Report No.

CDOT-DTD-R-2006-7

CDOT FOUNDATION DESIGN PRACTICE AND LRFD

STRATEGIC PLAN

Professor Nien-Yin Chang, P.E.,Ph.D.

University of Colorado at Denver and Health Science Center

February 2006

COLORADO DEPARTMENT OF TRANSPORTATION

RESEARCH BRANCH
The contents of this report reflect the views of the author(s), who is(are) responsible
for the facts and accuracy of the data presented herein. The contents do not
necessarily reflect the official views of the Colorado Department of Transportation or
the Federal Highway Administration. This report does not constitute a standard,
specification, or regulation. The preliminary design recommendations should be
considered for only conditions very close to those encountered at the load test sites and
per the qualifications described in Chapter 6. Use of the information contained in the
report is at the sole discretion of the designer.
1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.
CDOT-DTD-R-2006-7
4. Title and Subtitle 5. Report Date
CDOT FOUNDATION DESIGN PRACTICE AND LRFD February 2006
STRATEGIC PLAN
6. Performing Organization Code

7. Author(s) 8. Performing Organization Report No.

Nien-Yin Chang, P.E., Ph.D. CDOT-DTD-R-2006-7
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
University of Colorado at Denver and Health Sciences Center
Campus Box 113, P. O. Box 173364, Denver, Colorado 80217 11. Contract or Grant No.
PG: 01HQ562
Contract Routing #:99HAA0046
12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered
Colorado Department of Transportation - Research Final Report
4201 E. Arkansas Ave.
Denver, CO 80222
14. Sponsoring Agency Code
Study # 80.12
15. Supplementary Notes
Prepared in cooperation with the US Department of Transportation, Federal Highway Administration
The Colorado Department of Transportation (CDOT) has adopted the AASHTO LRFD Bridge Design
Specifications for the design of all highway bridges. This prompt implementation of the LRFD code was
possible because of the history of LFD and ASD designs and the uniformity of the structural materials, concrete
and steel. The implementation of geotechnical LRFD is hampered by the non-uniformity of the geological
materials, soils and rocks, which requires the evaluation of the state-specific resistance factors. The process of
evaluation and calibration of resistance factors is complex and time-consuming. It is based on the theory of
probability and reliability. It requires the creation of a property database through a long-term data collection,
the evaluation of property statistics, the formulation of probability density function for each property of each
geological material, the determination of probability of failure, and the selection of reliability of index. In
civil engineering designs the failure probability is determined by a code committee and, in the AASHTO code,
the failure probability is determined to be 1/10,000.

Fostering a smooth process for the design of bridges and bridge foundations requires the implementation of
both structural and geotechnical LRFD procedures. The FHWA has set the Year 2007 as the target time for the
full implementation of the AASHTO LRFD Bridge Design Specifications and it is urgent for CDOT to devote
effort to the implementation of geotechnical LRFD, which requires both financial and time commitment from
all concerned parties, particularly the CDOT Central Administration. This study has found that the CDOT
geotechnical practice is severely deficient in manpower and field test equipment. Without immediate remedial
actions the quality of CDOT geotechnical services will be of great concern. The formation of an LRFD
Committee with members from the CDOT Administration, Bridge Branch, Materials Laboratory (Geotechnical
and Soil-Rockfall Programs), and Research Branch; structural and geotechnical engineering communities; and
academia is recommended to strategize the implementation of the geotechnical LRFD procedures.
17. Keywords 18. Distribution Statement
load and resistance factor design (LRFD), allowable No restrictions. This document is available to the
stress design (ASD), load factor design (LFD), public through the National Technical Information
standard penetration test (SPT), finite element analysis, Service, Springfield, VA 22161
bridge design, piers
19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of Pages 22. Price
None None 207
Form DOT F 1700.7 (8-72) Reproduction of completed page authorized
CDOT Foundation Design Practice and LRFD Strategic Plan

Professor Nien-Yin Chang, P.E.,Ph.D.

University of Colorado at Denver and Health Science Center

Report Number CDOT-DTD-R-2006-7

Sponsored by the
Colorado Department of Transportation
In Cooperation with the
U.S. Department of Transportation
Federal Highway Administration

February 2006

Colorado Department of Transportation

Research Branch
4201 E. Arkansas Ave.
Denver, CO 80222
(303) 757-9506
ACKNOWLEDGEMENTS

The opportunity to lead the study on the very important issue of the implementation of
geotechnical load and resistance factor design is highly appreciated and it presented a
great challenge to the Principal Investigator, Dr. Nien-Yin Chang of the Center for
Geotechnical Engineering and Sciences at the University of Colorado at Denver and
Health Sciences Center. A great deal was learned about the strategic plan for the
implementation of geotechnical LRFD at the Colorado Department of Transportation.
This study on “CDOT Foundation Design Practice and LRFD Strategic Plan” came about
during a causal discussion with Dr. Hsing-Cheng Liu, Manager of the CDOT
Geotechnical Program, about the need for the implementation of geotechnical LRFD by
2007 as recommended by the Federal Highway Administration. The discussion
contribution from research panel members, Dr. Hsing-Cheng Liu of the Geotechnical
Program, Mr. C. K. Su of Soils and Rockfall, Dr. Trever Wang of Bridge, and Dr. Naser
Abu-Hejleh and Mr. Richard Griffin of Research and the funding from the Colorado
Department of Transportation and the Federal Highway Administration are gratefully
appreciated.

iii
EXECUTIVE SUMMARY

Load and resistance factor design (LRFD) was adopted by AASHTO to replace the
allowable stress design method (ASD). As opposed to the ASD, where all uncertainties
are accounted for in a factor of safety, the LRFD approach applies separate factors to
account for each uncertainty in load and resistance. This provides a reliable and rational
approach consistent with the safety requirement for highway structural and substructure
designs. In the LRFD method, the external loads are multiplied by load factors, while the
soil resistances are multiplied by resistance factors. A limit state is a condition beyond
which a structural component ceases to fulfill it design function. The limit states, which
must be evaluated in the AASHTO LRFD specification, include strength and service
limit states. The strength limit state ensures that the design procedure provides adequate
resistance (or margin of safety) against geotechnical and structural failures. The service
limit state ensures that the function of the structure under normal service conditions
performs satisfactorily (i.e., deformations is less than its tolerance). Hence, the
foundation design procedure requires the estimation of the nominal response (ultimate
strength and deformation) of the highway foundation when subjected to loading.

Foundation deformations can be evaluated from in-situ load tests and analytical methods
(e.g., the finite element method and simple geotechnical analysis). Since the evaluation
of foundation displacements by LRFD are performed in accordance with the service limit
state, where load and resistance factors are both equal to unity, the methodologies used to
estimate settlement and lateral deflection are identical for LRFD and ASD. The
implementation of LRFD in CDOT design procedure with strength limit state requires 1)
proper evaluation of soil strength and deformability, 2) establishment of the soil property
database, and 3) evaluation and calibration of resistance factors.

Some of the methods employed in the prediction of the ultimate soil resistance are
empirical, e.g., friction angle based on standard penetration test (SPT) blow counts,
whereas others are rational based on classical theoretical soil mechanics (bearing capacity
based on measured soil friction angle and cohesion). CDOT uses an empirical formula,
Denver Magic Formula (DMF), to estimate the nominal strength of soils and rocks in the
pier design as follow:

Qallowable = 0.5 N kips/ft2 assuming a factor of safety of 2.5.

The formula has been adopted extensively by the Denver geotechnical community in the
design of drilled shafts and driven piles for many decades starting with F. H. Chen
(1988). CDOT has adopted it in the deep foundation designs for many years and there
has been a lack of documented drilled shaft failures. This implies that the method might
be somewhat conservative for the deep foundation design in Colorado and some
calibration is needed for the implementation of the method in the LRFD deep foundation
design in order to evenly apply the risk factor to the different strength contributing
factors. Because of the nature of SPT tests, the blow count most likely reflects the
undrained shear strength of soils or soft rocks. It is important to note that CDOT is not
the first one to apply this method in assessing the bearing capacity of deep foundations.

iv
In fact, the following formulas for the allowable end bearing capacity were proposed by
different researchers and practitioners assuming a factor of safety of 2.5:

1.5 N ksf (Meyerhoff, 1956),

1.0 N ksf (Terzaghi and Peck, 1967),

0.5 N ksf (Reese, Touma, and O’Neill, 1976), and

0.37 N ksf (Strounf and Butler, 1975)

The current DMF adopts a formula similar to the one proposed by Reese, Touma and
O’Neill. This formula gears towards ASD, for it predicts the allowable soil and rock
resistances using the SPT blow count (N) alone. There is nothing wrong with
implementing DMF in the LRFD foundation designs. However, to implement DMF in
the LRFD bridge substructure design in Colorado requires the calibration against
Colorado soils and rocks. Two possible approaches are suggested. While it deviates
somewhat from the recommended FHWA/AASHTO LRFD design approach, it can be
calibrated and improved for use in deep foundation designs with data from PDA (pile
driving analyzer) and full-scale load tests. For CDOT foundation design practice, two
approaches are recommended:

First, stay the course of using DMF. This approach may exert less impact to the current
CDOT foundation design practice. However, it still needs systematic calibration. A
significant number of load tests are required to generate a database for the ultimate
strength of Colorado piers and soils and rocks. The data should include SPT blow count
N and soil/rock properties including index properties, strengths, and compressibility.
When a sufficient database is established, the strength and compressibility can be related to
the blow count, N, and, if necessary, index properties of soils and rocks and used in
predicting the bearing capacity and settlement for different foundations. This field
ultimate foundation capacity can be compared with the nominal foundation capacity
calculated from analytical methods for the purpose of evaluating the resistance factors.
Additionally, the correlation between the N value and ultimate strength can be used for a
more reliable N value-based design method.

Second, adopt the FHWA/AASHTO recommended LRFD design practices with an

attempt to evaluate the Colorado-specific resistance factors. While the approach allows
us to gain some technical support from other states, some changes in geotechnical
practice in CDOT are needed. This requires new field and laboratory equipment for the
evaluation of the strength and deformability of Colorado soils and rocks and the
associated resistance factors because of the distinctive characteristics of the subsurface
material in the Colorado. Initially, the values of the resistance factors recommended in
the AASHTO design specifications can be used, while a designer needs to exercise great
caution by taking the unique properties of the Colorado soils and rocks into account.
During the transition period the LRFD design can be compared to the ASD. Eventually

v
Colorado needs to have its own resistance factors for geotechnical and bridge
substructure designs.

In sum, the implementation of geotechnical LRFD is logical and unavoidable. The

current DMF-based design needs to move toward the LRFD-based DMF design during
the transition and eventually the AASHTO LRFD procedures. Internally CDOT has
already adopted the AASHTO LRFD Bridge Design Specifications in the Bridge Branch.
To allow a smooth design process of bridge superstructures and substructures requires the
implementation of geotechnical LRFD. Technically, the LRFD procedures are more
rational than the ASD and LFD alternatives, for they apply a uniform risk to all design
factors. The FHWA has recommended that all state DOTs implement the LRFD
procedures by 2007.

Whichever approach CDOT chooses to enforce, the complete implementation requires

significant monetary and time investments. New subsurface investigation procedures,
laboratory and field testing equipment, and full-scale foundation load tests are needed.
This laboratory/field testing equipment and full-scale testing will provide the information
for the database needed for the LRFD implementation. Additionally, a geotechnical
LRFD training is needed to facilitate a smooth transition from ASD to LRFD. Careful
planning will minimize the potential delay of project delivery. This may require,
initially, a parallel design effort using both ASD and LRFD procedures to avoid any
delay in project delivery and also provide information for comparison and calibration.

vi
IMPLEMENTATION PLAN

To successfully implement the LRFD procedures in geotechnical investigation and

design, the following steps are strongly recommended:

1. Establish a geotechnical LRFD Committee, a think-tank group, to formulate the

strategy and plan for an effective implementation. The committee membership
should include some responsible persons from the CDOT Central Administration,
Bridge Branch, Geotechnical Program, and Soil-Rock Fall Program; one
representative from each of the structural and geotechnical consulting industries;
and a couple from academia knowledgeable in LRFD, probability and reliability.
2. Hire additional geotechnical engineering and investigation staff required for the
quality service delivery, public safety and implementation of LRFD procedures in
geotechnical design and investigation.
3. Create a position(s) for foundation engineers, who are responsible for all
foundation designs. The person(s) should have the overlapping capability of
structural and geotechnical engineers to effectively communicate with both
geotechnical and structural staff.
4. Upgrade the field investigation facility by purchasing CPT, PMT, GE, and VST
equipment.
5. Activate these excellent laboratory testing apparatuses and place them in a
production line to generate urgently needed laboratory data of soils and rocks for
the formulation of resistance factors for Colorado-specific geological materials.
6. Select deep foundation as the first foundation type for geotechnical LRFD
implementation and formulate detailed procedures for the evaluation of all
necessary state-specific resistance factors for all geological materials.
7. Continue to practice DMF but establish an enhancement program for establishing
resistance factors and calibration, while exploring the feasibility of using other
alternative design methods.
8. Develop in-house full-scale load test capability to check the design
recommendation and calibrate the geotechnical LRFD recommendation.

vii
TABLE OF CONTENTS
1. Introduction..........................................................................................................1
1.1 History of LRFD Development ...........................................................................1
1.2 Significance of the Implementation of CDOT Geotechnical LRFD
Procedures............................................................................................................1
1.3 Research Goal and Objectives .............................................................................2

2. Fundamentals and Development of LRFD Specifications ...............................3

2.1 Development History of LRFD Specifications....................................................3
2.2 ASD versus LRFD ...............................................................................................5
2.2.1 Allowable Stress Design………………………………………………………5
2.2.2 Load and Resistance Factor Design (LRFD) ....................................................5
2.3 Fundamentals of LRFD .......................................................................................5
2.3.1 Probability Density Function, Probability of Failure, Reliability, and
Reliability Index ...............................................................................................5
2.3.2 Limit State.........................................................................................................8
2.3.3 Bias Factor ........................................................................................................8
2.3.4 Calibration.........................................................................................................9

3. Survey of the Status of Geotechnical LRFD Implementation at

Different State DOTs............................................................................................10
3.1 Introduction..........................................................................................................10
3.2 Survey Questionnaires .........................................................................................10
3.3. Survey Results and Discussion ...........................................................................10
3.3.1 Staff...................................................................................................................10
3.3.2 Laboratory and Field Testing and Full-Scale Load Test Facilities...................11
3.3.3 Geological Materials.........................................................................................12
3.3.4 Geotechnical Investigation and Design Practices .............................................12
3.3.5 Status of LRFD Implementation.......................................................................13

4. CDOT Geotechnical Investigation, Design Practices, and Personnel

Resources ............................................................................................................14
4.1 Geotechnical Personnel Resources ......................................................................14
4.1.1 CDOT Geotechnical and Soil/Rock Fall Program Organization......................14
4.1.2 Geotechnical Personnel Resources Among Neighboring States
and Comparison ................................................................................................14
4.2 Field and Laboratory Testing Facilities ...............................................................14
4.2.1 CDOT Existing Geotechnical Investigation Facilities......................................14
4.2.2 Recommended Addition to New Geotechnical Investigation
Equipment .........................................................................................................15
4.2.3 Geotechnical Testing Facilities.........................................................................16
4.3 Geotechnical Investigation and Design Practices ................................................16
4.3.1 CDOT Geotechnical Investigation and Design Practices .................................16
4.3.2 Geotechnical Investigation and Design Practices in Neighboring
States .................................................................................................................17

viii
4.3.3 Comparison and Recommendations .................................................................17
4.4 Geotechnical LRFD Implementation...................................................................18
4.4.1 CDOT Status.....................................................................................................18
4.4.2 Status of LRFD Implementation in Neighboring States ...................................19

5. Status and Need for Implementation of Geotechnical LRFD at

CDOT ..................................................................................................................21
5.1 Introduction..........................................................................................................21
5.2 CDOT Geotechnical Design and Investigation Practices, and LRFD
Needs....................................................................................................................21
5.2.1 Colorado-Specific Shortcomings ......................................................................21
5.2.2 It Is Imperative, Not a Choice….......................................................................22
5.3 Implementation of Substructure LRFD in Colorado ...........................................23
5.3.1 Introduction.......................................................................................................23
5.3.2 Long-Term Effort..............................................................................................23
5.3.3 Deep Foundation Recommended for Substructure LRFD
Implementation .................................................................................................24

6. Strategic Plan for the Implementation of Geotechnical LRFD at

CDOT ..................................................................................................................25
6.1 Manpower ............................................................................................................25
6.2 Laboratory, Field, and Full-Scale Load Test Facilities........................................25
6.2.1 Laboratory.........................................................................................................25
6.2.2 Field Test and Exploration Equipment .............................................................25
6.2.3 Full-Scale Load Test Equipment ......................................................................25
6.3 Substructure Foundation for LRFD Implementation...........................................26
6.4 Implementation Plan and Strategy .......................................................................26

7. Deep Foundation Design .....................................................................................28

7.1 Introduction........................................................................................................28
7.2 Material Parameter Selection...............................................................................30
7.2.1 Introduction.......................................................................................................30
7.2.2 Young’s Modulus of Soils (Es) .........................................................................30
7.2.3 Methods for Determining Es .............................................................................30
7.2.4 Empirical Correlations of Es with Shear Modulus (G)………………………..35
7.2.5 Coefficient of Horizontal Subgrade Reaction (kh)............................................37
7.2.6 Methods for Determining kh .............................................................................38
7.2.7 Empirical Correlations ......................................................................................39
7.2.8 Relationship Between Es and kh ........................................................................40
7.2.9 Es for Numerical Models ..................................................................................41
7.2.10 Estimation of cu ...............................................................................................42
7.2.11 Ramberg-Osgood (RO) Parameters for NIKE3D ...........................................57
7.3 Broms Method .....................................................................................................62
7.3.1 Fundamental Theory and Assumptions ............................................................62
7.3.2 Lateral Ground-line Deflections (y0) at Working Loads ..................................66
7.3.3 The Ultimate Lateral Resistances of Piles (Pult) ...............................................66

ix
7.3.4 Coefficient of Horizontal Subgrade Reaction (kh)............................................71
7.3.5 Comparisons Between Broms Method and Test Results..................................76
7.3.6 Summary and Conclusions ...............................................................................80
7.4 Analyses Using LPILE Program..........................................................................81
7.4.1 Introduction.......................................................................................................81
7.4.2 Theoretical Background of LPILE....................................................................82
7.4.3 Input Data..........................................................................................................91
7.4.4 Evaluation of LPILE Program Using Field Test Results..................................94
7.4.5 Summary and Conclusions ...............................................................................97

8. Summary, Conclusions, and Recommendations for Further Study ...............99

References …………………………………………………………….……….…108
Appendix A – Survey Results……………………………………………………A-1
Appendix B – Finite Element Method and Calibration………………………..B-1
Appendix C – Finite Element Analyses of Rock Socketed Drilled Shafts…….C-1

x
LIST OF FIGURES
Figure 2-1 Probability Density Function for Load…….. ............................................6
Figure 2-2 Probability Density Function for Resistance..............................................6
Figure 2-3 Probability of Failure .................................................................................7
Figure 2-4 Definition of Reliability....……………………………………...……......7
Figure 7-1 The Effective Pre-consolidation Pressure (σ’c) vs. Liquidity
Index (LI) .................................................................................................31
Figure 7-2 Correlation Factor Kc used in Eu = Kccu...................................................34
Figure 7-3 Plot of Ratio of Ei to Cu from Unconfined Compression Tests
of Clays ....................................................................................................35
Figure 7-4 Ratio of Shear Modulus (G) to cu for Saturated Clays.............................37
Figure 7-5 Soils Modeled by Winkler’s Assumptions...............................................38
Figure 7-6 Flow Chart for Determination of Es .........................................................42
Figure 7-7 Correction Factor, λ for Vane Shear Tests...............................................44
Figure 7-8 Idealized Pressure-Volume Curve from Menard-type Prebored
Pressuremeter Test ...................................................................................45
Figure 7-9 Variation of Undrained Shear Strength Ratio with OCR for
Five Soils .................................................................................................46
Figure 7-10 Undrained Shear Strength Ratio vs. OCR for Marine Clay ...................47
Figure 7-11 Undrained Shear Strength Ratio vs. OCR..............................................51
Figure 7-12 Eu/cu vs. Shear Stress Ratio....................................................................54
Figure 7-13 Eu/cu vs. OCR at Two Shear Stress Levels ............................................54
Figure 7-14 Back-figured Es for Piles in Clay ...........................................................56
Figure 7-15 Typical Backbone Curve........................................................................60
Figure 7-16 Typical Stress-Strain Curve ...................................................................61
Figure 7-17 Effects of γy on Backbone Curves..........................................................61
Figure 7-18 Effects of α on Backbone Curves ..........................................................62
Figure 7-19 Effects of r on Backbone Curves ...........................................................62
Figure 7-20 Failure Modes for Short and Long Piles ................................................63
Figure 7-21 Soil Reaction and Bending Moment Distribution for Short
Piles.......................................................................................................63
Figure 7-22 Soil Reaction and Bending Moment Distribution for Long
Piles.......................................................................................................64
Figure 7-23 Flow Chart for Determining Behavior of Piles ......................................65
Figure 7-24 Pile Subjected to the Lateral Load .........................................................67
Figure 7-25 Lateral Deflections at Ground Surface...................................................67
Figure 7-26 Solution Procedures of Pult for Short Piles .............................................69
Figure 7-27 Graphical Solution of Pult for Short Piles...............................................70
Figure 7-28 Graphical Solution of Pult for Long Piles ...............................................72
Figure 7-29 Ground-line Deflections from Back-figured kh for Dunnavant
Data .......................................................................................................74
Figure 7-30 Ground-line Deflections from Back-figured kh for Reese &
Welch Data............................................................................................75
Figure 7-31 Ground-line Deflections by Broms Using Parameter Set 1 for
Dunnavant Data ....................................................................................78

xi
Figure 7-32 Ground-line Deflections by Broms Using Parameter Set 2 for
Dunnavant Data ....................................................................................79
Figure 7-33 Ground-line Deflections by Broms Using Parameter Set 1 for
Reese & Welch Data.............................................................................80
Figure 7-34 Element from Beam-Column Bar ..........................................................84
Figure 7-35 Schematic Diagram of p and y ...............................................................86
Figure 7-36 (a) Typical p-y Curve and (b) Resulting Soil Modulus.........................87
Figure 7-37 Transformed Sections for (a) Compression and Balanced
Failures and (b) Tension Failure ...........................................................89
Figure 7-38 Stresses and Forces in Pile Section ........................................................90
Figure 7-39 Typical Interaction Diagram of Pile.......................................................91
Figure 7-40 Pile Cross Sections.................................................................................92
Figure 7-41 Ground-line Deflections Using LPILE for Dunnavant’s Test
Results...................................................................................................96
Figure 7-42 Ground-line Deflections Using LPILE for Reese & Welch
Data .......................................................................................................98

xii
LIST OF TABLES

Table 7-1 Correlation Factor αc a ...............................................................................32

Table 7-2 Modulus of Elasticity from Undrained Shear Strength .............................35
Table 7-3 Values of Km = Gmax/cu ..............................................................................36
Table 7-4 Shear Modulus from Undrained Shear Strength .......................................36
Table 7-5 Values of k*s1 in ton/cu. ft for Square Plates, 1 ft x 1 ft and for
Long Strips, 1 ft Wide a, Resting on Pre-compressed Clay .....................39
Table 7-6 Coefficient of Horizontal Subgrade Reaction (kh) from cu .......................40
Table 7-7 Relationships Between Elasticity and Coefficient of Subgrade
Reaction for Clays....................................................................................41
Table 7-8 Approximated Undrained Shear Strength of Clays from SPT ..................43
Table 7-9 Undrained Shear Strength of Clays from CPT ..........................................43
Table 7-10 Database for CIUC Test...........................................................................48
Table 7-11 Ratio of E50 to cu for Clays ......................................................................52
Table 7-12 Overlapping Ranges and Average Values of E50/cu ................................52
Table 7-13 Ratio of Ei/cu ............................................................................................52
Table 7-14 Comparison of Advantages and Limitations of In-situ versus
Laboratory Testing for Cohesive Soils .................................................57
Table 7-15 Output from RAMBO for Clay with ρ = 3.80 slug/ft..............................59
Table 7-16 The Coefficient n1 .............................................................................................................................. 72
Table 7-17 The Coefficient n2....................................................................................72
Table 7-18 Pile Parameters from Dunnavant and Reese & Welch Tests ..................73
Table 7-19 Ground-line Deflections from Back-figured kh for Dunnavant...............74
Table 7-20 Ground-line Deflections from Back-figured kh for Reese & Welch.......74
Table 7-21 kh Using Terzaghi and Davisson for Dunnavant Data ............................76
Table 7-22 kh Using Terzaghi and Davisson for Reese & Welch Data.....................76
Table 7-23 Ground-line Deflections by Broms Method Using Parameter
Set 1 for Dunnavant Data......................................................................77
Table 7-24 Ground-line Deflections by Broms Method Using Parameter
Set 2 for Dunnavant Data......................................................................77
Table 7-25 Ground-line Deflections by Broms Using Parameter Set 1 for
Reese & Welch Data.............................................................................81
Table 7-26 Soil-Modulus Parameter k for Clays .......................................................93
Table 7-27 Strain at 50% of The Maximum Stress (ε50) for Clays............................93
Table 7-28 Ground-line Deflections (y0) for Pier in Uniform Soil and
Two-layered Soil Using LPILE with Dunnavant Data .........................95
Table 7-29 Ground-line Deflections (y0) for Pile Using LPILE with Reese
& Welch Data .......................................................................................97

xiii
1. INTRODUCTION

1.1 History of LRFD Development

The fundamental LRFD procedures and criteria were in place in Canada in early 1970’s.
Later in 1989, AASHTO decided to develop its own LRFD specifications. The 2002
AASHTO LRFD Bridge Design Specifications and the 2001 NHI lecture manual on
“Bridge Substructure Design” detail the LRFD procedures and criteria for bridge
substructure designs. If any state DOT were to adopt all the resistant factors and load
factors as provided, it could simply accept the recommended values of load and
resistance factors in its bridge substructure designs. This may not yield reasonable
designs for the states with soil and rock properties distinctive from those used in the
formulation of the resistance factors. Each state is likely to have some peculiar soils and
rocks with properties significantly different because of the distinctive geological
environment in which the soils and rocks are formed. Colorado soils and rocks can be
quite different from those the code resistance factor values are based. This shows the
need for the evaluation of the resistance factors specific to the Colorado soils and rocks.

The Colorado Department of Transportation (CDOT) uses a great number of drilled

shafts in supporting bridge abutments and piers. The drilled shaft design becomes an
important issue and the drilled shaft LRFD design procedures would assume a high
priority in the LRFD implementation. Drilled shafts derive their capacities from end
bearing and side friction. In the ASD design, a single factor of safety is used to address
the uncertainty of both components of drilled shaft capacity. The Osterberg’s Cell load
tests show that they seldom reach their ultimate values simultaneously. So it is illogical to
impose the same factor of safety. In the LRFD procedures, a uniform risk is applied to all
design parameters. The risk is established by a specification committee, not by an
individual design engineer. In the AASHTO specifications, the design risk is established
at one ten thousandth (or 0.0001). This is much more logical and rational.

1.2 Significance of the Implementation of CDOT Geotechnical LRFD Procedures

The implementation of geotechnical LRFD procedures is critical to the success of the
implementation of the AAASHTO LRFD Bridge Design Specifications. CDOT's Bridge
Branch has been practicing the AASHTO LRFD procedures for some time. To facilitate
the smooth and integrated LRFD bridge superstructure and substructure design, it is
logical to accelerate the implementation of the geotechnical LRFD procedures. It is more
difficult to implement the Colorado specific geotechnical LRFD because the Colorado
specific geological materials make it mandatory to determine their resistance factors in
advance of the geotechnical LRFD implementation.

The majority of superstructure construction materials like concrete and structure are
manufactured under strict quality control. As a result, the Colorado concrete is quite
similar in engineering properties, if not identical, to those from anywhere else in the
country or the world. Their properties are quite uniform irrespective of where the
concrete and steel are made and the evaluation of their Colorado-specific resistance
factors becomes unnecessary. Thus, the implementation of the AASHTO LRFD structure
code is much less time-consuming. Meanwhile, geological materials are of natural

1
occurrence and their properties are quite random. Because of the spatial and manufacture
process randomness, their properties can be quite different from those of other states and
the Colorado-specific resistance factors become a necessity for the implementation of the
geotechnical LRFD in Colorado.

The delay in the implementation of geotechnical LRFD is caused by the need for the
evaluation of the Colorado-specific resistance factors. Many other states face similar
problems. The response from nearly 30 out of 50 states shows that only a limited number
of state DOTs have begun their effort toward the implementation of geotechnical LRFD.
In fact, Colorado is ahead and can take a leadership role at least among the Rocky
Mountain States. The FHWA recommended the year 2007 as the target time for the
implementation of geotechnical LRFD.
For Evaluation of
1.3 Research Goal and Objectives
The goal of this study is to facilitate the CDOT implementation of geotechnical LRFD for
bridge substructure and restructure design. To realize this goal requires the
accomplishment of the following research objectives:

• Understand the current geotechnical investigation and design practices.

• Prioritize the areas for the implementation of geotechnical LRFD design
procedures and identify desirable methods for the strength evaluation of high
priority geostructures.
• Identify laboratory and field tests for evaluation of soil/rock properties.
• Identify the enhancement need for geotechnical investigation and testing facilities.
• Evaluate the sufficiency of geotechnical personnel resources.
• Establish the material property database for the formulation of the probability
density function.
• Evaluate the resistance factor for each strength parameter of soils and rocks.
• Establish the field test database, if possible, for the calibration of resistance
factors.

It is important to understand that the implementation of geotechnical LRFD is a

tremendous task that requires significant time and monetary investment on the part of
CDOT.

2
2. FUNDAMENTALS AND DEVELOPMENT OF LRFD SPECIFICATIONS

2.1 Development History of LRFD Specifications

The Ontario Ministry of Transportation (and Communication) decided in late 1970’s to
develop its own bridge design code rather than continue to use AASHTO Standard
Specification for Highway Bridges. It also decided to base its specification on
probabilistic limit states. Statistical Reliability of a bridge component is based on the
mean values of the applied loads and resistance parameters and their standard deviations.
In 1979 the first edition of the Ontario Highway Bridge Design Code (OHBDC) was
released to the design community as North America’s first calibrated, reliability-based
limit state specification. The OHBDC was updated and released in 1983 and 1993 with
the companion volume of commentary.

The LFD, and later LRFD, was adopted by AASHTO to replace the Allowable Stress
Design Method (ASD) in the 1970’s. As opposed to the ASD, where all uncertainty is
embedded in the factor of safety (FS), the LRFD approach applies separate factors to
account for uncertainties in loads and resistances based on the reliability theory. This
provides a reliable and uniform approach for the design of highway structures and
achieves more consistent level of safety in structure and substructure designs. In the
LRFD method, external loads are multiplied by load factors while the soil resistances are
multiplied by resistance factors.

In 1986 a group of five state bridge engineers from California, Colorado, Florida,
Michigan and Washington met in Denver and drafted a letter to the Subcommittee on
Bridges and Structures including their concerns that the AASHTO Specification was
falling behind the times. This sowed the seed for the new LRFD Specification. In
August 1986 the NCHRP Project 20-7/31 on “Development of Comprehensive Bridge
Specification and Commentary” was initiated. The main project objectives include:

• Review of philosophy of safety and coverage of other specification.

• Review of ASSHTO documents for inclusion in a standard specification.
• Assess the feasibility of a probabilistic-based specification.
• Prepare an outline for a revised AASHTO Specification for Highway Bridge
Design.”

Three possibilities exist for the task on feasibility:

• Allowable stress design (ASD) treats each load on a structure with equal
statistical variability.
• Load Factor Design (LFD) recognizes the difference in statistical variability
among different loads by using different multipliers for different loads.
• Reliability-based design, such as the procedure adopted in OHBDC takes into
account the statistical variability by using the mean and the standard deviation (or
the coefficient of variation) of all loads and resistance parameters. Given a set of
loads and resistance parameters the process can calculate the “probability of
failure.” This probability of failure is “not to exceed 0.0001.” Alternatively the

3
 process can target a quantity called “reliability index” which can be related to the
probability of failure.

The NCHRP Project 12-33 entitled “Development of Comprehensive Specification and

Commentary” was initiated in July of 1988. The project progress schedule was as
follows:

• The first draft showing the extent of coverage and organization was released in
April 1990.
• The second draft showing preliminary set of load and resistance factors was
released in April 1991.
• The third draft with 2,000 comments was released in April 1992.
• The fourth draft was submitted in March 1993 and was accepted as a ballot item
at the May 1993 meeting of the Subcommittee of Bridge and Structures. The
process featured “two rounds of trial designs.”

This leads to the publication of the “Load and Resistance Factor Design Specification for
Highway Bridges,” and further to the effort on the “North American Highway Bridge
Design Specification.”

The objective of a design based on reliability theory, or probability theory, is to separate

the load distribution from the resistance distribution such that the area of overlap is
sufficiently small, say one in 10,000.The probability-based LRFD specification is
advantageous in the following areas:

• More uniform level of safety throughout the system.

• Measurement of safety as a function of variability of loads and resistances.
• Designers will have an estimate of the probability of meeting or exceeding the
design criteria during the design life.

The major disadvantages are the increased design effort, the start-up cost in re-training
and the effort and time required during the transition.

The implementation of an LRFD design code requires:

• Establishment of a targeted transition period for changing from ASD to LRFD

design.
• Overlapping knowledge in geotechnical and structural engineering for all design
engineers.
• Requirement for all design geotechnical and structural engineers to enhance their
knowledge in probability and reliability.

In a reliability-based code, a designer is required to calculate the value of “reliability

index” provided by the design and compare it to the code-specified tolerable value. This
effort requires a designer, geotechnical or structural, to possess a sound knowledge of
probability and reliability. In principle it is possible, through the calibration of load and

4
resistance factors by trial designs, to develop load and resistance factors so that the
design process looks very much like the existing load factor design. In conclusion, an
LRFD design code requires the development of the load factors and resistance factors
such that the overlapping area between the load and resistance probability diagrams is no
greater than the value accepted by a design code, such as one in 10,000.

2.2 ASD versus LRFD

As discussed earlier, the ASD procedure collectively accounts for the uncertainty of all
design loads and resistances in one single factor of safety, while the LRFD procedure
applies a load factor to each load and resistance factor to each resistance (or strength)
parameter to account for the uncertainty in loads and resistances. The procedures are
compared as follows:

2.2.1 Allowable Stress Design (ASD)

• ASD: Rn/FS ≥ ∑Qi
• Resistance ≥ Effects of Loads
• Limitations
o Does not adequately account for the variability of loads and resistance
o Does not embody a reasonable measure of strength
o Subjective selection of factor of safety

2.2.2 Load and Resistance Factor Design (LRFD)

• LRFD: R = φ Rn ≥ ∑ηi γi Qi = Q

where Rn = nominal strength (e.g. ultimate bearing capacity); ∑Qi = nominal load
effect; FS = factor of safety; Rn = nominal resistance; φ = statistically-based
resistance factor; ηi = load modifier to account for ductility, redundancy and
operational importance; γi = statistically-based load factor; Qi = load effect.

• Limitations
o Require the availability of statistical data and probabilistic design
algorithms
o Resistance factors vary with design methods
o Require the change in design procedure from ASD

• Load and resistance can be modeled by a normal or log normal probability

density function based on its distribution characteristics.

2.3 Fundamentals of LRFD

2.3.1 Probability Density Function, Probability of Failure, Reliability, and Reliability
Index
Reliability-based design takes into account the statistical variability by using the mean
and the standard deviation (or the coefficient of variation) and the probability density
functions of all loads and resistance parameters. Given a load, Q (the sum of factored
load), and resistances, R, load factors, γi’s, and resistance factors, φ, can be calculated so

5
that the design meets the required “probability of failure” as specified in the LRFD
design guide. This probability of failure is selected by the code development professional
through extensive research. It is chosen “not to exceed 0.0001” by AASHTO or other
value specified by the code development group. This leads to the application of more
rational design factors. Alternatively the process can target a quantity called “reliability
index” related to the probability of failure.

An LRFD design code requires the development of the load factors and resistance factors
such that the overlapping area between the load and resistance probability diagrams is no
greater than one in 10,000, the code accepted probability of failure. So to formulate the
LRFD procedures requires the probability density functions (pdf) for all loads and
resistance parameters, as sampled in Figures 2.1 and 2.2 for load and resistance factor.
This means all loads and resistance parameters are considered random variables and their
pdf will most ideally derived from the data collection.

Figure 2.1 Probability Density Function for Load

Figure 2.2 Probability Density Function for Resistance

Probability of failure as shown in Figure 2.3 is defined as the probability that a design
load, Qm, (=∑ηi γi Qi) exceeds a selected value of material resistance, Rm (= φ Rn ), or R –
Q becomes negative:

Probability of failure = Pr {φ Rn ≤ ∑ηi γi Qi}

where Rn = nominal strength (e.g. ultimate bearing capacity); ∑Qi = Qn = nominal load
effect; Rn = nominal resistance; φ = statistically-based resistance factor; ηi = load
modifier to account for ductility, redundancy and operational importance; γi =
statistically-based load factor; Qi = load effect.

6
 Figure 2.3 Probability of Failure

Reliability expresses the probability of success and is expressed as follow:

Reliability = 1 - Pr (failure)
= probability of success
= Pr {φ Rn ≥ ∑ηi γi Qi}

As shown in Figure 2.4, reliability index, β, is defined as the difference between the mean
value of resistance and the mean value of design load divided by the standard deviation
of the difference between resistance and design load, i.e.

β = (Rm – Qm)/σ(R-Q) ……………………………………………………. (1)

where σ(R-Q) = (σR2 + σQ2)1/2 = standard deviation of (R-Q).

Figure 2.4 Definition of Reliability

For the probability of failure 1 x 10-4, β equals to 3.57 for a lognormal distribution, and
3.72 in normal (or bell shape) distribution. In the pavement design, β of 2.5 is frequently
adopted to define the designed service life, i.e. probability of failure of 1 x 10-2.

The designer will have to calculate the value of β for the design and then compare the
value to the code specified tolerable value. Thus, the designer’s sound knowledge of

7
reliability theory will be very beneficial to the LRFD design decision process. It is also
possible to develop load and resistance factors by trial designs through calibrating load
and resistance factors. The above equation yields:

β = (Rm – Qm)/σ(R-Q) and

Rm = Qm + β (σR2 + σQ2)1/2 = λ R .

Reliability-based design requires that the factored resistance to be equal to or greater than
the sum of factored loads, thus:

φ R = Q = ∑ γi Qi.

Thus, Rm = 1/φ λ ∑ γi Qi and

φ = λ (1/Rm) ∑ γi Qi = λ (∑ γi Qi) / (Qm + β ( σR2 + σQ2)1/2).

An acceptable value of reliability is specified in the code. A β value of 2.0 would imply
that approximately 97.3% of the values being included under the bell-shaped curve. Both
load and resistance factors must be evaluated. It can be accomplished by choosing the
values for load factors and then calculate the value for the resistance factors. Some
resistance factors and load factors specified in the AASHTO LRFD Bridge Design
Specifications are included in Appendix C.

2.3.2 Limit state Before an appropriate procedure is selected for estimating the nominal
strength in an LRFD design specification, the limit state must be defined. The limit state
is a condition beyond which a structure or its component ceases to fulfill the function for
which it is designed. Depending on the type of structures, the limit state includes, but not
limited to, service limit state (settlement, lateral displacements, etc) and strength limit
state (bearing capacity, etc). The service limit state is performance-based, which
specified the tolerable performance limit for settlement, lateral displacement, and wall
tilt, etc. The strength limit state ensures that the design procedure provides adequate
resistance (or margin of safety) against structure and substructure failures. Hence, the
design procedure of foundations requires the estimation of the nominal response (ultimate
strength and deformation) of the highway foundation in service. Foundation
deformations (e.g., load-settlement or p-y curve) are established from load tests and/or
analytical methods, such as finite element analysis and rational geotechnical analysis).

2.3.3 Bias factor A bias factor is the ratio of measured “mean1” resistance to the
predicted nominal resistance from the method defined by code. A bias factor reflects the
effects of all potential sources of errors in evaluating the resistance and therefore is the
product of the bias factors from all parameters involved in its evaluation. All parameters
are considered random variables. The calibration of resistance factors requires the

1
The ratio of mean value to nominal value obtained by following a code-defined procedures. In LRFD
method, all loads and resistances are considered “random variables” and, therefore, the term “actual value”
does not carry much meaning.

8
evaluation of the average value of a bias factor (to account for uncertainty due to errors in
predicted resistance) and the coefficient of variation of the random variable (to account
for uncertainty due to variability in predicted resistance). The values of bias factors and
coefficient of variation for the resistance from various types of tests on soils (e.g., SPT,
angle of internal friction, PDA, static load tests) have been reported. Due to limited
performance data, engineering judgments are used to estimate the statistical parameters of
different resistances required in the calibration. Bias factors can be greater than 1.00
because of the variation in testing methods (never two different soils test methods yield
the same strength) and the prediction methods for the nominal strength (engineering
judgments will have to be exercised for the lack of definitive methods available for its
estimate and even the rational methods are not necessarily accurate, like the classical
bearing capacity theory for both shallow and deep foundations.) While the estimate from
the result of large-scale tests is usually more reliable than the prediction from the
empirical rules (like Denver Magic Formula, DMF), it could still vary among cases.

2.3.4 Calibration The process of assigning appropriate values to resistance factors and
load factors is called calibration. The resistance factors developed for the 1997a LRFD
specifications were calibrated using a combination of reliability theory, fitting to ASD,
and engineering judgment. It is generally true in the ASD method that a higher FS is used
for empirically based method as opposed to the more rational methods. Calibration by
fitting to the ASD method was used in conjunction with reliability-based calibration to
ensure that the designs were comparable with accepted engineering practice. In situations
when sufficient data were not available, the reliability requirement is relaxed. Calibration
using reliability theory is preferred because it permits the selection of a target reliability
index that reflects the reliability and the failure probability. The value of resistance
factor chosen for a particular design procedure and limit state from a reliability-based
calibration can take into account the following factors:

• Variability of the soil and rock properties (geologic conditions),

• Method and extent of field exploration,
• Type, extent, accuracy, conditions, and suitability of in-situ and laboratory
testing,
• Type of geotechnical problems, and
• Reliability of methods used for predicting resistance.

The calibration procedures involve the following tasks:

• Selection of geotechnical problems for LRFD development,

• Establishment of the statistical database for resistance parameters of soils and
rocks,
• Formulation of resistance models (probability density function),
• Development of the reliability analysis procedures,
• Selection of target reliability index, and calculation of resistance factors.

9
3. SURVEY OF THE STATUS OF GEOTECHNICAL LRFD
IMPLEMENTATION AT DIFFERENT STATE DOTS

3.1 Introduction
The Federal Highway Administration (FHWA) recommended the implementation of the
AASHTO LRFD Bridge Design Specifications by the Year 2007. As shown in the
survey, state DOT’s are in the different stages of the LRFD implementation. Majority
either has already implemented or are implementing the structural part of the AASHTO
LRFD Bridge Specifications. However, most have not started the implementation of the
geotechnical LRFD for bridge substructures and foundations, and are still practicing ASD
or LFD procedures. Only very few have fully implemented the geotechnical LRFD
procedures, like Oklahoma, South Carolina, Washington, and Florida, etc. It is
encouraging to learn that, with only one exception, most respondents indicated the plan to
implement geotechnical LRFD by 2007.

3.2 Survey Questionnaires

Under the CDOT sponsorship, the University of Colorado at Denver and Health Sciences
Center (UCDHSC) sent the questionnaires with 29 questions to all state DOT’s. It was
very grateful that the majority of state DOT’s (28 out of 50) responded to the
questionnaires, although the returns were slower than expected. The respondents took
time beyond their regular call of duty to answer the questionnaires. Their efforts are
greatly appreciated. The questionnaires are shown in Table A.1. The distribution of
questions is as follow: one question each about the size of staff, major geological
materials, and organizational structure, two about laboratory and field test facilities, nine
about geotechnical investigation, design and business practices, and nine about status of
implementation of geotechnical LRFD procedures.

3.3. Survey Results and Discussion

The results of survey on the status of implementation of geotechnical LRFD procedures
are shown in Table A.2. Discussions are divided into the following five areas: staff,
facilities, geological materials, geotechnical investigation and design practices, and status
of LRFD implementation.

3.3.1 Staff
The size of geotechnical staff varies widely among the respondents and it ranges from 2
in Rhode Island and 138 in N.Y. with an average of 32.27. The CDOT geotechnical staff
of 20 covering geotechnical engineers, geologists, drilling crews, laboratory technicians,
and CAD specialists is much smaller then the average among all respondents and also
smaller than Kansas with 43 including two mechanics and Wyoming with 26. This shows
that the CDOT Geotechnical and Soil-Rock Fall Programs are grossly understaffed. To
assume the leadership role in the implementation of geotechnical LRFD among the
Rocky Mountain States in the next few years, it is necessary to enhance the geotechnical
staff.

10
3.3.2 Laboratory and Field Testing and Full-Scale Load Test Facilities
CDOT laboratory testing facility is comparable to the average of all respondents. This
indicates Colorado has sufficient laboratory equipment for the operation at the current
level. When implementing the geotechnical LRFD procedures, it mostly likely needs to
enhance the laboratory testing facility to meet the challenge of increased workload.

In the subsurface investigation, most respondent states use standard penetration test
(SPT), cone penetration test (CPT), and geophysical exploration (GE) in their subsurface
explorations. Undisturbed sampling using Shelby tube is also frequently adopted.
Occasionally, trenching and test pits are also used to expose the stratification of the
subsurface conditions.

CDOT uses nearly strictly SPT in its exploration program. It is of general practice to
begin an exploration program for a large project with geophysical exploration for an
approximate delineation of subsurface conditions. The geophysical exploration is then
followed by SPT and CPT for more detail identification of the subsoil layers through both
visual inspection in SPT and the strength characteristics in both SPT and CPT. The
former is not as ideal as a strength indicator, but it provides specimens for visual
inspection. The latter is more ideal as a strength indicator, but it doesn’t provide
specimens for inspection or testing. Performing both in parallel would be very ideal
practice for the situation where detail delineation of strength and stratification is
necessary. For design pressure-meter tests will be most favorable for assessing the in-situ
properties of soils and rocks for use in bridge substructure and foundation designs. PMT
is performed in a borehole after the undisturbed sampling. It becomes obvious that CDOT
is deficient in the following field test devices:
• Cone penetrometer,
• Pressure meter test device,
• Vane shear test, and
• Geophysical exploration equipment.

The implementation of geotechnical LRFD procedures requires full-scale load test results
for calibration of the selected design methodology and the resistance factors. CDOT has
in the past performed some deep foundation load tests. The load test effort accelerated
because of the TREX project where a great number of drilled shafts are used as support
for bridge piers and abutments. CDOT also uses a large number of driven piles as pier
and abutment foundations. It is recommended to add the following to the CDOT
geotechnical facilities list:
• Deep foundation load test facilities for vertical, lateral, and torsional loads.
• Expand the available PDA capability to cover CAPWAP capability for assessing
the vertical load capacity and the load-settlement relationship of driven piles.
Very few states have the capability of performing own load tests. They are usually
performed by load test companies. When implemented, CDOT would be able to perform
a greater number of tests for LRFD design calibration at a much reduced cost.

11
3.3.3 Geological Materials
Geological materials vary widely because of different geological formations in different
regions. Because of different soils and rocks the responding states have a wide variety of
geotechnical problems and use different approach in their geotechnical investigation,
design and testing.

3.3.4 Geotechnical Investigation and Design Practices

As shown in Section 3.3.2, CDOT uses nearly exclusively SPT in its subsurface
exploration. In terms of geotechnical design, it uses DMF for deep foundation design
assisted by PDA. In other words, the CDOT geotechnical investigation equipment is very
deficient and it is in a great need of adding some new investigation equipment, like CPT,
PMT and GE devices even before the implementation of geotechnical LRFD procedures.

Among the respondent states, about 35 percent of geotechnical investigation and design
is contracted out to the geotechnical consulting firms. During the CDOT re-engineering,
even higher percentage of geotechnical work was contracted out when Colorado was
effluent in funding. Now the World has turned and Colorado is short on cash, it might be
necessary to redistribute the engineering investigate and design fee and allocate a greater
portion to the internal investigation.

The survey results on retaining walls, bridge foundations, and slope stabilization are
summarized as follows:
• Five earth retaining mechanisms are investigated: reinforced concrete cantilever
wall (28), gravity wall (5), MSE w/ block facing (24), MSE w/ full-height rigid
panel (20), and Misc (5). Two types of MSE walls contribute 44% for total
retaining wall practice, while RC cantilever walls still maintains a significant
percentage of 28%.
• The bridge foundation support comes in three major types: drilled shaft, driven
pile, and shallow foundation with 21, 50, and 22 percent of total bridge
foundation practices. The deep foundation together contributes to 71 percent of
bridge foundation support. Thus, the deep foundation constitutes the major
foundation support for highway bridges.
• Slope stabilization mechanism comes in six different types: benching (20.2),
retaining wall (16.1), geosynthetics (11.4), scaling (5), anchors (3.1), and
micropile (0.4). Benching and retaining walls are as frequently used to stabilize
the retaining walls and geosynthetic slope reinforcement contributes to 11.4% of
the slope stabilization practice.

Note: The sum of percentage does not equal 100%. This is due to the data deficiency.

In summary, major geotechnical foundation practices include deep foundation; MSE

walls, RC cantilever walls and shallow foundation constitute major foundation practices,
which are the areas of emphasis in the implementation of geotechnical LRFD design
procedures.

12
3.3.5 Status of LRFD Implementation
The survey result on the status of the LRFD code implementation is summarized as
follows:
• Structural practices
o LFD: 7 (26.9% of respondents)
o ASD/LFD: 3 (11.5%)
o ASD: 3 (11.5%)
o LRFD: 6 (23.1%)
o LFD/LRFD: 3 (11.5%)
o ASD/LRFD: 2 (7.7%)
o ASD/LFD/LRFD: 2 (7.7%)

The above statistics reveal among the respondents to this question, only six states
have completely implemented the structural LRFD. Fifty percent of the
respondents either fully or partially implement the LRFD and other fifty percent
have not begun their implementation effort. In other words, near one half of the
respondent states have not implemented the AASHTO LRFD Bridge Design
specifications for bridge superstructure. In this respect, CDOT Bridge Branch is
ahead of absolute majority of the state DOT’s in practicing the LRFD code for
superstructure design. This seems natural for CDOT Bridge being one of the five
original proponents of the LRFD code.

• Geotechnical practices
o ASD: 16 (59.3% of respondents)
o LFD: 3 (11.1%)
o ASD/LFD: 2 (7.4%)
o ASD/LRFD: 3 (11.1%)
o LRFD: 3 (11.1%)
The above statistics reveal that less than 22.2% of the respondent states either
have implemented or begun to implement the geotechnical LRFD and more than
three quarters of all respondents have not yet attempted the implementation of the
geotechnical LRFD for bridge substructure design. This implies that the state
DOT’s are not rushing to the implementation LRFD code in their geotechnical
designs. This may indicate that the implementation of geotechnical LRFD is
experiencing significant difficulties. Reasons could include insufficient
manpower, equipment and/or the realization of the need for state-specific (or
regional) resistance factors for geological materials. All except one respondent
intent to implement the geotechnical LRFD in 2007 and one state in 2008. The
one exception indicated that they would not implement the geotechnical LRFD
unless they were forced upon.

In summary, the implementation of the AASHTO LRFD Bridge Design Specifications is

taking shape slowly, particularly the geotechnical LRFD for substructure design. To
successfully implement the specifications requires: 1) intensive educational training
program, 2) additional manpower, 3) additional equipment for geotechnical investigation
and testing. All except two plan to implement the LRFD code by the end of 2007.

13
4. CDOT GEOTECHNICAL INVESTIGATION, DESIGN PRACTICES AND
PERSONNEL RESOURCES

4.1 Geotechnical Personnel Resources

4.1.1 CDOT Geotechnical and Soil-Rockfall Program Organization
Currently, CDOT has twelve professionals working in the geotechnical program and
eight in the soil/rock fall program with a total of 20 persons as follows:
• Geotechnical Program has 12 persons including H.C. Liu as the program manager
with the following technical teams:
o Five-geotechnical engineers including Liu.
o Drilling crew of six technicians (3 senior and 3 junior drillers).
o One AutoCAD specialist.
• Soils/Rock Fall Program has 8 persons including C. K. Su as the program
manager with the following technical teams:
o Four professional in total (two geotechnical engineers including Su, senior
geologist, and geologist). The geologists specialize in rock fall problems.
o Four soil lab technicians.
In summary, CDOT has seven geotechnical engineers, two geologists, six drillers, four
soil laboratory technicians and one CAD specialist.

4.1.2 Geotechnical Personnel Resources Among Neighboring States and Comparison

It is unfortunate to find that only two neighboring states, Kansas and Wyoming,
responded to the questionnaires. Kansas DOT has 41 persons in its geotechnical staff and
Wyoming DOT 26 with an average between the two states of 33.5. The average
from all respondents is 32.3. Both averages are much larger than the CDOT geotechnical
staff of 20. These statistics show that CDOT is grossly understaffed. To provide quality
and safe geotechnical service, it must have additional staff, particularly when CDOT is
anticipating the accelerated effort in the implementation of the geotechnical LRFD.

4.2 Field and Laboratory Testing Facilities

4.2.1 CDOT Existing Geotechnical Investigation Facilities
While CDOT already has significant geotechnical investigation and testing equipment, to
effectively implement the LRFD procedures in geotechnical investigation and design
requires some additional equipment:
1. Field strength tests
• Standard Penetration Test (SPT)
• Cone Penetration Test (CPT)
• Pressure Meter Test (PMT)
• Vane Shear Test Device (VST), and
• Geophysical exploration equipment.
2. Laboratory property and strength tests
• Unconfined compression test (UCT)
• Direct shear tests (DST)
• Triaxial compression test (TCT)
• CBR tests
• Vheem Stabilometer test apparatus for R values.

14
 3. Cyclic triaxial test apparatus for resilient modulus test
4. Laboratory compressibility (or swell) test
• Oedometer tests
• Denver swell tests
5. Index property and classification
6. Standard (or Modified) Proctor Compaction tests

CDOT has the adequate equipment for the laboratory index property, compaction,
permeability, consolidation, and strength tests to maintain the status quo. When the
demand for the geotechnical services exceeds the capability to deliver the investigation
results for use in the bridge substructure and foundation designs, it is necessary to acquire
additional equipment, particularly when seeking the implementation of geotechnical
LRFD.

4.2.2 Recommended Additions to New Geotechnical Investigation Equipment

To enhance the quality of geotechnical service and accelerate the implementation of
geotechnical LRFD, CDOT must address the problem of the shortage of its field
geotechnical investigation and testing equipment. The following are the recommended
purchases:

• Cone penetration test (CPT) facility,

• Pressure meter test (PMT) facility,
• Vane shear test (VST) facility,
• Geophysical exploration equipment (GEE), and
• Full-scale load test facility for determining the in-situ load capacity and
deformation (or displacement) of drilled shafts and driven piles.

The survey shows that most state DOTs use CPT, PMT, GEE, and, sometime, VST in
addition to SPT CDOT uses exclusively. Devices are there for an effective geotechnical
investigation and it is puzzling that CDOT still does not use any of the above equipment.
In general, the site investigation begins with geophysical exploration to delineate the
subsurface strata and is followed by the joint use SPT and CPT to further define the
subsurface condition. When grossly undesirable soils or rocks are detected, undisturbed
sampling is followed to secure samples for laboratory study. Sometimes, it is necessary to
perform vane shear tests to assess the undrained shear strength of clayey soils. If the field
mechanical properties are needed, then PMT is performed. The CDOT subsurface
investigation will be greatly enhanced with the above equipment purchase.

The full-scale load test facilities should include the equipment for performing load tests
of piers and driven piles under vertical, lateral and torsional loads. The full-scale tests
will provide the database urgently needed for the calibration of LRFD procedures in
Colorado. CDOT has already begun to perform some full-scale tests of deep foundations
with the out-of state technical assistance. If CDOT were to develop its own in-house
capability for the full-scale tests, more load tests can be performed more cost effectively.
In the long run, it would save significantly the cost for conducting such tests under a
strict CDOT quality control. It is also recommended to expand the use of PDA to

15
CAPWAP to assess the vertical load capacity and performance of driven piles. To
implement the above recommendation will require the addition of geotechnical staff.

4.2.3 Geotechnical Testing Facilities

The survey shows: 1) CDOT laboratory testing facilities are comparable to those of the
questionnaire’s respondents. Although some states do have much better facilitated soils
testing laboratory; 2) Most state DOTs have much better field investigation facilities.
SPT, CPT and GE (geophysical exploration) are routinely used in the subsurface
exploration program. CDOT greatly lacks the subsurface exploration equipment, like
CPT, GEE, VST, and PMT, as discussed above. To effectively implement the
geotechnical LRFD code, CDOT needs to enhance its field exploration facilities.

4.3 Geotechnical Investigation and Design Practices

4.3.1 CDOT Geotechnical Investigation and Design Practices
Currently, CDOT geotechnical investigation and design use nearly exclusively the SPT.
The in-house triaxial and direct shear test devices are idled. This could be the result of re-
engineering and insufficient manpower to engage the equipment in the geotechnical
investigation. To implement the geotechnical LRFD, CDOT needs to activate these
apparatuses for the foundation investigation. Besides, as the nation’s pavement design
increasingly advances toward the mechanistic approach, the need for the capability to test
base course, sub-base, subgrade and pavement materials for their resilient moduli will
only increase with time. Thus, CDOT needs to acquire the quality cyclic triaxial test
capability for the evaluation of the mechanical properties and strength of materials
involved in the design and construction of the pavement.

CDOT has based most of its deep foundation design on DMF, where the allowable
bearing capacity is assumed to be equal to N/2 with the inherent factor of safety of 2.0 to
2.5. The side frictional capacity is recommended as 10% of the end bearing capacity.
First, DMF does not strictly belong to Denver. Many other researchers and practitioners
(Meyerhof, 1959; Terzaghi and Peck, 1967; Reese, et al, 1976; and Stroud and Butler,
1975; Chen, 19????) had used similar approaches, as outlined in the Executive Summary.
Chen promoted the use of the DMF in the design of drilled shafts in clay and clay shale in
Colorado because of its simplicity. However, as the state-of-the-art of geotechnical
engineering practices advances, more rational approaches have become available. In fact,
the survey of the fifty states in the United States in which 28 states responded reveals that
Colorado is the only state uses nearly exclusively the SPT blow count in the geotechnical
investigation and design. While this simplistic approach has a good track record of
success due to the lack of documented failures, it is certainly a good idea to check “how
conservative it is or unconservative it could be.” Besides, the DMF approach is geared
toward the ASD procedures and needed to be expanded to address the settlement (or
serviceability) issues.

The past Colorado experience indicates that the procedure can be either too conservative
or unconservative and a more rigorous approach or a calibrated DMF is needed to
enhance the rationality of the design method. Deficiencies of the DMF method are
summarized in the following:

16
 • The blow count N does not reflect the static property of soils because of the
impact nature of the Standard Penetration Test.
• DMF is an ASD design based on the bearing capacity. The settlement has never
been addressed. Most deep foundations in CDOT are used for bridge support and
settlement becomes critical in the maintenance of sufficient bridge deck
clearance. So, the approach needs enhancement for settlement assessment.
• The blow count N does not reflect well the effect of stress history on the strength
of clayey soils or clay shale and the N-based DMF might not provide a rational
estimate of end bearing strength.
• An effective settlement evaluation method using N value is needed, particularly
for bridge foundation on clay and clay shale. The method eventually needs
calibration for it effectiveness and reliability.
• After proper calibration against the Colorado soils and rocks, the N value can still
be used for bearing capacity and settlement computations, while seeking more
rational options.

4.3.2 Geotechnical Investigation and Design Practices in Neighboring States

Kansas and Wyoming are the only two states bordering Colorado that responded to the
questionnaires. Neither of them exclusively uses the blow count approach in their design
of bridge substructures and other foundations. Wyoming has begun to implement the
geotechnical LRFD. In this respect, CDOT is behind and needs to plan for complete
LRFD implementation.

4.3.3 Comparison and Recommendations

It was interesting to note that none of the states that returned the survey questionnaires
indicated the use of blow count method (or DMF) as practiced in CDOT on a routine
basis. Instead, most use the conventional foundation design methods. This points to two
possible directions for the future foundation design practice in CDOT:
1) Enhance, calibrate and continue to use the DMF approach, and
2) Examine alternative approaches that are more rational for the foundation
investigation and design.
No matter which direction CDOT eventually adopts, it is necessary to develop the in-
house capability in the area of field investigation and testing and full-scale foundation
load tests.

CDOT leads the PDA effort in Colorado, while local companies begin to develop PDA
capability and beyond. To maintain the leadership role, CDOT will need to keep up with
and advance beyond its current PDA practices and utilizes all available pile driving
technology in the pile design. Field tests, full-scale tests, and pile driving technology will
provide the information urgently needed for the development of resistance factors and
calibration of foundation design methodology adopted in Colorado. In sum, the following
areas are recommended for improvement in CDOT geotechnical investigation and design
practices aiming at the full implementation of geotechnical LRFD:

• Hire additional geotechnical engineers and technicians to enhance its current

practice, to maintain roadway safety, to perform full-scale load tests, and to

17
 prepare for the manpower need when the geotechnical LRFD implementation
effort is initiated.
• Enhance the CDOT facilities for laboratory and field investigation and testing.
• Calibrate and enhance the DMF design approach for geotechnical LRFD
procedures.
• Select the method for adoption in the implementation of CDOT geotechnical
LRFD.
• Evaluate the Colorado-specific resistance factors for the implementation of
the AASHTO LRFD Design Specifications for bridge substructures. During
the transition, adopt the resistance factors from the AASHTO specifications
with caution and use the full-scale load test, whenever possible and the ASD
design as calibrators.
• Close communication with the Bridge Branch is needed to foster the smooth
transition for the full implementation of the AASHTO LRFD Bridge Design
Specifications.
• Close communication with the geotechnical industry and government agencies
in Colorado for the geotechnical LRFD implementation.
• Retrain all geotechnical staff in CDOT and the Colorado geotechnical industry
through continuing education.

The implementation of a new design concept usually takes time because of the time-
consuming human adjustment. The CDOT Bridge is one of the five initiators and
pioneers for the development of the LRFD code. It is only reasonable to expect CDOT to
take pride in participating in the full implementation of the LRFD code. Besides using
more rational procedures, the implementation will enhance CDOT technical
communication with agencies at both federal and state levels.

4.4 Geotechnical LRFD Implementation

4.4.1 CDOT Status
CDOT needs to take pride in implementing the LRFD bridge design code. The CDOT
Bridge Branch participated in the study on the need for the development of a LRFD
bridge design code. It was one of the five pioneer states who met in Denver and decided
that the United States needed its own rational bridge design code after the Canada
developed and implemented its own LRFD code in 1970’s. After a comprehensive study,
the committee recommended the development of our own version. The committee action
leads to the current AASHTO LRFD Bridge Design Specifications, which is
recommended for implementation at all state DOTs. The effort took nearly twenty years.
CDOT has adopted the LRFD procedures in the bridge superstructure design. To allow a
smooth bridge design process in CDOT, it needs to implement the geotechnical LRFD
procedures for bridge substructures. Besides the technical superiority of the LRFD
procedures, CDOT will benefit significantly from practicing the national specifications,
when the need occurred for the consultation with other state DOTs on technical design
issues, when the same design specifications are practiced in all state DOTs.

Because of the distinctive geological materials, it is necessary to select the design

methodology, and evaluate the resistance factors for state-specific geological materials

18
when any state decides to adopt the geotechnical LRFD procedures. The evaluation of
resistance factors will take time and money. This may be why state DOTs are slow and
reluctant in the geotechnical LRFD implementation, as revealed in the survey results.
This delay may reflect the implementation difficulty. The extent of geotechnical
investigation depends on the following influencing factors:

• Importance of a project.
• Types of structures.
• Types of soils and rocks.
• Types of foundations.

To take advantage of collective wisdom, a geotechnical LRFD implementation

committee can be formed to formulate the details. Different issues can be discussed
openly in order to draw unbiased conclusions. The implementation strategy is then
brought to a meeting with key representatives from engineering consulting companies
and government agencies.

It is observed that CDOT geotechnical staff only provides design parameters to the
Bridge Branch, which is responsible for the design of superstructures. It is unclear where
the foundation design task is positioned. It is important to create a number of positions
responsible for foundation design. The positions can be located in either Geotechnical
Program or Bridge Branch.

In sum, the CDOT Geotechnical and Soil-Rockfall programs are much understaffed
with 20 highly qualified engineers, geologists and experienced technicians. Collectively
they are responsible for all subsurface investigation, laboratory soil testing, field
investigation and roadway safety, like rock fall and slope slides, etc. The Soils Unit is
equipped with excellent production-oriented laboratory testing equipment for index
property, R-value (Hveem stabilometer), CBR, Standard Proctor, permeability,
consolidation, unconfined compression, direct shear, and triaxial compression and
resilient modulus tests, etc. Some high quality laboratory apparatuses, like triaxial test,
resilient modulus test, and direct shear test apparatuses are left idle because of the lack of
manpower after re-engineering effort. It is critical to the geotechnical LRFD
implementation to activate these excellent laboratory testing apparatuses and place them
in production line to generate urgently needed laboratory data of soils and rocks for the
formulation of resistance factors for Colorado-specifically geological materials.

The field investigation still relies on the conventional Standard Penetration Test
equipment while the equipment for cone penetration tests, geophysical exploration tests,
and pressure meter tests are readily available and accepted as quality standard field test
equipment in geotechnical consulting companies and government agencies. CDOT needs
to accelerate its effort in the performance of field test and full-scale load tests.

4.4.2 Status of LRFD Implementation in Neighboring States

The Kansas DOT bases its structural design on LFD procedures and its geotechnical
design on ASD procedures. The Wyoming DOT does not believe that there is a major

19
advantage in switching over to the LRFD design procedures, in either bridge
superstructure or substructural design and currently practices LFD for both. However, it
had attempted the LRFD design procedures on a few structures. Although both DOTs
have some suspicion about the benefit of switching over to the LRFD codes as yet, but
had begun to pave the way for the eventual implementation of LRFD procedures in both
geotechnical and structural designs.

Since CDOT is already practicing the LRFD code in structural designs, and has taken
action to look into the feasibility of its implementation in geotechnical designs. In this
comparison, CDOT is somewhat ahead in the overall LRFD implementation. CDOT is
one of the five pioneer states in recommending the LRFD development, it seems
reasonable to carry this pioneer spirit forward and take a strong initiative and leadership
role in the implementation of the AASHTO LRFD Bridge Design Specifications. To
complete this transition, CDOT needs to provide additional geotechnical human resource,
field testing equipment, and full-scale load test capability to accelerate its LRFD
implementation effort in the bridge substructure design. While the task is not simple, it is
a worthwhile and rewarding effort.

20
5. STATUS AND NEED FOR IMPLEMENTATION OF GEOTECHNICAL LRFD
AT CDOT

5.1 Introduction
The implementation of geotechnical LRFD procedures is critical to the adoption of the
AAASHTO LRFD Bridge Design Specifications. The CDOT Bridge Branch has been
practicing the AASHTO LRFD procedures for some time. To facilitate the smooth and
integrated LRFD bridge superstructure and substructure designs, it is logical to accelerate
the implementation of the geotechnical LRFD procedures. It is more difficult to
implement the geotechnical LRFD because the need for the Colorado-specific geological
materials make it mandatory to determine their resistance factors in advance of the
geotechnical LRFD implementation.

The majority of superstructure construction materials like concrete and structure are
manufactured under strict quality control. As a result, Colorado concrete is quite
similar in engineering properties, if not identical, to concrete from anywhere else in the
country or the world. Their properties are quite uniform irrespective of where the
concrete and steel are made and the evaluation of their Colorado-specific resistance
factors becomes unnecessary. Thus, the implementation of the AASHTO LRFD structure
code is much less time-consuming. Meanwhile, geological materials are of natural
occurrence and their properties are quite random. Because of the spatial distribution and
manufacture process randomness, their properties can be quite different from those of
other states and the Colorado-specific resistance factors become necessary for the
implementation of the geotechnical LRFD in Colorado. This causes the delay in the
implementation of geotechnical LRFD. Many other states face the same dilemma. The
response from 28 states to the questionnaire shows that only a limited number of state
DOT’s have begun their effort toward the implementation of geotechnical LRFD. In fact,
Colorado is ahead and can still provide the geotechnical experience to other states
bordering Colorado. FHWA recommended the year 2007 as the target time for the
implementation of geotechnical LRFD, as reflected in the responses to the question on
the timing of the implementation.

As far as the implementation of geotechnical LRFD at the state level, among the states
that responded to the survey, Oklahoma and South Carolina have implemented the
geotechnical LRFD, and among the states that did not respond to the survey, at least
Washington and Florida have implemented the LRFD procedures in retaining wall design
and deep foundation design, respectively. The response from the states bordering
Colorado, like Kansas and Wyoming, indicates CDOT is ahead in the structural LRFD
implementation. CDOT can share its LRFD development experience with the states
bordering Colorado.

5.2 CDOT Geotechnical Design and Investigation Practices and LRFD Needs
5.2.1 Colorado-Specific Shortcomings The resistance factors used in the AASHTO
1997a LRFD specifications were calibrated for typical subsurface (geologic) conditions
and for methods widely used in predicting the nominal resistance. AASHTO (1991)
summarizes special geologic conditions encountered in the U.S that need modification of

21
resistance factors. Therefore, to ensure the successful implementation of geotechnical
LRFD in Colorado, it is important to generate resistance factors for CDOT methods not
calibrated in or accounted for in the LRFD specifications or generate the “Colorado-
specific” resistance factors for the design methods outlined in the AASHTO code.
Some current CDOT methods for predicting the nominal resistance need to be calibrated
when implementing LRFD procedures. The use of the Denver Magic Formula (DMF
based on blow count) to predict the allowable end bearing and side friction for deep
foundation is an example. The method, while widely used in CDOT, has never been
rigorously calibrated for Colorado geological materials for either the bearing capacity
evaluation or settlement computation, critical to maintaining the clearance in highway
bridges.

5.2.2 It Is Imperative, Not a Choice To improve the Colorado practice in substructure

designs, two possible approaches can be taken:

1) Improve the current CDOT practice by formulating the correlation between the
blow count and ultimate bearing capacity and the settlement assessment for deep
foundation through using full-scale load tests and/or laboratory testing;
2) Adopt the ASHTO LRFD procedures and specifications after reevaluating the
resistance factors for Colorado-specific geological materials;
3) Adopt the AASHTO LRFD code in the design while checking the design against
the ASD or LFD design and evaluating the Colorado-specific resistance factors.

While the first approach imposes less impact to the Colorado’s geotechnical practices, it
does have the following drawbacks:

1. While the DMF works “magically” well (for both driven pile and drilled shaft), it
deviates from the FHWA recommended AASHTO LRFD design approach.
Besides, the lack of reported failures could be an indication of DMF procedures
being excessively conservative and lack of safety uniformity among design cases.
2. N values are usually used directly in the design of shallow foundations in granular
soils. Because of its inability in closely reflecting the stress history of claystone
and clayshale, critical to the settlement and heave calculation, it is used only in
bearing capacity estimation.
3. Generation of the resistance database for the Colorado-specific soils and rocks
will be time-consuming.

The adoption of the AASHTO LRFD specifications will mean significantly revising the
current CDOT design practice. While this will be more in line with the FHWA
recommendations and allow us to gain technical support from other states in time of
need, it has the following difficulties:

1. Require revision in CDOT geotechnical investigation practices,

2. Require changes in substructure design practices and, new test equipments, as
outlined in Chapters 3 and 4.

22
 3. Due to the unique characteristics of the subsurface materials in the Colorado like
clayshale and expansive/collapsible soils, new databases are required for their
engineering properties in an effort to evaluate the Colorado-specific resistance
factors for its geotechnical LRFD implementation.

CDOT bridge engineers are practicing the LRFD design procedures, while it studies the
feasibility and the need for the implementation geotechnical LRFD. The specifications
are recommended for implementation at the state level by FHWA. While there may be
some contributing factors like shortage of appropriate equipment for subsurface
investigation and laboratory testing, insufficient manpower, etc, the main reason for the
implementation delay is the lack of material strength database for evaluating the
resistance factors for the Colorado-specific geological materials. This proposed study
aims at examining the current CDOT practice in geotechnical investigation and design,
determine the pros and cons of the practice, and recommend the best strategy for the
implementation of the geotechnical LRFD procedures in Colorado. The implementation
must not adversely affect the current Project Delivery System. To foster a smooth CDOT
design mission, the implementation of the geotechnical LRFD procedures is imperative,
not a choice.

5.3 Implementation of Substructure LRFD in Colorado

5.3.1 Introduction Colorado is one of the five states drafted a letter that laid the
foundation for the development of the AASHTO LRFD Bridge Design Specifications.
Having implemented it in the bridge superstructure design, the CDOT implementation of
geotechnical LRFD for substructures will produce a unified approach to bridge and
substructure design. The implementation of LRFD substructure design procedures will
take a significant investment of time and money. This long-term effort for complete
implementation will take about five years. The smooth transition and continuity are
critical to the successful implementation that does not adversely affect the project
delivery. Once the decision to implement the substructure LRFD procedures is made by
the LRFD committee involving the CDOT Central Administration, Bridge Branch, and
Geotechnical and Soil-Rock Fall Programs, etc, the data collection for soil and rock
resistance and properties and full-scale load tests should be initiated to provide the
critically needed database for calculating the resistance factors for Colorado-specific soils
and rocks. A data center should be established as the repository of all laboratory, field
and full-scale load test data. It will provide the database needed for the continual revision
for the values of resistance factors and the continual comparison of these values with the
database used in establishing the standard resistance factors in the AASHTO code.

5.3.2 Long-term Effort The complete implementation of LRFD design procedures for
substructures is estimated to take five years. It covers the following tasks:
• Establish Substructure LRFD Implementation Committee to oversee the
development of the substructure LRFD procedures.
• LRFD procedure training to psychologically prepare all design staff for the
implementation of foundation LRFD procedures.
• Establish the LRFD data center as the repository for all data for the evaluation of
resistance factors.

23
 • Select the substructure type for implementation,
• Select the design method and equations needed for the design,
• Determine the laboratory and/or field test methods for the evaluation and
collection of the design parameters.
• Purchase all necessary equipment for the laboratory, and field testing for soil and
rock properties and the full-scale load test, if deemed necessary.
• Statistical analyses of all data to formulate the probability density functions (pdf)
for all design parameters.
• Adopt the load factors, the associated database, and the pdf’s for all loads from
the AASHTO data bank.
• Evaluate the resistance factors corresponding to the risk factor of 0.0001, as
selected by the AASHTO LRFD committee.
• Design the foundation with the newly obtained resistance factors and compare it
with the ASD (LFD) design, or field full-scale load test results for the calibration
of the design.
• Promulgation of the new design procedures throughout CDOT, consulting
industry, and government agencies involved in the design of transportation
structures.
• The above calibration effort will continue for a period of time till all design staff
are satisfied with the design.
• Recommend the resistance factors and LRFD procedures for adoption.
• Expand the procedure to the next foundation type (S) recommended for LRFD
implementation and repeat the above development cycle till the complete
implementation of substructure LRFD design procedures.

The above listed tasks are to be enhanced by the LRFD Committee and divided into
several different phases for execution.

5.3.3 Deep Foundation Recommended for Substructure LRFD Implementation

The national survey shows that the frequency of the foundation types used as bridge
abutment and pier support is 21, 51, and 28% for drilled shafts, driven piles, and shallow
foundations, respectively. The deep foundation (drilled shafts and driven piles) has 72%
percent chance of being adopted as bridge abutment and pier support. Colorado uses a
large number of drilled shafts to support bridges, noise barriers and sign/signal posts and
another great number of driven piles for as bridge support. Thus, it seems reasonable to
recommend the deep foundation as the first candidate for geotechnical LRFD
implementation, which can then be subsequently extended to retaining walls, and other
foundation types.

24
6. STRATEGIC PLAN FOR THE IMPLEMENTATION OF GEOTECHNICAL
LRFD AT CDOT
Colorado was one of the five states that pioneered the development of LRFD bridge
design code in a meeting that was held in Denver. Besides all the technical merits of the
LRFD bridge design code, it seems to be reasonable to carry this pioneer spirit to fruition
by fully implement the AASHTO Bridge Design Specifications. Since the Bridge Branch
has already implemented the LRFD code, it also seems reasonable to accelerate the
implementation of geotechnical LRFD. The following outlines the recommended strategy
for the implementation of the substructure LRFD design procedures in three different
areas: manpower, facility, and technical implementation strategy.

6.1 Manpower
As the survey shows, the CDOT Geotechnical Program and Soils-Rockfall Program are
severely understaffed. The average from 28 states excluding Colorado is 32.27 and the
average of Kansas and Wyoming is 34.5. This shows that CDOT geotechnical programs
are severely understaffed to carry out even the current routine tasks. So it is
recommended to drastically increase the geotechnical staff soon.

6.2 Laboratory, Field, and Full-Scale Load Test Facilities

6.2.1 Laboratory The survey shows that CDOT laboratory testing facility is comparable
to those of the other respondent states is sufficient for carrying out the routine tasks at
current level. However, when the work load increases due to the substructure LRFD
implementation, there would be need to purchase additional equipment. It is noticed that
triaxial test and direct shear test apparatuses are idle due to the lack of manpower. This
situation needs correction.

6.2.2 Field Test and Exploration Equipment The survey shows that CDOT is very
much behind the national average in terms of the types of field tests employed in the field
subsurface investigation and its practice is far behind the state-of-the art practice in the
field investigation. It is quite unfair to expect the quality subsurface investigation to
provide the quality data for an effective bridge substructure design, when the
Geotechnical Program does not even have minimum field test and exploration equipment.
It is strongly urged to purchase the equipment for the following field test and exploration:
• Geophysical exploration,
• Cone penetration test,
• Menard pressure test, and
• Vane shear test.

6.2.3 Full-Scale Load Test Equipment The survey shows that very few states have the
capability of performing full-scale load tests. However, this is before their plan for
implementing the substructure LRFD is initiated. The state-specific resistance factors
required the calibration of the design computation against a reliable calibration standard.
In the case of substructure designs, nothing is more reliable than the full-scale load test
results. Thus, it is most definite that CDOT needs to perform more full-scale load tests to
provide the information for the calibration of the substructure designs. An Octerberg’s
cell load test, a bottom-up load test, costs, in an average, $100,000. If CDOT plans to

25
perform 100 OC load tests. The total cost amounts to ten million dollars. The need for the
top-down, lateral, and torsional load tests will significantly increase the expenditure. It
seems to be wise and reasonable investment to develop the in-house capability for
different kinds of full-scale load tests. This allows the in-house quality and cost control of
load tests. CDOT may even be able to export the professional service to the neighboring
states to recover some of the cost of the equipment purchase and development. An
alternative will be for all the neighboring states to jointly purchase the equipment for the
purpose of sharing the cost of equipment purchase and development.

6.3 Substructure Foundation for LRFD Implementation

Eventually all substructure foundations should be designed using the LRFD code. Initially
CDOT can select the most frequently used foundation type for implementation. The
survey shows the deep foundation (drilled shafts and driven piles) is used 72 times out of
100. Colorado also uses the deep foundation as the major foundation type for bridge
substructures. Thus, it is recommended to select the deep foundation as the first
foundation type for the LRFD code implementation and the implementation strategy in
Section 6.4 is formulated with this in mind.

6.4 Implementation Plan and Strategy

The task for the implementation of substructure LRFD code is immense. It will take
significant investment of time and money. Thus, it needs careful planning and
execution. This implementation strategy should not be formulated singularly by the
research group at the University of Colorado at Denver and Health Sciences Center
(UCDHSC). It should be the collective effort of a Substructure LRFD Committee (or
LRFD Committee) with members from all concerned parties to be identified. The
following implementation plan and strategy is presented for your “target shooting”
aiming at producing an executable plan and strategy.

Plan and Strategy for DOT Implementation of Geotechnical LRFD

• Close examination of the Plan and Strategy recommended for implementation by

the UCDHSC research group,
• Establish Substructure LRFD Implementation Committee to formulate the
substructure LRFD implementation pan and strategy. The committee should
include members from all concerned parties, particularly the CDOT Central
Administration, Bridge Branch, Geotechnical Program and Soil-Rock Fall
Program, etc,
• LRFD procedure training to psychologically and technically prepare all design
staff for the implementation of foundation LRFD procedures,
• Establish an LRFD data center as the repository for all data to be used in the
evaluation of resistance factors and calibration of a design,
• Select the substructure type for implementation,
• Select the design method and equations needed for the design,
• Determine the laboratory and/or field test methods for the evaluation and
collection of the design parameters,
• Hire additional geotechnical engineering staff,

26
 • Purchase all necessary equipment for the laboratory, and field testing for soil and
rock properties and the full-scale load test, if deemed necessary.
• Statistical analyses of all data to formulate the Colorado-specific probability
density functions (pdf) for all design parameters.
• Compare the Colorado database of all design parameters with the corresponding
AASHTO database to examine the difference,
• Adopt the load factors, associated database, and pdf’s for all loads from the
AASHTO data bank.
• Evaluate the resistance factors corresponding to the risk factor of 0.0001, as
selected by the AASHTO LRFD committee.
• Compare the Colorado-specific resistance factors with the AASHTO resistance
factors.
• Design the foundation with the Colorado-specific resistance factors and compare
the design with the ASD (LFD) design, and/or full-scale load test results for the
design calibration.
• Promulgation of the new design procedures throughout CDOT, consulting
industry, and government agencies involved in the design of transportation
structures.
• The above calibration effort will continue for a period of time till a satisfactory
design is reached.
• Recommend the resistance factors and LRFD procedures for adoption.
• Expand the procedure to the next foundation type(s) recommended for LRFD
implementation and repeat the above development cycle till the complete
implementation of substructure LRFD design procedures.
• Technology transfer through an educational program and in-house training.

The complete implementation of the geotechnical LRFD procedures will require

significant investment of time and money. The author of this report anxiously anticipates
significant reference of this research report, particularly the plan and strategy, in the
CDOT implementation effort.

27
7. DEEP FOUNDATION DESIGN

7.1 Introduction
In Colorado, most bridge abutments and piers are supported on drilled piers, rock socket
piers, or driven piles. The national survey reflects the same trend with 72% goes for the
deep foundations, drilled shafts (21%) and driven piles (51%). Thus, selecting piers and
piles for geotechnical LRFD implementation is a reasonable choice. The implementation
should start from single and then to group piers or piles.

CDOT has adopted DMF in the evaluation of vertical pier capacity. The lack of failure
cases indicates the DMF approach could be conservative. CDOT also has used LPILE
program in the in the design of pier under lateral loads with good success. However, with
the intent of implementing geotechnical LRFD, it is necessary to assess the true (or
ultimate) pier capacity through full-scale pier (or pile) load tests. It is very encouraging to
learn in a meeting that CDOT is committed to performing in-situ pier and pile load tests
for construction projects and has already done some as presented in the research report
(Abu-Hejleh, etc, 2003). These field tests form an irreplaceable precious database for the
calibration of nominal capacities based on the selected design methodology and
formulation. The statistical sample size is increasing, it is still much too small to
recommend for implementation. The sample size issue was discussed in a recent paper
(McKay, 2004). However, the load test effort can not cease, but continue to allow the
enhancement of database.

In an attempt to present an alternative methodology for the assessment of deep

foundation capacity, the NIKE3D-SSI research group in the Center for Geotechnical
Engineering Science at the University of Colorado at Denver and Health Science Center
has devoted a good portion of its research effort on the finite element analysis of deep
foundation. When calibrated and found effective, this method can be used to predict the
bearing capacity, settlement, lateral deformation, torsional rotation of deep foundation.
Results are shown in Appendix B.

So far, the research group has used the NIKE3D program developed at the Lawrence
Livermore National Laboratory and is trying LSDYNA program from the LSTC, Inc.
This finite element programs can become formidable design tools, when found to be
effective, because numerical tests do not cost much to perform. It can also be used to
generate the database for the deep foundation bearing capacity and performance for use
in evaluating the resistance factors and performance limits. The following are the areas
covered by this research group:
• Drilled Shafts in Hard Clays (completed). The study covers the performance of
singular load of vertical, horizontal, moment, or torsional load. When the choice
of material parameters and the delineation of soil strata are properly conducted,
the comparison with the existing full-scale load tests shows some encouraging
results as presented in later section of this report. This indicates that the
numerical analysis may be an effective tool for assessment of deep foundation
capacity and performance and can assist in generating and check the database for
LRFD implementation. Excel programs are being finalized for use by design

28
 engineers to assess the bearing capacity and performance of short and long piers
under lateral and torsional wind loads.
• Drilled Shafts in Sands (completed). The study assumed the sand to be elastic
material. This is only qualitative and preliminary purpose for pier under singular
load, of vertical, lateral, moment and torsional. Another study is being pursued
with more rigorous effort in simulating the behavior of sand and thus, true
behavior of deep foundation in sand.
• Drilled Shafts in Hard Clays under Combined Loads of vertical, horizontal,
moment and torsional loads. (on going, completion expected May, 2005). This
study aims to investigate the deep foundation performance under the combined
load of torsional, axial, lateral and moment loads. The cross-load type effects are
expected and this study intends to investigate the effect.
• Vertical Load Capacity of Drilled Shafts (ongoing, completion expected
December, 2005). Osterberg cell (OC) load tests are being used throughout the
country, in which the load is applied from the bottom of a deep foundation. The
test provides the load-displacement curve for both end bearing and side friction.
Problems needing investigation are two: the simultaneous failure in both end
bearing and side friction is not observed and the effect of overburden on the true
top-down side friction is not well understood. This study aims to answer these
two questions and propose a viable mechanism with which the true top-down
bearing capacity (end bearing plus side friction capacities) can be evaluated
given the OC load test results.

The NIKE/SSI research group urgently needs the funding for research and the results of
full-scale load tests for calibration of the numerical analysis results. When proven
effective, the numerical analysis can be used to generate the urgently needed database for
the Colorado-specific resistance factor evaluation. To evaluate the load capacity and
performance limit for deep foundation, different methods are available depending on the
loading conditions. Deep foundations are designed to resist any combination of vertical,
lateral, moment and torsional loads. The design can follow the following steps:
• Define the loading condition.
• Survey available design tools including the numerical analysis, select a minimum
of one method, and use it to determine the nominal capacity and the associated
performance limit.
• Compare the analysis result with the full-scale test result to establish the
confidence in the chosen analysis method.
• Establish a database for full-scale load tests for use in the calibration of the design
method and recommended LRFD resistance factors.
• In case of driven piles under vertical load, design methods include Terzaghi and
Meyerhoff bearing capacity equations and the capacity using PDA and CAPWAP.
The design calculations can be compared to static load test results to gain
confidence in the selected design method. It is well known that Meyerhoff’s
method gives much higher capacity than Terzaghi’s method.
• During the transition, the ASD design can serve as a calibrator for the LRFD
design.

29
This section reports the study on drilled shafts in hard clay under lateral load, moment, or
torsional load. The parallel study on the piers under combined loading is progressing and
is expected to complete in May, 2005. This study is much more involved and
complicated. Discussion in this section on singular load includes:
• Material parameter selection,
• Brom’s method and its comparison with numerical analysis method,
• LPILE method, and its comparison with numerical analysis method, and
• Finite element method of socketed piers in clay shale will be presented in
Appendix B on Socketed Piers.

7.2 Material Parameter Selection

7.2.1 Introduction
In a numerical analysis, material properties are needed for each material. A laterally
loaded pile problem involves reinforced concrete for piles and surrounding soils.
Concrete is modeled as an elastic material. Concrete properties are more consistent and
predictable as compared to soils. Consequently, this chapter section is devoted to soil
parameters for elastic soils with two parameters: Young’s modulus (Es) and Poisson’s
ratio (νs). Poisson’s ratio of 0.35 and 0.45 are used for low and high moisture clay and
clayshale, respectively.

Young’s modulus of soil is difficult to evaluate. Section 7.2.2 synthesizes the literature
survey on the elasticity of soils. Methods of determining Es and its correlation with
undrained shear strength of soils (cu) are presented. In laterally loaded pile problems, the
concept of subgrade reaction is usually used. Section 3.3 is devoted to the determination
of coefficient of horizontal subgrade reaction (kh) and its relation with undrained shear
strength of soils. The relationship between Es and kh, the ratios of Es to cu, the elasticity of
clays, and finally the Ramberg-Osgood (RO) parameters for soils are all synthesized in
Section 7.2.

7.2.2 Young’s Modulus of Soils (Es)

This section discusses the methods for evaluating the Young’s modulus of soils, Es.
Laboratory and in-situ tests are commonly used to determine Es. Sometimes, empirical
correlations are also used to estimate Es using undrained shear strength of clays or shear
modulus. The back calculations of Es from full-scale lateral load tests are considered to
provide the best fit value, Es, for soils, which is calculated directly from the lateral load-
deflection curves from the tests.

7.2.3 Methods for Determining Es

Literatures give the following methods for determining Es:
1) Back calculation from full-scale pile tests,
2) In-situ tests,
3) Laboratory tests,
4) Empirical correlations with undrained shear strength, and
5) Empirical correlations with shear modulus

30
7.2.3.1 Full-Scale Pile Tests
This method is accepted to be the most reliable method to determine Young’s modulus of
soils. Effects of piles installation and pile-soil interaction are taken into account
automatically. Poulos and Davis (1980) recommended using ground-line deflection at
working load to back analyze the secant modulus and use the linear part of load-
deflection curve to back figure the tangent modulus. They also commented that a tangent
modulus should give a more logical load-deflection relationship.

7.2.3.2 Laboratory Testing

Unconfined compression (UC) and unconsolidated undrained (UU) triaxial tests are
frequently used in determining the modulus of elasticity and strength of clayey soils
under undrained conditions, like in the design of laterally loaded piles. In the situation
when Es and undrained shear strength data are not available, Es can be estimated using
EQ (7.1) and results from the Oedometer test as recommended in the Canadian
Foundation Engineering Manual (1985):

E s = Cσ c' (7-1)

where σ’c is effective preconsolidation pressure from Oedometer test or Figure 7-1 and
C is a constant:
• C = 80 for stiff clays,
• C = 60 for medium clays, and
• C = 40 for soft clays.

Figure 7-1 Effective Preconsolidation Pressure (σ’c) vs. Liquidity Index (LI)
(Prakash and Sharma, 1990)

31
7.2.3.3 Correlation with Shear Modulus
Shear modulus can be related to elasticity of soils as shown in EQ (7-2).

E s = 2(1 + ν )G (7-2)
where Es is modulus of elasticity of soil
G is shear modulus evaluated in laboratory or in-situ tests.
ν is Poisson’s ratio

7.2.3.4 In-situ Tests

Cone penetration test (CPT)
Many in-situ tests can be used for determining elasticity of soils. The Cone Penetration
Test (CPT), pressuremeters, and plate load tests are commonly used. Es in tsf can be
empirically related with the cone tip bearing resistance by

E s = α c qc (7-3)

where αc is correlation factor depending on soil type and the cone bearing resistance and
can be obtained from Table 7-1
qc is the cone tip bearing resistance, tsf

Table 7-1 Correlation Factor αc a

Soil Type Resistance qc, tsf αc

Lean clays (CL) <7 3 to 8
7 to 20 2 to 5
> 20 1 to 2.5
Plastic clays (CH) < 20 2 to 6
a
Adapted from Corps of Engineers (1990).

Pressuremeter Test (PMT)

The pressuremeter modulus may be evaluated from Corps of Engineers 1990 Manual

(1 + ν )∆P ( R po + ∆R pm )
Ep = (7-4)
∆R p
where ν is Poisson’s ratio
∆P is change in pressure measured by the pressuremeter, tsf
Rpo is radius of probe, inches
∆Rpo is change in radius from Rpo at midpoint of straight portion of the pressure
meter curves, inches
∆Rp is change in radius between selected straight portions of the pressuremeter
curve, inches
Plate load test (PLT)
The modulus of elasticity can be estimated from the results of plate load tests by
32
 (1 − ν 2 )
Es = BI w (7-5)
∆ρ
( )
∆q

where Es is elasticity of soils, psi

ν is Poisson’s ratio
∆ρ/∆q is slope of settlement versus plate pressure, inches/psi
B is plate diameter, inches
Iw is influence factor, for circular plates, Iw = π/4

7.2.3.5 Empirical Correlations with Undrained Shear Strength of Soil

Many previous studies showed that the elasticity of soil is strongly related to the
undrained shear strength of clayey soils. Some empirical correlations between Es and Cu
are summarized as follows:
• In 1951, Skempton suggested that the secant modulus of the soils (E50) was
related to the unconfined compressive strength of soils (qu). He proposed the ratio
of E50 to qu were in the range of 25 to 100. Consequently, E50 is in between 50 to
200 times cu.
• Poulos (1971a) used elastic model to study behavior of piles subjected to lateral
loads. In his study, he proposed to use 15 to 95 times cu as the secant modulus of
soils. The lower value is for soft clays and the higher value is for stiff clays.
• Banerjee and Davies (1978) reported the value of E50 was in between 100 to 180
times cu for soft clay.
• Sullivan (1980) reported values of E50 for overconsolidated clays were 100 to 250
times cu from UU triaxial tests.
• Davies and Budhu (1986) suggested the ratio of secant modulus to undrained
shear strength for stiff clays in the range of 500 to 1000. In their review of values
of E50/cu, they found that Simons (1976) reported the ratio ranges from 40 to 3000
for cu from laboratory tests and Bjerrum (1972) gave the values of 500 to 1500 for
cu from vane shear tests.
• Bowles (1988) suggested using the elasticity of clays as functions of undrained
shear strength of clays in finite element analyses. He gave the expressions to
calculate Es for different types of clays as shown below:
o Normally consolidated sensitive clay:
Es = 200Cu to 500Cu

o Normally consolidated insensitive and lightly overconsolidated clay:

Es = 750Cu to 1200Cu

o Heavily overconsolidated clay:

Es = 1500Cu to 2000Cu

• Corps of Engineers (1990), in Engineering Design series: Settlement analysis,

gave the relationships between undrained modulus (Eu) and undrained shear
strength of clays. EQ (7-6) shows this relation:

33
 E u = K c cu
(7-6)
where Kc is correlation factor and can be found in Figure 7-2.

Figure 7-2 Correlation Factor Kc used in Eu = KcCu (Corps of Engineers 1990)

Few studies had been carried out to relate initial modulus of elasticity with undrained
shear strength. Matlock et al (1956) and Reese et al (1968) studied the relationships
between the initial or tangent modulus of clays (Ei) and undrained shear strength from
unconfined compression tests. Matlock et al found that Ei were in range of 40 to 200
times the undrained shear strength of clays from Lake Austin. Reese et al studied stiff
clays at Manor, Texas and found that Ei were in between 100 and 200 times cu. Plot of
ratio of Ei to cu along the depth from these two studies is shown in Figure 7-3.
Poulos and Davis (1980) reported the ratio of Ei to cu ranged from 250 to 400 using lower
and higher values for soft and stiff clays respectively. These empirical relations were
supposed to use in laterally loaded piles problems.

In 1989, Stokoe found that values of Ei were 2000 times undrained shear strength from
laboratory test results of soils at small strain levels.

34
Figure 7-3 Plot of Ratio of Ei to cu from Unconfined Compression Tests of Clays (Reese
et al 2000)

7.2.4 Empirical Correlations of Es with Shear Modulus (G)

A few studies had been carried out to determine correlations of shear modulus and
undrained shear strength of clays. Seed and Idriss (1970) studied soil moduli and
damping factors for dynamic analyses. In their study, they collected ratio of in-situ shear
modulus to undrained shear strength of saturated clays from many sources and plotted
into one graph (Figure 7-4). The ratios of G/cu depend on shear strain levels.

In 1988, Weiler studied shear modulus of clays at small strain levels. He provided the
ratios of maximum shear modulus to undrained shear strength of clays obtained from CU
triaxial tests. These ratios depend on plasticity index (PI) and overconsolidation ratio
(OCR) of clays and are shown in Table 7-3. Correlations of shear modulus and
undrained shear strength from various studies can be summarized in Table 7-4.

Table 7-2 Modulus of Elasticity from Undrained Shear Strength

No Study Correlation Remarks

1 Skempton E50 = 50 to 200cu cu from unconfined compression tests
(1951)
2 Poulos (1971) E50 = 15 to 95cu Lower values for soft clays and
higher values for stiff clays
E50 = 40cu For average value

35
3 Bjerrum (1972) E50 = 500 to 1500cu cu from vane shear tests
4 Simons (1976) E50 = 40 to 3000cu cu from laboratory tests
5 Banerjee and E50 = 100 to 180cu For soft clays and E50 assumed to
Davies (1978) vary linearly with depth
6 Sullivan (1980) E50 = 100 to 250cu For OC North Sea clays and cu from
UU triaxial tests
7 Davies and E50 = 500 to 1000cu For stiff clays
Budhu (1986) cu from standard triaxial undrained
test
8 Bowles (1988) Es = 1500 to 2000cu For heavily overconsolidated clays
Es = 750 to 1200cu For normally consolidated
insensitive and lightly OC clays
Es = 200 to 500cu For normally consolidated sensitive
clays
9 Corps of Eu = KcCu Eu is undrained modulus and Kc is
Engineer correlation factor from Figure 3-2
(1990)

10 Matlock et al Ei = 40 to 200cu Ei is initial or tangent modulus and cu

(1956) from unconfined compression tests
11 Reese et al Ei = 100 to 200 cu cu from unconfined compression tests
(1968)
12 Poulos and Ei = 250 to 400cu Lower values for soft clays and
Davis (1980) higher values for stiff clays
13 Stokoe (1989) Ei = 2000cu From soil samples subjected to very
small strains

Table 7-3 Values of Km = Gmax/cu (Weiler, 1988)

Plasticity Index Overconsolidation Ratio, OCR
1 2 5
15-20 1100 900 600
20-25 700 600 500
35-45 450 380 300

Table 7-4 Shear Modulus from Undrained Shear Strength

No Study Correlation Remark

1 Seed and G = KCu G is in-situ shear modulus
Idriss (1970) K is factors depending on strain levels
and can be determined from Figure 7-4.
2 Weiler (1988) Gmax = KmCu Gmax is maximum shear modulus
Km is factors from Table 3-3
cu is undrained shear strength from CU
tests

36
Figure 7-4 Ratio of Shear Modulus (G) to cu for Saturated Clays (Seed and Idriss 1970)

7.2.5 Coefficient of Horizontal Subgrade Reaction (kh)

The concept of subgrade reaction has been using in predicting lateral deformations of
piles under lateral loading. This concept is based on Winkler’s study in 1867. In his
study, he modeled soil behavior by independent series of linear springs along the depth of
the soils. The Winkler model is shown in Figure 7-5. Coefficients of horizontal subgrade
reaction (kh) represent stiffness of springs. When pile subjected to lateral load, pile will
deform laterally from the original position y. The resistance from soils is represented by
pressure q can be related to lateral deformation y by EQ (7-7).

q = kh y (7-7)

where q is soil reaction in force per square length

kh is coefficient of horizontal subgrade reaction in force per cubic length
y is lateral deflection of pile in length.

Evaluation of kh is required for solving piles under lateral loadings. There are many ways
to evaluate kh. Among these are:
1) Full-scale lateral-loading tests on piles
2) Plate load tests
3) Empirical correlations with soil properties

37
 Ground surface Pile

kh1

Foundation Soils

khi

khn

Figure 7-5 Soils Modeled by Winkler’s Assumptions

7.2.6 Methods for Determining kh

7.2.6.1Full-Scale Lateral-Loading Tests on Piles
The lateral load test may be the best method for determining the represented in-situ
horizontal subgrade reaction modulus of soils, kh, where a pile is instrumented to measure
soil pressures and associated deflections. The test results are used to calculate kh.
However, the method is both expensive and time-consuming.

7.2.6.2 Plate Load Tests

Plate load tests can be undertaken on both horizontal and vertical plates. These tests may
be tested using either circular or square plates. Many studies had been carried out to
evaluate kh from plate load tests. Terzaghi (1955) is the pioneer to evaluate coefficients
of horizontal subgrade reaction. He suggested determining kh from the results of
horizontal plate load tests using 1-feet wide square plate. He proposed the expression to
calculate kh from the coefficient of vertical subgrade reaction from 1 ft x 1ft plate (k*s1).
The expression is shown in EQ (7-8).

k s*1
kh = (7-8)
1.5D

where kh is coefficient of horizontal subgrade reaction in ton per cubic feet

k*s1 is coefficient of vertical subgrade reaction and given in Table 7-5.
D is pile diameter in feet

38
In 1961, Vesic studied behavior of infinite beam resting on isotropic elastic soil. He
proposed an expression to determine coefficient of vertical subgrade reaction k for
infinite piles from the plate load tests using square plate with a width D. The ks can be
applied to relatively long piles with a width D subjected to lateral loads.
Broms (1964a) suggested using Vesic’s expression to calculate kh for relatively
long piles by assuming ks and kh values are the same. Broms (1965) proposed the
expression to calculated kh for relatively long piles from plate load tests as shown in EQ
(7-9).

0.4 k 0 B
kh = (7-9)
D

where k0 is the coefficient of vertical subgrade reaction for a square or circular plate with
the side or diameter of B
D is pile diameter

Table 7-5 Values of k*s1 in ton/cu. ft for Square Plates, 1 ft x 1 ft and for Long Strips, 1 ft
Wide a, Resting on Pre-compressed Clay (Terzaghi 1955)

Consistency of clay Stiff Very Stiff Hard

Values of qu, ton/sq. ft 1-2 2-4 >4
Range of k*s1, square plates 50-100 100-200 > 200
Proposed values, square plates 75 150 300 b
a
For rectangular plates with width 1 ft and length L ft: ks1 = k*s1 (L + 0.5)/1.5L
b
Higher values should be used only if they were estimated on the basis of adequate test
results.

7.2.7 Empirical Correlations

There were a few studies on relationships between kh and soil properties. As mentioned
in previous section, Terzaghi (1955) proposed the procedure for calculating kh from plate
load test. In 1970, Davisson used nondimensional solutions to determine lateral load
capacity of piles. He suggested the expression to relate kh with undrained shear strength
of clay (cu) and pile diameter (D) as shown in EQ (7-10).

67cu
kh = (7-10)
D

It is easy to show that Terzaghi and Davisson’s expressions are the same. From Table 3-
5, k*s1 can be converted to qu and then cu as shown in EQ (7-11).

k s*1 = 50qu = 100cu (7-11)

Substitute EQ (7-11) into EQ (7-8), we get EQ (7-12).

39
 67cu
kh = (7-12)
D

The difference between Terzaghi and Davisson is Terzaghi suggested to use

representative value of kh for a range of qu or cu while Davisson used typical value of kh
for given value of cu. Table 3-6 summarizes kh formulae from both Terzaghi and
Davisson.

Table 7-6 Coefficient of Horizontal Subgrade Reaction (kh) from cu

No Study Correlation Remark

1 Terzaghi (1955) k* k*s1 from Table 7-5
k h = s1 and D is pile
1.5D
diameter
2 Davisson (1970) 67cu cu is undrained shear
kh = strength
D
D is pile diameter

7.2.8 Relationship Between Es and kh

Many attempts were devoted to study the relationships between elasticity and coefficient
of subgrade reaction. Based on Skempton’s works in 1951, Broms (1964a) mentioned
that the coefficient of subgrade reaction for cohesive soils is approximately proportional
to the unconfined compressive strength of the soil. Work by Terzaghi (1955) also
showed the relation of coefficient of subgrade reaction with unconfined compressive
strength of clays. Many studies show that elasticity of soils can be related to undrained
shear strength of the soils (Section 7.2.2). It is possible to relate Es with kh.
Vesic (1961) proposed the coefficient of vertical subgrade reaction in terms of elasticity
of soil, Poisson’s ratio of soil, bending stiffness of pile, and pile width as shown in EQ
(7-13).
1

0.65 ⎛⎜ E s D 4 ⎞ 12 E s
ks = ⎟ (7-13)
D ⎜⎝ E p I p ⎟ 1 −ν 2
⎠

where ks is coefficient of vertical subgrade reaction

Es is modulus of elasticity of soil
Ep is modulus of elasticity of pile
Ip is moment of inertia of pile section
ν is Poisson’s ratio of soil
D is pile width or diameter

Based on Timoshenko and Goodier (1951)’s theory of elasticity, Broms (1964a)

suggested to calculate coefficient of subgrade reaction from circular or square plate load
tests and relate kh with secant modulus of elasticity of clays as shown in EQ (7-14).

40
 1.25 E50
k0 = (7-14)
B(1 − ν 2 )

where k0 is coefficient of subgrade reaction from circular plate load test

E50 is the secant modulus of elasticity of the soils
B is plate diameter or width
ν is Poisson’s ratio of the soils

Substitute k0 from EQ (7-14) into EQ (7-9), we get

0.5 E 50
kh = (7-15)
D(1 − ν 2 )

Pike and Beikae (1984) reviewed the solutions of piles subjected to lateral loads from the
previous studies. Based on Beikae’s work in 1982, they proposed the relationships
between kh and E50 for clays as shown in EQ (7-16a) to (7-16c).

2.3E 50
kh = (7-16a)
D

2.0 E 50
kh = (7-16b)
D

1.8 E 50
kh = (7-16c)
D

EQ (7-16a), (7-16b), and (7-16c) are for Poisson’s ratio of 0.0, 0.33, and 0.5,
respectively.

Table 7-7 Relationships between Elasticity and Coefficient of Subgrade Reaction for
Clays

No Study Relationship Remark

1
1 Vesic (1961) For relatively long
0.65 ⎛⎜ E s D 4 ⎞
12
⎟ Es piles
ks =
D ⎜⎝ E p I p ⎟ 1 −ν 2
⎠
2 Broms (1964a) 0.5 E50 For relatively long
kh = piles
D(1 − ν 2 )
3 Pyke and Beikae 2 E50 For µ = 0.33
(1984) kh =
D

7.2.9 Es for Numerical Models

41
As shown in the preceding sections, modulus of elasticity of soils can be determined by
many methods. Among those, the most reliable method is to perform pile load test in the
field and back analyze the soil modulus. If the full-scale pile test is not performed, the
evaluation of Es from in-situ or laboratory testing is recommended. If the undrained
shear strength is available, the empirical relationships between Es and cu can be used with
caution. Es is a function of undrained shear strength of clays. Coefficients of horizontal
subgrade reaction can be related to both undrained shear strength and elasticity of soils.
Consequently, Es, kh, and cu are interrelated. From previous sections, a flow chart, Figure
7-6, is drawn for determination of Es. Figure 7-6 clearly shows that that method IV and V
can be regrouped to calculate kh from cu. It is widely accepted that undrained shear
strength of clays is important parameter for analyses of piles under lateral loads in clay.
The subsequent sections are devoted to the evaluation of undrained shear strength, Cu/σvc’
and Es/cu ratios, and Es for analyzing pile performance in clays.

I
Es by pile load tests

II
Es by in-situ tests

Determination of Es
III
Es by laboratory tests

IV

Es by cu

Es by kh

Figure 7-6 Flow Chart for Determination of Es

7.2.10 Estimation of cu
7.2.10.1 In-situ Tests
Knowing N value from Standard Penetration Tests, undrained shear strength of clays can
be estimated from Table 7-8. In case that CPT (Cone Penetration Tests) data is available,
cu can be estimated from Table 7-9.

42
Table 7-8 Approximated Undrained Shear Strength of Clays from SPT

SPT Penetration N Values Estimated Consistency Estimated Range of cu,

a
ksf
<2 Very Soft (extruded < 0.25
between fingers when
squeezed)
2-4 Soft (molded by light 0.25 – 0.50
finger pressure)
4-8 Medium (molded by 0.50 – 1.00
strong finger pressure)
8 - 15 Stiff (readily indented 1.00 – 2.00
only with great effort)
15 - 30 Very stiff (readily 2.00 – 4.00
indented by thumbnail)
> 30 Hard (indented with > 4.00
difficulty by thumbnail)
a
The Canadian Foundation Engineering Manual does not recommend the relationship
with N.
Source: Prakash and Sharma (1990) and Terzaghi and Peck (1967), Design Manual
NAVFAC DM7.1 (1982), and Canadian Foundation Engineering Manual (1985).

Table 7-9 Undrained Shear Strength of Clays from CPT a (McCarthy 2002)

General qc/pa from CPT Consistency Approximated cu, tsf

Description Index, CI = 1.0 –
LI b
Very soft <5 < 0.5 < 0.25
Soft 5 to 0.5 to 0.25 – 0.5
Medium 15 0.75 0.5 – 1.00
Stiff 15 – 30 0.75 – 1.0 1.00 – 2.00
Very stiff 30 – 60 1.0 – 1.5 2.00 – 4.00
Hard > 60 > 1.5 > 4.00
a
Data compiled from various sources.
( w − PL)
b
LI = is liquid index.
( LL − PL)
The undrained shear strength can also be evaluated from the result of vane shear tests as
follow (Coduto, 1999):

6λT f
cu = (7-17)
7πd 3
43
where cu is undrained shear strength in psf
Tf is torque at failure in lb-ft
d is vane diameter in ft
λ is correction factor for time effect from Figure 7-7

Pressuremeter can be used for determining undrained shear strength of clays as shown in
EQ (7-18) provided by Roberson (1986).

PL − P0
cu = (7-18)
5.5

where PL is the maximum pressure in kg/cm2

VL is the maximum volume corresponding to PL (Figure 7-8)
P0 is the initial pressure of the pressuremeter in kg/cm2 and the corresponding
volume is V0 (cm3)

Alternatively with EQ (7-4), the pressuremeter modulus, Ep, can be estimated from the
slope of the linear portion of the pressure-volume curve in (Figure 7-8) by

⎛ ∆P ⎞
E p = 2.66(V0 + Vm )⎜ ⎟ (7-19)
⎝ ∆V ⎠

where V0 is initial volume of the measuring cell

Vm is volume change read on the volumeter at a pressure corresponding to the
mean pressure in the pseudoelastic range (estimated from mean of V0 and Vf)
∆P/∆V is slope of the pressure-volume curve as the Section AB, where Point A
refers to the initial stage and Point B the upper limit of linear portion.

Figure 7-7 Correction Factor, λ for Vane Shear Tests (Coduto, 1999)

44
Figure 7-8 Idealized Pressure-Volume Curve from Menard-type Prebored Pressuremeter
Test (Prakash and Sharma, 1990)

7.2.10.2 Laboratory Tests

Many laboratory tests can yield the undrained shear strength of clays and consolidated
undrained (CU) triaxial tests are preferred (Bardet 1997) including isotropically-
consolidated undrained triaxial compression tests (CIUC). Chen and Kulhaway (1993)
indicated that unconsolidated undrained (UU) triaxial tests gave a better estimate of
undrained shear strength than unconfined compression tests (UC). However, because of
their simplicity, UC tests are more frequently used commercially than CU and UU tests.
This section is devoted to the empirical, laboratory-based correlations of cu with others
soil properties. cu from undrained shear strength can be estimated from plasticity index
(PI) and the effective preconsolidation pressure (σ’c) (Design Manual NAVFAC DM 7.1
1982).

cu = σ c' (0.11 + 0.0037 PI ) (7-20)

where σ’c is determined from Oedometer tests or Figure 7-1. For normally consolidated
natural clays, the effective vertical overburden pressure (σ’v) can be replaced with σ’c in
EQ (7-20) to estimate cu (Skempton 1948; Bjerrum and Simons 1960).
Based on the database for CIUC and UU tests, Chen and Kulhawy (1993) suggested the
correlation between σ’c, atmospheric pressure (pa) and liquidity index, LI, (See also
Figure 7-1 or Table 7-9):

σ c' = p a 10 (1.11−1.62 LI ) (7-21)

The undrained shear strength of clays can also be estimated from the effective
overburden stress (σ’vc or σ’vo) and overconsolidation ratio (OCR). Ladd et al. (1977) and

45
Jamiolkowski et al. (1985) suggested the following relation between cu/σ’vc and OCR as
follows
cu
= S (OCR) m (7-22)
σ vc
'

⎛c ⎞
where S = ⎜⎜ u' ⎟⎟ is the ratio for normally consolidated clay and m is the strength
⎝ σ vc ⎠ nc
increase exponent.

Ladd et al. (1977) found that the average value for five different clays from the direct
simple shear tests is 0.8. Jamiolkowski et al. (1985) reported that S = 0.23 ± 0.04 for
clays having plasticity index < 60%. Davies and Budhu (1986) suggested using S = 0.3
for many clays. EQ (7-22) can be plotted on both semi-log scale as shown in Figure 7-9
and log-log scale as shown in Figure 7-10.

Figure 7-9 Variation of Undrained Shear Strength Ratio with OCR for Five Soils (Foott
and Ladd 1973).

46
Figure 7-10 Unddrained Shear Strength Ratio vs. OCR for Marine Clay (Jamiolkowski et
al. 1985).

Chen and Kulhawy (1993) studied the undrained shear strength interrelationships among
CIUC, UU, and UC tests. A database for CIUC and UU tests was developed as shown in
Table 7-10 and plotted in Figure 7-11, where S and m equal to 0.4 and 0.58, respectively
with R2 = 0.77 as shown in EQ (7-23). This curve or equation can be used to estimate
undrained shear strength of clays at particular OCR and overburden pressure.

cu
= 0.4(OCR ) 0.58 (7-23)
σ '
vc

47
Table 7-10 Database for CIUC test a

b b
Site Soil description Depth, m OCR cu/σ'vc Source

Boston Medium blue clay 12.2 4 0.8 Ladd and Lambe (1963)
18.3 1.8 0.48
27.4 1 0.32
Lagunillas High plasticity clay 6.2 1.3 0.4 Ladd and Lambe (1963)
6.4 1.3 0.41
Kawasaki High plasticity clay 20.5 1.2 0.49 Ladd and Lambe (1963)
25 1.2 0.45
35 1.2 0.48
Beaumont High plasticity clay with fissures 0 to 8 7.5 1.58 Mahar and O'Neill (1983)
and slickensides
Montgomery Light grey sandy clay with 9 to 17 4 1.14 Mahar and O'Neill (1983)
desiccation
Hamilton Firm to stiff grey silty clay 3 to 6 3.2 1.16 Ismael and Klym (1978)
6 to 9 2.2 0.73
11 1.2 0.78
Firm to stiff grey silty clay 4.6 3.2 1.18
(surface desiccated and fissured) 7.2 2.5 0.73
10.8 1.5 0.78
15.2 1.1 0.6
Lackland Expansive black to grey clay 0 to 3 5 1.48 Johnson and Stroman
3 to 6 4.8 1.12 (1984)
Fissured expansive clay shale 6 to 9 6.5 1.6
9 to 12 8.5 1.47
Rio de Janeiro, Soft grayey clay 2 to 4 2.1 0.62 Ramalho-Ortigao et al.
Guanabara Bay 4 to 6 2 0.51 (1983)
6 to 8 1.7 0.46
8 to 10 1.7 0.44
South Padre Medium to stiff clay 6 to 12 1.2 0.6 Focht and Drash (1985)
Island 15 to 18 1.1 0.51
8.2 1.2 0.58
14.6 1.2 0.46
19 6.4 0.52

48
Table 7-10 Database for CIUC test (Cont.) a

b b
Site Soil description Depth, m OCR cu/σ'vc Source

St. Alban Soft to medium silty clay 2.1 2.1 0.99 Roy et al. (1982)
5.5 2.3 0.85
7.2 2.4 0.93
Boston Lean and moderately sensitive 18 1.7 D'Appolonia (1972)
blue clay 12 1.4
3.2 0.72
1 0.32
Laboratory Overconsolidated kaolinite 9 1.28 Duncan and Seed (1966)
result 9 1.03
Hackensack Varved clay 7.9 to 15 1.8 0.6 Saxena (1978)
Valley
Santa Barara Firm Pleistocene clay 20 to 60 1.6 0.36 Quiros and Young (1988)
Channel Hard silty clay 100 to 140 1.2 0.26
Lakeland Cohesive slimes 0 to 33 1.1 0.41 Ladd (1991)
San Francisco Soft grey clay (New Bay Mud) 6 to 10 1.4 0.43 Clough and Denby (1980)
Bay Mud 10 to 15 1.3 0.44
San Francisco Sandy clay 6 to 9 1.4 0.55 Clough and Denby (1980)
Soft grey clay 9 to 12 1.2 0.49
Boston Marine illitic blue clay 4 0.91 Kinner and Ladd (1973)
2 0.55
1 0.31
Anacostia 4 to 6 2.1 0.46 Mayne and Frost (1986)
6 to 9 2.1 0.32
Tuckerton Dark grey silty clay 16 8 2.03 Koutsoftas and Fischer
17 5.2 1.17 (1976)
Dark grey plastic clay 18 to 23 4 1.16
Ottawa Leda clay - moderately OC clay 6 to 9 3.1 1.08 Coates and Mcrostie
with high plasticity and 9 to 12 2.2 1.02 (1963)
sensitivity 12 to 15 2 0.95 Eden and Crawford (1957)
15 to 18 2 0.7
18 to 21 1.6 0.68

49
Table 7-10 Database for CIUC test (Cont.) a

b b
Site Soil description Depth, m OCR cu/σ'vc Source

Madingley Grey fissured Gault clay with 3 to 4 20 2.33 Windle and Wroth (1977a)
heavily OC clay 4 to 6 18 2.27
6 to 7 14 2 Coop and Wroth (1989)
Windle and Wroth (1977b)
Southeastern Very stiff clay with high 15.2 6.5 0.87 Endley et al. (1979)
Texas plasticity 18.3 5.8 0.75
21.3 2.9 0.64
Empire Fine grey clay 36.6 1.2 0.27 Cox et al. (1979)
Chicago Hard silty clay 3.7 17 2.5 Holtz and Baker (1972)
9 20 2.22
10 22 2.35
11.6 39 2
Gulf of Mexico Soft plastic clay 0 to 6 3.5 0.86 Fenske (1956)
Firm to stiff plastic clay eith silt 18 to 32 1.1 0.34
and sand 32 to 50 1.2 0.29
Skabo Plastic clay with high salt 10.6 to 16 1.2 0.45 Ladd and Lambe (1963)
content
Gota Valley Lilla Edet clay - marine 4 to 6.8 12 1.44 Bjerrum and Wu (1960)
late-glacial plastic clay with 10 to 12.3 1.8 0.76
high sensitivity 16.2 to 18 1.5 0.52
10.8 to 13 1.5 0.32
Drammen Soft silty clay with thin seams 5 to 12 1.3 0.34 Simons (1960)
of silt and fine sand 18 1.1 0.33
Sault Ste. Marie Varved glacial lake clay with 9 1.2 0.23 Wu (1960) and Wu et al.
floccuent structure (1962)
Kars Cemented Leda clay 2.5 to 6 7 1.47 Raymond (1972)
6 to 12 2.5 0.72

a
Adapted from Chen and Kulhawy (1993).
b
Values were taken from either single point or an average over a certain depth.

50
 3.5

2.5

2
cu/σ vc
'

1.5

0.5788
y = 0.3976x
1
2
R = 0.7709

0.5

0
1 10 100

OCR

Figure 7-11 Undrained Shear Strength Ratio vs. OCR from developed from the database
in Chen and Kulhawy 1993

7.2.10.3 Analysis of Es/cu Ratio

In section 7.2.2, relationships between Es and cu were given. From Table 7-2, the ratio of
E50 to cu can be grouped into two ranges based on values as shown in Table 7-11. E50/cu
ratio by Simons (1976) is not included in Table 7-11 because the ratio is out of range.
Ratios by Corps of Engineers are also not shown because this method requires more
detailed information of soils. Note that grouping was done regardless of type of clays
and test procedures. For each range, the common range (i.e. the range that can be found
in every sub-range except ratios by Poulos) of E50/cu can be obtained and shown in Table
7-12.

From Table 7-2, the range and average values of Ei/cu can be shown in Table 7-13. The
common range for Ei/cu ratio is 100 to 200. It is obviously that Ei/cu is greater than
E50/cu. The common range of E50/cu from range I give the reasonable values compare
with Ei/cu. By the same token, the average range of Ei/cu is suggested. The ratios of E50/cu
or Ei/cu are approximate and can be used as a guide when only undrained shear strength is
available.

51
Table 7-11 Ratio of E50 to cu for Clays a

Range Sub-range Study Ratio of E50/cu Remark

No No
1 Poulos (1971) 15 to 95 Piles in uniform,
I elastic soils
2 Skempton (1951) 50 to 200 Bearing capacity
of clays and cu
from unconfined
compression tests
3 Banerjee and Davies 100 to 180 For soft clays and
(1978) Es vary linearly
with depth.
4 Sullivan (1980) 100 to 250 For OC clays and
cu from UU
triaxial tests

1 Bjerrum (1972) 500 to 1500 cu from vane

II shear tests
2 Davies and Budhu 500 to 1000 For stiff clays
(1986) and cu from UU
triaxial tests
b
3 Bowles (1988) 200 to 2000 Lower values for
NC clays and
higher values for
heavily OC clays
a
Ratio of Es/cu by Simons (1976) and Corps of Engineers are not shown.
b
Elasticity by Bowles is assumed to be secant elasticity.

Table 7-12 Overlapping Ranges and Average Values of E50/cu

Range No Overlapping Average values

range of E50/cu of E50 /cu
I 50 to 180 115
II 180 to 500 340
III 500 to 1000 750

Table 7-13 Ratio of Ei/cu

Study Minimum Value of Ei/cu Maximum Value of Ei/cu

Matlock et al (1956) 40 200
Reese et al (1968) 100 200
Poulos and Davis (1980) 250 400
Average 130 265 a
52
a
Round off to whole number

The Ratio of Es/cu (Es is the secant modulus of elasticity at a designated strain level)
depends on shear stress level as shown in Figure 7-12. Eu in Figure 7-12 represents the
secant modulus of elasticity of clays. The Eu/cu ratios were plotted versus the shear stress
level (τh/cu) from CK0U direct simple shear tests on seven NC clays. From Figure 7-12,
Eu/cu ratios decrease with increasing values of shear stress ratios. Eu/cu also decrease
with increasing plasticity and organic content of clays (soil 1 to soil 7). This trend can
also be seen in Figure 7-2.

The ratios of Es to cu are dependent of OCR. Figure 7-13 shows Eu/cu (Eu is
secant modulus) were plotted with OCR at two shear stress levels. The values of Eu/cu
can be either increased or decreased for low OCR. But for high OCR, ratios of Eu/cu tend
to decrease with OCR (See also Figure 7-2).

53
Figure 7-12 Eu/cu vs. Shear Stress Ratio (Ladd et al. 1977)

Figure 7-13 Eu/cu vs. OCR at Two Shear Stress Levels (Ladd et al. 1977)

54
Es/cu depends on many factors: shear stress, plasticity of clays, organic content, and stress
history (OCR). These factors need to be considered in selecting the ratio of Es/cu. Based
on analyses of Es/cu in this section, these recommendations for selecting Es/cu can be
drawn below:

1) In case of no cu data, the undrained shear strength of clays can be estimated by

procedures in Section 7.2.10. Then Es/cu or Ei/cu can be estimated from 2).
2) In case of only undrained shear strength data from laboratory or in-situ tests are
available, the values of E50/cu of 50 to 180 and the values of Ei/cu of 130 to 265
can be used with great care.
3) If cu, OCR, and PI data is available, elasticity proposed by Corps of Engineers
(1990) is recommended.

7.2.10.4 Modulus of Elasticity for Clays

If possible, lateral pile load tests are recommended and the elasticity of soils is back-
figured. These Es are most reliable values for developing the numerical model, which
needs to be calibrated before being used to predict behavior of piles under lateral loads.
Data from field tests are used for calibrating numerical models. Then values of Es can be
back-figured. Jamiolowski et al. (1985) mentioned that back-analysis from full-scale
loading tests are the most reliable assessment of geotechnical design parameters for piles
under lateral loads (Poulos and Davis 1980). Figure 7-14 shows the Es versus cu plot of
piles in clays, where Es values were back-figured from a number of published pile-test
results. It is seen that Es increases with increasing cu for both driven piles and drilled
shafts until Cu reaches some limits whereon Es become constant. For drilled shafts, this
limit is 135 kN/m2 and 95 kN/m2 for driven piles. At the same value of Cu, Es is larger for
driven piles than drilled shafts.

Soil moduli from laboratory tests usually are much lower than those from back-figured
(Jardine et al. 1986, 1985, 1984, and Poulos and Davis 1980). This may be caused by
samples disturbance and nonhomogeneities of natural soils. The limitation of laboratory
methods increases when dealing with heavily OC clays, conditions at small strains
(Jamiolkowski et al. 1985), and fissured stiff clays (Callanan and Kulhawy 1985). In-situ
tests also have their own limitations with the greatest being the interpretation of data
based on empirical correlations. Table 7-14 shows comparison of advantages and
limitations of in-situ versus laboratory testing for cohesive soils. The earliest in situ test
method is plate load test. Es from plate load tests were 1.8 to 4.8 times greater than those
determined from undrained triaxial tests (Marland 1971a). Poulos and Davis (1980)
commented that Es from plate load tests at various depths gave satisfactory load-
deflection predictions for piles.

Pressuremeter tests (PMT) give stress-strain curve that have potential for use in
predicting performance of both axially and laterally loaded foundations (Huang 1995).
Fryman, et al. (1975) reported the use of pressuremeter test results produced good
agreement between calculated and measured behavior. Self-boring pressuremeter (SBP)
gives soil moduli for use in the calculation of both vertical settlement and lateral

55
deformation of soils. SBP can shear soils in the horizontal direction and yields a
horizontal shear modulus (Jamiolkowski et. al. 1985).

Dilatometer tests (DMT) can also be used in determining Es for clays. Some studies
were reported to apply DMT with laterally loaded piles (Davidson and Boghrat 1983 and
Robertson et al. 1989).

The direct methods for determining Es of soils are laboratory testing, in-situ tests, and
back-analysis. Each method has advantages and limitations. Many factors should be
taken into consideration on selecting one of these methods. Cost-benefit, resources,
times, nature of problems, and nature of soils at site play the important roles in the
decision.

Figure 7-14 Back-figured Es for Piles in Clay (Poulos and Davis 1980)

56
Table 7-14 Comparison of Advantages and Limitations of In-situ versus Laboratory
Testing for Cohesive Soils a

In Situ Testing Undisturbed Laboratory Testing

Advantages
1. Response of large volume of soil. 1. Known soil type, nature, and
2. Response to natural environment, physical features.
i.e., in situ temperature and no 2. Well-defined boundary conditions.
stress relief. 3. Well-defined stress paths.
3. More economical and less time- 4. Strain rates can be specified.
consuming. 5. Strictly controlled drainage
4. Semi-continuous profile. conditions.
5. Can be carried out in soil in which
undisturbed sampling is still
impossible or unreliable.
Limitations
1. Unknown effects of advancing 1. Expensive and time-consuming.
devices in the ground. 2. Effects of sample disturbance may
2. Poorly defined stress and strain be difficult to identify and
boundary conditions. minimize.
3. Cannot control drainage conditions. 3. Small, discontinuous test
4. Strain fields are nonuniform and specimens.
strain rates are higher than those 4. Discontinuous nature of
anticipated in the foundation. information obtained.
5. Unavoidable stress relief.
a
Compiled from Ladd et al. 1992 and Jamiolkowski et al. 1985.

7.2.11 Ramberg-Osgood (RO) Parameters for NIKE3D

This section presents the method for the evaluation of Ramberg-Osgood (RO) parameters
and their sensitivity. There are five important parameters in the Ramberg-Osgood model:
reference shear stress (τy), reference shear strain (γy), stress coefficient (α), stress
exponent (r), and bulk modulus (K). Figure 7-15 shows its typical backbone curve. RO
parameters can be determined using RAMBO software. Input information for RAMBO
includes soil type (sand or clay) and density.

Table 7-15 shows output from RAMBO for clay with density of 3.80 slug/ft3, where Vs is
shear wave velocity from Cross-Hole test and Gmax is maximum shear modulus. If shear
wave velocity data are provided, Gmax and other parameters can be found. Shear modulus
(G) can be estimated by using 85 to 90 % of Gmax. Shear modulus can be related to bulk
modulus by EQ (7-24).

2G (1 + ν )
K= (7-24)
3(1 − 2ν )

57
where ν is Poisson’s ratio.

The effects of RO parameters on the shape of backbone curve are investigated. The
backbone stress-strain curve for monotonic loading of Ramberg-Osgood is described by
EQ (7-25):

r −1
γ τ τ
= (1 + α ) (7-25)
γ y τy τy

where γ is shear strain

τ is shear stress
γy is reference shear strain
τy is reference shear stress
α is stress constant and α ≥ 0
r is stress exponent and r ≥ 1.0

The typical stress-strain (backbone) curve for Gmax = 1436400 psf, τy = 515.34 psf, γy =
0.000359, α = 1.257, and r = 2.441 is shown in Figure 7-15, where the slope of the
backbone curve is secant shear modulus (G) defined by τ/γ at given shear strain level,
Gmax is the maximum value of shear modulus defined by τy/γy, reference shear strain
defined as the strain at the point of maximum curvature in backbone curve, and reference
shear stress is the shear stress at reference γy.

58
Table 7-15 Output from RAMBO for Clay with ρ = 3.80 slug/ft3

Vs Gmax α R γy τy
(ft/sec) (psf) (psf)
100 38000 1.257 2.441 0.0003588 13.6
150 85500 1.257 2.441 0.0003588 30.7
200 152000 1.257 2.441 0.0003588 54.5
250 237500 1.257 2.441 0.0003588 85.2
300 342000 1.257 2.441 0.0003588 122.7
350 465500 1.257 2.441 0.0003588 167.0
400 608000 1.257 2.441 0.0003588 218.1
450 769500 1.257 2.441 0.0003588 276.1
500 950000 1.257 2.441 0.0003588 340.8
550 1149500 1.257 2.441 0.0003588 412.4
600 1368000 1.257 2.441 0.0003588 490.8
650 1605500 1.257 2.441 0.0003588 576.0
700 1862000 1.257 2.441 0.0003588 668.0
750 2137500 1.257 2.441 0.0003588 766.9
800 2432000 1.257 2.441 0.0003588 872.5
850 2745500 1.257 2.441 0.0003588 985.0
900 3078000 1.257 2.441 0.0003588 1104.3
950 3429500 1.257 2.441 0.0003588 1230.4
1000 3800000 1.257 2.441 0.0003588 1363.4
1050 4189500 1.257 2.441 0.0003588 1503.1
1100 4598000 1.257 2.441 0.0003588 1649.7
1150 5025500 1.257 2.441 0.0003588 1803.0
1200 5472000 1.257 2.441 0.0003588 1963.2
1250 5937500 1.257 2.441 0.0003588 2130.2
1300 6422000 1.257 2.441 0.0003588 2304.1
1350 6925500 1.257 2.441 0.0003588 2484.7
1400 7448000 1.257 2.441 0.0003588 2672.2
1450 7989500 1.257 2.441 0.0003588 2866.4
1500 8550000 1.257 2.441 0.0003588 3067.5
1550 9129500 1.257 2.441 0.0003588 3275.4
1600 9728000 1.257 2.441 0.0003588 3490.2
1650 10345500 1.257 2.441 0.0003588 3711.7
1700 10982000 1.257 2.441 0.0003588 3940.1
1750 11637500 1.257 2.441 0.0003588 4175.3
1800 12312000 1.257 2.441 0.0003588 4417.2

59
 τ

Gmax

G
τy 1

γy γ
Figure 7-15 Typical Backbone Curve

Effect of γy on Backbone Curves

This section discusses the effect of reference shear strain on backbone curves. Using the
same data as the base curve, Figure 7-16, except γy, two new backbone curves can be
drawn with γy = 0.01 and 0.0001 respectively. Figure 7-17 shows as the reference shear
strain increases; the backbone curve flattens itself, which also results in the decrease of
Gmax when τy remains constant and γy increases or τy/γy decreases.

Effect of α
Effect of α (alpha) can be seen from Figure 7-18. As α increases, the backbone curve
flattens. The initial tangent shear modulus Gmax remains the same, but the curve beyond
reference point become flatter as α increases.

Effects of r on Backbone Curves

Stress exponent (r) is varied from 2.441 in base curve to 1.75 and 3.9. Figure 7-19 shows
that the effect of r is similar as the effect of α, but at a much greater degree. The above
sensitivity study can assist in the calibration process. Besides, the further study on the
RO parameters for different soils will be useful to the finite element analysis of
geostructure behavior and soil-structure interaction.

60
 8000
7000
6000 τy = 515.34 psf
5000 γy = 0.000359
, psf 4000 α = 1.257
3000
r = 2.441
2000
1000
0
0 0.01 0.02 0.03 0.04 0.05
γ

Figure 7-16 Typical Stress-Strain Curve (Base Curve)

8000

7000
γy = 0.000359
6000

5000 γ y = 0.0001
, psf

4000

3000

2000
γy = 0.01
1000

0
0 0.01 0.02 0.03 0.04 0.05

Figure 7-17 Effects of γy on Backbone Curves

61
 8000

7000

6000 α = 1.257
5000 α = 0.5
⎠ , psf

4000

3000

2000

1000
α = 1.75
0
0 0.01 0.02 0.03 0.04 0.05

Figure 7-18 Effects of α on Backbone Curves

8000
7000 r = 2.441
6000 r = 1.75
5000 r = 3.9
⎠, psf

4000
3000
2000
1000
0
0 0.01 0.02 0.03 0.04 0.05
γ

Figure 7-19 Effects of r on Backbone Curves

7.3 Broms Method

7.3.1 Fundamental Theory and Assumptions
In this section, Broms theory and assumptions are presented. Broms method is used to
predict the behavior of laterally loaded piles. It is based on the subgrade reaction concept
to calculate lateral deflections of piles.

7.3.1.1 Fundamental Theory

Broms proposed his theory in 1964. The objectives were to calculate the ultimate lateral
resistances (Pult), lateral ground-line deflections (y0) at working loads, and maximum
bending moments (Mmax) in pile. The working load is defined at 1/3 to ½ of the ultimate

62
lateral resistance (Pult). Broms proposed the failure mechanism for relatively short and
long piles. Figures 7-20-a and b show the failure modes for long and short piles,
respectively. For long piles, the failures take place when pile section yields (or cracks) at
the location of the maximum bending moment. For short piles, the piles are considered
to fail when they rotate and the soils along the piles fail.

Figure 7-20 Failure Modes for Short and Long Piles (Broms 1964a)

Soil reaction distributions and maximum are shown in Figures 7-21 and 22, respectively.
The ultimate lateral load and the maximum bending moment can be calculated from soil
reaction distribution. The soil reaction and bending moment distribution for short and
long piles are shown on Figures 7-22 and 7-22, respectively.

Figure 7-21 Soil Reaction and Bending Moment Distribution for Short Piles
(Broms 1964a)

63
Figure 7-22 Soil Reaction and Bending Moment Distribution for Long Piles (Broms
1964a)

For short piles, soil reaction distributions are assumed to be linear with zero soil pressure
from ground surface to depth of 1.5D. The point of rotation is the point at which soil
reaction changes to the opposite sign. The point of rotation is located at mid-way of the
distance ‘g’ in Figure 7-21. Bending moment distributions can be calculated from soil
reaction distributions. In Figure 7-21 and 7-22, the distance ‘f’ is the distance from depth
of 1.5D to the depth of maximum bending moment. cu is undrained shear strength of clay.

For long piles, the soil reaction distribution is assumed to be constant (9cuD) up to the
depth of maximum bending moment. Then the distribution is nonlinear below this depth.
k D
Behavior of piles can be specified by dimensionless length βL (β = 4 h ), where L is
4EI
length of pile below ground surface, EI the effective stiffness of piles, kh the coefficient
of horizontal subgrade reaction, and D pile diameters. In general, βL is greater than 2.25
for long piles and less than 2.25 for short piles. Also in case of short piles, Mmax are less
than Mult, for long piles, Mmax and Mult are equal. Flow chart for determining behavior of
piles is shown on Figure 7-23.

7.3.1.2 Assumptions
Broms method is simple method for solving plies under lateral loads. Consequently, the
following simplified assumptions are made:
1. Working load is 1/3 to ½ of ultimate lateral resistance.
2. At working loads, lateral forces vary linearly with lateral deflections.
3. Short pile (or rigid pile) fails when pile rotate as a unit and soil around pile fails.
4. Long pile (flexible pile) fails when plastic hinge is formed and Mmax reaches at
Mult.

64
In addition to general assumptions, there are following assumptions for piles in cohesive
soil:
5. Broms theory can be applied to driven piles into saturated clays.
6. The coefficients of subgrade reaction in both vertical and horizontal directions are
the same.
7. The coefficient of subgrade reaction is constant with depth.

One needs to take note of the above assumptions and limitations when using Broms
theory.

Start

βL < 2.25 βL > 2.25

Calculate
βL

Yes
Mult < Short Piles
Mmax

No

Set Mmax = Mult

Long Piles

65
 Figure 7-23 Flow Chart for Determining Behavior of Piles

7.3.2 Lateral Ground-line Deflections (y0) at Working Loads

This section discusses equations and graphical solutions for calculating y0. In some
situations, lateral deflections at working loads control the design of laterally loaded piles
or drilled shafts. As mentioned before, working load equals to 1/3 to ½ of Pult. In the
range of working loads, the lateral deflections are assumed to vary linearly with the
applied loads Figure 7-24 shows the pile subjected to the lateral load. The lateral
deflections at working load of piles can be calculated using both equations and graphical
solutions. The lateral deflections of short piles depend on the magnitude of loads, pile
dimensions, and soil properties. For short piles with βL < 1.5, the equation for
calculating y0 is shown in EQ (7-26).
e
4 P(1 + 1.5 )
y0 = L (7-26)
k h DL
where P is lateral load
e is the eccentricity or distance from ground surface to the point of load
application
L is the length of pile below ground surface
kh is the coefficient of horizontal subgrade reaction
D is pile diameter

For long piles with βL > 2.5, y0 can be determined from

2 Pβ (eβ + 1)
y0 = (7-27)
k∞ D
where k∞ is the coefficient of subgrade reaction for long piles.

For piles with βL up to 5.0, Broms provided graphical solutions for lateral deflections as
shown on Figure 7-25. In this graphs, ratio of e to L (e/L) is required. The calculation of
lateral deflections is based on the subgrade reaction concept. The important soil property
is the coefficient of subgrade reaction. This soil parameter will be discussed later.

7.3.3 The Ultimate Lateral Resistances of Piles (Pult)

Lateral deflections require the values of ultimate lateral loads of piles (Pult). Ultimate
lateral loads are based on hypothesized soil reaction distributions. The calculation of Pult
is based on behavior of the pile. The procedure to determine

66
 P
e
Ground Surface

D
L

Figure 7-24 Pile Subjected to the Lateral Load

Figure 7-25 Lateral Deflections at Ground Surface (Broms 1964a)

67
pile behavior was given in the previous section. The ultimate lateral resistance of pile
depends on pile behavior. Pult for short and long piles are different depending on failure
mechanism and behavior.

7.3.3.1 Short Piles

Short piles fail when they rotate as a unit and the soils around pile fail. Failure
mechanism is shown in Figure 7-20. Based on the soil reaction distribution, the bending
moment distribution can be determined. The distance f in Figure 7-21 can be calculated
from:

Pult
f = (7-28)
9cu D

The maximum bending moment can be determined by

M max = Pult (e + 1.5 D + 0.5 f ) (7-29)

Mmax is also function of the distance g as follow

M max = 2.25cu Dg 2 (7-30)

The pile length needs to be checked with EQ (7-31):

L = 1 .5 D + f + g (7-31)

The solution procedure is iterative and follows the following steps:

1. Assume ‘f’ and calculate Pult from EQ (7-28).
2. Calculate Mmax and ‘g’ from EQ (7-29) and (7-30), respectively.
3. Compute calculated L from EQ (7-21) and compare with known L. If the
calculated L is not equal to known L, value of ‘f’ needs to be adjusted.
4. Repeat from step 1 until calculated L is equal to known L.

Figure 7-26 shows the flow chart of the solution procedures for short piles.

68
 Start

Try ‘f’

Compute Pult
from EQ (4-3)

Compute Mmax and g from EQ (4-4)

and (4-5), respectively

No
Check L

Yes

End

Figure 7.26 Solution Procedures of Pult for Short Piles

69
 Figure 7-27 Graphical Solution of Pult for Short Piles (Broms 1964a)

7.3.3.2 Long Piles

Failure mechanism of long piles can be seen from Figure 7.20. Soil reaction and bending
moment distributions are shown on Figure 7-22. For long piles, the ultimate lateral
resistances can be expressed in term of the distance f.

Pult = 9cu Df (7.32)

The maximum bending moments in the piles are equal to ultimate moments of pile
sections and can be related to Pult and ‘f’ as shown in EQ (7-33).

M max = M ult = Pult (e + 1.5 D + 0.5 f ) (7-33)

From EQ (7-32) and (7-33), substitute Pult from EQ (7-32) into EQ (7-33) and rearrange
the terms. Then we get
4.5cu Df 2 + 9cu D(e + 1.5 D) f − M ult = 0 (7-34)

EQ (7-34) is a quadratic equation. This equation is solved for f as shown in EQ (7-35).

− b ± b 2 − 4ac
f = (7-35)
2a

70
where a = 4.5cuD
b = 9CuD (e+1.5D)
c = -Mult

Using positive number of f, Pult can be determined by either EQ (7-32) or (7-33).

Broms also provided the graphical solutions for Pult. Solutions for short and long piles
are shown on Figures 7-27 and 7-28, respectively.

7.3.4 Coefficient of Horizontal Subgrade Reaction (kh)

The coefficient of horizontal subgrade reaction is the most important soil parameter for
calculating lateral deflections using Broms method. Broms assumes the vertical and
horizontal coefficients of subgrade reaction to be the same. He proposed the method to
estimate the coefficient of subgrade reaction for long piles based on plate load test
results. The coefficient of subgrade reaction for long pile with diameter or width D can be
determined:

αK 0
k∞ = (7-36)
D
where α is defined as:
1 / 12
⎛ K0D4 ⎞
α = 0.52⎜⎜ ⎟⎟ ,
⎝ EI ⎠
“K0” is the coefficient for circular plate,
“EI” is the stiffness of the circular plate.
The coefficient α can be approximately determined by

α = n1n2 (7-37)

where n1 and n2 can be found from Table 7-16 and 7-17, respectively.

71
 Figure 7-28 Graphical Solution of Pult for Long Piles (Broms 1964a)

Table 7-16 The Coefficient n1 (Broms 1964a)

Unconfined Compressive Strength qu, tsf Coefficient n1

Less than 0.5 0.32
0.5 to 2.0 0.36
Greater than 2.0 0.40

Table 7-17 The Coefficient n2 (Broms 1964a)

Pile Material Coefficient n2
Steel 1.00
Concrete 1.15
Wood 1.30

The coefficient K0 can be determined from plate load tests. The coefficient of subgrade
reaction is defined by q/d0, where q is the intensity of the applied load and d0 is the
deflection of the circular plate. For Poisson’s ratio of 0.5, K0 can be computed with
known secant modulus of elasticity E50 by

K 0 = 1.67 E50 (7-38)

72
Skempton (1951) found that E50 was approximately equal to 25 to 100 times the
unconfined compressive strength of clay. The coefficient K0 can be expressed in term of
qu as

K 0 = 40qu to160qu (7-39)

K0 from EQ (7-39) can be used for determining k∞ in EQ (7-40).

The coefficients of horizontal subgrade reaction can also be estimated using

Terzaghi (1955) and Davisson (1970) methods are discussed earlier in this chapter.

7.3.4.1 Back-figured kh
The coefficient of horizontal subgrade reaction can be back figured from the load-
deflection curves of pile lateral load tests. Here the results from Dunnavant and Reese &
Welch test will be analyzed. Only the linear portion of load-deflection curves was used.
Loads up to 180 and 80 kips were used for Dunnavant and Reese & Welch data,
respectively. Pile parameters from Dunnavant and Reese & Welch are shown in Table 7-
18 and the back calculated kh values are 278 and 464 tcf, respectively.

Table 7-18 Pile Parameters from Dunnavant and Reese & Welch Tests
Pier or Pile Parameters Dunnavant Reese & Welch
Diameter (D), ft 6.0 2.5
Penetrated Length (L), ft 37.5 42.0
Effective Stiffness (EpIp), lb-in2 5.65x1012 2.01x1011
Eccentricity (e), ft 0.92 0.25

Ground-line deflection (y0) results from Dunnavant and Reese & Welch computed from
back-figured kh are shown in Tables 7-19 and 7-20, respectively. Lateral loads versus
ground-line deflections are plotted for Dunnavant and Reese & Welch and shown on
Figure 7-29 and 7-30, respectively.

73
Table 7-19 Ground-line Deflections from Back-figured kh for Dunnavant Data

Lateral Load y0 by Broms using y0 from Dunnavant

P back-figured kh Result
(kips) (in) (in)
0 0.000 0.000
25 0.013 0.022
65 0.034 0.037
85 0.044 0.040
135 0.070 0.066
180 0.093 0.093
250 0.130 0.296
300 0.156 0.589
350 0.182 0.909
400 0.208 1.291
410 0.213 1.634

Table 7-20 Ground-line Deflections from Back-figured kh for Reese & Welch Data

Lateral Load y0 by Broms using y0 from Reese &

P back-figured kh Welch result
(kips) (in) (in)
0 0.000 0.000
22 0.034 0.026
32 0.049 0.040
43 0.066 0.075
63 0.096 0.183
83 0.127 0.461
100 0.153 0.927

450

400

350

300
Load; P, kips

250

200 Dunnavant data

150

100
Broms using back-figured kh

50 Back-figured kh = 278 tcf

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Ground-line deflection; y0, in

Figure 7-29 Ground-line Deflections from Back-figured kh for Dunnavant Data

74
 120

100

80

Load ;P, kips

60
Reese & Welch data

40 Broms using back-figured

kh

20 Back-figured kh = 464 tcf

0
0.000 0.200 0.400 0.600 0.800 1.000

Ground-line deflection; y0 ,in

Figure 7-30 Ground-line Deflections from Back-figured kh for Reese & Welch Data

7.3.4.2 kh from Terzaghi and Davisson Methods

The methods for evaluating kh using Terzaghi and Davisson methods were presented
earlier in this Chapter. This section summarizes the results from both methods using
Dunnavant and Reese & Welch data. For Dunnavant data, soil parameters were divided
into two sets based on soil layers used in determining soil parameters. The first set is
based on averaging soil properties over 40 ft. The second set uses the averaged soil
properties over top 15 ft. This 15-ft layer was recommended in Broms method as
“critical depth (Lc)” or the depth wherein any change in kh will not affect lateral ground-
line deflection or the maximum bending moment by more than 10%. Lc is defined as
L/Lβ or 1/β. Table 7-21 shows kh using Terzaghi and Davisson methods on Dunnavant
data.

For Reese & Welch data, soil parameters were also divided into two sets. The first and
second sets are based on averaging soil properties over 42 and 20 ft, respectively. The
20-ft layer was the value recommended by Reese & Welch as significant depth. The
resulting kh using both Terzaghi and Davisson methods is shown in Table 7-22.

From Table 7-21, kh from set No.1 is closer to back-figured kh than ones from set No. 2.
kh using Terzaghi and Davisson are not much different. From Table 7-22, kh from set
No.1 and 2 are not much different. This is because of the undrained shear strength along
the depth for Reese & Welch site are almost uniform. Anyway, for both Dunnavant and
Reese and Welch data, back-figured kh are much higher than those from the Terzaghi
and Davisson methods.

75
Table 7-21 kh Using Terzaghi and Davisson for Dunnavant Data a

Set No Soil property Terzaghi Davisson b

1 cu, tsf 1.48 1.48

(average over 40 ft) kh, tcf 16.67 16.53

2 cu, tsf 0.97 0.97

(average over 15 ft) kh, tcf 8.33 10.83

a
kh from back analysis is 278 tcf
b
kh = 67cu/D

Table 7-22 kh Using Terzaghi and Davisson for Reese & Welch Data a

Set No Soil property Terzaghi Davisson b

1 cu, tsf 1.24 1.24

(average over 42 ft) kh, tcf 40 33.23

2 cu, tsf 1.06 1.06

(average over 20 ft) kh, tcf 40 28.41

a
kh from back analysis is 464 tcf
b
kh = 67cu/D

7.3.5 Comparisons Between Broms Method and Test Results

This section compares the results of Broms method using kh from different methods, and
with test results from both Dunnavant and Reese & Welch tests.

7.3.5.1 Dunnavant’s Pier Test

Using kh from Table 7-21, ground-line deflections can be calculated. The results from
parameter sets 1 and 2 are shown in Tables 7-23 and 7-24, respectively. Data in Tables 7-
23 and 7-24 are plotted in Figures 7-31 and 7-32, respectively. The results show the best
method for evaluating kh is back calculation and set 1 provides better result than set 2.

7.3.5.2 Reese & Welch’s Pile Test

This section presents the results using Broms method and kh from different methods and
compare with Reese & Welch test data. From Table 7-22, kh from set 1 and 2 are about
the same. Only results from set 1 are presented. Ground-line deflections from parameter
set 1 are shown on Table 7-25. Data from Table 7-25 are plotted on Figure 7-33. From
the results, the best method for evaluating kh is back analysis of kh. The coefficients of
horizontal subgrade reaction from set 1 are better than kh from set 2.

76
Table 7-23 Ground-line Deflections by Broms Method Using Parameter Set 1 for
Dunnavant Data

Lateral Load, P y0 in inches using Broms and kh by y0 from

Terzaghi set 1 Davisson set 1 back-figured kh Dunnavant data
(kips) (in) (in) (in) (in)
0 0 0 0.000 0
25 0.17 0.17 0.013 0.022
65 0.43 0.43 0.034 0.037
85 0.56 0.57 0.044 0.040
135 0.9 0.9 0.070 0.066
180 1.19 1.2 0.093 0.093
250 1.66 1.67 0.130 0.296
300 1.99 2.01 0.156 0.589
350 2.32 2.34 0.182 0.909
400 2.65 2.68 0.208 1.291
410 2.72 2.74 0.213 1.634

Table 7-24 Ground-line Deflections by Broms Method Using Parameter Set 2 for
Dunnavant Data

Lateral Load, P y0 in inches using Broms and kh by y0 from

Terzaghi set 2 Davisson set 2 back-figured kh Dunnavant data
(kips) (in) (in) (in) (in)
0 0.00 0.00 0.000 0.000
25 0.33 0.26 0.013 0.022
65 0.86 0.66 0.034 0.037
85 1.13 0.87 0.044 0.040
135 1.79 1.38 0.070 0.066
180 2.39 1.84 0.093 0.093
250 3.32 2.55 0.130 0.296
300 3.98 3.06 0.156 0.589
350 4.65 3.57 0.182 0.909
400 5.31 4.08 0.208 1.291
410 5.44 4.19 0.213 1.634

77
 450

400

350

300
Load, kips

250

200

150 Dunnavant Data

100
Broms using back-figured kh

Broms using Terzaghi set 1

50

Broms using Davisson set 1

0
0 0.5 1 1.5 2 2.5 3
Gound-line deflection, in

Figure 7-31 Ground-line Deflections by Broms Using Parameter Set 1 for Dunnavant
Data

78
 450

400

350

300
Load; P, kips

250

200

150 Dunnavant data

Broms usign back-figured kh

100

Broms using Terzaghi set 2

50

Broms using Davisson set 2

0
0 1 2 3 4 5 6
y0, in

Figure 7-32 Ground-line Deflections by Broms Using Parameter Set 2 for Dunnavant
Data

79
 120

100

80
Load, kips

60

Reese & Welch Data

40

Broms using back-figured kh

20 Broms using Terzaghi set 1

Broms using Davisson set 1

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Ground-line deflection, in

Figure 7-33 Ground-line Deflections by Broms Using Parameter Set 1 for Reese & Welch
Data

7.3.6 Summary and Conclusions

Broms proposed the method to determine the ultimate lateral resistances (Pult), lateral
ground-line deflections (y0), and the maximum bending moments (Mmax) in piles. Broms
method is simple but required a number of assumptions. Piles can be divided into two
types based on their behavior and lengths. Failure modes of

80
Table 7-25 Ground-line Deflections by Broms using Parameter Set 1 for Reese & Welch
Data

Lateral Load, P y0 in inches using Broms and kh by Reese & Welch

Terzaghi set 1 Davisson set 1 back-figured kh Data
(kips) (in) (in) (in) (in)
0 0.000 0.000 0.000 0.000
22 0.210 0.240 0.034 0.026
32 0.300 0.350 0.049 0.040
43 0.410 0.470 0.066 0.075
63 0.600 0.680 0.096 0.183
83 0.790 0.900 0.127 0.461
100 0.950 1.090 0.153 0.927

both types are shown in Figure 7-20. Lateral ground-line deflections were calculated
based on subgrade reaction concept. The most important soil parameter is the coefficient
of horizontal subgrade reaction. Broms method can only determine the lateral deflection
at working load, 1/3 to ½ of Pult; consequently it requires Pult before calculating lateral
deflections. The ultimate lateral resistance and maximum bending moment can be
determined using hypothesized soil reaction distributions. To determine Pult and Mmax, the
undrained shear strength (cu) of clay is required.

Broms provided graphical solutions for lateral deflections and ultimate lateral resistances
of piles. Equations for calculating these deflections and ultimate loads including
maximum bending moments in the piles were also given.
Broms (1964a), Terzaghi (1955), and Davisson (1970) proposed the methods for
estimating the coefficient of horizontal subgrade reaction. kh also can be back calculated
from lateral load tests on piles. Of all methods for evaluating kh, the back calculation of
kh from pile load test is the best method. Both Terzaghi and Davission method
underestimate the value of kh. The back-figured kh is 17 and 10 times greater than the
values from Terzaghi method Dunnavant and Reese & Welch data, respectively.

Broms method gives good estimation of lateral deflections at working loads if the
coefficients of horizontal subgrade reaction are estimated correctly. It is incapable of
computing the ultimate ground-line lateral deflection. Broms method using Terzaghi’s
and Davisson’s methods of calculating kh grossly overestimates the ground-line lateral
displacement.

7.4. Analyses Using LPILE Program

7.4.1 Introduction
LPILE program is developed to solve the problems of piles or drilled shafts under lateral
loads. Piles of different materials and cross sections can be analyzed. It has a capability to
analyze piles in different soil types, such as soft clay, saturated stiff clay, unsaturated stiff
clay, saturated sand, and dry sand and can calculate deflection, shear, bending moment,
and soil response along length of piles.

81
The constitutive relation of the soil around a pile is represented by p-y curve, where p
represents soil reaction pressure per unit length of pile and has the unit of pound per
square inch, and y the lateral deflection of pile at the same location. The p-y curves can
model the nonlinear behavior of soils and the pile behavior is governed by a beam-
column equation.

Soil, pile, boundary conditions, and loading data are required as input data. For soil data,
soil properties of different layers, with or without water table, are needed. Pile type,
cross section, and material properties are required. Boundary conditions at pile heads are
required in terms of loads (shear and moment) and deformations (displacement and
rotation of pile head). Two full-scale pile tests used in Section 7.3 were also used to
calibrate the LPILE program.

7.4.2 Theoretical Background of LPILE

LPILE uses the beam-column equation (Hetenyi in 1946) and the solution mechanism, p-
y method (McClelland and Focht, 1958), to solve the beam-column equation. Much effort
has been devoted to improve the p-y method (Lymon C. Reese). In 1986, the first version
of LPILE was released to the market. To solve the 4th order differential beam-column
equation, finite difference and p-y methods are required. The beam-column equation, p-y
curves, and bending characteristics of pile are explained in this section.

82
7.4.2.1 The Beam-Column Equation
The beam-column equation is the governing equation for a pile subjected to axial and
lateral loads. Hetenyi gave the solution to this equation in 1946, which was subsequently
enhanced by many researchers. Figure 7-34 shows an element cut from a beam-column
bar on elastic foundation subjected to horizontal load and a pair of compressive forces Px
acting in the center of gravity of the end cross-sections of the bar. The equilibrium of
moments (ignoring second-order terms) leads to EQ (7-40) as shown below:

( M + dM ) − M + Px dy − Vv dx = 0 (7-40)

or

dM dy
+ Px − Vv = 0 (7-41)
dx dx

Differentiating EQ (7-41) with respect to x, we get

d 2M d 2 y dVv
+ Px − =0 (7-42)
dx 2 dx 2 dx

It is noted the following identities:

d 2M d4y
= EI
dx 2 dx 4

dVv
=p
dx

p = − E py y

where Epy is soil modulus or the secant modulus of the p-y curves.

Substituting the above equations into EQ (7-42), the following equation is obtained:

d4y d2y
EI + Px + E py y = 0 (7-43)
dx 4 dx 2

83
 p = -Epyy

Figure 7-34 Element from Beam-Column Bar (Reese and Van Impe 2001)

The shearing force in the plane normal to the deflection line can be shown in term of Vv
and Px as shown in EQ (7-44):

Vn = Vv cos S − Px sin S (7-44)

where S is the angle between the element axis and the vertical axis.

For small angle of S, it is assumed that cosS = 1 and sinS = tanS = dy/dx. Then, EQ (7-
44) becomes:

dy
Vn = Vv − Px (7-45)
dx

If the distributed force W per unit length along the upper portion of a pile is applied, the
complete form of the equation used in LPILE program is shown in EQ (7-46).

d4y d2y
EI + Px − p +W = 0 (7-46)
dx 4 dx 2

where EI is flexural rigidity or bending stiffness of the pile

y is lateral deflection of the pile
x is the depth of soil below ground surface
Px is axial load on the pile
p is soil reaction or response per unit length of pile
W is distributed load along the length of the pile

84
In addition to the beam-column equation, other auxiliary expressions for analyzing piles
under lateral loads are shown in EQ (7-47) to EQ (7-50).

d4y
p = EI = − E py y (7-47)
dx 4

d3y dy
EI 3
+ Px =V (7-48)
dx dx

d2y
EI 2 = M (7-49)
dx

dy
=S (7-50)
dx

where Epy is soil modulus or the secant modulus of p-y curve

V is shear force in the pile
M is bending moment in the pile
S is slope of the axis of the pile

In deriving the differential equation, the following assumptions are made (LPILE
Technical Manual 2000):
1. The pile is straight and has a uniform cross section.
2. The pile has a longitudinal plane of symmetry; loads and reactions lie in that
plane.
3. The pile material is homogeneous.
4. The proportional limit of the pile material is not exceeded.
5. The modulus of elasticity of the pile material is the same in tension and
compression.
6. Transverse deflections of the pile are small.
7. The pile is not subjected to dynamic loading.
8. Deflections due to shearing stresses are small.

EQ (7-46) can be solved by numerical techniques. LPILE uses finite difference method
and with the aid of p-y method.

7.4.2.2 p-y Curves

The p-y method was proposed by McClelland and Kocht in 1958. This method requires
p-y curves at different depths of soil deposit. p is defined as the resultant of soil reaction
or response at certain depth with a dimension of force per unit length, and y the lateral
deflection of pile. The schematic diagram of p and y is shown in Figure 7-35. p-y curves
are mechanism to model soil behavior. p-y curves depend on soil type, depths along the

85
pile, and pile stiffness. Typical p-y curve and resulting soil modulus (Epy) is shown in
Figure 7-36.

The development of p-y curves requires both analytical and experimental procedures.
The analytical methods are used for determining soil reaction (p). Lateral deflections are
measured from full-scale lateral load tests on piles. Inclinometers are used to determine
lateral deflections of piles. Soil reactions or responses are calculated from bending
moments along the pile and bending moment is computed from the product of pile
curvature and the bending stiffness, obtained directly from pile tests.

Figure 7-35 Schematic Diagram of p and y (Reese and Allen 1977)

86
Figure 7-36 (a) Typical p-y Curve and (b) Resulting Soil Modulus (Reese and Impe
2002)

7.4.2.3 Ultimate Bending Moments (Mult)

In designing piles or drilled shafts, piles must satisfy both structural and geotechnical
requirements. For structural purpose, drilled shaft should be design as a compression
member subjected to bending and axial load. The ultimate bending moment and bending
stiffness of piles are required. The ultimate bending moment of the section is the
maximum bending moment to be resisted by the cross section before failure.
Three possible modes of failure in reinforced concrete piles subjected to combined
bending and axial load are: tension failure, balanced failure, and compression failure.
The tension failure occurs when the steel on the tension side yields, compression failure
occurs at the initial crushing of the concrete at the compression side, and the balanced
failure is the simultaneous failure in tension and in compression.

87
For pile subjected to combined compression and bending, the ultimate bending moment
of the pile can be determined using LPILE, or the analytical solution, or Whitney’s
approximate solution. The Whitney’s solution is empirical and approximate. The circular
pile is transformed to idealized equivalent rectangular and circular sections for
compression and tension failures, respectively, as shown in Figure 7-37. Figure 7-38
shows stresses and forces in the equivalent rectangular section as shown in Figure 7-37
(a).

The circular pile has outside diameter D, the diameter of reinforcement cage Ds, the total
area of reinforcement Ast, and the gross area Ag.

For compression failure, the equivalent rectangular has:

- h = 0.8D
- b = Ag/0.8D
- hs = 2Ds/3
- As’ = As = 0.5Ast

For tension failure, the equivalent circular section has:

- Ds’ = 0.75Ds
- As’ = As = 0.4Ast

For balanced failure, circular pile is transformed to the equivalent rectangular as

shown in Figure 7-37 (a). The ultimate bending moment for a circular pile subjected to
eccentric axial load can be determined using the transformed rectangular section and
expressed in EQ (7-51).

88
 (a) Equivalent Rectangular Section for Compression and Balanced Failures

(b) Equivalent Circular Section for Tension Failure

Figure 7-37 Transformed Sections for (a) Compression and Balanced Failures and (b)
Tension Failure

89
Figure 7-38 Stresses and Forces in Pile Section (Navy 2000)

_ ab _ _
M ult = Qnb eb = 0.85 f c' bab ( y − ) + As' f s' ( y − d ' ) + As f y (d − y ) (7-51)
2

ab is the depth of the equivalent stress block and expressed in EQ (7-52).

0.0003d
a b = β 1cb = β 1 (7-52)
fy
0.003 +
Es

The value of the stress block depth factor β1 can be determined from:

90
 β 1 = 0.85 0 < f c' ≤ 4000 psi

f c' − 4000
β 1 = 0.85 − 0.05( ) 4000 psi < f c' ≤ 8000 psi
1000

β 1 = 0.65 f c' > 8000 psi

where Qnb is the axial load corresponding to the balanced condition

eb is the horizontal eccentricity of Qnb from the centroid of pile section
fc’ is the ultimate compressive strength of concrete
b is the width of the pile section
cb is the distance to neutral axis for the balanced condition
_
y is the distance to the centroid of the section

The ultimate bending moment for pile under combined axial load and bending moment
can be designed using interaction diagram. The interaction diagram represents the
relationship between axial loads and the ultimate bending moments of pile section as
shown in Figure 7-39. The procedure for determining interaction diagrams can be found
in reinforced concrete design textbooks. However, LPILE also provide the interaction
diagram for a given pile section.

Figure 7-39 Typical Interaction Diagram of Pile (Reese and Impe 2002)

7.4.3 Input Data

LPILE program requires different input data grouped into three categories: pile data, soil
data, and boundary conditions and loading data. The detail of how to input these data can
be found in LPILE user’s manual.

91
7.4.3.1 Pile Data
In LPILE, the required basic pile data consist of pile dimension, material properties, and
cross-sectional details. A general description of the data is shown below:

1) Types of Pile Cross Sections: The following nine different types of cross sections
or piles (Figure 7-40) can be analyzed using LPILE:
- Rectangular or square section: this section can be reinforced concrete pile,
- Circular section for reinforced concrete bored pile or drilled shaft,
- Circular with shell but without core: it can be the composite pile constructed
using steel pipe and filled with reinforced concrete,
- Circular with shell and core: this section can be the composite section constructed
using steel pipe filling with hollow reinforced concrete,
- Circular steel pipe,
- Circular prestressed concrete pile,
- Hollow circular prestressed concrete pile,
- Solid square prestressed pile with chamfered corners,
- Hollow square prestressed pile with chamfered corners.
2) Diameter: this data corresponds to the outside diameter (D) of the pile section.
3) Total Pile Length: this number represents the total length of pile.
4) Pile Length above the Ground surface: this data is required when the pile with the
eccentricity load will be analyzed.
5) Moment of Inertia: the effective moment of inertia of pile section is specified.
6) Area: the cross-sectional area of pile section is entered.
7) Modulus of Elasticity: this data corresponds to the Young’s modulus of elasticity of
the pile section.

If reinforced concrete pile section is analyzed, the material properties and the
reinforcement details will be required. The material property data consists of the ultimate
compressive strength of concrete (fc’), the yield strength of reinforced steel (fy), the
modulus of elasticity of reinforced steel (Esteel). The rebar number, the number of rebars,
concrete coverage are used for defining the reinforcing details.

Figure 7-40 Pile Cross Sections

92
7.4.3.2 Soil Data

LPILE divides a soil deposit into a number of layers with the layer information:

1) Soil Types: There are five general types of soils and one type of rock that users may
specify: soft clay, stiff clay with water table at layer top, stiff clay without water table at
layer top, sand, silt, and weak rock
2) Unit Weight: Effective unit weight for soil at each depth is needed. LPILE will
linearly interpolate values of unit weight for soil between two specified soil depths.
3) Soil-Modulus Parameter k: This is the constant k used in the equation Epy = kx. Epy is
soil modulus and x is depth below the ground surface. The parameter k has a unit of
force per cubic length and depends on the type of soil. LPILE user’s manual provides the
numerical values of k as shown in Table 7-26.
4) Undrained Shear Strength: Undrained shear strength (cu) values are required for
clays and silts at each depth. cu is not needed for sands.
5) Angle of Internal Friction of Soil: Values of the angle for sands or silts are entered at
each depth in degree. This input data is not required for clays and rocks.
6) Soil Strain: Values of strain at 50% of the maximum stress (ε50) are for clays and silts
and not for sand. The values are assigned at each soil depth. The values of ε50 are given in
Table 7-27 and provided by LPILE user’s manual.

Other rock properties and parameters are required for piles in weak rocks. They are
modulus of elasticity, unconfined compressive strength, rock quality designation, and
parameter krm. The parameter krm ranges between 0.0005 and 0.00005 (LPILE user’s
manual 2000).

Table 7-26 Soil-Modulus Parameter k for Clays (LPILE User’s Manual 2000)

Clay Type Average Undrained Shear Parameter

Strength (cu), psi k, pci
Soft 1.74 – 3.47 30
Medium 3.47 – 6.94 100
Stiff 6.94 – 13.9 500
Very Stiff 13.9 – 27.8 1000
Hard 27.8 – 55.6 2000

Table 7-27 Strain at 50% of The Maximum Stress (ε50) for Clays (LPILE User’s Manual
2000)

Clay Type Average Undrained Shear ε50

Strength (cu), psi
Soft 1.74 – 3.47 0.020
Medium 3.47 – 6.94 0.010
Stiff 6.94 – 13.9 0.007
Very Stiff 13.9 – 27.8 0.005
Hard 27.8 – 55.6 0.004
93
7.4.3.3 Boundary Conditions and Loading Data
The following five boundary conditions can be assigned:

1) Shear and Moment: Values of applied lateral load (Pt) and applied moment (Mt) at
the pile head can be specified. This condition indicates that the pile head is free to rotate
and move in lateral direction (free-head condition). The positive values of lateral load are
defined when applied from left-to-right. The positive values of moments are defined
when applied clockwise.

2) Shear and Slope: Applied lateral load (Pt) and the pile head rotation (θt) can be
specified. The pile head rotation can be assigned in term of the angle between the pile
axis and the vertical axis. This angle has a unit of radian. The rotation is positive when
the pile head rotates counterclockwise. The lateral load is positive when applied from
left-to-right. This boundary condition implies that the pile head is fixed (fixed-head
condition). This means the lateral movement of the pile head is allowed but no pile head
rotation is permitted.

3) Shear and Rotational Stiffness: Applied lateral load (Pt) and a value of rotational
stiffness are assigned. The rotational stiffness at the pile head is defined as a ratio of
applied moment (Mt) at the pile head to the pile head rotation (θt). The values for
rotational stiffness are always positive. In this boundary condition, a fixed-head
condition will be specified if a large value of rotational stiffness is assigned. If the user
intends to simulate an elastically restrained type of pile-head connection, this boundary
condition can be served the user’s intension.

4) Displacement and Moment: The lateral pile head displacement (yt) or deflection and
the pile head rotation (θt) are specified. The displacement is positive when applied from
left-to-right. The positive moment is defined when applied clockwise. This boundary
condition represents the free-head condition.

5) Displacement and Pile Head Rotation: Value of the lateral pile head displacement
(yt) and the pile head rotation (θt) are assigned in this boundary condition. The positive
lateral pile head displacement is considered positive when applied from left-to-right. The
pile head rotation is positive when the pile head rotates counterclockwise. This boundary
condition implies the free-head condition.

Values of axial loads applied at the pile head may be input after specifying the boundary
conditions and corresponding loads. Axial loads are only utilized to account for
secondary moments produced when the pile is laterally deflected (P-∆ effects).

7.4.4 Evaluation of LPILE Program using Field Test Results

Two full-scale lateral load tests on piles were used for calibrating LPILE program. These
full-scale test results were also used previously to evaluate Broms method. The first test
was conducted on 6-foot diameter pier by Dunnavant in 1986. The second test was
performed on 2.5-foot diameter pile by Reese and Welch in 1975.

94
7.4.4.1 Dunnavant’s Pier Test
Two soil systems were investigated. The first soil system is uniform soil layer. The
second system is two-layered soil. For two-layered soil, the upper and lower layers are
18-ft and 19.5-ft thick, respectively. The ground-line deflections using LPILE for pier in
uniform soil and two-layered soil are shown in Table 7-28. The ground-line deflections
for pier in both soil systems using LPILE are plotted with Dunnavant’s measured results
as shown in Figure 7-41.

Table 7-28 Ground-line Deflections (y0) for Pier in Uniform Soil and Two-layered
Soil using LPILE with Dunnavant Data
Lateral load y0 from LPILE y0 from LPILE y0 from
for pier in 2-layered soilfor pier in uniform soil Dunnavant
(kips) (in) (in) (in)
0 0.000 0.000 0
25 0.011 0.008 0.022
65 0.044 0.025 0.037
85 0.065 0.036 0.040
135 0.130 0.069 0.066
180 0.202 0.104 0.093
250 0.344 0.170 0.296
300 0.470 0.226 0.589
350 0.621 0.290 0.909
400 0.803 0.366 1.291
410 0.843 0.383 1.634

From Figure 7-41, ground-line deflections for pier in uniform soil fit the initial curve
better than those for pier in two-layered soil. However, two-layered soil model tends to
predict the nonlinear behavior of pier better than uniform soil model.

95
 450

400

350

300

250
P, kips

200

Dunnavant Data
150

LPILE: 2-layered soil

100
LPILE: uniform soil

50

0
0 0.5 1 1.5 2

y0, in

Figure 7-41 Ground-line Deflections using LPILE for Dunnavant’s Test Results

7.4.4.2 Reese & Welch’s Pile Test

For Reese & Welch case, the two-layered soil was used in the analysis. The upper layer is
unsaturated stiff clay with the thickness of 20 feet and the bottom layer is saturated stiff
clay with thickness of 22 feet. Ground-line deflections using LPILE for Reese & Welch’s
pile are shown in Table 7-29. These deflections were plotted along with Reese &
Welch’s results as shown in Figure 7-42.

As shown in Figure 7-42, the ground-line deflections from LPILE show good agreement
with results from Reese & Welch data; however, the analysis underestimated the ultimate
ground-line deflection.

96
Table 7-29 Ground-line Deflections (y0) for Pile using LPILE with Reese & Welch Data

Lateral Load y0 from LPILE y0 from Reese

& Welch Data
(kips) (in) (in)
0 0.000 0.000
22 0.024 0.020
32 0.054 0.040
43 0.102 0.065
63 0.231 0.187
83 0.415 0.452
100 0.615 0.942

7.4.5 Summary and Conclusions

LPILE program was developed on the basis of the beam-column equation and p-y curves.
The beam-column equation can be solved by both analytical and numerical procedures
for simple cases. In LPILE, finite difference method was used in solving the beam-
column equation. The concept of p-y curves was also employed in analyzing piles under
lateral loads. The p-y curve concept was initially developed by McClelland and Focht. p-
y curves were used to model the nonlinear soil and pile behavior. The nonlinear behavior
of pile under lateral load depends partly on both soil and pile stiffness. The nonlinear
behavior of soil is represented by an unconnected series of nonlinear springs. The pile
stiffness depends on cross-sectional area and material properties of pile. The pile
stiffness indicates the capacity of pile in resisting bending moment. The ultimate bending
moment of pile section can be calculated for a given cross section and material property
of pile.

97
 120

100

80
P, kips

60

Reese & Welch Data

40
LPILE

20

0
0.0 0.2 0.4 0.6 0.8 1.0

y0, in

Figure 7-42 Ground-line Deflections using LPILE for Reese & Welch Data

Input data for LPILE can be grouped into three categories i.e. pile data, soil data, and
boundary conditions Loading data. LPILE allow users to analyze nine different cross-
sectional shapes of piles. Five soil conditions and one rock are available in LPILE
program. Five boundary conditions and loading data can be represented the conditions
and loads at the top of piles.

Two full-scale lateral load tests on pier (Dunnavant, 1986) and pile (Reese & Welch,
1075) were used for evaluating LPILE performance of piles under lateral loads. The
ground-line deflections are in good agreement with the value at working load, and were
underestimated at the ultimate load. Soil parameters used in LPILE can be estimated
correctly using undrained shear strength (cu) from unconsolidated undrained (UU) triaxial
test.

98
8. SUMMARY, DISCUSSIONS, CONCLUSIONS, AND RECOMMENDATIONS
FOR FURTHER STUDY

Summary

In 1986 a group of five state bridge engineers including Colorado met in Denver and
drafted a letter to the Subcommittee on Bridges and Structures including their concerns
that the AASHTO Specification was falling behind the times. In August of the same year,
the NCHRP Project 20-7/31 on “Development of Comprehensive Bridge Specification
and Commentary” was initiated. The effort resulted in the current AASHTO LRFD
Bridge Design Specifications. One might say, “Colorado is one of the five pioneer states
in initiating the study on probabilistic and reliability design specifications for bridges that
resulted in the AASHTO LRFD Bridge Design Specifications.” CDOT needs to carry this
pioneer spirit forward with the complete implementation of the LRFD code. While its
implementation for superstructure design is already in place, the implementation for
substructures is not so straightforward. It requires the evaluation of the Colorado-specific
resistance factors. The task requires systematic laboratory and field testing of strength
and deformability of soils and rocks, and field full-scale load tests on some selected
foundations. The implementation of the LRFD code in geotechnical investigation design
and investigation will need major investment of time and money and it will take a few
years depending on the amount of investment. The national survey shows that over 70%
of foundation for bridge support is deep foundation with drilled shafts (21%) and driven
piles (51%). This statistic does not differ much from the CDOT practice.

Discussions

Extended discussions were held between the members of the study panel and the P.I. of
this research project. The essentials are presented as follows:

Questions from the Study Panel and PI’s Responses

Question In the current political environment the increase in staff is highly unlikely.
Therefore, in addition to the "increase staffing" option, please provide an alternate
strategy that does not require increase in staff and state-owned equipment. In other
words, please provide an alternative that relies on outsourcing of additional testing
needed.

The implementation of the geotechnical LRFD in Colorado requires the Colorado-

specific resistance factors, which, in turn, needs the creation of database for all
geotechnical material parameters for each geological materials, soils and rocks. This will
take:
• Choice of a design method(s).
• Evaluation of material parameters for the chosen design method.
• Establishment and maintenance of the database for all material parameters.

99
 • Determine the probability density function for each design parameter with
sufficient statistical sample size.
• Formulation of resistance factor evaluation procedures.
• Calibration of the resistance factors.
• Development of the geotechnical LRFD procedures by adopting the Colorado-
specific resistance factors.
• Training of the CDOT personnel in the use of geotechnical LRFD procedure.
• Technology transfer from CDOT to the professional engineering community and
local government agencies.
• Adoption of the Colorado-specific geotechnical LRFD.

If the current political environment were strictly financial, then outsourcing does not
make sense because it might cost even more to the Colorado tax payers. But if CDOT
were to bypass the solution of its geotechnical staffing and field testing equipment
problems, then, to meet the FHWA mandate deadline, it has no choice but to outsource
all its geotechnical LRFD tasks. In the long run, the CDOT Geotechnical and Soil-
Rockfall programs will lose the design capability and become engineering regulatory
design evaluation units.

Question: In regards to Implementation Recommendations Nos. 4 and 5: How will the

purchase and use of CPT, PMT, GE, and VST equipment move us toward LRFD?

PI Response: So far the geotechnical investigation at CDOT relies almost exclusively on

SPT, the roughest method of field investigation. The use of CPT, etc will improve the
precision of material parameter evaluation, and, thereby, enhance the reliability of the
LRFD geotechnical design process.

Question: In your report you state that we should focus on SPT and DMF, so you need to
explain your strategy of expanded use of CPT, PMT, GE, and VST.

PI Response: Since SPT has been the only field investigation tool available to CDOT, all
foundation designs are based on blow count, N. This method of foundation design is
locally called the Denver Magic Formula (DMF) approach. The kind of general practice
is not observed anywhere else in the country as shown in the national survey of state
DOT’s. So CDOT is so entrenched in the DMF practice, it is very difficult and unrealistic
to stop the practice and switch to a more rational practice immediately. Thus, it is advised
to continue the practice and transition the design procedure to the eventual geotechnical
LRFD procedures. The development of the new LRFD procedure requires the use of the
field test devices more precise than SPT, which include PMT, CPT, VST and GE, etc
devices for more reliable assessment of the material parameters and subsurface
stratification. The acquisition of field test equipment is dictated by the funding
availability, and can be pursue in a financially sound time frame. When the equipment
funding is not available, the job will obviously have to be outsourced to different
geotechnical companies or universities.

Question: You recommend that CDOT should be moving toward a rational method; what

100
rational method do you recommend? Why?

PI Response: Two types of design methods are available: empirical and rational. The
former is based mainly on the observation and the latter is mechanistic. There are times
where mechanistic and empirical methods can coexist to serve as mutual check. Using the
empirical method without serious verification can be either un- or over-conservative. In
the design of deep foundation, the DMF method is empirical based on experience and
observation, while the Terzaghi’s bearing capacity, for instance, is a rational mechanistic
approach. In the former the allowable bearing capacity is given as N/2 ksf (kips per
square foot) and in the latter the bearing capacity of a deep foundation includes two
terms: side friction and end bearing, both are functions of angle of internal friction and
cohesion of soils, fundamental Mohr-Coulomb strength parameters of soils and rocks that
can be evaluated in laboratory. The blowcount in the Denver Magic Formula is an all-
inclusive parameter and not a fundamental engineering parameter. Therefore the design
approach based on the fundamental engineering parameters is more rational. It is good to
move toward a rational design approach to enable the communication among all design
engineers, in/out of state.

Question: Absent a load test at each location, how will soil parameters and boring logs
be used to establish failure probability curves?

PI Response: Two separate issues are raised: probability density function (pdf), and
nominal strength of soils or rocks. The nominal strength of a foundation is evaluated
based on a selected design method (or equation) for computing the nominal strength,
which involves the material parameters and a boring log aids in the selection of material
parameters by identifying the material type. The former is obtained in a laboratory/ field
test and the latter revealed in a subsurface exploration program. If an engineer is
equipped with the boring log and soil parameters for the specific location then the design
is straight forward by using the parameters provided and the equation selected.

The stratification, soil/rock types, and associated material parameters vary in a random
fashion and probabilistic approaches are needed to rational describe their variation, i.e. a
probability density function (pdf) is needed for each parameter. Probabilistic approaches
also require calibration against field load tests, which are usually expensive and can be
performed at all sites. Thus, the database of field load test results is extremely valuable in
serving as the calibrator for the rational approach and also the LRFD procedures when a
sufficient statistical sample size is achieved. Thus, in the Colorado geotechnical LRFD
(G-LRFD) effort, more field load tests should be performed to enhance the database to be
used in the calibration of G-LRFD procedures.

Additionally even a theoretical formulation involves some uncertainty because of the

simplifying assumptions to make the problem mathematically manageable. Like in the
deep foundation design, there are several formulations for bearing capacity computation
with Meyerhoff's solution giving the highest capacity, while the Terzaghi’s, the lowest or
most conservative and the load tests can provide precision check. A load test must be
physically sound and conducted accurately to be useful in serving as a calibrator. For

101
instance, questions for the Osterberg Cell load test (O-C load test) include: Can it provide
an accurate top–down load capacity? How can it serve reliable settlement assessment for
deep foundations? The questions need to be investigated before the extensive use of the
O-C load test for the top-down behavior prediction.

Question: Your recommendation for a database is probably a good one, but we need to
know what kind of analysis will be performed at some future date to make sure we are
collecting and storing the right data.

PI Response: You are exactly right. This is why I recommend the formation of the G-
LRFD Committee to delineate the approach to the problems of identification of analysis
method, parameters needed, and the method of testing, etc.

Question: Also database maintenance is costly and we would need to demonstrate that
we would be getting something of sufficient value to justify the investment. Is our
current standard drilling practice sufficient for a viable database or do we need to perform
additional tests at each hole?

PI Response: CDOT uses nearly exclusively SPT in its drilling practice. The blowcount
reflects approximate strength through it correlation with the field load tests, and provides
samples, which are, in principle, good only for examination of soil types and index
property tests. Because of the shortage of geotechnical staff, CDOT has been unable to
utilize its excellent laboratory strength test facility to aid in the evaluation of strength
parameters needed in the foundation design. In sum, the current CDOT geotechnical
investigation practice is held to the minimum and needs improvement for the effective
implementation of the geotechnical LRFD design procedures. The geotechnical
investigation should be extended from SPT to cover laboratory and field testing,
whenever possible. In the CDOT quest for the implementation of G-LRFD procedures
the establishment and maintenance of material property databases are optional, but are
needed.

It is recommended to establish and maintain the databases on a pool-fund basis among

different states because all states face the same problem and the cost will reduce as more
states participate in the G-LRFD effort. Besides, a financially-sound implementation time
frame can be established to avoid unbearable financial burden.

Question: You recommended that we use CPT, PMT, GE, and VST tests. When should
we perform these tests? On what type of soils and rocks? How often?

PI Response: The following tests are routinely required: SPT, CPT (soils only) and
PMT. VST is recommended for the investigation of soft and medium clay deposits and
GE in a big project. In a big project, a series of GE tests will delineate subsurface
stratification on a large-scale basis. This can then be followed by SPT and CPT tests for
general exploration at selected locations. Pros and cons of SPT and CPT are outlined as
follow:

102
 SPT CPT

Pros 1. Easy to conduct. 1. Easy to conduct.

2. Equipment is readily available 2. Excellent delinea-
3. While somewhat disturbed, it tion of strength and
does provide samples, at least, for stratification
examination and maybe for testing

Cons 1. Results do not reflect the static 1. No sample for properties.

strength of soil or rock examination or testing.
2. Detail delineation of layers is not
possible.

Thus, it is best to perform SPT and CPT in parallel side by side to provide subsurface
information for the selection of other laboratory/field tests needed in further
investigations to provide accurate material parameters for the critical subsoils and rocks.

PMT (pressure meter test) and VST (vane shear test) are usually requested by a design
engineer to provide more reliable properties and/or strength data at the locations
considered critical to the design project. PMT can be performed on both rock and soils to
provide stress-strain-strength parameters, coefficient of earth pressure at rest, Young's
modulus, compressibility, etc. VST provides the undrained shear strength of clayey soils.
Situations do occur when VST is performed to provide the in-situ strength of clay,
particularly in a stability investigation. Not all tests are required at all sites.

Question: There are several foundation design models available. Which one do you
recommend we use? The answer may not only depend on accuracy but also on the cost
and ability to collect and test rock samples to derive the parameters for the model.

PI Response: The answer to this question involves the following factors: effectiveness
and reliability of each design model, effort needed to provide the design material
parameters, designer’s preference, and cost, etc. A design engineer will have to select one
or more theories for comparison. In the deep foundation design, among many methods,
one may pick DMF, Terzaghi method, and/or Meyerhoff method, etc. Each method
comes with some deficiencies. For instance, DMF lacks the backing of a serious study for
its effectiveness for Colorado soils and rocks and the blowcount reflects only the bulk or
approximate strength of soils or rocks and not the basic engineering property. Because of
the tremendous experience in using DMF approach, it might be good for CDOT to
critically examine its effectiveness during its transition to the general G-LRFD
procedures. Additionally the Terzaghi and/or Meyerhoff methods ought to be
investigated for their effectiveness in Colorado geological environment. It is a very
difficult problem to answer and the choice is very personal. To avoid the answer being
tainted by the PI’s personal opinion, it is recommended to allow the G-LRFD Committee
to jointly make the unbiased selection of the design method.

103
Question: How will soil parameters and boring logs be used to establish failure
probability curves? Will the data tell us how variable our rocks are?

PI Response: The boring logs are used to identify the stratification of subsoils and rocks.
Each soil and rock can be tested for its design material parameters. When sufficient
statistical sample size is achieved for a parameter, its probability density function (pdf)
can be formulated. The pdf’s of all design parameters are entered into a selected design
method to compute the pdf for the nominal strength. The nominal strength probability
density curve can then be compared to the load pdf to determined the failure probability
as outlined in Chapter 2 on “Fundamental of LRFD.” The data will show the variability
of soils and rocks as demonstrated by the histograms, pdf curves, and the mean, standard
deviation and coefficient of variation of a parameter.

Question: Your recommendation of an LRFD foundation committee is a good one, but

they need a starting point. Visualize CDOT in five year after we have been collecting
data on every hole we drill. What analyses should we perform on the database? Toward
what objectives?

PI Response: The establishment of geotechnical design criteria, procedures, and theory

should be objective and not biased by the decision of a single individual. Many methods
(or theories) are available for the deep foundation design. To avoid a biased choice, the
decision and selection should be made by the collective wisdom of the G-LRFD
Committee through serious open deliberation. Thus, this report has not attempted to make
the choice for the State of Colorado. Instead, the committee will assume the task. This
report serves as a good starting point for the G-LRFD Committee to begin its work and
deliberation.

If serious about the implementation of the G-LRFD design procedures, CDOT must
invest, whether in-house or outsourcing. The tentative tasks for the committee as
proposed in this study, when it convenes include the following:

• Decide the core committee with the membership from FHWA representative,
CDOT Bridge, Geotechnical Program and Research and UCD representative.
• Convene the core committee meeting to choose the LRFD Committee
membership from structural and geotechnical consulting engineering industries
and other collaborative government agencies.
• Convene the first big committee meeting to establish agenda and progress
schedule, and assign responsibility, etc.
• Select a foundation type (s) for first implementation of the Geotechnical LRFD
(G-LRFD) procedures, design theories (or models, methods), design parameters,
types of tests and test procedures, and the Database Center (DC).
• I recommend the deep foundation as the first foundation type for the G-LRFD
implementation. Once the process progresses to an extent that the committee is
ready to move onto another foundation type or geotechnical problem, then the
implementation effort for the new foundation type can be gradually phased in.

104
 • DC will be responsible for updating and analyses of database, formulation of
probability density function (pdf), evaluation of nominal strength, and evaluation
of resistance factors based on the reliability theory. The database center will
report to the LRFD Committee on a quarterly basis of the progress made.
• DC will inform the LRFD Committee to call a meeting(s) to discuss the status of
data collection and analysis and timing for the initiation of the effort for the
evaluation of resistance factors and eventual implementation of G-LRFD
procedure for the selected foundation type(s).
• LRFD Committee will initiate another geotechnical problem(s) for G-LRFD
implementation and process continue until the full implementation is achieved.
• CDOT can initiate a pool fund study among willing participating states, like the
neighboring states and a separate committee can be established to carry out the
pool fund study effort. DC can expand its effort to cover the data collection and
analysis for the data from the out-of-state participants.

CDOT is still leading the G-LRFD implementation effort among the neighboring
states, and can continue to take a leadership role if it initiates the effort soon before
others begin to catch up.

Question: Just because there is a specific testing device available, doesn't mean we buy
one and use it on every borehole sample. We need to decide if it is cost-effective. The
data collection plan must be based on some vision of how we are going to use it.

PI Response: You are right again. The purchase choice of field test equipment must be
based on its applicability to the Colorado geological materials, the CDOT fiscal strength,
and the chosen design method. Not all equipment is good for projects, soils and rocks.
Funding dictates any project activity and its extent. Under the Colorado budget constraint
as pointed out by the CDOT Research, the purchase choice must be prioritized and the
actual purchase is priority based. The equipment most appropriate to the Colorado
geological and budget environments is first purchased and so on. The sole objective of
performing different types of tests is to enhance the data reliability, and better and more
reliable design of a foundation structure. Structural LRFD is much more straight-forward,
because of much less material uncertainty than the material uncertainty in the
geotechnical design. Thus, to implement geotechnical LRFD, we need to have a much
better understanding of material properties by establishing the statistical database of
material parameters obtained from more reliable laboratory and/or field tests for all
relevant materials. One needs to bear in mind that the implementation of G-LRFD is not
be free. It is a major financial and budget commitment to the CDOT Administration and
the Colorado tax payers. While out-sourcing might be an alternative, somehow the CDOT
needs to keep some in-house expertise to be cost effective and to lead the engineering
industry in Colorado.

There are three different levels of G-LRFD procedure calibration, from simplest to most
sophisticated, as follows:

105
 • Level I is referred to as a mean value first order and second moment (MVFOSM)
procedure and the value of the reliability so evaluated is least precise.
• Level II is advanced first order and second moment (AFOSM) method. It
provides more precise reliability.
• Level III, a fully probabilistic and most complex method, requires the knowledge
of pdf of each random material parameter and the correlation between all
parameters. It is not used in the general calibration of the G-LRFD procedure.

CDOT’s G-LRFD implementation can begins at the Level I for now and advance to a
more advanced procedure in the future. This would allow the G-LRFD implementation
effort to proceed and progress.

Conclusions

1. The Geotechnical and Soil-Rockfall Programs are severely understaffed. The

national statistics give an average geotechnical staff size of 32.27; the average for
the Colorado neighboring states is 34.3. This manpower deficiency can hamper
the delivery of quality service of the programs and staff morale. Most
importantly, the safety of the traveling public could be affected in an event of
sinkholes, bridge foundation collapses, rockfalls and landslides, etc.
2. The CDOT foundation design is nearly exclusively based on the result of the
standard penetration test (SPT) and so called Denver Magic Formula. The
survey indicated that no other states in the Union practice the DMF-type design
procedures. Instead, they all based their foundation designs on a rational design
approach more in line with what AASHTO requires.
3. The survey shows that CDOT has laboratory testing facility comparable in
quality to those of other states. However, it is severely deficient in the field
investigation equipment for geophysical exploration (GE), cone penetration test
(CPT), Menard pressure meter test (PMT), and vane shear test (VST), which are
routinely used in other states in geotechnical investigation.
4. CDOT does not have any staff responsible for foundation design. This results in
significant operational chaos in the bridge substructure design. In a reasonable
operation, the geotechnical programs provide subsurface investigation results and
design soil parameters, and the bridge staff provides design requirements to
foundation engineers, who are responsible for the foundation design. It might be
a good idea to create a foundation engineer position.
5. CDOT does not have any in-house capability for full-scale foundation load tests.
This can be economical when a large number of tests are recommended for both
design calibration and evaluation of resistance factors for Colorado-specific soils
and rocks.

Recommendation for further study

1. Establish a G-LRFD Committee, a think-tank group, to formulate the strategy and

plan for an effective G-LRFD implementation. The committee membership

106
 should include some responsible persons from the CDOT Central Administration,
Bridge Branch, Geotechnical Program, and Soil-RockFall Program; one
representative from each of the structural and geotechnical consulting industries;
and an academic representative knowledgeable in LRFD, probability and
reliability.
2. Hire additional geotechnical engineering and investigation staff required for the
laboratory and field testing, quality service delivery, and implementation of G-
LRFD procedures.
3. Create a foundation engineer position(s) responsible for all foundation designs.
The person(s) should have the overlapping capability of structural, geotechnical
and foundation design to effectively communicate with both geotechnical and
structural staff.
4. Upgrade the field investigation facility by purchasing PMT, CPT, GE, and VST
equipment whenever the budget permits.
5. Activate the excellent laboratory testing apparatuses and place them in a
production line to generate urgently needed laboratory data of soils and rocks for
the formulation of resistance factors for Colorado-specific geological materials.
6. Select the deep foundation as the first foundation type for G-LRFD implementa-
tion and formulate detailed procedures for the evaluation of resistance factors for
all Colorado-specific geological materials. The practice can be extended and
gradually phased in to other areas of geotechnical investigations and designs.
7. Continue to practice and calibrate DMF, while G-LRFD implementation effort
progresses. The possibility of enhancing DMF for Colorado geotechnical design
and investigation should be explored, while the implementation of a G-LRFD
design procedure more in line with the AASHTO procedures progresses.
8. Develop in-house full-scale load test capability to check the design recommenda-
tion and calibrate the G-LRFD recommendation.

107
 REFERENCES

AASHTO (2001). Standard Specifications for Structural Supports for Highway Signs,
Luminaires and Traffic Signals. 4th ed. Washington D.C.: American Association of
State Highway and Transportation Officials.

AASHTO (2004). AASHTO LRFD Bridge Design Specifications 3th ed.” Washington
D.C., American Association of State Highway and Transportation Officials.

ACI Committee 318 (1992). Building Code Requirements for Reinforced Concrete
(ACI 318-89) and Commentary (ACI 318R-89). Revised ed. Detroit, Michigan:
American Concrete Institute.

Banerjee, P. K., & Davies, T. G. (1978). The Behaviour of Axially and Laterally
Loaded Single Piles Embedded in Nonhomogeneous Soils. Geotechnique, Vol. 28,
No. 3, 309-326.

Barker, R. M., J.M. Duncan, K.B. Rojiani, P.S.K. Ooi, C.K. Tan, and S.G. Kim
(1991), “Manuals for the Design of Bridge Foundations: Load Factor Design
Criteria for Highway Structure Foundations,” NCHRP Report 343, Transportation
Research Board, National Research Council, Washington, D.C.

Benjamin, J.R., and C.A. Cornell, (1970), “ Probability, Statistics, and Decision for
Civil Engineers,” McGraw Hill, New York, N.Y.

Bhushan K., Haley, S. C., & Fong, P. T. (1979). Lateral Load Tests on Drilled Piers
in Stiff Clays. J. of Geotechnical Engineering Div., Proc. of The ASCE, Vol. 105, No.
GT8, August, 969-985.

Broms, Bengt B. (1964a). Lateral Resistance of Piles in Cohesive Soils. J. of the Soil
Mechanics and Foundations Div., Proc. of the ASCE, Vol. 90, No. SM 2, March, 27-
63.

Broms, Bengt B. (1964b). Lateral Resistance of Piles in Cohesionless Soils,” J. of the

Soil Mechanics and Foundations Div., Proc. of the ASCE, Vol. 90, No. SM 3, May,
123-156.

Broms, Bengt B. (1965). Design of Laterally Loaded Piles,” J. of the Soil Mechanics
and Foundations Div., Proc. of the ASCE, Vol. 91, No. SM 3, May, 79-99.

Chen, Yit-Jin, & Kulhawy, F. H. (1993). Undrained Strength Interrelationships

among CIUC, UU, and UC Tests. J. of Geotechnical Engineering Div., Proc. of The
ASCE, Vol. 119, No. 11, November, 1732-1750.

108
Davies, T. G., & Budhu, M. (1986). Non-linear Analysis of Laterally Loaded Piles in
Heavily Overconsolidated Clays. Geotechnique, Vol. 36, No. 4, 527-538.

Davisson, M. T. (1970). Lateral Load Capacity of Piles. Highway Research Record,

No. 333, 104-112.

Degroot, D. J., & Sheahan, T. C. (1995). Laboratory Methods for Determining

Engineering Properties of Overconsolidated Clays. Transportation Research Record,
No. 1479, 17-25.

Dunnavant, Todd W. (1986). Experimental and Analytical Investigation of The

Behavior of Single Piles in Overconsolidated Clay Subjected to Cycle Lateral Loads.
Ph.D. Thesis. Houston, Texas: University of Houston.

Dunnavant, T. W., and O’Neill, M. W. (1989). Experimental p-y Model for

Submerged, Stiff Clay. J. of Geotechnical Engineering Div., Proc. Of The ASCE, Vol.
115, No. 1, January, 95-114.

Foott, R., & Ladd, C. (1973). The Behavior of Atchafalaya Test Embankments During
Construction. Research Report No. R73-27. Cambridge, MA: Massachusetts Institute
of Technology.

Huang, An-Bin. (1995). In Situ Testing in Overconsolidated Clays. Transportation

Research Record, No. 1479, 35-42.

Jamiolkowski, M., Ladd, C., Germaine, J. T., & Lancellotta, R. (1985). New
Developments in Field and Laboratory Testing of Soils. Proc. of the 11th Int.
Conference on Soil Mechanics and Foundation Engineering: Vol. 1, San Francisco,
57-153.

Ladd, C. C., Foott, R., Ishihara, K., Schlosser, F., & Poulos, H. G. (1977). Stress-
Deformation and Strength Characteristics. Proc. of the 9th Int. Conference on Soil
Mechanics and Foundation Engineering: Vol. 2, Tokyo, 421-494.

FHWA, Publication No. FHWA HI-98-032, May 2001.

Mahar, L. J., & O’Neill, M. W. (1983). Geotechnical Characterization of Desiccated

Clay. J. of Geotechnical Engineering Div., Proc. of The ASCE, Vol. 109, No. 1,
January, 56-71.

Mayne, P. W., Kulhawy, F. H., & Traumann, C. H. (1992). Experimental Study of

Undrained Lateral and Moment Behavior of Drilled Shafts During Static and Cyclic
Loading. Research Report No. EPRI TR-100221, Project 1493-4. Ithaca, NY: Cornell
University.

109
McClelland, B., & Focht, Jr., J. A. (1956). Soil Modulus for Laterally Loaded Piles.
J. of the Soil Mechanics and Foundations Div., Proc. of the ASCE, Vol. 82, No. SM
4, October, 1081-1 to 1081-22.

McVay, M.C., C.L. Kuo, W.A., and W.A. Singletary (1998), “Calibrating Resistance
Factors in the Load and Resistance Factor Design for Florida Foundations,” Final
Report, State Project No.: 99700-3339-010, Florida Department of Transportation.

Nawy, E. G. (2000). Reinforced Concrete: A Fundamental Approach. 4th ed. Upper

Saddle River, NJ: Prentice Hall.

Nowak, A. S. (199), “Calibration of LRFD Bridge Design Code,” NCHRP Report

368, NCHRP, TRB, Washington D. C.

O’Neill, M. W., & Reese, L. C. (1999). Drilled Shafts: Construction Procedures and
Design Methods. Report No. FHWA-IF-99-025. Washington D.C.: Federal Highway
Administration.

Pakowsky, S.G. (2004), “Load and Resistance Factor Design for Deep Foundations,”
NCHRP Report 507, TRB, Washington, D.C.

Pike, R, & Beikae, M. (1984). A New Solution for the Resistance of Single Piles to
Lateral Loading. In Langer, J. A, Mostley, E. T., & Thompson, C. D. (Eds.), Laterally
Loaded Deep Foundations: Analysis and Performance, ASTM STP 835 (3-20).
American Society for Testing and Materials.

Poulos, H. G. (1971). Behavior of Laterally Loaded Piles: I-Single Piles. J. of the Soil
Mechanics and Foundations Div., Proc. of the ASCE, Vol. 97, No. SM 5, May, 711-
731.

Poulos, H. G., & Davis, E. H. (1980). Pile Foundation Analysis and Design. NY:
John Wiley & Sons.

Prakash, S., & Sharma, H. D. (1990). Pile Foundations in Engineering Practice. NY:
John Wiley & Sons.

Randolph, M. F. (1981). The Response of Flexible Piles to Lateral Loading.

Geotechnique, Vol. 31, No. 2, 247-259.

Reese, L. C., & Allen, J. D. (1977). Drilled Shaft Manual: Volume II-Structural
Analysis and Design For Lateral Loading. Washington D.C.: Offices of Research and
Development, U.S. Department of Transportation.

Reese, L. C., & Matlock, H. (1960). Generalized Solutions for Laterally Loaded Piles.
J. of the Soil Mechanics and Foundations Div., Proc. of the ASCE, Vol. 86, SM 5,
October, 63-91.

110
Reese, L. C., & Van Impe, W. F. (2001). Single Piles and Pile Groups Under Lateral
Loading. Natherland: A.A.Balkema.

Reese, L. C., & Welch, R. C. (1975). Lateral Loading of Deep Foundations in Stiff
Clay. J. of Geotechnical Engineering Div., Proc. of The ASCE, Vol. 101, No. GT7,
July, 633-649.

Reese, L. C., Wang, S. T., Isenhower, W. M., & Arrllaga, J. A. (2000). LPILE Plus
4.0 for Window: Technical Manual. Texas: Ensoft.

Reese, L. C., Wang, S. T., Isenhower, W. M., & Arrllaga, J. A. (2000). LPILE Plus
4.0 for Windows: User’s Manual. Texas: Ensoft.

Seed, H. B., & Idriss, I. M. (1970). Soil Moduli and Damping Factors for Dynamic
Response Analyses. Research Report No. EERC 70-10, Earthquake Engineering
Research Center. Berkeley, CA: University of California at Berkeley.

Terzaghi, K. (1955). Evaluation of Coefficients of Subgrade Reaction. Geotechnique,

Vol. 5, No. 4, 297-326.

Vesic, A. B. (1961). Bending of Beam Resting on Isotropic Elastic Solid. J. of the

Engineering Mechanics Div., Proc. of the ASCE, Vol. 87, No. EM 2, April, 35-53.

Yegian, M., & Wright, S. G. (1973). Lateral Soil Resistance – Displacement

Relationships for Pile Foundations in Soft Clays. The Fifth Annual Offshore
Technology Conference, Houston, Texas, Vol. II, Paper No. 1893, 663-676.

111
APPENDIX A SURVEY RESULTS

Table A.1 LRFD Survey Questionnaires

1) How many people are there in your respective geotechnical and structural engineering
divisions? Are the divisions combined? Is the size of your staff sufficient?
2) What kind of projects does your division do?
† Bridge
† Slope Stability
† Road/Highway Construction
† Other
3) What are the predominant soil and rock types in your state?
† Clay
† Sand
† Clay shale
† Sandstone
† Igneous/Metamorphic rock
† Loess
4) What codes do you currently use in designing geotechnical structures?
† ASD
† LFD
† LRFD
5) What codes do you currently use in designing substructures or superstructures?
† ASD
† LFD
† LRFD
6) Do all projects require subsurface investigation?
† Yes, what type? _____________________________________
† No
7) What lab tests are normally required for a geotechnical project?
† Swell/Consolidation
† Unconfined Compression Test
† Triaxial
† Direct or Simple Shear
† Standard and Modified Proctor
† Atterberg Limits
† Grain Size Analysis
† Other ______________________________________
Is your facility sufficient? If not why?
8) What types of insitu tests are required for geotechnical projects?
† Conventional static load test. If yes, what type?
† Osterberg Load Cell Test
† Statnamic Load Test
† Dynamic Load Test; Type of test ______________________________
† Others ___________________________________________________
Is your facility sufficient? If not why?
9) Do you contract laboratory or insitu testing to local firms?
† Yes ____________% contracted?
† No
10) What are your standard procedures for subsurface investigation?

A-1
11) What procedures do you follow to decide the types of tests required for a project?
12) Do you contract any design projects to local firms?
† Yes ____________% contracted?
† No
13) What is the supporting relationship between the structural and geotechnical divisions?
14) Do you use any “magic formulas” used in geotechnical design? For instance, in
Colorado, the end-bearing pressure for drilled piers is usually estimated as N/2 ksf.
15) Has your DOT begun to implement or perform research on LRFD design concept?
† Yes
† No
† Structural
† Geotechnical
16) If you have not implemented LRFD in geotechnical design, what is your implementation
schedule?
17) If you have not implemented LRFD in structural design, what is your implementation
schedule?
18) In your opinion, is the LRFD method a better method for geotechnical design? If yes,
why? If no, why?
19) If LRFD is used for geotechnical design, are your load and resistance factors state
specific? If not, do you have plans to generate state specific factors?
20) If LRFD is used in geotechnical design at your DOT, then how long has it been in place?
21) If LRFD is used in geotechnical design, do you use it for all projects?
22) The main objective of our research is to develop a geotechnical database required for
evaluation of resistance factors for eventual implementation of LRFD in geotechnical
engineering. What do you think would help us in achieving this research objective?
23) If you have implemented LRFD in geotechnical design, have you experienced resistance
from the geotechnical consulting industry? If yes, how did you address the issue?

A-2
Table A.2 Survey results on status of implementation of geotechnical LRFD
Primary
ID State Department Respondent Title
1 North Dakota Geotechnical Section Jon Ketterling Geotechnical Engineer
2 Oregon Geo/Hydro Tim Potter HQ Unit Manager
3 Georgia Bridge Design Paul V. Liles Jr. State Bridge Engineer
4 Ohio Geotechnical Engineering Steven Sommers Geotechnical Program Coordinator
5 Oklahoma Bridge Division Greg Allen Assistant Bridge Engineer
6 Utah Geotechnical Division Darin Sjoblom Geotechnical Engineer
7 Indiana Geotechnical Section Steve Morris Geotechnical Engineering Group Leader
South
8 Carolina Bridge Design, Geotech Section Tim Adams for Jeff Sizemore Former Bridge Design Geotechnical Engineer
9 Pennsylvania Transportation Scott Christie Chief Bridge Engineer
10 Texas Bridge Division Mark McClelland Geotechnical Branch Manager
11 Minnesota Bridge Office David Dahlberg Bridge Design Unit Leader
12 Wyoming Geology Program Mark Falk Project Geologist
13 Connecticut Soils and Foundations Leo Fontaine Trans. Principal Engineer
Geotechnical Section/Bridge Geotechnial Engineer, Structural Special Assignments
14 Missouri Design Unit Kevin W McLain & Bryan Harnagel Engr.
Bridge Design & Pvmt & Geotech
15 Louisiana Design Kim Martindale, Kelly Kemp Pvmt & Geotech Des. Engr Adminstrator, Bridge Design
South
16 Dakota Geotechnical Engineering Activity Kevin Griese Geotechnical Engineer
17 Mississippi Materials Division, Bridge Division James Williams, Mitch Carr Materials, Bridge Engineer
18 Arkansas Bridge, Geotechnical Stewart Linz, Johnathan Annable Staff Bridge Design Engr, Staff Geotech Engr
19 Tennessee Structures Division Edward P Wasserman Civil Engineering Director Structures Div.
20 Alabama Transportation Fred Conway Bridge Engineer
21 Maryland Geotechnical Explorations Division Mark Wolcott Chief
22 Nevada Transportation Parviz Noori Assistant Materials Engineer
23 Hawaii Transportation Herbert Chu Engineer V
24 Rhode Island Bridge Design Anthony Palombo Sr. Civil Engineer
25 N.Y. Transportation D. Walton Assoc. Soils Engineer
26 N.J. Transportation Harry A. Capers Manager, Structural Engineering
27 Kansas Bridge/Geology Loren Risch Bridge Design Engineer
28 VT Transportation Chris Benda Soils and Foundation Engineer

A-3
# people in Geotech div Field Lab Drill
geotech independent? Size of geotech staff enough? Engineers Geologists Techs Techs Crew Drafters
6 yes no 3 0 0 0 3 0
75 yes no 24 19 0 0 6 4
36 yes no 6 1 2 2 12 1
21 yes no 8 1 0 3 7 2
0 yes no 3 0 3 5 8 0
10 yes no 4 1 1 1 3 0
21 yes no 9 4 0 3 4 1
19 no no 8 1 0 6 4 0
20 no yes 20 0 0 0 0 0
125 no n/a 5 0 0 0 0 0
25 yes yes 6 3 0 1 12 0
26 yes yes 1 14 0 2 8 1
20 yes yes 7 0 0 0 2 0
47 no no 6 2 2 2 21 0
32 no not answered 16 0 1 8 6 1
12 yes yes 4 1 0 2 5 0
20 yes yes 6 2 0 2 10 0
16 yes yes 3 1 1 1 10 0
28 yes no 6 7 0 3 10 2
60 yes no 6 1 18 2 16 1
42 yes no 11 0 4 7 20 0
20 yes no 7 0 0 2 3 0
8 yes no 3 0 2 3 0 0
2 no no 2 0 0 0 0 0
138 yes yes 55 12 21 17 30 3
22 no no 8 4 0 0 11 0
41 yes no 6 12 4 4 10 3
12 yes no 3 1 0 2 6 0

A-4
 Soil and Rock Types Major geotech problems in the state
clay settlement, low bearing capacity, frost heave, landslides
clay, sand, clay shale, sandstone, igneous/metamorphic rock, peat earthquake, ladslides, rockfall
clay, sand, clay shale, sandstone, igneous/metamorphic rock landslides, karst topography, river scour
soft, weak pavement subgrades- landslide- mines (surface and underground)- embankment
clay, clay shale, sandstone, a-4, a-6, glacial till support/settlement- water bearing sandstone- weatherable shales, mudstones and siltstones
clay, sand, clay shale, sandstone, igneous/metamorphic rock, loess, limestone blank
settlement, potentially liquefiable soils, landslide, collapsible/expansive soils, rockfall, highly
clay, sand, clay shale, sandstone, igneous/metamorphic rock, lake bottom sediments organic soils
clay, sand, clay shale, limestone, glacial till soft and wet subgrade, peat and marl deposits, landslides
setlement and liquefaction of soft/loose sedimentary deposits in the coastal plan, adequate
pile penetration in limestone areas due to lateral loads, unsupported length and scour, rock
clay, sand, clay shale, igneous/metamorphic rock, limestone, marl socket penetrat. for drilled shafts, length of socket vs. rock quality
clay, sand, clay shale, sandstone, limestone limestone, solution cavities
clay, sand, clay shale, limestone variable materials, construction errors
clay, sand , clay shale, sandstone, igneous/metamorphic rock, loess, peat deposits,
till, carbonate deposits peat deposits (settlement/slope stability), high groundwater
Clay, Clay Shale, igneous/metamorphic rock swelling soils, slope stability, collapsing soils, rockfall
west & east uplands are glacial tills. central valley has deposits of sand & silt over
varved clay deposits. Bedrock types are sedimentary (arkose) & igneous (basalt) The upper half to 2/3's of the central valley have significant deposits of compressible soils
rock in the central valley, & metamorphic (schist and gneiss) in the uplands (varved clays). The New Haven harbor area has significant deposits of organic silt.
clay, sand, clay shale, loess, limestone, dolomite landslides, settlement at bridge ends, karst topography
clay, sand, loess soft soils, lack of bearing layers, variability of profiles, setup issues
Clay shale, sandstone, igneous/metamorphic rock, glacial till pierre shale for highway subgrades
clay, sand, loess high volume change clays, soft soils, landslides
slope failues in clay and weathered shale slopes, subgrade stability for grass root (lowfill)
clay, sand, sandstone, shale, limestone, dolomite widening projects
sinkholes, colluvium, high plasticity clays, water sensitive silts, poor quality control during
clay, sand, sandstone, limestone construction
clay, sand, clay shale, sandstone, igneous/metamorphic rock, loess blank
clya, sand, clay shale, sandstone, igneous/metamorphic rock sink holes, landslide, slough
clay, sand, igneous/metamorphic rock, gravel and cobbles boulders/cobbles, kaliche, expansive soils, hydrocollapsible soils
clay, sand, and basalt rockfall, shallow embankment slipage
no major geotechnical problems, challenging site conditions: high ground water tablesoft
Clay shale, sandstone, igneous/metamorphic rock, glacial deposits with sand, silt, organic soils posing negative skin friction for piles and settlement for embankments.
gravel, cobble,s and boulders. Liquefaction may be a problem.
galcial lake beds with deepsilt and clay deposits with organics, highly variable rock elevations
within a short distance, boulders obstructing deep foundation construction, staged
clay, sand , clay shale, sandstone, igneous/metamorphic rock, and glacial soils. construction required for structure replacement.
clay, sand, clay shale, snadstone,Ingeous/Metamorphic rock, limestone. settlement, rock slope (cut), large driven piles.
clay, sand, clay shale, sandstone, loess swelling soils, collapsing solils, very soft alluvial soils.
Soil and Rock Types slope instability (slides)

A-5
 Tests for subsurface exploration Lab tests req'd for geotech project
some swell/consol, unconfined, triaxial, some direct or simple shear, standard and modified proctor, atterberg
spt, shelby tubes limits, grain size, specific gravity, some permeability
swell/consol, unconfined, triaxial, direct or simple shear, standard and modified proctor, atterberg limits, grain size
cpt, spt, trenching, very little geophysical analysis, specific gravity, permeability
swell/consol, unconfined compression, triaxial, std. And modified proctor, atterberg limits, grain sz analysis,
spt, geophysical, trenching specific gravity, permeability
swell/consol, unconfined, triaxial, direct/simple shear, std and modified proctor, atterberg limitis, grain sz analysis,
cpt (very little), spt, geophysical (very little) permeability
unconfined, triaxial, direct/simple shear, standard/modified proctor, atterberg limitis, grain size analysis, specific
cpt, spt, geophysical gravityl, resilient modulus
swell/consol, unconfined, triaxial, direct or simple shear, standard and modified proctor, atterberg limits, grain size
CPT, SPT, Geotphysical, Trenching/test pits analysis, specific gravity, permeability
SPT Unconfined, atterberg, grain size analysis, moisture content
CPT, SPT, PDA consolidation, triaxial, atterberg, grain size, specific gravity
cpt, spt unconfined, triaxial, atterberg, grain size analysis, specific gravity, permeability
swell/consol, unconfined, triaxial, direct/simple shear, std and modified proctor, atterberg, grain size, specific grav,
Texas CPT permeability
cpt, spt, geophysical (occasionally) swell/consol, unconfined, triaxial, direct/simple shear, specific gravity
swell/consol, unconfined, some triaxial, direct/simple shear, atterberg limits, grain size analysis, some specific
cpt, spt, geophysical, trenching/test pits gravity and permeability
CPT-rarely, SPT-predominantly, Geophysical-occasionally, Trenching/test
pits-occasionally Atterberg Limits,Grain Size Analysis
cpt, spt, geophysical swell/consol, unconfined, direct or simple shear, atterberg,
cpt, spt swell/consol, unconfined, triax, direct/simple shear, standard and modifed proctor, atterberg, grain size
california retractable plug tube sampler - 2 7/8" pk rod driven with a 490 lb
hammer swell/consol unconfined compression test, direct or simple shear, atterberg limits, grain size analysis
spt, shelby tube swell/consolidation, unconfined compression, standard and modified proctor, atterberg, grain size, specific gravity
spt unconfined, triaxial, atterberg, grain size
spt, geophysical, shelby swell/consol, unconfined, direct/simple shear, standard and modified proctor, atterberg, grain size, specific gravity
swell/consol, unconfined, triaxial, direct or simple shear, standard and modified proctor, atterberg, grain size,
cpt, spt specific gravity
swell/consolidation, unconfined compression test, triaxial, direct or simple shear, standard and modified proctor,
cpt, spt, geophysical, trenching/test pits, DMT atterberg, grain size analysis, specific gravity, permeability, resistivity
spt swell/consolidation, direct or simple shear, atterberg limits, grain size analysis, specific gravity
swell/consolidatioin, unconfined compression, direct shear, compaction (standard or modified Proctor), Atterberg's
spt, trenching and test pits limits, grain size analysis, and specific gravity test.
spt, cpt, field vane shear tests, trenching/test pits and occassional
geophysical exploration. undisturbed sampling, ground water observatoin,
lab testing, liquefaction analysis when needed. consolidation, unconfined compression, triaxial, Proctor compaction, Atterberg's limit tests, grain size,
consolidation, unconfined compression, triaxial, Proctor compaction, Atterberg's limit tests, grain size, specific
SPT, geophysical exploration and undisturbed sampling. gravity, pH and resistivity on backfill for MSE walls.
spt, cpt, pmt, and undisturbed sampling. Consolidation, unconfined compression, triaxial, Atterberg's limits, Grainsize analysis, specific gravity.
swell/consolidation, unconfined compression, triaxial, direct/simple shear, Proctor (standard and modified),
spt, cpt, geophysical and continuous sampling Atterbderg's limits, grain size analysis, specific gravity, and permeability
survey, boring, inclinometers, monitoring wells, modeling spt, vane shear, Menard pressure meter test.

A-6
 Contract lab/insitu test?
Lab facility sufficient? (%) Types of insitu tests?
yes 0 not answered
yes 10 osterberg load cell, dnamic pile load test (PDA, CAPWAP)
yes 30 osterberg load cell test, vertical pile load test
osterberg load cell, vertical pile load, lateral pile load, dynamic pile load (ASTM
yes - AMRL certification 80 D4945, CAPWAP), Menard Pressuremeter
plate load test, dynamic plie load, menard pressuremeter, dilatometer, texas cone
yes penetrometer
0
not enough technicians 40 pile driving analyzer
osterberg load cell, vertical pile load, lateral pile load test, dynamic pile load, PDA,
yes 35 Statnamic, dilatometer
50 blank
n/a 50 osterberg load cell, vertical pile load test
yes 50 osterberg, vertical , lateral, dynamic (PDA), menard
yes 0 none
We do not employ a lab technician. We have a fully equipped lab, Osterberg Load Cell Test, Vertical Pile Load Test, Lateral Pile Load Test, Dynamic
but have found it more efficient to outsource these activities. 70 Pile Load Test
no - don't have own load fram for unconfined, need new resilient
modulus software and hardware 7.5 Osterberg load cell, dynamic pile load test, statnamic, SCPTU
not answered 85 osterberg, vertical pile, lateral pile, dynamic pile - PDA in house
yes 0 none
yes 10 osterberg, vertical pile, dynamic pile (PDA)
yes 0 menard pressuremeter (special projects only)
no - no triaxial equipment, too few staff, staff not formally trained in
testing 60 blank
could use more help 50 osterberg load celll, vertical pile load, dynamic pile load
no, need modern computerized testing equipment 25 plate load, osterberg load cell, lateral pile, dilatometer
yes 0 dynamic pile load - PDA
plate load test, Osterberg's load test, vertical pile load test, dynamic pile load test
no 90 (PDA)
Osterberg's load test, vertical pile load test, lateral pile load test, dynamic pile driving
no. depend heavily on contracted laboratory testing service 100 analyzer (PDA)
Osterberg's laod tests, vertical pile load test, dynamic pile driving (PDA), parallel
yes, worry about losing it for financial reasons. 0 seismic fro determining unknown pile length.
Osterberg's load test, vertical pile load test, lateral pile load test, dynamic pile driving
no. The lab personnel may not receive adequate training. 95 analyzer (PDA), Menard pressure meter test.
Osterberg's load test, vertical pile load test, Menard pressure meter test, vane shear
yes 30 test, Iowa MHS
unconfined compression, triaxial, Atterberg's limits, grrain size
analysis, specific gravity. 0 Osterberg's load test, vertical pile load test, PDA, Menard pressure meter test.

A-7
 Cantiliver Gravity MSE w/ block MSE w/ full ht panel Miscellaneous
Types of retaining walls? (%) (%) facing (%) (%) (%)
cantilever, MSE w/ block facing, MSE w/ full height panel 10 0 80 10
cantilever, gravity, mse w/block facing, mse w/ full height panel 3 5 40 50 0
cantilever, gravity, mse w/ block facing, mse w/ full height panel 20 15 0 5 5
cantilever, gravity, mse w/ block facing, mse w/ full height panel, miscellaneous 15 1 70 1 13
cantiliver, gravity, mse w/ block facing, mse w/ full height panel 0 0 0 0 0
0 0 0 0 0
cantilever, MSE w/ 5' x 6' R.C. panels 30 0 0 70 0
cantilever (temporary shoring sheet pile), MSE (temporary for staged construction), misc.
(gabion), 80 0 0 15 5
cantilever, MSE with full height panel 50 0 0 50 0
MSE with block facing, MSE w/ full height panel, misc 0 0 5 75 20
cantilever, gravity, MSE w/ block facing, MSE w/ full ht, miscellaneous 85 2 10 2 1
cantilever, gravity, MSE w/ block facing 10 30 60 0 0
Cantilever, gravity (incl. Semi-gravity cantilever, prefab modular), MSE w/ block facing,
MSE w/ full height panel ( including precase partial height panels), miscellaneous 5 35 20 30 10
cantilever, MSE w/ block facing, gabions, cribwalls 9 0 90 0 1
cantilever, MSE w/ block facing 10 90
MSE w/ block facing, MSE w/ full height panel 20 80
cantilever, gravity, MSE w/ block facing 90 5 5
cantilever, MSE w/ block facing 25 75
cantilever, gravity, MSE w/ block facing, MSE w/ full height panel, bins, cribs and tie-back
walls 30 20 10 30 10
cantilever, gravity, MSE w/ block facing
cantilever, gravity, MSE w/ block facing, MSE w/ full height panel, misc. 65 5 5 20 5
cantiliver, MSE w/ block facing 35 65
cantilever, gravity, MSE with block facings 0
cantilever, gravity, MSE w/ block facings, MSE w/ full height rigid panel, and misc. 50 15 10 15 10
cantilever, MSE w/ block facings, soldier pile and laggings, precast walls 15 0 20 0 65
cantilever (including 35% T walls), MSE walls w/ full-height rigid panels, sheet pile/soldier 15 (sheet
piles, tieback, soil nail 50 35 piles,etc)
cantilever, MSE w/ block facings, MSE w/ full-height rigid panels 10 1 3 4 0
cantilever, gravity, MSE w/ block facing, MSE w/ full-height rigid panels, Misc. 90 2 1 3 4

A-8
 Drilled Driven pile Helical Pier Shallow
Types of foundations for bridges shaft (%) (%) (%) (%)
driven pile 100

0 5 75 0 20
drilled shaft, driven pile, shallow 20 70 0 10
drilled shaft, driven pile, shallow 15 80 0 5
drilled shaft, driven pile 95 5 0 0
0 0 0 0
drilled shaft, driven pile, shallow 5 80 0 15
drilled shaft, driven pile, pile footings - depends on the location - Coastal plains,
shafts - 20%, driven pile - 80%. Piedmont, shafts - 80%, piles 15%, pile footings 5% 0 0 0 0
drilled shaft, driven pile, shallow 5 45 0 50
drilled shaft, driven pile 70 30 0 0
drilled shaft, driven pile, shallow 5 85 0 10
drilled shaft, driven pile, shallow 15 30 0 55
drilled shaft, driven pile, shallow, micropiles (5%) 10 25 0 65
drilled shaft, driven pile, shallow 10 65 0 25
drilled shaft, driven pile 20 80
drilled shaft, driven pile 20 80
drilled shaft, driven pile 5 95
drilled shaft, driven pile, shallow 2 75 23
drilled shaft, driven pile, shallow 10 50 40
drilled shaft, driven pile, shallow 35 40 25
drilled shaft, driven pile, shallow 10 50 40
drlled shaft, driven pile, shallow 50 10 40
drilled shaft, driven piles, and shallow foundation 60 20 20

drilled shaft, driven piles, shallow 20 50 0 30

drilled shaft, driven piles, shallow, micropiles 10 40 0 40
drilled shaft, driven piles, shallow 20 40 0 40
drilled shaft, driven piles, and shallow foundation 45 45 0 10
drilled shaft, driven piles, and shallow foundation 10 40 0 50

A-9
 Micropile Geosynthetics Grouting Scaling Benching Anchors Retaining Wall
Slope stabilization (%) (%) (%) (%) (%) (%) (%)
geosynthetics, benching, retaining wall 70 20 10
scaling, anchors, retaining wa 0 0 0 80 0 15 5
geosyn, bench., anchors, ret w 0 5 0 0 5 10 80
geosynth, benching, retain wall 0 15 0 0 60 0 25
geosynthetics, benching, retaining wall 0 0 0 0 0 0 0
0 0 0 0 0 0 0
retaining wall 0 0 0 0 0 0 100
geosynthetics, retaining wall 0 80 0 0 0 0 20
geosynthetics, benching, retaining wall 0 0 0 0 0 0 0
most often by removal and replacement, occasionally w. geosynthetics,
anchors, walls 0 0 0 0 0 0 0
geosynthetic, benching, other - retaining wall, flatten slopes, lower grade,
remove and replace, drainage - 75% is these methods 0 10 0 0 10 0 5
micropile, scaling, anchors, dewatering (20%), reconstruction of fill and toe
berms (50%) 10 0 0 10 0 10 0
not a significant problem 0 0 0 0 0 0 0
geosynthetics, benching, removal and replacement 0 5 0 0 95 0
geosynthetics, benching, anchors, retaining wall
geosynthetics, scaling
geosynthetics, benching 20 80
excavate and replace w/ upgraded material 0
benching, retaining wall 0 70 30
geosynthetics, benching, retaining wall 0
micropiles, geosynthetics, benching, retaining wall 2 38 20 40
scaling, retaining wall 50 50
geosynthetics, scaling, benching, retaining walls
geosynthetics, benching, anchor, retaining walls 0 20 0 0 20 30 30
geosynthetics, anchors, retaining walls, soil nail walls 0 40 0 0 0 23 35
geosynthetics,retaining walls, 2:1 slope without treatment. 0 2 0 0 0 0 18
benchingl 0 0 0 0 100 0 0
geosynthetics, benching, retaining wall 0 8 0 0 90 0 2

A-10
 Codes used in geotech Codes used in Contract design Geotech and structural
design structural design projects? (%) coordinate? Magic formulas?
ASD LFD 20 yes no
ASD, LRFD n/a 20 yes no
ASD, LFD ASD, LFD 40 yes no
ASD LFD 90 no no
LRFD LRFD yes no empirical charts for drilled shaft design foundation using TX CPT data
n/a n/a
LFD LRFD 90 yes no
case studies used to apply typical load transfer values that would be misleading
ASD, LRFD ASD, LFD yes with our local experience (SPT97 not effect. For design. Piles in marl)
LRFD LRFD 80 yes no
ASD and TxDOT
procedures ASD, LRFD varies same division specific design procedures developed by TxDOT
ASD LFD, LRFD 15-40 yes no
LFD LFD 10 yes blank
ASD, LRFD (for highway
bridge des. By ASD, LRFD
consultants) (highway bridge) 75
drilled shafts in shale - ultimate EB = qu, drilled shafts in rock ultimate EB =
ASD LFD yes 0.15qu
ASD, LFD, Su/p ratios, limiting adhesion values in drilled shafts & piles, experience based
ASD LRFD 10-15 yes - road and bridge setup predictions
ASD LFD, LRFD 10 yes no
<10 for geotechnical
ASD ASD, LFD services yes no
LFD LFD 15 yes no
LFD, LRFD
(except fdn
ASD design) 60 yes no
9000 psi end bearing on piles; min drill shaft socket = 2D in soil, 1D in rock; drill
ASD, LFD ASD 20 yes shaft FS = 3; pile load test to 2P
ASD ASD 10 yes no
asd asd 55 yes no
LRFD LRFD 80 yes no
ASD LFD 90 n/a (no geotechnical div) no
We do our best. When
structure by consultants
and geotechnical by
state the communication
ASD LRFD (not fully) 20 is much more difficult. no
ASD LRFD 90 yes no
ASD LFD yes yes n/a
ASD ASD/LFD/LRFD 30 yes no

A-11
 DOT implementing or researching LRFD? Geotech implem schedule? Structural implem schedule?
no none none
yes - both geotech & struct. by 2008 by 2005
no sept. 2007 sept. 2007
no July 2007 July 2007
yes - both geotech and structural blank blank

no, maybe in the next 2 years start

yes - structural, but not geotechnical thinking about n/a
No calibration, only AASHTO factors n/a blank
yes, geotech and structural
20% currently, full implementation
yes - structural none by 2007
sent in their implementation
yes, both geotech and structural schedule, see file see file
when the feds require it - we have
yes - structural no - geotechnical when we are forced to done a few test structures

yes, both, geotech started research with University of

Missouri-Columbia depends on results from MU all bridge structures end of 2005
selected projects only, so far only 3
structural only none projects done w/ LRFD
currently some superstructure
design, full implementation in
structural only October 1, 2007
no - Geotech no 2007
when suitable design software and
no - geotech none design aids are developed
small, can't keep up w/ workload, start in earnest Jul 1 '03, fully
yes - structural, no-geotech prefers not to retrain implem. Oct. '06
NO 2007 2007
yes - both 2007 2007
yes - structural none considering - no schedule yet
yes- geotechnical and structural implemented for two years implemented for two years
no n/a none
yes in both structure and geotechnical n/a implemented
yes, structural n/a implemented
yes, both in structural and geotechnical might be 2005 n/a
yes, structural n/a in transition

A-12
 LRFD better method?
have not researched it thoroughly but it appears to model loads and resistances better than ASD
No - uncertain load and resistance factors. For example, correlation for driven piles, with ASD is very low, and leads to
less conservative design for not much cost savings.
no - extreme unfamiliarity with the subject
Don't know enough about it yet to form an opinion. Might be helpful if over conservative designs can be reduced due to
accurate load and resistance factors. If one has to use generic factors, it seems that we come up with as using FS on
final values
I don't know

No, in order to know the actual strength of the soil need to take hundreds of samples and run hundreds of triaxial tests.
Not practical. Better to estimates on the conservative side and use ASD than try to use LRFD
prefer ASD. However, just begun to learn how to associate a resistance factor with a factor of safety. Basically with
LRFD taking the "magic" gained by years of field experience out of equation. Maybe that's a good thing….
This is a trick question. Most of LRFD was really implemented in LFD. There are the same problems in both areas. So
LRFD is not the main problem. LFD is the start of the problems
not correlated against our geotech methods or experience, unable to conclude if better or worse
yes. LRFD gives an opportunity to take credit for better investigations. The risks and unknowns can be applied to the
most appropriate areas through the load and resistance factors. LRFD can be calibrated to real test results
most of the geologists have taken the FHWA LRFD course. At this time, it does not appear that there are major
advantages to switching to LRFD in geotechnical design
different approach, at this point in time-don't think it is better or worse that working stress. From geotech perspective
much of the code based on the working stress approach. still needs time to mature and fully develop the approp.
resistance fact..
not worked with it enough to have an opinion
no - very difficult to devleop resistance factors specific to state's geology. Geotech analysis involves accuracy that is far
less than can be achieved in structural engr. Soil-structure interaction is crucial
Haven't had enough experience yet to answer
no, difficulty in detmining representative resistance factors for a vast array of soil types in Mississippi
blank
Geotech branch no opinion. Structural branch believes that LRFD essentially same as that in the Standard Spec. 2002
No. Not too much has been done in geotech in the current LRFD.
yes, because LRFD is based on realistic design parameters
not sure
no, the current resistance factors need to be state specific. It is still empirical in nature.
No, not yet, because the database of load and resistance factors has not been developed. Also because LFD is used for
substructure and superstructure design and ASD is used for foundations. Our consultants use ASD for folundations. With
LRFD, there may be a risk that engineering judgment and experience are bypassed in geotechnical designs.
Conceptually LRFD has the potential for better design, but at the current stage of development, LRFD seems more costly.
Once AASHTO introduces design specifications to address deep foundationdesigns and earth pressures for LRFD use.
no. highlight a misunderstanding of the unit of geotechnical design.
yes, it will provide more rational approach to design

A-13
 Load and Resist factors state specific? If use LRFD then how long? If implemented LRFD any resistance from consulting?
we do not have plans at this time, however if we change over to
LRFD we will need specific facotrs for ND soils. SD and MN
have similar soils and we could possibly get together with them n/a blank
not so far, but we will develop some just starting n/a
not used not used not implemented
n/a - getting all soils consultants on board is key because tey wil be main
developers of all this data and have most of the equipment used in
Don't know if we would go state specific, but I doubt we would. n/a testing
yes, used with TX CPT tests. The rest are from LRFD spec. five years - not used for bridges on the off-system no

N/A N/A don't know

State specific factors should be determined through calibration.
Don't see that I the near gurture unless a research project is It becomes easier when the owner is requesting the work. Details were
funded one year individually checked by reviewing plans for the initial projects
yes three years, all geotech projects some but only because results were more conservative
n/a n/a n/
no plans at this time n/a n/a
blank blank blank
Globally no, however, there are situations or specific designs where
Consultants have been using it on their projects there is a reluctance to use the methods/resistance factors
Not in the forseeable future. for the past 4 years. recommended by the code.
research at MU will help generate and implement state specific
factors
will require several sets of different factors for each geologic area
- marsh, prarie and upland deposits.
haven't had enough experience yet to answer n/a n/a
n/a n/a blank
blank blank blank
not determined at present. Have scheduled a workshop in the fall
of 2003 to discuss this. n/a
Will likely develop state specific facotrs and verify all designs
using ASD n/a n/a
n/a n/a n/a
blank blank blank
no. no plan to generate the state-specific resistance factors. two years no
n/a n/a n/a
tried on a couple of projects in the last two years.
On one unfamiliarity slowed the design and was
working on the state specific resistance factors where feasible. abandoned. no
no n/a n/a
n/a n/a n/a
will probably involve in a regional basis. n/a n/a

A-14
APPENDIX B FINITE ELEMENT METHOD AND CALIBRATION

B.1 Introduction

This chapter is devoted to calibration of finite element models. Each finite element
program must be calibrated for its ability to predict the behavior of a structure that it tries
to simulate. NIKE3D was selected to analyze the behavior of drilled shafts subjected to
lateral loads. Finite element mesh is generated using TrueGrid software. Griz program
will be used to visualize the results of NIKE3D analyses. Microsoft Excel will be used to
process and calculate results from NIKE3D. Process in using finite element program is
shown in Figure B-1.
Two material models will be used for representing pile and soil in NIKE3D
models. One is elastic material model for pile and soil. Another is Ramberg-Osgood (RO)
material model for soil.
In finite element analyses, calibration is the most critical. Calibration requires
mesh or model adjustment. Full-scale or model test results are also required to check the
model’s validity. Two full-scale, pier load tests were selected to calibrate the models.
One is 6-ft diameter pier tested by Dunnavant in 1986. Another is 2.5-ft diameter
concrete drilled shaft tested by Reese and Welch in 1975. Both test results were also used
in investigating Broms method and LPILE program as discussed in chapter 4 and 5,
respectively.
Comparisons between Broms, LPILE, and NIKE3D will be given in this chapter.
Advantages and limitations of each method will also be discussed.

B.2 Preprocessor: TrueGrid

Finite element methods require mesh generation software to generate mesh representing
the problems. TrueGrid is mesh generation software. TrueGrid can generate meshes
which is compatible to NIKE3D. TrueGrid will produce input files for NIKE3D.
Physical problems are converted to mesh systems. Pile dimensions and properties,
soil properties, and boundary conditions are required as input data. Loading conditions,
finite element analysis options, and load curves are necessary input data. Input files for
TrueGrid are stored in text-file format.
Once TrueGrid was completed, the output file called ‘trugrdo’ was produced.
trugrdo file contains necessary information in standard format for NIKE3D analysis.

B-1
Figure B-1 Finite Element (FE) Analysis Processes

B.3 FE Software: NIKE3D

NIKE3D will be used as a finite element program in this thesis. NIKE3D is nonlinear,
implicit, three-dimensional finite element software. It can be used in analyzing solid and
structures of many materials. Two material models are used: elastic and Ramberg-
Osgood models. Elastic material represents reinforced concrete pile and soil in the initial
analyses. Ramberg-Osgood material is used to represent the nonlinear behavior of soil.
Material parameters for elastic model are (1) Young’s modulus of elasticity (E), (2)
Poisson’s ratio (ν), and (3) mass density ( ρ). Material properties for Ramberg- Osgood

B-2
model are (1) reference shear strain (γy), (2) reference shear stress (τy), (3) stress
coefficient (α), (4) stress exponent ( r ), and (5) bulk modulus (K). These properties for
RO models are generated using RAMBO software at given type of soil and mass density
as explained in Chapter 3.
Using ‘trugrdo’ as an input file for NIKE3D, the result from the analysis is
contained in output file called ‘n3plot’.

B.4 Postprocessor: Griz

Once output from NIKE3D is obtained, software ‘Griz’ program is required to visualize
results of NIKE3D analyses contained in ‘n3plot’ files. Results of NIKE3D analysis are
presented in colored diagrams. Numerical results of NIKE3D analyses can be extracted
and saved as text files. MS Excel then is used for processing numerical results from text
files.

B.5 Calibration of NIKE3D with Available Field Test Results

Before using NIKE3D to predict the behavior of drilled shafts, it needs to be calibrated
against model or full-scale test results. Two full-scale pier load tests are used in
calibrating NIKE3D models.
Calibration is an iterative process. An initial, simple mesh is generated. Input data
are assigned to the initial mesh. NIKE3D then analyzes the problem. Results of NIKE3D
are compared with full-scale test results. If there are unacceptable differences between
NIKE3D and test results, the adjustment of the mesh is required. Improvement of mesh
and comparing with test results need to be done iteratively. The acceptable final mesh is
the one that can produce acceptable difference between NIKE3D’s and test results.
Initially pile and soil were considered elastic. Results from elastic-model run were
compared with full-scale test results. Mesh adjustment can be done if required. When the
elastic runs are successful, nonlinear analyses are performed using Ramberg-Osgood
model for soil. Again, mesh adjustment may be necessary. This process is repeated until
acceptable results from NIKE3D are obtained.
Flow chart for developing finite element model is shown in Figure B-2.
Calibration process is the process shown in the dashed block.

B.5.1 Dunnavant’s Pier Test

The test site was at the University of Houston. Dunnavant’s pier is a reinforced concrete
drilled shaft constructed in saturated stiff clay. A 6–feet diameter pier was constructed
and flooded before the test was performed. Pier was reinforced with 24 #11 bars. Average
bending stiffness was 3.92x1010 lb-ft2. Penetration length was 37.5 ft below the ground
surface. Lateral loads were applied on drilled shaft at 11 inches above the ground surface.
The lateral displacement of the pier was measured by the inclinometer installed along the
center of the pier. Water table was found at depth of 6 ft below the ground surface.
The soils at test site were natural soils containing two strata. The first stratum was
24-ft deep from ground surface called Beaumont clay. The upper 10-ft layer of the first

B-3
stratum showed some secondary structure. The second stratum underlying the first
stratum was a very thick layer of Montgomery clay.
Unconsolidated undrained (UU) and consolidated, isotropic, undrained (CIU)
triaxial, vane shear (VST), and cone penetration (CPT) tests were used to determine the
strength of soil. UU tests were conducted on soil samples up to the depth of 18 ft. Vane
shear and CIU tests were performed in upper 5 ft. CPT were conducted along the whole
length of the pile. Undrained shear strength data are shown in Figure B-3. From Figure
B-3, undrained shear strength were found to increase with depth. Undrained shear
strength from VST and CPT were greater than those from UU and CIU tests.
Young’s modulus of elasticity in laboratory was defined as the secant modus at 20
% of the failure stress difference (E20). Young’s modulus of elasticity data provided by
Dunnavant were UU triaxial and Cross-Hole tests (CHT) in the upper 18 ft. In addition,
elasticity data provided by Marhar and O’Neill (1983) were included. Marhar and
O’Neill used UU triaxial, presuremeters (PMT), and Cross-Hole tests. Elasticity from UU
and PMT were reported up to depth of 40 ft below the ground surface. For CHT, data
were presented up to depth of 60 ft below the ground surface. In general, elasticity
increased with depth as shown in Figure B-4. From Figure B-4, CHT gave the highest
values of elasticity compared to Young’s modulus of elasticity from other tests. Elasticity
from UU tests gave the lowest values. Unit weight data were also provided as shown in
Figure B-5. Piles will be modeled using elastic material. Soil will be modeled using either
elastic or Ramberg-Osgood materials.
Ground-line deflection data from Dunnavant’s thesis is shown in Figure B-6. The
curve can be divided into two portions: linear and nonlinear portions. Linear portion of
the curve begins from origin up to the load of 180 kips. Loads greater than 180 kips show
non-linearity of pile behavior. Dunnavant reported that this could be caused by tensile
crack in the pile probably occurred at the depth about 20 ft below the ground surface.
Load was applied up to 425 kips until it stopped because of the load frame buckling.
Deformed shapes of the pier at different loads are shown in Figure B-7. Guided
by the deformed shapes of the pier, Dunnavant revealed: (Dunnavant 1986)
1. At large pile-head displacements the tip of the pile did not undergo large
displacements (did not “kick out”).
2. Based upon inclinometer measurements, pile curvatures at the 300- and 400-kip
loads were large enough to produce compressive concrete strains up to 0.0013 and
0.0020, respectively, assuming flexure about the centroidal axis. Significant
tensile cracking and nonlinear compressive stress-strain behavior probably
occurred at about 20-ft depth below ground surface. This hypothesis is supported
by the observation that the slope of the unloading curve is lower than the slope of
the initial loading curve.

Dunnavant noted that there was translation of deflection shapes at 400 kips
relative to shape at 300 kips. This translation can occur only in restrained head pile under
lateral load. It has been observed that extensive concrete cracking and crushing probably
occurred at 400 –kips load at about 20-ft depth below the ground surface (Dunnavant
1986). In Dunnavant’ s thesis, he believed that the pile at this depth acted as a partial
hinge. Consequently, the capacity of the pile to transmit moment on that section was

B-4
reduced. Then, this hinge condition would resist pile tip to kick out as shown in Figure B-
7.
From Dunnavant’ s results, it is concluded that pile failure occurred at the load of
250 kips.

B-5
Figure B-2 The Development of Finite Element (FE) Model

B-6
FigureB-3 Undrained Shear Strength from Dunnavant’s Test Site (Dunnavant 1986)

B-7
Figure B-4 Young’s Modulus of Elasticity by Dunnavant (Dunnavant 1986)

B-8
Figure B-5 Total and Dry Unit Weight by Dunnavant (Dunnavant 1986)

B-9
 450

400

350

300
load, kips

250

200

150

100

50

0
0 0.5 1 1.5 2

y0, in

Figure B-6 Ground-line Deflections from Dunnavant Test

B-10
Figure B-7 Deformed Shapes of Dunnavant’s Pier at Different Loads (Dunnavants 1986)

B.5.1.1 Elastic Soil Model

In these models, both soil and pile were considered elastic. Elastic properties for drilled
shaft can be calculated using Dunnavant data. Modulus of elasticity of reinforced
concrete pile (Ep) was 6.162x108 psf. Poisson’s ratio (νp) of 0.20 for pile was used. Pile
mass density (ρp) was 4.50 slug/ft2.
Two two-layered and one three-layered-soil meshes were investigated. The first
two-layered-soil mesh of 18-ft plus 19.5-ft layers was selected. Young’s modulus of
elasticity for each layer was calculated. Unit weight and mass density were computed for
each layer. Elasticity by UU tests, PMT, CHT, average of UU and PMT, and average
between UU and CHT were separately calculated and input into NIKE3D. Each case was
run separately by NIKE3D and its result is shown in Figure 6-8. From Figure 6-8, it is
clear that the elasticity from CHT gives the best prediction of ground-line deflections in
the initial part of the curve. Result using the elasticity by UU tests overestimates the

B-11
ground-line deflections of the pier. Result using the elasticity by PMT gives ground-line
deflections in between results using elasticity by UU and CHT.
The second two-layered-soil mesh consists of 24-ft plus 13.5-ft layers. In this
model, elasticity by UU, PMT, and CHT were used. Figure B-9 shows ground-line
deflections from these models. From Figure B-8 and B-9, it shows that the two-layered-
soil mesh with 18-ft plus 19.5-ft gives the better results than 24-and-13.5-ft mesh in
general.

B.5.1.2 Ramberg-Osgood (RO) Soil Model

For Ramberg-Osgood soil models, soils were represented using RO material but piles
were modeled using elastic material. Using results from elastic soil models as guideline,
the primary results show that 18-and-19.5-ft mesh is better. It is believed that the first 6-
ft-deep soil layer would affect more on ground-line deflections of the pile. In RO soil
models, Two-layered and three-layered soil systems were investigated. Two-layered-soil
mesh consists of 24-ft plus 13.5-ft layers. Three-layered system consists of 6-ft, 12-ft,
and 19.5-ft layers.
Modulus of elasticity from CHT was used for determining RO parameters. Shear
(G) and bulk (K) moduli were calculated from E20 for a given Poisson’s ratio. Ratio of
shear modulus to maximum shear modulus (Gmax) of 90 % was used. Other RO
parameters can be determined using Gmax.
Figure 6-10 shows ground-line deflections using 2- and 3-layered-soil meshes.
For 2-layered mesh, ground-line deflections agree well with test results up to the load of
180 kips. For 3-layered mesh, the results from NIKE3D show good agreement with both
2-layered systems and test results up to the load of 50 kips. But 3-layered system gives
better results with loads greater than 230-kips load. The two-layered-soil mesh is
appropriate for linear portion of load-deflection curve. On the other hand, the three-
layered mesh gives the good agreement in general. Based on RO model results, division
of soil into appropriate number of layers is critical for results of FE models.
Deformed shapes of the pile at different loads are shown in Figure B-11. These
show behavior of relatively stiff pile.

B-12
 450

400

350

300
load, kips

250

200 Dunnavant data

E20 from UU and PMT

150
E20 from UU tests

100 E20 from PMT

E from Cross-Hole
50
E from UU and CH tests

0
0 0.5 1 1.5 2
y0, in

Figure B-8 NIKE3D Models of Ground-line Deflections using Different Secant Moduli
with 2-layered Elastic Soil (18 and 19.5-ft layers)

B-13
 450

400

350

300
load, kips

250

200

150 Dunnavant data

E20 from UU tests

100
E from CH tests

50 E20 from PMT

0
0 0.5 1 1.5 2 2.5 3 3.5
y0, in

Figure B-9 NIKE3D Models of Ground-line Deflections using Different Secant Elasticity
with 2-layered Elastic Soil (24 and 13.5-ft layers)

B-14
 450

400

350

300

250
load, kips

Dunnavant data
200
3-layered RO model

2-layered RO model
150

100

50

0
0 0.5 1 1.5 2
y0, in

Figure B-10 NIKE3D RO Models of Ground-line Deflections with 2- and 3-layered Soil
Meshes

B-15
 5.00

0.00

-5.00

-10.00

-15.00
depth, ft

-20.00
P = 139 kips

P = 300 kips
-25.00
P = 400 kips

-30.00

-35.00

-40.00
-0.200 0.000 0.200 0.400 0.600 0.800 1.000

deflection, in

Figure B-11 NIKE3D RO Models of Deformed Shapes of the Pile at Different Loads
using Two-layered Soil Mesh

B.5.1.3 Evaluation of Soil Young’s Modulus

In this section, Young’s modulus of soil from will be evaluated using three different
ways. Elasticity was defined as the secant modulus determined at 20% of the failure
stress difference. To find out the ratios between elasticity from different methods, soil
profile was divided into three different meshes. These meshes are same as those used in
FE models. The result is shown in Table B-1. Ratios were calculated for each sub-layer.
Then the average from each sub-layer is used.
From Table B-1, average ratio of elasticity from PMT and CHT to UU tests is 2.9
and 15.5, respectively. The average ratio of CHT to PMT is 5.5. The maximum and
minimum ratios of PMT to UU tests are 4.7 and 1.9, respectively. Maximum and
minimum ratios for CHT to UU tests show the values of 21.5 and 10.6, respectively. The
values of 7.5 and 3.7 were found for CHT to PMT.

B-16
 Ratios of elasticity to undrained shear strength also were calculated as shown in
Table B-2. Only results from UU tests were computed. The average, minimum, and
maximum ratios of E/cu are found to be 152, 133, and 185, respectively.

Table B-1 Relationships between Elasticity from Different Methods (Dunnavant data)

Elasticity
Layered Depth by EPMT/EUU ECH/EUU ECH/EPMT
Soil UU tests PMT Cross-Hole
Mesh Tests (CHT)
(ft) (ksf) (ksf) (ksf)
2-layered mesh 0 to 18 338.4 738.0 4104.0 2.2 12.1 5.6
of 18-ft plus 22-ft 18 to 40 432.0 1386.0 7311.0 3.2 16.9 5.3
2-layered mesh 0 to 24 344.6 983.3 3654.0 2.9 10.6 3.7
of 24-ft plus 16-ft 24 to 40 374.4 1752.0 8064.0 4.7 21.5 4.6
3-layered mesh 0 to 6 217.5 500.0 3774.9 2.3 17.4 7.5
of 6-ft, 12-ft, and 6 to 18 335.0 635.0 4417.4 1.9 13.2 7.0
22-ft 18 to 40 440.0 1512.7 7386.5 3.4 16.8 4.9
Average 2.9 15.5 5.5
Max 4.7 21.5 7.5
Min 1.9 10.6 3.7

Table B-2 Ratios of Elasticity to cu by UU Tests (Dunnavant data)

Depth Elasticity cu E/cu

(ft) (psf) (psf)

0 200000 1080 185

5 250000 1660 151
10 310000 2230 139
15 365000 2740 133
Average 152
Max 185
Min 133

B.5.2 Reese and Welch Test Results

The test site was in Houston. A 2.5-ft diameter drilled shaft was constructed in stiff clay.
The shaft was reinforced with 20 #14 bars. Average bending stiffness was 1.39x109 lb-ft2.
Penetration length was 42 ft. Loads were applied at a height of 3 inches above the ground
surface. Water table was found at 18 ft below the ground surface.
Soils at this site were called Beaumont clays. UU triaxial tests were conducted on
undisturbed samples. Reese and Welch reported the most significant layer of soil was the
upper 20 ft. The average undrained shear strength in this layer is 1.1 tsf. Between 20 and
33 ft, the average undrained shear strength was 1.0 tsf. From the depth of 33 to 42 ft, the

B-17
average undrained shear strength was 1.65 tsf. Undrained shear strength is sown in Figure
B-12. Undrained shear strength data was scattered but tend to increase with depth.
Soil Young’s modulus of elasticity was defined as the secant modulus at 50 % of
the maximum principal stress difference (E50). UU triaxial tests were used to determine
soil Young’s modulus of elasticity on both vertical and horizontal samples. No significant
difference in elasticity was found between vertical and horizontal samples. Elasticity data
fell in a wide range from 2000 to 5000 psi between the first 20 ft. Below 20 ft, few data
were shown and elasticity tended to decrease with depth. Modulus of elasticity is shown
in Figure B-13.
Unit weight and water content data were provided as shown in Table B-3.
Piles will be represented with elastic material model. Both elastic and Ramberg-
Osgood models were used for soil. Ground-line deflections of pile are shown in Figure B-
14.

B-18
Figure B-12 Undrained Shear Strength from Reese and Welch’s Test Site (Reese and
Welch 1975)

B-19
Figure B-13 Modulus of Elasticity from Reese and Welch (Reese and Welch 1975)

Table B-3 Total Unit Weight and Water Content from Reese and Welch

Depth Total unit weight Water content

(ft) (pcf) (%)
0 123.6 18
1.3 123.6 18
3.4 119.7 22
20 121.7 20
42 126.8 15

Source: Adapted from Reese and Van Impe (2001)

B-20
 120

100

80
load, kips

60

40

20

0
0.000 0.200 0.400 0.600 0.800 1.000

y0, in

Figure B-14 Ground-line Deflections from Reese and Welch Test

B.5.2.1 Elastic Soils

As explained in Dunnavant test, soils and piles were considered elastic. From RO results
in Dunnavant ’s pier, results showed that the 3-layered mesh gave the better results in
general. Hence, the 3-layered soil mesh will be used as an initial mesh. Three-layered soil
mesh consists of 5-ft, 15-ft, and 22-ft layers. In Dunnavant results, the elasticity from UU
tests (EUU) gives the values much smaller than the elasticity from CHT (ECH). So, the
upper bound of EUU was used. Elasticity of 7488 ksf was assigned to the first 5-ft layer.
The elasticity of the middle layer was 7200 ksf. The elasticity for bottom layer was 5458
ksf. As mentioned in previous section, EUU is much smaller than ECH. It is assumed that
ECH is 10 times greater than EUU. Then, ECH or E10UU was assigned to all three layers.
Figure 6-15 shows ground-line deflections using both the elasticity from UU tests (EUU)
and the elasticity with 10 times greater than those from UU tests (E10UU).

B-21
 From Figure B-15, it shows that ground-line deflections from NIKE3D results
using E10UU agree well with test results up to 45-kip load. Then, this elasticity will be
used in RO soil model.

B.5.2.2 Ramberg-Osgood (RO) Soils

The nonlinear RO model was used only to represent soils. Elastic material represented for
piles. E10UU was used for calculating RO parameters. These RO parameters were assigned
to the three-layered mesh. Results are shown in Figure B-16. Results from elastic run are
also shown in Figure B-16. It is found that both elastic and RO models gave similar
results.
After reviewing undrained shear strength and elasticity data, it showed some
controversy between these two data. Undrained shear strength tends to increase with
depth. But elasticity shows decrease with depth. This relation is in contrary to the
previous studies that show the undrained shear strength is proportional to elasticity of the
soil. Then, the ratios of elasticity to cu are used instead of the elasticity from UU tests.
The ratio of elasticity to cu for upper 20-ft layer was found to be 200. This ratio will be
used for all layers.
The new mesh is composed of 20-ft, 13-ft, and 9-ft layers. Elasticity of the first 20
ft was used as a basis for elasticity and bulk moduli in other layers. The ratio of E50 to cu
for the first 20 ft of 200 was used. This ratio was also used for determining E50 of middle
and bottom layers. Based on elasticity, shear modulus, maximum shear modulus, and
bulk modulus can be computed. Shear modulus (G) is assumed to be 85% of maximum
shear modulus.

B-22
 120

100

80
load, kips

60

Reese & Welch data

40 Elastic and EUU

Elastic and E10UU

20

0
0.000 0.200 0.400 0.600 0.800 1.000

y0, in

Figure B-15 NIKE3D Elastic Models of Ground-line Deflections for Different Secant
Elasticity with 3-layered Soil (5-ft, 15-ft, and 22-ft layers)

B-23
 140

120

100

80
load, kips

60 Reese & Welch data

Elastic soil & E10UU

40
RO & E10UU

20

0
0.000 0.200 0.400 0.600 0.800 1.000

y0, in

Figure B-16 NIKE3D Elastic and RO Models of Ground-line Deflections with 3-layered
Soil (5-ft, 15-ft, and 22-ft layers) using 10 times Elasticity from UU Tests

The average cu for the 20-ft layer is 1.1 tsf. cu of 1.0 and 1.65 tsf were used in
middle and bottom layers, respectively. Results from RO analysis for the three-layered
mesh are shown in Figure B-17 (case I). The results show potential to give good
agreement with test results.
The average values of cu were used as in Figure B-17. It required stiffer soils to
improve the agreement of the results. It is worth studying effects of soil stiffness or
elasticity on load-deflection curve. Two cases were studied. First is to increase stiffness
of middle layer (case II). Second is to increase stiffness of the first layer (case III). For
stronger middle layer, cu of 1.4 tsf was used. For stronger first layer, cu of 1.3 tsf was
used. These two values of cu are still in the range between minimum and maximum
values. Figure B-18 shows the results from three cases.
From Figure B-18, it shows that top layer has more effect than the middle layer. It
is also clear that the agreement between RO and test results can be improved. The

B-24
calibration process was continued to improve the results. Two additional cases were
performed. The first case is to increase soil stiffness for all layers (case IV). Upper bound
of cu from each layer was used in this case. The second case is same as the first case but
using stiffer soil in the top layer (case V). Results from case I, case IV, and case V are
plotted in Figure B-19. The results from case IV and case V are similar. It shows that
both cases improve the results.
Using case V’s results, deformed shapes of the pile can be plotted in Figure 6-20.
These show the behavior of relatively long pile.

B-25
 120

100

80
load, kips

60

40 Reese & Welch data

Ramberg-Osgood soil

20

0
0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600

y0, in

Figure B-17 NIKE3D RO Model of Ground-line Deflections with 3-layered Soil (20-ft,
13-ft, and 9-ft layers) using E50/cu = 200 (case I)

B-26
 120

100

80
load, kips

60

Reese & Welch data

40 Average soil stiffness

(case I)
Stiffer middle layer soil
(case II)
Stiffer upper layer soil
20
(case III)

0
0.000 0.500 1.000 1.500 2.000

y0, in

Figure B-18 Effect of Soil Stiffness (Modulus) on Load-Deflection Curve

B-27
 120

100

80
load, kips

60

40 Reese & Welch data

RO & EUU (case I)

20 RO & upper bound EUU (case IV)

RO & stronger upper layer (case V)

0
0.000 0.500 1.000 1.500 2.000

y0, in

Figure B-19 NIKE3D RO Models of Ground-line Deflections for Different Soil Stiffness

B-28
 5

-5

-10

P = 25.2 kips
-15
depth, ft

P = 62.4 kips
-20 P = 100 kips

-25

-30

-35

-40

-45
-0.20 0.30 0.80 1.30 1.80
deflection, in

Figure B-20 NIKE3D RO Model of Deformed Shapes of the Pile at Different Loads (case
V)

B.6 Comparison of Broms, LPILE, and Finite Element

Analyses

In this section, results from Broms, LPILE, and NIKE3D will be compared. Both test
results from Dunnavant and Reese & Welch were compared. Results from Broms and
LPILE were already presented in Chapter 4 and 5, respectively. In this section, only
selected results from Broms and LPILE are presented.
For Broms method, results using back-figured kh were selected in both Dunnavant
and Reese & Welch cases.
For LPILE results, the 2-layered mesh of 18-ft and 19.5-ft was chosen for
Dunnavant case. The two-layered soil mesh consisted of 20-ft and 22-ft layers were
selected for Reese and Welch case.

B-29
 In NIKE3D, the three-layered RO model of 5-ft, 13-ft, and 19.5-ft layers was
chosen for Dunnavant comparison. Three-layered RO model consisted of 20-ft, 13-ft, and
9-ft layers with stronger upper soil layer was used in comparison for Reese and Welch
case.
The results of comparison are shown in Figure B-21 and B-22 for Dunnavant and
Reese & Welch data, respectively. Comparisons of ground-line deflections between
Broms method, LPILE, and NIKE3D for Dunnavant and Reese & Welch cases are shown
in Table B-4 and B-5, respectively.
Broms method using back-figured kh shows good agreement at loads up to 180
and 43 kips for Dunnavant and Reese & Welch results, respectively.
NIKE3D gives as good results as LPILE does in Dunnavant test. But LPILE
shows better agreement with test data than NIKE3D does in Reese and Welch test. This
could be because of Reese and Welch test is part of database used in developing LPILE.

B.7 Pros and Cons of Broms, LPILE, and NIKE3D

Advantages and limitations of each method will be discussed. Each method has its
advantages and disadvantages. Table B-6 shows pros and cons of three methods.
Item No.1 “Material represented soil mass” refers to models used by each method.
For example, linear soil in Broms method means soil is modeled with single linear spring.
In LPILE, soil is represented with unconnected series of nonlinear springs. Soil is
replaced by Ramberg-Osgood nonlinear material model in NIKE3D.
Item No.2 “Solving procedure” refers to procedures used in solving laterally
loaded pile problems. In Broms method, solution procedure is easy and simple. Solution
procedure for LPILE is a bit complicate than Broms. But it is still easy to get the results.
More complicate in solution procedure is required in NIKE3D. Mesh generation and
material selection make the problem more complicate.
In item No.3, soils are modeled using discrete linear and nonlinear springs for
Broms and LPLIE, respectively. These methods consider soils as discrete materials. In
NIKE3D, soils are represented by continuum mass.
In item No.4, soil is assumed to be uniform in Broms method. For LPILE and
NIKE3D, soil can be divided into a number of layers.
Item No.5 relates to soil conditions in the problem domain. In Broms method,
clay is saturated and overconsolidated. LPILE was developed using some test results. It is
clear that soil test sites should have similar properties as the soil used in developing
LPILE. NIKE3D can be applied to more general sites of any kind of soils.
In item No. 6, Broms method can be used to predict only the linear behavior of
load-deflection relationships. LPILE and NIKE3D are able to predict the load-deflection
relationships at any load.
Item No.7 considers the capacity of methods in predicting the deformed shape of
the pile. Broms method is not able to predict the deformed shapes of the piles but both
LPILE and NIKE3D are.

B-30
 450

400

350

300
load, kips

250

200 Dunnavant data

150 NIKE3D RO model

100 LPILE

50 Broms using backfigured kh

0
0 0.5 1 1.5 2

y0, in

Figure B-21 Comparisons between Broms, LPILE, and NIKE3D for Dunnavant Case

B-31
 120

100

80
load, kips

60

Reese & Welch data

40
NIKE3D RO model

LPILE
20

Broms using backfigured kh

0
0.0 0.2 0.4 0.6 0.8 1.0

y0, in

Figure B-22 Comparisons between Broms, LPILE, and NIKE3D for Reese and Welch
Case

B-32
Table B-4 Comparisons of Ground-line Deflections (y0) between Broms Method, LPILE, and NIKE3D for Dunnavant Case

Lateral Load y0 from Broms Method LPILE Program NIKE3D

Dunnavant
(kips) (in) y0, in % Difference y0, in % Difference y0, in % Difference
0 0.000 0.000 0.00 0.000 0.00 0.000 0.00
25 0.022 0.013 -0.02 0.011 -50.00 0.017 -22.73
65 0.037 0.034 -0.01 0.044 +18.92 0.047 +27.03
85 0.040 0.044 +0.02 0.065 +62.50 0.066 +65.00
135 0.066 0.070 +0.03 0.130 +96.97 0.112 +69.70
180 0.093 0.093 0.00 0.202 +117.20 0.158 +69.89
250 0.296 0.130 -4.92 0.344 +16.22 0.236 -20.27
300 0.589 0.156 -25.52 0.470 -20.20 0.293 -50.25
350 0.909 0.182 -66.11 0.621 -31.68 0.356 -60.84
400 1.291 0.208 -139.86 0.803 -37.80 0.421 -67.39
410 1.634 0.213 -232.21 0.843 -48.41 0.434 -73.44

Table B-5 Comparisons of Ground-line Deflections (y0) between Broms Method, LPILE, and NIKE3D for Reese & Welch
Case

Lateral y0 from Broms Method LPILE Program NIKE3D

Load Reese & Welch
(kips) (in) y0, in % y0, in % Difference y0, in %
Difference Difference
0 0.000 0.000 0.00 0.000 0.00 0.000 0.00
22 0.020 0.034 +68.23 0.024 +20.00 0.085 +325.00
32 0.040 0.049 +22.35 0.054 +35.00 0.131 +227.50
43 0.065 0.066 +1.18 0.102 +56.92 0.191 +193.85
63 0.187 0.096 -48.47 0.231 +23.53 0.367 +96.26
83 0.452 0.127 -71.92 0.415 -8.19 0.597 +32.08
100 0.942 0.153 -83.76 0.615 -34.71 0.863 -8.39

B-33
Table B-6 Pros and Cons of Broms, LPILE, and NIKE3D

No. Items Broms Method LPILE Program NIKE3D

Pros Cons Pros Cons Pros Cons
1 Material represented Soil Linear Nonlinear Nonlinear
Mass
2 Solution Procedure Simple Simple Complicate
3 Continuity of Soil Mass No No Yes
4 Layered Soil No Yes Yes
5 Application to Various Limited Limited Any soil
Soil Conditions soil soil conditions
conditions conditions
similar to
data base
6 Load-Deflection Curves Only linear Whole range Whole range
portion of load of load
7 Deformed Shapes of Piles No Yes Yes

B-34
B.8 Conclusions

In previous sections, the development of finite element (FE) model was explained. This
process consists of preprocessing of FE analysis input information, FE analyses, and
post-processing of FE analysis results. Preprocessing is mesh generation using TrueGrid.
FE analyses are performed by NIKE3D. Post-processing is visualization and data
processing. Visualization can be done with Griz. Data processing can be obtained with
MS Excel.
Calibration of NIKE3D was conducted using two pile-load-test results from
Dunnavant (1986) and Reese & Welch (1975). Calibration process requires iterative
procedure. Mesh or model adjustment are essential steps in FE development.
Dunnavant’s test site was in Houston, Texas. On-site soils were saturated,
overconsolidated clays. Soil and pile data were provided. In Dunnavant case, the pile
failed at the load of 250 kips. So the results used up to 250-kip load are used in the
calibration. Both elastic and RO models were investigated. Two 2-layered soil meshes
using elastic soil were studied. In elastic models, elasticity from CHT gave the best fit for
test data up to 180 kips for both layered systems. Both 2-layered and 3-layered soil
meshes were analyzed using RO model. Results from both soil meshes with RO model
gave good agreement with test results for load up to 250 kips.
Reese & Welch site was in Austin, Texas. Soils at the site were overconsolidated
clays with water table at the depth of 18 ft below ground surface. Two cases were
analyzed on elastic 3-layered soil systems. Five RO models (case I to case V) were
investigated. It was found that the upper layer of soil had most effect on ground-line
deflections of the pile.
Results from Broms, LPILE, and NIKE3D were compared. Broms method’ s
results were calculated using three different methods in assessing kh of the soil. They are
back-figured kh, Terzaghi, and Davisson. Broms method using either Terzaghi or
Davissons’ methods in evaluating kh fails to predict ground-line deflections of the pile in
both Dunnavant and Reese & Welch tests. Results from Broms using back-figured kh
have good agreement with linear portion of load-deflection curves for both test data.
NIKE3D showed better results than LPILE did in Dunnavant test for the load up
to 250 kips. But LPILE showed better results for load beyond 250 kips. As already
known, pile failed after 250-kip load. So, the prediction of load-deflection curve for load
after 250 kips by LPILE is in question.
LPILE showed better results in Reese and Welch test excepting the last point at
load of 100 kips. This leads to the question of using LPILE to predict pile behavior after
pile was loaded to a certain load. NIKE3D showed ability of prediction of load-deflection
curve and gave the results closer to test data. NIKE3D also appeared to have potential in
giving better results if more soil data are provided.
From the comparisons of two test results, NIKE3D showed good results for both
cases. While LPILE leaves us with question about how good it can predict the behavior
of pile nearing failures.
NIKE3D can predict more insight into pile-soil interface behavior than either
LPILE or Broms method. NIKE3D can provide shear force at tip of the pile for static
equilibrium verification but LPILE does not have this ability. Soil pressure distribution

B-35
around and along the pile can also be obtained by NIKE3D. It has ability to predict pile
behavior under combined loadings and other more complicated problems.

B-36
APPENDIX C. FINITE ELEMENT ANALYSES OF ROCK SOCKETED
DRILLED SHAFTS UNDER LATERAL LOAD, MOMENT AND TORSION

C.1 Drilled Shafts for Sign Pole Structures

Sign pole structures are subjected to winds. Wind loads can induce lateral force,
moment, and torsion imposed on sign pole foundations. Usually drilled shafts are used as
foundations. This chapter will discuss the drilled shaft performance under lateral load,
moment, and torsion. Finite element method was used in analyzing drilled shafts under
lateral load, moment or torsion.
It is usually assumed that piles or drilled shafts supporting sign pole structures fail
by lateral loads or moments. The failure criteria of piles under lateral loads or moments
are either lateral deflections or ultimate moments of the piles. But the piles also are
subjected to torsions. There is a possibility that torsions could control the design or the
any combination of lateral loads, moments, and torsions. While this thesis would not
accomplish this goal of pile performance under the above combined loads, it is important
to unfold the mystery pile behavior. To design piles or drilled shafts, failure modes under
loads need to be known. Failure modes of piles under lateral loads and moments are
same. Failure modes for piles under torsions are not clear. It could be either axial rotation
of pile or pile itself fails under torsion.

C.2 Rock Socketed Drilled Shafts

In this chapter, the rock socketed drilled shaft under different loads is analyzed. Concrete
and soil properties are presented. Parameters for NIKE3D are also given.
Drilled shaft is assumed to have a diameter of 4 ft and total length of 40 ft. The
bottom 10 ft of the shaft is situated in the rock as shown in Figure C-1. Top of drilled
shaft is at ground surface. Failure criteria for drilled shafts under lateral loads, moments,
and torsions are presented.
The 4-ft diameter drilled shaft of 40-ft length with bottom 10 ft socketed in the
rock was analyzed. This drilled shaft is assumed to situate in uniform medium and hard
clay layers overlain over the rock. Average undrained shear strength of 0.75 and 1.75 tsf
were assigned for medium and hard clay layers, respectively. This soil-and-pile system
will be used in the FE analyses.

C-1
 As the rock properties, Young’s modulus and Poisson’s ratio are chosen as
250,000 ksf and 0.25, respectively. Soil and rock properties are summarized in Table C-
1. Ultimate compressive strength (fc’) of concrete is 5000 psi. Concrete has Young’s
modulus of 4,050,000 psi. The steel reinforcement consists of 19 #14 bars with yield
strength (fy) of 60,000 psi.
Two different material models were used. Rock and concrete were assumed
elastic. Clays were assumed to behave nonlinearly like Ramberg-Osgood (RO) materials.
Elastic parameters of rock and concrete are shown in Table C-2. Table C-3 shows RO
parameters for clays.
In NIKE3D, the coefficient of friction (µ) between clay and pier is assumed to be
0.35 for both medium and hard clays. The interface between rock and pier has the
coefficient of 1.0.
Drilled shafts under lateral loads or moments can fail in the ultimate bending
moment, the lateral deflections, or shaft-top rotation. The shaft used in analyses has the
ultimate moment capacity of the section of 4000 kip-ft. AASTHO (1998) recommends
that the allowable lateral deflections of piles should not exceed 1.5 inches. As a rule of
thumb, the maximum pile-top rotations should not be greater than 1.0 degree.
The current design code has not defined the failure criteria for drilled shafts under
torsions. Anyway in this thesis, shafts under torsions can fail in either the ultimate shear
resistance of concrete at shaft surfaces or the axial rotation. The limiting shear stress of
concrete pile caused by torsion is 2.4 f c' as recommended in the ACI Building Codes
(1989). Table C-4 shows failure criteria for shafts under lateral loads, moments, and
torsions.

C-2
 Drilled Shaft

30 ft
D = 4 ft

Clay

10 ft Rock

Figure C-1 Rock Socketed Drilled Shaft used in Analyses

Table C-1 Soil and Rock Properties

Material cu E ν Unit E/cu

(tsf) (ksf) Weight
(pcf)
Medium clay 0.75 181 0.35 122 115

Hard clay 1.75 657 0.35 129 180

Rock 575 250,000 0.25 150 207

C-3
Table C-2 Elastic Parameters for Rock and Concrete

Materail E Density ν
(ksf) (slug/ft3)
Rock 250,000 4.66 0.25

Concrete 670,000 4.5 0.20

Table C-3 RO Parameters for Clays

Clay τy γy α r K
(psf) (ksf)
Medium 28.28 0.000359 1.257 5.0 201

Hard 101.90 0.000359 1.257 5.0 723

Table C-4 Failure Criteria for Drilled Shafts

Failure Criteria Lateral Load Moment Torsion

Ultimate Moment 4000 4000 N/A

(kip-ft)
Lateral Deflection 1.5 1.5 N/A
(in.)
Shaft-top Rotation 1.0 1.0 N/A
(degree)
Limiting Shear Stress N/A N/A 24,400
of Pile Section (psf)
Axial Rotation N/A N/A 1.0
(degree)

C.3 Drilled Shaft under Lateral Load

Drilled shafts are founded in both medium and hard clays. The soil, rock, and concrete
properties outlined in previous section are used. Lateral load was applied at the top of
drilled shaft.
As mentioned in previous section, pier under lateral load can fail in three modes:
ultimate bending moment, excessive lateral deflection, or pier-top rotation. This section
will present the results from NIKE3D analyses in order to determine the failure modes of
the piers under lateral loads.
Lateral loads and ground-line deflections can be plotted and shown in Figure C-2.
From this graph, pier in hard clay deflects less than the one in medium clay at the same
load. Load of 155 kips is required to produce the deflection criterion of 1.5 inches for pier

C-4
in medium clay. For pier in hard clay, it is required 260 kips to reach at 1.5-inch lateral
deflections.
Load and shaft-top rotation relationship is shown in Figure C-3. Pier in hard clay
rotates less than pier in medium clay. At the loads that produce 1.5-inch deflection, shaft-
top rotations of 0.35 and 0.37 are found for piers in medium and hard clays, respectively.
These mean the failures of the shafts are not controlled by shaft rotations.
Deformed shapes of drilled shaft in medium clays are shown in Figure C-4.
Figure C-5 shows deformed shapes of drilled shaft in hard clay. Deformed shapes have
the same pattern for piers in both medium and hard clays. Points of zero deflection or
fixity are at the depth of 30 ft below ground surface for both cases.
Soil pressure distributions for piers in medium and hard clays are plotted in
Figure C-6 and C-7, respectively. The soil pressures have the same pattern in both
medium and hard clays. The maximum soil pressures occur at the depth of 30 ft below
ground surface for both cases. Rock pressures were changed to negatives values.
Shear force diagrams can be calculate from soil pressures. Figure C-8 and C-9
show shear force diagrams of pier in medium and hard clay, respectively.

C-5
 350

300
medium clay

250 hard clay

200
load, kips

150

100

50

0
0.0 0.5 1.0 1.5 2.0 2.5
y0, in

Figure C-2 Ground-line Deflections for Rock Socketed Piers under Lateral Loads

C-6
 450

400
medium clay

350
hard clay

300
load, kip

250

200

150

100

50

0
0.0 0.2 0.4 0.6 0.8

shaft-top rotation, degree

Figure C-3 Shaft-top Rotations for Rock Socketed Piers under Lateral Loads

C-7
 0

-5

-10

-15
depth, ft

-20
P = 70.4 k
-25 P = 121.1 k

P = 171 k
-30
P = 215.4 k

-35

-40
-0.5 0.0 0.5 1.0 1.5 2.0 2.5
deflection, in

Figure C-4 Deformed Shapes of Drilled Shaft under Lateral Loads in Medium Clay

C-8
 0

-5

-10

-15
depth, ft

P = 71.2 kips
-20
P = 109.5 kips

P = 202.6 kips
-25

P = 306.5 kips

-30 P = 403.9 kips

-35

-40
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

deflection, in

Figure C-5 Deformed Shapes of Drilled Shaft under Lateral Loads in Hard Clays

C-9
 0

-5

-10 P = 70.4 kips

P = 121.1 kips
-15
P = 171 kips
depth, ft

-20 P = 215.4 kips

-25

-30

-35

-40
-10000 -5000 0 5000 10000 15000 20000

pressure, psf

Figure C-6 Soil Pressure Distributions along the Pier under Lateral Loads in Medium
Clay

C-10
 0

-5

-10
P = 71.2 kips

P = 109.5 kips
-15
P = 202.6 kips
depth, ft

P = 306.5 kips
-20
P = 403.9 kips

-25

-30

-35

-40
-10000 -5000 0 5000 10000 15000 20000

pressure, psf

Figure C-7 Soil Pressure Distributions along the Pier under Lateral Loads in Hard Clay

C-11
 0

-5

-10

P = 70.4 kips
-15
P = 121.1 kips
depth, ft

-20 P = 171 kips

P = 215.4 kips
-25

-30

-35

-40
-150 -100 -50 0 50 100 150 200 250

shear, kips

Figure C-8 Shear Force Diagrams of Pier under Lateral Loads in Medium Clay

C-12
 0

-5

-10

-15
depth, ft

-20

P = 71.2 kips
-25
P = 109.5 kips

-30 P = 202.6 kips

P = 306.5 kips

-35
P = 403.9 kips

-40
-100 0 100 200 300 400 500
shear, kips

Figure C-9 Shear Force Diagrams of Pier under Lateral Loads in Hard Clay

From Figure C-8 and C-9, the maximum shear forces occur at ground surface and
decrease with depth until reaching minimum values at soil/rock interfaces for both piers
in medium and hard clay. Shear forces in the rock were changed to negative values.
Bending moment diagrams can be obtained from shear force diagrams as shown
in Figure C-10 and C-11 for pier in medium and hard clay, respectively. Bending
moments increase from zero values at ground surface to maximum values at soil/rock
interfaces and then decrease with the depth of rock
Maximum bending moments occur at points of fixity at soil/rock interfaces for
both cases. Loads and maximum bending moments in the piers can be plotted in Figure
C-12. The ultimate moment resistance of the pier is 4000 kip-ft. From Figure C-12, loads
of 180 and 230 kips produce the maximum bending moments in the piers for medium and
hard clay, respectively.

C-13
 0

-5 P = 70.4 kips

P = 121.1 kips
-10
P = 171 kips

-15 P = 215.4 kips

depth, ft

-20

-25

-30

-35

-40
0 1000 2000 3000 4000 5000 6000

moment, kip-ft

Figure C-10 Bending Moment Diagrams of Pier under Lateral Load in Medium Clay

C-14
 0
P = 71.2 kips

-5 P = 109.5 kips

P = 202.6 kips

-10 P = 306.5 kips

P = 403.9 kips
-15
depth, ft

-20

-25

-30

-35

-40
0 2000 4000 6000 8000

moment, kip-ft

Figure C-11 Bending Moment Diagrams of Pier under Lateral Loads in Hard Clay

C-15
 350

300

250
Medium clay

200 Hard clay

load, kips

150

100

50

0
0 1000 2000 3000 4000 5000 6000
M max, kip-ft

Figure C-12 Maximum Bending Moments in the Piers under Lateral Loads

C.4 Drilled Shaft under Moment

In this section, moment will be applied at shaft-top. Both rock socketed drilled shafts in
medium and hard clays were analyzed. Soil, rock, and concrete properties are the same as
lateral-load cases.
Failure modes of drilled shafts under moments are same as those under lateral
loads. From NIKE3D results, the shaft behaviors can be investigated. Same results
similar to piers under lateral loads can be produced for piers under moments.
Ground-line deflections of the piers can be plotted in Figure C-13. At deflection
of 1.5 inches, moments of 3700 and 5500 kip-ft are required to produce this deflection for
piers in medium and hard clay, respectively. Figure C-14 shows shaft-top rotations for
piers under moments. For the same moments that produce the 1.5-inch deflection, the
rotations of 0.45 and 0.55 degree are found for medium and hard clay, respectively.

C-16
 Deformed shapes of piers in medium and hard clay are shown in Figure C-15 and
C-16, respectively. From figure C-15, it is found that points of zero deflection are at the
depth of 32 ft below ground surface. For pier in hard clay, the points of zero deflections
are in between 28 to 32 ft below the ground surface.
Soil pressure distributions along the piers are plotted in Figure C-17 for pier
medium clay and Figure C-18 for pier in hard clay. The distribution patterns are the same
for piers under lateral loads and those under moments. The maximum soil pressures occur
at soil-rock interfaces for both piers under moments and lateral loads.
Figure C-19 and C-20 show shear force diagrams of pier in medium and hard
clay, respectively. Shear forces increase from zero at ground surface to the maximum
values at the depth of 32 ft below ground surface for pier in medium clay. For pier in
hard clay, the same distributions can be seen. The maximum shear forces occur at depths
of 28 to 32 ft below ground surface for pier in hard clay.
Bending moment diagrams of piers in medium and hard clay are shown in Figure
C-21 and C-22, respectively. Bending moment diagrams for both cases have the same
pattern. The maximum bending moments at ground surface decrease nonlinearly with
depth to minimum values at pier tips. The maximum bending moments equal to applied
moments at pier top.

C-17
 8000

7000
medium clay

6000
hard clay

5000
moment, kip-ft

4000

3000

2000

1000

0
0.0 0.5 1.0 1.5 2.0 2.5
y0, in

Figure C-13 Ground-line Deflections of the Piers under Moments

C-18
 10000

9000

8000
Medium clay
7000
Hard clay
applied moment, kip-ft

6000

5000

4000

3000

2000

1000

0
0 0.2 0.4 0.6 0.8 1 1.2

rotation, degree

Figure C-14 Shaft-top Rotations for Piers under Moments

C-19
 0

-5

-10

-15
M = 1552.3 kip-ft
depth, ft

M = 2785.9 kip-ft
-20
M = 3997 kip-ft

-25 M = 4690.7 kip-ft

M = 6898.2 kip-ft

-30

-35

-40
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

deflection, in

Figure C-15 Deformed Shapes of Pier under Moments in Medium Clay

C-20
 0

-5

-10

-15
depth, ft

M = 2621.9 kip-ft
-20
M = 4294 kip-ft

-25
M = 5975.6 kip-ft

M = 6816.6 kip-ft
-30

-35

-40
-0.5 0.0 0.5 1.0 1.5 2.0 2.5
deflection, in

Figure C-16 Deformed Shapes of Pier under Moments in Medium Clay

C-21
 0

-5

-10 M = 1552.3 kip-ft

M = 2785.9 kip-ft
-15
M = 4690.7 kip-ft
depth, ft

-20 M =6898.2 kip-ft

-25

-30

-35

-40
-15000 -10000 -5000 0 5000 10000 15000 20000 25000

pressure, psf

Figure C-17 Soil Pressure Distributions along the Pier under Moments in Medium Clay

C-22
 0

-5

-10 M = 2621.9 kip-ft

M = 4294.0 kip-ft

-15 M = 6816.6 kip-ft

depth, ft

M = 8378.6 kip-ft
-20

-25

-30

-35

-40
-8000 -6000 -4000 -2000 0 2000 4000 6000

pressure, psf

Figure C-18 Soil Pressure Distributions along the Pier under Moments in Hard Clay

C-23
 0

-5
M = 1552.3 kip-ft

-10 M = 2785.9 kip-ft

M = 4690.7 kip-ft
-15
M = 6898.2 kip-ft
depth, ft

-20

-25

-30

-35

-40
-300 -250 -200 -150 -100 -50 0
shear, kips

Figure C-19 Shear Force Diagrams of Pier under Moments in Medium Clay

C-24
 0

M = 2621.9 kip-ft
-5
M = 4294.0 kip-ft

-10 M = 6816.6 kip-ft

M = 8378.6 kip-ft
-15
depth, ft

-20

-25

-30

-35

-40
-250 -200 -150 -100 -50 0
shear, kips

Figure C-20 Shear Force Diagrams of Pier under Moments in Hard Clay

C-25
 0

-5

-10

-15
depth, ft

-20

-25

-30 M = 1552.3 kip-ft

M = 2785.9 kip-ft

-35 M = 4690.7 kip-ft

M = 6898.2 kip-ft

-40
0 1000 2000 3000 4000 5000 6000 7000 8000

moment, kip-ft

Figure C-21 Bending Moment Diagrams of Piers under Moments in Medium Clay

C-26
 0

-5

-10

-15
depth, ft

-20

-25

M = 2621.9 kip-ft
-30
M = 4294.0 kip-ft

M = 6816.6 kip-ft
-35

M = 8378.6 kip-ft

-40
0 2000 4000 6000 8000 10000

moment, kip-ft

Figure C-22 Bending Moment Diagrams of Piers under Moments in Hard Clay

C.5 Shaft under Torsion

Torsions were applied at the top of drilled shafts. Drilled shafts under torsions could be
failed by maximum shear stresses in shafts, maximum shear stresses in the soil, and
excessive axial rotation of the shafts.
This section will be devoted to investigate the behavior of rock socketed drilled
shafts under torsions. Total of five piers under torsions were investigated. Four cases
were analyzed by NIKE3D. The first two cases (interfaced piers) contained interfaces
between clay (or rock) and piers. The second two cases (glued piers) glued clay (or rock)
and piers together. The last pier was fixed-end pier under torsion without soil (or rock)
and analyzed using mechanics. This pier is shown in Figure C-23.

C-27
 4 ft

40 ft T

Figure C-23 Fixed-end Pier under Torsion without Soil

Axial rotations of piers under torsions can be shown in Figure C-24. There is no
difference between interfaced pier in medium and hard clays. Results from both glued
cases are the same. Fixed-end pier serves as lower bound of the rotations. This case gives
the highest rotation for the given torsion. Glued piers give the lowest rotation for the
same torsion. For the rotation of 1.0 degree, fixed-end pier requires torsion of 3000 kip-ft.
Torsions of 3400 and 3800 kip-ft are required to product one-degree rotation for
interfaced and glued piers.
Figure C-25 shows the maximum shear stresses in the piers for all cases. For the
limiting shear stress of 24,400 psf, fixed-end pier shows the lowest required torsion of
307 kip-ft. The glued piers required the highest torsion of 3700 kip-ft in order to produce
the limiting shear stress. For interfaced piers, torsion of 1950 kip-ft can produce the
limiting value of 24,400 psf.
From NIKE3D glued pier results, the maximum shear stresses in soil occur at
ground surface. The maximum shear stresses decrease drastically with the distance from
pier surface. The maximum shear stresses in the soil can be plotted along the distance
from pier surface and shown in Figure C-26.

C-28
 5000

4500

4000

3500

3000
torsion, kip-ft

2500

2000
medium
1500
hard

1000 no soil

glued: medium
500
glued: hard

0
0.0 0.5 1.0 1.5
rotation, degree

Figure C-24 Axial Rotations of Piers under Torsions

C-29
 5000

4500

4000

3500
medium clay

3000 hard clay

torsion, kip-ft

no soil
2500 glued: medium
glued: hard
2000

1500

1000

500

0
0 20000 40000 60000 80000
max. shear stress, psf

Figure C-25 Maximum Shear Stresses in Piers under Torsions

C-30
 90

T = 4725.9 kip-ft
80

T = 3524.4 kip-ft
70
T = 1002.7 kip-ft

60
max. shear stress, psf

Max. shear stresses at pier surface

50
30116 psf for T = 4725.9 kip-ft
22424 psf for T = 3524.4 kip-ft
40 6290 psf for T = 1002.7 kip-ft

30

20

10

0
0 5 10 15 20 25
distance from pier surface, ft

Figure C-26 Maximum Shear Stresses in the Soil along the Radial Direction for Glued
Pier

The maximum shear stresses in soil along the depth for glued pier can be plotted
in Figure C-27. Figure C-27 shows that the maximum values at ground surface decrease
to very little values at depth of 12 ft below ground surface.

C-31
 0

depth from ground surface, ft -5

-10

T = 4725.9 kip-ft

-15
T = 3524.4 kip-ft

T = 1002.7 kip-ft
-20

-25

-30
0 10000 20000 30000 40000
max. shear stress, psf

Figure C-27 Maximum Shear Stresses in the Soil along the Depth for Glued Pier

C.6 Conclusions

This chapter presented the results of analyses of rock socketed drilled shafts under lateral
load, moments, and torsions, separately. The shaft has 4-ft diameter and 40-ft length with
the bottom 10 ft embedded in the rock Shaft, soil, and rock properties were given.
For shafts under lateral loads, ground-line deflections, shaft-top rotations,
deformed shapes, soil pressures, shear force diagrams, and bending moment diagrams
were obtained. The failure mode for drilled shaft under lateral load in medium clay was
excessive lateral deflection at failure load (Pf) of 170 kips. The failure mode for drilled
shaft under lateral load in hard clay was maximum bending moment with the failure load
(Pf) of 230 kips.
For shafts under moments, the same results as those under lateral loads were
observed. The failure mode for shaft under moment in medium clay was excessive lateral

C-32
deflection at failure moment (Mf) of 3700 kip-ft. For drilled shaft in hard clay, it failed by
maximum bending moment (Mf) of 4000 kip-ft.
Axial rotations and maximum shear stresses in the piers for frictional interface,
glued interface, and fixed-end piers were given. The maximum shear stresses in the soil
at different radial distances and depths for glued piers are also provided. For drilled
shafts under torsions in both medium and hard clay, shafts failed by maximum shear
stresses in concrete at failure torsions (Tf) of 1950 kip-ft. Table C-5 shows the capacities
of drilled shafts under different loads.

Table C-5 Capacities of Drilled Shafts under Different Loads

Failure Mode Shafts in Medium Clay Shafts in Hard Clay

for

Lateral Loads Lateral deflection at Pf = 170 Max. moment at Pf = 230

kips kips
Moments Lateral deflection at Mf = 170 Max. moment at Mf = 230
kips kips
Torsions Max. shear stress at Max. shear stress at
Tf = 1950 kip-ft Tf = 1950 kip-ft

For drilled shafts under lateral loads or moments in clays, shaft capacities depend
on strength and stiffness of clays. Shaft capacities under torsions are independent of
strength and stiffness of clays but depend on coefficients of friction between soil and
pier.

C-33