Professional Documents
Culture Documents
Abstract: The Federal Highway Administration 共FHWA兲 mandated the use of the load and resistance factor design 共LRFD兲 approach in
the U.S. for all new bridges initiated after September 2007. This paper presents the bridge deep foundation practices established through
a nationwide survey of more than 30 DOTs in 2008. Highlighted by this study are the benefits of the LRFD as well as how the flexibility
of its usage is being exploited in design practice. The study collected information on current foundation practice, pile analysis and design,
pile drivability, pile design verification, and quality control. Since this is the first nationwide study conducted on the LRFD topic following
the FHWA mandate, the status on the implementation of LRFD for bridge foundation design was also examined. The study found that: 共1兲
more than 50% of the responded DOTs are using the LRFD for pile design, while 30% are still in transition to the LRFD; and 共2兲 about
30% of the DOTs, who use the LRFD for pile foundations, are using regionally calibrated resistance factors to reduce the foundation costs.
DOI: 10.1061/共ASCE兲BE.1943-5592.0000118
CE Database subject headings: Load and resistance factor design; Piles; Bridge foundations; Design; Deep foundations; Pile
foundations.
Author keywords: LRFD; Piles; Bridge foundations; Analysis; Design; Deep foundations.
The data for the study reported in this paper was collected
Foundation Practice
through a web-based survey developed in January 2008 and dis-
tributed to the DOT officials from different states as well as Fig. 1 presents a summary of results obtained for the common
FHWA engineers. A total of 33 fully completed responses were foundation practice in different states. Included in Fig. 1 are the
> 45 m SPT KS WV
NV UT In-house KY VA
DP, OEP, < 15 m No response
CO MO No response
dia. 1 to 1.5 m DS, CIDH, < 20 m
α/β/Nord No response DP, HP,
TN NC
Deep Deep No response 310x79;
Deep 360x132 mm
DS, CIDH, OK DP, PSCP, SC
DP, HP, SPT
dia. 1.2 to 3 m dia.40 to 60cm
360x132 mm No response No response
β/SPT α/Nord No response > 30 m
AZ AR 10 m DP, PSCP,
NM MS AL GA DP, HP, 30 to 60 cm
310x79;
TX LA 360x174 mm
No response PSCP, 30 to 40 cm
AK No response
> 25 m Nord/SPT
DS, CIDH,
30 m
Nord/SPT/ In-house dia. 1 to 1.5 m FL
DP, PSCP,
α/SPT
dia. 0.5 to 1.5 m
In-house
HI
Fig. 1. U.S. map summarizing the typical soil formations, average depth to bedrock, commonly used deep foundation categories, types and sizes,
and static analysis methods used in different states
most common soil formations; average depth to bedrock; the most shallow foundations in different soil types, while shallow founda-
commonly used category of deep foundations, pile types and tions are used primarily for low volume bridges in shallow bed-
sizes; and the static analysis methods used in pile design. Respon- rock or gravel. Among the deep foundation users, 76% of them
dents were allowed to identify up to three different soil forma- use driven piles, 18% use drilled shafts, and 6% use a combina-
tions for each state. Consequently, the soil formation shown for tion of both types. Figs. 2 and 3 highlight the percentage of usage
each state in Fig. 1 was based on the predominant soil noted in of different types of driven piles and drilled shafts, respectively.
the survey response. For a few respondents who opted not to Among the driven pile users, all respondents indicated that they
answer this question and the state DOTs who did not respond to use steel H-piles, while 80% of them use closed-end pipe piles,
the survey, the soil formation shown was based on geological 40% use open-end pipe piles, 32% use precast/prestressed con-
maps of Belcher and Flint 共1946兲. crete piles, 8% use precast concrete piles, 4% use timber piles,
For questions about the commonly used in situ and laboratory and about 12% of the respondents reported to be using other pile
tests used to define soil parameters, 94% of the respondents re-
ported to be using the standard penetration test 共SPT兲, 52% use
the cone penetration test 共CPT兲, 16% use the vane shear test, and 100% 100%
around 20% perform some other uncommon tests. Note that the 90%
respondents were allowed to identify an unlimited number of tests 80%
80%
in this category. Furthermore, all of the respondents reported to be
70%
performing basic laboratory soil tests such as Atterberg limits and
60%
soil classification, while 42% stated that they perform triaxial
tests, 35% reported to be performing the unconfined compression 50%
40%
test, 29% perform the direct shear test, and 16% perform some 40%
32%
other uncommon laboratory tests. Despite the subjective nature of 30%
the test, the survey confirmed that the majority of respondents 20%
12%
depend on SPT tests to determine the basic soil parameters. 10% 8%
4%
The next set of questions gathered information about the use 0%
of different foundation types associated with bridges, the most Steel Closed Open Pre- Precast Timber Other
commonly used categories of deep foundations, as well as details H-piles End Pipe End Pipe stressed Concrete Piles Types
about the commonly used types of driven piles and drilled shafts.
For the choice of foundation type, about 91% of the respondents Fig. 2. Distribution of the most commonly used driven pile types for
indicated that they more frequently use deep foundations than bridge foundations
0%
CIDH Soldier piles CFA Micropiles The questions for this section were aimed to understand the deep
foundation design and analysis process practiced by different state
Fig. 3. Distribution of the most commonly used drilled shaft types agencies. This section began with questions directed at determin-
for bridge foundations ing the pile resistance criterion in cohesive and cohesionless soils.
In cohesive soils, it was found that 88% of the respondents de-
pend on both skin friction and end bearing, 6% of the respondents
types 共e.g., monotube piles, tapered tube piles, and a combination depend only on skin friction, and 6% of them indicated that they
of prestressed concrete and steel H-piles兲. Among the drilled shaft ignore end bearing only when the average SPT N-value is less
users, it was found that 83% of them use cast-in-drilled-hole than 12 blows per 30.5 cm 共1 ft兲. In cohesionless soils, it was
shafts 共CIDH兲, 50% use soldier piles, 33% use continuous flight found that 87% of the respondents depend on combining the re-
auger, and 33% use micropiles. sistances obtained through skin friction and end bearing, 9% de-
Furthermore, the respondents were asked to identify the most pend only on skin friction, and 4% include the resistance from
common pile size共s兲 that they use. This information is included in end bearing only when the average SPT N-value is greater than 25
Fig. 1, which reveals that the steel H-piles and precast/prestressed blows per 30.5 cm 共1 ft兲.
concrete piles are more commonly used on the East Coast of the Furthermore, the most preferred method for designing deep
United States where the soil formation is mainly composed of foundations was found to be the ASD method, which was con-
coastal plain and glacial tills. On the West Coast, where the soil firmed by 45% of the respondents and the primary reasons for this
profile is mainly composed of alluvium soil, open-end pipe piles choice are their familiarity with the ASD method and possible
are more commonly used. Most of the states in the Midwest use increase in foundation cost when the AASHTO-LRFD specifica-
steel H-piles, where the main soil profile is composed of glacial tions are used due to the built-in conservatism. However, the
tills. In specific areas of the West, the CIDH shafts are also widely LRFD approach was reported to be the most commonly used
used which is presumably due to the seismic requirements and the method due to the imposed FHWA mandate. About 52% of survey
possibility of forming dependable plastic hinges in this founda- respondents are currently using LRFD, while 33% of them are in
tion shaft. a transition stage from ASD to LRFD, and about 15% of respon-
The bridge foundation practice reported by FHWA-EFLHD dents are still using ASD with a factor of safety ranging from 2 to
and the Alberta Infrastructure and Transportation was similar to 2.5.
MS AL GA
TX
AK
AK LA
FL
HI
11% 9%
10% 6% 0%
0% WEAP PDA CAPWAP Other
α β CPT λ Nordlund SPT In-house Other
Fig. 6. Most commonly used dynamic analysis methods for the de-
Fig. 5. Most commonly used static analysis methods for the design sign of deep foundations
of deep foundations
lund 1963兲. About 40% of the respondents use the SPT method
Fig. 4 shows the current extent of LRFD implementation in the 共Meyerhof 1976/1981兲 and 6% of them use in-house methods for
design of bridge foundations in the United States. Fig. 4 was piles in cohesionless soils. Most of the respondents chose the
created by combining the 31 responses collected from the state Nordlund’s method to be the most accurate static method for
DOTs with the results of the AASHTO-LRFD OC survey 共Moore sandy soils, and the ␣-method as the most accurate static method
2007兲 for the 19 DOTs who did not respond to the recent survey. for clayey soils. Note that the survey permitted multiple answers
Even though the FHWA mandate was implemented in October for this particular set of questions. Complete descriptions of the
2007, 15 state DOTs who responded to the survey and 11 of those different static analysis methods identified above may be found in
who only responded to the AASHTO-OC survey are believed to Hannigan et al. 共1998兲.
be either still using the ASD method or in a transition stage to the For dynamic analysis methods, all respondents reported to be
LRFD approach for designing bridge foundations. Among the using the wave equation analysis program 共WEAP兲 for determin-
DOTs who responded to be using the LRFD method for founda- ing pile capacity. The most commonly used software for this pur-
tion design, 46% are using regionally calibrated resistance factors pose is GRLWEAP 共Pile-driving analyzer; PAK user’s manual
that are based on SLT database and reliability theory, 23% are 1999兲, in which 41% of them define the soil parameters using the
using regionally calibrated factors by fitting to ASD factor of soil type based method, 28% use the SPT N-values based method,
safety, while 31% of them are using the geotechnical resistance 21% use the Driven software 共Hannigan et al. 1998兲, and 10%
factors as specified in the current AASHTO Specifications follow some other unspecified approaches. Fig. 6 represents the
共AASHTO 2007兲. Information on the assumed risk or probability use of other dynamic analysis methods, 74% of the respondents
of failure 共p f 兲 for the LRFD approach to pile foundations was also indicated they use the pile driving analyzer 共PDA兲 and the case
requested. Since they are currently not using LRFD or in transi- pile wave analysis program 共CAPWAP兲, while 16% of them use
tion to LRFD, 15 of the respondents did not answer this question. other uncommon methods. Information collected on the fre-
Another nine respondents reported that they are unaware of the quently used dynamic formulas in the design of deep foundations
assumed probability of failure. However, four respondents indi- is presented in Fig. 7. About 57% of respondents prefer the
cated that they rely on a p f value of less than 1/100 for the piles, FHWA-modified Gates formula, 21% of them use the Engineering
while three of them reported to be using a p f value in the range of News Records 共ENR兲 formula, 14% use the Gates formula, and
1/5,000 to 1/1,000. Although no specific question was asked re- more than 43% use in-house formulas. Complete descriptions of
lating the pile redundancy to p f , the reported variation in p f is
believed to be due to the assumed redundancy for the pile.
The next set of questions gathered information about the dif- 100%
ferent static analysis methods, dynamic analysis methods, and
90%
dynamic formulas that are used for determining pile capacity. It
was found that about 2/3 of the respondents employ a combina- 80%
tion of static and dynamic methods. Respondents who reported to 70%
be using this combination essentially deploy the static analysis 60% 57%
methods for determining the number of piles and dynamic analy- 50%
sis methods or dynamic formulas for finalizing the penetration 43%
40%
length.
As shown in Fig. 5, the most common static analysis method 30%
21%
used for piles in cohesive soils is the ␣-method at 42% 共Tomlin- 20% 14%
son 1957; API 1974兲. About 32% respondents reported to be 10%
using the -method 共Esrig and Kirby 1979兲, 11% use the CPT 0%
method 共Nottingham and Schmertmann 1975兲, and 9% follow the FHWA ENR Gates In-house
-method 共U.S. Army Corps of Engineers 1992兲. The same Fig- Modified Gates
ure also reveals that the most popular static analysis method for
piles in cohesionless soils is Nordlund’s method at 63% 共Nord- Fig. 7. Most commonly used dynamic formulas for deep foundations
NMa Steel H-piles -method Nordlund P+C+W ENR, G, F-G 0.35 0.45 N/A
NV Steel H-piles ␣-method Nordlund Not used Not used 0.35 0.25 N/A
PAa Steel H-piles -method Nordlund P+C+W Not used 0.5 0.5 0.5
UTa Steel H-piles ␣-method Nordlund Not used Not used 0.5 0.7 0.7
WA Steel H-piles In-house In-house WEAP F-G 0.5 0.5 0.5
a
Pile Load Test Database available.
b
P + C + W = PDA, CAPWAP, and WEAP.
c
F-G = FHWA Modified Gates method.
d
G = Gates method.
the different dynamic formulas identified above may be found in point of contrast between the FHWA-EFLHD bridge foundation
Hannigan et al. 共1998兲. More than 80% of the respondents chose design practice and that of the state DOTs is that FHWA-EFLHD
the dynamic analysis methods to be more accurate and reliable considers the static analysis methods to be more accurate than the
than static analysis methods and dynamic formulas in determining dynamic methods as long as the soil strength parameters are ad-
the capacity of deep foundations. equately obtained. On the other hand, the Alberta Infrastructure
In this part of the survey, the different extreme load types used and Transportation pile design practice was quite different from
in the design of bridge foundations as well as the different meth- that of the state DOTs in that they have already implemented the
ods used for estimating the lateral displacement demand of piles LRFD approach for the design of bridge foundations and depend
were also inquired. For the extreme load consideration, it was on the Canadian Highway Bridge Design Code for determining
found that 37% of the respondents account for scour load in their the appropriate LRFD resistance factors. They mainly use the
design, 25% include earthquake load, 20% account for loads due static analysis methods in the design of bridge deep foundations
to collision, and 18% use a combination of extreme loads. All and the appropriate static method for a given site is reported to be
respondents reported that lateral displacement is a design consid- selected by the consulting firm chosen for the project. They also
eration for piles. However, the method used for determining the indicated to be using the ENR dynamic formula for design veri-
pile displacement demand varied, as 72% of the respondents use fication.
the p-y curves, 14% of them use the Broms method 共Broms
1964兲, and 14% use other methods such as the FB-Pier strain
Regionally Calibrated LRFD Resistance Factors
wedge theory, point of fixity method, and empirical methods. In
the context of lateral demand on piles, it is worth noting that the As previously indicated, among those who have already imple-
expected pile displacement is significantly influenced by the abut- mented the LRFD approach to their design of bridge foundations,
ment type chosen for the bridge. For example, integral abutments 12 DOTs use regionally calibrated resistance factors that are more
designed without any expansion joints are becoming an increas- suitable for the local soil types. These resistance factors were
ingly popular choice among various DOTs. When bridges are calibrated based on local SLT data and using statistical ap-
designed with integral abutments, the superstructure loads in the proaches based on the reliability theory. Of those DOTs, who are
longitudinal and traverse direction may not only be distributed still in transition to the LRFD, six of them have adopted prelimi-
over more number of supports thus increasing the design effi- nary, regionally calibrated resistance factors, which were estab-
ciency of the foundations, but they also increase the pile redun- lished using their local design and construction experience as well
dancy 共Mistry 2005兲. as the recommended load factors to retain the factor of safety
In general, the FHWA-EFLHD pile design practice was found used for the ASD method. This information, together with details
to be similar to that of the state DOTs. The FHWA-EFLHD indi- about deep foundation practice and design methodology, is sum-
cated that they are still in a transition stage from ASD to LRFD marized in Table 1 for the respondents who indicated the use of
and that they are currently using the resistance factors from the regionally calibrated resistance factors. Of the respondents, nine
2007 AASHTO LRFD Specifications for the LRFD approach. For of them noted that they have a collection of SLT data for their
the design of bridge pile foundations, they indicated the use of state and they are identified in the table.
static analysis methods such as ␣-method, -method, and Nord- Note that the first-order second moment 共FOSM兲 and the first-
lund. As for the pile design verification, the FHWA-EFLHD re- order reliability method are two common reliability approaches
sponse noted that the use of the SLT less frequently and the that can be used for the LRFD calibration. According to Allen et
preferred approach being the dynamic analysis methods 共i.e., al. 共2005兲, FOSM is the most straightforward technique, in which
PDA and CAPWAP兲 and dynamic formulas 共i.e., Engineering the random variables are represented by their first two moments,
News Records or ENR, Gates, and FHWA-modified Gates兲. A i.e., the mean 共兲 and standard deviation 共兲, while the coeffi-
cient of variation is defined by / . Paikowsky et al. 共2004兲 Furthermore, it is noted that the in-house methods lead to higher
performed the LRFD calibration for the resistance factors using resistance factors than those determined for routinely used static
both methods and concluded that the difference between them is analysis methods.
around 10%, with FOSM providing slightly conservative resis- Table 4 compares the mean values of the regionally calibrated
tance factors. Moreover, the 2007 AASHTO-LRFD specifications resistance factors from Tables 2 and 3 with those reported in the
are based on the FOSM assuming lognormal distribution density National Corporative Highway Research Project 共NCHRP兲 report
functions for the loads and resistances. From the specific ques- number 507 by Paikowsky et al. 共2004兲 and the 2007 AASHTO-
tions related to the selected reliability approach used for the LRFD specifications. In most cases, it is observed that the
LRFD calibration, it was found that most tend to follow the AASHTO recommendations are more conservative than those
FOSM approach because it was used in the 2007 AASHTO- proposed by Paikowsky et al. 共2004兲. Furthermore, the mean of
LRFD specifications. the reported regionally calibrated resistance factors by different
All the reported regionally calibrated resistance factors for dif- DOTs in all cases are equal to or greater than the recommended
ferent soil and pile types were examined using statistical means, values in NCHRP 507 and the AASHTO guidelines. In some
in order to establish a representative mean and standard deviation cases, the regionally calibrated factors are twice as high as those
so that benefits of the reported resistance factors could be real- recommended for design practice by AASHTO. Therefore, it is
ized. The minimum sample size 共N兲 for these statistical analyzes
clear that the LRFD regional calibration could result in higher
was limited to three. For all data sets, the mean and standard
resistance factors as previously shown, thus optimizing the pile
deviation were determined as a function of the soil type, pile type,
design and reducing the foundation costs.
and static analysis method. For those data sets with N ⬍ 3, only
The benefits of using regionally calibrated resistance factors in
the average resistance factors were determined. All of these re-
pile design can be realized through a simple design example.
sults are summarized in Tables 2 and 3. Fig. 8 presents the histo-
Consider a bridge pier with a maximum factored axial load of
grams and frequency distributions as well as the confidence
intervals 共CIs兲 for the probability density functions 共PDFs兲 of the 8,900 kN 共2,000 kips兲 that is to be designed with a deep founda-
reported resistance factors for steel H-pile in different soil types. tion consisted of 18.3-m 共60-ft兲-long steel HP 254 mm 共depth兲
As shown in Fig. 8, the PDFs of different data sets follow a ⫻ 62.4 kg/ m 共self-weight兲 共10 in⫻ 42 lbs/ ft兲 connected with a
normal distribution, while the probability of the upper and lower concrete cap. The soil formation at the site is firm to very firm
limits of the true mean of each data set were found within 95% glacial clay with silt seams and boulders, with medium soil vari-
confidence intervals. As indicated by the standard deviations in ability. If the ␣-API 共API 1974; Coduto 2001兲 static analysis
Table 2, it appears that the mean of the reported regionally cali- method is used for the design of the pile foundation, the axial
brated resistance factors for a given soil type is somewhat consis- capacity of a single pile is 567 kN 共127 kips兲. Using the LRFD
tent, especially for the steel H-piles. Also, note in this table that resistance factor of 0.35 provided in the 2007 AASHTO specifi-
the mean resistance factor of the open-end pipe piles is greater cations, the design capacity of a single pile is 198.5 kN 共42 kips兲,
than those established for the steel H-piles and CIDH shafts. requiring a minimum of 45 piles. If the average regionally cali-
However, it is noted that the resistance factor of the pipe piles is brated resistance factor of 0.47 for the ␣-API method is used from
based only on two data points. For resistance factors reported for Table 3, the design capacity of each pile is 266.5 kN 共60 kips兲,
different static analysis methods in Table 3, smaller standard de- needing a total of only 34 piles. This 25% reduction in number of
viations are again observed, indicating less variations in the re- piles will significantly reduce the foundation cost as it reduces the
ported resistance factors for a given method and soil type. construction costs of both the piles and the pile cap.
Table 3. Mean Values and Standard Deviations of the Reported Regional Resistance Factors according to Different Static Analysis Methods and Soil
Types
Reported factors in sand Reported factors in clay Reported factors in mix soil
Static analysis method N Mean SD N Mean SD N Mean SD
Nordlund 11 0.5 0.12 N/A 4 0.53 0.17
SPT method 3 0.45 0.25 N/A 3 0.53 0.11
␣-method N/A 6 0.47 0.19 N/A
-method N/A 4 0.49 0.13 N/A
CPT-method N/A 3 0.45 0.17 N/A
In-house 3 0.62 0.11 4 0.63 0.10 3 0.62 0.11
Percent
50
20
Static of reported
Frequency
0 0
0.0 0.2 0.4 0.6 0.8 1.0 Clay ␣-method 0.45 0.35 0.47
Reported Resistance Factors in Sand Soils -method 0.2 0.25 0.49
In-house N/A N/A 0.63
0.0 0.2 0.4 0.6 0.8 1.0
1
Normal - 95% CI
5 Mean 0.4854 99 5
StDev 0.1502
95 also addressed in this section of the survey and it was found that
80 more than 50% of the respondents believe that the pile capacity is
Percent
4 N 12 4
50
unaffected by relaxation in cohesive soils. The responses of those
20
Frequency
3 5 3
who indicated that the pile capacity is affected by relaxation were
1
0.0 0.2 0.4 0.6 0.8 1.0
grouped based on the most predominant soil formation that they
2
R.F.
2 indicated for their region. Accordingly, 40% of respondents noted
that the change in pile capacity would not exceed 5% of the value
1 1 expected at end of driving 共EOD兲 in silty sands, 6% indicated that
the effect varies from 5 to 10% in sandy soils, and only 2%
Normal
0 0 assumed that the effect exceeds by 20% of the pile capacity in
0.0 0.2 0.4 0.6 0.8 1.0 silts. Similarly, as for the influence of soil setup on the pile ca-
Reported Resistance Factors in Clay Soils pacity, 34% of the respondents indicated that the pile capacity
increases above 20% of the value at EOD due to setup in clays
0.0 0.2 0.4 0.6 0.8 1.0
1 and silty clays, 25% indicated that the effect varies from 5 to 10%
Normal - 95% CI
5 Mean 0.5531 99 5 in glacial tills and clays, 25% indicated that the effect varies from
StDev 0.1299 95 11 to 20% in clayey silts, 6% indicated that the soil setup effect
N 8 80
4 4 does not exceed 5% on the pile capacity, and 10% of the respon-
Percent
50
dents assume that the pile capacity is unaffected by soil setup in
Frequency
20
3 5 3 cohesionless soils.
1
0.0 0.2 0.4 0.6 0.8 1.0
2 R.F. 2
Design Verification and Quality Control
1 1 This final section of the survey focused on design verification and
Normal
quality control issues. All but one respondent indicated that they
0 0 perform field tests on 5–10% of the installed piles to verify the
0.0 0.2 0.4 0.6 0.8 1.0
design capacity. Among several different techniques, about 45%
Reported Resistance Factors in Mixed Soils of respondents have used the SLT for design verification, while
others predominantly use the dynamic monitoring approach espe-
Fig. 8. Histograms, frequency, and 95% CI of the reported regional
cially for friction piles 共i.e., use of WEAP, PDA, and/or CAP-
LRFD resistance factors for steel H-pile in different soil types
WAP兲. Usually for small projects, the WEAP analysis is
Pile Drivability
In the design stage, the pile penetration length required in the 15%
field is estimated using the chosen static analysis method. How- 24% WEAP and CAPWAP
ever, the penetration length may be readjusted at the construction-
site using the pile driving data. It was found that 31% of the Dynamic Formulas
respondents depend on a WEAP analysis and field observations to Static Methods
readjust the pile penetration length, 20% employ dynamic formu- 4%
31% Drive until Refusal
las to make the adjustments, and 15% rely on the initial static
6%
analysis results and make no adjustments to pile length in the field Drive until Bedrock
共see Fig. 9兲. Interestingly, 6% of the respondents noted that they
Combination of Methods
drive the pile until refusal, while 4% prefer to drive the piles until 20%
the bedrock is reached. The remaining 24% of respondents indi-
cated that they use a combination of the aforementioned methods
depending on the site conditions and design requirements. Fig. 9. Methodologies used for readjusting the pile penetration
Effects of soil relaxation and setup on the pile capacity were length