RECALIBRATION OF THE

GRLWEAP LRFD RESISTANCE

FACTOR FOR OREGON DOT
Final Report

SPR 683

RECALIBRATION OF THE GRLWEAP LRFD RESISTANCE
FACTOR FOR OREGON DOT
Final Report


SPR 683



by

Trevor Smith, Ph.D., P.E., Andrew Banas, Max Gummer, and Jaesup Jin
Civil and Environmental Engineering
Portland State University
Portland, OR 97207


for

Oregon Department of Transportation
Research Section
200 Hawthorne Ave. SE, Suite B-240
Salem OR 97301-5192

and


Federal Highway Administration
400 Seventh Street, SW
Washington, DC 20590-0003



February 2011

Technical Report Documentation Page
1. Report No.
FHWA-OR-RD-11-08
2. Government Accession No.


3. Recipients Catalog No.


5. Report Date
February 2011
4. Title and Subtitle
Recalibration of the GRLWEAP LRFD Resistance Factor for Oregon DOT
6. Performing Organization Code

7. Author(s)
Trevor D. Smith, PhD. PE, Andrew Banas, Max Gummer, Jaesup Jin
8. Performing Organization Report No.

10. Work Unit No. (TRAIS)


9. Performing Organization Name and Address
Department of Civil and Environmental Engineering
Portland State University
PO Box 751
Portland, OR 97207

11. Contract or Grant No.
SPR 683
13. Type of Report and Period Covered

Final Report

12. Sponsoring Agency Name and Address
Oregon Department of Transportation
Research Section and Federal Highway Administration
200 Hawthorne Ave. SE, Suite B-240 400 Seventh Street, SW
Salem, OR 97301-5192 Washington, DC 20590-0003

14. Sponsoring Agency Code

15. Supplementary Notes

16. Abstract

The Bridge Section of the Oregon Department of Transportation (ODOT) is responsible for the design of all bridge structures and routinely
uses GRLWEAP for controlling pile driving stresses and establishing capacity from the bearing graph. The LRFD resistance factor, , for
GRLWEAP sets the factored amount of nominal capacity in the LRFD inequality. Foundation conditions throughout Oregons Willamette
Valley and the Portland metropolitan area are predominantly sand, silt, and clay. Steel pipe and H section foundation piles typically are of
sufficient length to be friction piles and exhibit set-up after the end of initial driving (EOID). The two objectives of this study were to build an
extensive database of driven pile case histories to include restrike conditions from the present available sources that reflect ODOTs diverse
soils and piles, and to establish factors for EOID and beginning of restrike (BOR) conditions using GRLWEAP to match ODOT practice.

A diverse group of existing databases, including the FHWA-built DFLTD and NCHRP 507 PDLT2000, were accessed and merged with new
cases from the literature to build a comprehensive database, called the Full PSU Master. Neither of these two national databases proved
always correct for the large amount of source input required, with the largest source of anomalies and missing data being the blow count,
especially at the BOR condition. Over 150 new cases were added to establish the Full PSU Master database containing 322 piles, with each
case placed into one of three input tiers for statistical profiling to assist in preserving quality for the calibration. The Full PSU Master
database then supplied 179 cases analyzed by FHWA static capacity software DRIVEN and by GRLWEAP for capacity prediction by the
bearing graph. These predictions generated bias mean and COV statistics for a range of ODOT selected scenarios. The 322 piles ranged up
to 40 inches in diameter, and up to 197 ft in embedment length. The 179 analyzed piles ranged up to 36 inches in diameter and 167 ft in
embedment length and had driving blow counts up to 100 BPI. This research showed similar trends for GRLWEAP capacity as that reported
in NCHRP 507 for CAPWAP capacity on the statistical effects from variables such as blow count ranges. Sub-grouping by blow count
revealed a clear decay in easy driving mean and COV parameters when blow counts were 2 BPI. Above 2 BPI, little difference was
found in these parameters, and no upper limit of statistical accuracy was identified. A clear difference in statistical sample characteristics
existed between piles supported in predominately cohesive soils to those in cohesionless soils, and also between pile types.

For the ODOT case of redundant piles in groups, a reliability index at 2.33 was used to establish resistance factors and / efficiency
measures. Statistics for an initial ten scenarios were generated, and the First Order Second Moment (FOSM) resistance factor at EOID and
BOR was reported, based on lognormal fits to the distribution. The final five ODOT selected scenarios to permit comparison to NCHRP
507 and to form a basis to design implementation measures underwent advanced Monte Carlo based probabilistic procedures using random
number generation and the lognormal tail fits to provide EOID and BOR factors. Recommended resistance factors from the visual tail fit
procedure on the likely best fit to ODOT practice scenario containing all soil and pile types were 0.55 and 0.4 for EOID and BOR
respectively. Recommendations were made for a separate implementation activity, including additional calibration work based on the Full
PSU Master including use of field measured hammer performance, CAPWAP based soil input parameters, and pile type.

17. Key Words
Driven piles, GRLWEAP, LRFD, resistance factor, database,
calibration, soil, DFLTD, NCHRP507, pile blow counts
18. Distribution Statement
Copies available from NTIS, and online at
http://www.oregon.gov/ODOT/TD/TP_RES/

19. Security Classification (of this report)

Unclassified
20. Security Classification (of this page)

Unclassified
21. No. of Pages
92 + appendices
22. Price


Technical Report Form DOT F 1700.7 (8-72) Reproduction of completed page authorized Printed on recycled paper
i
SI* (MODERN METRIC) CONVERSION FACTORS
APPROXIMATE CONVERSIONS TO SI UNITS APPROXIMATE CONVERSIONS FROM SI UNITS
Symbol When You Know Multiply By To Find Symbol Symbol When You Know Multiply By To Find Symbol
LENGTH LENGTH
in inches 25.4 millimeters mm mm millimeters 0.039 inches in
ft feet 0.305 meters m m meters 3.28 feet ft
yd yards 0.914 meters m m meters 1.09 yards yd
mi miles 1.61 kilometers km km kilometers 0.621 miles mi
AREA AREA
in
2
square inches 645.2 millimeters squared mm
2
mm
2
millimeters squared 0.0016 square inches in
2
ft
2
square feet 0.093 meters squared m
2
m
2
meters squared 10.764 square feet ft
2

yd
2
square yards 0.836 meters squared m
2
m
2
meters squared 1.196 square yards yd
2

ac acres 0.405 hectares ha ha hectares 2.47 acres ac
mi
2
square miles 2.59 kilometers squared km
2
km
2
kilometers squared 0.386 square miles mi
2

VOLUME VOLUME
fl oz fluid ounces 29.57 milliliters ml ml milliliters 0.034 fluid ounces fl oz
gal gallons 3.785 liters L L liters 0.264 gallons gal
ft
3
cubic feet 0.028 meters cubed m
3
m
3
meters cubed 35.315 cubic feet ft
3

yd
3
cubic yards 0.765 meters cubed m
3
m
3
meters cubed 1.308 cubic yards yd
3
NOTE: Volumes greater than 1000 L shall be shown in m
3
.

MASS MASS
oz ounces 28.35 grams g g grams 0.035 ounces oz
lb pounds 0.454 kilograms kg kg kilograms 2.205 pounds lb
T short tons (2000 lb) 0.907 megagrams Mg Mg megagrams 1.102 short tons (2000 lb) T
TEMPERATURE (exact) TEMPERATURE (exact)
F Fahrenheit (F-32)/1.8 Celsius C C Celsius 1.8C+32 Fahrenheit F
*SI is the symbol for the International System of Measurement
i
i



ACKNOWLEDGEMENTS
The authors of this report wish to thank all those individuals and agencies that have generously
shared data, expertise, and time during the course of this research. The contributions of
undergraduate students Jane Wallis and Jaclyn Schaefer were invaluable on the research team.
Special thanks go to the members of the Technical Advisory Committee: Jan Six, Thomas
Braibish, and Steve Soltesz of ODOT, Tony Allen of WSDOT, Robert Miner of Robert Miner
Dynamic Testing, Inc., and Dr. Naser Abu-Hejleh of FHWA. Their collective input, experience,
and willingness to review the research progress and results contributed greatly to its success.
A debt of thanks goes to all of those who assisted by providing new case histories and/or by
supplying data to resolve case history anomalies, including: Eric Thibodeau (NHDOT), Jay
Wang (LADOT), Mark Morvant (LADOT), Jeffrey Sizemore (SCDOT), Tom Shantz
(CALTRANS), Jeffrey Horsfell (WISDOT), Larry Jones (FLDOT), Darrin Beckett (KYDOT),
Dave Dundas (Ontario Ministry of Transportation), Lynn C. Doyle (Geotechnical Engineering
Testing, Inc.), Houston Walker (TNDOT), Jim Scott (URS CORP), Erik Trum (Multiconsult)
and Denise Whelan (MMSD). Special thanks are owed to those contributors who unselfishly
shared valuable databases: Prof. James Long (Univ. of Illinois), Prof. Mike McVay (Univ. of
Florida), and Prof. Roy Olson (Univ. of Texas). Also thanks to Professors Kyle Rollins and
Steve Bartlett, Dr. Mike Hollaway, and Dr. Bengt Fellenius.
Finally, much praise is due to Judy Fox and Beth Richardson at GRL, Inc. for their help in
supplying biannually the correct versions of GRLWEAP. And to Elise Smith for her diligence,
fortitude, and attention to detail above and beyond the call in editing and proofing.
DISCLAIMER
This document is disseminated under the sponsorship of the Oregon Department of
Transportation and the United States Department of Transportation in the interest of information
exchange. The State of Oregon and the United States Government assume no liability of its
contents or use thereof.

The contents of this report reflect the view of the authors who are solely responsible for the facts
and accuracy of the material presented. The contents do not necessarily reflect the official views
of the Oregon Department of Transportation or the United States Department of Transportation.

The State of Oregon and the United States Government do not endorse products of
manufacturers. Trademarks or manufacturers names appear herein only because they are
considered essential to the object of this document.

This report does not constitute a standard, specification, or regulation.
iii

iv
RECALIBRATION OF THE GRLWEAP LRFD RESISTANCE FACTOR
FOR OREGON DOT
TABLE OF CONTENTS
1.0 INTRODUCTION................................................................................................................. 1
1.1 OVERVIEW AND NCHRP 507 ........................................................................................ 1
1.1.1 Overview..................................................................................................................... 1
1.1.2 The NCHRP 507 Report.............................................................................................. 3
1.2 BACKGROUND AND PHASE 1 STUDY........................................................................ 4
1.2.1 GRLWEAP.................................................................................................................. 6
1.3 ODOT LRFD ISSUES IN PRACTICE .............................................................................. 8
1.4 FIRST ORDER SECOND MOMENT AND MONTE CARLO METHODS.................... 9
2.0 DRIVEN AND GRLWEAP SOFTWARE........................................................................ 13
2.1 DRIVEN APPLICATION RULES................................................................................... 13
2.2 GRLWEAP APPLICATION PROCEDURES................................................................. 15
2.2.1 Overview................................................................................................................... 15
2.2.2 GRLWEAP application............................................................................................. 15
3.0 PSU DATABASE DEVELOPMENT................................................................................ 21
3.1 SOURCES......................................................................................................................... 21
3.1.1 The PDLT2000.......................................................................................................... 21
3.1.2 The FHWA DFLTD Database................................................................................... 22
3.1.3 Other Databases......................................................................................................... 23
3.1.4 Other Sources ............................................................................................................ 24
3.2 FULL PSU MASTER DATABASE................................................................................. 24
3.2.1 Input Quality Tier ...................................................................................................... 25
3.2.2 Case Histories Overview........................................................................................... 28
3.3 CASE HISTORY ANOMALIES AND ERRORS ........................................................... 30
3.4 SUBSURFACE INFORMATION, QUALITY, AND INTERPRETATION................... 33
4.0 BEARING CAPACITY METHODOLOGY.................................................................... 35
4.1 EOID AND BOR MODELING AND BLOW COUNTS................................................. 35
4.2 OUTPUT RANKING....................................................................................................... 36
4.3 GENERAL SENSITIVITY COMMENTS....................................................................... 37
4.4 PILE BLOW COUNTS .................................................................................................... 38
5.0 STATISTICAL PROFILING OF RESULTS .................................................................. 41
5.1 REANALYSES OF NCHRP 507 CASE HISTORY SET................................................ 41
5.2 FULL POPULATION SAMPLE ANALYSIS................................................................. 42
5.3 SCENARIO DEVELOPMENT........................................................................................ 47
5.4 DISTRIBUTION TESTING............................................................................................. 49
v
6.0 OUTLIER EXAMINATION ............................................................................................. 51
6.1 INPUT TIER AND OUTPUT RANK.............................................................................. 51
6.2 BLOW COUNT INVESTIGATION................................................................................ 53
6.3 TIME TO RESTRIKE AND TIME TO LOAD TEST INVESTIGATIONS................... 56
7.0 RESISTANCE FACTOR DEVELOPMENT................................................................... 59
7.1 FIRST ORDER SECOND MOMENT (FOSM) ........................................................... 59
7.2 ODOT SCENARIOS AND MONTE CARLO RELIABILITY ................................... 62
7.3 LRFD FACTORED CAPACITY OPTIMIZATION........................................................ 67
8.0 CONCLUSIONS AND RECOMMENDATIONS............................................................ 71
8.1 CONCLUSIONS............................................................................................................... 71
8.2 RECOMMENDATIONS AND FUTURE WORK........................................................... 75
9.0 REFERENCES.................................................................................................................... 79
APPENDICES
APPENDIX A: FULL PSU MASTER DATABASE
APPENDIX B: LOGNORMAL DISTRIBUTION EOID/BOR STATISTICAL CHECKS FOR
SCENARIO A, F, G, AND I (BOR)
APPENDIX C: SCENARIO A STATISTICAL EOID/BOR PDF AND CDF PLOTS
APPENDIX D: SCENARIO F STATISTICAL EOID/BOR PDF AND CDF PLOTS
APPENDIX E: SCENARIO G STATISTICAL EOID/BOR PDF AND CDF PLOTS
APPENDIX F: SCENARIO I STATISTICAL EOID/BOR PDF AND CDF PLOTS

List of Tables

Table 1.1: Recommended resistance factors from NCHRP 507..............................................................................2
Table 2.1: Acceptable damping factors for variable soil layers ............................................................................17
Table 3.1: Pile case histories in PDLT2000 by pile type and driving time ...........................................................22
Table 3.2: Source of data for pile case histories for resolution of errors and anomalies ....................................25
Table 3.3: Summary of the input tier rubric utilized for analysis ........................................................................26
Table 3.4: Geographic distribution of Full PSU Master pile case histories .........................................................28
Table 3.5: Breakdown of all 322 piles in the Full PSU Master by pile and soil type...........................................30
Table 3.6: Number of piles and anomalies found in the 156 from PSU PDLT2000 Master ................................31
Table 4.1: Summary of output ranking requirements ...........................................................................................36
Table 5.1: Comparison between NCHRP 507, PSU matched piles, and all analyzed 179 piles in the Full
PSU Master.......................................................................................................................................................41
Table 5.2: Statistical characteristics for GRLWEAP based capacity of the 175 qualified cases by soil and
pile type*...........................................................................................................................................................45
Table 5.3: Base statistics for Scenarios A-H compared to NCHRP 507*.............................................................48
Table 6.1: Statistical breakdown of Tier 1, Tiers 1 & 2a, and Tiers 1 & 2 for EOID and BOR........................51
Table 6.2: Top input Tier 1 & 2a statistical breakdown by soil and pile type* ...................................................52
Table 6.3: Top output ranks 1 & 2 statistical breakdown by soil and pile type* ................................................53
Table 6.4: Blow count ranges for EOID and BOR in Full PSU Master ...............................................................55
Table 6.5: Pile counts complying with current AASHTO and ODOT TR and TST times .................................56
Table 7.1: Statistical load parameters used for calibration...................................................................................60
Table 7.2: FOSM resistance values and efficiencies for =2.33 and =3.0...........................................................61

vi
List of Figures

Figure 1.1: Example of a GRLWEAP bearing graph....................................................................................................7
Figure 1.2: Probability of failure and reliability index (William, et al. 1998).............................................................10
Figure 2.1: Example of DRIVEN provided soil resistance to GRLWEAP.................................................................16
Figure 3.1: Growth of cases in the Full PSU Master by input tier ..............................................................................27
Figure 3.2: Distribution of Full PSU Master pile case histories worldwide................................................................28
Figure 3.3: Number of pile case histories in Full PSU Master by state ......................................................................29
Figure 3.4: Source of Full PSU Master pile case histories by region .........................................................................29
Figure 3.5: Source of anomalies between databases for Full PSU Master..................................................................32
Figure 4.1: PSU Case History #293 - effects to the bearing graph of stiffer restrike cushion ....................................35
Figure 4.2: Illustration of DFLTD presentation in digital feet for restrike..................................................................39
Figure 5.1: Predicted BOR capacity to predicted EOID capacity from GRLWEAP ..................................................42
Figure 5.2: Frequency of set-up ratio in the Full PSU Master.....................................................................................44
Figure 5.3: EOID histogram.....................................................................................................................................46
Figure 5.4: BOR histogram......................................................................................................................................46
Figure 5.5: Lognormal distribution fit to Scenario A at EOID....................................................................................50
Figure 6.1: All EOID and BOR qualified 175 values plotted to blow count by soil type ........................................54
Figure 6.2: All EOID and BOR qualified 175 values plotted to blow count in 2 BPI intervals...............................55
Figure 6.3: Restrike and load test time ratios by soil type...........................................................................................57
Figure 6.4: Restrike and load test time ratios by pile type ..........................................................................................57
Figure 7.1: Scenario A Monte Carlo reliability method CDF fits at EOID.................................................................64
Figure 7.2: Factored BOR plotted to factored EOID for Scenario A..........................................................................68
Figure 7.3: Factored BOR plotted to factored EOID for Scenario F...........................................................................68
Figure 7.4: Factored BOR plotted to factored EOID for Scenario G..........................................................................69
Figure 7.5: Set-up implementation on cohesive soils factored BOR Scenario I to factored EOID for
Scenario G..........................................................................................................................................................70
Figure 7.6: Set-up implementation on cohesionless soil factored BOR Scenario J to factored EOID for
Scenario G..........................................................................................................................................................70


vii
viii


1.0 INTRODUCTION
1.1 OVERVIEW AND NCHRP 507
1.1.1 Overview
In Oregon, the Bridge Section of the Oregon Department of Transportation (ODOT) is
responsible for satisfactory design, based on current standards of practice, of all bridge
structures, including bridge foundations, within the states highway system. The Bridge Section
also plays a significant leadership role in the distribution and implementation of new and
emerging bridge technologies. For implementation of Load Resistance Factor Design (LRFD)
reliability-based principles for bridge foundations, the American Association of State Highway
and Transportation Officials (AASHTO) and the Federal Highway Administration (FHWA)
require LRFD adoption through the AASHTO bridge design code. The LRFD reliability-based
principles created a precedent in the areas of driven pile foundation policy and implementation
from national load test databases to fit highly regionalized standards of practice. The
widespread adoption by the nations departments of transportation (DOTs) of geotechnical
LRFD principles, specified in the AASHTO code, has been difficult in the case of deep
foundations. Difficulty around LRFD principles for driven pile design, including resistance
reduction factors (called resistance factors) for dynamic methods, has caused multiple and
local implementation concerns. These have led to successive revisions to the AASHTO code,
and a fourth edition was subsequently released (AASHTO 2007). The AASHTO Bridge Design
Specification, 2006 Interim Revision drew heavily from the work performed for both drilled
shafts and driven pile foundations by the National Cooperative Highway Research Program
(NCHRP), referred to hereafter as NCHRP 507 (Paikowsky 2004).
The well-known Wave Equation of Pile driving (GRLWEAP
1
) software (Pile Dynamics, Inc.
2005) calculates the induced stress and displacement waves traveling along the pile for a single
hammer blow. The program computes the pile blow count, axial stresses in the pile and
transferred energy from the hammer for each capacity analyzed. The program then allows the
user to indirectly determine bearing capacity at the time of driving or restrike using observed
blow count. In the most recent version, GRLWEAP features include improved set-up models, as
well as use of the pile static capacity DRIVEN software code (Mathias and Cribbs 1998) for soil
resistance distribution. The Phase 1 portion of this research effort for ODOT (Smith and
Dusicka 2009), included a survey of Northwest state DOTs current pile field acceptance
practice relating to GRLWEAP. The surveys objective was to assess both the overall use of
GRLWEAP in Northwest practice and the effects of applying the AASHTO resistance factors
using GRLWEAP to bridge designs in the Northwest. The surveys results showed a strong
opinion among DOT practitioners to support increasing the AASHTO reported GRLWEAP

1
1
Only GRLWEAP version of the wave equation of pile driving was used in this research and no other code is either
endorsed or approved.

resistance factors, , in the code when used on both initial driving and restrike to capture pile
capacity. The objective of this study is to provide new calibrated resistance factors for ODOT
field practice for the end of initial driving (EOID) and the beginning of restrike (BOR) driving
conditions.
ODOT typically uses the GRLWEAP method for verifying the predicted nominal resistance of
driven pile deep foundations. According to the Pacific Northwest DOT survey for the resistance
factor of GRLWEAP in AASHTO codes, approximately 80% of respondents believed the
value of 0.4 (Table 1.1) for redundant piles with a 1% probability of failure is too conservative,
and 37.5% of those surveyed do not use 0.4, but have set separate and higher values (Smith and
Dusicka 2009). The purpose of this research was to determine the appropriate resistance factors
for the GRLWEAP method using an extended high-quality pile load test database, including data
from the NCHRP 507 study, the FHWA DFLTD (Raghavendra, et al. 2001) database, and other
sources. For completeness, the GRLWEAP method is discussed in Section 1.2.1 and Section 2.

Table 1.1: Recommended resistance factors from NCHRP 507
Mean S. D. COV Resistance Factor, Method Time of
Driving
# of
Cases
P
f
=1.0%,
=2.33
P
f
=0.1%,
=3.0
EOID 125 1.626 0.797 0.49 0.65 0.45 CAPWAP
BOR 162 1.158 0.393 0.339 0.65 0.50
EOID 128 1.084 0.431 0.398 0.55 0.40
Dynamic
Measure-
ment
Energy
Approach BOR 153 0.785 0.29 0.369 0.40 0.30
ENR general 384 1.602 1.458 0.91 0.25 0.15
Gates general 384 1.787 0.848 0.475 0.75 0.55
Dynamic
Equation
FHWA Modified general 384 0.94 0.472 0.502 0.40 0.25
GRLWEAP EOID 99 1.656 1.199 0.724 0.40 0.25

The Phase 2 effort described in this report had two primary objectives: to build an extensive
database from the present sources that reflects ODOT diverse soils, and to calibrate the
resistance factors for the GRLWEAP method for use with EOID and BOR conditions. To
achieve the primary objectives, the tasks described below for this Phase 2 were performed:
- Searched and reviewed current national DOT LRFD recalibration efforts and the
available published procedures discussing LRFD calibration, as well as the references
adopted in AASHTO Bridge Design Specification (AASHTO 2006) for the resistance
factors.
- Reviewed and recompiled the static and dynamic tested pile case histories from the
existing databases made available by ODOT, and generated a new Portland State
University (PSU) database for this study.
2
- Performed static analysis with the DRIVEN software program and the bearing graph
evaluation of capacity with the GRLWEAP software program.

- Searched and expanded the database with new documented cases. Conducted a statistical
analysis to profile the database and design practice scenarios.
- Recalibrated the resistance factor by using the reliability-based methods for the wave
equation analysis and recommended resistance factors for approved scenarios.
FHWA has collected extensive deep foundation load test information and compiled a database
with entries up to 2004, primarily drawn from state DOT load tests. The database offered more
than 1,000 load test entries from 1985 to 2004 and was obtained by the research group in
preparation for this Phase 2 recalibration study. The information collected included soil boring
logs, laboratory data, and the field load test results for both driven pile and drilled shaft deep
foundations, including pile driving logs up to the EOID, as well as the restrike-driving log when
a pile is re-driven after some delay, referred to as BOR driving. J. Long (Long, et al. 2009)
accessed five independent databases to identify statistical means and standard deviations for
seven capacity predictive methods. Some of these databases included EOID and BOR driving
data that permitted examination of the GRLWEAP produced bearing graph, as well as CAse Pile
Wave Analysis Program (CAPWAP) predictions of nominal capacity. For both these methods,
Long illustrated the improvement in predictive ability achieved by moving from EOID to BOR
blow counts, something NCHRP 507 did not illustrate for the GRLWEAP bearing graph. The
improvement in both mean and standard deviation reported by Long when comparing EOID to
BOR was statistically better for GRLWEAP than for CAPWAP.
1.1.2 The NCHRP 507 Report
The NCHRP 507 project was performed to provide updated resistance factors for a new
AASHTO LRFD Bridge Design Specification edition that would reflect current practice in
geotechnical design and construction (Paikowsky 2004). The resistance values and methodology
in codes existing at that time were thought to be significantly conservative and would increase
foundation costs. As the first step of the NCHRP 507 project, high quality databases containing
case histories of piles tested to failure were compiled and reliability-based methods were used
for the calibration of the resistance factors. Then statistical analyses were performed for all
acceptable methods including the static analysis by -method, Nordlunds method, and dynamic
analysis for six different methods: CAPWAP, Energy Approach, ENR, Gates, FHWA modified
Gates, and the GRLWEAP method.
For the reliability-based method, the target reliability index was recommended according to the
condition of pile redundancy. NCHRP 507 recommended resistance values from the First Order
Reliability Method (FORM) rather than the First Order Second Moment (FOSM). The former is
called the advanced FOSM method for the case of a lognormal distribution of the variables and a
non-linear limit state function.
3
The NCHRP 507 work statistically established resistance factors by setting dead load to live
load ratios, site variability, pile redundancy, soil types, and the quality of dynamic testing; all of
which are, in fact, regional factors. The AASHTO code was an attempt to provide a basis for
uniform implementation of LRFD. The NCHRP 507 report recommended resistance factors for
static and dynamic analyses methods and recommended the number of dynamic tests according
to the site variability. Table 1.1 presented the recommended resistance factors for dynamic

analysis extracted from NCHRP 507. Resistance factors, , at BOR condition in GRLWEAP
method were not shown in the NCHRP 507 main report text but can be located in the appendix
of the report. It appears likely that the values originated with the earlier FHWA report using
GRLWEAP at both EOID and BOR (Rausche, et al. 1997). ODOT obtained the NCHRP 507
database with modifications, including quality metrics and pile selection strategy, from the
Washington State Department of Transportation (WSDOT) that was employed in its own
resistance factor calibration of the pile driving dynamic formula. The Phase 1 study for ODOT
(Smith and Dusicka 2009) revealed that considerable information was missing for approximately
two-thirds of the NCHRP 507 pile load test database on restrike.
1.2 BACKGROUND AND PHASE 1 STUDY
The LRFD methodology at the ultimate limit state calls for the load to be factored up by load
factors () assigned to the load source (e.g. dead, live, wind) and compared to a reduced nominal
resistance, R
nk
, employing a resistance factor, . The inequality to be satisfied in LRFD based
design is set out in the equation below:

nk k ij ij R Q

s
(1-1)
where: Q
ij
= Structural load from each source condition

ij
= Magnification load factor set by the code
R
nk
= Nominal strength-based resistance established by a defined method

k
= Resistance factor for the defined resistance method

Q
ij
and R
nk
are not deterministic but are random variables; therefore, the calibration of

ij
and
k

to foundation design employs a fixed Reliability Index value, , quantifying risk for the
foundation. Deep foundations with a high redundancy in a group have a 1/100 probability of
failure, which sets the target at 2.3. The NCHRP 507 required low redundancy foundation
groups (< 4 piles) to have a stricter 1/1000 probability of exceedance with at 3 as the target
reliability. The calibration reported here for ODOT driven piles uses the pile redundancy target
of 2.33 and AASHTO approved
ij
load factors. The calibration by matching to Allowable
Stress Design (ASD) is only a mathematical based design equivalency to calculate the same
number of piles. It uses no reliability theory and should not be used as a valid calibration for .
The appropriate resistance value (determined by any method) to satisfy the inequality of
Equation 1-1 is a function of the structures proportion of live load (LL) and dead load (DL) and
the AASHTO code load factors,
ij
.
4
Given the mandated October 1, 2007 implementation date for adoption of the AASHTO bridge
code, the number of individual state DOTs making efforts to implement the code with load
modifications has increased. Continued research efforts to meet local needs, as well as bridge
code changes, are occurring. The concerns of state foundation engineering practitioners caused
major revisions to the code, resulting in the subsequent release of a fourth edition of the code
(AASHTO 2007). In parallel to the AASHTO/FHWA effort, the last five years have seen a
growth in LRFD research material published in foundation engineering journals and conference

proceedings. It is evident that LRFD implementation is proceeding slowly. Better recognition
of the need for regional and local standards of practice, together with improvements in statistical
competency to assist local implementation of the AASHTO recommendations, is beginning to
gain momentum.
Tony Allen (Allen 2005a) provided a clear presentation of axial capacity, load, and resistance
factor historical development, as well as the difficulties relating to driven pile foundations. In
addition, a calibration of the resistance factor was undertaken by WSDOT (Allen 2005b) for the
Modified Gates pile driving formulae based on reinterpretation of the same pile driving
databases accessed by NCHRP 507. Both of these reports were identified by ODOT as key
reference studies in the recalibration effort. However, ODOT had extensive experience and
successful application of GRLWEAP. The evaluation of the nominal static capacity by ODOT
for each pile was most often performed in the field using GRLWEAP models, while the
AASHTO code presented resistance factors for this technique in its LRFD methodology.
GRLWEAP models of the pile driving hammer, driving accessories, pile, and soil by a viscous
mass-spring system is widespread and constitutes the industry standard internationally.
The Phase 1 study identified the current trends in recalibration, key resources to assist in
designing implementation procedures, and the role played by GRLWEAP in assessing nominal
capacity. In reality, piles are often driven at locations for which no borehole exists at that
position to provide concise subsurface conditions. Subsurface conditions, together with pile size
and type, dictate the axial capacity. This limits the direct application and reliability of static
methods derived from borehole testing of samples to those physical locations coinciding with
borehole locations. In all calibration efforts, the basis for statistical comparison between
different prediction methods is the static load test capacity using Davissons interpretation.
Recall that predicting Davissons capacity is not the intent of either the static capacity analytical
methods or the dynamic methods. (AASHTO 2009 mandates a different interpretation for piles
with diameters in the range 24 to 36 in.) Therefore, in addition to analytical modeling errors, a
bias to Davissons capacity is introduced and statistically reported from any database studies.
In addition to NCHRP 507, significant documentation of correct consistent design procedures for
pile foundations had recently appeared from FHWA (Hannigan, et al. 2006). This two-volume
report offered a comprehensive exploration of pile design and analysis in Volume I and field
testing quality control and field dynamic testing in Volume II. The key communication
connections to the structural engineer, bridge engineer, and contractor were well presented, as
were the decisions concerning construction capacity verification testing. Both volumes were
directed toward ASD and did not directly incorporate LRFD principles. A simple LRFD
example was presented in Volume I of Appendix G, following the discussion of the LRFD
structural origins.
A series of studies in the Phase 1 effort reported the full effect of switching design
methodologies from ASD to LRFD linked to ODOTs design and field procedure. The first
study of Phase 1 surveyed the U.S. Northwest States DOTs to determine each DOTs standard
of practice GRLWEAP recommended

value. The survey results found strong support for a
LRFD resistance factor recalibration effort of GRLWEAP. Summary findings included:
5
- 60% of the surveyed Northwest DOTs utilized GRLWEAP.

- 80% of the surveyed Northwest DOTs considered a
n
at 0.4 to be conservative.
- 750 bridges were expected to be designed for the Northwest in the next 10 years.
- 60% of the surveyed Northwest DOTs were willing to assist in a recalibration effort.
The second part of the Phase 1 effort studied the economic impact to ODOT via the cost
difference between bridge foundations designed with ASD vs LRFD utilizing ODOTs standard
of practice. The study concluded that with GRLWEAP as the design tool, a 30% increase to
foundation costs could be expected when utilizing LRFD design compared to ASD design. Pile
capacity, whether established by static analyses, dynamic testing in the field at the time of
driving, or load testing, was governed by the soil layer(s) shear strength around the pile
perimeter and at the pile tip. Time and economic constraints dictated that only limited
geotechnical sampling, testing, and logging of boreholes were performed. By assigning a higher
resistance factor, AASHTO declared that the pile driving analyzer (PDA) signal matching
technique with CAPWAP technology was more reliable; however, use of this technology was
cost prohibitive for many bridge piling contracts.
Use of the AASHTO and the FHWA approved GRLWEAP bearing graph for nominal capacity
has two distinct advantages over signal matching with PDA and CAPWAP analysis. First,
according to the AASHTO code, site statistical variability work does not have to be completed
since the inspector assesses each pile at the time of driving. Second, the bearing graph check is
a deliberate activity using observed field pile and hammer performance to determine that each
pile meets the limit state axial nominal capacity. Geotechnical site variations are established by
both site investigations and by the design team applying their own local experience, knowledge,
and judgment. These variations can then be incorporated in region-specific input of soil
parameters into GRLWEAP, including soil side and tip quake and viscous damping parameters,
as well as soil layering across the site. This study was based only on default parameters to
ensure recommended resistance factors can be applicable to a broad range of pile types and
subsurface conditions.
1.2.1 GRLWEAP
GRLWEAP calculates the induced stress and displacement waves traveling along the pile after a
single hammer blow and reports pile permanent set after elastic rebound, called quake (Q). It
further assists in decisions about pile drivability and reports the change in static equivalent
bearing capacity at the time of driving (R
ult
), from changes in field blow count by means of a
bearing graph. For illustration purposes only, Figure 1.1 presents an example of the bearing
graph with a range of possible capacities as a function of driving blow counts. When the field
driving blow count is recorded, the graph can be used to read the static R
ult
predicted by the
program at the time of driving.
6



Figure 1.1: Example of a GRLWEAP bearing graph
Establishing bearing capacity from GRLWEAP becomes more difficult by the large number of
site-specific variables, the modeling complexities, and the sensitivity of the output to all driving
components, particularly the hammer efficiency. These site specific variables include the
equivalent soil springs elastic quake movement (Q
i
) and the soil springs viscous damping
values (J
i
) for each soil supported pile element. The typical application of GRLWEAP is often at
two stages, both in design and construction: the first stage is during the pre-bid period to
establish that the pile designed by static methods can be driven by available equipment, and the
second stage is after the chosen contractor selects the final production hammer and driving
accessories. At the second stage, the field bearing graph and hammer stroke to capacity plots,
which control final penetration depths, are made available to the agency field inspector for
confirmation of capacity and acceptance of the pile.
7
It is well known that pile long-term capacity will often show a capacity gain, called set-up, and
occasionally show relaxation when the capacity drops. Under these conditions, the use of
measured driving blow counts at EOID could yield either conservative or unconservative
capacity results and the pile should be restruck after a waiting period (often a minimum of 24
hours) to give a more representative blow count. This is called the BOR blow count, and any
associated use of a bearing graph must be established from the appropriate GRLWEAP model
bearing graph. Even when static load tests have been conducted at the site, AASHTO
recommends that GRLWEAP be used to extrapolate these load test results to the production
piles. GRLWEAP features include improved set-up models, as well as the use of the static pile
capacity DRIVEN software code (Mathias and Cribbs 1998) for input of soil resistance
distribution data to GRLWEAP.

1.3 ODOT LRFD ISSUES IN PRACTICE
NCHRP 507 allowed some reliance on local judgment and experience in the application of
any field verification procedures to establish factored nominal resistance. Two disturbing
features in the report emerged that limited flexibility for implementation of dynamic testing to
local conditions: first, the elimination of soil types as a variable when selecting resistance
factors for GRLWEAP and second, the previously reported absence of the GRLWEAP-BOR
combination resistance factor to be used after pile set-up has occurred. To capture the
transportation agencies standard of practice, the AASHTO code acknowledges that regional
implementation can proceed after local recalibration efforts are complete.
No agency implementation policy is discussed in the AASHTO code to assist in the transition
from ASD to LRFD. Historically, the ODOT Bridge Section has generally followed the
recommendations contained in all past and current FHWA manuals for driven pile design and the
code requirements set by AASHTO. For most Oregon bridges, ODOT requires a minimum of
one logged and sampled borehole per pier, with limited laboratory shear strength testing
conducted. Foundation conditions throughout the Willamette Valley, coastal development
regions, and the Portland metropolitan area are predominantly sand, silt, and clay. Bridge
foundation piles, typically steel pipe and H section piles, are of sufficient length to be primarily
friction piles and these soils are known to exhibit set-up after EOID.
The statistical studies reported in NCHRP 507 to establish resistance factors for the AASHTO
code used default soil and hammer parameters in GRLWEAP and had no restrike condition
included to capture any known soil set-up that may be large in Oregon soils. The AASHTO
reported values for static analysis are generally low, and static analyses are now relegated for
use to establish preliminary pile sizes and lengths, for contract purposes only. However, static
analysis do form the basis of pile side shear to pile tip capacity ratios used in the GRLWEAP
program and are most often established from uncertain SPT blow counts that statistically have a
large coefficient of variation (COV) between 15% to 40% (Duncan 2000).
The concern for predicting the ultimate capacity at restrike is related to the pile gain set-up in
resistance governed by uncertainty from geotechnical soil characteristics. About 70% of the
Willamette Valley subsurface consists of flood deposits, clay and silt. In these soils, the capacity
of a driven pile increases with time after driving, and the increased capacity is indicated by
restrike blow count. After the selected contractor proposes hammer and driving accessories and
other specific details become known, the inspectors graphs are prepared showing required blow
count versus hammer stroke for a given static resistance. ODOT provides the soil input
parameters for GRLWEAP in the contract specifications. ODOT has routinely used GRLWEAP
for capacity at EOID and occasionally at BOR if significant set-up was expected and the EOID
measured capacity was low. In ASD, both EOID and BOR capacity values were used with a
recommended factor of safety of 2.5.
8
Within any DOT, the accumulated foundation engineering knowledge base helps establish the
accepted standard of practice. Pile design is set by site investigation results for which the
amount of data, data quality, and interpretation are locally and regionally specific. Much of the
NCHRP 507 research findings removed the insensitive silt soil category, which is common in

Oregon, with no specific recommendations offered for this soil type. The AASHTO code
cautions using the published resistance factors when piles are over 24 in in diameter, most
likely because a limited number of 24 in diameter piles are in the database used to establish the
resistance factor. The most likely reason is that no piles over this diameter were used in the
statistical calibration. However, these large sizes are not typical in present ODOT practice.
1.4 FIRST ORDER SECOND MOMENT AND MONTE CARLO
METHODS
LRFD design separates uncertainties into two independent variables: load and resistance,
expressed with separate load and resistance factors. Satisfactory design requires that the
factored down resistance for a pile should be larger than the linear combination sum of the
factored loads.
The sum of the factored loads may represent the possible largest statistical acceptable load
combination. The nominal resistance (similar to ultimate capacity as defined in ASD) is
established from the code-approved procedure. The resistance factor, , is less than one, and
applied to reduce the measured nominal resistance. The adoption of factors to increase the load
and to reduce the resistance is based on probability theory to model uncertainty. Therefore,
failure based on reliability methods is defined when the load exceeds the resistance, i.e.
expressed by the area in which the two probability density functions (PDF) for load and
resistance overlap. This overlap will be controlled by the resistance factor, because the PDF of
source loads usually have much less variation than the resistance distribution. For this study, the
resistance factor is calculated using reliability theory with both the FOSM, and more advanced
probabilistic Monte Carlo random number based method (Allen, et al. 2005).
The FOSM method uses the first terms in a Taylor series expansion of the performance
function to estimate the expected value and variance of the function. It is called a second
moment method because the variance is a form of the second moment and is the highest order
statistical result used in the analysis (Baecher and Christian 2003).
The limit state function is represented by the safety margin and can be defined as:
0 Q R Q) g(R,
i i n
> =
(1-2)
where Q
i
is the load,
i
is the load factor g is the limit state function and, if g is less than zero,
failure is predicted.
9
This function makes the two variable distributions (shown on the left of Figure 1.2) merge to one
distribution (shown on the right of Figure 1.2) and illustrates the probability of failure as when
the safety margin is less than zero.


Figure 1.2: Probability of failure and reliability index (William, et al. 1998)
To establish the statistical parameter for the resistance, the ratio of the measured to the predicted
capacity is established, called the bias factor, , from a population of case histories. In this
study, the measured value is the Davissons pile failure capacity secured from the database.
The predicted value from the GRLWEAP analysis is based on the bearing graph with the
measured blow counts used both at the EOID condition and at the BOR condition, and is further
discussed in Section 4.1.
The distance from the failure region, Pf, to the mean value of the limit state function, g, can be
expressed, using mean bias, , and COV, as |o, where o is the standard deviation (s.d.) of the
limit state function, and | is the reliability index. Pf is typically represented by the reliability
index parameter, |, a function of both the load statistics and resistance statistics, i.e. the
acceptable magnitude of | called the target reliability depends on the desired value of Pf for the
pile. To permit final calculation of resistance values, some assumptions are needed. Dead load
(Q
D
) to live load (Q
L
) ratios ranging from 2 to 5 have previously been investigated since this
range is typical for bridges and similar structures (Allen 2005a) and the final is typically found
insensitive to this ratio. When only dead and live loads are considered, the resistance factor, ,
can be found by FOSM method (See Section 7.1) the following equation:


( )
( )( )} {
2
QL
2
QD
2
R T QL
L
D QD
2
R
2
QL
2
QD
L
L
D D
R
COV COV 1 COV 1 n 1 exp
Q
Q
COV 1
COV COV 1
Q
Q
+ + + |
|
|
.
|

\
|
+

+
+ +
|
|
.
|

\
|
+

=
) 3 1 (

where:
R
,
QD
,
QL
= resistance, dead and live load bias factor,

D
,
L
= dead and live load factor,
10
Q
D
/Q
L
= dead to live load ratio,

COV
R
, COV
QD
, COV
QL
= coefficient of variation of the resistance,
dead and live load factors, and
|
T
= target reliability index.

In the above equation, all statistical values, except
R
and COV
R
, are generally taken at the given
code values. In this study the statistical values for load follow NCHRP 507 (Appendix B) and
the resistance
R
and COV
R
are taken from the result of the GRLWEAP database analysis
performed for a variety of scenarios. However most of the resistance factors recommended for
use from NCHRP 507 are the more advanced FORM method.
After careful review of source pile driving data, more sophisticated and accurate calibration
using the AASHTO endorsed Monte Carlo probabilistic procedures can proceed. This procedure
requires high quality PDFs, and has been discussed by Allen, et al. (2005). Variability in the
total load from dead and live load sources, as well as the variability of resistance, are expressed
in the form of the safety margin and are generated by random number Monte Carlo procedures
according to the distribution mean and COVs. The safety margin defines risk of failure that
arises from the fitting of the lower
R
bias resistance distribution tail (where predictions are
unconservative) and upper total load bias tail. Intermediate steps include constructing the PDF
of the calculated
R
resistance bias curves (from GRLWEAP measured capacities), conversion to
cumulative distribution function (CDF), and finally via each case,
R
creation of the standard
normal variable (SNV) plots.
This research incorporated the recommendations and the example offered by WSDOT (Allen
2005b and Allen, et al. 2005) using lognormal best fits from the following three approaches:
regressed fitting all the case history data points, regressed fitting by dropping data points from
the upper (conservative) tail, and finally, fitting the lower tail by visual adjustment. Random
number generation provided both the dead and live load distributions and the resistance
distributions, and established the final safety margin distribution with the preset dead to live load
ratio. By iteration the appropriate resistance factor, , was found to produce the target
reliability index |
T
value. These probabilistic procedures used Excel

spreadsheet computation
for convenience.
11


12












2.0 DRIVEN AND GRLWEAP SOFTWARE
FHWA (Hannigan, et al. 2006) endorses the dual use of DRIVEN 1.2 software (Mathias and
Cribbs 1998) to calculate the likely pile static capacity resistance distribution and format the
input file to GRLWEAP for bearing graph capacity calculation. This research employed all
default parameters and options in both codes consistent with EOID and at BOR conditions.
These procedures are discussed by FHWA and were diligently followed in derivation of the
GRLWEAP capacities in this study. It is essential that any use of the wave equation to establish
LFRD capacity with the recommendations contained in this report use only GRLWEAP as other
programs contain differences that affect the bearing graph and capacity. Application discussion
and, when required, rules for consistency in this research are discussed below.
2.1 DRIVEN APPLICATION RULES
DRIVEN performs static analysis computations utilizing Alpha method (Tomlinson 1980) for
cohesive soil and Nordlunds method (1963, 1979) for cohesionless soil provided by FHWA
(Hannigan, et al. 2006). Both the Alpha and Nordlund method are current methods and appear
in the AASHTO code (AASHTO 2009) for static analysis capacity prediction. A consistent
approach was taken by the research group in analyzing standard penetration test (SPT) blow
count, N, data for input into DRIVEN for cohesionless soil layers. N values can be input directly
into DRIVEN to calculate an effective friction angle for the layer using Meyerhofs method.
DRIVEN accepts a maximum of five N values, creating a limitation in a layer which might
include much more than five blow counts. Rather than picking five representative N values for
input, the research group calculated effective friction angle values using the full set of reported N
values for each layer as the DRIVEN program input. Analyzing the full set of values allowed for
a more accurate representation of the layer, especially considering the inherent variability of SPT
test results. In addition, obvious outliers could be removed prior to determining an effective
friction angle from SPT results.
The applicability of the SPT to cohesive soils is largely agreed as minimal. The FHWA
(Hannigan, et al. 2006) recommends the SPT solely for use with cohesionless soils. However, in
many instances there was no other data other than the N values available for clay or silt layers.
In the absence of other soil data, N values were used to estimate undrained shear strength using
correlations published by Bowles (Bowles 1996). Bowles lists a range of uncorrected N values
corresponding to a range of unconfined compressive strength values. Shear strength values
based on SPT data were used only in absence of any other available soil data.
13
Default parameters were utilized for all inputs following the DRIVEN manual recommendations
and those given by AASHTO and FHWA driven pile manuals. Some additional interpretation
and judgment rules were required for DRIVEN analysis that went beyond the reference manual
set. These included:


- The program could not export open ended pipe (OEP) piles to GRLWEAP. As the
majority of OEP pile to penetration depth ratio fell within the FHWA plugged condition
they were modeled closed end.
- H-Pile toe areas were taken to be the area of the metal while the side friction
development utilized plugged box perimeter dimension.
- Only square concrete piles are supported in the DRIVEN program. The pile case
histories recorded in this researchs database were predominantly frictional piles;
therefore, the outside perimeters of non-square shaped concrete piles were transformed
into an equivalent sized square pile to retain proper area for side friction development
during DRIVEN analysis.
- DRIVENs prime purpose use was to calculate likely shaft and tip resistance distributions
and the GRLWEAP input file. The program input prompts recommending capping
cohesionless soil friction angles was disregarded and the gain/loss factors were not used.
Due to the methods accepted for static capacity analysis there were internal restrictions placed
on the reported capacities. These were as follows:
- For cohesionless soil, Nordlunds method for side friction calculations limits the
acceptable range of friction angle to the 20 to 45 range. Values outside this range were
not allowed.
- Pile end bearing capacities were limited following the recommendations of Meyerhof
found in Table 7.2 in the DRIVEN manual. The program did not consider any capacities
larger than those recommended by Meyerhof.
14
After creation of a soil profile distribution, the calculated side friction and end bearing was
available and the tabulated output capacity screen was examined to determine each individual
soil layers contribution to side friction capacity, and the overall distribution of ultimate
capacity. A general soil type category was assigned to the case for ease of case history
organization. Each case was then placed into one of three subsurface categories: Clay, Sand, and
Mixed. Cases that reported cohesive soils contributing to more than 80% of a piles total
capacity were considered Clay case histories. Cases that reported cohesionless soils contributing
to more than 80% of a piles total capacity were considered to be Sand case histories. Layered
soil cases that fell in between these brackets were considered Mixed cases. Predominant bearing
conditions along a piles shaft and at the toe, together with the distribution of side friction
between the layers and toe, were recorded for later use in the GRLWEAP analysis. All the
recommendations offered by FHWA and the DRIVEN manual for a pile to plug in cohesive soils
were followed for both EOID and BOR driving conditions.

2.2 GRLWEAP APPLICATION PROCEDURES
2.2.1 Overview
The basic principle of the Wave Equation Analysis for Pile driving is the expression, in finite
difference form, of the governing partial differential equation modeling the transfer of kinetic
energy into the pile to overcome the static and dynamic soil resistance. In the driving process,
the hammer kinetic energy is delivered to the pile in the form of compressive force pulse (Pile
Dynamics, Inc. 2005). At the bottom of the pile, the force pulse reflects and moves up to the top
again. While the energy travels from the hammer to the bottom of pile and back to the top of
pile, there are energy losses in pile, soil, and driving system including hammer, cushion, and
helmet. This transmission of energy will produce a permanent set at the pile into the supporting
soil. The principle of energy transfer is expressed from the GRLWEAP manual as follows:
Kinetic Energy (E) Energy losses (E
loss
) = Resistance (R
n
) permanent set (s) (2-1)
The penetration resistance, N, is the inverse of the permanent set, s, calculated by the pile and is
expressed in either blows per foot (BPF) or blows per inch (BPI). This report uses the BPI
expression of driving resistance exclusively. Wave equation modeling requires the following
input to be available from every candidate pile case study:

- Input for pile stiffness and energy loss: pile size, elastic modulus, specific weight.
- Input for soil resistance and damping energy loss: percent friction contribution, quake
and damping factors.
- Input for hammer energy and efficiency: hammer type (often in the hammer library).
- Input for driving system energy loss: helmet weight, cushion stiffness, coefficient of
restitution, thickness, and elastic modulus.
GRLWEAP then provides an estimation of bearing capacity from a bearing graph and represents
the relationship between equivalent static capacity at the time of driving and the pile
penetration resistance, expressed as blow count, N. For this study, GRLWEAP 2005, with all
updates and issued biannually up to July 2010, was supplied by Pile Dynamics, Inc. and used to
predict the ultimate capacity with the field observed blow counts reported in the database at
EOID, and at BOR following a time delay. Section 4.1 discusses specific details of using the
assembled database to calculate these EOID and BOR capacities.
2.2.2 GRLWEAP application
15
Guiding the analysis process were the GRLWEAP 2005 Manual, the DRIVEN Manual, the two
FHWA volumes on driven pile foundations (Hannigan, et al. 2006), and the original work
performed by Goble, Rausche, and Likins (GRL) that formed the basis of the NCHRP 507
recommendations around the use of GRLWEAP (Rausche, et al. 1997). On occasions when the
manuals were unclear, or omitted necessary steps, the standard interpretation for ODOT was
determined through discussion with select research Technical Advisory Committee (TAC)

members. At key stages in program development, systematic steps, including careful review and
sensitivity studies, were performed, as well as other activities to avoid gross modeling errors.
Importing the DRIVEN soil profile input into GRLWEAP requires the User select from the file
menu Open Pre 2002 input file (*.GWI) option. The file produces the DRIVEN soil profile
and graphically represents the individual soil layers capacity contribution relative to one
another. As the soil profile is constructed by DRIVEN, no alterations to soil properties are
permitted in GRLWEAP. Therefore, any sensitivity analyses with variations of soil strength
parameters, water table location, or profile geometry required a separate DRIVEN input file be
produced. Hammer type is selected from the program pull down menus, and hammer model
details and performance information are found in the GRLWEAP library. If the hammer library
did not list the required hammer, a similar hammer type with matching energy/power and ram
weight was selected. If no similar match to a hammers energy/power and ram weight was
available, the hammer with the closest energy/power rating was selected. Default values for all
hammer accessories and helmet weight were used unless specific values were provided in the
field records. Hammer efficiencies, pressure, and stroke were never altered from their default
values.
Figure 2.1 shows the GRLWEAP pile and hammer details and the distribution of pile resistance
provided by DRIVEN for one of the piles analyzed.


16
Figure 2.1: Example of DRIVEN provided soil resistance to GRLWEAP

Both the hammer and pile cushion provide protection to the hammer and pile and are key
components in GRLWEAP modeling as they modify the delivered energy. Appropriate pile
cushion input proved occasionally problematic for concrete piles and rarely was the hammer
cushion or pile cushion properties reported in the source data. If multiple pile cushion options
were available, then each option was selected successively and a sensitivity analysis was
performed to assist in gauging the sensitivity of the prediction. Purpose designed rules beyond
the GRLWEAP Manual that guide hammer cushion selections were discussed with the TAC
member, Robert Miner. These discussions provided the following guidance:
- If multiple hammer cushion leads were available through the GRLWEAP library, a lead
was selected that was 4-6 in larger than the diameter of the pile being driven. This
countered the affect of a worn or damaged lead and insured proper contact.
- If no hammer cushion parameters were available in the GRLWEAP library, a cushion
from the closest matching hammer with a matching ram weight was selected
- According to Robert Miner, Robert Miner Dynamic Testing Inc., likely pile cushion
thickness should be no more than 12 in. An error was discovered for the Conmaco 300-C
hammer aluminum cushion in the GRLWEAP library, which reported an unreasonable
cushion thickness of 29 in.
GRLWEAP captures the dynamic effects of the soil/pile interaction through the use of its quake
and damping factors. Quake is a measure of the elastic recoverable slip between the soil and pile
after a single hammer strike. These quake and damping parameters are directly influenced by soil
stiffness, and the FHWA (Hannigan, et al. 2006) suggests there may also be a difference
between the EOID and BOR condition. These parameters and their role in GRLWEAP analysis
are explained in detail in the GRLWEAP manual. Following the manual recommended
procedures, and in consultation with the TAC, a research quake and damping selection
methodology was developed. Shaft damping (j
s
) utilized a weighted average method to
determine the appropriate single j
s
value for modeling the soil conditions along the entire length
of a piles shaft. GRLWEAP recommends j
s
=0.05sec/ft for cohesionless soils and j
s
=0.20 sec/ft
in cohesive soils and interpolation was employed for intermediate soil conditions. The Phase 1
effort (Jackson 2008), in consultation with ODOT and Robert Miner, developed the Table 2.1
recommendations of acceptable damping factors for variable soil layers. Each case history
logged soil layer description along the length of a piles shaft was assigned its appropriate j
s
value from the table.

Table 2.1: Acceptable damping factors for variable soil layers
Soil Type
j
s

sec/ft
j
s

sec/m
Clay 0.20 0.65
Silty Clay 0.17 0.55
Clayey Silt 0.16 0.50
Clayey Sand 0.11 0.35
Silty Sand 0.10 0.30
Sand 0.05 0.15
17


A weighted single j
s
value was calculated by multiplying a layers appropriate j
s
value by its
percent contribution to the piles total side friction capacity. The summation of the resulting
partial j
s
values yields the correct weighted average used for calculation. Toe damping, j
s
, was
always set at 0.1sec/ft, as recommended by the GRLWEAP manual.

Shaft quake (q
s
) was always set to 0.1in for all soil/pile types as recommended by the
GRLWEAP manual. The toe quake, q
t
, was calculated based on the predicted soil strength of the
end bearing soil layer. GRLWEAP recommends non-displacement piles in all soil types utilize a
toe quake value of 0.1 in. Displacement pile recommendations use a toe quake value equaling
the pile diameter divided by 120 for dense soil and pile diameter divided by 60 for soft soil.
No further information was available for assessing the break point between soft and dense soil
strengths. The FHWA reference table (Sabatini, et al. 2002) was utilized in development of the
subsurface profile to set the criteria for the different densities. The breakpoint for cohesionless
soils was selected to be friction angles of 34, while cohesive soils utilized undrained shear
strength of 1500 psf to classify a soil as either dense (D/120) or loose (D/60). No linear
interpolation was performed on the toe quake parameter based on the degree of soil density.
The last input requires the User to define the expected percent allocation of total capacity
between a piles side and toe. The DRIVEN calculated static capacity allocation was always
used. DRIVENs enforced end bearing capacity cap has the potential to alter this distribution.
However, the majority of piles analyzed in the database were frictional piles; therefore, the data
were not skewed by the limitation.
As stated in Section 2.1 in all cases the recommendations of both the DRIVEN software and
GRLWEAP software default parameters were selected to conduct the analysis, unless field
records provided in the database showed a different parameter, e.g. thickness and type of
hammer cushion. In summary these GRLWEAP defaults were:
- The detailed soil resistance distribution option was used and imported from DRIVEN
(see Figure 2.1) as well as the percent of capacity from side shear from the Drive
option only.
- Unless indicated otherwise, all default hammer efficiencies and accessories were used. If
multiple hammer cushion leads were indicated in GRLWEAP then a lead 4 in to 6 in
larger than the pile diameter was used. If no hammer cushion parameter were available
in the library a cushion was selected from the closest matching hammer ram weight was
used.
- Following the Phase 1 recommendations of Mr. Robert Miner, a more detailed
breakdown of side damping values was used for intermediate soil types between clays
and sands as shown in Table 2.1. A single side damping value was determined based on
each layers percentage contribution to static friction capacity.
18
- The toe quake break point in moving from Dia/60 to Dia/120 was not defined by any
manual. This was taken as 34 in cohesionless soils and 1500 psf in cohesive soils.

19
- Driving strength Loss Factors were left as 0.0 and no Set-Up Gain/Loss Factors were
ever used. All recommendations contained in DRIVEN and the FHWA pile manuals for
pile plugging of H and pipe piles were followed.

20


3.0 PSU DATABASE DEVELOPMENT
The quantity and quality of information required for GRLWEAP pile analysis was considerable.
The first key task for the research team was to search and build a complete and full case history
detail database. ODOT offered two databases as the primary sources to begin gathering
available pile case histories and build a master database: PDLT2000 and Deep Foundation Load
Test Database (DFLTD). An Excel

spreadsheet version of PDLT2000 was created by WSDOT

for LRFD calibration of the Modified Gates equation resistance factor and supplied to the
research team. DFLTD compiled by FHWA was supplied by ODOT to the research team. (For
clarity, any database names appear throughout this report in italics.) Two other large databases,
called FL database and FHWA database, were secured and, together with DFLTD and
PDLT2000, enabled researchers to build a quality, purpose designed, case history set of pile data.
The additional databases helped ensure full confidence for this calibration effort by allowing
cross checks for errors, including additional cases histories, and resolving anomalies. For
consistency, names of the source database were not altered or renamed. The new database
constructed for this research effort, called Full PSU Master database is described in Section 3.2.
3.1 SOURCES
3.1.1 The PDLT2000
The PDLT2000 database was compiled from approximately 77 different project sites by
Paikowsky (1994) and was used for the LRFD performance evaluation of the dynamic methods
contained in NCHRP 507. The database contained 389 driven pile cases, which appeared to be
sorted from data provided mostly by Pile Dynamics, Inc. The database included some or all of
the following pile case history data pertinent for this study:
- geographical location by state, province and country, and site reference numbers,
- pile type, size, length and penetration depth,
- soil type along the side or tip of the pile,
- hammer type, rated and delivered hammer energy,
- measured blow counts at EOID, BOR, or EOR,
- soil quake and damping parameters on pile tip and side,
- CAPWAP summary analysis data, and/or
21
- pile capacities, such as Davissons Criteria or DeBeer, determined by the static load test
results.


Of the 389 case histories, the database had a total 210 driven pile cases with separated driving
times. Among these were 83 pile cases having data at the time of EOID and BOR, and 73 cases
with BOR only, offering a total of 156 pile qualified cases known to be restruck. These are
shown in summary form breakdown in Table 3.1 below. However, complete soil information for
each layer, pile driving logs, the time delay from EOID to BOR, or the static load test complete
result, did not exist in the database for all of the 156 piles. These were significant ommissions
when attempting DRIVEN and GRLWEAP for capacity and which required a considerable
search effort to secure the missing data.

Table 3.1: Pile case histories in PDLT2000 by pile type and driving time
Number of Tested Piles
Pile Type # Driving Time #
Pipe Pile 71 EOID only 54
H-Pile 37
Con'c Pile 98 BOR only 73
Timber 2
Monotube 2 EOID & BOR 83
Total 210 Total 210

The Full PSU Master database (Section 3.2) included 156 qualified driven pile case histories
from the PDLT2000 database shown in Table 3.1. These 156 piles include a subset of 91 piles
matched to the 99 piles reported in the NCHRP 507 study.
For DRIVEN and GRLWEAP software application, field subsurface profiles and shear strengths
were required; however, the PDLT2000 included only soil type information along the shaft and
at the toe, with no indication of layering or soil properties. There were some parameters for
which the PDLT2000 was found to be consistently erroneous, or inexact, when compared to the
values recorded for the same pile in other databases and from other reports and articles. These
included the pile blow count, modulus of elasticity, pile length, and the penetration depth.
3.1.2 The FHWA DFLTD Database
The DFLTD developed by the FHWA (Raghavendra, et al. 2001) consisted of over 1000 load
tests on driven and drilled deep foundations gathered and updated over a 15-year period with
soil profiles and detailed foundation test data. Updates appeared to have stopped at, or about, the
year 2000. The database did have a useful function for user-interface query applications, which
made it easy for a User download of all of the pertinent information, e.g., location, pile and
hammer details, driving logs, load-test, and soil-test data. Futhermore, the database included
several analysis methods to predict the static bearing capacity with Alpha, Beta, Nordlund,
Gates, and Coyles static capacity methods, and the load test failure defined from Hansen,
Davissons criteria, DeBeer, and Max Curvature interpretation methods.
22


The information offered in this database supplemented very well the information provided in
PDLT2000. However, after securing other materials, DFLTD was also shown to have anomalies,
which were likely caused by simple input key-stroke mistakes during the compilation of the
source data from the large number of project sites and state DOTs. One limitation of the
database was that pile driving records showed the blow counts by depth, but had no calendar
and/or clock time delay shown of the driving log for the restrike or other driving interuptions.
When generating the Full PSU Master database, the blow counts were checked, where possible,
with other databases, e.g. FL database, FHWA database, and are introduced below. DFLTD
also provided additional case histories not contained in PDLT2000.
3.1.3 Other Databases
With the assistance of Florida DOT, a load-test database established in 1994, in the LOTUS-
123

format, was converted to Excel

spreadsheet format and was called the FL database. The

database contained 72 driven pile load-tests gathered from 1985 to 1991 in Florida and 120
drilled shaft data from 1962 to 1989 at approximately 43 different project sites. It contained pile
and hammer information, driving blows, load-settlement results, insitu test results, and the
predicted capacity from the SPT91

program for the driven piles. Most information in the data

seemed to have been used to build the Florida entries in DFLTD or to have the same background
of data since all the information shown in both was similar.
The University of Florida provided an additional Florida database. The Deep Foundations
Database was a product of the University of Florida and the Florida DOT. Prof. McVay of the
University of Florida sent a copy of this database to the research group. This was a Microsoft
Access

document that included: pile type and dimensions, boring log data, driving log, and pile
load test results for a total 627 driven piles and drilled shafts. Not all of the driven pile case
histories were included in the PDLT2000 or the DFLTD, but most of the piles in the Deep
Foundations Database with restrike and load test information were included in these two
databases. However, three new pile case histories were located in this database and entered in
the Full PSU Master. Additionally, several case histories with anomalies and missing data were
resolved by locating their records in the Deep Foundations Database.
23
From the cooperation of Prof. James Long at the University of Illinois, a database reporting
almost 200 pile load test details, called the FHWA Database, was provided to the research
group. Among the 200 pile cases, 99 pile case histories met requirements and were used for the
wave equation calculations contained in the earlier FHWA effort at determining pile driveability
and capacity (Rausche 1997). Included in these requirements in this earlier study were that blow
counts at BOR should be less than 30 BPI and comparable time delay between EOID and BOR
should be found between EOID and the static load test. The FHWA database showed a wide
variety of information about the driven pile load tests, except the subsurface information for each
soil layers. This FHWA database consisted of data in five areas: pile details, pile location,
hammer, hammer ram and helmet weight, and penetration depth at the EOID and the BOR. Soil
layer details from the FHWA database could be crosschecked with the DFLTD data by matching
the static capacity and the dynamic capacity test results. A total of 25 pile cases could be
identified that were not in the PDLT2000 database, but they lacked water table information.
Further, among the 200 data in the FHWA database, aside from the pile cases shown in

PDLT2000 and DFLTD, 21 additional piles showed up as new cases. These pile cases had almost
all the field test data, but no soil information of any kind. Unfortunately, this research effort was
unable to locate the additional spreedsheets cited by the FHWA report (Rausche 1997) and
therefore, these 21 additional piles could not be analysed.
Prof. Roy Olsen of the University of Texas at Austin supplied the research group the results of
large numbers of California load tests conducted by Caltrans in a Microsoft Access

document.
Each pile case history included project name and location data, site investigation and laboratory
testing results, and static load test capacity. The database included 28 driven piles that were
restruck and load-tested, but no blow counts at EOID or BOR were included in this document. A
limited number of the piles were identified in the Full PSU Master database from previous
sources; but the majority was inducted into the Full PSU Master as new pile case histories.
3.1.4 Other Sources
Additional pile case histories were found by an extensive review of recent geotechnical literature
and requests to various DOTs for new pile case histories that were not present in any database
previously reviewed by the research team. Considerable effort was spent in identifying new
complete case histories, which required as a minimum for this research:
- A soil profile including a basis to determine shear strength from either in situ or
laboratory testing,
- Pile type, size, and length,
- Full driving hammer and accessory details,
- Field blow counts at EOID and BOR, and
- Davissons interpretation of pile capacity from the static load test.
Significant additional case histories, which qualified for analysis, were found in the states of
Texas, South Carolina, Michigan, Utah, and Massachusetts. Other states providing incomplete
case history documentation that subsequently could not be analyzed were North Carolina and
Louisiana. In general, any private consultant and non-state transportation agency published case
history rarely contained sufficient information for full analysis, but the data were included in the
Full PSU Master.
3.2 FULL PSU MASTER DATABASE
24
The research group created two Microsoft Excel

spreadsheets containing separate tabs for the

DRIVEN input, GRLWEAP input, Summary, Output, and Notes and References. Each of the
fully qualified case histories was analyzed using DRIVEN and GRLWEAP and the results
summarized for the purposes of statistical calibration of the resistance factor for EOID and BOR.
The two created spreadsheets were the PSU PDLT2000 Master database and the Full PSU
Master database. The PSU PDLT2000 Master database contained the 156 driven pile case
histories extracted from the PDLT2000 database supplemented by additional details from the
DFLTD. The Full PSU Master reached a total sum of 322 driven piles from a number of

different sources identified above and included all the PSU PDLT2000 Master cases. PDLT2000
and DFLTD cases contributed over 50% of the total number of case histories finally entered into
the master. A breakdown of all sources included in the Full PSU Master database is shown in
Table 3.2. There was considerable overlap between the numbers of pile case histories because
data were tracked to more than one source, i.e. the total sum of the case histories in Table 3.2 is
greater than the total number of case histories present in Full PSU Master.

Table 3.2: Source of data for pile case histories for resolution of errors and anomalies
Source of Pile Case History Pile Case Histories in Full PSU Master
PD/LT2000 156
DFLTD
102
Prof. James Long
28
Data sent by state DOT
18
Data for state DOT project, but not sent by DOT
61
Scholarly articles
60
TOTAL represents overlap between sources
425

3.2.1 Input Quality Tier
Allen et al. make clear (Allen, et al. 2005) that for any quality LRFD calibration, the statistical
quantity and quality of case history data must be assessed and reported. In this study, the
assumptions made to study effects of unresolved anomalies were expected to affect the quality
and confidence of the case history capacity prediction. These may, if not tracked and assessed,
erode the quality of the calibration resistance factor by increasing statistical coefficients of
variation. For this reason, each pile case history included in Full PSU Master database was
assigned an input tier number that described the level of reliance on assumptions in input to
analyze the case history in both the DRIVEN and GRLWEAP software. Input tiers ranged in
value from 1 to 3, where a pile case history with the most complete set of input parameters was
assigned to Tier 1 and a pile case history with the most incomplete set of input parameters that
could not be analyzed without further key information (e.g. pile blow count) was assigned to
Tier 3. Each of the Tiers 1 and 2 had sub-categories with more specific requirements, as
described in the following sections. The input tier classification of a pile was related implicitly
to the quality and completeness of the data in the pile case history, which changed according to
new information obtained during the course of the research.
A summary of the tier rubric developed is presented in Table 3.3. If anomalies were discovered
during initial entry of a case history into the Full PSU Master, their severity would be reflected
in the initial tier assigned. Once flagged, efforts were made to resolve the error.
25
Pile case histories placed in input Tier 1 could be analyzed in both DRIVEN and GRLWEAP
with no assumptions and were broken down into three sub-categories: 1a, 1b, and 1c. Piles
placed in Tier 1a had full soil data, including measured internal angles of friction and/or
laboratory shear strengths of cohesive soils, the hammer type and driving accessory details were
known, and the pile information was known including composition, size, and driving state

(plugged or unplugged). In addition to this, Tier 1a piles had Davissons capacity and
PDA/CAPWAP capacity available with no anomalies reported in either the bore log or driving
log that might significantly affect prediction. The number of BPI at EOID or BOR for Tier1a
had to be less than 10.

Table 3.3: Summary of the input tier rubric utilized for analysis
Tier Definition
a
Full soil data is reported, including measured internal angles of friction and/or laboratory shear
strengths of cohesive soils. Hammer type and driving accessory details are known. All pile
information is known including composition, size, and driving state (plugged or unplugged).
Davissons capacity and PDA/CAPWAP capacity is available (possibly reported in DFLTD). No
anomalies are reported in either the bore log or driving log that might significantly affect prediction.
b
All criteria is met for Tier 1a except no PDA/CAPWAP information is reported, or pile is driving
harder and 10 BPI < N < 15 BPI
1
c
Relaxation detected with blow count EOID > BOR and should be confirmed, or
Very hard driving with N >15 BPI.
a
Typically DRIVEN can be performed; however, GRLWEAP cannot be routinely performed but is
attempted with assumptions. All pile and hammer information and details are known, but there may be
some anomalies to resolve, data missing, or anomalies in EOID or BOR blow counts. Some soil
strength properties with most other key properties known. 2
b
Typically neither DRIVEN nor GRLWEAP can be routinely performed. All criteria are met for Tier
2a, except assumptions on cohesive soil shear strength values are required, or there is a lack of water
table information for granular soil.
3
No soil data or hammer information known. Non-typical soil types, e.g. sandstone. No field blow
counts or load test results are available.

Dynamic testing to determine the ultimate static pile capacity requires that the driving system
should mobilize most of the available soil resistance acting on the pile. However, when pile
penetration resistances approach 10 BPI, the soil resistance may not be fully mobilized at, or
near, the pile toe. Blow count effects are discussed in Section 6.2. In these circumstances,
FHWA reports that dynamic test capacities tend to produce lower bound capacity estimates
(Hannigan, et al. 2006). Pile displacement during dynamic testing must be large enough to fully
mobilize the side shear and end bearing components to determine reasonable ultimate pile
bearing capacity. However, matching to Davissons capacity may not increase bias to the
conservative side necessarily as Davissons is a restricted settlement based criteria. Piles in Tier
1b met all of Tier 1a requirements but had either no PDA/CAPWAP information or the pile was
driving harder so that the number of blow counts at EOID or BOR ranged from 10 to 15. Tier 1c
piles were either very hard driving, with the number of blow counts greater than 15 at EOID or
BOR, or relaxation was detected such that EOID blow counts were greater than those at the BOR
condition.
26
Piles in Input Tier 2 could only be analyzed with certain assumptions and were broken down into
subcategories, 2a and 2b. Piles in input Tier 2a could typically be used in DRIVEN without
assumptions but needed to follow key assumptions to be used in GRLWEAP. Input Tier 2a piles
were those cases for which all pile and hammer information and details were known, but they
may have had some anomalies to resolve such as the BOR and/or EOID blow counts. Other
small anomalies included moderate assumptions to analyze soil profile, e.g. extrapolation of soil

profile from the existing bore log data to a depth past the piles tip, or poor correlation between
the bore logs SPT count and driving logs installation record.
Tier 2b piles could not be performed in either DRIVEN or GRLWEAP without key assumptions,
e.g. assumptions on cohesive soil shear strength values (values were based on CPT or SPT data,
unless layer was not significant to bearing capacity), or for granular soil water table information.
Pile case histories in Input Tier 3 could not be analyzed by DRIVEN or GRLWEAP due to
missing or fatally anomalous input parameters. For the most part, piles in Input Tier 3 were
missing soil and hammer data. Some pile case histories were assigned to Input Tier 3 due to a
lack of blow count data at either EOID or BOR despite a complete set of values in other
parameters.
Missing data and data anomalies were continuously resolved for the Full PSU Master entries as
new information became available. The assigned input tier for a specific pile could change over
time often improving its ranking with the addition of data that allowed the pile to be analyzed
with fewer assumptions. The quality improvement of the data set as a whole could be observed
and shared by the research team by observing the number of piles in each tier. The change in the
number of piles in each tier over the course of the research task can be seen in Figure 3.1 for the
Full PSU Master piles throughout 2009 and early 2010 when a significant effort at data
gathering took place.


27
Figure 3.1: Growth of cases in the Full PSU Master by input tier

As the figure illustrates, in addition to the total number of pile case histories in the Full PSU
Master database increasing, the number of piles in the upper two tiers steadily increased over the
course of this 15 month period. This trend of increasing quality of input parameters was a
product both of the deliberate search for new pile case histories with high-quality input data and
the ongoing endeavor to resolve anomalies and locate missing data. Input tiers were a means of
organizing data by level of reliance or degree of completion for input parameters. Following
analysis, output results were also organized by level of confidence in prediction including
reliance on the data. These output ranks are discussed in Section 4.3. Both input tiers and output
ranks are recorded in the Full PSU Master spreadsheet.
3.2.2 Case Histories Overview
The construction of the Full PSU Master dataset was to ensure a reasonable match to piles by
their pile/soil breakdown to those driven for ODOT bridges. The sources of the pile case
histories included in the Full PSU Master spreadsheet were, however, unevenly distributed
throughout the U.S. and other countries. The large majority of cases were within the U.S., but
there were also a number of cases from Canada. A breakdown by percent is given in Table 3.4
and Figure 3.2 below.
Table 3.4: Geographic distribution of Full PSU Master pile case histories
Location
Number of
Pile Case Histories
Percentage of Total
U.S. 267 83%
Canada 29 9%
Unknown 9 2.8%
All other countries 18 5.6%


28
Figure 3.2: Distribution of Full PSU Master pile case histories worldwide


Figure 3.3: Number of pile case histories in Full PSU Master by state
Piles from the U.S. formed the vast majority of case histories. Figure 3.3 and Figure 3.4 break
down the number of case histories per state and by the five FHWA regions. Clearly, the
Northwest was under-represented. It should be noted that due to the FHWA regional division of
the country, there were a considerable number of piles in California that were listed under the
Southwest Region.

29
Figure 3.4: Source of Full PSU Master pile case histories by region

Approximately one-third of the original 156 piles identified in PDLT2000 had insufficient data
for satisfactory DRIVEN and GRLWEAP analysis and were in Tier 3. It should be noted that of
the 322 piles in the Full PSU Master database, approximately one-third were also in Tier 3.
No useful soil profiles to obtain properties were contained in PDLT2000. Therefore, heavy
reliance was placed on DFLTD and additional databases from the Universities of Illinois and
Florida, Florida Department of Transportation (FDOT), and some published literature to secure
SPT data and, occasionally, CPT data. As introduced in Section 2.1, three general subsurface
soil types were used for categorization: Clay, Sand and Mixed. Soil types, broken down by both
pile type and soil category, were assigned to the case history for ease of organization using an
80% threshold contribution from the DRIVEN calculated capacity and are shown in Table 3.5.

Table 3.5: Breakdown of all 322 piles in the Full PSU Master by pile and soil type
Pile Type
Major Contributing
Soil Type
Concrete
Pile
H-Pile
Closed End
Pipe Pile
Open Ended
Pipe Pile
Other
Total Cases for
Soil Type
Sand 62 19 17 4 1 103
Clay 17 5 10 1 0 33
Mix 14 9 16 5 1 45
Unknown 54 24 38 20 5 141
Pile Totals 147 57 81 30 7
Total Piles
322

The Full PSU Master database contained 81 cases where the calendar and clock time were
known at the time of EOID and at the BOR, as well as the time for static load test. This
knowledge permitted an examination of cases where restrike or load testing may have occurred
either comparatively early or late for representative set-up to occur. A total of 84 piles reported
multiple restrike blows, but few qualified for DRIVEN and GRLWEAP analysis.
Cross examination of DFLTD and PDLT2000 showed a total of 72 of the original 156 piles in
PDLT2000 had anomalies, and 29 piles in PDLT2000 had no site I.D for any follow-up
investigation. Of most concern was the misreporting on pile blow count, N, especially at BOR.
A total of 28 piles had more than one anomaly. After resolution of errors and anomalies, 103 of
the 156 PDLT2000 entries were analyzed, and these are discussed in the following section.
Based on thorough examination, the most reliable source, judged by multiple source data
confirmation, was from DFLTD. Case history summary details for each case history contained
in the Full PSU Master are given in Appendix A.
3.3 CASE HISTORY ANOMALIES AND ERRORS
30
It was essential that the most accurate data were used for reliable calibration, as well as the best
possible case history population match. The majority of pile case histories in the Full PSU
Master had some identification and some limited geographical location data, which allowed for

matching pile case history between the two principal sources of data: PDLT2000 and DFLTD.
The type and dimensions of the pile, blow counts at EOID and BOR, and other pile
characteristics were available for most piles in both databases. Soil data in the PDLT2000 were
confined to an assigned value of soil type along the shaft and soil type at the toe with no values
for strength, unit weight, or water table. Therefore, additional soil data from other sources was
critical for use of DRIVEN. In cases where piles from the PDLT2000 were matched with piles
in the DFLTD, soil data were obtained from DFLTD by matching a pile to a specific soil profile.
DFLTD often contained sets of site investigation data on multiple boreholes when determining
the applicable soil profile to the piles identity. Matching soil data in the Full PSU Master to a
pile case history in the DFLTD was achieved by comparing the driving log for the pile and the
log of the SPT blow counts. The profile of SPT values in the boring log most applicable to the
pile correlated quite well to the blow counts in the piles driving log. For example, areas of hard
driving were seen to match with depth (or roughly with depth in the case of slightly different
elevations) between the boring log and the driving log indicating that the soil properties were
similar and could be modeled in DRIVEN. If possible, the missing data for each case history
were found in other databases, such as FHWA database, or even from original site documents
obtained by a thorough search of geotechnical literature, private companies, and state or local
agencies. Unfortunately, not all pile case histories from the PDLT2000 included in the Full PSU
Master could be matched to piles in the DFLTD.
There were several piles in the Full PSU Master data set for which each database reported a
different value for a specific parameter. For example, the PDLT2000 reported a blow count of 7
BPI at BOR for a pile, while the DFLTD and source drive log reported 16 BPI on the same fully
matched pile. For the most part, the majority of these anomalies were present in the
PDLT2000-based cases of the Full PSU Master database as the pile case histories added to this
original data set were less likely to have more than one source from which a disagreement in
data values could result. It was often the case that a pile case history had more than one
anomaly. The percentage of piles with anomalies and the breakdown by anomaly types for the
PSU PDLT2000 Master database is shown below in Table 3.6 and Figure 3.5. The nature of an
anomaly varied, though most of them resulted from a difference in blow count values from one
or more sources.

Table 3.6: Number of piles and anomalies found in the 156 from PSU PDLT2000 Master
Pile Type
Major Contributing
Soil Type
Concrete
Pile
H-Pile
Closed End
Pipe Pile
Open Ended
Pipe Pile
Total Cases for
Soil Type
Sand 50 14 12 0 76
Clay 10 2 9 1 22
Mix 12 5 13 4 34
Pile Totals 72 21 34 5 132
31



Figure 3.5: Source of anomalies between databases for Full PSU Master
Missing and anomalous data were first cross-checked with available databases. If the anomaly
or missing data were still unresolved but sufficient site and location data were available to find
the original project with the appropriate state or local agency, then direct inquiries were made.
Some data could be obtained by using sources not associated with the pile itself. For instance,
values such as the location of a water table or a range of typical cohesion values in local clays
was obtained by searching the United States Geological Survey (USGS) data or directly from
local geotechnical practitioners familiar with the piles locale.
The structured approach for cross checking was the result of considerable planning and
discussion, and all changes for anomaly resolution were carefully documented. For a pile case
history with more than one source for a disputed value, if all sources were relatively equal in
quality, then the value found in more than one source was determined to be more accurate. Two
sources reporting the same data value were taken as more reliable than one other source
reporting a different value. When more than two sources for a parameter were available, it was
seldom that all of them disagreed on an anomalous value; usually two or more shared a value.
There were instances in which all the available sources disagreed on the value of a parameter. In
these cases, a thorough familiarity with encountering anomalies and cross-checking them with
other sources led to a consistent approach to which source was most likely to be correct.
32
A reported value from the DFLTD was preferred over the PDLT2000 value, all other
circumstances being equal. This was due to the numerous times in which a value from the
PDLT2000 had disagreed with a value from the DFLTD or had been proven erroneous by the
consistent approach of matching the pile record to another database or even original documents
such as driving logs. The PDLT2000 contained anomalous values such as pile length,
penetration depth, and pile elasticity modulus. Matching values in order to resolve anomalies
also revealed many cases in which blow counts were not accurately reported for EOID and BOR
in the PDLT2000. In cases where the DFLTD value appeared to be erroneous or when another
source disagreed with the DFLTD but agreed with the PDLT2000 database, the PDLT2000 value
was used. Of the 74 PDLT2000 matched piles in DFLTD, 18 had anomalies from the states of

AL, FL, LA, SC, and WI. Of these states, the highest percentage of any state contribution to
PDLT2000 with anomalies was in SC with 30%.
For parameters in which several had missing or anomalous data, a sensitivity analysis was
performed in order to determine the criticality of the value. Sensitivity analyses were a critical
portion of the research effort to establish that assumptions made would not have statistically
significant effect on results. If the value in question was found to produce less than ten percent
difference between the GRLWEAP reported capacity values, then the analysis was considered
relatively insensitive to the parameter. It was determined that in the majority of cases, the
calculated GRLWEAP capacity was largely insensitive to values for the water table. However,
GRLWEAP was quite sensitive to values for hammer energy and driving accessories. Therefore,
the resolution of missing hammer data was given more priority than that of missing water table
data. The input tier value for a pile with missing hammer data was lower than for a pile with
missing water table data (see Section 4.3).
In summary, of all AASHTO declared nominal capacity methods, the GRLWEAP bearing graph
analysis was the most demanding for input data quality and quantity. Before calibration efforts
began, any existing databases used were carefully reviewed and data were validated.
3.4 SUBSURFACE INFORMATION, QUALITY, AND
INTERPRETATION
The Full PSU Master database was created from several different source databases and other
reports and articles; often data sets collected for research had a variety of different research foci
and were not designed to perform GRLWEAP analysis. The quality and quantity of site
investigation data varied across the pile case histories in the final database. This was expected
given that these piles ranged in time over the past thirty or more years and across a vast
geographical area. The original purpose of the PDLT2000 was for the capacity prediction of
driven piles as measured by PDA dynamic methods and did not include extensive analysis and
discussion of the importance of soil types and specific characteristics. Therefore, soil data from
the PDLT2000 was extremely limited, and little attention was paid to obtaining additional soil
data other than that provided by the sources of this database. Piles in the PDLT2000 not
identified in the DFLTD did not have sufficient soil data for DRIVEN analysis. This scenario
was also true for piles from other databases, which included little or no soil data. As a result, the
soil investigation quality of some cases in the Full PSU Master database was limited, so
sensitivity analyses were conducted.
33
The DRIVEN program required a relatively modest amount of soil information, provided by
most geotechnical explorations, including water table, soil unit weight, and strength parameters
for either cohesive or cohesionless soils. In order to perform DRIVEN analysis, prior to final
analysis in GRLWEAP, a certain minimum amount of soil information was required. For those
pile case histories with soil profile data available, there were key parameters either absent or for
which certain assumptions were still necessary.

34
For the majority of pile case histories, even when no other soil data were available, the soil type
was available. Unfortunately, pile capacity could not be established by designating soil as
sand or clay in DRIVEN. Additional soil documentation was necessary. When more
specific data could not be secured, assumed values were used based on similar cases or from
typical values. Typical values for soil unit weight from standard reference texts (Coduto 2001)
were assumed applicable for certain soil layers in which a specific value was unavailable.
However, as with other soil data, strength parameters for each pile case history were not always
available. When present, they consisted predominantly of results from the SPT and, in some
cases, the CPT. Laboratory data for soil strength parameters, such as Triaxial test results, were
rarely present in the DFLTD or any of the other source databases and seldom reported in other
sources of pile case histories. However, on some occasions, results from Atterberg limits testing
were available from the DFLTD. In summary, soil strength parameters for the majority of the
piles in the Full PSU Master were derived from in situ test results in the absence of laboratory
testing results.
The applicability of specific in situ test results to a piles soil profile was hard to assess. The
distance from a site investigation borehole to the pile was not included in any of the databases,
and was rarely mentioned or included in other sources. For example, the DFLTD soil
investigation data was matched to the piles driving log based on SPT data, but the actual
distance of the referenced boring log to the pile was unknown. In addition to distance concerns,
both the SPT and the CPT had shortcomings when used to calculate soil strength parameters, and
neither test performed equally well for all soil types. For the majority of pile case histories, the
only soil data available came from one or both of these two in situ tests. The majority of the in
situ soil strength values were in the form of SPT blows per foot. These blow counts were
uncorrected for energy or for overburden effects in the DFLTD. In other sources, the nature of
the N value was often unspecified, in which case they were assumed uncorrected.
For some of the pile case histories, CPT data were provided and often reported in DFLTD in
addition to SPT data. DFLTD-provided CPT data also included sufficient data to obtain
undrained shear strength in cohesive layers. These data were included in the Full PSU Master,
and were preferred to obtaining cohesive strength values using the SPT (not endorsed by
FHWA). It should be noted that the DFLTDs CPT analyses occasionally provided soil strength
values that were high to the point of being erroneous in comparison with the soils density and
consistency according to SPT results. In these cases, the cohesive soil strength was derived from
SPT data using the method described previously. In some cases CPT data were provided in the
absence of any other soil data, and a consistent approach was taken to deriving soil strength
parameters.


4.0 BEARING CAPACITY METHODOLOGY
4.1 EOID AND BOR MODELING AND BLOW COUNTS
For the complete pile case history sets found in the Full PSU Master database input Tiers 1 and
2, GRLWEAP analyses provided a bearing graph as discussed in Section 1.2.1. Analyses
required careful evaluation and separate analysis for a total of double the number of qualified
cases, as each required both the EOID and BOR capacity predictions. As well as changes in blow
count causing capacity changes viewed from the same bearing graph, different graphs for EOID
and BOR can result from changes to hammer models and/or substitution of a well-used stiffer
cushion on concrete piles, as endorsed by the FHWA. Default GRLWEAP parameters called for
a plywood pile cushion of 6 thickness with an elastic modulus of 30 ksi at EOID on concrete
piles. Due to driving effects, BOR analysis on concrete piles always followed the rule-of-thumb
discussed in FHWA (Hannigan, et al. 2006) that a used pile cushion was inserted of usually half
the thickness and twice the stiffness relative to its EOID properties. The BOR pile cushions
were assumed to be 3 thick plywood cushion with an elastic modulus of 60 ksi. Stiffer cushions
will allow a larger amount of driving energy to be delivered to the pile and will lower the blow
count required to overcome the same static capacity. It is more likely on BOR that a stiffer
cushion will show a similar or higher blow count compared to EOID indicating pile set-up.
Further, a reduction in observed blow count at BOR, for the same hammer, may not necessarily
indicate relaxation. This difference in bearing graphs and the relationship between blow count
and blows for EOID and BOR is illustrated in Figure 4.1 for one of the prestressed concrete PSC
piles in the Full PSU Master: a 16.5 octagonal prestressed concrete pile.


35
Figure 4.1: PSU Case History #293 - effects to the bearing graph of stiffer restrike cushion

Comprehensive discussion on the User input requirements to GRLWEAP were discussed in
Section 2.2 covering pile, hammer, driving accessories and the soil model. In summary, all EOID
and BOR modeling used only default parameters for the hammer, accessories, and the soil
distribution given by DRIVEN.
The recalibration effort utilized the ratio between Davissons criteria measured load test capacity
and the corresponding GRLWEAP capacity prediction at both EOID and BOR condition with:
(4-1)

So, expresses the GRLWEAP capacity of each case history as either a safe or unsafe
prediction. Clearly, a case history value at EOID or BOR that is less than 1 is unconservative,
while a value greater than 1 is a conservative prediction compared to the Davisson measurement.
The average from a sample of case histories is called the bias and has previously been
designated
R
for resistance factor calculation, but hereafter will be termed mean for the full
sample statistical mean ratio of bias.
4.2 OUTPUT RANKING
In accordance with the quality metric guidelines suggested by AASHTO (Allen, et al. 2005), the
GRLWEAP output was assigned a qualitative output rank ranging from 1 to 4 designed by the
research group to record confidence levels in results. Ranking of the outputs allowed researchers
to accomplish the following: summarize the analysis process, tag critical assumptions, and flag
problematic cases for further examination including discussion with selected members of the
TAC and conduct possible re-analysis. Table 4.1 summarizes the output ranking designed for
this study.

Table 4.1: Summary of output ranking requirements
Rank Definition
1
Good
No key assumptions were required for analysis and no anomalies present in output. Typically in
soft soils GRLWEAP capacity approximately equals DRIVEN capacity, and for harder soils
GRLWEAP capacity is less than DRIVEN.
2
Acceptable
Possible minor anomalies present, or non critical assumptions made, e.g. hammer type not clear
but matched from library, assumed water table in sand, GRLWEAP capacity greater than
CAPWAP. Davissons capacity is significantly greater than the DRIVEN capacity.
(Disagreement to Davissons is possible in end bearing piles or hard driving which may not be
mobilizing significant side friction.)
3
Poor
No hammer cushion match. Significant anomalies. Pile showed relaxation on BOR. Bearing
graph overly sensitive to damping and/or side friction to total capacity percent distribution.
4
Ineligible
Lacking information to become a statistical , e.g. lacking Davissons capacity or field blow
count.

Placing pile cases into both input tiers (Section 3.2.1) and output ranks enabled researchers to
target and troubleshoot problematic cases and address any trends that showed during statistical
analysis.
36

4.3 GENERAL SENSITIVITY COMMENTS
As discussed previously, GRLWEAP is a finite difference based pile driving modeling program
that reports permanent pile set in units of length, and calculates from the set inverse the pile
advancement rate in BPI. Any pile placed in Tier 1 or Tier 2 had EOID and BOR blow count
values known, and the GRLWEAP bearing graph established predicted equivalent static
capacity. The stepping up of static capacity by GRLWEAP, with the bearing graph option,
produced the graph and caused the output to be relatively insensitive to minor alterations in soil
properties and variations of pile length. Holding the same soil type constant would require
significant differences to the center of soil resistance depth and/or the percent allocated to side
shear to modify the predicted capacity. However, sensitivity analyses were always performed
when assumptions were required. A threshold of 10% difference in predicted capacity values
was used to separate those cases overly sensitive to assumptions from those that were not overly
sensitive. Site investigations should always comply with the minimums recommended by
FHWA (Hannigan, et al. 2006) to produce satisfactory GRLWEAP capacities. The statistics
reported in Section 5 show a correlation between input quality of the source data and output
quality confidence of the predictions. Sensitivity analyses were always used to determine what
the affect of a missing piece of input information had on a cases calculated GRLWEAP
capacity.
Values for the location of the water table were also assumed for both cohesive and cohesionless
soils based on the relative insensitivity of pile capacities calculated using GRLWEAP to the state
of the soil as saturated or unsaturated. The resulting sensitivity was determined for each pile,
when necessary, on a case-by-case basis. Though DRIVEN can be insensitive to the water table
location and soil unit weight, the calculation of pile capacity in the program was strongly
dependent on the input soil strength parameters. The general trends found were the following:
- Water table location only mildly affected results in cohesionless soils.
- Shear strength variations showed little effect on capacity as the percent allocated to side
shear and the shape of the distribution may not change.
- GRLWEAP capacity proved to be relatively insensitive to slight variations in pile tip
depth.
37
The trends discovered during the analytical process were consistent with those reported in Phase
1 research effort (Jackson 2007). The ability to study by sensitivity analysis the effect of
assumptions on a case-by-case basis enabled researchers to promote more cases to a higher
output rank than to demote cases to a lower rank.

4.4 PILE BLOW COUNTS
Of all case history variables, the GRLWEAP nominal bearing capacity was most sensitive to pile
blow count, which was the easiest to reliably document. However, there was no consistent
definition of the restrike blow count in any pile national standard or design manual. The source
databases confirmed this lack of clarity. A large variation in installation condition requirements,
calendar times, and field practice norms existed in the database due to the large span in years and
geographic locations from which the cases originated. Fortunately, for the purposes of resistance
factor calibration, the inherent variability of national standards was captured in the Full PSU
Master by keeping a broad database. For the best possible resistance factor accuracy, any known
errors in the restrike blow count were detected and corrected. DFLTD had no separation
between EOID and BOR blow count, but it did present the complete driving record.
Among the 156 piles based on PDLT2000 listed in the Full PSU Master sheet, 74 piles were
matched in the DFLTD database. These offered high quality data confidence when the data
matched exactly in each database. There were some blow count assumptions made to use the 74
piles in Tier 1 and Tier 2 for this calibration analysis. Driving records in DFLTD represented
the blows that increased depth by one-foot intervals up to a maximum integer value with
subsequent values in decimal format. It was assumed the integer values were for initial driving
and the decimal values were for restrike. A particular problem existed with the 82 piles not
found in DFLTD. Some had listed BOR blow count but did not list the EOID blow count in
PDLT2000. The research team worked to independently secure the missing EOID blow count.
38
DFLTD had a group of field entries for hammer and PDA delivered energy as well as entries for
EOID, BOR, and EOR driving blow counts. In all cases the database value showed 0. Without
confirmation for these EOID and BOR field entries, the research group was required to use the
decimal depth assumption discussed above in interpretation of the full driving log. The blows
given by the decimal feet format indicated restrike had begun and represented an equivalent
number of the blows per foot (this was also supported by additional data from Florida), not the
actual blows between the two decimal depths. This is illustrated below in Figure 4.2.


Figure 4.2: Illustration of DFLTD presentation in digital feet for restrike
The penetration resistance at EOID and BOR in BPI was then simply calculated from the BPF.
Following this assumption, the blow count of 32 in the figure was the penetration resistance at
EOID when moving from 89 to 90 ft, and the blow count of 24 was the BOR count in the unit of
BPF (but it travelled only 1 in as shown in Figure 4.2) in the DFLTD database. Those
corresponded respectively to blow counts of 2.67* and 2 BPI in the PDLT2000 where the
database authors asterisk denoted the blow count based on blows per real foot of penetration.
After making the separation between EOID and BOR using the assumption above, only 16 of the
74 matched piles, without other anomalies, showed the blows at EOID and BOR reported in both
databases were exactly the same. Of the remaining matched 58 piles between the two databases:
- 10 piles had the EOID blow count missing in PDLT2000, but the blow count at BOR was
the same in both databases,
- 8 piles had the BOR blows missing in DFLTD,
- 17 piles had one of the blow counts at EOID and BOR different between both databases,
and
- 23 piles had neither of the blow counts for EOID and BOR the same in both databases.

39
This illustrated the difficulty encountered in relying solely on blow counts in a single database
and suggested that other recalibration efforts using this data may contain flaws. For each pile
case matched by any source in DFLTD, blow count was determined by employing the decimal
feet entry assumption described above.

40
In the Full PSU Master dataset only 19 case histories had multiple restrikes and load test clock
times in Tier 1 and Tier 2. These blow count series carried the designation BOR1, BOR2,
BOR3, etc., corresponding to their sequential order and were recorded on the Full PSU Master
Spreadsheet Summary tab. Therefore, without knowing both restrike calendar day and load
test calendar day, no decision could be made on choosing the multiple restrike blow number
matching the load test for the large majority of piles.
In theory, it is always appropriate to leave the longest possible time before restrike to capture
sufficient regain in capacity, but with production scheduling constraints in the field, contractors
will often be reluctant to remobilize a crane and will likely seek restrike close to the minimum
established in the specifications. Currently ODOT allows a minimum of one day. So, the best
population sample fit to practice might be after a first restrike, typically of one to two days only.
It is likely, with the associated repeat rise in excess pore water pressure for cohesive soils, and
large soil fabric changes that piles undergoing multiple restrikes do not belong to the same
GRLWEAP capacity statistical population as those at BOR1. It can also be argued that a pile at
BOR2 time would not have the same driving blow count BPI as if a BOR1 had been conducted
with the same time delay from EOID.
Following TAC discussions and the wide variation in case history locations and dates that load
tests were conducted, the following rules were applied:
- For the 19 cases with multiple restrike values: the first beginning of restrike blow count
series was selected for use in analysis.
- For cases that required driving log interpretations, the guidelines (Hannigan, et al. 2006)
from the FHWA were followed, with a restrike recorded within the first 6 in or a
maximum blow count of 50 BPI, whichever came first.
- Restrike was taken from DFLTD when depth began recording in decimal feet format.
For consistency, when encountering anomalies or confusing data, the GRLWEAP capacity was
established using the above rules for restrike to establish BOR.

5.0 STATISTICAL PROFILING OF RESULTS
5.1 REANALYSES OF NCHRP 507 CASE HISTORY SET
To confirm the validity of the research groups GRLWEAP predictions and to reveal any serious
modeling violations, a comparison was made with the 99 pile case histories reported in NCHRP
507 using the PDLT2000 Master. The NCHRP 507 authors reported that the 99 case histories
for GRLWEAP calibration were inherited from the earlier FHWA study conducted by the
authors of GRLWEAP (Rausche, et al. 1997). These were assumed to contain only well defined
case histories, which all included PDA/CAPWAP work, and would have placed them all in Tier
1 and Tier 2 in this study. The BOR analysis of these 99 piles was contained in the NCHRP 507
report appendix and not discussed in the main text. No description was offered in NCHRP 507 of
any input and output case history quality ranking structure or the specific definition of BOR
used. The research reported here compared the mean standard deviation (s.d.) and coefficient
of variation (COV) utilized in NCHRP 507s calibration of GRLWEAP to those produced in this
study.
Due to the lack of full and complete source data for identification purposes, a small portion of
the original NCHRP 507 cases could not be identified. Therefore, the current research
comparison included a limited number of case history data from outside the original 99 case
history set, but still within the NCHRP 507s PDLT2000 database. The inclusion of these data
allowed for a comparable size population of 91 cases for statistical profile comparison. The Full
PSU Master database for this research grew to 322 cases with 179 cases in Tier 1 and Tier 2
qualified to produce values. A summary of the results from the 99 piles reported in NCHRP
507 and the 91 NCHRP 507 matched cases, along with the full 179 cases, are shown in Table
5.1.

Table 5.1: Comparison between NCHRP 507, PSU matched piles, and all analyzed 179 piles in the Full
PSU Master
EOID BOR
Source
Soil
Type
Pile
Type
#
cases S.D. COV S.D. COV
Original GRL-99 Series as
Reported in NCHRP 507
99 1.656 1.199 0.724 0.939 0.399 0.425
PSU Recalculation of Original
NCHRP 507 Source Data
91 1.661 1.264 0.761 1.009 0.599 0.594
PSU Tier 1 and 2 Source Data
All All
179 1.633 1.27 0.78 0.995 0.47 0.47


41
The re-analysis of 91 cases from the original NCHRP 507 source data and the full analysis of the
complete Tier 1 and Tier 2 source data cases at EOID matched closely to the reported EOID
values in NCHRP 507. The similarities in the EOID reported results placed credence on the

present research modeling. The mean at both EOID and BOR showed improvements for the
Full PSU Master 179 piles due to the considerable effort expended on identifying and correcting
anomalies in the PDLT2000 and DFLTD databases. The Full PSU Master s.d. results at BOR did
not correspond as closely to the reported NCHRP 507 appendix values, as shown in italics in
Table 5.1. But the current research improved significantly the mean at BOR, which was no
longer unconservative. It can be seen that all BOR statistics displayed a much more accurate
mean value with a significantly tighter statistical variation from that mean expressed by the
COV. Considering the analytical methodology and no use of outlier definitions, the discrepancy
between the two results was not of concern
It was inferred from the NCHRP 507 report that the data reported at EOID and BOR were not
produced entirely by NCHRP 507 researchers, but made available to the NCHRP 507 study
group by GRL, Inc. (Paikowsky 2004). Despite the fact that not all the source data were made
available to this current research team, the close match was excellent verification of the present
studys analytical equivalency. It also confirmed that any assumptions made for individual case
histories were defensible. It should be noted the statistics offered in NCHRP 507 may also have
had outliers removed beyond +/- 2 s.d. prior to calibration which improved the statistics
characteristics. In this study, statistics calculations using the 91 cases from the PSU PDLT2000
database and the 179 cases from the Full PSU Master had no outliers removed.
5.2 FULL POPULATION SAMPLE ANALYSIS
The Full PSU Master database contained 194 cases categorized in Input Tier 1 and Tier 2. After
GRLWEAP analysis, 179 cases produced valid case values placed in output Ranks 1, 2, or 3
and shown above in Table 5.1. These cases made up the full statistical population sample
available to provide LRFD resistance factors for GRLWEAP. The missing 15 cases were placed
in output Rank 4 because they lacked either a documented Davissons capacity or a reported
blow count. The comparison between the EOID and BOR capacities predicted by GRLWEAP
for the 179 pile case histories is illustrated in Fig. 5.1.


42
Figure 5.1: Predicted BOR capacity to predicted EOID capacity from GRLWEAP

GRLWEAP indicated a drop in capacity for 25 of these 179 cases. The percentage drop from
EOID to BOR capacity varied from 1.2% to 43%. If a GRLWEAP predicted capacity difference
of 5 % is considered significant less than 5% difference may indicate little, to no, confirmed
capacity change. By this measure seven piles of the 25 piles were shown likely to be capacity
neutral between EOID and BOR. It suggested that of the 179 cases, 18 piles exhibited possible
relaxation based on GRLWEAP predicted capacities.
To provide a calibration best representing typical Oregon soils, considerable effort was made
to attain quality cohesive soil case histories. When completed, this offered a total of 38 clay
sites and 43 mixed soil sites available in Input Tier 1 and Tier 2 that should exhibit significant
pile set-up on restrike. A series of statistical steps were taken to produce samples of the
population for analysis and that were representative of the ODOT standard of practice. In
constructing samples, consideration was given to the following:
1. Examine trends observed in the full population sample data set,
2. Study population statistical outliers and any mismatched cases requiring a filter
criteria, and
3. Select a subset group of scenarios extracted from the full population that best represents
ODOT site investigation, design analysis, and field installation standard of practice.
43
Some concern existed on the possibility that gross data entry errors compiling the source
databases (PDLT2000, DFLTD) were responsible for the large number of anomalies discussed in
Section 3. Unfortunately, significant discrepancies in a GRLWEAP capacity prediction can arise
from erroneous reporting of field blow count, which was the largest single source of anomalies
reported above between source databases. Errors on the blow count low side reduce predicted
capacity, raise mean , and increase the COV, especially for the EOID. To help identify these
possible errors, a simple blow count-based BOR/EOID set-up ratio (SR) breakdown was
performed that used the same hammer on restrike. This allowed a closer study on a case-by-case
basis and ensured no rogue case histories distorted the statistics. This is shown below in Figure
5.2 that indicates of the total 322 case histories, a large majority had consistently low SR (SR <
5) and were highly unlikely to contain any errors.


Figure 5.2: Frequency of set-up ratio in the Full PSU Master
44
The four single cases with SR > 30 had ratios of 32, 48, 63, and 166. Of these, two were from
Utah in highly sensitive clays, and this soil type did appear to be a mismatch to the stable soil
types in the remainder of the dataset. The remaining two were in Texas and South Carolina. In
the case of the Texas pile, the boring log stopped 18 feet short of the pile tip, was ranked in the
low input Tier 2b, and ranked poorly on output in Rank 3. The South Carolina pile showed the
SR of 166 and delivered the predicted GRLWEAP capacity 7 times higher moving to BOR from
the EOID condition; however, CAPWAP suggested only 8% capacity increase moving to the
BOR restrike capacity. With South Carolina having the largest percent of anomalies of any state
DOT in the source data (Section 3), this was treated as an EOID or BOR data entry keyboard
blow count error and/or a hammer efficiency issue at the BOR condition. All four piles were
flagged and did not feature in any scenario for statistical study. This established the total number
of valid cases for the resistance factor calibration effort to be 175, which was 75% more than
reported in NCHRP 507. The final statistical parameters, broken down by pile type and soil type,
for the 175 qualified case histories are shown in Table 5.2 below.

Table 5.2: Statistical characteristics for GRLWEAP based capacity of the 175 qualified cases by soil and
pile type*
Soil Pile # of EOID BOR
**

Type Type Cases S.D. COV S.D. COV
All 34 1.94 1.42 0.73 1.1 0.61 0.55
Concrete 15 1.21 0.53 0.44 0.71 0.24 0.33
CEP 15 2.52 1.83 0.73 1.31 0.59 0.45
OEP 0 - - - - - -
Clay H-pile 4 2.45 1.19 0.49 1.8 0.96 0.53
All 98 1.27 0.66 0.52 0.9 0.46 0.51
Concrete 60 1.21 0.61 0.5 0.76 0.36 0.48
CEP 18 1.44 0.79 0.55 1.11 0.43 0.39
OEP 2 1.96 0.13 0.07 1.24 0.08 0.07
Sand H-pile 18 1.23 0.61 0.5 1.12 0.59 0.52
All 43 1.9 1.47 0.77 1.12 0.41 0.36
Concrete 14 1.22 0.34 0.28 0.92 0.32 0.35
CEP 15 2.89 2.09 0.72 1.24 0.38 0.31
OEP 5 1.37 0.39 0.29 1.1 0.36 0.33
Mixed H-pile 9 1.62 0.76 0.47 1.25 0.51 0.41
All Soils All 175 1.555 1.1 0.71 0.993 0.47 0.47
*
All GRLWEAP capacities used the 2005 version, with updates, normalized by Davisson reported capacities in the databases accessed.
**
The time between EOID and BOR is known for 64 of the 175 cases and varied between 16 hours and 93 days (see also Section 6.3)

Overall, without any filtering of the data, EOID analysis displayed a much larger statistical
spread than at the BOR condition. The conservative nature of EOID capacity, the considerable
variations in the cohesive soils, and frequent large capacity well above Davissons in sands were
determined to be the driving influence of this scatter. The Sand cases comprised over 50% of the
total sample and the COV, expressing the spread, was slightly lower than the Clay and Mixed
sites, and overall the COV did not improve between EOID and BOR. Sands also showed a less
conservative bias compared to Clays and the Mixed cases. Frequency of occurrence
histograms were plotted to visually display both EOID and BOR population trends in Table 5.2
broken down into soil type, and these are shown in Figures 5.3 and 5.4 respectively. Attention
should focus on the overall fit and also the tail shapes fit, particularly high and low
predictions, as these dictate the COV and the difference between FOSM and Monte Carlo based
resistance values predicted in Section 7. Monte Carlo based calibration results were driven by
the lower portion of the distribution, where capacity predictions were unconservative and risk
of failure was higher.
45
The predominance of the Sand site cases in the lower bias tail (high capacity predictions
compared to Davisson capacity) are seen in both Figures 5.3 and 5.4. Normal and lognormal
curve fits and associated statistics were shown to visually determine the likely statistical
distribution types for both populations and to compare to NCHRP 507. Visual examination
clearly showed both EOID and BOR populations were represented better by lognormal
distributions than by normal distributions, in agreement with previous calibration efforts. The
tightening of the BOR histogram, with a lower COV, displayed the more reliable prediction at
restrike of the Davisson capacity with all soils, especially capturing the influence of pile set-up.


Figure 5.3: EOID histogram

Figure 5.4: BOR histogram
46


5.3 SCENARIO DEVELOPMENT
A series of different scenarios were designed to study existing pile capacity LRFD statistical
filters, input tier differences, outlier definitions, and ODOT practice. Statistical filters and
presentation formats appearing in NCHRP 507 for CAPWAP based nominal capacity were
selected and used primarily for comparison. Both GRLWEAP and CAPWAP use the finite
difference based wave theory, and it was of interest to determine whether both methods exhibited
similar statistical trends. The NCHRP 507 visual presentation of the changes with blow count
suggested a dramatic increase in data scatter for pile capacity predicted by CAPWAP in easy
driving conditions at and below 2 BPI. This was adopted to detect its effect and is discussed in
detail in Section 6.2.
Scenario A contained the full 175 case history population after removal of the four high
SR cases and was the best match to NCHRP 507s reported work, but now with over 70%
more case histories. It did not employ any kind of filtering of data. This set of case
histories was selected to be the control scenario, and the histograms for the cases were
shown in Figure 5.2 and Figure 5.3 for EOID and BOR respectively.
Scenario B contained only those cases in top Tier 1 and Tier 2a that had the best possible
site investigation source data and required no key assumptions in the driving analysis.
Scenario C used the same NCHRP 507 arbitrary +/- 2 s.d. range tail outliers filter and
removed cases beyond this range out of Scenario A. This disqualified only nine cases as
the tails on Scenario A were evidently already well conditioned with few extreme values.
Scenario D used the 2 BPI cut off from NCHRP 507 to modify Scenario A.
Scenario E employed the 2 BPI cutoff indicated in NCHRP 507 and the +/- 2 s.d.
outlier definition to modify the higher input tier quality cases extracted from Scenario B.
Scenarios A to E were designed to examine these different NCHRP 507 filter effects and draw
conclusions on the changes in base and COV statistics. However, the following two additional
scenarios moved closer to ODOT practice.
Scenario F modified Scenario B and was designed to reflect current ODOT practice by
using the easy driving cutoff at 2 BPI and, recall, used the higher input quality Tier 1 and
Tier 2a without any outlier definitions.
Scenario G modified Scenario F by also including the Tier 2b cases when output ranked
in the very satisfactory highest Rank 1 and were, therefore, shown to be insensitive to
assumptions.
47
Scenario H was an attempt to study the present combined AASHTO and FHWA
guidelines on restrike (TR) and time to static load test (TST) by enforcing wait times
extracted from the total of 69 piles in Scenario F that reported calendar TR and TST
times (see also section 6.3).

The base statistics, including the number of qualifying cases, mean, , s.d., and COVs for these
different scenarios, A through H, are presented in Table 5.3. The table illustrates well the
statistical effect on mean and COV to that each filter had on each scenario. The baseline
NCHRP 507 reported statistics for EOID and BOR from CAPWAP and GRLWEAP are given at
the top of the table again for reference.

Table 5.3: Base statistics for Scenarios A-H compared to NCHRP 507*
Pile Resistance Prediction Method Case # of Piles Mean S.D. COV
EOID 125 1.626 0.797 0.490
CAPWAP
BOR
**
162 1.158 0.393 0.339
EOID 99 1.656 1.199 0.724
NCHRP-507
GRLWEAP
BOR 99 0.939 0.399 0.425
EOID 175 1.555 1.102 0.708
PSU Scenario A Tier 1-2
BOR 175 0.993 0.468 0.472
EOID 82 1.685 1.100 0.653
PSU Scenario B Tier 1-2a
BOR 82 1.030 0.447 0.434
EOID 163 1.372 0.669 0.488
PSU Scenario C
Tier 1-2 +/-2
S.D.
BOR 165 0.938 0.399 0.425
EOID 138 1.269 0.614 0.484
PSU Scenario D Tier 1-2 BPI>2
BOR 162 0.967 0.459 0.475
EOID 65 1.394 0.603 0.433
PSU Scenario E
Tier 1-2a BPI>2
+/-2 S.D.
BOR 73 0.970 0.397 0.409
EOID 69 1.406 0.596 0.423
PSU Scenario F Tier 1-2a BPI>2
BOR 79 1.012 0.429 0.424
EOID 94 1.328 0.564 0.425
PSU Scenario G
Tier 1-2a, 2b
Rank 1 BPI>2
BOR 114 0.985 0.430 0.437
EOID 58 1.330 0.570 0.429
PSU Scenario H
Tier 1-2a, 2b
Rank 1, BPI>2
TR and TST
Filters
BOR 69 0.974 0.408 0.419
*
All GRLWEAP capacities used the 2005 version, with updates, normalized by Davisson reported capacities in the databases accessed.
**
The time between EOID and BOR was known for 64 of the 175 cases and varied between 16 hours and 93 days (see also Section 6.3)

48
The two highlighted rows in the table represent the closest fit to NCHRP 507s main reported
work, which was the basis for AASHTO LRFD Manuals (2006, 2009) at EOID. The Scenario
As expanded total pile count of 175, without any filters or outliers, had a slightly less
conservative at EOID than the 99 piles reported in NCHRP 507. However, the BOR of 0.993
did reveal less unconservatism with almost an exact match to the Davissons capacity when
compared to BOR statistics in the appendix of NCHRP 507. Recall Scenario G was created by
loosening input quality restrictions on likely ODOT match Scenario F by including the lower
input Tier 2b cases (which required more assumptions) that were in Rank 1. This demonstrated

the value of output quality ranks by gaining a considerable number of cases. Loosening the tier
restriction added close to 50% more cases and showed no degradation of statistical quality
compared to Scenario F at either EOID or the BOR condition. Case history count in Scenario G
was comparable to those used for GRLWEAP in NCHRP 507. Scenario H followed the much
stricter guidance on TR and TST delays set by AASHTO and FHWA and showed improvement
of EOID and BOR conditions. However, this was not a fair assessment when applied to cases
whose installation and load test took place before the code requirements. It did not apply to
ODOT practice and was not used in the more advanced Monte Carlo Reliability Method
calibration presented in Section 7.2
5.4 DISTRIBUTION TESTING
Pile resistances are statistically random variables and calibration of the LRFD resistance factor
requires assumptions on the shape of their PDF distribution. Scenarios A to F were examined to
determine how their populations fitted to both a normal and lognormal distributions. The
statistical analysis software Minitab16 (Minitab 2009) was used to create each scenarios PDF,
CDF, and SNV, described for calibration purposes by Allen (Allen 2005b; Allen, et al. 2005). A
scenarios population distribution can be visually scanned utilizing the PDF and CDF plots.
Normal distributions show linear on the SNV plot and both normal and lognormal distribution
fits were tested by Minitab.
49
The rigorous Anderson Darling (AD) test and the P-tests use the SNV plots to determine
statistical distributions fits employing a null hypothesis theory. The null hypothesis being the
distribution does fit in either the normal or the lognormal distribution. The P-value is a measure
of the risk associated with false rejection of the null hypothesis and offers indications that the
data fits a specified distribution. If the P-value is less than 0.05, then there is a greater than 95%
chance that the specified distribution is incorrect and the null hypothesis can be rejected without
risk. The AD test was employed to determine the goodness of fit accuracy of the distribution
to model each scenarios sample and was used in conjunction with the P-value. The AD and P-
values for all scenarios at EOID and BOR for both normal and lognormal distributions were
examined. This is shown for Scenarios A at the EOID on Figure 5.5 for the lognormal fit. A true
lognormal would appear as a straight line on the SNV to log bias plot. The lognormal EOID
and BOR set for each of the final implementation scenarios of interest to ODOT (Section 7.2)
can be seen in Appendix B.

50
10.0 1.0 0.1
3
2
1
0
-1
-2
-3
EOI D Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc 0.2884
Scale 0.5469
N 172
AD 2.298
P- Value < 0.005
Lognor mal Pr obabi l i t y Pl ot f or EOI D
Scenario A
95% Confidence Interval

Figure 5.5: Lognormal distribution fit to Scenario A at EOID
To aid interpretation of the standard normal plots, the +/-5% variant from the mean was plotted
represented by the two lines beyond the mean. In general, all scenarios populations fit the
lognormal distribution with one exception: Scenario A at EOID produced a P-value < 0.005.
Graphically, the visual degree of scatter in the data at both upper and lower EOID tails was the
cause of the rejection. Consistent with all other published LRFD calibration work and for the
purpose of this research, all scenarios, including Scenario A at EOID, were modeled as a
lognormal distribution fit.

6.0 OUTLIER EXAMINATION
As with any population sample, a careful examination for those data that do not belong in the
sample, improves validity. A variety of driven pile characteristics were investigated to determine
their influence on the overall sample and to help detect mismatched case histories. When any
trend was discovered, it was considered as a possible filter before final calibration analysis
proceeded. The +/- 2 s.d. outlier imposed on all NCHRP 507 sets (Paikowsky 2004) was
examined, but not considered a valid outlier definition, as it produced artificial tail
modifications. This outlier approach was not employed in any final implementation scenario
selection (see Section 7). After selecting the fit to a log normal distribution the tail controlled
accurate recalibration efforts. The removal of extreme data points without any other justification
was not easily defensible. All the efforts to identify mismatched case histories are reported
below and include input tiers to output ranks, blow count ranges, the time to restrike (TR), and
time to static load test (TST).
6.1 INPUT TIER AND OUTPUT RANK
A statistical breakdown was performed on both the populations input tiers and the output ranks
to assess if clear trends existed and to help determine the usefulness of both criteria. Table 6.1
presents a breakdown of data at EOID and BOR performed on three input tier subset variations
for Scenario A: Tier 1 only, Tier 1 and 2a only, and all Tier 1 and Tier 2 cases.

Table 6.1: Statistical breakdown of Tier 1, Tiers 1 & 2a, and Tiers 1 & 2 for EOID and BOR
EOID BOR
Input
Tier
Soil
Type
Pile
Type
# of
Cases
S.D. COV S.D. COV
1 All All 55 1.63 1.01 0.62 1.01 0.46 0.46
1 and 2a All All 86 1.64 1.02 0.62 1.04 0.44 0.42
1 and 2 All All 175 1.56 1.10 0.71 0.99 0.47 0.47

51
As previously discussed, the input tiers were used to evaluate the quality and completeness of a
cases source input data. As expected, the comparison from Table 6.1 showed that the inclusion
of less complete source data produced degradation of the statistical parameters, but only at
EOID. Little difference was evident when including Tier 2a with Tier 1 at EOID, while the BOR
parameters remained relatively constant. Tailoring the recalibration effort to ODOT standard
field practice required the inclusion of case histories that were well defined. The use of any
questionable source data would induce an undesirable skew in the calculated resistance factor.
Trends displayed on the EOID analysis in Table 6.1 suggested that adding Tier 2a to Tier 1 was
acceptable, and that the combined use of Tier 1 and Tier 2a could form an appropriate basis to
study ODOT practice.

Table 6.2 displays the statistical and COV breakdown for EOID and BOR by soil and pile type
for the top input Tier 1 and Tier 2a. Table 6.3 displays the same breakdown statistics but for the
higher output Ranks 1 and 2. Only 50% of these cases were in the top Tier 1 and Tier 2a, but
almost 80% of the total cases were in the excellent and good output rank. It is apparent no
serious degradation of statistics arose from increasing the number of reasonable assumptions
made throughout the analysis. No clear trends were determined between the different qualitative
output ranks between each input tier and sub tier in Table 6.2. This was expected as the issues
relating to assigning a case to an output rank were varied.

Table 6.2: Top input Tier 1 & 2a statistical breakdown by soil and pile type*
EOID BOR

**

Soil Type Pile Type
# of
Cases
S.D. COV S.D. COV
All 15 1.87 0.68 0.36 1.20 0.62 0.52
Concrete 4 1.59 0.93 0.59 0.73 0.35 0.48
CEP 8 1.73 0.41 0.24 1.10 0.35 0.32
OEP 0 - - - - - -
Clay
H-pile 3 2.62 0.50 0.19 2.09 0.21 0.10
All 42 1.31 0.56 0.42 0.88 0.36 0.41
Concrete 29 1.19 0.51 0.43 0.74 0.22 0.30
CEP 6 1.31 0.44 0.33 1.12 0.44 0.39
OEP 2 1.96 0.13 0.07 1.24 0.08 0.07
Sand
H-pile 5 1.79 0.69 0.39 1.28 0.50 0.39
All 29 1.99 1.47 0.74 1.18 0.38 0.33
Concrete 9 1.29 0.39 0.30 1.03 0.33 0.32
CEP 12 2.63 2.06 0.78 1.13 0.36 0.31
OEP 1 1.88 - - 1.43 - -
Mixed
H-pile 7 1.82 0.75 0.41 1.41 0.45 0.32
ALL All 86 1.64 1.02 0.62 1.04 0.44 0.42
*
All GRLWEAP capacities used the 2005 version, with updates, normalized by Davisson reported capacities from the databases accessed.
52
**
The time between EOID and BOR was known for 64 of the 175 cases and varied between 16 hours and 93 days (see also Section 6.3)

Table 6.3: Top output ranks 1 & 2 statistical breakdown by soil and pile type*
EOID BOR
**

Soil Type Pile Type
# of
Cases
S.D. COV S.D. COV
All 25 2.31 1.59 0.69 1.12 0.61 0.55
Concrete 11 1.28 0.61 0.48 0.64 0.19 0.29
CEP 9 3.21 1.96 0.61 1.44 0.51 0.35
OEP 0 - - - - - -
Clay
H-pile 5 2.93 1.16 0.40 1.59 0.71 0.45
All 76 1.37 0.69 0.50 0.87 0.43 0.50
Concrete 52 1.26 0.63 0.50 0.75 0.36 0.48
CEP 12 1.71 0.84 0.49 1.18 0.51 0.43
OEP 2 1.96 0.13 0.07 1.24 0.08 0.07
Sand
H-pile 10 1.41 0.66 0.47 1.06 0.42 0.40
All 34 2.13 1.58 0.74 1.11 0.39 0.35
Concrete 9 1.32 0.34 0.26 0.90 0.37 0.41
CEP 13 3.16 2.12 0.67 1.21 0.36 0.30
OEP 5 1.37 0.39 0.29 1.10 0.36 0.33
Mixed
H-pile 7 1.81 0.75 0.41 1.21 0.48 0.40
ALL All 135 1.74 1.27 0.73 0.98 0.53 0.55
*
All GRLWEAP capacities used the 2005 version, with updates, normalized by Davisson reported capacities from the databases accessed.
**
The time between EOID and BOR was known for 64 of the 175 cases and varied between 16 hours and 93 days (see also Section 6.3)

6.2 BLOW COUNT INVESTIGATION
The pile driving process is an energy balance of two phenomena: an input of relatively known
energy at the top of a pile and the surrounding soils ability to absorb this energy. The soil
absorbs this delivered energy when the tip is advanced through the displaced soil. A basic
assumption of this model is that the pile, hammer, and soil be considered a lumped mass-spring
system. GRLWEAP employs the Smiths lumped mass model to approximate the external pile
supporting soil resistance. If a pile is driving too easily, the large energy loss developed by the
displaced soil and the complex large dynamic strain pattern over time and distance introduces
possible plug and other violations to the assumptions in the model. NCHRP 507 reported that at
easy driving, the complex interaction and energy loss due to the high velocity and acceleration
displacement of the soil at the pile tip is unable to be captured accurately. Further, the
GRLWEAP manual stated that during easy driving of less than 2 BPI, inaccurate results may be
produced so reduced energy should be employed. The GRLWEAP manual also warned against
sustained hard driving conditions, as did the FHWA manual, inducing a hammer serviceability
issue rather than affecting the accuracy of a capacity prediction. Damage to the pile and/or
hammer can occur under these conditons, which typically causes an upper range of 15 to 20 BPI
to be recommended by the hammer manufacturer.
53
NCHRP 507 did not analyze blow count spread versus GRLWEAP predicted capacity, but did
report this spread for CAPWAP. The report text stated that, measured by mean and COV, a
statistical change at 4.5 BPI occurred, which visually was better expressed from 2 BPI. As both

GRLWEAP and CAPWAP are dynamic methods using the lumped mass model, a similar
relationship between and blow count may exist for GRLWEAP. Figure 6.1 shows the blow
counts for the 175 cases, at both EOID and BOR, plotted against their respective values to help
determine if statistical accuracy trends existed between BPI and the case-by-case calculated .
Figure 6.1 does clearly show in a similar trend to CAPWAP, GRLWEAPs inability to
accurately and reliably predict capacity at the lower blow counts. The data displayed supports
the idea that as blow counts increased, there was less permanent set between each successive
blow count and the soil-pile-hammer model behaved more predictably when modeled by
GRLWEAP. To determine a reasonable quantitative break point in the data, the blow counts
shown in Figure 6.1 were placed into 2 BPI incremental ranges, following the format used in
NCHRP 507, and each range average and s.d. calculated.


Figure 6.1: All EOID and BOR qualified 175 values plotted to blow count by soil type
54
The number of cases in each range, mean , and the vertical bar in the range expressing +/- 1 s.d.
are shown in Figure 6.2, together with the overall sample mean and s.d. range with a dashed
line for all 175 piles. Each pile case history generated two data points: one for EOID and one for
BOR.


Figure 6.2: All EOID and BOR qualified 175 values plotted to blow count in 2 BPI intervals
There were insufficient data points to make reasonable comparisons for every 2 BPI increments
after 12 BPI. Therefore, Figure 6.2 shows the BPI categories greater than 12 to less than or
equal to 16 combined, and greater than 16 to less than or equal to 22 combined to correct this
lack of data. The 46 instances where pile blow counts were greater than 22 were plotted at 22
BPI for presentation clarity. All blow count ranges above 2 BPI showed relatively consistent
statistical variation up to the last recorded data point of 100 BPI. This fell into line with both the
GRLWEAP manual recommendations and NCHRP 507s interpretation for CAPWAP.

Table 6.4: Blow count ranges for EOID and BOR in Full PSU Master
EOID BOR Total
BPI
# Cases Total % # Cases Total % # Cases Total %
>0 to 10 207 78% 162 59% 369 68%
>10 to 15 19 7% 36 13% 55 10%
>15 to 20 12 5% 25 9% 37 7%
>20 to 30 13 5% 24 9% 37 7%
>30 to 40 7 3% 10 4% 17 3%
>40 6 2% 18 7% 24 4%
Sum 264 100% 275 100% 539 100%

55
Table 6.4 offers a breakdown of the blow count ranges between EOID and BOR from the Full
PSU Master database for all input tiers and output ranks. (Recall that 179 of the 322 cases
received complete DRIVEN and GRLWEAP analyses.) The difference in case numbers between

the EOID and BOR count was due to cases having no reported blow counts at either EOID or
BOR.
Prior to 2009, ODOT set their acceptable range for driving blow counts at 3 BPI 15 to avoid
easy driving at the lower limit and possible hammer damage at the upper limit. Field practice
reported by TAC members confirmed high blow counts were tolerated for the strictly limited
time duration of the restrike. These statistics do not support implementing any upper cutoff for
capacity by GRLWEAP; in fact, 29% of the case history piles have pushed hammer blow counts
greater than the 15 BPI limit on restrike for the BOR blow count series. As reported in the
GRLWEAP manual, an upper cutoff represents a hammer serviceability concern rather than
accurate capacity determination concern. As the blow counts greater than 15 BPI were field
recorded values in the database, no use of an upper BPI cut off for the purpose of the resistance
factor recalibration was used.
6.3 TIME TO RESTRIKE AND TIME TO LOAD TEST
INVESTIGATIONS
The contribution of soil set-up has been shown to statistically improve the GRLWEAP
prediction of a piles ultimate capacity when using BOR. The research group investigated the
database pile case histories level of adherence to current AASHTO and ODOT required
minimum wait time to restrike (TR) and time to static load test (TST) times. Among the 175
piles accepted for resistance calibration, 64 in Tier 1 and Tier 2 had TR and TST recorded in
the Full PSU Master database. This pile count is shown in Table 6.5, broken down by pile and
soil type.

Table 6.5: Pile counts complying with current AASHTO and ODOT TR and TST times
Con CEP HP OEP Total
Clay 8 6 3 0 17
Sand 18 9 2 2 31
Mix 6 4 4 2 16
Total 32 19 9 4 64

56
The distribution of soil and pile type as well as the statistical averages at EOID and BOR for the
64 cases was similar to that reported in the full population sample of the 175 cases in Table 5.2.
This allowed observations made from the smaller 64-pile subset to be considered appropriate for
the full population sample with relative confidence. AASHTO recommends the required wait
time before restrike based on one of the following: clean sands at 1 day, silty sands at 2 days,
sandy silts at 3-5 days, and clays at 7 days (AASHTO 2009). ODOT bridge foundation
construction specifications typically require a one-day minimum set-up period before a restrike
can be performed. AASHTO mandates a minimum of five days after initial driving has passed
before a static load test can be conducted.

The times stated by both ODOT and AASHTO were compared to those recorded in the database
for the 64 piles, and the majority of cases did not meet both the TR and TST requirements. At
restrike, six piles failed ODOTs wait time requirement, while 16 piles failed AASHTOs wait
time recommendation. Further, 48 out of 64 piles failed to wait the AASHTOs recommended
required time after initial driving before the static load test was conducted. The 64 cases were
examined to determine if any trends could be detected to assist identifying mismatched cases that
were restruck too early and, therefore, captured little set-up or were load tested much later.
Figures 6.3 and 6.4 presents the ratio of TR to TST times compared to their corresponding BOR
case-by-case values broken down by soil type and by pile type respectively. Load testing
would have occurred after restrike for TR/TST < 1 and restrike would have occurred after the
load test for TST/TR < 1.


Figure 6.3: Restrike and load test time ratios by soil type

Figure 6.4: Restrike and load test time ratios by pile type
57
The majority of the piles that did not meet the standard TR and TST time requirements were in
cohesive soils. A cloud of conservative, high , cohesive soil pile cases that had been restruck

58
early and load tested late (TR/TST<< 1), was prominent in Figure 6.3. The majority of this cloud
was made up of CEP and HP pile sections. Also five sites showed very late restrike performed
well after the static load test (TST/TR<< 1).
Calibration for the resistance factor by the FOSM method uses lognormal distributions, which
are sensitive to both upper and lower tails. The more advanced, and appropriate, Monte Carlo
based recalibration method, employed using data from the lower tail, was not as concerned with
the conservative ( > 1) side of the distribution; therefore, these cases did not affect the
calculated value and did not have the same degree of concern for this study. The presence of
unconservative sand cases with low , and shown in Figure 6.3, was similar in ratio to the
statistics seen in the full 175 case distributions; however, this group of sand cases did not violate
the TR and TST requirements.
Without any compelling, consistent, and clear trends, all the cases that violated the TST and TR
requirement set forth through ODOT and AASHTO were included. The rationale to include
these cases was the following:
- Cases that violated their TR and TST were conservative because they failed to capture
adequate set-up and, therefore, did not negatively influence the calibration effort.
- A large majority of the cases that violated TR and TST were cohesive. By eliminating
them from the calibration effort, the soil type balance of case history data would no
longer have represented the broad distribution of soils found in Oregon.
- Only 64 of the 172 cases available for analysis had TR and TST values. This would
unduly have eliminated a quality subset of data from analysis.
A large variation in installation condition requirements and field practice norms existed in the
database due to the differences in TR and TST time and geographic location of the case histories.
The inherent variability of national piling standards was captured by keeping a broad database
that should produce more representative and generally applicable conservative resistance
factor results.
In summary, no cases were removed from the dataset before calibration began for reasons of
falling outside any arbitrary s.d. range, input tier position, or restrike and load test delay timing.
When grouped by blow counts, cases were shown to dramatically increase their COV and for
the lower blow counts; therefore, a minimum cut-off of 2 BPI was adopted.


7.0 RESISTANCE FACTOR DEVELOPMENT
Section 5.3 of this report discussed the composition of eight scenarios, A through H, which were
used to produce the bias statistics to examine differences in input tiers, the use of NCHRP 507
outlier definitions, and likely matches to ODOT practice. The resistance factor calibration
employing these statistics was performed with the procedures outlined by AASHTO (Allen, et al.
2005). Both the FOSM equation and the Monte Carlo probabilistic procedures outlined in
Section 1.4 were used. The statistical breakdown and the outlier studies, discussed in Section 5
and 6, did reveal some differences between sites, mainly those deriving resistance in granular
materials to those sites predominately deriving resistance in cohesive soils. Pile set-up, and the
associated increased capacities, are likely to arise in cohesive soils, but are also detected in
granular soils. If restrike decisions are made based on poor EOID capacity, then different
resistance factors should be available at BOR derived from the restrike population statistics.
These differences, expressed in the field by blow count changes, are the key to appropriate
project-by-project decisions around capturing set-up and optimizing pile acceptance under
LRFD. With the amount of set-up possible in piles founded in mainly cohesive soils being
higher, two additional practice scenarios were created to assist in implementation of this
researchs findings. The breakout of cohesive and cohesionless soils from Scenario A could not
be done with equal confidence as some doubt was statistically detected (see section 5.4) on the
appropriateness of the lognormal assumption. Scenario A also contained a large number of
poorly defined Tier 2b case histories. Therefore, both new scenarios were derived from Scenario
G and are listed below.
Scenario I was extracted from Scenario G but contained only those sites that derived at least 20%
DRIVEN predicted capacity from cohesive soils.
Scenario J, for comparison purposes, was also extracted from Scenario G but contained only
sites with at least 80% of resistance from granular soils.
Recall Scenario G (Section 5.3) had over 100 case histories, matched well to ODOT practice,
which follows standard AASHTO and FHWA recommendations, and used blow counts > 2BPI
to remove easy driving mismatches. A total of ten scenarios, A through J, were examined with
simple closed form FOSM, with Scenarios A, F, G, I, and J also undergoing the considerably
more complex Monte Carlo reliability method.
7.1 FIRST ORDER SECOND MOMENT (FOSM)
59
When the bias ratio is taken as lognormal distributed, the closed FOSM calibrated resistance
can be quickly calculated by the closed form Equation 1-3 from Section 1.4, which is repeated
here.

( )( | | ) | |
2
L
2
D
2
R L
L
D
D
2
R
2
L
2
D
L
L
D
D R
COV COV 1 COV 1 ln exp
Q
Q
COV 1
COV COV 1
Q
Q
+ + + |
|
.
|

\
|
+
|
.
|

\
|

+
+ +
|
.
|

\
|
+
|
.
|

\
|

=
t

Contributions to total load from Q
D
and Q
L
sources are included, expressed by their bias and
COV. These were set at the values required by ODOT, were used in Phase 1 (Smith and Dusicka
2009), and for consistency were the values most often selected by LRFD researchers for driven
pile studies (Long 2009; Paikowsky 2004). For redundant pile groups of five or more, the target
reliability index was 2.33, and for fewer than five piles, the target reliability was 3.0. The full
set of statistical parameters used in this study is shown in Table 7.1.

Table 7.1: Statistical load parameters used for calibration
Load Type DL LL
Distribution Lognormal Lognormal
Monte Carlo Value 800 kips 400 kips
Bias 1.05 1.15
COV 0.10 0.20

DL
1.25

LL
1.75

More superior capacity methods should predict capacity more accurately and precisely, and they
may be quickly compared by their COVs or by the efficiency ratio, /, which normalizes the
LRFD capacity reduction factor by the detected method bias. A high COV corresponds to a low
efficiency value, and the / and COV efficiencies can be used to rank the optimum method:
either EOID capacity or BOR capacity to show the method with the least over-capacity in the
field. To aid implementation, this optimization is further explored in Section 7.3. Table 7.2
reports all the FOSM calculated resistance values and / efficiencies for both reliability
values of 2.33 and 3.0, and also for comparison, the reported NCHRP 507 values by GRLWEAP
and CAPWAP at the EOID and BOR condition. In Table 7.2, the NCHRP 507 BOR statistics are
shown in italics to denote they did not appear in the text of that report.
60
The FOSM resistance values shown for the baseline Scenario A, and from NCHRP 507 at EOID
(forming AASHTO recommendations), are highlighted and are very similar, but they do, in fact,
illustrate some differences. The parameters under which NCHRP 507 generated the FOSM
resistance factors most likely had used the strict +/- 2 s.d. outlier cutoff to improve COVs.
Further, the original GRL based work (Rausche, et al. 1997) appears to have applied a 10 BPI
blow count cap to select acceptable case histories. If these different constraints are recognized,
the Scenario A statistics with 175 piles were slightly superior to the 99 piles of the original
NCHRP 507 work and reflected the considerable effort the current study expended for the
resolution of anomalies and the inclusion of only carefully selected, and well documented, case
histories in the Full PSU Master database.

Table 7.2: FOSM resistance values and efficiencies for =2.33 and =3.0
FOSM
S.D. COV Factor / Efficiency
Pile Resistance
Prediction Method
Case
# of
Piles
Mean
=2.33 =3.0 =2.33 =3.0
EOID 125 1.626 0.797 0.490 0.59 0.42 0.36 0.26
CAPWAP
BOR 162 1.158 0.393 0.339 0.58 0.45 0.50 0.39
EOID 99 1.656 1.199 0.724 0.36 0.23 0.22 0.14
NCHRP
507
GRLWEAP
BOR 99 0.939 0.399 0.425 0.39 0.29 0.42 0.31
EOID 175 1.555 1.102 0.708 0.35 0.22 0.23 0.14
Scenario A Tier 1 and 2
BOR 175 0.993 0.468 0.472 0.38 0.27 0.38 0.27
EOID 82 1.685 1.100 0.653 0.43 0.28 0.26 0.17
Scenario B Tier 1 and 2a
BOR 82 1.030 0.447 0.434 0.42 0.31 0.41 0.30
EOID 163 1.372 0.669 0.488 0.50 0.36 0.37 0.26
Scenario C
Tier 1 and 2
+/-2 S.D.
BOR 165 0.938 0.399 0.425 0.39 0.29 0.42 0.31
EOID 138 1.269 0.614 0.484 0.47 0.33 0.37 0.26
Scenario D
Tier 1 and 2
BPI>2
BOR 162 0.967 0.459 0.475 0.36 0.26 0.38 0.27
EOID 65 1.394 0.603 0.433 0.57 0.42 0.41 0.30
Scenario E
Tier 1 and 2a
BPI>2
+/-2 S.D.
BOR 73 0.970 0.397 0.409 0.42 0.31 0.43 0.32
EOID 69 1.406 0.596 0.423 0.59 0.43 0.42 0.31
Scenario F
Tier 1 and 2a
BPI>2
BOR 79 1.012 0.429 0.424 0.42 0.31 0.42 0.31
EOID 94 1.328 0.564 0.425 0.56 0.41 0.42 0.31
Scenario G
Tier 1 and 2a+
Tier 2b Rank1
BPI>2
BOR 114 0.985 0.430 0.437 0.40 0.29 0.41 0.30
EOID 58 1.330 0.570 0.429 0.55 0.40 0.42 0.30
Scenario H
Tier 1 and 2a
BPI>2 & TR,
TST
BOR 69 0.974 0.408 0.419 0.41 0.30 0.42 0.31
EOID 43 1.464 0.595 0.406 0.64 0.47 0.44 0.32
Scenario I
Clay &
Mixed
Tier 1 and
2a+Tier 2b
Rank1 BPI>2
BOR 56 1.123 0.457 0.407 0.49 0.36 0.44 0.32
EOID 51 1.214 0.515 0.424 0.51 0.37 0.42 0.31
Scenario J
Sands
Tier 1 and
2a+Tier 2b
Rank1 BPI>2
BOR 58 0.852 0.358 0.421 0.36 0.26 0.42 0.31

61
Measured by / efficiencies alone, both Scenario A and NCHRP 507 indicated using restrike
capacity to optimize the number and/or pile depth, but was much less clear under all other
scenarios. Similarity was seen in the / efficiencies at BOR of all scenarios, and was in close
agreement with NCHRP 507. The BOR values were generally lower than the EOID values.
As expected, the EOID on Scenario A offered the lowest and / efficiency of any in the table
due to the previously reported poorly defined upper tail inflating the COV. Efficiencies on all
other scenarios for both EOID and BOR were generally twice that reported by NCHRP 507 at
EOID. Of significance was the effect shown in Scenario C of employing the +/- 2 s.d. NCHRP
507 outlier definition. From the direct comparison to Scenario A, which used the data in all
qualified tiers, the factor was inflated by 42%, up to 0.501, for EOID at the of 2.33, by the
artificial removal of extreme tail data at both ends.

Not illustrated in the table were the individual soil type specific effects when constructing
Scenario G from Scenario F. The statistical effects of adding the lower Tier 2b Clay sites to
Scenario F was identical to adding Tier 2b Mixed sites to Scenario F; the COV at EOID
increased slightly and BOR was unchanged. But the opposite was found when adding the Tier
2b Sand sites to Scenario F. This supported treating Clay and Mixed sites as a single sample, as
in Scenario I, and treating Sand-only sites to form Scenario J. A clear difference in statistical
sample characteristics existed between piles supported in predominately cohesive soils to those
in cohesionless soils since they were showing as different populations.
7.2 ODOT SCENARIOS AND MONTE CARLO RELIABILITY
To recap, Scenario A represented all 175 acceptable Tier 1 and Tier 2 cases that generated
separate values and did not employ any kind of data filtering. With a good match to NCHRP
507, this scenario was selected as the baseline for comparison. Scenarios F and G were
determined likely to be the best fit to the ODOT standard of practice requiring a single at
EOID and a single at BOR, with high quality site investigation data and field supervision and
documentation during pile driving. These scenarios eliminated the statistical scatter associated
with easy driving (eliminating less than and equal to 2 BPI) and accepted only high tier source
data. In the interest of developing a larger population for calibration analysis, Scenario G
included Tier 2b cases that received an output Rank 1 classification. Output Rank 1 signified
that a sensitivity analysis was conducted on the missing piece of source information and resulted
in less than a 10% difference in the GRLWEAP predicted ultimate capacity. As stated earlier,
Scenarios B through E were for studying the effects of a variety of published filters and outliers
and not designed to represent ODOT practice. Scenario H was a derivative of Scenario G with
those cases removed violating AASHTO current guidelines on TR times and TST times in all
soil groups, irrespective of input tier or output rank. This was overly strict as current AASHTO
code requirements would not have been in force at the time the load tests were conducted.
Scenario I at BOR was selected for further investigation as it allowed GRLWEAP analysis in
Oregons cohesive soils for the benefit of soil set-up. Scenario J for cohesionless sands was kept
as this could be used to compare to the calibrated resistance factor effects for cohesive soils used
in Scenario I. Examination of trends in soil types throughout the previous figures all showed
cohesive soils displaying set-up at BOR, while the mean illustrated the cases behave
conservatively in Scenario I. (Recall, this scenario included all the Clay sites as well as the
Mixed sites.) After review, ODOT selected four scenarios: A, F, G at both EOID and BOR for
full Monte Carlo Reliability based calibration of , and Scenario I. The cohesionless Scenario J
was also of interest.
62
In past studies (Paikowsky 2004; Allen 2005b; Long, et al. 2009), generally, there was a 5% to
10% increase found in a calculated from FOSM to those found from reliability based methods;
however, this may not always be the case as the change depends on the distribution of the data.
The same dead and live load statistical parameters, reported previously in Table 7.1 for FOSM,
and a constant live load to dead load ratio of two were selected for use during Monte Carlo
analysis. Previous studies had shown that altering the load ratio resulted in little variation to
resistance calculation results (Allen 2005b; Long, et al. 2009). The Monte Carlo calibration

effort on the approved scenarios was driven by the data located in the lower tail sections and
was used to construct the most accurate safety margin distribution, where g 0. The high
unconservative predicted resistance (low ) case histories were therefore of most interest, but as
pointed out by Allen (Allen, et al. 2005), due to the nature of geotechnical data, there can often
be insufficient data to confidently model this portion for any calibration effort. The Monte Carlo
technique extrapolates the trends displayed in the lower tail of each scenario and fills the gaps
in data smoothly to establish the distribution of the safety margin. As previously described, the
standard minimum for redundant pile groups of 2.33 was selected for analysis and
corresponded to a probability of failure of 1/100 representative of the level of acceptable risk for
an individual pile within a pile group large enough to have sufficient redundancy. A modified

value was repeatedly selected until the calculated 2.33 was reached, at which point the
acceptable had been determined and the calibration process was complete. The Monte Carlo
technique utilized multiple sets of 10,000 randomly generated variables for loads and resistance
sufficient for calculation to this of 2.33. Calculations were conveniently conducted on Excel

spreadsheets. Any alterations in the random variable statistics affected the calculations and
even a re-execution could change as different random number sets were used. This variation
was accounted for by running at least three iterations for a single input value. If the calculated
consistently produced < 2.33, which was unconservative because of unacceptable probability
of failure, then would be reduced by 0.01 and the iteration process repeated until the target
2.33 was met.
63
The Monte Carlo reliability method was performed on each of the four final ODOT selected
scenarios: A, F, G, and I. For each of these scenarios, the probability density function (PDF),
cumulative distribution function (CDF), and standard normal variable (SNV), for both EOID and
BOR driving conditions are shown in Appendices C through F. Three curves are shown on each
of the standard normal to log bias plots and represent the lognormal regressed fit to all data, the
best lognormal regressed fit to the unconservative lower tail, and the best visual adjusted fit to
the lower tail. Following Allen (Allen, et al. 2005) this last visual adjustment was achieved by
manipulation of the mean resistance and COV from the scenario under consideration to better
match the lower tail and was considered the most superior of all the three fits. Shown in Figure
7.1 is the EOID plot on Scenario A illustrating this technique. Here the improved fit to the tail
reduced the COV and consequently increased the significantly for the 2.33. Lowering the
COV had the visual effect of rotating the line counterclockwise up from the regressed fit to all
data line and produced a higher resistance factor. For each scenario at EOID and BOR, the
statistical parameters were modified and the Monte Carlo procedure repeated each time until the
minimum target of 2.33 was confirmed for the three fits.


Figure 7.1: Scenario A Monte Carlo reliability method CDF fits at EOID
The study of Scenarios F and G were of more significance for ODOT implementation, as they
represented the likely match to practice (single without considering soil type) especially at
EOID. The three lowest data point case-by-case values on Scenario F, in both EOID and BOR
conditions (shown in Appendix D) all came from sands in Florida where Davissons criteria had
underreported capacities compared to the GRLWEAP capacities. Scenario Gs lower tail in
Appendix E was better populated and conditioned to the lognormal distribution by the additional
25 case histories found from Tier 2b having output Rank 1.
64
For each scenario at both EOID and BOR conditions, a summary of the FOSM and Monte Carlo
and / results for the three curve fits described above, including the cohesionless sites
forming Scenario J for comparisons, is shown in Table 7.3.

Table 7.3: All scenarios FOSM and Monte Carlo fit based and / efficiencies for = 2.33
FOSM Monte Carlo
Model and Curve Fit
#
Cases
Mean
S.D. COV / /*
EOID 175 1.555 1.102 0.708 0.35 0.23 0.36 0.23
Fit to All
Data
BOR 175 0.993 0.468 0.472 0.38 0.38 0.39 0.39
EOID - 1.309 0.541 0.413 - - 0.59 0.38
Fit To
Tail
BOR - 0.993 0.468 0.472 - - 0.39 0.39
EOID - 1.380 0.650 0.471 - - 0.54 0.35
Scenario
A
Best
Visual Fit
BOR - 0.910 0.410 0.451 - - 0.39 0.39
EOID 69 1.406 0.596 0.423 0.59 0.42 0.62 0.44
Fit to All
Data
BOR 79 1.008 0.427 0.424 0.42 0.42 0.45 0.45
EOID - 1.380 0.558 0.405 - - 0.64 0.46
Fit To
Tail
BOR - 0.978 0.386 0.395 - - 0.46 0.46
EOID - 1.380 0.610 0.442 - - 0.59 0.42
Scenario
F
Best
Visual Fit
to Tail
BOR - 0.960 0.410 0.427 - - 0.42 0.42
EOID 94 1.328 0.564 0.425 0.56 0.42 0.59 0.44
Fit to All
Data
BOR 114 0.985 0.430 0.437 0.40 0.41 0.43 0.44
EOID - 1.308 0.532 0.407 - - 0.6 0.45
Fit To
Tail
BOR - 0.985 0.430 0.437 - - 0.43 0.44
EOID - 1.280 0.550 0.430 - - 0.57 0.43
Scenario
G
Best
Visual Fit
BOR - 0.960 0.430 0.448 - - 0.41 0.42
EOID 43 1.464 0.595 0.406 0.64 0.44 0.68 0.46
Fit to All
Data
BOR 56 1.123 0.457 0.407 0.49 0.44 0.51 0.45
EOID - 1.251 0.335 0.268 - - 0.80 0.55
Fit To
Tail
BOR - 1.049 0.382 0.364 - - 0.52 0.46
EOID - 1.230 0.300 0.244 - - 0.83 0.57
Scenario
I
Clay &
Mixed
Best
Visual Fit
BOR - 1.080 0.450 0.417 - - 0.49 0.44
EOID 51 1.21 0.515 0.424 0.51 0.42 0.55 0.45
Fit to All
Data
BOR 58 0.85 0.358 0.421 0.36 0.42 0.39 0.46
EOID - 1.19 0.486 0.409 - - 0.51 0.42
Fit To
Tail
BOR - 0.84 0.339 0.406 - - 0.39 0.46
EOID - 1.17 0.510 0.436 - - 0.55 0.45
Scenario
J
Sands
Best
Visual Fit
BOR - 0.82 0.360 0.439 - - 0.36 0.42
* All Monte Carlo efficiencies are calculated using the mean bias fit to all the data in that scenario

The superior visual fit to the tail data are highlighted in each scenario and represents the highest
degree of confidence in the modeling process between the three cases. (Since the two tail fits did
not use all the available individual cases history values, the number of case histories was not
displayed.)
65
Data points to the left of the CDF log normal fit line were viewed as unconservative relative to
the curve. The best visual fit to the tail allowed for a more accurate model the lower tail section

without allowing an anomalous data point to skew the overall trend. It can seen from Figure 7.1
that at EOID for Scenario A, a single low, rogue, data point dominated the lognormal
regression with all the data. This pile case had been placed in a low input Tier 2b, ranked very
low in an output Rank 3, and with noted suspicions of incorrect EOID blow count, possibly not
properly recorded in the original database. The much better visual tail fit raised the calculated
resistance factor compared to all other scenarios significantly, as shown in Table 7.3, up to 0.54.
The Scenario A FOSM at EOID had been depressed severely by this single data point. The
BOR on Scenario A data, shown in Appendix C, was already well fitted to the lower tail with
little change to mean and COV being required. The creation of cohesive sites in Scenario I
permit a better targeted EOID and BOR calibration applicable to ODOT for high set-up sites.
This better targeting was also true for Scenario I in the EOID plot where a large reduction in
COV was possible to fit the lower tail at EOID, with corresponding improvements in the
resistance factor. However, few case history data points were present in the lower tail to count
this at the same degree of confidence compared to other EOID scenarios. Other observations
were:
- Studying / efficiencies of the Fit to All Data with the exception of the large
improvement in moving to BOR in Scenario A, almost no change was found indicating
no clear improvement, or decay, in statistical efficiency by moving to a restrike.
- Scenario F and G were comparable and illustrated no decay in statistics by including
those soils in Tiers 2b from all sites where assumptions were required to complete
DRIVEN and GRLWEAP. Comparing Monte Carlo to FOSM calibrations for Fit to All
Data in these scenarios showed improvements in between approximately 6% to 7% at
EOID and 5% to 6% at BOR.
- In all scenarios for the Fit to All Data, the presence of a few individual high values at
EOID from high set-up soil sites raised the COV and degraded the FOSM and the
Monte Carlo . The advantages in Scenario A of the Monte Carlo fit to the lower tail was
illustrated by removing these few data points, which showed a corresponding increase
of 50% from 0.35 by FOSM to 0.54 with the Monte Carlo Best Visual Fit to the tail
method.
- The Best Visual Fit to the tail showed little change in Scenario G and F compared to
Fit to All Data and confirmed a well-conditioned lognormal fit, but did have a 22%
improvement in at EOID for the cohesive sites in Scenario I.
- Both Scenario G and F had very similar EOID and BOR efficiencies and appeared
GRLWEAP capacity neutral. However, both the cohesive Scenario I and Sand Scenario
J revealed higher EOID efficiencies.
- Increases in factors were found for both EOID and BOR when the higher input quality
tiers were applied in Scenario F and G compared to Scenario A.
66
Design efficiency is most often reported by the / ratio when comparing methods to establish
the highest usable capacity under LRFD design principles. This quickly permits a comparison
between scenarios and EOID and/or BOR choices for design. However, this efficiency measure
can be skewed by extreme data points and should be viewed as an average for the data set.

7.3 LRFD FACTORED CAPACITY OPTIMIZATION
Implementation decisions are made on a project-by-project basis by ODOT and make use of
design office issued bearing graphs keyed to the field hammer/pile performance match observed
by the inspector. A choice must be made on underperforming piles during initial driving: either
to continue driving deeper for increased capacity or to delay and monitor restrike blow count
after approved delay in those soils expected to set-up.
It was reported previously that for all scenarios, at BOR was less than at EOID. The
GRLWEAP capacity most often increases on restrike, simply because of the blow count
increase. However using the resistance factor at the BOR conditions, that is lower than the EOID
condition, it is possible that the calculated factored BOR resistance is lower than the factored
EOID condition. The LRFD inequality was given in Equation 1-1 and is repeated here:

If the load factors are held constant on the left hand side of the equation for the purpose of
comparing different nominal capacity methods then the foundation engineer, to minimize cost,
seeks the maximum on the right hand side of the equation. This has been addressed in previous
work by simply comparing efficiency measurements, / , or just seeking the lowest COV.
However, both run a risk of being skewed by a limited number of extreme case histories in the
sample, especially when no outliers have been removed, such as the +/- 2 s.d. used in NCHRP
507.
Table 7.3 shows the final statistical parameters and number of cases for ODOT final Scenarios F
and G, and Scenarios I and J in cohesive and cohesionless soils, respectively, compared to
baseline Scenario A, at both EOID and BOR. For ODOT Scenario F and G, for all soil and pile
types, these efficiency results were not conclusive between EOID or BOR as each / value
was similar and each factored capacity, on average, was neutral between EOID and BOR.
67
To offer a clearer representation and assist interpretation of the / efficiencies, each case
history was viewed as a data point around the factored equivalency relationship, R
kn
(BOR) =
R
kn
(EOID). This presentation is found in Figures 7.2, 7.3, and 7.4, using the most appropriate
Monte Carlo best visual fit to tail , broken out into Clay, Sand, and the Mixed soil sites
respectively for the Scenarios A, F, and G.


Figure 7.2: Factored BOR plotted to factored EOID for Scenario A

68
Figure 7.3: Factored BOR plotted to factored EOID for Scenario F


Figure 7.4: Factored BOR plotted to factored EOID for Scenario G
69
These figures can be compared to Figure 5.1 and nicely illustrated that the range of factored
capacity was not equally distributed but was at the lower end for each scenario, with few piles of
GRLWEAP factored capacity beyond 1000 kips. The role played by the likely higher set-up
cohesive soils was the rationale for the creation of Scenario I and can be presented in a way to
assist in making field decisions. For example, at the time of initial driving the pile was a sample
member of Scenario G and, when checked for LRFD factored capacity, R
kn
(EOID), might be
found deficient. By choosing to restrike, the pile now was a member of Scenario I and the
calibration at BOR was now valid (Table 7.3). Figure 7.5 plots Scenarios G and Is factored
resistances, R
kn
, from their respective EOID and BOR and predicted GRLWEAP EOID and
BOR capacities, R
nk
, on each of the 56 case histories contained in Scenario I at restrike. In a
similar approach, Figure 7.6 plots the corresponding Scenario G factored EOID resistance to
Scenario J factored BOR resistance set for cohesionless soils.

70

Figure 7.5: Set-up implementation on cohesive soils factored BOR Scenario I to factored EOID for Scenario G

Figure 7.6: Set-up implementation on cohesionless soil factored BOR Scenario J to factored EOID for Scenario G
Judging by the larger number of piles in the database above the R
kn
(EOID) = R
kn
(BOR)
equivalency line in Figure 7.5, on average for these two scenarios in the database, the factored
capacity for Clay and Mixed soil sites was more likely to increase on restrike. However, it
should be made clear any decision on whether or not to restrike should be made on a case-by-
case basis depending upon the expectation that set-up will occur. These decisions are to be
furthered explored for ODOT specific practice in a separate implementation Phase 3.

8.0 CONCLUSIONS AND RECOMMENDATIONS
For this research, a diverse but complimentary group of existing databases were accessed and
merged and then cross-checked for anomalies. Over 150 new cases were added to the ODOT
supplied PDLT2000 and DFLTD databases to establish a new Full PSU Master database with
322 piles. Each case history was placed into one of three input tiers for statistical profiling to
assist in preserving quality and repeatability. The final Full PSU Master database supplied 179
cases. All cases were fully analyzed by DRIVEN and GRLWEAP, version 2005 with all updates,
to generate bias mean and COV statistics for a range of scenarios and for resistance factor
calibration. These scenarios were constructed to study the effects from NCHRP 507 removal of
arbitrary outliers in the tails, the easy driving mismatched cases, restrike and load test calendar
times, and the difference between cohesive and cohesionless soils. The 322 piles ranged up to 40
inches in diameter and up to approximately 200 ft in embedment length. The 179 analyzed piles
ranged up to 36 inches in diameter and approximately166 ft in embedment length with driving
blow counts up to 100 BPI.
8.1 CONCLUSIONS
It was previously stated that a large variation in installation condition requirements and field
practice norms existed in the database due to the differences in time to restrike (TR) time to
static test (TST) and the geographic location of the case histories. The inherent variability of
national piling standards was captured by keeping a broad database to produce more
representative, generally applicable, and conservative resistance factor results.
From database construction and statistical profiling, it was concluded:
- The qualified 179 cases in the expanded Full PSU Master represented a broad spread of
soil and pile types typical of ODOT practice. The sand sites represented approximately
54% of the cases, and 46% of the sites were considered to have significant cohesive soil
contributions.
- Locating a considerable new set of case histories to expand the database allowed
researchers to capture and record the calendar and clock times for restrike and the load
test on 69 cases. This proved valuable for study purposes and is recommended for all
future database building efforts.
71
- Subsurface information was generally poor across the source databases and caused over
75% of the 179 cases analyzed to be placed in lower input Tier 2, which required some
assumptions for analyses. Sensitivity analysis proved GRLWEAP capacities for these
piles were relatively insensitive to the missing detailed input, including water table and
shear strength, provided that at least SPT data and the soil log were available.

- Input tiers and Output ranks proved an invaluable aid to sorting data quality and assisting
case-by-case interpretation and should always be employed in LRFD calibration.
- The single largest source of anomalies and missing data was the field-reported blow
count, especially for the BOR condition. The restrike blow count had no clear definition
across the source databases. For the few cases when BOR was absent but the driving log
was available, a definition consistent with FHWA of averaging the blows for the first 6 in
was used. This provided sufficient movement to capture considerable set-up capacity and
mobilize shaft friction.
- Neither of the two largest source national databases, PDLT2000 or DFLTD, was always
correct on every piece of the large amount of source input required for DRIVEN and
GRLWEAP. For any database calibration work, independent verification from original
sources was recommended. From extensive cross checking in this research of the two
national databases, DFLTD was considered to be the more accurate and more useful as it
contains detailed soil logs.
The 175 qualified cases included 91 cases extracted from DFLTD and PDLT2000 matching the
original 99 reported in NCHRP 507s appendix. Key general guidance notes from the DRIVEN
and GRLWEAP analysis used in this research were:
- All default parameters were used in both EOID and BOR models to establish the
GRLWEAP bearing graph. The GRLWEAP restrike options and the set-up and gain/loss
factors were never used.
- FHWA endorses the use of a used cushion in BOR modeling with concrete piles. They
recommend a used pile cushion of half the thickness and twice the stiffness (Hannigan, et
al. 2006). This recommendation was always followed.
- A weighted average by percent contribution to resistance from DRIVEN for the side
damping, j
s
, in layered soils was always used. Damping, quake, and percent side friction
was held constant between EOID and BOR models.
- A uniform first restrike blow count was selected for those piles containing multiple
restrikes.
72
NCHRP 507 gave considerable attention to the needs of dynamic capacity evaluated by the
CAPWAP-based signal matching technique to establish implementation recommendations using
the efficiency ratio /. These new studies here showed similar trends for GRLWEAP capacity
on the statistical effects from variables such as blow count ranges and the time to restrike when
determining nominal capacity. However, the objective of this research was not to compare to
other calibration efforts, only to NCHRP 507as a reference work of interest. ODOT does use the
PDA/CAPWAP technology for piles under certain conditions at select locations. PDA and
CAPWAP results are used to develop pile acceptance criteria and work hand in hand with
GRLWEAP to reduce uncertainty from variability in site conditions. It should be understood
CAPWAP capacity bias and COV statistics are different from GRLWEAP, and alone should
not be the basis of selecting the most efficient method. Unlike limited piles selected for
CAPWAP, every driven pile has a recorded field blow count and a predicted bearing capacity is
available from the bearing graph generated by GRLWEAP analysis.

Statistically a driven pile may belong to one, or both, of two categories for the purposes of
capacity determination by any dynamic method: at the EOID condition and at the BOR condition
from restrike. Both these categories are shown by all dynamic methods to possess different
mean bias and COVs that change the resistance factor that can, from this research, be used in
either situation for LRFD design. For this research the database qualifications by soil type, pile
type, driving blow count and the load test and restrike times were kept as broad as possible
without eroding the statistical quality of the data for GRLWEAP capacity. The expected higher
set-up capacity gain would be those piles founded in cohesive soils contributing large capacity
from side friction; however, piles in cohesionless soils also show significant set-up.
In this study, Scenario A for EOID was established as a broad match to NCHRP 507 and formed
the baseline to measure to other scenarios of interest and likely ODOT practice scenarios. Using
an input tier structure, key statistical subgroup scenarios were examined. These scenarios applied
filters which included: re-applying the +/- 2 s.d. outlier definition from NCHRP 507, using the
AASHTO time to restrike and time to load test recommendations, and exploring the low blow
count mismatched cases. From the results of these scenarios, it was concluded:
- Across the full database sites defined by Scenario A choosing a restrike to partially
capture the set-up gain in strength reduced the positive bias from EOID and lowered the
COV; both were desirable in providing a more accurate and reliable prediction. For all
scenarios shown in Table 7.2 the restrike lowered the bias and illustrated set-up
occurred in all cases, including the Sand, Scenario J, sites.
- Sub-grouping by blow count revealed a clear decay in easy driving mean bias and COV
parameters with blow counts 2 BPI. Above 2 BPI, little difference was found in these
parameters and no upper limit was identified. This was in agreement with the NCHRP
507 report presentation for CAPWAP (but was reported in the report text to be 4.5 BPI).
- In the database, the time to restrike (TR) and time to static load test (TST) were known
on 64 of the 179 piles analyzed. Of these, only 19 had multiple pile restrikes; therefore,
few candidates were available to choose the option of the BOR closest to TST. A
uniform first BOR designation was adopted for consistency and also to avoid changes in
soil properties and repeat consolidation effects on the 19 piles, which would have tainted
the restrike data.
- A study of TR/TST and TST/TR ratios offered valuable insight to assist interpretation
and identify possible mismatched case histories. Premature BOR restrike times captured
insufficient set-up gain, had low capacity and high , and were ignored by the Monte
Carlo Reliability method option focused on the lower tail. However, they did feature in
FOSM calibration by distorting the calibration statistics, especially in EOID for Scenario
A.
- Both GRLWEAP and DRIVEN manuals must be very well understood by foundation
practitioners and, in addition, all the parameters for which the LRFD calibration was
conducted. Recall that recalibration efforts in all scenarios used default parameters.
73
- The diverse Scenario A Best Visual Fit yielded the EOID factor of 0.54 from 175
cases, including 18 exhibiting relaxation.

ODOT practice is quite diverse and often project-specific with dependency on private-sector
consultants and subcontractors for site investigation, pile design, and field monitoring. This
diversity required keeping case history scenario definitions as broad as possible in order to be
inclusive, rather than narrowing the parameters for the sole purpose of improving COV to raise
the value. Narrowing the scenario data range of qualifications has the penalty of generating a
larger number of pile/soil/driving combinations with each requiring factor, which complicates
implementation and reduces the number of cases available for the statistics. The principal
division in this study between EOID and BOR conditions for ODOT implementation came from
separating cohesive sites from the cohesionless sites. Three likely practice scenarios are of
interest to ODOT:
1. Scenario F used the higher Tier 1 and Tier 2a input quality only.
2. Scenario G used Tier 1 and Tier 2a from Scenario F plus those in Tier 2b with output
Rank 1 showing insensitivity to assumptions.
3. For restrike, Scenario I used those sites where cohesive soils generated greater than 20%
of predicted capacity based on DRIVEN and were likely to display considerable set-up.
These final three scenarios, and the Scenario J created for sand sites, were statistically explored
by FOSM and rigorous Monte Carlo reliability methods in addition to the control Scenario A.
Conclusions from the FOSM and Monte Carlo reliability method calibrations for these scenarios
were:
- Generally, gains from the FOSM resistance factor to the more accurate Monte Carlo
method factor were 4% to 6%.
- Expressed by s.d. changes a tighter statistical distribution and a lower bias value was
found for the BOR condition compared to EOID for all scenarios.
- Extreme outlier data points inflated the FOSM statistical COVs which can be better
modeled by the lower tail Monte Carlo fits.
- The choice of conducting a restrike on a pile was shown in all cases to reduce the
compared to EOID, except for FOSM in Scenario A.
- Often the gain in restrike resistance, R
nk
, was somewhat negated for LRFD compliance
by the reduction in the resistance factor, since the same Davisson capacity is used as the
target in calculating the bias.
- Application of the superior best visual fit to the tail in the Monte Carlo method focused
on the higher risk unconservative predictions of capacity. This showed a 51%
improvement in the baseline Scenario A at EOID up to 0.54 over the poor FOSM of
0.35 on all 175 cases.
74
- At the EOID condition, ODOT Scenarios F and G were very comparable, irrespective of
which method was used for , but with the advantage that Scenario G had approximately

40% more cases compared to Scenario F, and both were superior to in Scenario A. At
the BOR condition, all scenarios were similar using Monte Carlo methods.
- Scenario G represented the broadest and best inclusive ODOT category for all piles in all
soils with 94 case histories used at EOID and 114 used at BOR (20 were removed at
EOID by the field blow count > 2 BPI requirement).
- The range of Monte Carlo values from the three fits for Scenario G at EOID was 0.57
to 0.6, and the most conservative value arose from the visual tail fit. It is possible other
state DOTs standards of practice using in-house databases may fit the upper end of this
range, or may be even higher.
- Cohesive soil sites in Scenario I yielded Monte Carlo best visual fit EOID and BOR
resistance factors of 0.83 and 0.49 respectively. This Monte Carlo best visual tail fit
raised by 30% at the EOID compared to the FOSM resistance value of 0.64. However,
these EOID safety margin high capacity tail fits, illustrated in Appendix F, did produce
unusually low COVs with few data points but well conditioned lognormal tails. They are
shown to be lowest COVs of any fit in all scenarios and would benefit from more data
populating the tail. The cohesionless sites in Scenario J yielded Monte Carlo based EOID
and BOR resistance factors of 0.55 and 0.36 at EOID and BOR respectively.
- Scenarios I and J represent the only breakout by soil type. Comparing the bias drop
from EOID to BOR shows equally significant time dependant capacity gain in the sand
sites compared to the cohesive sites. This is in agreement with the CAPWAP based soil
specific changes in bias in both NCHRP 507 and the work for Wisconsin DOT (Long,
et al. 2009)
8.2 RECOMMENDATIONS AND FUTURE WORK
This study showed GRLWEAP, under well defined conditions, was a reliable measure of
nominal capacity for a broader array of soil and pile types than had previously been recognized
for EOID and BOR conditions. To provide a more focused ODOT practice group of parameters,
Scenarios G, I, and J, are offered around which to establish implementation recommendations in
a Phase 3. The Scenario I created from Scenario G captured the set-up in cohesive soils by
merging clays and mixed soils to establish the EOID and BOR factors.
Based on the results of Scenarios G the EOID Monte Carlo resistance factor for all soils and
pile types was calibrated to be 0.57, and restrike BOR resistance factor was 0.41. For
implementation most past investigators follow AASHTO stated step increments of 0.05. From
this present research study that leads to recommendations of 0.55 at EOID and 0.4 at BOR. To
recap, Scenario G possessed these features:
75
- Had no outlier definition, abandoned the +/- 2 s.d. NCHRP 507 outlier approach, and had
a number of cases comparable to NCHRP 507.

- One case history was removed from the full dataset after examination of the ratio of
restrike to load test times (TR/TST) in Figure 6.3. This case was in clay, re-struck less
than 24 hours (in violation of ODOT standard), and left 51 days before load test.
- Removed piles driving at and under 2 BPI as mismatched and offered 94 piles at EOID
for calibration and 110 for the BOR calibration. The difference in case histories arose
from cases being removed at EOID with the easy driving mismatch.
- Contained piles in Sands, Clays and Mixed sites, of diameter up to 36 in and 142 ft
embedment length and comprised of closed end pipe (CEP), open end pipe (OEP), H
section steel and prestressed concrete (PSC) sections.
- Counted the best site investigation and case history reporting practices specified by input
Tier 1 and Tier 2a definition.
- Included input Tier 2b cases placed in output Rank 1 since they were shown to be
insensitive on the bearing graph to typical assumptions made concerning the water table
and/or shear strength ranges.
- Kept the source sample as representative as possible to the broad pile and soil types
found across Oregon including CEP, OEP, concrete piles, clays, sands, and mixed soils,
with 8 piles exhibiting relaxation coming from the highest Tier 1 and 2a input quality.
Considerable time elapses between pile installation and the full design axial loading. Those sites
exhibiting set-up will reach close to their maximum capacity after dissipation of excess pore
water pressure in cohesive soils, and by reconstitution of the soil fabric. If restrike is delayed and
sufficient time for the AASHTO recommendation of 75% set-up is allowed to occur, then the use
of effective stress analysis methods for BOR may be appropriate. However, drained analysis in
clays is not supported in DRIVEN but are supported by AASHTO. Fortunately, it has been
reported that dynamic methods do capture the set-up more quickly than static load tests would.
The bias presented in Table 7.3 reveals differences between cohesionless soil sites and the
cohesive soil sites for both EOID and BOR. As expected higher bias is revealed for the cohesive
Scenario I compared to cohesionless Scenario J. Measured by GRLWEAP bias reductions
from EOID to BOR the amount of average strength gain for the sites in each of these scenarios is
comparable.
This research has provided baseline statistics and built a comprehensive database that will permit
more scenario sub-sets to be examined for calibration, including the effects of pile type, use of
field hammer observed and PDA monitored performance, and the derivation of CAPWAP
resistance factors for comparison to those presented here from the Full PSU Master.
This research showed cohesive and cohesionless pile GRLWEAP based capacities belong in
different populations. The mixing of both in a single scenario raises the combined COV and
lowers resistance factors. Any future work should consider separation by soil type. The
following recommendations for future study and final implementation work in a Phase 3 are
offered:
76
- The differences between steel and concrete piles are worthy of additional studies.
Additional scenarios should be created and calibrated to include the effects of

hammers, cushion differences on steel and concrete piles, and field documented hammer
performance. Special attention should be given to the common ODOT field condition of
pipe piles driven into sands.
- Implementation would benefit from a study of the effects to GRLWEAP capacities from
the simpler soil side friction distribution models and the effects of a single side damping,
Js, values specific to either cohesive (Scenario I) soil only and cohesionless (Scenario J)
soil.
- Additional efforts should be made to update the master annually with new data when
available, to locate missing information, and to raise the number of case histories to
better populate the tails and permit their values to be considered. Calibration in new
scenarios should treat cohesive and cohesionless soils as belonging to two different
statistical population sets.
- Those cases that report CAPWAP measured damping and quake values should be
reanalyzed in GRLWEAP with these values, and recalibrated.
- A database of drained soil properties should be compiled. Consultants and agency testing
laboratories should be allocated sufficient resources to measure drained properties on all
soils as well as undrained shear strength for cohesive soils. No reliance should be placed
on the SPT as source data for cohesive strengths.
- The use of CAPWAP and wave trace PDA monitoring technology will likely continue for
ODOT projects. Those cases from the Full PSU Master database with CAPWAP
capacities should undergo calibration for comparison to NCHRP 507. Implementation
decisions around GRLWEAP and associated CAPWAP reported capacities can then be
better understood.
- It is likely that larger pile sizes will become more common. New scenarios should be
created to investigate larger pile diameters effects, pile type, and calibration.
- Custom designed input tiers and output ranks are encouraged for all future calibration
efforts. This study showed the inclusion of less complete source data case histories
produced a degradation of statistical parameters more pronounced at the EOID condition.
- A carefully planned Phase 3 for implementation of this work should be undertaken for
both agency and private sector consultants to understand and adhere to the calibration
controlling parameters and modeling techniques. Decisions around those conditions that
merit a restrike should be identified. Changes to more recent AASHTO code releases can
be examined and incorporated into implementation activities. The PI and research group
should be prominent in preparation and delivery across ODOT regions of this
implementation package, including preparation of a PowerPoint presentation.

77


78


9.0 REFERENCES
AASHTO. AASHTO Bridge Design Specification, 3
rd
Edition, American Association of State
Highway Transportation Officials, Washington, DC. 2004.
AASHTO. AASHTO Bridge Design Specification, 2006 Interim Revision, American Association
of State Highway Transportation Officials, Washington, DC. 2006.
AASHTO. AASHTO Bridge Design Specification, 4
td
Edition, American Association of State
Highway Transportation Officials, Washington, DC. 2007
AASHTO. AASHTO Bridge Design Specification, 2009 Interim Revision, American Association
of State Highway Transportation Officials, Washington, DC. 2009.
Allen, T.M. Development of Geotechnical Resistance Factors and Downdrag Load Factors for
LRFD Foundation Strength Limit State Design. National Highway Institute, Federal Highway
Administration, Report No. FHWA-NHI-05-052. 2005a.
Allen, T.M. Development of the WSDOT Pile Driving Formula and Its Calibration for Load and
Resistance Factor Design LRFD. Report No. WA-RD 610.1. Washington State Department of
Transportation and Federal Highway Administration, US Department of Transportation.
Washington, D.C. 2005b.
Allen, T.M., A.S.Nowak, and R.J.Bathurst. Calibration to Determine Load and Resistance
Factors for Geotechnical and Structural Design. Transportion Research Circular, E-C079,
September 2005, Transportation Research Board, Washington, DC. 2005.
Baecher, G. B. and J. T. Christian. Reliability and Statistics in Geotechnical Engineering, John
Wiley & Sons. 2003.
Bowles, J.E. Foundation Analysis and Design. 5th edition, McGraw-Hill. 1996.
Coduto, D. Foundation Design: Principles and Practices. Prentice-Hall, New Jersey, US. 2001.
Duncan, M. J. Factors of Safety and Reliability in Geotechnical Engineering. ASCE. Journal of
Geotechnical and Geoenvironmental Engineering. Vol. 126, No.4. April, 2000.
Hannigan, P., G. Goble, G. Likins, and F. Rausche. Design and Construction of Driven Pile
Foundations, Volumes 1 and 2. National Highway Institute, Federal Highway Administration,
US Department of Transportation, Washington, DC. 2006.
79
Jackson, B. N. LRFD: Case Studies for ODOT Bridge Pile Foundations. Thesis in partial
fullfillment of the Master of Science degree. Portland State University. 2008.

Long, J. H., J. Hendrix, and D. Jaromin. Comparison of Five Different Methods for Determining
Pile Bearing Capacities. Report #0092-07-04, Wisconsin Highway Research Program.
Wisconsin Department of Transportation. 2009.
Long, J.H., J. Hendrix, and A. Baratta. Evaluation/Modification of IDOT Foundation Piling
Design and Construction Policy. Research Report ICT-09-037. Illinois Center of Transportation.
2009.
Mathias, D. and M. Cribbs. DRIVEN 1.0: A Microsoft Windows Based Program for
Determining Ultimate Vertical Static Pile Capacity. Report No. FHWA-SA-98-074, Federal
Highway Administration, Washington, DC. 1998.
Minitab. Minitab16 Statistical Software, Minitab, State College, Pennsylvania. 2009.
Nordlund, R.L. Bearing Capacity of Pile in Cohesionless Soils. ASCE. Journal of Soil
Mechanics and Foundation Engineering, Vol. 89, SM3. 1963.
Nordlund, R.L. Point Bearing and Shaft Friction in Sand. 5
th
Annual Fundamentals of Deep
Foundation Design. University of Missouri - Rolla. 1979.
Paikowsky, S.G. A Simplified Field Method for Capacity Evaluation of Driven Piles. Report No.
FHWA-RD-94-042, Federal Highway Administration, Washington, DC. 1994.
Paikowsky, S. G. Load and Resistance Factor Design (LRFD) for Deep Foundations. NCHRP
Report 507. National Cooperative Highway Research Program, Transportation Researcd Board
of the National Academies, Washington, DC. 2004.
Pile Dynamics, Inc., GRLWEAP: Wave Equation Analysis of Pile Driving. GRL Engineers
Inc., Cleveland, Ohio. 2005.
Raghavendra, S., C.D. Ealy, A. F. Dimillio, and S.R. Kalaver. User Query Interface for the Deep
Foundations Load Test Database. Transportation Research Record: Journal of the
Transportation Research Board. Transportation Research Board of the National Academies,
Washington, DC. Volume 1755. 2001.
Rausche, F., G. Thendean, H. Abou-matar, G.E. Likins, and G.G. Goble. Determination of Pile
Driveability and Capacity from Penetration Tests, Vol. 1-3. Report No. FHWA-RD-96-179.
Federal Highway Administration, Washington, DC. 1997.
Sabatini, P.J., R.C. Bachus, P.W. Mayne, J.A. Schneider, and T.E. Zettler. Geotechnical
Engineering Circular No. 5: Evaluation of Soil and Rock Properties. FHWA-IF-02-034. Federal
Highway Administration, Washington, DC. 2002.
80
Smith, T.D. and P. Dusicka. Application of LRFD Geotechnical Principles for Pile Supported
Bridges in Oregon: Phase 1. OTREC Report TT-09-01. Oregon Transportation Research and
Education Consortium. Portland State University. Portland. 2009.

Tomlinson, M. J. Foundation Design and Construction. 4
th
Ed. Pitman Advanced Publishing,
Boston, MA. 1980.
81
William, J. L., E.P. Voytko, R.M. Barker, J.M. Duncan, B.C. Kelly, S.C. Musser, and V. Elias.
Load and Resistance Factor Design (LRFD) for Highway Bridge Substructures. Report No.
FHWA HI-98-032. Federal Highway Administration, Washington, DC. 1998.

82


APPENDICES





APPENDIX A:


FULL PSU MASTER DATABASE





(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-1
Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
1
NE HP - 1042 75-73 2.83 8 2.83 2-a 1 CL SP
Mix
302 304 297 408 Delmag D-30
2
NE PSC- 12"sq 65-65 5 5 1.00 2-a 1 CL SP
Mix
525 358 391 437 Delmag D-30
3
NE PSC- 14"sq 65-66 9.17 6 0.65 2-a 3 CL SP
Mix
687.15 378 517 487 Delmag D-30
4 NE CEP - 12.75" 70-66 2.5 5 2.00 2-a 1 CL SP Mix 404.5 292 287 405 Delmag D-30
5 IA HP - 1489 120-118 3.33 10 3.00 2-a 1 SP-SC SC Sand 1881 928 372 536 Kobe K25
6 IA CEP - 14" 100-95 5.83 10 1.72 1-a 1 SP-SC SC Sand 1210 650 390 443 Kobe K25
7 OK CEP - 26" 63.3-61.3 4.58 3.75 0.82 2-b 1 CL CL Clay 743 598 744 657 MKT DE-110C
8 OK PSC - 24"oct 64.3-63.3 4.42 6 1.36 2-b 2 CL CL Sand 1000 760 733 978 MKT DE-110C
9 OK RC - 24"sq 63.3-57 25.3 14.8 0.58 2-b 3 CL CL Clay 868 1700 1843 1555 MKT DE-110C
10 OR PSC- 20"sq 135-126 9.58 92.8 9.69 2-b 2 SP/SM MH Sand 6177 1360 1303 2270 Delmag D 46-23
11 ME CEP - 18" 120-100 1.42 3 2.11 1-a 1 SP SW Mix 809 440 349 518.00 Kobe K45
12 ME CEP - 18" 80.5-71.5 1.42 3 2.11 1-a 1 SP SP Sand 758 408 358 575.00 Kobe K45
13 ME CEP - 18" 60-52 1.33 2 1.50 1-a 1 SP SP Sand 476 342 329 453.00 Kobe K45
14 CO CEP - 12.75" NA-127.7 3.5 3.8 1.09 3 316 KC-25
15 CO CEP - 12.75" NA-127.7 3.67 4 1.09 3 368 KC-25
16 MO CEP - 14" 87-83.5 3 7 2.33 2-b 1 SP GP Sand 518 330 327 429.00 ICE 640
17 MO CEP - 14" 65-61.75 1.42 3 2.11 2-b 1 SP GP Sand 247 209 186 323.00 ICE 640
18 WA CEP - 48" 160-34 19.7 16 0.81 1-c 3 ML SP Mix 4217 1300 1503.5 1443.00 CONMACOC300
19 WA CEP - 48" 158-16 26.75 17 0.64 1-c 3 SW/SM SP Sand 3160 1000 1310 1229.00 CONMACO300
20 AL PSC- 18"sq 67-65-67.5 1.5 7 4.67 1-a 2 SM/OL SW Sand 672 370 290 762.00 Kobe K45
21 AL PSC- 18"sq 77-75 3.5 7 2.00 1-a 2 SM/OL SM Sand 2470 550 509 729.00 Kobe K45
22 AL PSC- 24"sq 67-64-64.8 2.83 6 2.12 1-a 2 SM/OL SM Sand 1904 625 425 660.00 Kobe K45
23 AL PSC- 24"sq 77-75-75.5 6.42 8 1.25 1-a 2 SM/OL SM Sand 4086 817 615 693.00 Kobe K45
24 AL PSC- 36"sq 74-73-73.25 7.67 5 0.65 1-c 3 SM/OH SM Mix 5316 1140 990 937.00 Delmag D 62-22
25 VT HP - 1473 95-75-75.75 4.17 9 2.16 1-a 2 GM GP Sand 1275 315 250 373.00 MKT DA 35B
26 VT HP - 1473 95-90-90.4 2.67 3 1.12 1-a 2 GM GP Sand 1934 345 176 197.00 MKT DA 35B
27 MN HP - 1473 100-96-96.1 1.83 25 13.66 2-b 1 ML SC Sand 655 765 460 880.00 ICE 90S
28 PA Monotube 37-23-23.67 5.42 13 2.40 2-b SC SP Sand 116 243 Delmag D 12
29 KT PSC- 14"sq 75-34.67-34.75 15.5 30.5 1.97 1-c 2 CL SC Sand 222 378 384 450.00 LINKBELT LB 520
30 LA PSC- 24"sq 103-84-84.33 1.67 4 2.40 2-b 1 OL CL Clay 447 400 253 634.00 VULCAN Vul 020
31 Ontario, Canada CEP - 9.6" NA-154.3-154.3 21.33 40 1.88 3 540 BirminghamB-400
32 Ontario, Canada CEP - 9.6" NA-NA-101.1 14 3 366 BirminghamB-400
33 Ontario, Canada CEP - 10.24" NA-NA-64.4 4.23 3 189 ICE 40S
34 Ontario, Canada CEP - 12.75 NA-NA-38.6 50 3 242 Delmag D12
35 Ontario, Canada CEP - 12.75 NA-NA-54 8 3 660 Delmag D 30-13
36 FL PSC- 24"sq 50.4-37.2 9.33 9.33 1.00 1-a 2 SP SP Sand 1107 610 869 962 Delmag D 46-02
37 FL PSC- 24"sq 43.5-28-28.5 5 8 1.60 1-a 2 SP SP Sand 434 453 556 805.00 Delmag D 46-02
38 FL PSC- 30"sq 135-NA-128.6 5 6.67 1.33 1-a 2 SP SP Sand 2784 900 1796 2612.00 CONMACO300E5
39 FL PSC- 30"sq 109-NA-110 3.3 5 1.52 1-a 2 SP SP Sand 2008 820 1310 2250.00 CONMACO300E5
40 Ontario, Canada HP - 1273 60-48.5-48.5 3.6 1.83 0.51 2-a 3 ML SP Mix 279 322 320 180.00 BirminghamB-400
41 Ontario, Canada CEP - 12.75" 52.2-48.2-48.2 1.4 5.56 3.97 2-a 2 ML SP Mix 332 330 144 307.00 BirminghamB-400
42 Ontario, Canada HP - 1273 100.2-90.5-90.5 5.3 10.31 1.95 2-a 1 ML SP Mix 587 612 368 448.00 BirminghamB-400
43 Ontario, Canada CEP - 12.75" 105.4-90-90.5 13.5 16.67 1.23 2-a 2 ML SP Mix 502 600 324 329.00 BirminghamB-400
44 Ontario, Canada T-TIMBER 44.4-41.6 4.9 2.53 0.52 2-a 4 ML SP Mix 379 122 BirminghamB-225
45 Ontario, Canada PSC- 12"sq 50-48 4.1 6 1.46 2-a 3 ML SP Mix 446 402 269 365.00 BirminghamB-400
46 NC PSC - 12"sq NA-NA-44.5 10 3 415
47 PA HP - 1253 70-NA-66 1.4 4.5 3.21 2-b 1 CH CH Clay 394 284 147 300.00 ICE 640
48 NA PSC - 20"sq NA-NA-36 4 3 380
49 NA PSC - 20"sq NA-NA-55 8 3 580
50 NA PSC - 20"sq NA-NA-86 7 3 620
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
(#1). Not in DFLTD. Pile info needed
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
Davisson's at EOID and BOR reported. Similar - EOID used
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
146 piles of BOR Final in PDLT2000 fromWSDOT
Comments
(#1). Not in DFLTD. Pile info needed
(#1)
(#1), (#2), (#3)
(#1), (#2), (#3)
(#1), (#2), (#3)
(#1), (#2), (#3)
Davisson's at EOID and BOR reported. Similar - EOID used
Davisson's at EOID and BOR reported. Similar - EOID used
Davisson's at EOID and BOR reported. Similar - EOID used
Davisson's at EOID and BOR reported. Similar - EOID used
No support of tapered timber piles in GRLWEAP


(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-2
Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
51 NA PSC - 20"sq NA-NA-94 10 3 600
52
NA PSC - 10"sq NA-NA-50 10 3 250
53
NA PSC - 10"sq NA-NA-58 5.5 3 270
54
VA PSC - 54"sq NA-NA-109 7.86 3 920
55 VT CEP - 12.75" NA-NA-99.5 12 3 360
56 Australia RC - 10.8"sq NA-NA-67.6 3.63 18.14 5.00 3 652
57 Australia RC - 10.8"sq NA-NA-67.6 1.95 6.68 3.43 3 558
58 Ottawa, Canada CEP - 9.625" NA-68.9 14 6 0.43 2-b 4 CL Glacial Till 271 Delmag D 30-32
59 FL VC - 24"sq 98-93.2 3.42 4 1.17 2-b 1 OH/ML SW Sand 1173 958 512 756.00 Vulcan 020
60 FL VC - 24"sq 83.9-56.03-58.83 3.3 3 0.91 2-a 1 SM SW Sand 935 870 747 890.00 Conmaco 300
61 FL VC - 24"sq 66.3-54.6 5 8 1.60 1-a 1 SM SW Sand 900 715 645 1056.00 Vulcan 020
62 FL PSC- 18"sq 65.2-61 3.17 6 1.89 2-b 1 SM SW Mix 147 315 245 491.00 VULCAN 010
63 FL VC - 24"sq 69.17-62 4 8 2.00 2-b 1 SM SW Mix 774 524 550 1018.00 Vulcan 020
64 FL VC - 24"sq 98-62 2.91 6 2.06 1-a 2 SC SC Sand 405 812 422 967.00 VULCAN Vul 020
65 FL VC - 24"sq 123.67-103.6 5.1 12 2.35 2-b 1 CH MH Clay 639 808 785 1400.00 Vulcan 020
66 FL VC - 24"sq 121.5-103 6 12 2.00 2-b 1 SC CL Sand 678 976 815 1336.00 Vulcan 020
67 FL PSC- 24"sq 84-NA-84.3 1 10 10.00 2-b 1 SP SC Sand 771 500 190 1195.00 VULCAN Vul 020
68 FL VC - 30"sq 71.1-54 12 12 1.00 2-b 1 SW SP Sand 1322 1250 1064 1193.00 ICE 200S
69 FL VC - 30"sq 106.2-NA-87.7 2 2 1.00 2-b 2 SC CH Sand 1703 1435 521 570.00 ICE 200S
70 FL VC - 30"sq 102-78.4 16 15.33 0.96 2-b 1 SM ML/CL Sand 1305 1515 1474 1755.00 ICE 200S
71 FL VC - 30"sq 101-83.4 3 8 2.67 2-b 1 SM ML/CL Sand 1361 643 637 1345.00 ICE 200S
72 FL VC - 30"sq 106-84.7 4.6 8 1.74 2-b 1 SW ML/CL Sand 1556 917 780 1306.00 ICE 200S
73 FL VC - 30"sq 102-72.9 8.75 20 2.29 2-b 1 SW ML/CL Sand 1450 1463 1130 1871.00 ICE 200S
74 FL VC - 30"sq 106-80 15.17 24 1.58 2-b 1 SP/SC CL Sand 1371 1410 1615 2467.00 ICE 200S
75 FL PSC- 24"sq 80-70.7 5.42 10 1.85 1-b 1 SM/SP ML Sand 958 960 800 1350.00 VULCAN Vul 020
76 Ontario, Canada HP - 1273 120.6-114.4 12 16 1.33 1-c 2 CL/ML SM/ML Mix 659 800 455 477.00 BirminghamB-400
77 Ontario, Canada CEP - 12.75" 108.2-107.2 39 76 1.95 1-c 2 ML SM/ML Mix 412 490 337 345.00 BirminghamB-400
78 Ontario, Canada HP - 1274 NA-20.1 30 20 0.67 3 354
79 Ontario, Canada HP - 1274 NA-25.7 22 10 0.45 3 556
80 Ontario, Canada HP - 1253 NA-25.2 38 25 0.66 3 410
81 Ontario, Canada CEP - 7.063" NA-NA 4 3 140
82 Brunswick HP - 1289 NA-NA-126 3 730
83 Brunswick CEP - 12.75" NA-NA-104 4.67 3 340
84 WI CEP - 12.75" 140.4-123 20.2 48 2.38 1-c 2 OL/MH/CH SW Clay 682 654 343 360.00 VULCAN Vul 200C
85 WI HP - 1263 NA-155.5 0.75 2.5 3.33 1-a 1 OL/MH/CH SW Mix 1097.4 302 100 267.00 Vulcan 010
86 WI CEP - 9.63" 165.92-142 0.83 5 6.02 2-b 1 CL-ML SW Clay 696 360 90 262.00 VULCAN Vul 010
87 WI CEP - 9.63" 145-142 0.83 16 19.28 2-b 1 CL-ML SW Clay 696 660 90 324.00 VULCAN Vul 010
88 WI CEP - 9.63" 154.58-144 0.5 11.36 22.72 1-b 1 CL-ML SW Mix 474 376 60 332.00 VULCAN Vul 010
89 WI CEP - 9.63" NA-139 1.17 10 8.55 2-b 1 CL-ML SW Clay 347 556 120 308.00 VULCAN Vul 010
90 WI CEP - 9.63" NA-NA-140 4.58 32 6.99 2-b 1 CL-ML SW Clay 678 596 256 347.00 VULCAN Vul 010
91 WI CEP - 9.63" NA-NA-156 1.5 30 20.00 2-b 1 CL-ML SW Mix 826 600 167 376.00 VULCAN Vul 010
92 FL PSC - 18"sq 32.78-19.33-20.63 7.6 5.33 0.70 1-c 3 SW SW Sand 212 274 496 410.00 ICE 640
93 FL PSC - 18"sq 34.25-17.4-17.6 9.17 20 2.18 1-c 2 SW SW Sand 132 230 537 758.00 ICE 640
94 FL PSC - 24"sq 92-NA-85.8 7 7.29 1.04 2-b 3 SW SW Sand 1710 855 1178 1400.00 Delmag D 46-32
95 FL PSC - 24"sq 65-NA-61.3 16.7 10.35 0.62 2-b 3 SP/GP SC Sand 846 1671 1678 1793.00 Delmag D 46-32
96 FL VC - 30"sq 121-92.8-94.5 3.36 5.21 1.55 2-b 3 CL SP Mix 2795 1006 1025 1676.00 CONMACO300E5
97 FL VC - 30"sq 116.9-88.3-89.3 3 3.7 1.23 2-b 3 SP SP Sand 2106 1162 935 1356.00 CONMACO300E5
98 FL VC - 30"sq 110.6-80.22-81 5.75 6.25 1.09 2-b 1 SP CL Sand 1244 1114 1775.5 2423.50 CONMACO300E5
99 FL VC - 30"sq 104.5-71.6-78.6 5.083 5.33 1.05 2-b 2 SP SP Sand 4035 1136 1394 1800.00 CONMACO300E5
100 FL PSC - 24"sq 67-57.25-58.5 5 4 0.80 2-b 3 MH SP Mix 792 752 844 689.00 Delmag D 46-32
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2).
(#1), (#2).
(#1), (#2).
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1)
Comments


(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-3
Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
101 FL PSC - 24"sq 62-46.1 7 18.75 2.68 2-b 4 SC/Limestone CL 1827 1066 Delmag D 46-32
102
FL PSC - 24"sq 39-NA-36 9 9.17 1.02 2-b 4 ML/OH
CL
566 Delmag D 46-32
103
SC PSC- 24"oct 66.5-61 6 5 0.83 1-c 1 SC
GW Sand
1276 1140 771 1094.00 VULCAN Vul 320 & 520
104
SC PSC - 16"sq NA-62-63 1.75 5.58 3.19 3

807
105 SC HP - 1489 NA-NA-66 3 897
106 SC OEP - 16" NA-NA-66 3 932
107 SC OEP - 24" 85-81 3.75 12.5 3.33 2-b 1 SW SP Mix 649.47 596 638 937 VULCAN Vul 512
108 SC HP - 1473 85-78-79 1.42 0.67 0.47 2-b 2 SW SP Mix 544 318 300 584.00 VULCAN Vul 512
109 FL PSC - 14"sq 115-90-91 5.67 6.5 1.15 2-b 2 SM-SC SC Sand 2023 842 226 551.00 VULCAN Vul 80C
110 FL CEP - 12.75" 90-83-88.33 2.25 4 1.78 2-b 2 SM-SC SC Sand 439 496 125 236.00 VULCAN Vul 80C
111 LA PSC - 24"sq NA-81.5 0.83 1.75 2.11 3 414
112 LA PSC - 30"sq NA-82-83 1.17 1.92 1.64 3 511
113 LA PSC - 30"sq NA-82-83 1.82 4.92 2.70 3
114 LA PSC - 36"cyl NA-81-82 1.25 2.83 2.26 3 542
115 LA PSC - 36"cyl NA-81 3.93 2.67 0.68 3 540
116 VA PSC- 24" sq 110-NA-105 0.6 6 10.00 2-b 4 SM SW Sand 3642 508 222 1283.00 Delmag D 46-32
117 OH CEP - 12" 40-NA-30 3.8 3.6 0.95 2-b 3 CL-ML CL-ML Sand 121 152 164 160.00 FEC 1500
118 MS HP - 1473 40-40 3 6 2.00 1-a 1 SM-SC SC Sand 182 500 213 350.00 Delmag D 19-32
119 SC PSC- 24" sq 90-90 3 83.3 27.77 2-b 2 CL CL Sand 1097 1066 667 2385.00 VULCAN Vul 520
120 SC HP - 1473 90-91 0.5 83.3 166.60 1-c 2 CL-CH CL-CH Clay 364 619 128 852.00 VULCAN Vul 520
121 SC PSC - 12"sq 91-88 2.6 26 10.00 2-b 2 CL-CH CL-CH Clay 210 360 334 531.00 ICE 640
122 AZ HP - 14117 51.3-50.5 20.4 15 0.74 2-b 3 SC GW Sand 481 1460 836 765.00 MKT DE 70B
123 AZ HP - 14117 65.6-50-51 25.6 6 0.23 2-b 3 CL-SC SC Sand 797 1281 737 455.00 MKT DE 70B
124 AZ CEP - 14" 30.75-22.4 65.3 13 0.20 2-b 3 CL-SC SC Sand 344 721 793 635.00 MKT DE 70B
125 AZ CEP - 14" 42.5-NA-24.8 34.5 16 0.46 2-b 3 SC GP Clay 391 668 585 525.00 MKT DE 70B
126 AZ PSC - 16"sq 32-NA-23 24.3 29 1.19 2-b 2 CL-SC SC Sand 639 952 696 908.00 MKT DE 70B
127 AZ PSC - 16"sq 41.6-19 56.3 12 0.21 2-b 3 SC GP Sand 538 1006 715 623.00 MKT DE 70B
128 WI CEP - 12.75" 161.25-NA-95.1 18 14 0.78 2-b 2 CL-ML SP Sand 74.41 438 266 322.00 Vulcan 010
129 Hong Kong PSC - 19.69"cyl NA-74.5 3.63 9.07 2.50 3 980
130 Hong Kong HP - 12120 NA-97.44 4.38 10.58 2.42 3 1065
131 CA OEP - 24" 50-40.3 0.92 1.25 1.36 1-a 1 SP CL Mix 687.74 685 365 480.00 Delmag D 62-22
132 CA CEP - 24" 64-60 1.5 2.5 1.67 1-a 1 SP CL Mix 1120 726 234 667.00 Delmag D 30-32
133 CA CEP - 24" 64-60 1.25 10.33 8.26 1-b 1 SP CL Mix 1120 885 200 1155.00 Delmag D 30-32
134 NY HP - 1042 NA-109.9 13 34 2.62 3 312
135 FL PSC - 24"sq 68.8-49.2 8 38 4.75 2-b 3 SP CL-ML Clay 1866 786 1093 1892.00 CONMACO C300
136 FL PSC - 20"sq 68-47.29 3.08 28 9.09 2-b 2 SP CL-ML Clay 1301 587 401 1124.00 CONMACO 160
137 FL PSC - 24"sq 60.5-27.4-27.9 11 26 2.36 2-b 3 SP CL-ML Sand 2473 1148 1135 1650.00 CONMACO C300
138 FL PSC- 24"cyl 43.5-26.7 14 16 1.14 2-b 4 SP/Limestone SP 3029 623 CONMACO C300
139 FL PSC - 20"sq 55-45.75-46.2 8 13 1.63 2-b 2 SM SP Sand 1636 583 695 1085.00 Vulcan 510
140 FL PSC - 20"sq 37-NA-36.4 3.92 7 1.79 2-a 1 SP SW Sand 570 368 480 853.00 Vulcan 510
141 FL PSC - 30"sq 101.75-72.6-73.5 4.75 5 1.05 2-a 2 SP SP Sand 3782 754 1105 1480.00 CONMACO C300
142 FL PSC - 30"sq 75-64.4-69 8.3 11.4 1.37 2-a 1 SP-ML SP Sand 3579 908 984 1382.00 ICE 200S
143 FL PSC - 30"sq 65-53.4-59 4.33 5.3 1.22 1-b 2 SM SM Sand 2269 778 732 933.00 ICE 200S
144 FL PSC - 18"sq 53.1-53.1 7.7 9 1.17 2-b 2 SP-SM SP-SM Sand 1201 312 481 563.00 ICE 640
145 FL PSC - 14"sq 76-76 2 10 5.00 2-a 2 SP SM Sand 700 398 210 592.00 ICE 640
146 FL PSC - 14"sq 69.65-69.5 5.55 13 2.34 2-b 2 CL-ML SM Clay 1734.73 360 354 564.00 ICE 640
147 NY CEP - 11.73" NA-NA-65.6 11 3 468
148 Brunswick HP - 1289 NA-NA-102.1 25 3 325
149 WI HP - 1263 NA-155.5 1 2 2.00 1-a 4 SP SM 1240.9 200
150 WI CEP - 14" NA-155.2 1.75 2.08 1.19 1-a 4 SP SM 1962.19 188
Define ASCON cap block properties. Predominant soil is
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
EOID hammer not confirmed. Assumed same as BOR
(#1), (#2)
(#1), (#2)
(#1), (#2)
BOR Hammer - Conmaco 100E5
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
CAPWAP less than 1/2 of BOR Capacity. No Davisson's criteria
No load test result known
Limestone dampening factors unknown
(#1), (#2)
(#1), (#2)
(#1), (#2)
BOR Hammer - VULCAN Vul 010
BOR Hammer - VULCAN Vul 010
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
Limestone dampening factors unknown
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
10 additional piles of multiple BORs in PDLT2000 fromWSDOT
Comments


(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-4
Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
151 FL PSC - 24"sq NA-NA-36 9.17 3 566
152
SC PSC - 24" oct 81.5-78
2.75 2
0.73 1-b 2 SP-SC
Limestone
Sand 1382 512 472 546.00 VULCAN Vul 320 & 520
153
Delft, Holland PSC - 9.7"sq NA-NA-35.8
2.4
3

68
154
Delft, Holland PSC - 9.7"sq NA-NA-60
3.14
3

226
155 Delft, Holland PSC - 9.7"sq NA-NA-60 6.35 3 236
156 CA CEP - 24" 60-56 2.17 1.11 0.51 1-c 3 SP CL Mix 766 694 487 370.00 Delmag D 46-32
157 TX CEP - 16" 67-63.75-68.42 1.33 6 4.50 1-b 1 ML CH Clay 105.49 194 95 252.00 LINKBELT LB 520
158 TX CEP - 16" 7874-79.5 1.5 16.67 11.11 1-b 2 ML CH Clay 182 218 102 295.00 LINKBELT LB 520
159 TX PSC - 16"sq 38-33.7-38 4 7 1.75 1-b 1 SP SP Sand 238.72 248 220 315.00 Delmag D 22
160 TX PSC - 16"sq 20-19.85-21.77 7.08 10 1.41 1-b 2 SP-CL SP Mix 120.47 356 286 330.00 LINKBELT LB 520
161 TX PSC - 16"sq 24-21.1-24.3 3.58 4 1.12 1-b 1 CL SP Mix 448.05 230 199.5 226.00 LINKBELT LB 520
162 MA PSC - 14"sq 97-87.4 29 27 0.93 2-b 1 SC-ML GC Clay 990 520 622 669.00 ICE 660
163 ME HP - 14117 150.25-135.4 13.6 28 2.06 2-b 1 SP-SC GW Sand 2082.34 900 794 906.00 Kobe K 45
164 AL OEP - 12.75" 50-48 4 7 1.75 2-b 2 SP SP Mix 453.13 270 190.3 262.00 Kobe K 13
165 AL OEP - 12.75" 140-86 2.8 3.4 1.21 2-b 2 SP SP Mix 1147.64 760 477 510.00 Delmag D 46-13
166 IA HP - 1489 115-93 3.6 9 2.50 1-b 2 SP SP Mix 1989.69 509 326 454.00 Kobe K 25
167 IA CEP - 14" 80-80 4.2 8 1.90 1-b 2 SP SM Mix 1300.21 375 340 416.50 Kobe K 25
168 OH HP - 1489 120-103 30 22 0.73 2-b 3 CL CL Mix 887.34 601 757 731.00 Vulcan 512
169 OH OEP - 18" 120-104 15 35 2.33 2-b 1 CL CL Mix 1684 720 703 787.00 Vulcan 512
170 OH HP - 1253 121-105 4.3 20 4.65 1-b 1 SM CL Sand 536.8 315 351 474.00 Vulcan 506
171 TN PSC - 14"sq 45-26 12 19 1.58 2-b 1 SM SC Sand 62 267 513 738.00 Delmag D 19-32
172 TX OEP - 24" 140-133 3.3 210 63.64 2-b 3 SM CH Clay 1007 1514 542 986.00 Raymond R 5/0
173 TX PSC - 20"sq 101.5-98 4.7 52 11.06 2-b 1 SM CH Mix 865 987 708 1294.00 Raymond R 5/0
174 Annacis, Canada PSC- 24"oct 96.79-75.6 8.1 9 1.11 2-b 3 SM SM Sand 1484.64 400 1300 1380.00 Menck MH 96
175 SC PSC - 16"sq 80-79.7 1.2 15 12.50 1-b 2 CL CL Mix 480.1 590 320 1123.00 Vulcan 520
176 FL PSC - 24"sq 85.55-49.5-54.8 7.9 10 1.27 1-a 1 SP SP Sand 2023.04 967 858 1168.00 Vulcan 020
177 WI CEP - 12.75" 161-143 30.8 80 2.60 2-b ML-CL CL 656 Vulcan 010
178 WI CEP - 9.625" 166.17-120 1.2 9.8 8.17 2-b 3 ML-CL CL Clay 244.45 580 155 322.50 Vulcan 012
179 WI CEP - 9.625" 155.42-142 0.9 20.8 23.11 2-b 3 ML-CL CL Mix 726 600 94 329.00 Vulcan 010
180 PA HP - 1274 70-61 100 26 0.26 2-b 3 SM SP Sand 412 580 571 524.00 ICE 640
181 PA HP - 1275 34-28 15 28 1.87 2-b 1 SM SP Sand 78 305 419 460.00 LINKBELT LB 520
182 PA HP - 1057 35-31.25 15 39 2.60 2-b 3 SM SP Sand 75 340 393 344.00 LINKBELT LB 520
183 PA HP - 1274 50-32.75 39 8 0.21 2-b 3 SM SP Sand 111 240 463 362.00 LINKBELT LB 520
184 PA HP - 1057 36-33.85 39 7 0.18 2-b 3 SM SP Sand 86 310 460 416.00 LINKBELT LB 520
185 PA HP - 1057 50-34.84-35.5 48 26 0.54 2-b 3 GM SP Sand 155 367 567 552.00 ICE 640
186 PA HP - 1274 50-35.58 23 35 1.52 2-b 1 GM SP Sand 212 480 611 633.00 ICE 640
187 SC OEP - 54" 90-80 8.5 6 0.71 1-c 3 SW SM Sand 5545.8 2950 3184 3352.00 APE 400u
188 SC OEP - 54" 90-81 2.83 10 3.53 1-a 3 SW-ML SM Sand 1957.67 2460 2890 4406.00 APE 400u
189 SC PSC- 24"oct 90-82.5 3.75 2.25 0.60 2-b 4 SP-CL SP Sand 2713 614 571.00 Vulcan 520
190 SC PSC - 24"sq 77-63.5 26.17 3 0.11 2-b 4 SP-CL SP 1333 1172 473.00 Vulcan 520
191 SC OEP - 24" 85-80.5 3 VULCAN Vul 512
192 Sc PSC - 24"oct 90-79 3 VULCAN Vul 512
193 SC HP - 1489 80-66 80 0.00 3 SC MH 950 CONMACO 100E5
194 SC PSC - 16"sq 80-62 1.75 6 3.43 1-a 2 SC MH Sand 368.25 560 237 683.00 CONMACO 100E5
195 SC OEP - 16" 80-66 80 3 SC MH 970 CONMACO 100E5
196 SC HP - 1473 80-66 3.67 23 6.27 1-b 4 SC MH Sand 199.47 425 617.00 CONMACO 100E5
197 SC HP - 1473 80-66 1 7 7.00 3 SC MH CONMACO 100E5
198 Los Angeles, CA PSC- 24"oct 95-95 19.8 20 1.01 2-b 4 CL-ML SM Delmag D 46-02
199 Boston, MA HP - 1274 90.54-90.54 7 24 3.43 3 CONMACO 160
200 Kontich, Belgium HP - 14142 196.86-196.86 30.7 95.3 3.10 3 Delmag D 36
No load test result known
(#1)
(#1)
No load test result known. Relaxation shown at BOR
No load test result known. Relaxation shown at BOR
(#1)
(#1)
(#3)
(#3)
No load test result known
Not analyzed due to being a reaction pile
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
(#1), (#2)
(#1), (#2)
(#1), (#2)
(#1), (#2)
BOR Hammer - Delmag D 62-22
Report recommended use Davisson's at BOR
Report recommended use Davisson's at BOR
Report recommended use Davisson's at BOR
Report recommended use Davisson's at BOR
Report recommended use Davisson's at BOR
5 piles fromTTI (Texas Transportation Institute) in 1973
25 piles fromProf. JimLong's FHWAdatabase matched with DFLTD
11 piles fromSCreports sent fromJeffrey Sizemore
21 piles fromProf. JimLong's FHWA data only
Comments


(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-5
Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
201 Kontich, Belgium HP - 14143 65.62-65.62 3.2 3 Delmag D36
202
Kontich, Belgium HP - 14144 55.78-55.78 2.9 5.3 1.83 3

Delmag D36
203
Kontich, Belgium HP - 14145 65.62-65.62 5.4 3

Delmag D36
204
China PSC - 31.5"cyl 59.06-59.06 0.6 3.8 6.33 3

Delmag D62
205 Duluth, MN OEP - 9.625" 145.33-145.33 1.5 5 3.33 3 Delmag D36-32
206 NewOrleans, LA CEP - 12.75" 70-70 0.5 3 6.00 3 Vulcan 06
207 NewOrleans, LA T - 16.5/8.5 70-70 0.5 4 8.00 3 Vulcan 06
208 NewOrleans, LA PSC - 14"sq 70-70 0.7 9 12.86 3 Vulcan 06
209 NewOrleans, LA CEP - 12.75" 70-70 0.5 3 6.00 3 Vulcan 06
210 Jakarta, Indonesia PSC - 15.75"sq NA-65.62-47.57 13.5 12.8 0.95 3 IHI 35
211 Mobile, AL G- 120.075 48-48 3.9 5.8 1.49 2-b 4 MH SP 180 CONMACO65
212 Mobile, AL CEP - 12.75" 65-65 3.7 7.1 1.92 2-b 1 MH SP Mix 456 340 189 226.00 CONMACO65
213 Mobile, AL Monotube 60-60 4.5 21 4.67 2-b MH SP 240 CONMACO65
214 Luling Bridge, LA PSC - 24"sq 84-84 0.8 4 5.00 3 Delmag D46-13
215 Luling Bridge, LA PSC - 30"sq 84-84 1.2 2 1.67 3 Delmag D46-13
216 Luling Bridge, LA PSC - 30"sq 84-84 1.9 4.9 2.58 3 Delmag D46-13
217 Luling Bridge, LA PSC - 36"cyl 84-84 1.3 2.8 2.15 3 Delmag D46-13
218 Luling Bridge, LA PSC - 36"cyl 84-84 3.8 2.7 0.71 3 Delmag D46-13
219 Martin/Bertie, NC PSC - 20"sq 62-57 3.33 3 SP SW 400 Kobe K 22
220 Carteret, NC PSC - 24"sq 80-73 3 SP-CL SP 386 CONMACO 160
221 Carteret, NC PSC - 20"sq 65-41 3 SP-CL SP 330 CONMACO 160
222 Dare, NC PSC - 30"sq 95-70 3 SW-CL ML 650 Delmag D 100-13
223 Bertie/Chowan, NC PSC - 30"sq 119-66 3 ML SW 1860 Raymond R 60x
224 Bertie/Chowan, NC PSC - 20"sq 46-32 3 SW SP 510 Conmaco 300
225 Sampson, NC HP - 1253 55-51 3 320 MKT DE 30B
226 Dare, NC PSC - 20"sq 66-54 3 SM SM 390 Vulcan 512
227 OTTAWA CEP - 9.625" NA-62.3 6 18 3.00 2-b 4 CL SW 502 Delmag D 30-32
228 OR HP - 1489 96-61-71 5.17 5.58 1.08 2-b 4 MH SP Vulcan 010
229 Boston, MA PSC - 16"sq NA-142.71 7 11 1.57 2-a 1 ML CL Mix 3183.09 876 688 770.00 ICE 1070
230 Boston, MA PSC - 16"sq NA-122 4 5 1.25 2-a 2 ML CL Mix 3192.05 1034 550 610.00 ICE 1070
231 Charlestown, MA CEP - 12.25" NA-64 5 7 1.40 2-a 2 GW GP Sand 897.74 640 326 360.00 Delmag D 19-42
232 Charlestown, MA CEP - 12.25" NA-74.1 5 8 1.60 2-a 2 GW GP Sand 979.46 607 464 516.00 Delmag D 30-32
233 Delaware PSC - 24"sq 75-66 4.08 10.7 2.62 1-b 1 SM SM Sand 1824.3 1150 782 1598.00 Delmag D 46-32
234 Delaware PSC - 24"sq 75-72 2.83 10.4 3.67 1-b 1 SM SM Sand 1158.89 1300 540 1700.00 Delmag D 46-32
235 Sandpoint, ID CEP - 16" 150.6-148 0.25 2 8.00 2-a 1 SM CL Mix 611.3 430 65 360.00 APE D36-32
236 Pearl Harbor, HA PSC - 24"oct 95-95 3.17 13 4.10 2-a 1 CH CH Clay 969.9 895 303 959.00 HPSI 1000
237 Pearl Harbor, HA PSC - 24"oct 106-105 2.5 2-a 4 CH CH Clay 1161 900 267 HPSI 1000
238 Pearl Harbor, HA PSC - 24"oct 99-98 2.17 2-a 4 CH CH Clay 1403 900 241 HPSI 1000
239 Pearl Harbor, HA PSC - 16.5"oct 66-65.6 3.75 50 13.33 2-a 3 CH CH Clay 515 450 375 892.00 HPSI 1000
240 Pearl Harbor, HA PSC - 24-oct 67-65.6 3.17 30 9.46 2-a 1 CH CH Clay 683.14 415 320 1132.00 HPSI 1000
241 Shanghai, China OEP - 36" 262-259 3 50 16.67 2-a 2 CL CH Sand 13400 3440 1838 2900.00 BSP HA-30
242 Shanghai, China OEP - 36" 262-259 3 50 16.67 2-a 2 CL CH Sand 13400 3777 1838 2900.00 BSP HA-30
243 Salt Lake City, UT CEP - 12.75" NA-NA-114 0.83 26.66 32.12 2-b CH CH 599 Junttan HHK6A
244 Salt Lake City, UT CEP - 12.75" NA-88 0.33 15.84 48.00 2-b CH CH 610 Junttan HHK6A
245 Salt Lake City, UT CEP - 24" NA-108.5 1.833 33.79 18.43 2-b CH CH 842 Junttan HHK6A
246 Salt Lake City, UT CEP - 12.75" NA-76 1 6.02 6.02 2-b CH CH 468 Junttan HHK6A
247 Salt Lake City, UT CEP - 16" NA-75 6.833 20.52 3.00 2-b TBD TBD 880 Junttan HHK6A
248 Salt Lake City, UT CEP - 12.75" NA-89.25 8.33 62.03 7.45 2-b TBD TBD 675 Junttan HHK6A
249 Salt Lake City, UT CEP - 12.75" NA-91 5.95 18.09 3.04 2-b TBD TBD 650 Junttan HHK6A
250 Salt Lake City, UT CEP - 16" NA-66 4.5 36.65 8.14 2-b TBD TBD 700 Junttan HHK6A
(#1)
(#1)
(#1)
(#1)
(#1)
(#1)
(#1)
(#1)
(#3). No hammer cushion data
(#3). No hammer cushion data
(#3)
(#3)
(#1)
(#1)
(#1)
(#1)
(#1)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
Gpile classification unknown
Monotube pile not analyzable
(#1)
(#1)
(#1)
(#1)
(#1)
(#1)
(#1)
(#3)
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
8 piles fromDevelopment of Resistance Factors for Axial Capacity of Driven Piles in North Carolina (2002) by Kim, Kyung Jun
1 pile from"Static or Dynamic Test - Which to Trust?" by Edde, Robert D. and Fellenius, Bengt H. Geotechnical News, Vol. 8, No. 4, December 1990, p. 28.
1 pile from"LRFD: Case Studies for ODOT Bridge Pile Foundations" Jackson, Bethany
4 piles from"Design and Construction of Driven Pile Foundations: Lessons Learned on the Central Artery/Tunnel Project" FHWA June 2006
2 piles fromGRL Report ""Determination of Pile Driveability and Capacity fromPenetration Tests" Vol. 1,2,3 May 1997 FHWA
1 pile from"Static Loading Test on a 45 mLong Pipe Pile in Sandpoint, Idaho" Canadian Geotechnical Journal 2003.
5 piles from"Offshore and onshore loading tests for the Ford Island Bridge, Hawaii" Seki, Mimura, and Smith DFI-98. Vienna, 1998
2 piles from""Installation and Loading Tests of Deep Piles in Shanghai Alluvium"" Pump, Korista, and Scott DFI-1998, Vienna
9 piles from"Measured Pile Setup During Load Testing and Production Piling" Attwool, Holloway, Rollins, Esrig, Sakhai and Hemenway Transportation Research Record 1663 Paper No. 99-1140
Comments


(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-6
Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
251 Salt Lake City, UT CEP - 12.75" NA-35 5.67 8.64 1.52 2-b TBD TBD 740 Junttan HHK6A
252
Ontario, Canada OEP - 12.75" 40-40-41 0.92 1.83 1.99 2-b 4 SW
SW
185 154 Drop 1.81T
253
Ontario, Canada HP - 12 53 45-44-45 1.33 2.08 1.56 2-b 4 SW
SW
177.43 155 Drop 1.81T
254 NewHaven, CT PSC - 14"sq 105-98.5 3.25 42 12.92 2-b 2 SP MH Sand 615 490 556 1073.00 HPSI 2000
255 NewHaven, CT PSC - 16"sq 120-114.5 3.75 46 12.27 2-b 2 SP MH Sand 1045 600 649 1462.00 HPSI 2000
256 NewHaven, CT CEP - 18" 125.5-119 15.25 12 0.79 2-b 3 SP MH Sand 858.5 680 971 907.00 HMC86
257 NewHaven, CT Monotube 100-94.5 2 5 2.50 2-b SP MH 623 HMC86
258 NewHaven, CT CEP - 12.75" 126-120.25 1.75 4 2.29 2-b 1 SP MH Sand 422 358 138 209.00 HMC86
259 NewHaven, CT CEP - 18" 153-141 2.5 2 0.80 2-b 3 SC-CL SC-CL Sand 1468 510 589 534.00 HPSI 2000
260 NewHaven, CT PSC - 16"sq 105-100 1.75 2 1.14 2-b 2 SC-CL SC-CL Sand 1030 640 424 566.00 HPSI 2000
261 NewHaven, CT Monotube 100-90-90.8 1 5 5.00 2-b SC-CL SC-CL 450 HPSI 2000
262 NewHaven, CT CEP - 24" 166-156.9 2.5 3 1.20 2-b 3 SC-CL SC-CL Sand 2654 828 650 700.00 HPSI 2000
263 NewHaven, CT PSC - 16"sq 120-115 2 3 1.50 2-b 2 SC-CL SC-CL Sand 1342 633 480 734.00 HPSI 2000
264 NewHaven, CT CEP - 18" 127-120-121.3 1 4 4.00 2-b 3 SC-CL SC-CL Sand 1096 440 290 833.00 HPSI 2000
265 Choctawhatchee, FL PSC - 24"sq 125-87.2 3 11 3.67 2-b 2 ML SC Sand 1237 807 776 1953.00 Delmag D 62-12
266 Orlando, FL CEP - 12.75" 175-NA-NA 10.17 3 784 Vulcan 80C
267 Cimarron, OK HP - 14117 113-NA-NA 16.67 3 770 MKT DE-110C
268 Socastee, SC PSC - 24"sq 85-NA-NA 19.17 83.3 4.35 3 1095 Vulcan 520
269 Unknown HP - 1253 75-NA-NA 7.5 10 1.33 3 374 ICE 640
270 Unknown HP - 1254 40-NA-NA 10 30 3.00 3 521 ICE 640
271 Unknown HP - 1255 80-NA-NA 12.67 20 1.58 3 378 ICE 640
272 Jacksonville, IL HP - 1253 NA-34.94-42.65 3 5 1.67 1-a 1 CL SM Mix 224 570 218 299.00 Delmag D 19-32
273 Boston, MA PSC - 16"sq NA-107.9 7 8 1.14 2-b 3 CL ML Clay 738 899 833 989.00 HPSI 2000
274 Boston, MA PSC - 16"sq NA-149 16 21 1.31 2-b 1 CL ML Clay 1088.9 854 860 911.00 ICE 1070
275 Kelowna, B.C., Canada CEP - 24" 170-144.7-146 13 19 1.46 2-b 4 ML ML Mix 2655 786 1104.00 Delmag D 62
276 Hampton, VA PSC - 24" 59.1-55 3 SW-ML CH 696 ICE 80S
277 LA OEP - 24" 109.9-53.15 2.17 9 4.15 3 302 Delmag D46-32
278 LA PSC - 24"sq 67.9-39 1.67 16 9.58 3 720 Delmag D46-32
279 LA PSC - 16"sq 77-32.5 0.75 8.33 11.11 3 224 ICE 60S
280 LA PSC - 16"sq 55.1-36.1 0.583 16.67 28.59 3 216 ICE 60S
281 LA PSC - 14"sq 59-43.3 0.83 5 6.02 3 224 ICE 60S
282 LA PSC - 16"sq 70-23 15.8 67 4.24 3 595 ICE 60S
283 Appalachicola Bay, FL PSC - 18"sq 68.2-64 4.17 12.5 3.00 2-b 2 CL CH Clay 179.3 400 360 719.00 CONMACO115
284 Appalachicola Bay, FL PSC - 24"sq 69-64.6 5.42 12.92 2.38 2-b 2 SC CL Sand 539.4 860 811 1504.00 Vulcan 020
285 Dodge Island, FL PSC - 30"sq 110-39.8 11 20 1.82 2-b 4 SM Sandstone 1118.2 1260 Conmaco 300
286 Sao Paulo, Brazil HP - 1293 183-177.16-178.15 5 3 CH CH Free Fall Hammer
287 North Platte River, WY HP - 1253 NA-70-72 13.17 12.75 0.97 2-b 3 SW SW Sand 283.6 280 392 387.00 Mitsubishi MH15
288 North Platte River, WY HP - 1473 NA-85-86 13.75 15 1.09 2-b 2 SW SW Sand 677.3 430 491 504.00 Mitsubishi MH15
289 ElkemMosjen, Norway PSC - 14"sq 128.6-125.7 1.5 11.5 7.67 2-b 4 SP SP Sand 742 Banut Superramrig
290 Jasper County, IN CEP - 14" NA-57.1 4.23 12.7 3.00 3 OH ML 309 ICE 42 S
291 Baton Rouge, LA PSC - 24"sq 45-40.5-41.17 4.7 3 0.64 2-a 3 CL CH Clay 484.7 460 PILECOD36-32
292 Alberta, Canada CEP - 12.75" NA-66.44 4 2.5 0.63 3 207 Hera 1500
293 Alberta, Canada CEP - 12.75" 53.4 4 2 0.50 3 187 Hera 1500
Hammer not in GRLWEAP library
(#2). SLT performed on adjacent pile. Length of pile unknown
(#2)
(#1), (#2), (#3)
(#1), (#2), (#3)
2 piles from"Load Capacity of Pipe Piles in Cohesive Ground" Diyaljee, Vishnu; Pariti, Murthy. Deep Foundations 2002 pp 1318-1334.
1 pile fromLouisiana DOT for Highland Road Job #06-CS-HC-0026 Pile TP-1, Station 57+03.56
(#1)
(#1)
(#1)
(#1)
3 piles fromFlorida Deep Foundations Database
Sandstone properties required
(#2),(#3)
(#1)
(#1)
Used reported BOR Davisson's based on chronology
EOID utilized 3855 kg drop hammer, not in library.
(#3)
(#1)
(#1)
Unknown pile type
(#1). BOR Hammer - Vulcan 010
(#1)
(#1)
(#1)
Drop hammer used. Not in GRLWEAP library
Drop hammer used. Not in GRLWEAP library
Unknown pile type
(#1)
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
2 piles fromMinistry of Transportation, Ontario, Canada
11 piles from""Performance Evaluation of a Large Scale Pile Load Testing Programin Light of Newly Developed LRFD Parameters"" Thibodeau and Paikowsky Geo Frontiers 2005 GSP 131
7 piles fromJimLong's additional FHWAdata not matched to either PDLT200 or DFLTD
1 pile from"Friction Bearing Design of Steel H-Piles" by James H. Long and Massimo Maniaci IDOT ITRC 1-5-38911 December 28, 2000
2 piles from"Design and Construction of Driven Pile Foundations: Lessons Learned on the Central Artery/Tunnel Project" FHWA June 2006
1 pile from"Pile Load Test for W.R. Bennett Bridge"" by Naesgaard, Uthayakumar, Esroy and Gillespie
1 pile from"Axial and Lateral Load Performance of Two Composite Piles and One Prestressed Concrete Pile" by Miguel Pando, George Filz, Carl Ealy and Edward Hoppe Transportation Research Record 1849 Paper No. O3-2912
6 piles from"On the Prediction of Long TermPile Capacity fromEnd-of-Driving Information" by Frank Rausche, Brent Robinson, and Garland Likins ASCE GSP 125 2004
Comments
1 pile from"Load Testing of a Closed-Ended Pipe Pile Driven in Multilayered Soil"
Kim, Daehyeon; Bica, Adriano Virgilio Damiani; Salgado, Rodrigo; Prezzi, Monica; Lee, Wonje Journal of Geotechnical and Geoenvironmental Engineering April 2009 pp 463-473
1 pile from"Comparison Between Static and Dynamic Pile Capacity. A Case Study FromNorway"
Tistel, J., Trum, E., Rnning, S., Alstad, M., Vik, A., SchramSimonsen, A.
2 piles from"Dynamic Pile Monitoring and Pile Load Tests in Unconsolidated Sands and Gravels, Wyoming" Schulte, Michael P. Proceedings of the 37th Annual Highway Geology Symposium, Helena, Montana, August 20-22 1986.
1 pile from"Correlation of CAPWAP with instrumented static load test on a steel Hpile" Paraiso, S.C., Costa, C.M.C., and Aleixo, L. The application of stress-wave theory to piles: science, technology and practice, pg 637


(#1) Incomplete or missing soil data Output Rank 4 does not qualify for recalibration
(#2) Incomplete or missing hammer data C/S/M Major soil contribution. Clay, sand, or mix.
(#3) Incomplete or missing blow count data All lines in bold represent Input Tier 1 & 2

A-7

Pile Length -
EOID
Embedded
Length - BOR
Length
DRIVEN
Capacity
Davisson's
Capacity
EOID BOR1 EOID BOR
(BPI) (BPI) (Kips) (Kips)
294
MI HP - 1253 61-53.4 7 9 1.29 1-b 1 CH
CH
Clay 279 418 184 201.00 Vulcan 50C
295
MI HP - 1254 60-53.2 6 13 2.17 1-b 1 CH
CH
Clay 279 418 174 222.00 Vulcan 50C
296
MI HP - 1255 61-53 4 7 1.75 1-b 1 CH
CH
Clay 279 418 131 181.00 McKiernan-Terry DE-30
297 MI CEP - 12" 80-65.4 3.5 5 1.43 1-b CH CH 775 Delmag D 12
298 MI OEP - 12" 72.75-56.4-NA 3 4 1.33 1-b CH CH 100
299 Oakland, CA CEP - 24" 78.5-70.7 3 SP-CL ML 1000 Delmag D46-32
300 Oakland, CA CEP - 24" 64-53 3 CH CL 756 Delmag D46-32
301 Oakland, CA CEP - 24" 75.4-66.5 3 CH CL 668 Delmag D46-32
302 Oakland, CA OEP - 40" 85.4-73.3 3 SP CL 1176 Delmag D62-22
303 Oakland, CA OEP - 40" 98-83 3 CH CH 1225 HPSI 2005
304 Oakland, CA OEP - 40" 105-91 3 ML-SP GP-CL 1031 HPSI 2005
305 Oakland, CA OEP - 40" 101-86 3 CL-GP CL 710 HPSI 2005
306 Oakland, CA CEP - 24" 64-60 3 CL-SP CL-SP 998 Delmag D30-32
307 Oakland, CA CEP - 24" 45-40 3 CL-SP CL 1000 HPSI 2005
308 Oakland, CA OEP - 24" 64-60 3 CL-GP CL 401 Delmag D46-32
309 Oakland, CA CEP - 24" 33.3-28.3 3 SM SP-SM 552 Delmag D62-22
310 Los Angeles, CA PSC - 14"sq 43-40 3 CL SC 363 Delmag D36-32
311 Oakland, CA OEP - 24" 43-43 3 SP CH 568 MKT 11B3
312 San Diego, CA PSC - 14"sq 34-34 3 SP CH 252 Delmag D36-32
313 San Diego, CA PSC - 14"sq 24-24 3 SP CH 228 Delmag D36-32
314 San Diego, CA PSC - 14"sq 17-17 3 SP CH 251 Delmag D36-32
315 San Jose, CA CEP - 14" 55-49 3 SP CH 340 Vulcan 80 C
316 Castroville, CA OEP - 72" 114-114 3 SP CH 1513 Delmag D80-23
317 San Francisco, CA OEP - 24" 64.5-47.5 3 SP CH 1007 Menck MHF5-10
318 San Francisco, CA OEP - 24" 70.5-51.5 3 SP CH 999 Menck MHF5-10
319 San Francisco, CA OEP - 16" 66.5-52.5 3 SP CH 954 Menck MHF5-10
320 San Francisco, CA OEP - 24" 70-53 3 SP CH 997 Menck MHF5-10
321 San Francisco, CA OEP - 24" 100-79 3 SP CH 209 Menck MHF5-10
322 San Francisco, CA OEP - 24" 74-49 3 SP CH 901 Menck MHF5-10
5 piles fromMichigan DOT
Hammer Type C/S/M
Output
Rank
Input
Tier
Set Up
Ratio
Pile Type Location
PSU
Pile
Number
Predominate Soil
Condition
Blowcounts
GRLWEAP
Capacity
Side Tip (ft) (Kips) (Kips)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3). BOR hammer - Delmag D62-22
(#3). BOR hammer - Delmag D 436-32 & D 62-22
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
(#3)
24 piles from Rollins Brown's M.S. Thesis, Appendix B, 2006.
Comments
???
(#2)



APPENDIX B:

LOGNORMAL DISTRIBUTION EOID/BOR STATISTICAL
CHECKS FOR SCENARIO A, F, G, AND I (BOR)



10.0 1.0 0.1
3
2
1
0
-1
-2
-3
EOI D Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc 0.2884
Scale 0.5469
N 172
AD 2.298
P- Value < 0.005
Lognor mal Pr obabi l i t y Pl ot f or EOI D
Scenario A
95% Confidence Interval
10.0 1.0 0.1
3
2
1
0
-1
-2
-3
BOR1 Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc - 0.1130
Scale 0.4570
N 172
AD 0.548
P- Value 0.156
Lognor mal Pr obabi l i t y Pl ot f or BOR1
Scenario A
95% Confidence Interval

B-1

10.0 1.0 0.1
3
2
1
0
-1
-2
-3
EOI D Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc 0.2418
Scale 0.4324
N 6
AD 0.193
P- Value 0.890
Lognor mal Pr obabi l i t y Pl ot f or EOI D
Scenario F
95% Confidence Interva
5
10.0 1.0 0.1
3
2
1
0
-1
-2
-3
BOR1 Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc - 0.08930
Scale 0.4357
N 7
AD 0.507
P- Value 0.195
Lognor mal Pr obabi l i t y Pl ot f or BOR1
Scenario F
95% Confidence Interva
5
B-2

10.0 1.0 0.1
3
2
1
0
-1
-2
-3
EOI D Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc 0.1890
Scale 0.4140
N 9
AD 0.388
P- Value 0.380
Lognor mal Pr obabi l i t y Pl ot f or EOI D
Scenario G
95% Confidence Interval
0
10.0 1.0 0.1
3
2
1
0
-1
-2
-3
BOR1 Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc - 0.1163
Scale 0.4362
N 11
AD 0.670
P- Value 0.078
Lognor mal Pr obabi l i t y Pl ot f or BOR1
Scenario G
95% Confidence Interval
0

B-3

B-4
10.0 1.0 0.1
3
2
1
0
-1
-2
-3
BOR1 Bias
S
t
a
n
d
a
r
d

N
o
r
m
a
l

V
a
r
i
a
b
l
e
,

z
Loc 0.02450
Scale 0.4255
N 5
AD 0.399
P- Value 0.353
Lognor mal Pr obabi l i t y Pl ot f or BOR1
Scenario I
95% Confidence Interval
2

APPENDIX C:
SCENARIO A STATISTICAL EOID/BOR PDF AND CDF PLOTS




C-1

C-1

C-2

C-2


APPENDIX D:
SCENARIO F STATISTICAL EOID/BOR PDF AND CDF PLOTS




D-1

D-1

D-2

D-2


APPENDIX E:
SCENARIO G STATISTICAL EOID/BOR PDF AND CDF PLOTS






E-1

E-2


APPENDIX F:
SCENARIO I STATISTICAL EOID/BOR PDF AND CDF PLOTS

























F-1

F-2