limit state. Each can be represented by a perform- MC Simulation or Probability
ance function of the form: Calculation to Get Statistical
Properties of Scalar R
g ( X ) = g ( X 1 , X 2 ,..., X n ) (11)
Back-calculated Beta Reliability
in which X is a vector of basic random variables (X1, vs Load Ratio Curves Assessment
X2,…,Xn) for strength and loads. The performance in Practice
function g(X) is sometimes called the limit state
function. It relates the random variables for the
limit-state of interest. The limit state is defined when Assign Target Betas
Review Target
g(X) = 0, and therefore, failure occurs when g(X) < 0 Betas in the
(see Figure 2). The reliability index β is defined as Literature and
the distance from the origin of the space of basic Practice
random variables (X1, X2,…, Xn) to the failure sur-
face at the most probable point. The most probable Calculate Load and Resistance Factors
failure point is that point on the limit state function
at which the probability density of the basic random
variables is greatest. This is also called the design Select Load and Resistance Factors
point, which is solved for iteratively (Thoft-
Christensen and Baker, 1982, Madge et al., 1985,
Ayyub and McCuen, 1997). This relationship can Adjust for Mean/Nominal Parameters
also be used to back calculate representative values
of the reliability index β from the current design
Case Study Designs for Comparison
practice.
In developing design code provisions for piles, it
is necessary to follow the current design practice to G(x)=0 GL(x)=0
ensure consistent levels of reliability over various
pile types. Calibrations of existing design codes are Contours of
needed to make the new design formats as simple as Failure fRS = fX(x)
possible and to put them in a form that is familiar to Region
users or designers. For a given reliability index β Safe
and probability characteristics for the resistance and Region
load effects, the partial safety factors determined by
the FORM approach might be different for different µS
failure modes for the same or differing component.
For this reason, calibration of the calculated partial
safety factors (PSF’s) is important in order to main-
tain the same values for all loads at different failure
modes. In the case of geotechnical codes, the cali- µR
bration of resistance factors is performed for a set of
load factors already specific in the structural code. Figure 2. Resistance factor analysis flow chart (after Ayyub
Thus, the load factors are fixed. In this case, the fol- and Assakkaf, 1999 and Ayyub et al., 2000).
lowing algorithm is used to determine resistance fac-
tors: (1) For a given value of the reliability index β,
probability distributions and moments of the load The present effort requiring the assembly of actual
variables, and the coefficient of variation for the re- case history database is by itself a major undertaking
sistance, compute mean resistance R using FORM. requiring the immense effort of database collection
(2) With the mean value for R computed in step 1, and analysis. Many difficulties are added when one
the partial safety factor φ is revised as: tries to obtain the calculated or “predicted” capacity
n for comparison with the measured static load-test
∑γ i µ Li value. In order to challenge the problem in its cardi-
φ= i =1
(12) nal aspects, one needs to address various issues,
µR which are further discussed in section 13.2.
where µLi and µR are the mean values of the loads The present effort was focused, therefore, on
and strength variables, respectively and γi, i = 1, calibrating the direct design and construction proc-
2,…, n, are the given set of load factors. ess. For the design, specific methods and correla-
tions were chosen and followed. Their results (i.e.
static capacity evaluation) were compared to the ac-
5.3 Loading conditions and load factors tual pile performance under load.
The actual load transformed from the superstructure In the dynamic analysis case, the database was
to the foundations is, by and large, unknown with utilized (along with the mechanics of the problem) to
very little long-term research focusing on the sub- guide for the controlling parameters, which were
ject. The load uncertainties are taken, therefore, as then calibrated. A short description of the principles
those used for the superstructure analysis. AASHTO used for the assessments of the three databases fol-
LRFD bridge design specifications provide five load lows.
combinations for the strength limit state. The use of
a load combination that includes lateral loading may 5.4.2 Driven piles – static analysis
be the restrictive loading condition for design. Pile The vast majority of the database case histories were
lateral capacity is usually controlled by service limit related to SPT and CPT field testing. Four correla-
state, and as such, was excluded from the scope of tions of soil parameters from SPT and CPT were
the present study, focusing on axial capacity of sin- identified. The case histories were divided on the
gle piles/drilled shafts. The load combination basis of soil condition (clay, sand, and mixed) and
strength I was applied with the following load fac- pile types (H pile, concrete piles, pipe piles). In
tors and lognormal distributions for live and dead summary, the given field conditions were used via
loads, respectively: various soil parameter identifications, and pile ca-
pacity evaluation procedures to determine capacities,
γL = 1.75 λQL = 1.15 COVQL = 0.2 (13) which were then compared to the measured static
γD = 1.25 λQD = 1.05 COVQD = 0.1 (14) capacity. Details of the analyses are presented in
section 7.
5.4 Conceptual evaluation of driven piles and 5.4.3 Driven piles – dynamic analysis
drilled shafts capacities The dynamic evaluation of driven piles is of cardinal
importance, as it is the most common way to deter-
5.4.1 Overview mine capacity during construction. The existing
Figure 3 presents a flowchart depicting the design
AASHTO specifications as described in section
and construction process of deep foundations. A
3.5.1 are especially problematic due to the incorpo-
complete application of LRFD requires a framework
ration of the factor λv, which tangles the design stage
that deals with each step of the process. For exam- – static evaluation, with the construction stage – dy-
ple; (i) the evaluation of the method by which a spe- namic evaluation. As such, a completely fresh look
cific field test (say SPT) was used to obtain the soils on the basis for dynamic calibrations and the
shear strength (say φ), (ii) the evaluation of the spe- controlling factors was required. The developments
cific method utilized the shear strength for the static
are described in more details by Paikowsky and
capacity of the pile, and for both, the assessment of
Stenerson (2000) and its major findings are provided
different sources of uncertainty. Independently, one
in section 8.
needs to evaluate the design verification process that
takes place during construction, i.e. static load test-
5.4.4 Drilled shafts – static analysis
ing and dynamic testing as it is used separately from Evaluation of the design of drilled shafts is difficult
the design to assess and modify the actual pile instal- as little reliable data are available for the separation
lation. Added to that, is the quality assurance (e.g. between the different capacity components (i.e. shaft
non-destructive testing of drilled shafts) and other and tip), and both are largely affected by the method
related issues. Previous LRFD developments using of construction. The following procedure was used
back analysis of ASD and judgment, addressed some therefore, for the evaluation of the “measured” skin
of the different issues (see e.g. Withiam et al., 1988).
 Field Exploration
& Testing

Dynamic Analysis
Geomaterial Strength & Static Analysis of Superstructure
of Driven Piles
Deformation Parameters Deep Foundations Loading
Evaluation
Deep Foundation
Type/Construction
Method
Laboratory
Deformation
Testing
and Bearing Capacity Superstructure
Settlement Vertical and Loading
Lateral Resistance Requirement
Single/Group
Design
• Geometry
• Configureation
• Installation
Criteria No

Design
Testing Verification/ ?
Completed • Material QC Construction OK
Modification
Substructure • Performance Monitoring
• Dynamic testing
• Driving • Static testing
• integrity Yes

Figure 3. The process of deep foundation design and construction

capacities. The shape of the load-displacement

curves was evaluated, and the shafts for which more 6.2 Failure criterion for statically loaded driven
than 80% of the total capacity was mobilized in a piles
displacement of less than 2% of the shaft’s diameter, Past work related to driven piles (Paikowsky et al.
were considered as those in which the resistance is 1994) have resorted to a “representative” static pile
based on friction, and comparisons were held with capacity based on the assessment of five interpreta-
the shaft resistance calculated values. The results of tion methods; Davisson’s Criterion (Davisson,
these procedures were compared to static analyses as 1972), Shape of Curve (similar to the procedure pro-
described in section 9. posed by Butler and Hoy, 1977), Limiting Total Set-
tlement to 25.4 mm and to 0.1B (Terzaghi, 1942),
and the DeBeer log-log method (DeBeer, 1970). A
6 DEEP FOUNDATIONS NOMINAL single representative capacity value was then calcu-
STRENGTH lated for the analyzed case as the average of the
methods considered relevant (i.e. provided reason-
able value). The development of a calibration in a
6.1 Overview framework suitable for future modifications requires
Probabilistic calibration of resistance factors for any that the evaluated resistance factors be based on an
predictive method utilizing a database is possible objective, reproducible procedure. In order to do so,
when the nominal geotechnical pile strength (i.e. the static capacity of each pile in database
static pile capacity) is defined and compared to the PD/LT2000 was evaluated utilizing all five afore-
outcome of the calibrated prediction method. The mentioned criteria and a representative capacity was
definition of ultimate static capacity given static load assigned for each pile. A statistical analysis was
test results (load-displacement relations) is not then carried out by determining the mean and stan-
unique, and the use of the term reference static ca- dard deviations of the ratio of the representative pile
pacity for calibration may be more appropriate than capacity to the method being evaluated. Figure 3
nominal strength. The static load test results depend shows the histogram and calculated distributions
on the load testing procedures and the applied inter- (normal and lognormal) for Davisson’s failure crite-
pretation method, often being subjective. The fol- rion. Davisson’s criterion was found to perform the
lowing sections examine each of these factors and its best overall and was therefore chosen as the single
influence on the reference static capacity. method to be used when analyzing load-
displacement curves. The method provides an ob-
 jective failure criterion and was also found to per- The effect of the test type was further investi-
form well for piles exceeding a diameter of 610mm gated utilizing a database containing information re-
(examined through 30 pile cases). The data pre- lated to 75 piles tested under slow maintained and
sented in Figure 4 demonstrates, however, that: (i) static-cyclic load testing procedures. In the static-
though small, a bias exists in the static capacity used cyclic procedure, the piles were loaded to failure us-
as a reference for the evaluation of predictive meth- ing a high loading rate and then unloaded. The proc-
ods for capacity evaluation of driven piles, and (ii) ess was repeated for four cycles. The testing
this bias (and other considerations) needs to be con- procedure and its interpretation method are pre-
sidered when evaluating the resistance factor to be sented by Paikowsky et al. (1999). A comparison be-
used for field static load tests. tween the pile capacity based on Davisson’s failure
45
criterion for the slow maintained tests and the static-
0.24
cyclic capacity is presented in Figure 5. The ob-
40
0.22 tained relations and the associated statistical infor-
0.2 mation suggest that there is no significant influence
35 on the static pile capacity based on the applied static
0.18
load rate. The static cyclic load test results were
30 0.16 also compared to the representative static pile capac-
Number of Pile-Cases

Relative Frequency
log-normal
0.14
ity (based on the aforementioned five methods) re-
distribution
25 mx = 1.018 sulting in a mean KSC of 1.023 and a standard devia-
^
mx = 1.013 0.12 tion of 0.057.
20 These evaluations led to the conclusions that
σlnx = 0.0829 0.1
Davisson’s pile failure criterion can be used as a
15 normal distribution 0.08
method to determine the reference pile capacity for
0.06 driven piles irrespective of the static load-testing
10 σx = 0.1010
0.04
procedure.
5 1400
Slow Maintained Static Load Test - Davissons Criterion (kN)

0.02
Equation Y = 0.924 * X
0 0 Number of data points used = 75
1200 R2 = 0.978351
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
Ratio of Representative Pile Capacity over the Pile
Capacity based on Davisson's Failure Criterion 1000

Figure 4. Histogram and frequency distributions of KSD for 800

186 PD/LT2000 pile-cases in all types of soils. (Paikowsky and
Stenerson 2000)
600

6.3 Load test procedure for statically loaded driven 400

piles Mean KSC = 0.930
Standard Deviation = 0.136
An additional factor to examine is the influence of 200 Max KSC = 1.215
the static load testing procedure (loading rate) on the Min KSC = 0.577
designated pile capacity. This influence was exam- 0
ined in two ways. 0 200 400 600 800 1000 1200 1400
Two detailed case histories from a research site Static Cyclic Load Test Capacity (kN)
in Newburyport, Massachusetts, were evaluated. A
pipe and pre-stressed concrete heavily instrumented 1 kN = 0.2248 kips

friction piles were tested over a lengthy period of

Figure 5. Comparison between pile capacity based on Davis-
time at a bridge reconstruction site. Both piles were son’s criterion for slow maintained load tests and static cyclic
tested using three types of static load testing proce- load test capacity for 75 piles, (Paikowsky et al., 1999).
dures: slow maintained (testing duration of about 45
hrs), short duration (testing duration of about 6-8
hrs), and static cyclic (testing duration of about 15 6.4 Failure criterion for statically loaded drilled
min.). Details about the piles and the testing are pre- shaft
sented by Paikowsky and Hajduk (1999, 2000) and Static load tests of small to medium capacity drilled
Paikowsky et al. (1999). The interpretation of the shafts (say up to 5 MN) are similar to that of driven
load-displacement relationships in both cases sug- piles. It is common, however, to design and build
gested that the test type had an insignificant influ- high capacity drilled shafts (10 MN and more) often
ence on the pile capacity, (referring to a failure crite- as an alternative to a large group of small capacity
rion irrespective of the displacement). driven piles (e.g. in the Northeast region of the
USA). The testing for capacity of such shafts is a 8.2 Methods of analysis
challenge that often requires alternative methods to
the common external reaction testing, e.g. Osterberg 8.2.1 General
load-cell (Osterberg, 1992), statnamic tests (Ber- Table 9 presents a summary of the major available
mingham and White, 1995, Middendrop and Biele- dynamic methods for evaluating pile capacity. The
feld, 1995) and drop weight dynamic testing. Addi- methods are subdivided according to the project
tional National Cooperative Highway Research stage (i.e. design vs. construction) and the need for
Program project, NCHRP-12-55 “Innovative Load data obtained through dynamic measurements. The
Testing Systems” headed by the author currently ex- incorporation of dynamic equations and WEAP re-
amines these issues. As part of this ongoing project, flects the state of practice.
the static load-test results of statically loaded drilled The methods that require dynamic measurements
shafts were examined utilizing the failure criteria can be broadly categorized as those that utilize a
previously described for driven piles, and the FHWA simplified analysis of an instantaneous pile capacity
criterion for drilled shafts (O’Neil and Reese, 1999). evaluation for each hammer blow, and those that re-
This criterion establishes the failure load as that as- quire elaborate calculations (i.e. wave matching),
sociated with a displacement of five percent of the traditionally carried out in the office.
diameter at the shaft, if plunging of the shaft cannot
be achieved. The results of this preliminary study 8.2.2 WEAP
are presented in Table 5, suggesting that the use of Based on Smith (1960), the use of the WEAP (Goble
the FHWA criterion provides a reliable and a simple and Rausche, 1976) during design is of great impor-
failure interpretation. For the presented LRFD cali- tance for achieving compatibility between the driv-
bration study, the FHWA failure criterion for drilled ing system, the pile, and the soil conditions. Driv-
shaft (i.e. load at a displacement of 0.05 B) was, ability study and pile stress analysis often determine
therefore, adopted. the pile type and geometry and the adequacy of the
proposed equipment. The evaluation of WEAP ef-
fectiveness for capacity predictions is difficult to as-
7 DRIVEN PILES – STATIC ANALYSIS sess as a large range of input parameters is possible
METHODS and the results are greatly affected by the actual field
conditions. Examination of the method through
Table 6 present a summary of the methods used for analyses making use of default values is probably
static capacity evaluation of driven piles. The asso- the best avenue. Other evaluations, including WEAP
ciated correlations used to evaluate the soil proper- analysis adjustments following dynamic measure-
ties from SPT and CPT tests are presented in Tables ments (e.g. matching energy) seem to be impractical
7 and 8 respectively. The tables were, by and large, in light of the other available methods, and lead to
prepared as part of the study conducted for the static questionable results regarding their quality and
pile capacity at the University of Florida and one meaning (Rausche et al. 1997, Rausche, 2000). The
presented by Nguygen et al. (2001). WEAP analyses are evaluated in this study in two
ways: (i) as a dynamic method for pile capacity pre-
diction, using WEAP default input values and the
8 DRIVEN PILES – DYNAMIC ANALYSIS pile's driving resistance at the end of driving (EOD)
METHODS compared to the static load test results and (ii)
WEAP as a pile design method examining the ana-
lyzed stresses at the design stage with the measured
8.1 Overview stresses during construction. Such evaluation leads
Prior to detailed analyses leading to the determina- to a strength factor (related to the allowed structural
tion of resistance factors, two components must be stresses in the pile) that is beyond the scope of this
established: (a) the type of the dynamic methods to manuscript.
be evaluated and (b) the conditions under which
these methods need to be examined. Sections 8.2 8.2.3 Dynamic Equations
and 8.3 address these issues, respectively. The choice of dynamic equations came to address
the state of practice and reflect a range in equation
Table 5. Evaluation of failure criteria for statically loaded drilled shafts.
Statistics for the Ratio between Drilled Shaft Capacity of different Interpretation Methods and the Representative Capacity
Davisson DeBeer Shape of Curve FHWA
# mx σx # mx σx # mx σx # mx σx
47 0.862 0.17 39 0.908 0.11 36 0.956 0.09 40 0.999 0.13
# - no. of cases mx=mean σx=standard deviation loads 0.85 to 20 MN diameter 0.3 to 1.5m length 5.3 to 58.5m
Table 6: Summary of the methods
Parameters
Methods Side resistance Tip resistance Constraints
required
Su; +Bearing layer must be stiff
α-Tomlinson
Db (bearing cohesive
(Tomlinson, 1980/1995)
qs = αSu embedment) + Number of soil layers ≤ 2
α-API (Reese et al.,
Su
1998) q p = 9 Su
β in cohesive (AASHTO,
qs = βσ’ OCR
1996/2000)
λ (US Army Corps of
qs = λ(σ’+2Su) Su Only for cohesive soils
Engineers, 1992)
β in cohesionless
βσ’ Dr
(Bowles, 1996)
Nordlund and Thurman sin(δ + ϖ ) qp =
qs = Kδ C Fσ ' φ
(Hannigan et al., 1995) cos ϖ αt N’q σ’
Meyerhof SPT (Meyer- qp = + For cohesionless soils
qs = k N N
hof, 1976/1981) 0.4D/BN’ + SPT data
Schmertmann SPT (Lai
qs = function(N) qp = fn(N) N SPTdata
and Graham, 1995)
Schmertmann CPT
(McVay and Townsend, qs = function(fs) qp = fn(qc) q c, f s CPT data
1989)
able through the years (e.g. Olsen and Flaate, 1967)
Table 7: Correlations of soil properties from SPT it was founded on a solid theoretical basis and is
Properties From SPT Reference used in construction in about half of the states in the
Peck, Hanson and USA. The equation's traditional formulation (e.g.
Thornburn: Massachusetts State Building Code, 1997) includes a
Figure 4.12
≈ 54 - 27.6034 exp(- factor of safety of 6, which need to be recognized.
φ
Kulhawy & Mayne, 1990

0.014N’)
Schmertmannϕ' Figure 4.13
The Gates equation (Gates, 1957), while empirical,
≈ tan-1 [ N / (12.2 + 20.3 σ') and Equa- was found to provide reasonable results in the past
] 0.34 tion 4.11 (e.g. Olsen and Flaate, 1967 and Long et al., 1998).
Terzaghi and Peck (1967): Equation The equation was further enhanced by Richard Che-
0.06 N 4.59 ney of the Federal Highway Administration,
Su (bar)
0.72 Equation (FHWA, 1988), based on a statistical correlation
Hara 1974: 0.29 N
4.60
OCR Mayne and Kemper Figures 3.9
provided by Olsen and Flaate (1967) and had the fol-
for clay ≈ 0.5 N / σ’o (σ’o in bar) and 3.18 lowing format:
Figures 2.13
Dr Gibbs and Holtz’s Figures
and 2.14 Ru = 1.75 × E × log 10 N − 100 (15)

Table 8. Correlations of soil properties from CPT where: Ru= Ultimate capacity (tons)
Properties From CPT Reference E = Gross energy of pile hammer, ft-lb
Figure 0.75E for drop hammers
Robertson and Campanella: 4.14
φ 0.85E for all other hammers
atan(0.1+0.38*Log(qc/σ')) and Eq.
N = Number of blows per inch
Kulhawy & Mayne, 1990

4.12
Theoretical: ( qc - σo ) / Nk Eq. 4.61
Su (bar)
qc and σo in bars. 8.2.4 The Case method
OCR Mayne: 0.29 qc / σ’o Figure The Case method (Goble et al., 1970 and Rausche et
for clay qc and σo in bars. 3.10 al., 1975) is often used in field evaluations, as it is
Jamiolkowski: 68 log(qcn) – 68 Figure
2.24
built into Pile Dynamics Inc.’s Pile Driving Ana-
q'c lyzer (PDA), the most commonly used in the USA.
qcn = (dimensionless) and Eq.
Dr Pa σ'0 2.20 The method is based on a simplified pile and soil
q'c = qc / Kq behavior assumptions (free end and plastic soil), re-
Kq = 0.9 + Dr/300 sulting in a closed form solution related to the im-
qc and σ'o in bars. pact and its reflection from the tip. With the years,
the method evolved to be implemented into at least
type and performance. While the Engineering News five different variations (GRL, 1999). The Case
Record (Wellington, 1892) was proven to be unreli-
Table 9. Summary of the available dynamic methods.

Category Method Advantages Disadvantages Comment

WEAP
- Equipment Match - Non unique Analysis - Required for Construction
(Smith, 1960,
Design Stage - Drivability Study - Performance sensitive to - Required Evaluation for
Goble et al.,
- Structural Stresses field conditions capacity predictions
1976)
ENR
- Sound Principles - Needs to be examined
(Wellington, - Unreliable
- Common use without a built in FS.
1892)
Dynamic Gates - Empirical - Depends on original - Found to be more reliable
Equations (Gates, 1957) - Common use database than other equations
FHWA version
- Correction based on
of Gates Eqn. - Depends on database - Was found to be reliable
additional data
(FHWA, 2000)
Signal Matching - Solid principle of
- Stationary soil forces
(e.g. CAPWAP) matching calculations - Office Method
- Expensive
(Goble et al., to measurements by - Found reliable at BOR
- Requires time
1970) imposing msd. B.C.
Case Method - Requires local calibration - Was found reliable with
(Goble et al., - Simplified Analysis - Presumed dependency of local calibration
Dynamic
1970, Rausche et - Field Method soil conditions found - How to obtain national or
Measurements
al., 1975) baseless international calibration?
Energy
Approach - Shows long-term capacity
- Simplified Analysis
(Paikowsky, which may not be present - Ideal for construction
- Field Method
1982, Paikowsky at EOD
et al., 1984))

method utilizes a damping coefficient (Jc) that is as- graphical location may be unwise and/or unsafe, the
sumed to be associated with soil type. The influence Case method was excluded from the examined dy-
of this factor on the predicted static capacity depends namic analyses.
on the reflected wave from the pile's tip, and hence
on the driving resistance. The case-damping coeffi- 8.2.5 The Energy Approach
cient was investigated through a back calculation (to The Energy Approach uses basic energy relations in
match the measured static capacity). Results, de- conjunction with dynamic measurements to deter-
scribed in the next section, suggest no correlation mine pile capacity; the concept was presented by
between the soil type and the case-damping coeffi- Paikowsky (1982) and was examined on a limited
cient. The common recommended practice suggests scale by Paikowsky and Chernauskas (1992). Exten-
the use of the method based on a specific site/area sive studies of the Energy Approach method were
calibration (GRL, 1999). This approach, in conjunc- carried out by Paikowsky et al. (1994), and Pai-
tion with the application of the method for maximum kowsky and LaBelle (1994). The underlying concept
resistance (RMX), has proven effective. Accumu- of this approach is the energy balance between the
lated experience on extensive jobs in the Boston area total energy delivered to the pile and the work done
(e.g. GTR 1997, 1998) has demonstrated the effec- by the pile/soil system. The basic Energy Approach
tiveness of the Case method, when calibrated. A sta- equation is:
tistical examination of local calibration was per-
formed in Florida by McVay et al. (2000). The E max (16)
results of this analysis suggested that for 48 cases, Ru =
(D − Set )
the ratio between the static pile capacity to the Case Set + max
method prediction at EOD was 1.344 ± 0.443 (mean 2
± 1 S.D.).
As no generic conditions exist for the use of the where Ru = maximum pile resistance, Emax = meas-
Case method, international or national calibrations ured maximum energy delivered to the pile, Dmax =
are unrealistic. As the projection of local calibration measured maximum pile top displacement, and Set =
(of good experience and practice) beyond the geo- permanent displacement of the pile at the end of the
analyzed blow, or 1/measured blow count. For fur- more details the results related to the dynamic analy-
ther details regarding the Energy Approach method ses of this study.
see Paikowsky et al. (1994) and Paikowsky (1995). The evaluation of static capacity from pile driving
is based on the concept that the driving operation in-
8.2.6 The signal matching techniques duces failure in the pile-soil system, (i.e. a very fast
The signal matching technique is often referred to as load test is carried out under each blow). Dynamic
post driving analysis or office method. With the analyses encounter three fundamental difficulties: (i)
availability of faster, portable computers, it became correct formulation of the penetration process (e.g.
reasonably simple to conduct the analysis in the field soil motion, soil plugging etc.), (ii) evaluation of the
as well, although the field method analyses cannot static resistance out of the total resistance need to
be carried out for each blow during driving. The re- overcome during penetration, and (iii) time depend-
sponse of the modeled pile-soil system (e.g. force at ent pile capacity, (Paikowsky, 1995). The parame-
the pile top) under a given boundary condition (e.g. ters controlling the accuracy of the dynamic predic-
measured velocity at the pile top) is compared to the tions therefore reflect the ability of each method to
measured response (force measured). The modeled address the above difficulties.
pile-soil system or, more accurately, the modeled Based on the concept of a pile loading to failure
soil that brings about the best match (visual graphi- under each blow, it has been traditionally assumed
cal match) between the calculated and measured re- that during high driving resistance (i.e. refusal) there
sponses, is assumed to represent the actual soil resis- is not sufficient pile penetration to mobilize the full
tance. The static component of that resistance is pile capacity (Chellis, 1961). As such, the dynamic
assumed to be the pile’s capacity, and reflects that methods are deficient under high driving resistance,
time of driving. The procedure of signal matching categorized as equal or above 12BPI (Blows Per
was first suggested by Goble et al. (1970), utilizing Inch) or approximately 5BPcm (Blows Per cm),
the computer program CAPWAP. Others developed (MHD, 1988).
similar analyses, (e.g. Paikowsky 1982; Paikowsky The soil type is also believed to constitute a major
and Whitman 1990) utilizing the computer code factor as soil damping parameters are commonly
TEPWAP. The TNO program was developed by employed to represent viscous resistance in the
Middendrop and van Weel (1986), which led to im- modeling of the soil’s dynamic behavior. This vis-
provements and the CAPWAPC program, which is cosity is assumed to be soil type dependent, and as-
used to date. sociated with intrinsic soil properties. High viscos-
ity values are expected for cohesive soils and low
viscosity values are expected, therefore, for cohe-
8.3 The controlling parameters
sionless soils. Naturally, under a given velocity, high
viscous values are associated with higher dynamic
8.3.1 Overview.
resistance, and logically should prove more difficult
Preliminary examination of the parameters control-
to accurately define the static resistance.
ling the performance of the dynamic analyses is car-
The effect of time is well recognized but poorly
ried out prior to a final detailed evaluation of these
quantified. Piles undergo a decrease or increase of
methods, leading to resistance factors. Such exami-
capacity with time, also known as relaxation and set-
nation influences the sub categorization of the dy-
up, respectively. While the resistance during driving
namic methods (according to the important control-
and its static component represent the conditions en-
ling parameters), hence, directing the user to utilize
countered during penetration, the major interest re-
the appropiate resistance factor according to the
mains the long-term ability of the pile to carry load
relevant conditions of the employed method. For ex-
during its service life. The examination of the dy-
ample, if soil type is a controlling factor and the ac-
namic method predictions with static load tests (of-
curacy of the signal matching method is largely af-
ten carried out long after the driving) therefore re-
fected by soil type, evaluation of the method for
mains valid. The predictions can be assessed in
different soil types will result in the development of
relation to the time in which the driving and/or the
resistance factors, depending on the soil type. Con-
dynamic data have been obtained (e.g. EOD or
versely, if soil type does not control the accuracy of
BOR).
the specific dynamic method, categorization based
The sections below provide a short summary of
on soil type is neither desired nor perused.
the process in which the importance of each of the
The following sections outline the logic used for
above three assumed controlling parameters was ex-
the preliminary examination of the controlling pa-
amined. The results are used to evaluate additional
rameters, the analyses and the results. The rationale
possible controlling factors, laying down the frame-
presented in this section follow previous studies by
work for the detailed evaluation of the dynamic
Paikowsky et al., (1994), Paikowsky (1995), Pai-
methods and the resulting resistance factors. More
kowsky et al. (1995), Paikowsky and Chernauskas
details are provided by Paikowsky and Stenerson
(1996). Paikowsky and Stenerson (2000) present in
(2000).
8.3.2 The effect of soil type case of multiple restrikes, only the last restrike is
The effect of soil type was examined in two ways: considered for the analysis).
(i) the correlation between the parameters assumed The presented results suggest that the time of
to be soil type dependent and soil type, i.e. damping driving significantly affect the performance of the
parameters and (ii) the accuracy of the predictive CAPWAP prediction, regardless of soil type. The
methods relative to the soil type. mean values for the BOR sets are closer to one,
Relationships were developed: (i) between soil while the mean values for the EOD are closer to two.
type and Smith damping parameters used in ap- The coefficients of variation (the ratio of the stan-
proximately 370 CAPWAP analyses, from dard deviation to the mean) show values of 0.33 and
PD/LT2000, for the tip and side pile resistances, re- 0.39 for BOR, while the EOD ratios are 0.55 and
spectively, and (ii) back-calculated Case damping 0.85. This suggests the existence of a substantial
factors for 290 cases, from the PD/LT database (Pai- scatter.
kowsky et al. 1994), required to obtain a match be- Further evaluation of the records is carried out on
tween the predicted capacity and the measured static the basis of driving resistance. The division between
capacity. All three presentations clearly indicate that cases for which the driving resistance is smaller or
no unique relationship exist between soil type and greater than 5BPcm, examines the aforementioned
damping parameters, suggesting that other mecha- notion of refusal and the expected accuracy of the
nism control the value required for a damping factor dynamic methods. The results shown in Table 10 in-
rather than the soil type. dicate that the cases for which the driving resistance
A summary of the statistics obtained when exam- is smaller than 5BPcm result in less accurate analy-
ining the accuracy of the signal matching technique ses with larger scatter, compared to the cases for
(specifically CAPWAP) based on soil type is pre- which the driving resistance was above 5BPcm.
sented in Table 10. The statistics shown are the Though driving resistance seems to be an important
mean and standard deviation of a normal distribution factor, clear understanding of its influence on the ac-
function for the ratio of the pile’s static capacity curacy of the dynamic methods calls for additional
(based on Davisson’s failure criterion) to the pile ca- investigation, which is briefly presented in a follow-
pacity obtained in the CAPWAP analysis. The major ing section.
categorization based on three soil types at the pile’s In summary, while the performance of CAPWAP
tip show no significant differences between clay and is not well correlated to soil type, other factors asso-
till vs. sand and silt. Although the cases for piles ciated with soil type may be important (e.g. low
found on rock provide different values, the numbers driving resistance in soft cohesive soils or gain of
are based on a small subset of 15 pile-cases, com- capacity with time) but soil type itself does not ap
pared to 100 and 265 pile-cases for the other catego- pear to be important. The data of Table 10 suggests
ries. that time of driving must be considered and driving
Table 10 provides further examination of time of resistance need to be further examined.
driving and driving resistances as subsets of the soil
type categorization. Two sets are examined based on 8.3.3 The effect of time
the time of driving: EOD and last BOR, (i.e. in the Penetration of piles in fine-grained soils causes

Table 10. CAPWAP analysis based on soil type categorization.

Clay & Till Sand & Silt Rock
Mean 1.352 1.517 0.930
Standard Deviation 0.723 1.085 0.172
Number of Cases 100 265 15
Time of Driving EOD BOR(last) EOD BOR(last) EOD BOR(last)
Mean 1.634 1.133 2.068 1.193 0.968 0.925
Standard Deviation 0.899 0.444 1.765 0.391 0.132 0.203
Number of Cases 45 40 77 116 7 7
Blow Count (BPcm) <5 ≥5 <5 ≥5 <5 ≥5 <5 ≥5 <5 ≥5 <5 ≥5
Mean 1.127 1.725 0.750 1.315 2.191 1.458 1.126 1.283 1.070 0.952 0.671 0.879
Standard Deviation 0.637 0.807 0.241 1.160 1.901 0.512 0.386 0.355 ----- 0.136 0.163 0.230
Number of Cases 35 35 11 10 64 13 74 40 1 6 3 3
Notes: 1. EOD = End of Driving
2. BOR(last) = Beginning of the last restrike
3. BPcm = Blows per centimeter
compression and disturbance, resulting in soil resis- serves as a mechanism to absorb energy, but, as it
tance during driving that differs from the long-term does not reflect the actual phenomenon, its correla-
pile capacity. Although factors such as thixotropy tion to physical properties (e.g. soil type) or time of
and aging contribute to this phenomenon, the most driving cannot be achieved. If the motion of the dis-
significant cause for gain of capacity with time is as- placed soil is a major factor contributing to the en-
sociated with the migration of pore water. Meas- ergy loss during driving, a substantial portion of the
urements carried out on a model (Paikowsky and dynamic resistance should be a function of two pa-
Hart, 1998) and full-scale piles (Paikowsky and Ha- rameters: (i) mass/volume of the displaced soil that
jduk, 1999, 2000) show that pore pressure at magni- is a function of the pile geometry, namely, small vs.
tudes similar to the total soil pressures create in large displacement piles, and (ii) acceleration of the
clays around the pile’s shaft zones of about zero ef- displaced soil, (especially at the tip) that can be con-
fective stresses, resulting in almost a complete loss veniently examined as a function of the driving re-
of frictional resistance. Paikowsky et al. (1995, sistance. A brief summary of the findings described
1996) examined the static and dynamic gain of ca- by Paikowsky and Stenerson (2000) regarding the
pacity with time based on radial consolidation; a above two factors follows.
normalization process was followed, allowing for a Soil Acceleration/Driving Resistance
comparison between different pile sizes. Table 11 To evaluate the blow count that identifies the
presents a summary of parameters describing the transition between 'easy' driving with high soil ac-
pile capacity gain with time based on static and dy- celeration values and 'hard' driving with low values,
namic testing. The presented data shows that while the ratio between the static capacity and the CAP-
the rate of capacity gain (normalized to the maxi- WAP prediction (KSW) is presented in Figure 6
mum capacity vs. time on log scale) is similar based against the measured blow count values. The data
on both analyses (Cgt = 0.389, Cgtd = 0.348) the as- were separated into intervals of 8 BP10cm (2BPI)
sociated time for achieving 75% of the maximum with the mean and standard deviation of each group
capacity (normalized for all piles to 254mm (1ft) di- graphed as a point and an error bar against the mid
ameter) is about 20 times greater. In other words, point blow count of the interval. For example, for
dynamic testing and analyses (namely CAPWAP) driving resistance between 0 and 8BP10cm there
while following the physical behavior of capacity were 42 cases with a mean of 2.506 and a standard
gain, exhibit this gain much faster than the actual deviation of 2.217 plotted at the center of the inter-
gain monitored by the static load test results. The val, i.e. at 4BP10cm. The data presented in Figure 6
ramifications of these conclusions are: (i) actual gain shows that for the first two intervals (up to
of capacity is much slower than that exhibited by the 16BP10cm) the predicted capacity was substantially
dynamic methods, (ii) scheduling of construction or lower than for all other intervals with a significantly
testing based on capacity gain should consider the higher scatter. After approximately 16 blows per
reason for time evaluation (i.e. actual loading in con- 10cm, the mean and standard deviation of the indi-
struction or dynamic testing as part of quality con- vidual intervals fall within the range of all cases.
trol), and (iii) at present, the dynamic methods The boundary of the dynamic method evaluation
evaluation should concentrate on the long term pile based on driving resistance was defined, therefore,
capacity. as 16BP10cm (4BPI).
Table 11. Summary of static and dynamic based capacity gain Blow Count (blows/inch)
data sets (Paikowsky et al. 1996). 0 2 4 6 8 10 12 14 16 18 20
KSW = CAPWAP or TEPWAP Predictions

Static Data 4
Dynamic Data 4.72
Total no. of cases = 382
Sets LTT and ALL DATA
Set PD/LTT
PUT/LTT
Load Test Results

no. of cases in
CGT t75* Cgtd t75** CGT t75** 3 8 blows/10cm interval
mean for all cases = 1.452
NO. 15 5 7 6 22 11 42
2
Avg. 0.389 385.0 0.348 21.3 0.376 186.6
66 25
64 38 32 15 16 8 16 60
Stdev. 0.119 226.3 0.068 7.9 0.106 237.9 1
* closed-ended piles only
standard deviation for all cases = 0.985
0
8.3.4 The effect of soil motion 0 8 16 24 32 40 48 56 64 72 80 88
Overview Blow Count (blows/10cm)
Paikowsky and Chernauskas (1996) had shown
Figure 6. The ratio of static capacity to dynamic signal match-
that the stationary soil assumption, under which the ing prediction, KSW versus blow count for all pile-cases in
soil/pile interaction models were developed, does PD/LT2000.
not reflect the physical phenomenon that occurs dur-
ing pile driving. The use of pseudo viscous damping Displaced Soil/Pile Area Ratio
 The volume of the displaced soil is identical to 4

KSW = CAPWAP or TEPWAP Predictions

the volume of the penetrating pile, except when pile no. of cases in Total no. of cases = 382
intervals of 175
plugging takes place (Paikowsky and Whitman, (a)

Load Test Results

3
1990). The piles, therefore, can be classified as mean for all cases = 1.452
small (e.g., H and unplugged open pipe) and large
(e.g., closed pipe and concrete) displacement piles. 2
Additional classification of open-pipe piles can take 139
place according to a tip-area ratio similar to that 1
111
18 12 11
3
76
7
used for soil samplers (Paikowsky et al., 1989). 5

As most soil displacement takes place at the tip standard deviation for all cases = 0.985
0
area, the classification of piles can be better served
by looking at the ratio between the pile’s embedded 0 175 350 525 700 875 1050 1225 1400 1575
Area Ratio, AR
surface area and the area of the pile tip (Paikowsky
4

KSW = CAPWAP or TEPWAP Predictions

et al. 1994): no. of cases Total no. of cases = 71
A Surface area in contact with soil (17) in intervals
AR = skin = (b)

Load Test Results

Atip Area of pile tip 3

According to this ratio, a pile that is traditionally re- mean for all cases = 1.460
ferred to as a “large displacement” pile can behave 2
like a small displacement pile if it is driven deep 16 21
10
enough. A quantitative boundary of AR = 350 be- 1 6 8
tween “small” and “large” displacement piles was 10

proposed by Paikowsky et al. (1994). standard deviation for all cases = 0.734
Figure 7a presents the relationship between the 0
area ratio and the ratio of the static capacity over 0 175350 525 1050 2100 3150
Area Ratio, AR
CAPWAP prediction (KSW) for 382 cases. The data
were also separated into area ratio intervals of 175, Figure 7. KSW versus area ratio, (a) for all pile cases in
with the mean and standard deviation of each group PD/LT2000 and (b) for 71 pile-cases with driving resistance
graphed as a point and an error bar against the mid exceeding 16 BP10cm at the EOD.
point area ratio of the interval. For example, for the
139 piles with area ratio between 175 to 350, the drilled shaft database and are described by Kuo et al.
mean was 1.656 and the standard deviation was (2001).
1.425 plotted at the center of the interval (i.e. at the (1) FHWA Method, i.e., Reese, L. C. and M. W.
area ratio of 262.5). Figure 7a suggests that piles O’Neill (1988) ”Drilled Shaft: Construction Proce-
with area ratios smaller than 350 present less accu- dures and Design Methods”, FHWA-HI-88-042 for
rate predictions and larger scatters compared to the Sands and Clays.
mean and the scatter of all cases. Above an area ratio For Sands, β method was used.
of 350, the mean and standard deviation of the indi- For Clays, α method was used. For Su, the SPT
vidual intervals fall within the range of all cases. As correlation given by Terzaghi and Peck (1967)
the driving resistance may also affect the data in was used.
Figure 7a, the influence of the area ratio was further (2) R &W Method, i.e., Reese, L. C. and S. J.
examined for piles with a driving resistance greater Wright (1977) “Construction Procedures and Design
than 16 BP10cm at the EOD. The 71 cases answer- for Axial Loading”, Drilled Shaft Manual HDV-22,
ing to this criterion are presented in Figure 7b. The for Sand.
data in Figure 7b suggests again that when excluding In Sand-Clay Mix Deposit, α method was used
the easy driving resistance effects, the accuracy of for Clays.
the dynamic predictions are still lower with a large (3) C & K Method, i.e., Carter, J. P. and F. H. Kul-
scatter for piles with area ratios smaller than 350. hawy (1988) “Analysis and Design of Foundations
The boundary of AR = 350 between “small” and Socketed into Rock,” Report Number EL-5918 for
“large” displacement piles was therefore confirmed Rock.
based on database PD/LT2000. (4) IGM Method (Intermediate Geomaterials), i.e.,
O’Neill, et al (1996) “Load Transfer for Drilled
Shafts in Intermediate Geomaterials” FHWA-RD-
9 DRILLED SHAFTS – STATIC ANALYSIS 95-172, and O’Neill, M. W. and L. C. Reese (1999)
METHODS ‘Drilled Shaft: Construction Procedures and Design
Methods”, FHWA-IF-99-025.
The following analysis method and correlations have The design assumed smooth rock socket for skin
been used for the static capacity evaluation of the friction and closed joints for end bearing.
10 LEVEL OF TARGET RELIABILITY Target reliability levels vary from one industry ap-
plication to another due to various factors including
implied reliability levels in currently used design
10.1 Target reliability and probability of failure practices, failure consequences, public and media
The utilization of the LRFD method requires the se- sensitivity, types of users and/or owners, design life
lection of a set of target reliability levels which de- of a structure, and other political, economic, and so-
termines the probability of failure and hence the cietal factors. For a general view see, for example,
magnitude of the load and resistance factors. The Whitman, 1984, and Becker, 1996. Two approaches
probability of failure represents the probability for for generating target reliability levels were used in
the condition in which the resistance multiplied by general: (i) calibrated reliability levels that are im-
the resistance factors will be less than the load mul- plied in currently used codes, and (ii) cost benefit
tiplied by the load factors. When fitting LRFD to analysis.
ASD, the issue is less significant as practically the The first approach was commonly used to de-
factors are established to conform (often conserva- velop reliability-based codified design such as the
tively) to existing factors of safety. When calibrating LRFD format. The target reliability levels according
for a database, however, the establishment of an ac- to this approach are based on calibrated values of
ceptable probability of failure is cardinal, including implied levels in a currently used design practice.
the question of new design versus existing state of The argument behind this approach is that a code
practice. An approximate relationship between prob- represents a documentation of an accepted practice,
ability of failure and target reliability for a log- and as such, it can be used as a launching point for
normal distribution was presented by Rosenbleuth code revision and calibration. Any adjustments in
and Estava (1972): the implied levels should be for the purpose of creat-
pf = 460(-4.3β) (18) ing consistency in reliability among the resulting de-
signs according to the reliability-based code. Using
and is commonly in use; e.g. Withiam (1998). the same argument, it can be concluded that target
Baecher (2001) shows, however, that this approxi- reliability levels used in one industry might not be
mation is not so accurate below β of about 2.5, and fully applicable to another industry.
Table 12 prepared by Professor Gregory B. Baecher, The second approach is based on cost-benefit
of the University of Maryland, provides the com- analysis. This approach was used effectively in deal-
parison between the “exact” numbers to the ap- ing with designs for which failures result in only
proximation and suggests significant errors, espe- economic losses and consequences. Since structural
cially in the zone of interest (β = 2 to 3) for failures might result in human injury or loss, this
foundation design. method might be very difficult to use because of its
need for assigning a monetary value to human life.
Table 12. Comparison between Rosenbleuth and Estava’s
approximation and series expansion (labeled “exact”), Although this method is logical on an economic ba-
(Baecher, 2001). sis, its main shortcoming is its need to measure the
Rosenbleuth value of human life and hence not commonly ap-
β “exact” pf Percent Error plied to structural/foundation applications.
and Estavas’ pf
2.0 8.4689E-2 2.2750E-2 272.3%
10.2.2 Calibration
2.5 9.8649E-3 6.2097E-3 58.9% A number of efforts for the purpose of calibrating a
3.0 1.1491E-3 1.3500E-3 -14.9% new generation structural design code resulted in the
development of target reliability levels (i.e., safety
3.5 1.3385E-4 2.3267E-4 -42.5% indices or β values).
4.0 1.5592E-5 3.1686E-5 -50.8% The general methodology for code calibration
4.5 1.8162E-6 3.4008E-6 -46.6%
based on specific reliability theories, using second-
moment reliability concepts, is outlined by Melchers
5.0 2.1156E-7 2.8711E-7 -26.3% (1987), and others. Melchers also notes that
5.5 2.4643E-8 1.9036E-8 29.5% frequently the information is insufficient for this
determination and one must make a "semi-intuitive"
6.0 2.8705E-9 9.9012E-10 189.9%
judgment in selecting βt values.
While the specific reliabilities will be a function
10.2 Concepts for establishing target reliabilities of the strength criteria needed for specific materials
and load combinations within designated structures,
10.2.1 General method of approach it is useful to have an indication of the range of pos-
The selection of target reliability levels is a difficult sible target reliability levels.
task as these values are not readily available and
need to be generated or selected (Payer et al. 1994).
10.3 Target reliability for structures joints to be 2.5. He reported that this value is on the
Ellingwood et al. (1980) present ranges for reliabil- low end, because of the reference wave values.
ity levels for metal structures, reinforced and pre- Madsen et al. (1986) discuss target reliability
stressed concrete structures, heavy timber structures, levels that were used by the National Building Code
and masonry structures, as well as discussions of is- of Canada (1977) for hot-rolled steel structures. The
sues that should be considered when making the target reliability values were selected as follows: βt
calibrations. Table 13 provides typical values for = 4.00 for yielding in tension and flexure, βt = 4.75
target reliability levels based on values provided by for compression and buckling failure, and βt = 4.25
Ellingwood et al. (1980). The target reliability lev- for shear failures. These values are larger than the
els shown in Table 14 were also used by Ellingwood values in Tables 13 and 14 because they reflect dif-
and Galambos (1982) to demonstrate the develop- ferent environmental loading conditions and possi-
ment of partial safety factors. bly different design life. Also, the Canadian Stan-
dard Association presented the following target
failure probabilities for developing design criteria
Table 13. Target Reliability Levels (based on Ellingwood et
al., 1980). for offshore installation in Canadian waters (Man-
Target Reliability sour et al. 1994): 10-5 per year for failures that re-
Structural Type sult in great loss of life or a high potential for envi-
Level (βt)
Metal structures for buildings ronmental damage; and 10-3 per year for failures
3
(dead, live, and snow loads) that result in small risk to life or a low potential for
Metal structures for buildings environmental damage.
2.5
(dead, live, and wind loads) Madsen et al. (1986) also discuss target reliability
Metal structures for buildings
1.75
levels that were used by the Nordic Building Code
(dead, live, snow, and earthquake loads) Committee (1978). The target reliability values
Metal connections for buildings
4 to 4.5 were selected depending on the failure consequences
(dead, live, and snow loads) of a building in the following ranges: βt = 3.1 for
Reinforced concrete for buildings less serious failure consequences, βt = 5.2 for very
(dead, live, and snow loads) serious failure consequences, and βt = 4.265 for
- ductile failure 3
- brittle failure 3.5 common cases.
The βt values are for structural members designed for 50 years of ser-
vice.
10.4 Geotechnical perspective
The review provided in section 10.3 suggests that
Table 14. Target Reliability Levels used by Ellingwood and typical target reliability for members and structures
Galambos (1982).
Target Reliability
relevant to bridge construction varies between 1.75
Member, Limit State
Level (βt) to 3.0 with a target reliability of 2.5 to 2.7 for rele-
Structural Steel vant bridges.
Tension member, yield 3.0 Barker et al. (1991) have provided the following
Beams in flexure 2.5 regarding target reliability index for driven piles
Beams in shear 3.0 (p.A-51): “Meyerhof (1970) showed that the prob-
Column, intermediate slenderness 3.5
ability of failure of foundations should be between
Reinforced Concrete
Beam in flexure 3.0 10-3 and 10-4, which corresponds to values of β be-
Beam in shear 3.0 tween 3 and 3.6. The reliability index of offshore
Tied column, compressive failure 3.5 piles reported by Wu, et al. (1989) is between 2 and
Masonry, unreinforced 3. They calculated that the reliability index for pile
Wall in compression, uninspected 5.0 systems is somewhat higher and is approximately
Wall in compression, uninspected 7.5 4.0, corresponding to a lifetime probability of failure
The βt values are for structural members designed for 50 years of
service.
of 0.00005. Tang et al. (1990) reported that offshore
piles have a reliability index ranging from 1.4 to 3.0.
Moses and Verma (1987) suggested target reli- Reliability indices for driven piles are summa-
ability levels in calibrating bridge codes (i.e. rized in Table 5.4 (see Table 15). Values of β be-
AASHTO Specifications). Assuming that bridge tween 1.5 and 2.8 are generally obtained for the
spans of less than 100 ft are most common, a βt of lognormal procedure. Thus a target value of
2.5 to 2.7 is suggested for redundant bridges, and a β between 2.5 to 3 may be appropriate. However,
βt of 3.5 for non-redundant bridges. piles are usually used in groups. Failure of one pile
Wirsching (1984) estimated the safety index im- does not necessarily imply that the pile group will
plied by the API specifications API RP2A (1989) for fail. Because of this redundancy in pile groups, it is
fixed offshore structures in fatigue of tubular welded felt that the target reliability index for driven piles
can be reduced from 2.5 to 3.0 to a value between
2.0 and 2.5.”
Table 15. Reliability Indices for Driven Piles (Barker et probability of failure is pf = 0.1%, correspond-
al., 1991) ing to a reliability index of β = 3.00.
Dead to Live Reliability Index, β
Load Ratio Lognormal Advanced
1.00 1.6 – 2.8 1.6 – 3.0 11 ANALYSIS RESULTS AND RESISTANCE
3.69 1.7 – 3.1 1.8 – 3.3 FACTORS

Zhang et al. (2001) used a first order reliability

11.1 Initial resistance factors calculations
method to evaluate the reliability of axially loaded
pile groups, designed using the traditional concept of The factors were evaluated using dead to live load
group efficiency. Group effects and system effects ratios ranging from 1 to 4. The obtained results for a
were identified to be the major causes that led to a bias of one and a coefficient of variation of 0.4 are
significantly greater observed reliability of pile presented in Figure 8, suggesting very little sensitiv-
foundations than the calculated reliability of single ity of the resistance factors to the DL to LL ratio. A
piles. Group effect relates to the combined action of similar trend was observed when using dead to live
any number of piles vs. a single pile. A system ef- load ratio of 10. The large dead to live load ratios
fect is the contribution of the superstructure stiffness represent a wide possibility of bridge construction,
to the load distribution and resistance. typically associated with very long bridge spans.
The calculated probability of failure of pile The relatively small influence of the dead to live
groups was found to be one to four orders of magni- load ratio on the calculated resistance factors suggest
tude smaller than that of single piles, depending on that: (i) the use of a ratio (DL/LL) of 2 or 2.5 as a
the significance of system effects (changing the sys- typical value is reasonable, and (ii) the obtained fac-
tem bias factor λs from 1 to 2). Based on their study tors are by and large applicable for large span
Zhang et al. (2001) state that the target reliability in- bridges.
dex, βT, for achieving a specified reliability level 0.8
should be different for an isolated single pile (βTS),
General Case
an isolated pile group (βTG), and a pile system (βTS). Bias = 1 COV = 0.4
They give the following recommendations based on
Resistance Factor, φ

their research: 0.6

(i) A βTG value of 3.0 requires a βTS of 2.0 to 2.8 β = 2.0
if no system effects are considered. Pf= 2.28%
(ii) A βTG value of 3.0 requires a βTS of 1.7 to 2.5 β = 2.5
pf= 0.62%
if a system effect factor of 1.5 is considered. 0.4
β = 3.0
Pf= 0.14%
10.5 Recommended target reliability 0 1 2 3 4 5
DL/LL - Dead to Live Load Ratio
10.5.1 General Range
Based on the presented review and data it seems rea- Figure 8. Calculated resistance factors for a general case show-
sonable to establish the target reliability between 2.0 ing the influence of the dead to live load ratio.
to 2.5 for pile groups, and as high as 3.0 for single
piles. The evaluation of the resistance factors in the 11.2 Driven piles – static analysis
presented study was carried out by using reliability
indices of 2.0, 2.5, and 3.0 associated with pf = Table 16 presents a summary of the analyses used
2.28%, 0.62%, and 0.14% respectively. Such ap- for static capacity evaluation of driven piles, com-
proach provides a reasonable range before final tar- pared with the nominal resistance based on Davis-
get reliability values are set. son’s failure criterion. The table includes statistical
parameters and resistance factors for a range of reli-
10.5.2 Recommended concept and targets ability index values, and a ratio of dead load to live
Combining the review and common practice, the fol- load of 2.0. The data in Table 16 was already proc-
lowing reliability indices and probability of failure essed and the provided statistics follow the applica-
were developed and recommended in conjunction tion of judgment and relevant statistical evaluation
with methods for capacity evaluation of single piles. for the exclusion of the extreme (outlier) cases. The
(i) For redundant piles, defined as five or more data in table 16 is correct for the present stage of
piles per pile cap, the recommended probabil- progress in the research, and does not necessarily
ity of failure is pf = 1%, corresponding to a re- contain the final values obtained and recommended
liability index of β = 2.33. in this study. Reviewing the information presented
(ii) For non-redundant piles, defined as four or in table 16, one can conclude that sub-categorization
less piles per pile cap, the recommended based on pile and soil type resulted in many cases of
Table 16. The performance of the driven piles’ static analysis methods – statistical summary and resistance factors.

No. Resistance factors for a

Soil Pile Details of Method Stand.
Of Design Method Mean COV given reliability index β
Type Type Application Dev.
Cases
2.00 2.50 3.00
4 β-Method 11.5 B;T&P(5) 0.61 0.3721 0.61 0.23 0.18 0.13
12 λ-Method 2B; T&P(5) 0.79 0.237 0.30 0.54 0.46 0.39
H-Piles 14 α-Tomlinson 2B; Hara 5(h) 0.76 0.2052 0.27 0.55 0.47 0.40
13 α-API 2B; T&P(5) 0.92 0.2668 0.29 0.64 0.55 0.46
8 Schmertmann’s SPT-97 mob 1.04 0.4264 0.41 0.58 0.47 0.38
17 λ-Method 2B; Hara (6h) 0.79 0.1975 0.25 0.60 0.52 0.44
Clay Concrete 17 α-API 2B; Hara (5h) 0.81 0.2106 0.26 0.60 0.52 0.44
Piles 4 β-Method 2B; Hara (5h) 0.70 0.091 0.13 0.67 0.60 0.54
11 α-Tomlinson 2B; Hara (5h) 0.76 0.1824 0.24 0.59 0.51 0.44
9 α-Tomlinson 2B; T&P (1) 0.70 0.196 0.28 0.50 0.43 0.36
Pipe 14 α-API 2B; T&P (1) 0.76 0.2204 0.29 0.53 0.45 0.38
Piles 6 β-Method 2B; T&P (1) 0.51 0.1683 0.33 0.33 0.28 0.23
13 λ-Method 2B; T&P (1) 0.59 0.0708 0.12 0.57 0.52 0.47
13 Nordlund 36; 11.5B,P(6) 0.83 0.2241 0.27 0.60 0.52 0.44
16 Meyerhof 0.81 0.2592 0.32 0.54 0.45 0.38
H-Piles
12 β-Method 36; 2B; P(5) 0.66 0.1980 0.30 0.45 0.38 0.32
18 Schmertmann SPT-97 mob 1.35 0.5805 0.43 0.72 0.59 0.47
28 Nordlund Method 36: 11.5B; P(6) 0.94 0.3384 0.36 0.58 0.48 0.39
Concrete 28 β-Method 36; 2B; P(5) 1.05 0.3465 0.33 0.68 0.57 0.48
Sand
Piles 28 Meyerhof Method 0.65 0.2990 0.46 0.33 0.26 0.21
33 Schmertmann’s SPT97 mob 1.22 0.5124 0.42 0.67 0.54 0.44
16 Nordlund 36; 2B P(5) 1.35 0.5805 0.43 0.72 0.59 0.47
Pipe 11 β-Method 36; 2B P(5) 0.96 0.3360 0.35 0.60 0.50 0.41
Piles 15 Meyerhof 0.73 0.2920 0.40 0.41 0.34 0.28
19 Schmertmann’s SPT97 mob 1.58 0.8216 0.52 0.71 0.56 0.43
α-Tomlinson/Nordlund/
13 36; 2B; P(5) 0.62 0.1364 0.22 0.50 0.43 0.38
Thurman
H-Piles 26 α-API/Nordlund/Thurman 36; 2B; P(5) 0.72 0.1872 0.26 0.53 0.46 0.39
17 β-Method/Thurman 36; 2B; P(5) 0.56 0.1288 0.23 0.44 0.38 0.33
40 Schmertmann’s SPT-97 mob 1.23 0.5412 0.44 0.65 0.52 0.42
α- Tomlinson/Nordlund/
23 36; 2B; P; Hara(5h) 0.91 0.3458 0.38 0.54 0.44 0.36
Thurman
36; 11.5B; Sch;
55 α-API/Nordlund/Thurman 0.86 0.2924 0.34 0.55 0.46 0.38
T&P(8)
Mixed Concrete 36; 11.5B; Sch;
63 β-Method/Thurman 0.84 0.2436 0.29 0.59 0.50 0.42
Soils Piles T&P(8)
34 DRIVEN 1.66 1.3446 0.81 0.43 0.31 0.21
74 Schmertmann’s SPT97 mob 1.97 1.2017 0.61 0.75 0.57 0.43
32 Schmertmann’s CPT 0.88 0.2288 0.26 0.65 0.56 0.48
α- Tomlinson/Nordlund/
9 36; 2B; P(5) 0.84 0.2772 0.33 0.54 0.46 0.38
Thurman
Pipe Piles 28 α-API/Nordland/Thurman 36; 2B; P(5) 0.84 0.2856 0.34 0.53 0.45 0.37
16 β-Method/Thurman 36; 2B; P(5) 0.56 0.1120 0.20 0.47 0.41 0.36
29 Schmertmann’s SPT97 mob 0.83 0.2324 0.28 0.59 0.51 0.43
small subsets with limited ability to provide conclu- Histogram and frequency distributions were pre-
sive method evaluation and many of the methods pared for the identified critical cases in order to ex-
over-predict the actual pile capacity. amine the match between the actual data and the
probability distribution functions. Figures 10
through 14 present some of the data along with the
11.3 Driven piles – dynamic analysis calculated normal and lognormal distributions.
11.3.1 The analyzed cases 11.3.3 Intermediate conclusions
The time of driving, driving resistance, and area ra- The data presented in Table 17 and Figures 10
tio proved to be the major controlling parameters of through 14 lead to several preliminary conclusions:
the dynamic methods. To facilitate the codes’ sepa- (i) The signal matching procedure generally under-
ration between design and construction, the PD/LT predicts the pile’s capacity. The method performs
2000 database was organized into these categories, very well for the BOR (last restrike) cases. (ii) The
followed by subcategories of methods that use and simple Energy Approach provides excellent predic-
do not use dynamic measurements, with subsets fol- tion for evaluating the pile’s capacity during driving
lowing the controlling parameters. Figure 9 presents (EOD). (iii) The above suggests that construction de-
the analyzed subsets, the number of cases in the set lays due to restrike and costly signal matching
and the normal distribution mean and standard de- analyses need to be examined in light of capacity
viation. time dependency and economical factors.
WEAP is utilized in the design stage. The analy-
sis is carried out for driving stress evaluation, lead-
ing to a load factor, not included in this paper. The 11.4 Drilled shafts – static analysis
use of the method for the evaluation of pile capacity Table 18 presents a summary of the analyses used
was examined through WEAP results for default in- for static capacity evaluation of the drilled shafts,
put values and the blow count at the EOD compared compared with the nominal resistance based on the
to the static load test results. The presented data was FHWA failure criterion. The data in table 18 were
provided by GRL Inc. (Hannigan et al., 1996). limited to the cases bound within the range of two
For the construction category, the cases without standard deviations around the mean of the initial
dynamic measurements evaluate the dynamic equa- analysis results. The resistance factors for the dif-
tions, specifically the ENR, the Gates, and the ferent target reliability values were calculated for the
FHWA version of the Gates equation. The cases ratio of dead load to live load of 2.0. The data in ta-
with dynamic measurements evaluates both, CAP- ble 18 is correct for the present stage of progress in
WAP and the Energy Approach methods. The dy- the research, and does not necessarily contain the fi-
namic methods are broken down into subsets based nal values obtained and recommended in this study.
on time of driving, driving resistance, and area ra- Reviewing the information presented in table 18, one
tios. Judgment and statistics guidelines were used for can conclude that the obtained resistance factors are
the inclusion or exclusion of a few cases. For exam- within the range of current practice and the sub-
ple, extreme CAPWAP under-predictions (beyond 2 categorization provides details regarding both the
S.D.) were observed at the EOD in one site. All the method of design and construction. The design
cases on that site included easy driving and large methods in general provide more accurate predic-
area ratios and if included in the general population tions than those for driven piles, as indicated by the
of data, the EOD statistics would have become 1.861 bias(?).
± 1.483 (mean±1S.D.). This site is included only in
the subcategory of blow count < 16 BP10cm and AR
< 350.
12 PRELIMINARY RECOMMENDATIONS OF
RESISTANCE FACTORS
11.3.2 The critical cases
The statistical analyses presented in Figure 9 allows
for the identification of the critical cases that require 12.1 Summary tables
calibration and development into resistance factors.
Tables 19 through 21 were developed based on the
For example, the CAPWAP cases include (i) all
cases and resistance factors presented in section 11.
data, (ii) EOD, (iii) BOR, and (iv) the worst combi-
By and large, the cases were integrated according to
nation of soil motion effect (Blow count < 16
similarity of calculated factors and extent of data on
BP10cm and AR < 350).
which the factors were based. For example, the
Table 17 presents a summary of the major catego-
BOR and general Energy Approach cases in the dy-
ries of the dynamic methods that are identified from
namic analyses resulted in similar resistance factors,
Figure 9 as the cases that require calibration for a re-
and hence, were combined into one category. The
sistance factor.
factors are also divided based on pile redundancy as
described in section 10.5.2.
 Dynamic Analysis

Construction Design

WEAP
Dynamic No Dynamic
Measurements Measurements
Drivability
Resistance/ Capacity
Dynamic WEAP
Equations Pile Stress
GRL Analysis
EOD – Default
ENR Equation Gates Equation FHWA – Mod. Gates WEAP Analysis GTR
0.267 ± 0.243 1.787 ± 0.848 0.940 ± 0.472 1.656 ± 1.199 WEAP / Dynamic
No. = 384 No. = 384 No. = 384 No. = 99 Measurements

EOD BOR (last) Load Factor

1.073 ± 0.573 0.833 ± 0.403
No. = 135 No. = 159

< 16 BP10cm ≥ 16 BP10cm < 16 BP10cm ≥ 16 BP10cm

1.306 ± 0.643 0.929 ± 0.688 0.876 ± 0.419 0.809 ± 0.290
No. = 62 No. = 73 No. = 32 No. = 127

Signal Matching Field Evaluation

(CAPWAP) Energy Approach
1.368 ± 0.620 0.894 ± 0367
No. = 377 No. = 371

EOD BOR (last) EOD BOR (last)

1.626 ± 0.797 1.158 ± 0.393 1.084 ± 0.431 0.785 ± 0.290
No. = 125 No. = 162 No. = 128 No. = 153

< 16 BP10cm ≥ 16 BP10cm < 16 BP10cm ≥ 16 BP10cm < 16 BP10cm ≥ 16 BP10cm < 16 BP10cm ≥ 16 BP10cm
1.843 ± 0.831 1.460 ± 0.734 1.176 ± 0.530 1.153 ± 0.354 1.227 ± 0.474 0.972 ± 0.359 0.830 ± 0.352 0.775 ± 0.274
No. 54 No. = 71 No. = 32 No. = 130 No. = 56 No. = 72 No. = 29 No. = 124

AR < 350 AR ≥ 350 AR < 350 AR ≥ 350 AR < 350 AR ≥ 350 AR < 350 AR ≥ 350
2.589 1.929 1.116 1.308 1.431 1.422 0.764 0.954
± 2.385 ± 0.698 ± 0.362 ± 0.796 ± 0.727 ± 0.888 ± 0.318 ± 0.396
No. = 37 No. = 22 No. = 22 No. = 10 No. = 39 No. = 23 No. = 19 No. = 10

AR < 350 AR ≥ 350 AR < 350 AR ≥ 350 AR < 350 AR ≥ 350 AR < 350 AR ≥ 350
1.717 1.181 1.178 1.110 1.054 0.926 0.736 0.851
± 0.841 ± 0.468 ± 0.379 ± 0.303 ± 0.459 ± 0.320 ± 0.249 ± 0.305
No. = 37 No. = 34 No. = 83 No. = 47 No. = 39 No. = 34 No. = 82 No. = 42

*All values represent the ratio of the static capacity based on Davisson’s
failure criterion over the dynamic methods prediction, mean ± 1 S.D.

Figure 9. Flow chart presenting the sub-grouping of the dynamic analyses according to the controlling parameters and
the resulting statistical parameters for a normal distribution function.
 60 20 0.16
log-normal

Number of Pile-Cases
0.15

Relative Frequency
15 distribution 0.12
55 mx = 1.626
0.14 mlnx = 0.384
σlnx = 0.444 normal distribution
50 10 0.08
0.13
σx = 0.797
45 0.12 5 0.04

0.11
40 0 0
log-normal 0.1 0 0.5 1 1.5 2 2.5 >3
Number of Pile-Cases

Relative Frequency
distribution
35
0.09 Ratio of Static Load Test Results over the Pile Capacity
mlnx = 0.233 Prediction using the CAPWAP method
30 σlnx = 0.387 0.08

mx = 1.368 0.07
Figure 12. Histogram and Frequency Distributions for all EOD
25 (125) CAPWAP pile-cases in PD/LT2000.
normal distribution 0.06
20
0.05
σx = 0.620
25
15 0.04 log-normal
distribution
20 0.16

Number of Pile-Cases
0.03

Relative Frequency
10 mlnx = 0.011
σlnx = 0.366
0.02 15 mx = 1.084 0.12
5 normal distribution
0.01
10 0.08
σx = 0.431
0 0
0 0.5 1 1.5 2 2.5 >3 5 0.04
Ratio of Static Load Test Results over the Pile
Capacity Prediction using the CAPWAP method 0 0
0 0.5 1 1.5 2 2.5 3
Figure 10. Histogram and Frequency Distributions for all
Ratio of Static Load Test Results over the Pile Capacity
(377) CAPWAP pile-cases in PD/LT2000. Prediction using the Energy Approach method

Figure 13. Histogram and Frequency Distributions for all EOD

55 (128) Energy Approach pile-cases in PD/LT2000.
log-normal
distribution 0.14
50
mlnx = -0.187 0.13
σlnx = 0.379 45
45 0.12
mx = 0.894 40
40 0.11 0.24
0.1 35
35
Number of Pile-Cases

normal distribution 0.09 0.2

Relative Frequency

30
Number of Pile-Cases

30
Relative Frequency
0.08
log-normal 0.16
σx = 0.367 0.07 25 distribution
25
0.06 20
mlnx = 0.100 mx = 1.158
σlnx = 0.295 0.12
20
0.05 normal distribution
15
15 0.04 0.08
σx = 0.393
0.03 10
10
0.02 0.04
5
5
0.01
0 0
0 0
0 0.5 1 1.5 2 2.5 >3
0 0.5 1 1.5 2 2.5 3
Ratio of Static Load Test Results over the Pile Capacity
Ratio of Static Load Test Results over the Pile Capacity Prediction using the CAPWAP method
Prediction using the Energy Approach method
Figure 14. Histogram and Frequency Distributions for all
Figure 11. Histogram and Frequency Distributions for all BOR-last (162) CAPWAP pile-cases in PD/LT2000.
(371) Energy Approach pile-cases in PD/LT2000.
Table 17. The performance of the dynamic methods: statistical summary and resistance factors.
Resistance Factors for a given
No. of Mean Standard Reliability Index, β
Method Time of Driving COV
Cases KSX Deviation
2.0 2.5 3.0
General 377 1.368 0.620 0.453 0.68 0.54 0.43
EOD 125 1.626 0.797 0.490 0.75 0.59 0.46
Dynamic Measurements

CAPWAP EOD - AR < 350 &

37 2.589 2.385 0.921 0.52 0.35 0.23
Bl. Ct. < 16 BP10cm
BOR 162 1.158 0.393 0.339 0.73 0.61 0.51
General 371 0.894 0.367 0.411 0.48 0.39 0.32
EOD 128 1.084 0.431 0.398 0.60 0.49 0.40
Energy
Approach EOD - AR < 350 &
39 1.431 0.727 0.508 0.63 0.49 0.39
Bl. Ct. < 16 BP10cm
BOR 153 0.785 0.290 0.369 0.46 0.38 0.32
ENR General 384 1.602 1.458 0.910 0.33 0.22 0.15
Gates General 384 1.787 0.848 0.475 0.85 0.67 0.53
Equations
Dynamic

General 384 0.940 0.472 0.502 0.42 0.33 0.26

FHWA
modified EOD 135 1.073 0.573 0.534 0.45 0.35 0.27
Gates EOD
62 1.306 0.643 0.492 0.60 0.47 0.37
Bl. Ct. < 16BP10cm
WEAP EOD 99 1.656 1.199 0.724 0.48 0.34 0.25
Notes: EOD = End of Driving BOR = Beginning of Restrike
AR = Area Ratio Bl. Ct. = Blow Count
ENR = Engineering News Record Equation BP10cm = Blows per 10cm
KSX = Ratio of the Static Load Test Results COV = Coefficient of Variation
to the predicted capacity

Tables 19 through 21 are preliminary, and by no following sections deal with various aspects associ-
means represent the final outcome of the present ef- ated with the recommended factors and means for
fort. These tables represent, however, the most sig- their evaluation.
nificant attempt to date to develop LRFD code for
deep foundations based on databases. The evalua- 12.2.2 Working stress design
tion of the presented factors is provided in the fol- The traditional factors of safety presented in Table 1
lowing section. can now be evaluated in light of the available data.
For example, the coefficient of variation for the
ENR equation and the WEAP analyses are 0.910 and
12.2 Evaluation of the resistance factors 0.724, respectively, which practically means that the
methods are unsuitable for the purpose of capacity
12.2.1 Overview prediction. The reduction in the factor of safety
Evaluation of recommended factors is a complex from 3.50 to 2.75 in Table 1 when adding WEAP
and extensive process. The aim of the process is to analysis to static calculations is therefore unfounded.
compare an existing code of practice to the recom- The use of unspecified CAPWAP (general case) in
mended new factors. Very often this evaluation table 1 again does not justify the reduction of the
cannot be done directly, as either the principles be- factor of safety to 2.25 even though the average pre-
hind the factors are different (e.g. WSD vs. LRFD), diction is conservative and hence the mean case with
or the applied methodology is not compatible (e.g. a FS = 2.25 relates to an over prediction ratio of 3.1
the design and construction combined factors of the (1.368 × 2.25). In comparison, the use of FS = 2.25
existing code). As a result, the evaluation can be with a specified CAPWAP at the BOR is reasonable
carried out in two major ways: and is associated with an acceptable probability of
(i) by independently analyzing design case histories failure for a single pile application (approximately
using the new factors and existing codes. In this 1.85%, see Figure 14). The use of a large factor of
way comparisons can be made, and if large enough safety for the static analysis is seen to be reasonable,
case histories are analyzed, conclusions can be de- as most of the methods over-predict the actual
rived regarding the effectiveness and overall per- capacity. For example, the prediction of a concrete
formance of the new recommendations. pile in clay results in a mean under-prediction ratio
(ii) searching for common factors that can be com- of about 0.8 (λ or α methods). The actual factor of
pared, for example establishing a connection be- safety in this case would be 2.8 (0.8x3.5), which is
tween resistance factors and factors of safety. The actually lower than that used for the general CAP-
Table 18. Results of static analyses of drilled shafts and relevant resistance factors.
Resistance factors for a
Capacity Design Construction No. of given reliability index β
Soil Type Mean COV
Component Method Method Cases
2.0 2.5 3.0
Mixed 32 1.71 0.60 0.59 0.43 0.32
FHWA
Casing 12 2.27 0.46 1.02 0.80 0.62
Slurry 9 1.62 0.74 0.43 0.30 0.21
Sand
Mixed 32 1.22 0.67 0.37 0.26 0.19
R&W
Casing 12 1.45 0.50 0.60 0.46 0.36
Slurry 9 1.32 0.62 0.43 0.32 0.24
Mixed 53 0.90 0.47 0.40 0.31 0.24
Clay FHWA Casing 13 0.84 0.50 0.35 0.27 0.21
Dry 40 0.88 0.48 0.38 0.30 0.23
Mixed 44 1.19 0.30 0.72 0.60 0.50
Skin Friction Casing 21 1.04 0.29 0.64 0.54 0.45
FHWA
+ Dry 12 1.32 0.28 0.83 0.70 0.58
End Bearing Sand + Slurry 10 1.29 0.27 0.82 0.69 0.58
Clay Mixed 44 1.09 0.35 0.60 0.49 0.40
Casing 21 1.01 0.42 0.49 0.39 0.31
R&W
Dry 12 1.20 0.32 0.70 0.58 0.48
Slurry 10 1.16 0.25 0.77 0.65 0.55
FHWA Mixed 129 1.32 0.69 0.38 0.27 0.20
All Soils
R&W Mixed 76 1.14 0.52 0.46 0.35 0.27
Mixed 46 1.23 0.41 0.61 0.49 0.39
C&K
Dry 29 1.29 0.40 0.65 0.52 0.42
Rock
Mixed 46 1.30 0.34 0.73 0.60 0.49
IGM
Dry 29 1.35 0.31 0.80 0.67 0.56
FHWA Mixed 11 1.09 0.51 0.44 0.34 0.26
Sand
R&W Mixed 11 0.83 0.54 0.32 0.24 0.18
Clay FHWA Mixed 13 0.87 0.37 0.46 0.38 0.30
FHWA Mixed 14 1.25 0.29 0.77 0.64 0.54
Sand & Clay
Skin R&W Mixed 14 1.24 0.41 0.61 0.49 0.39
FHWA Mixed 39 1.08 0.41 0.53 0.43 0.34
All Soils
R&W Mixed 25 1.07 0.48 0.46 0.36 0.28
C&K Mixed 16 1.18 0.46 0.53 0.41 0.32
Rock
IGM Mixed 16 1.25 0.37 0.67 0.54 0.44

Table 19. Recommended resistance factors for the critical dynamic cases.
Resistance factor, φ φ/Mean KSX
Method Case Redundant Non-Redundant Redundant Non-Redundant
β = 2.33 β = 3.0 β = 2.33 β = 3.0
pf = 1.0% pf = 0.1% pf = 1.0% pf = 0.1%
General 0.60 0.45 0.431 0.314
Signal EOD, AR<350,
0.40 0.25 0.158 0.089
Dynamic Matching Bl. Ct.<16BP10cm
Measurements BOR 0.65 0.50 0.561 0.440
Energy General 0.40 0.30 0.470 0.358
Approach EOD 0.55 0.40 0.489 0.369
ENR General 0.25 0.15 0.162 0.094
Dynamic Gates General 0.75 0.55 0.409 0.297
Equations FHWA
General 0.40 0.25 0.383 0.277
modified
WEAP EOD 0.40 0.25 0.236 0.151
Notes: β = Reliability Index pf = Probability of Failure COV = Coefficient of Variation
EOD = End of Driving BOR = Beginning of Restrike Bl. Ct. = Blow Count
AR = Area Ratio ENR = Engineering News Record Equation
BP10cm = Blows per 10cm Non-Redundant = Less than five piles under one pile cap.
Redundant = Five piles or more under one pile cap.
Table 20a. Reesistance factors for drilled shafts – total resis- Table 21. Resistance factors for driven piles – static analyses
tance Resistance Factor
Resistance Factors Redun-

Pile Type
Soil Design Non-
Soil Design Construction non- dant
redundant Type Method redun-
Type Method Method redundant
βT = 2.33 dant
βT = 3.0 β=
β = 3.0
R&W, 2.33
Slurry α - API & Tomlinson
FHWA 0.35 0.25 Clay
R&W Mixed β, λ
Sand 0.55 0.45
R&W, Nordlund, β
Casing Sand
FHWA 0.55 0.40 Schmertman SPT97 mob

Concrete Piles
FHWA Mixed α - Tomlinson/Nordlund/
Thurman
Clay FHWA All 0.35 0.25 α- 0.50 0.40
Slurry & Dry 0.85 0.70 Mixed
API/Nordlund/Thurman
FHWA Casing 0.65 0.50 Soils
β - Thurman
Sand + Mixed 0.75 0.60 Schmertman CPT
Clay Slurry & Dry 0.75 0.60 0.60 0.45
Schmertman SPT mob
R&W Casing 0.50 0.35 Meyerhof/Sand or DRIVEN in
Mixed 0.60 0.45 0.30 0.20
Mixed Soils
All R&W Mixed 0.45 0.30 Clay α - API & Tomlinson, λ
Soils FHWA Mixed 0.35 0.25 α - Tomlinson/Nordlund/
C&K Mixed & Dry 0.60 0.45 Thurman
Rock
IGM Mixed & Dry 0.75 0.60 Mixed α - 0.50 0.40
Pipe Piles
Soils API/Nordlund/Thurman
Table 20b. Resistance factors for drilled shafts – skin resistance β - Thurman
Resistance Factors φ Schmertman SPT 97 mob
Soil Design Construction Non- Nordlund, β
Redundant Sand 0.60 0.45
Type Method Method redundant Schmertman SPT97 mob
β = 2.33 Meyerhof/Sand or β in Clay 0.30 0.20
β = 3.0
FHWA, α - API & Tomlinson, λ
All Soils Mixed 0.45 0.35 Clay
R&W Schmertman SPT 97 mob
C&K Mixed 0.50 0.35 Sand Nordlund, Meyerhof
Rock α - Tomlinson/Nordlund/ 0.50 0.40
IGM Mixed 0.65 0.50
H Piles

Mixed Thurman
WAP case. The major conclusion, therefore, is that Soils α-
API/Nordlund/Thurman
the value of a factor of safety by itself does not rep- β in Sand or Mixed Soils 0.45 0.40
resent the economics of the method (or the progres- Schmertman SPT97 mob in Sand or
siveness of the code), and databases are a necessary 0.60 0.45
Mixed Soils
tool to assess any design methodology.
12.2.4 Existing specifications
12.2.3 FOSM vs. FORM The existing resistance factors of the AASHTO
As the existing AASHTO specifications are based specifications for dynamic evaluation of driven
on FOSM (see section 5.1) one aspect that needs to piles, are limited and connected to static evaluation
be checked is the relationship between the factors methods. The factors presented in table 19 are novel
obtained by FOSM to those obtained by the current in their approach and categorization. Detailed com-
methodology (FORM). Figure 15 presents such re- parisons between the two are, therefore, not possible.
lationship for the different categories of the analyzed General comparison between the factors presented in
methods for all three databases and for a reliability table 19 to other sources (e.g. Australian code) sug-
index of β = 2.33. The data in Figure 15 suggests gest that the magnitude of the proposed resistance
that the use of FORM results in resistance factors factor is comparible.
consistently higher than those obtained by using The resistance factors presented in table 20 can
FORM. The ratio between the two suggests that as a be compared to the existing specification with the
rule of thumb, FOSM provides resistance factors ap- application of the λv factor. When doing that, the
proximately 10% higher than those obtained by proposed parameters are reasonably in agreement
FORM. Two practical conclusions can be obtained with the existing specifications. The resistance fac-
from the observed data; (i) First evaluation of data tors in Table 21 are somehow lower than those pro-
can be done by the simplified closed form FOSM vided by the existing specifications, but within the
approach and the obtained resistance factors are con- expected range. The categorization into construction
servative (on the lower side), and (ii) the obtained methods needs to be further evaluated in light of the
parameters in this study (i.e. tables 16 through 21) obtained results and current design methodology,
can be directly compared to the current specifica- which does not specify a pre-bidding construction
tions or other LRFD codes based on FOSM. method for drilled shafts. A more realistic compari-
 Figure 15. Comparison between resistance factors obtained using the First Order Second Moment (FOSM) vs. those
obtained by using First Order Reliability Method (FORM) for a target reliability of β = 2.33.

0.80
y = 1.1267x
FOSM =
0.70
No. of points = 160
Mean = 1.148
0.60 Std. Dev. = 0.039
Resistance factors using FORM

0.50 Driven Piles Static Analysis

Driven Piles Dynamic Analysis
Drilled Shafts
0.40
FOSM = FORM
all data points
0.30 Linear

0.20

0.10

0.00
- 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
Resistance factors using FOSM

QD
γD + γL
son between the presented to current parameters can QL (19)
FS =
be achieved via analyses of case histories using fac- Q 
φ D + 1
tors from both sources. Q
 L 
using DL/LL = 2, γL = 1.75, and γD = 1.25 for which
12.2.5 Evaluation of the design methods efficiency the resistance factors were calculated, results with:
The resistance factors alone do not provide a meas-
ure for the evaluation of the efficiency of the design F.S. = 1.4167/φ (20)
methods. Such efficiency can be evaluated through Different analysis cases were chosen to illustrate
the bias factor (mean of the ratio of the measured the approximated factors of safety associated with
over predicted), and its coefficient of variation or the the calculated resistance factors. The factors are
ratio of the resistance factor to the bias factor, i.e. presented in table 22 along with the mean over-
φ/mean bias, as proposed by McVay et al (2000). prediction ratio (calculated F.S. x the bias of the
This ratio is provided, for example, in table 19 for method), which in effect represents both the actual
the dynamic methods resistance factors. The effi- mean factor of safety and a measure of the economic
ciency values in Table 19 suggest that overall the efficiency of the method – the lower the value, the
higher efficiency is obtained by the signal matching smaller number of deep foundations are required,
analyses for the last restrike, followed by the Energy and the higher economic value of the method. Ex-
Approach at the end of driving (0.581 vs. 0.507). amining these factors, one clearly sees that the
This measure needs to accompany prescribed resis- method that provided the highest φ/λ ratio also pro-
tance factors in order to avoid a misconception of vided the lowest F.S. x value, and so on. The fac-
the existence of a correlation between the economy tors of safety presented in table 22 for β = 3.0 (lower
of a method and a high resistance factors,(when values) are in line with what one would expect
compared to others). Similarly, such misconceptions (ranging from 2.61 to 5.63, avg. 3.73). The use of
exist between the economy of a method and the lower factors of safety for redundant piles (β = 2.33)
lower level of factor of safety, where a mean factor provided factors of safety ranging from 2.13 to 4.0
of safety (F.S. x bias) represents the economic value (avg. 2.81), judged to be adequate as well.
of the method (the lower the better).
12.2.7 Actual probability of failure
12.2.6 Equivalent factors of safety One advantage of using a large database is that the
The fact that the resistance factors using FORM are probability of failure (or the risk) can be directly
similar to those obtained by FOSM, allows to use calculated from the available data, rather than using
simplified relationship between resistance factors the calculated distribution function. The procedure
and factor of safety based on Barker et al., 1991: is done by applying a certain resistance factor to the
Table 22. Resistance factors and associated factors of safety along with efficiency measures for sample methods.
β = 2.33
γL = 1.75, γD = 1.2, DL/LL = 2
β = 3.00
Pile Type or Soil Type or Method of
Category φ FS FS x λ
Construction State Analysis φ/λ
resistance factor of actual
efficiency
factor safety mean FS
0.54 0.67 2.62 2.13
PPC Clay α - API
0.44 0.54 3.22 2.61
Static Methods 0.60 0.57 2.36 2.48
PPC Sand β
Driven Piles 0.48 0.46 2.95 3.10
α - API 0.47 0.56 3.01 2.53
Pipe Mixed
Nordlund/Thurman 0.37 0.44 3.83 3.22
0.65 0.56 2.18 2.52
All BOR CAPWAP
0.50 0.43 2.83 3.28
Dynamic
0.55 0.51 2.58 2.80
Methods All EOD Energy Approach
0.40 0.37 3.54 3.84
Driven Piles
0.38 0.35 3.73 4.00
All EOD FHWA mod Gates
0.27 0.25 5.25 5.63
0.44 0.39 3.22 3.67
Mixed All R&W skin
0.31 0.27 4.57 5.21
Static Methods 0.60 0.49 2.36 2.90
Mixed Rock C&K total
Drilled Shafts 0.45 0.37 3.15 3.87
0.78 0.62 1.82 2.27
Mixed Sand & Clay FHWA skin
0.63 0.50 2.25 2.81
calculated resistance (capacity), and examining the way it is perceived by others and discuss a possible
number of cases that exceed the actual capacity framework that can address the difficulties. Duncan
(nominal strength). An example of the process as (1992) reported that one conclusion of a long-term
applied to study on LRFD for geotechnical applications was
some of the dynamic methods is presented in table that resistance factors for pile design were thought
23. The data in table 23 suggests that the recom- best to be based on judgment and common sense. A
mended factors presented in table 19 would result in National Research Council report agreed with the
target reliabilities higher (lower pf) than those the above view, saying, “this … subjective approach re-
factors were calculated for using the distribution flects the general lack of robust data sources from
functions. which a more objective set of factors can be de-
rived” (NRC, 1995). Eurocode 7, Geotechnics of
Table 23. Calculated probability of failure based on direct the time takes a similar view, stating: “It might
utilization of database PD/LT 2000 for selected prediction sometimes be helpful to use statistical methods.
methods. However, it is emphasized that this will rarely lead
pf(%)
Resis-
CAPWAP FHWA
directly to (resistance factor) values since these de-
tance CAP- CAP
EOD
Energy
Mod pend on an assessment of the field situation….”
factor WAP WAP Approach (CEN-European Committee for Standardization,
AR > 350 Gates
φ General BOR EOD
BL > 16 General 1990). NCHRP 24-17 CFP – (Research Problem
0.5 0.27 0 2.70 1.56 10.42 Statement of the presented work) “… the resistance
0.4 0 0 0 0 3.13 factors do not account for the variability of the site
0.33 0 0 0 0 0.78 conditions and the number of load tests conducted”
# of
cases 377 162 37 128 384 and Task 7 as part of resistance factors determina-
used tion required to “…consider the variability of … and
the method of engineering property determination in
the calibration”.
13 DISCUSSION - KNOWLEDGE BASED
DESIGNS 13.2 Framework for LRFD design for deep
foundation
13.1 Statement of problem In order to adopt LRFD for geotechnical purposes
An attempt is made in the presented research to de- one requires addressing the aforementioned issues.
velop resistance factors based on databases. Some of Many of the affecting factors are in fact being con-
the difficulties associated with the process required sidered in the design (e.g. previous experience) but
early decisions to be made (see section 5.4), and the need an established framework to allow future pro-
complexity of the issues become obvious when de- gress. A proposed solution is based on establishing
veloping and evaluating the obtained information. It knowledge based factors for both the design and the
is important, therefore, to review the subject in the construction (independent) capacity evaluation
methods. These factors can be accounted for by a
modifying constant ξ to be multiplied by the resis- extensive measures are taken to control construction
tance factor. quality, and multiple pile testing results are available
(e.g. integrity and capacity of drilled shafts, static
ξφ R > γL (22) and dynamic tests of driven piles). The high quality
control also relates to the number of tested deep
Where: foundations as a ratio of the number of piles in-
ξ = ξ1ξ2ξ3ξ4 ≤ ξlimit stalled per a substructure. ξ4 to be used during the
ξ1 =factor adjusting for the variability of site design to account for previous experience accumu-
conditions lated either on a specific construction site or from a
ξ2 =factor adjusting for the quality of soil pa- specific construction technology. Low values are
rameter estimates used if no previous experience is known at the site
ξ3 = factor adjusting for construction quality con- and a new unfamiliar construction technology is
trol used. High values mean that previous deep founda-
ξ4 =factor adjusting for previous site or construc- tion testing results similar to the one designed (type
tion experience and installation) are available.
ξlimit = an upper limit on the factor that will be
determined from computing the components
of ξ and judgment. The limit should have 14 CONCLUSIONS
some real value larger than 1.0, such as
1.10. Based on the presented data and analyses the follow-
ing major conclusions are derived:
A short description of the knowledge based fac- • The compilation of large databases allows for the
tors follows. ξ1 applies for both the spatial variation quantitative assessment of the pile capacity
of soil properties and stratification across a site, and evaluation methods both during the design and
for the extent to which that variation has been cate- the construction phases. In addition, databases
gorized in the subsurface investigation program. The allow (combined with the application of mechan-
factor is relevant for the deep foundations capacity ics principles) determining controlling parame-
evaluation both during design and construction ters of the capacity evaluation methods that re-
stages. Low values mean that the site is more erratic quire calibration. Databases are essential
than normal, or that little exploration and testing has therefore for the examination of any design
been done. Average values reflect normally variable methodology and hence enable to test the factors
soil conditions with adequate soil investigation pro- used by the Working Stress Design (WSD)
gram. High values means that the site is more uni- methodology (i.e. validity and economic of the
form than normal, and that an extensive program of assigned factors of safety), and the development
boring and testing has been conducted. Extent of ex- of other methodologies such as the Load and Re-
ploration can be evaluated via number of borings per sistance Factor Design (LRFD).
substructure unit. ξ2 applies only to deep founda- • LRFD facilitates a design methodology, which is
tions capacity methods employing calculations based better suitable for geotechnical applications than
on soil parameters the factor accounts for the manner WSD. The ability to determine design factors
in which soil parameters are estimated from field while quantifying the significance of their out-
and laboratory test data, and the resulting exactness come is a powerful tool in engineering. The de-
of those estimates. Low values mean that the corre- velopment of LRFD resistance (as well as load)
lation between soil parameters and the measure- factors utilizing reliability-based calibration and
ments they are based upon is poor (e.g. when esti- databases is a major necessary step for objective
mating undrained shear strength of soft to medium quantification of the parameters. More so, it al-
clay from SPT values), and thus the confidence in lows for a meaningful utilization of the LRFD
parameter values is low. Average values reflect principles (in contrast to parameter fitting to
normally variable soil conditions with adequate soil WSD) and sets a base for future development
investigation program. High values mean that the that rationalizes design and leads to a more eco-
correlation is good (e.g., when parameters are esti- nomic construction.
mated based on laboratory test of undisturbed sam- • The build-up of databases related to design
ples or direct insitu testing like a field vane test), and methods is by itself not sufficient in the context
thus the confidence in parameter values is high. ξ3 of LRFD utilization for geotechnical applica-
applies for the extent of measures taken to control tions. Variability in the parameters used in the
construction quality and testing the integrity and ca- design methods, site condition, construction
pacity of the constructed deep foundations.  Low quality and previous experience are all important
values mean that few measures are to control con- contributing factors. A more complete design
struction quality, and no static, dynamic or integrity based on LRFD needs to consider subsurface
testing results are available. High values mean that variability, site-specific technology and previous
 experience, as well as amount and type of testing REFERENCES
during construction. A framework for such ap-
proach is presented, further development how- AASHTO, 1994. LRFD Bridge and Construction Specifica-
ever require the combination of additional data- tions. American Association of State Highway and Trans-
portation Officials, Washington, DC.
bases (e.g. for correlation between soil ASSHTO, 1996/2000. LRFD Bridge Design Specifications, 2nd
parameters and field tests) and subjective meas- Edition--2000 Interim Revisions. Washington, DC.
ures. AASHTO, 1997. Standard Specifications for Highway Bridges.
American Association of State Highway and Transportation
Officials, Washington, DC., 16th Edition (1996 with 1997
ACKNOWLEDGEMENTS interims).
AASHTO, 1999. LRFD Bridge and Construction Specifica-
The presented research was sponsored by the tions. American Association of State Highway and Trans-
American Association of State Highway and Trans- portation Officials, Washington, DC.
portation Officials (AASHTO), under project 24-17, ACI, 1995, Building Code Requirements for Reinforced Con-
in cooperation with the Federal Highway Admini- crete. American Concrete Institute, Detroit.
stration (FHWA). The panel of the research project AISC, 1994. Load and Resistance Factor Design. Manual of
Steel Construction, American Institute of Steel Construc-
and Mr. David Beal of the NCHRP are acknowl- tion, Chicago, IL.
edged for their stimulating demands. Mr.’s. Jerry API, 1989. Draft Recommended Practice for Planning, Design-
DiMaggio, Al DiMillio and Carl Ealy of the FHWA ing and Constructing Fixed Offshore Platforms – Load and
are acknowledged for their interest, concern and Resistance Factor Design. API RP2A-LRFD, American Pe-
support. Dr.’s Gregory Baecher and Bilal Ayyub troleum Institute, Dallas, TX.
from the University of Maryland contributed to sec- ASCE, 1993. Minimum Design Loads for Buildings and Other
Structures. ASCE 7-93 (formerly ANSI A58.1).
tions 3.1 through 3.4, section 5.2, and sections 10.1 AUSTROADS, 1992. AUSTROADS Bridge Design Code, Na-
through 10.3 and performed the calculations of the tional Office, AUSTROADS, Surry Hills, NSW, Australia.
presented resistance factors based on FORM. Dr.’s Ayyub, B.M. 1999. Cost-Benefit Analysis of Risk-Informed
Mike McVay and Bjorn Birgisson of the University Design of Structural Systems. American Society of Me-
of Florida, and Dr. Ching Kuo of Geostructures chanical Engineers, the Research Committee on Risk Tech-
nology, Washington, DC.
compiled the static analyses databases and carried Ayyub, B. & Assakkaf, I. 1999. LRFD Rules for Naval Surface
out the analyses related to the material presented in Ship Structures: Reliability-Based Load and Resistance
sections 4.1, 4.2, 7.1, 9, 11.2, and 11.4. Dr. Frank Factor Design Rules. Naval Surface Warfare Center,
Rausche of Goble, Rausche, Likins (GRL) and As- Carderock Division, U.S. Navy.
sociates provided the data pertaining to the evalua- Ayyub, B., Assakkaf, I., & Atua, K. 1998a. Development of
tion of GRLWEAP as the WEAP method for dy- LRFD Rules for Naval Surface Ship Structures: Reliability-
Based Load and Resistance Factor Design Rules, Part III –
namic pile capacity evaluation. Mr. Kirk Stenerson Stiffened and Gross Panels. Naval Surface Warfare Center,
researched the performance of the dynamic analyses Carderock Division, U.S. Navy.
as part of his studies at the University of Massachu- Ayyub, B., Assakkaf, I. & Atua, K. 2000. Reliability-Based
setts Lowell. The help of Ms. Mary Canniff in the Load and Resistance Factor Design (LRFD) of Hull Girders
preparation of the manuscript is highly appreciated. for Surface Ships. Naval Engineers Journal, ASNE, May
2000.
Ayyub, B., Assakkaf, I., Atua, K., Engle, A., Hess, P.,
DISCLAIMOR Karaszewski, Z., Kihl, D., Melton, W., Sielski, R., Sieve,
M., Waldman, J., & White, G. 1998b. Reliability-based
The research presented here is part of an ongoing Design of Ship Structures: Current Practice and Emerging
project, which is aimed at rewriting the AASHTO Technologies. T&R Report R-53, Society of Naval Archi-
Deep Foundation Specifications. The presented pa- tects and Marine Engineers.
Becker, D. 1996. Eighteenth Canadian Geotechnical Collo-
rameters are preliminary and do not reflect in any- quium: Limit States Design for Foundations. Part I. An
way the final recommended or approved parameters. overview of the foundation design process, Canadian Geo-
The opinions and conclusions expressed or implied technical Journal, Vol. 33, pp. 956-983.
in the paper are those of the authors and are not nec- Barker, R., Duncan, J., Rojiani, K., Ooi, P., Tan, C., & Kim. S.
essarily those of the Transportation Research Board, 1991. Manuals for the Design of Bridge Foundations,
NCHRP Report 343. Transportation Research Board, Na-
the National Research Council, the Federal Highway tional Research Council, Washington, DC.
Administration, the American Association of State Barker, R., Duncan, J., Rojiani, K., Ooi, P., Tan, C., and Kim,
Highway and Transportation Officials, or the indi- S. 1991. Manuals for the Design of Bridge Foundations,
vidual states participating in the National Coopera- NCHRP Report 343. Transportation Research Board, Na-
tive Highway Research Program. tional Research Council, Washington, D.C.
Bermingham, P., and White, J. 1995. Pyrotechnics and the Ac-
curate of Prediction of Statnamic Peak Loading and Fuel
Charge Size. First International Statnamic Seminar, Van-
couver, BC, Canada.
Bowles, J. 1996. Foundation Analysis and Design. McGraw-
Hill, New York.
Butler, H.D., & Hoy, H.E. 1977. Users Manual for the Texas Hajduk, E., Paikowsky, S., Holscher, P., & Barends, F. 2000.
Quick-Load Method for Foundation Load Testing. Federal Accelerations of a Driven Pile and the Surrounding Soil.
Highway Administration, Office of Development, Report Proceedings, 6th International Conference on the Applica-
No. FHWA-IP-77-8, Washington, DC. tions of Stress-Wave Theory to Piles, September 11-13,
Canadian Geotechnical Society. 1992. Canadian Foundation 2000. Sãn Paulo City, Brazil.
Engineering Manual, 3rd Edition. Bi-Tech publishers, Ltd., Hamilton, J. & Murff, J. 1992. Selection of LRFD resistance
Richmond, British Columbia, Canada. factors for pile foundation design. Proceedings Structures
CIRIA 63. 1977. Rationalization of Safety and Serviceability Congress ’92. American Society of Civil Engineers, San
Factors in Structural Codes. Construction Industry Research Antonio, ASCE, NY.
and Information Association, SWIP 3AU, Report 63, Lon- Hannigan, P., Goble, G., Thendean, G., Likins, G., and
don. Rausche, F. 1995. Design and Construction of Driven Pile
Chellis, R.D. 1961. Pile Foundations, McGraw Hill 2nd ed. NY. Foundation. Federal Highway Administration, Washington,
Davisson, M. 1972. High Capacity Piles. Proceedings, Soil D.C.
Mechanics Lecture Series on Innovations in Foundation Hannigan, R., Goble, G., Thendean, G., Likins, G., & Rausche,
Construction. American Society of Civil Engineers, Illinois F. 1996. Design and Construction of Driven Piles Founda-
Section, Chicago, 81-112. tions – Volume II (Working Draft). U.S. DOT Federal
DeBeer, E. 1970. Proefondervindellijke bijdrage tot de studie Highway Administration. Washington, DC.
van het grandsdraagvermogen van zand onder funderinger Hansen, B. 1953. Earth Pressure Calculation. Danish Technical
op staal. English version, Geotechnique, Vol. 20, No. 4, Press. Copenhagen.
387-411. Hansen, B. 1956. Limit Design and Safety Factors in Soil Me-
DiMillio, A. 1999. A Quarter Century of Geotechnical Re- chanics. Bulletin No. 1, Danish Geotechnical Institute, Co-
search, Federal Highway Administration (FHWA), publica- penhagen.
tion No. FHWA-RD-98-139, 160 pp. Hansen, B. 1966. Code of Practice for Foundation Engineering.
Ellingwood, B., Galambos, T., MacGregor, J., & Cornell C. Bulletin No. 22, Danish Geotechnical Institute, Copenha-
1980. Development of a Probability-Based Load Criterion gen.
for American National A58. National Bureau of Standards Hölscher, P. 1995. Dynamical Response of Saturated and Dry
Publication 577, Washington, DC. Soils. Delft: Delft University Press.
Ellingwood, B., Galambos, T., MacGregor, J., & Cornell, C. Hölscher, P. and Barends, F. 1996. In-situ Measurement of
1982a. Probability Based Load Criteria – Assessment of Soil-Motion near the Toe of a Dynamically Loaded Pile. In
Current Design Practices. Journal of the Structural Divi- F.C. Townsend, M. Hussein & McVay, M. (ed), Proceed-
sion, ASCE, Vol. 108, No. ST5, 959-977. ings, 5th International Conference of the Application of
Ellingwood, B., Galambos, T., MacGregor, J., & Cornell, C. Stress-Wave Theory to Piles, September 11-13, 1996. Or-
1982b. Probability Based Load Criteria – Load Factors and lando, FL. 26-36.
Load Combinations. Journal of the Structural Division, Japanese Geotechnical Society. 1998. Foundation Design Stan-
ASCE, Vol. 108, ST5, 978-997. dards in the World – Toward Performed-Based Design –
Eurocode 7: “Geotechnical Design – Part I: General Rules”, Research Committee on present & future of Japanese Foun-
European committee for standardization, Central Secre- dation Design and Soil Investigation in view of Interna-
tariat, Brussels, 1997. tional Equivalency.
FHWA, 1988. FHWA Guide Specifications for Driven Piles. Kulhawy, F. et al. 1988-1994. Reliability-Based Foundation
Federal Highway Administration. Design of Transmission Line Structures. Report EL-5507.
Freudenthal, A. 1947. Safety of Structures. Transactions of the Electrical Power Research Institute, Palo Alto.
ASCE. Vol. 112, 125-180. Kulhawy, F. et al. 1996. Proceedings, Uncertainty in the Geo-
Galambos, T. & Ravindra, M. 1978. Properties of Steel for use logical Environment. Madison, ASCE, NY.
in LRFD. Journal of Structural Engineering. ASCE, Vol. Kulhawy, F. and Mayne, P. 1990. Manual on Estimating of
104, No. 9, 1459-1468. Soil Properties for Foundation Design. Electric Power Re-
Gates 1957. Empirical Formula for Predicting Pile Bearing Ca- search Institute. Palo Alto, California.
pacity. Civil Engineering, Vol. 27, No. 3, 65-66. Kuo, C., McVay, M., and Birgisson, M. 2002. Calibration of
Goble, G. 1999. Geotechnical Related Development and Im- LRFD Resistance Factors for Drilled Shaft Design, Trans-
plementation of Load and Resistance Factor Design portation Research Board (TRB), Annual Meeting, January,
(LRFD) Methods. NCHRP Synthesis 276, Transportation Washington, DC.
Research Board, National Research Council, National Lai, P. and Graham, K. 1995. Static Pile Bearing Analysis Pro-
Academy Press, Washington, DC. gram. SPT94. http://www.dot.state.fl.us/structures/manuals/
Goble, G., Likens, G., & Rausche, F. 1970. Dynamic Studies spt94.zip
on the Bearing Capacity of Piles – Phase III, Report No. 48. Long, J., Bozkurt, D. & Kerrigan, J. 1998. Value of Methods
Division of Solid Mechanics, Structures, and Mechanical for Predicting Axial Pile Capacity. Transportation Research
Design. Case Western Reserve University. Board Paper No. 99-1333, 76th Annual Meeting, January,
Goble, G. & Rausche, F. 1976. Wave Equation Analysis of Pile 1999. Washington, DC.
Driving-WEAP Program. Vol. 1-4, FHWA #IP-76-14.1 Madsen, H.O., S. Krenk, and N.C. Lind. 1986. Methods of
through #IP-76-14.4. Structural Safety, Prentice Hall, Englewood Cliffs, New
GRL. 1999. Pile-Driving Analyzer, PAK Users Manual. Goble, Jersey.
Rausche, Likins and Associates, Inc. Mansour, A.E., P.H. Wirsching, B.M. Ayyub, and G.J. White.
GTR. 1997. Dynamic Pile Testing Report, Central Ar- 1994. Probability Based Ship Design Implementation of
tery/Tunnel Project C07D2, I-90/Airport Interchange Arri- Design Guidelines for Ships, Ship Structures Committee
vals Tunnel – Phase I, East Boston, Massachusetts. Geo- Draft Report, U.S. Coast Guard, Washington, D.C.
sciences Testing and Research, Inc. North Chelmsford, Melchers, R. E. 1987. Structural Reliability Analysis and Pre-
MA. diction. Ellis Horwood Limited, UK.
GTR. 1998. Dynamic Pile Testing Report, Central Ar- Meyerhof, G. 1970. Safety Factors in Soil Mecanics. Canadian
tery/Tunnel Project C07D2, I-90/Airport Interchange Toll Geotechnical Journal, Vol. 7, No. 4, pp. 349-355.
Plaza, East Boston, Massachusetts. Geosciences Testing
and Research, Inc., North Chelmsford, MA.
Meyerhof, G. 1976/1981. The Bearing Capacity and Settlement Paikowsky, S. 1984. Use of Dynamic Measurements for Pile
of Foundations. Technical University of Nova Scotia. Hali- Analysis. Including PDAP-Pile-Driving Analysis Program,
fax, Canada. GZA Inc., Newton, Massachusetts.
McVay, M., Birgisson, B., Zhang, L., Perez., A & Putcha S. Paikowsky, S. 1995. Using Dynamic Measurements for the
2000. Load and Resistance Factor Design (LRFD) for Capacity Evaluation of Driven Piles. Civil Engineering
Driven Piles Using Dynamic Methods – A Florida Perspec- Practice, Journal of the Boston Society of Civil Engineers
tive. Geotechnical Testing Journal, ASTM, Vol. 23, No. 1, Section/ASCE. Vol. 10, No. 2, 61-76.
55-66. Paikowsky, S. & Chernauskas, L. 1992. Energy Approach for
McVay, M. and Townsend, F. 1989. Short course/ Workshop Capacity Evaluation of Driven Piles. 4th International Con-
Guide for PC Software Support for Design and Analysis of ference on the Application of Stress-Wave Theory to Piles.
Deep and Shallow Foundations from In-situ Tests (or PL- The Hague, Netherlands. 595-601.
AID manual). University of Florida. Gainesville, Florida. Paikowsky, S. & Chernauskas, L. 1996. Soil Inertia and the
MHD. 1988. Standard Specifications for Highways and Use of Pseudo Viscous Damping Parameters. 5th Interna-
Bridges. Massachusetts Highway Department. tional Conference on the Application of Stress-Wave The-
Middendorp, P. and Bielefeld, M. 1995. Statnamic Load Test- ory to Piles. Orlando, FL. 203-216.
ing and the Influence of Stress Wave Phenomena, Proceed- Paikowsky, S. & Hajduk, E. 1999. Design and Construction of
ings of the First International Statnamic Seminar, Vancou- an Instrumented Test Pile Cluster. Research Report submit-
ver, BC, Canada, pp. 207-220. ted to the Massachusetts Highway Department, September,
Middendorp, P. and van Weel, P. 1986. Application of Charac- 1999. Boston, Massachusetts.
teristic Stress-Wave Method in Offshore Practice. Proceed- Paikowsky, S. & Hajduk, E. 2000. Theoretical Evaluation and
ings of the 3rd International Conference on Numerical Full Scale Field Testing Examination of Pile Capacity Gain
Methods in Offshore Piling. with Time. Research Report to be submitted to the Massa-
Moses, F. 1985. Implementation of a Reliability-Based API chusetts Highway Department, December, 2000. Boston,
RP2A Format, Final Report. API PRAC 83-22. American Massachusetts.
Petroleum Institute. Paikowsky, S. & Hart, L. 2000. Development and Field Testing
Moses, F. 1986. Development of Preliminary Load and Resis- of Multiple Deployment Model Pile (MDMP). Research
tance Factor Design Document for Fixed Offshore Plat- Report to be submitted to the Federal Highway Administra-
forms, Final Report. API-PRAC 95-22, American Petro- tion, April, 2000. FHWA-RD-99-194 Washington, DC.
leum Institute. Paikowsky, S. & LaBelle, V. 1994. Examination of the Energy
Moses, F., and D. Verma, 1987, Load Capacity Evaluation of Approach for Capacity Evaluation of Driven Piles. US
Existing Bridges, NCHRP Report 301, Transportation Re- FHWA International Conference on Design and Construc-
search Board, Washington, D.C. tion of Deep Foundations, December 6-8, 1994. Orlando,
National Research Council. 1995. Probabilistic Methods in FL. Vol. II, 1133-1149.
Geotechnical Engineering. Washington. Paikowsky, S., LaBelle, V. & Hourani, N. 1996. Dynamic
National Research Council of Canada. 1977. National Building Analyses and Time Dependent Pile Capacity. 5th Interna-
Code of Canada. Ottawa. tional Conference on the Application of Stress-Wave The-
Nguygen, T., McVay, M., Birgisson, B., and Kuo, C. 2002. ory to Piles. Orlando, FL. 325-339.
Uncertainty in LRFD phi, φ, Factors for Driven Prestressed Paikowsky, S., LaBelle, V. & Mynampaty, R. 1995. Static and
Concrete Piles, Transportation Research Board (TRB), An- Dynamic Time Dependent Pile Behavior. Research Report
nual Meeting, January, Washington, DC. submitted to the Massachusetts Highway Department, No-
Nordic Building Code Committee. 1978. Recommendations for vember, 1995. Boston, Massachusetts.
Loading and Safety Regulations for Structural Design, Re- Paikowsky, S., Operstein, V., and Bachand, M. 1999. Express
port No. 36, Copenhagen, Denmark, 148 pp. Method of Pile Testing by Static Cyclic Loading. Research
Nowak, A. 1993. Calibration of LRFD Bridge Design Code. Report submitted to the Massachusetts Highway Depart-
Department of Civil and Environmental Engineering Report ment, October, 1999. Boston, Massachusetts.
UMCE 92-25. University of Michigan, NCHRP 12-33. Paikowsky, S., Regan, J., and McDonnell, J. 1994. A Simpli-
Nowak, A. 1999. Calibration of LRFD Bridge Design Code. fied Field Method for Capacity Evaluation of Driven Piles.
Department of Civil and Environmental Engineering Report FHWA Report No. FHWA-RD-94-042, September, 1994.
UMCE 92-25. University of Michigan, NCHRP 12-33. Paikowsky, S. & Stenerson, K. 2000. “The Performance of the
Olsen, R. & Flaate, K. 1967. Pile Driving Formulas for Friction Dynamic Methods, their Controlling Parameters and Deep
Piles in Sand. ASCE JSMFC, Vol. 93, SM6, Nov. 1967, Foundation Specifications” keynote lecture in Stress Wave
279-297. 2000, The Sixth International Conference on the Applica-
O’Neill, M., and Reese, L. 1999. Drilled Shafts: Construction tion of Stress-Wave Theory to Piles, September 12, 2000,
Procedures and Design Methods, Volume II Chapters 9-19, São Paulo, Brazil, Balkema, pp. 281-304.
Appendices. US DOT, Federal Highway Administration, Paikowsky, S. & Whitman, R. 1990. The Effect of Plugging on
Washington, DC. Pile Performance and Design. Canadian Geotechnical
O’Neill, M., Townsend, F., Hassan, K., Buller, A., and Chang, Journal, Vol. 27, No. 4, 429-440.
P. 1996. Load Transfer for Drilled Shafts in Intermediate Paikowsky, S., Whitman, R. & Baligh, M. 1989. A New Look
Geomaterials, Federal Highway Administration (FHWA),, at the Phenomenon of Offshore Pile Plugging. Marine Geo-
Publication No. FHWA-RD-95-172, 184 pp. technology, Vol. 8, No. 3, 213-230.
Ontario Ministry of Transportation and Communication. 1992. Payer, H., Huppmann, H., Jochum, C., Madsen, H., Nittinger,
Ontario Highway Bridge Design Code and Commentary, K., Shibata, H., Wild, W., and Wingender, H. 1994. Ple-
3rd ed. nary Panel Discussion on How Safe is Safe Enough? Struc-
Osterberg, J. 1992. The Osterberg Load Cell for Testing tural Safety and Reliability, Schueller, Shinozuka and Yao
Drilled Shafts and Driven Piles, report to the Federal High- (eds.), Balkema, Rotterdam, Netherlands, pp. 57-74.
way Administration, Washington, DC. Rausche, F. 2000. Personal communication. February 2000.
Paikowsky, S. 1982. Use of Dynamic Measurements to Predict Rausche, F., Goble, G. & Likens, G. 1975. Bearing Capacity of
Pile Capacity Under Local Conditions. M.Sc. Thesis, Dept. Piles from Dynamic Measurements, Final Report. Ohio De-
of Civil Engineering Technion-Israel Institute of Technol- partment of Transportation, Ohio DOT-05-75.
ogy.
Rausche, F., Thendean, F., Abou-matar, H., Likins, G., and
Goble, G. 1997. Determination of Pile Driveability and Ca-
pacity From Penetration Tests. Vol. 1-3, Final Report.
FHWA-RD-96-179, Federal Highway Administration.
Reese, L.C., Wang, S.C. and Arrellaga, J. 1998. APILE Plus
3.0 Manual. EnSoft, Inc. Austin, Texas.
Siu, w., Parimi, S. & Lind, N. 1975. Practical Approach to
Code Calibration. Journal of the Structural Division.
ASCE, Vol. 101, No. ST7, 1469-1480.
Smith, E. 1960. Pile Driving Analysis by the Wave Equation.
Journal of Soil Mechanics and Foundations, American So-
ciety of Civil Engineers, August 1960. 35-61.
Standards Association of Australia. 1995. Australiam Stan-
dards, Piling-Design and Installation. Homebush, NSW.
Tang, W. 1993. Recent development in geotechnical reliability.
Probabilistic methods in Geotechnical Engineering. Li, K.
& Lo, S-C. (Eds.), Balkema, Rotterdam. 3-28.
Tang, W., Woodford, D., and Pelletier, J. 1990. Performance
Reliability of Offshore Pile. 22nd Annual Offshore Technol-
ogy Conference, Paper No. OTC 6379, Houston, TX.
Terzaghi, K. 1942. Discussion of the Progress Report of the
Committee on the Bearing Value of Pile Foundations. Pro-
ceedings, ASCE. Vol. 68: 311-323.
Tomlinson, M.J. 1980/1995. Foundation Design and Construc-
tion, 6th Edition. Longman Scientific & Technical. Essex,
England.
US Army Corps of Engineers, No.7, EM 1110-1-1905. 1992.
Bearing Capacity of Soil, Technical Engineering and De-
sign Guides. Adapted by ASCE--2000. ASCE Press, New
York.
Wellington 1892. Discussion of “the Iron Wharf at Fort Mon-
roe, VA. By J.B. Cuncklee. Transactions, ASCE Vol. 27,
paper No. 543, Aug. 1892, 129-137.
Whitman, R. 1984. Evaluating Calculated Risk in Geotechnical
Engineering. The Seventeenth Terzaghi Lecture. Journal of
Geotechnical Engineering, ASCE, 110(2): pp. 145-188.
Wirsching, P. H. October 1984. "Fatigue Reliability for Off-
shore Structures." J. of Structural Engineering, American
Society of Civil Engineers, New York, Vol. 110, No. 10, pp
2340-56.
Withiam, J., Voytko, E., Barker, R., Duncan, M., Kelly, B.,
Musser, S., and Elias, V. 1998. Load and Resistance Factor
Design (LRFD) of Highway Bridge Substructures. FHWA
Publication no. HI-98-032, July, Washington, D.C.
Zhang, L., Tang, W. and Ng, C. 2001. Reliability of Axially
Loaded Driven Pile Groups, Journal of Geotechnical and
Geoenvironmental Engineering, vol. 127, no. 12, Decem-
ber.