Terms and Conditions of Use

:
Visit our
companion site
http://www.vulcanhammer.org
this document downloaded from
vulcanhammer.net
Since 1997, your complete
online resource for
information geotecnical
engineering and deep
foundations:
The Wave Equation Page for
Piling
Online books on all aspects of
soil mechanics, foundations and
marine construction
Free general engineering and
geotechnical software
And much more...
All of the information, data and computer software
(“information”) presented on this web site is for general
information only. While every effort will be made to insure
its accuracy, this information should not be used or relied on
for any specifc application without independent, competent
professional examination and verifcation of its accuracy,
suitability and applicability by a licensed professional. Anyone
making use of this information does so at his or her own risk
and assumes any and all liability resulting from such use.
The entire risk as to quality or usability of the information
contained within is with the reader. In no event will this web
page or webmaster be held liable, nor does this web page
or its webmaster provide insurance against liability, for
any damages including lost profts, lost savings or any
other incidental or consequential damages arising from
the use or inability to use the information contained
within.
This site is not an offcial site of Prentice-Hall,
Pile Buck, the University of Tennessee at
Chattanooga, or Vulcan Foundation
Equipment. All references to sources of
software, equipment, parts, service
or repairs do not constitute an
endorsement.





ANALYSIS OF LATERALLY LOADED DRILLED SHAFTS IN ROCK







A Dissertation

Presented to

The Graduate Faculty of The University of Akron







In Partial Fulfillment

of the Requirements for the Degree

Doctor of Philosophy









Ke Yang

May, 2006
ii




ANALYSIS OF LATERALLY LOADED DRILLED SHAFTS IN ROCK





Ke Yang



Dissertation



Approved: Accepted:


Advisor Department Chair
Dr. Robert Liang Dr. Wieslaw Binienda


Committee Member Dean of the College
Dr. Wieslaw Binienda Dr. George Haritos


Committee Member Dean of the Graduate School
Dr. Ernian Pan Dr. George Newkome


Committee Member Date
Dr. Yueh-Jaw Lin


Committee Member
Dr. Chien-Chung Chan



iii




ABSTRACT


Drilled shafts socketed into rock are widely used as foundations for bridges and
other important structures. Rock-socketed drilled shafts are also used to stabilize a
landslide. The main loads applied on the drilled shafts are axial compressive or uplift
loads as well as lateral loads with accompanying moments. Although there exist several
analysis and design methods especially for rock-socketed drilled shafts under lateral
loading, these methods were developed with assumptions without actual validations with
field load test results. Some of the methods have been found to provide unsafe designs
when compared to recently available field test data. Therefore, there is a need to develop
a more rational design approach for laterally loaded drilled shafts socketed in rock.
A hyperbolic non-linear p-y criterion for rock is developed in this study that can be
used in conjunction with existing computer programs, such as COM624P, LPILE, and
FBPIER, to predict the deflection, moment, and shear responses of a shaft under the
applied lateral loads. Considerations for the effects of joints and discontinuities on the
rock mass modulus and strength are included in the p-y criterion. Evaluations based on
comparisons between the predicted and measured responses of full-scale lateral load tests
on fully instrumented drilled shafts have shown the applicability of the proposed p-y
iv
criterion and the associated methods for determining the required input of rock
parameters.
In addition to the development of a hyperbolic p-y criterion for rock, a method for
predicting lateral capacities of drilled shafts in rock and/or soils is developed for
assessing the safety margin of the designed shafts against the design loads. A computer
program LCPILE is developed using VC++ to facilitate computations. An elastic solution
based on a variational approach is also developed for determining drilled shaft elastic
deflection due to applied lateral loads in a two-layer soil layer system. The computational
algorithm was coded in a Mathematica file for easy application.
Finally, Briaud’s method for deriving p-y curves of rock from pressuremeter or
dilatometer test results is evaluated using available field test data. A modification to the
Briaud’s method is recommended for applications in rocks.
v



DEDICATION


To my wife, my father and mother, who made all of this possible, for their endless
encouragement and patience.




























vi






ACKNOWLEDGEMENTS


As I approach the end of my graduate tenure, I am left with profound feelings of
accomplishment, excitement, and gratitude to a number of people whose support has been
significant in completing this dissertation.
My advisor Professor Robert Y. Liang, has been a constant source of inspiration,
motivation and guidance throughout the past three and a half years. I have learned so
much from his keen insight, his research and problem solving abilities, and his amazing
energy. What I learned from him is not only the knowledge, but also the way to be a
positive and energetic person.
Thanks in large part to the kindness and mentoring provided by the committee
members: Dr. Wieslaw Binienda, Dr. Ernian Pan, Dr. Yueh-Jaw Lin, and Dr. Chien-
Chung Chan. Additionally, the great help from Dr. Jamal Nusairat is highly appreciated.
My wife, parents, and brother have been extremely supportive of my studies, and
have always been encouraging and understanding. I am also grateful to my wonderful
friends in Akron, Ohio. Their aid has been invaluable during the period of this work.




vii



TABLE OF CONTENTS

Page
LIST OF TABLES ……………………………………………………………………..xiii
LIST OF FIGURES …………………………………………………………………….xvi
CHAPTER
I. INTRODUCTION......................................................................................................... 1
1.1 Statement of the Problem.................................................................................. 1
1.2 Objectives.......................................................................................................... 3
1.3 Scope of the Work............................................................................................. 4
1.4 Outlines of the Dissertation............................................................................... 6
II. LITERATURE REVIEW............................................................................................ 8
2.1 Analysis Methods of Laterally Loaded Rock-Socketed Drilled Shafts ............ 8
2.1.1 Introduction.......................................................................................... 8
2.1.2 Brief Description of the Existing Methods ........................................ 10
2.1.3 Comments on the Existing Analysis Methods ................................... 23
2.2 Bedrocks of Ohio ............................................................................................ 26
2.2.1 The Distribution of Rocks in Ohio..................................................... 26
2.2.2 The Shales in Ohio............................................................................. 30
2.2.3 Limestone and Dolomite.................................................................... 34
viii
2.3 Mechanical Characteristics of Rock and Its Classifications ........................... 35
2.3.1 The types of rocks .............................................................................. 35
2.3.2 Features for Rock Characterization.................................................... 35
2.3.3 Rock Mass Classifications ................................................................. 38

III. LATERAL LOAD TEST RESULTS AND DATA ANALYSES ........................... 49
3.1 Lateral Load Test at Dayton............................................................................ 49
3.1.1 Test Site.............................................................................................. 49
3.1.2 Test Setup and Procedure................................................................... 52
3.1.3 Lateral Load Test Results................................................................... 53
3.2 Lateral Load Test at Pomeroy-Mason............................................................. 57
3.2.1 Test Site.............................................................................................. 57
3.2.2 Test Set-up and Test Procedure.......................................................... 62
3.2.3 Lateral Load Test Results................................................................... 64
3.3 Methods for Deriving P-y Curves From Lateral Load Test Results ............... 70
3.3.1 Introduction........................................................................................ 70
3.3.2 Method for Deriving Deflection from Strain Gage Readings............ 72
3.3.3 Determination of Moment Profiles .................................................... 73
3.3.4 Methods for Deriving P (Net Resistance) .......................................... 74
3.3.5 Evaluation Using Field Test Data ...................................................... 81
3.3.6 Evaluation Using Hypothetical Cases................................................ 88
3.3.7 Optimum Strain Gage Spacing........................................................... 90
3.3.8 Effect of Measurement Error ............................................................. 92
3.3.9 Conclusions on Methods for Deriving p-y Curves ............................ 94
ix
3.4 Analyses of the Load Tests ............................................................................. 95
3.4.1 Evaluation of Experimental P-y Curves............................................. 95
3.4.2 Evaluation of Reese Interim Rock p-y Criterion.............................. 100
3.5 Summary and Conclusions............................................................................ 104
IV. LATERAL CAPACITY OF DRILLED SHAFTS IN ROCK............................... 105
4.1 3-D Finite Element Modeling and Validation............................................... 106
4.1.1 3-D FEM Modeling.......................................................................... 107
4.1.2 Validation of the 3D FEM Model .................................................... 111
4.2 Failure Modes of Rock Subjected to Loading from
Laterally Loaded Drilled Shafts .................................................................... 115

4.2.1 Failure of Isotropic and Homogeneous Rock
without Overburden Soils................................................................ 115
4.2.2 Failure of Jointed Rock without Overburden Soils.......................... 124
4.2.3 Failure of Isotropic and Jointed Rock with Overburden Soils......... 129
4.3 Rock Strength Criteria................................................................................... 129
4.3.1 Hoek-Brown Criterion ..................................................................... 130
4.3.2 Converting Hoek-Brown Criterion to Mohr-Coulomb Criterion..... 132
4.4 Side Shear Resistance ................................................................................... 132
4.4.1 Empirical Equations for Axially Loaded Drilled Shaft in Rock...... 134

4.4.2 Suggested Empirical Equation for
Side Resistance in Horizontal Direction........................................... 136
4.5 Ultimate Resistance of Rock Mass ............................................................... 138
4.5.1 Ultimate Resistance of Rock Near Surface...................................... 138
4.5.2 Ultimate Rock Resistance at Great Depth........................................ 140
x
4.5.3 Ultimate Resistance of Jointed Rock ............................................... 141
4.6 New Method for Predicting Lateral Capacity of Drilled Shafts in Rock...... 142
4.6.1 Free Head Boundary ........................................................................ 142
4.6.2 Fixed Head Boundary ...................................................................... 145
4.7 Ultimate Reaction of Soils ............................................................................ 148
4.7.1 Ultimate Resistance of Clay............................................................. 148
4.7.2 Ultimate Resistance of Sand ............................................................ 150
4.7.3 Ultimate Resistance of c-φ Soils ...................................................... 151
4.8 Validation of the Derived Capacity Prediction Method................................ 152
4.9 Summary and Conclusions............................................................................ 155
V. ELASTIC SOLUTION OF LATERALLY LOADED
DRILLED SHAFTS IN ROCK............................................................................... 157
5.1 Determination of Rock Mass Deformability................................................. 157
5.1.1 Introduction...................................................................................... 157
5.1.2 Goodman Jack Test .......................................................................... 158
5.1.3 Selection of In-Situ Test Method..................................................... 163
5.1.4 Existing Empirical Equations........................................................... 165
5.1.5 Parameters for Determination of E
m
................................................ 167
5.1.6 Proposed Empirical Equation........................................................... 169

5.1.7 Recommended Methodology for Determination
of Deformation Modulus of Rock .................................................... 170
5.2 Initial Modulus of Subgrade Reaction of Rock Mass ................................... 171
5.2.1 Introduction...................................................................................... 171
5.2.2 FEM Model and Parametric Study................................................... 173
xi
5.2.3 Suggested Empirical Equation......................................................... 179
5.2.4 Validation of the Empirical Equation .............................................. 180
5.3 Numerical Solution for Laterally Loaded Piles in
A Two Layer Soil Profile.............................................................................. 184
5.3.1 Introduction...................................................................................... 184
5.3.2 Definition of the Problem ................................................................ 186
5.3.3 Variational Solution ......................................................................... 187
5.3.4 Validations ....................................................................................... 191
5.3.5 Methods for Determining Input Parameters..................................... 195
5.3.6 Case Study........................................................................................ 195
5.4 Summary and Conclusions............................................................................ 198
VI. P-Y CRITERION FOR ROCK MASS .................................................................. 200
6.1 General Shape of P-y Curve in Rock ............................................................ 200
6.2 Determination of p
u
....................................................................................... 202
6.2.1 Failure Modes .................................................................................. 202
6.2.2 P
u
Near Surface ................................................................................ 204
6.2.3 P
u
at Great Depth.............................................................................. 205
6.2.4 P
u
of Jointed Rock............................................................................ 206
6.3 Initial Tangent to P-y Curve K
i
..................................................................... 206
6.4 Rock Mass Properties.................................................................................... 207
6.4.1 Strength Parameters ......................................................................... 207
6.4.2 Rock-Shaft Interface Strength.......................................................... 209
6.4.3 Rock Mass Modulus E
m
................................................................... 209
xii
6.5 Construction of P-y curves for Rock Mass ................................................... 209
6.6 Comparison of the Proposed P-y Criterion with
That of Gabr et al. (2002).............................................................................. 210
6.7 Case Studies .................................................................................................. 211
6.7.1 Dayton Load Test............................................................................. 213
6.7.2 Pomeroy-Mason Load Test .............................................................. 217
6.7.3 Load Tests at North Carolina........................................................... 221
6.8 Conclusions................................................................................................... 226
VII. DERIVING P-Y CURVE FROM DILATOMETER TESTS............................... 227
7.1 Pressuremeter and Dilatometer ..................................................................... 227
7.1.1 Modulus of Rock Mass .................................................................... 232
7.1.2 Limit Pressure .................................................................................. 232
7.1.3 Undrained Shear Strength ................................................................ 232
7.1.4 Friction Angle .................................................................................. 234
7.2 Deriving p-y Curves from Pressuremeter/Dilatometer Test Results............. 238
7.3 Evaluation ..................................................................................................... 244
VIII. SUMMARIES AND CONCLUSIONS............................................................... 252
8.1 Summaries..................................................................................................... 252
8.2 Conclusions................................................................................................... 253
8.3 Future Studies................................................................................................ 255
REFERENCES............................................................................................................. 257
xiii



LIST OF TABLES
Table Page
2-1 Engineering Properties of Rocks in Ohio………………………………

33
2-2 Simplified Rock Classification…………………………………………

36
2-3 Factors influencing Rock Mass Behavior………………………………

37
2-4 Classification Parameters and Their Rating
(After Bieniawski, 1976)……………………………………………….


41
2-5 Rock Mass Rating System (Bieniawski, 1989) ………………………...

42
2-6 GSI Ranges for General Rocks (Marinos and Hoek, 2000) ……………44
2-7 GSI Ranges for Typical Sandstones (Marinos and Hoek, 2000)……….

45
2-8 GSI Ranges for Typical Siltstones, Claystones
and Clay Shales (Marinos and Hoek, 2000) …………………………...


46
2-9 GSI Ranges for Typical Limestones (Marinos and Hoek, 2000)……….

47
2-10 GSI Estimates for Heterogeneous Rock Masses
Such as Flysch (Marinos and Hoek, 2000) …………………...………..


48
3-1 RMR Ratings and GSI Values of Rock at Dayton Test Site……………

51
3-2 RMR Ratings and GSI Values of Rock Based on Boring S-9………….

59
3-3 RMR Ratings and GSI Values of Rock Based on Boring S-10………...

62
3-4 Compiled Lateral Load Test Database………………………………….

82
3-5 Cumulative Shaft Head Deflection Errors
based on Various Methods……………………………………………...



87
xiv
3-6 Cumulative Moment Errors based on Various Methods……………….

87
3-7 Soil Parameters of Hypothetical Cases…………………………………

88
3-8 Cumulative Errors of p-y Curves of Hypothetical Cases……………….

89
3-9 Rock Properties for LPILE Analysis…………………………………...

100
3-10 Test Drilled Shafts Information………………………………………...

102
4-1 Rock Mass Properties…………………………………………………..

112
4-2 Summary of Rock Strength Criteria……………………………………

130
4-3 Values of Constant m
i
for Intact Rock
(After Marinos and Hoek, 2000)………………………………………..


131
4-4 Roughness Classes (After Pells et al., 1980) …………………………..

135
4-5 Recommended Values of K by Kulhawy et al. (1983)
and Kulhawy (1991) …………………………………………………...


151
4-6 Recommended Values of δ by Kulhawy et al. (1983)
and Kulhawy (1991) …………………………………………………...


151
4-7 Test Drilled Shaft Information………………………………………….

153
4-8 Input rock mass parameters of the load tests…………………………...

153
4-9 Comparison of Lateral Capacities of Test Drilled Shafts………………

154
5-1 Values of Constant K(ν) for β=45°……………………………………..

160
5-2 Values of T* (Heuze and Amadei, 1985) ……………………………...

163
5-3 Empirical Equations for Estimating the
Deformation Modulus of Rock Mass…………………………………...


166
5-4 The Strength of the Relation between E
M
and Parameters
(Kayabasi et al. 2003) ………………………………………………….


168
5-5 Properties of Rock Masses in Ironton-Russell………………………….

168
5-6 Predictions and Ratios of Predicted over
Measured Modulus of Rock Masses……………………………………

169
xv
5-7 Summary of Lateral Load Test Drilled Shafts………………………….

181
5-8 Modulus of Rock Masses Based on Pressuremeter Test……………….

182
5-9 Modulus of Rock Masses Based on Empirical Equation……………….

182
5-10 Measured and Predicted Initial Modulus of Subgrade Reaction……….

183
5-11 Summary of Definitions Related to Subgrade Reaction Theory……….

186
6-1 Comparison of P-y Criteria……………………………………………..

212
6-2 Input Rock Mass Parameters of Dayton Load Test…………………….

213
6-3 Input Rock Mass Parameters of Pomeroy-Mason Load Test…………..

217
6-4 Input Rock Mass Parameters of I-40 Load Test………………………..

222
6-5 Input Rock Mass Parameters of I-85 Load Test………………………..

223
7-1 Preliminary estimates of φ’
CV
(Robertson and Hughes, 1986)…………

236
7-2 The Rheological Factor α for
Various Soils (Baguelin et al., 1978) …………………………………..


239
xvi



LIST OF FIGURES
Figure Page
1-1 Drilled shaft and soil models of p-y analysis…………………………...

3
1-2 Flow chart of the work…………………………………………………

5
2-1 Distribution of ultimate lateral force
per unit length (after Carter and Kulhawy 1992) ………........................


13
2-2 Rock-shaft model (a) Shaft and soil/rock mass system;
(b) Coordinate system and displacement components;
(c) Shear force V(z) and moment M(z) acting on shaft
at z (Zhang et al. 2000) ………………………………………………...




16
2-3 Consideration of yielding of soil and/or rock mass
by decomposition of loading (after Zhang et al. 2000) ………………..


18
2-4 Components of rock mass resistance (Zhang et al. 2000) ……………..

19
2-5 Typical forces on wedge ……………………………………………….

23
2-6 The elastoplastic and brittle behavior of rock mass ……………………

25
2-7 Geological map of Ohio, showing the pattern of
surface rocks across the state ………………………………………......


28
2-8 Cross section through the rocks of central Ohio from the
Indiana-Ohio border to the Ohio River (taken from
Feldmann et al., 1996) ………………………………………………....



28
2-9 Generalized column of bedrock units in Ohio………………………….

29
2-10 Classification of shales for embankment construction
(Wood and Deo, 1975) ………………………………………………...


32
3-1 Soil and rock layer profiles at Dayton test site ………………………...

51
xvii
3-2 Instrumentation of load test at Dayton …………………………………

53
3-3 Load-deflection curves at the top of shafts……………………………..

54
3-4 Deflection-depth curves of shaft #3…………………………………….

55
3-5 Deflection-depth curves of shaft #4…………………………………….

55
3-6 Compression strain profiles of shaft #4………………………………...

56
3-7 Tension strain profiles of shaft #4……………………………………...

57
3-8 Boring S-9 ……………………………………………………………...

60
3-9 Boring S-10……………………………………………………………..

61
3-10 Instrumentation and load test setup……………………………………..

64
3-11 Measured load-deflection curves at loading point of
Pomeroy-Mason test……………………………………………………


65
3-12 Deflection-depth profiles of drilled shaft #1
at Pomeroy-Mason test…………………………………………………


66
3-13 Deflection-depth profiles of drilled shaft #2
at Pomeroy-Mason test…………………………………………………


67
3-14 Tension strain profiles of test shaft #1 of Pomeroy-Mason test………..

68
3-15 Compression strain profiles of test
shaft #1 of Pomeroy-Mason test………………………………………..


68
3-16 Tension strain profiles of test shaft #2 of Pomeroy-Mason test………..

69
3-17 Compression strain profiles of
test shaft #2 of Pomeroy-Mason test……………………………………


69
3-18 Procedure for reducing moment data to p using piecewise
polynomial (after Dunnavant, 1986) …………………………………..


75
3-19 Linear shape functions………………………………………………….

77
3-20 Elimination of outlier moment profiles…………………………………


83
xviii
3-21 Comparison of deflections from strain readings
and inclinometer for test PomS1………………………………………..


83
3-22 P vs. depth profile of test MaumeeS1…………………………………..

85
3-23 P vs. depth profile of test CDOTC1…………………………………….

86
3-24 Definition of deflection prediction error………………………………..

86
3-25 Comparison of p-y curves………………………………………………

90
3-26 Optimum spacing of strain gages……………………………………….

92
3-27 Four types of moment error profiles……………………………………

93
3-28 Cumulative p-y curve errors due to moment errors…………………….

94
3-29 Experimental p-y curves for shaft #4 of Dayton load test……………...

96
3-30 Experimental p-y curves for shaft #1 of
Pomeroy-Mason load test………………………………………………


96
3-31 Experimental p-y curves for shaft #2 of
Pomeroy-Mason load test………………………………………………


97
3-32 Equivalent load combinations…………………………………………..

98
3-33 Prediction of load-deflection curves of shaft #4 of
Dayton load test using experimental p-y curves………………………..


98
3-34 Prediction of load-deflection curves of shaft #1 of
Pomeroy-Mason load test using experimental p-y curves……………...


99
3-35 Prediction of load-deflection curves of shaft #2 of
Pomeroy-Mason load test using experimental p-y curves……………...


99
3-36 Prediction of load-deflection curve of shaft #4 of
Dayton load test using Reese weak rock p-y criterion………………….


102
3-37 Prediction of load-deflection curve of shaft #1 of
Pomeroy-Mason load test using Reese weak rock p-y criterion………..


103
3-38 Prediction of load-deflection curve of shaft #2 of
Pomeroy-Mason load test using Reese weak rock p-y criterion………..


103
xix
4-1 Finite element meshes of a drilled shaft-rock system ………………….

107
4-2 Finite elements for (a) drilled shaft,
(b) surrounding rock, and (c) outside boundary of rock………………..


108
4-3 Cap model: yield surface in the p-t plane (ABAQUS, 1998)…………..

110
4-4 Drilled shaft dimension and rock profiles
of the load test at Dayton……………………………………………….


111
4-5 Mesh convergence……………………………………………………...

113
4-6 Comparison of load-deflection curves at shaft head……………………

114
4-7 Comparison of deflection profiles at
the load of (a) 1126 kips and (b) 705 kip ………………………………


114
4-8 The forward movement of rock mass…………………………………..

116
4-9 Front view of upward movement of rock mass………………………...

117
4-10 Y direction stress distribution in xz plane………………………………

118
4-11 Maximum shear stress distribution on yz plane………………………..

119
4-12 Proposed wedge type failure model for the top layer of rock…………..

119
4-13 Stress distribution at Y direction of in-depth rock layer………………..

121
4-14 The predefined cracks…………………………………………………..

122
4-15 The stress redistribution at Y direction of
in-depth rock layer after crack………………………………………….


122
4-16 Friction distribution on the shaft-rock interface………………………..

123
4-17 Suggested stress distribution at failure at great depth ………………….

124
4-18 Tensile stress on xz plane for jointed rock mass case………………….

127
4-19 Maximum shear stress concentrations for
jointed rock mass case………………………………………………….


127
4-20 Concrete and rock joint…………………………………………………

134
xx
4-21 Displacement behavior of drilled shafts in rock
(Johnston and Lam, 1989a)………………………… ………………….


137
4-22 Increased normal stress due to lateral load……………………………..

137
4-23 The Rock-Shaft Model………………………………………………….

143
4-24 Lateral capacity calculation models for drilled shafts in rocks…………

144
4-25 Capacity of rigid drilled shaft at fixed head boundary…………………

146
4-26 Capacity of intermediate length drilled shaft
at fixed head boundary………………………………………………….


147
4-27 Capacity of long drilled shaft at fixed head boundary………………….

148
4-28 Distribution of lateral reaction stresses…………………………………

149
5-1 The schematic of loading of Goodman Jack (Heuze, 1984) …………...

159
5-2 Modulus reduction vs. hydraulic pressure for various borehole
diameters (a) undersize – the pressure must be decreased by about
14% for a given E
app
/E
act
as recalculated by Heuze et al.(1985).
(b) oversize holes…………………………………………………..…...




162
5-3 Correction for platen bending of the jack
(after Heuze and Amadei, 1985) …………………………………….....


163
5-4 Proposed empirical equation using GSI………………………………..

171
5-5 Initial modulus of subgrade reaction for various moduli of rock………

175
5-6 Initial modulus of subgrade reaction for various Poisson’s ratio………

175
5-7 P-y curves along with depth ……………………………………………

176
5-8 Initial modulus of subgrade reaction for
various rock-shaft interface frictions…………………………...………


177
5-9 K
i
varies with shaft diameter…………………………………………...

178
5-10 Initial modulus of subgrade reaction for
various shaft-rock relative stiffness………………………………….…



179
xxi
5-11 Comparison of FEM computed and predicted
subgrade reaction modulus……………………………………………..


180
5-12 Comparison of load-deflection curves of
North Carolina load test………………………………………………...


181
5-13 Validation of empirical equation using field test data………………….

184
5-14 Two-layer soil profile with four possible variations……………………

187
5-15 Soil-pile system…………………………………………………………

187
5-16 Comparison with Reese and Matlock
solutions– varying soil stiffness ………………………………………..


192
5-17 Comparisons with Davisson and Gill method
for free head condition …………………………………………………


193
5-18 Comparison with Davisson and Gill solution
for fixed head condition ………………………………………………..


194
5-19 Shaft and soil profiles of the case study………………………………...

196
5-20 Interpretation of subgrade reaction modulus of rock…………………...

197
5-21 Comparison of shaft head deflections…………………………………..

198
6-1 P-y curves deduced from shaft #4 of load test at Dayton………………

201
6-2 P-y curves deduced from Shaft #2 of
load test at Pomeroy-Mason…………………………………………….


201
6-3 Schematics of a hyperbolic p-y curve…………………………………..

202
6-4 Failure mode for rock near ground surface……………………………..

203
6-5 Failure mode of rock at great depth…………………………………….

204
6-6 Hyperbolic p-y curves of Dayton site…………………………………..

214
6-7 Comparison of load-deflection of
test shaft #4 at Dayton load test………………………………………...


215
6-8 Comparison of load-Maximum moment of
test shaft #4 at Dayton load test………………………………………...

215
xxii
6-9 Comparisons of deflection-depth
curves of shaft #4 at Dayton test………………………………………..


216
6-10 Comparisons of moment-depth
curves of shaft #4 at Dayton test………………………………………..


216
6-11 Hyperbolic p-y curves of Pomeroy-Mason site………………………...

219
6-12 Comparison of load-deflection at the
loading point for Pomeroy-Mason test…………………………………


219
6-13 Comparison of load-Maximum moment
of Pomeroy-Mason load test……………………………………………


220
6-14 Comparisons of deflection-depth curves
of shaft #2 at Pomeroy-Mason test……………………………………..


220
6-15 Comparisons of moment-depth curves
of shaft #2 at Pomeroy-Mason test……………………………………..


221
6-16 Layout of I-40 load test…………………………………………………

222
6-17 Layout of I-85 load test…………………………………………………

222
6-18 Comparison of load-deflection curves of I-40 short shaft………...……

224
6-19 Comparison of load-deflection curves of I-40 long shaft………………

224
6-20 Comparison of load-deflection curves of I-85 short shaft……………...

225
6-21 Comparison of load-deflection curves of I-85 long shaft………………

225
7-1 Menard G-Am pressuremeter from RocTest…………………………...

228
7-2 PROBEX-1 Dilatometer from RocTest………………………………...

228
7-3 Calibration for volume and pressure losses…………………………….

229
7-4 Typical pressuremeter/dilatometer test curve…………………………..

231
7-5 Determination of limit pressure from
inverse of volume versus pressure………………...……………………


233
7-6 P-y curves from pressuremeter (Baguelin et al., 1978) ………………..

240
xxiii
7-7 Steps for constructing p-y curve from pressuremeter test……………...

241
7-8 Determination of the critical depth (Smith, 1983) ……………………..

243
7-9 Reduction factor for depth
within critical depth (Smith, 1983)……………………………………..


244
7-10 Shaft and soil profiles of the case study………………………………...

245
7-11 Dilatometer test results of the case study
(after Cho et al., 2001)………………………………………………….


246
7-12 P-y curves from dilatometer tests of Cho et al. (2001)…………………

246
7-13 Comparison of the measured and
predicted deflections at shaft top……………………………………….


247
7-14 Derived p-y curves from dilatometers for I-40 short shaft……………..

248
7-15 Derived p-y curves from dilatometers for I-40 long shaft……….……..

248
7-16 Derived p-y curves from dilatometers for I-85 short shaft……………..

249
7-17 Derived p-y curves from dilatometers for I-85 long shaft……….……..

249
7-18 Comparison of load-deflection curves of I-40 short shaft………….…..

250
7-19 Comparison of load-deflection curves of I-40 long shaft…………..…..

250
7-20 Comparison of load-deflection curves of I-85 short shaft………….…..

251
7-21 Comparison of load-deflection curves of I-85 long shaft……..………..

251




1



CHAPTER I
INTRODUCTION

1.1 Statement of the Problem
Drilled shafts socketed into rock are widely used as foundations for bridges and
other important structures. Rock-socketed drilled shafts are also used to stabilize slopes.
The main loads applied to the drilled shafts are axial compressive or uplift loads as well
as lateral loads with accompanying moments. For axially loaded drilled shafts socketed
into rock, numerous research efforts have been conducted in the past, especially for the
determinations of side shear resistance.
However, for laterally loaded drilled shafts, there is a lack of validated, rational
analysis and design methods. It has been a customary practice to adopt the techniques
developed for laterally loaded piles in soil to solve the problem of rock-socketed drilled
shafts under lateral loading (Gabr, 1993). This practice has created erroneous designs and
often leads to excessive socket length. Although there exist several analysis and design
methods specially for rock-socketed drilled shafts under lateral loading, including those
by Carter and Kulhawy (1992), Reese (1997), Zhang et al. (2000), and Gabr et al. (2002);
however, these methods were developed with limited validations against field lateral load
test data.
2
Several researchers (such as DiGioia and Rojas-Gonzalez, 1993; Dykeman and
Valsangkar, 1996; and Cho et al., 2001) have evaluated the methods of Carter and
Kulhawy (1992) and Reese (1997) using their lateral load test data. They concluded that
the two methods provided very unconservative results. Zhang et al. (2000)’s analysis
method has not been evaluated by others due to the complexity of the method and the
needs of a computer program. The assumption of an elastic-perfectly plastic model for
rock masses by Zhang et al. (2000) prohibited its wide application, especially for weak
rock masses which can not be exactly characterized by elastic-perfectly plastic model.
In addition to the need to develop analysis methods for predicting lateral shaft
deflections, there is also a need for the development of methods for estimating lateral
capacity of shafts. Very few methods (Carter and Kulhawy, 1992; and To et al., 2003)
have been developed for estimating lateral capacity of piles in rock. Additionally, several
other researchers, such as Reese (1997) and Zhang et al. (2000), have proposed methods
to predict the ultimate lateral resistance of rock per unit shaft length. However, Carter and
Kulhawy’s (1992) method does not consider the effect of secondary structures of rock
mass; and the method of To et al. (2003) is only suitable for two sets of parallel joints and
rigid piles.
The lack of validated design methods stimulates the need to develop a more rational
design approach for laterally loaded drilled shafts in rock. Additionally, a method for
predicting ultimate capacity of drilled shafts in rock mass needs to be developed.
The p-y method has been widely and successfully used for the design of laterally
loaded drilled shafts in soils for decades. This method is based on a numerical solution of
a physical model based on a beam on Winkler foundation, shown in Fig. 1-1. The
3
structural behavior of the drilled shafts is modeled as a beam, while the soil-shaft
interaction is represented by discrete, non-linear springs.

Figure 1-1 Drilled shaft and soil models of p-y analysis

It is believed that the non-linear p-y approach is the best way to analyze and design
laterally loaded drilled shafts in rock, if a rational and appropriate p-y criterion for rock
masses can be developed. Therefore, this study will focus on the development of p-y
curves for rock masses. There are two ways that one can generate p-y criterion: one is to
use fundamental rock properties along with basic rock-shaft interaction mechanics; and
the second is to derive the p-y curves using in-situ tests, such as pressuremeter or
dilatometer tests.

1.2 Objectives
The objectives of this research are to:
1) Develop a method to predict the lateral capacity of drilled shafts in rock masses.
2) Develop a simple elastic solution for predicting deflections of laterally loaded drilled
shafts in rock in a two-layer soil/rock profile.
H M
p
y
p
y
p
y
p
y
y
z
4
3) Develop a p-y criterion for rock mass.
4) Identify the best field or laboratory test methods for determining rock mass properties
used in the developed p-y criterion. Necessary correlations between rock properties
and the p-y curve parameters will also be established.
5) Perform full-scale field lateral load tests on fully instrumented drilled shafts to obtain
reliable and comprehensive field test data for validating the p-y criterion.
6) Review and recommend a best approach for deriving site specific p-y curves of rock
mass from pressuremeter or dilatometer test data.

1.3 Scope of the Work
The work involved in this study mainly consists of two parts: one is the theoretical
work to develop a p-y criterion for rock and evaluation of an existing approach for
deriving site specific p-y curves using dilatometer tests; the other is to conduct full-scale
field lateral load tests. A lateral capacity prediction method and elastic solution for
estimating shaft deflection under lateral loads are developed as well. Additionally, an
evaluation is carried out to identify the most suitable method for deriving p-y curves
based on test results of a fully instrumented lateral load test. Specifically, the work to be
done are outlined as follows and depicted in Figure 1-2.
A literature review on the design and analysis methods of laterally loaded drilled
shafts and piles in rock is performed. The types of rock of interest and rock classification
systems are identified. To construct a hyperbolic p-y curve, it is necessary to obtain the
initial slope and ultimate rock reaction. The determination of ultimate rock reaction
involves a 3D finite element analysis to identify the possible failure modes of rocks
5
subjected to movement of laterally loaded drilled shafts. Rock strength criteria and side
shear resistance between rock and shaft are carefully identified. To obtain the initial slope
of p-y curves, a correlation between the slope and deformation modulus of rock masses is
established using a parametric study based on a 3D FEM model. Finally, based on the
ultimate rock reaction and initial slope of p-y curve, a hyperbolic p-y criterion for rock
mass is proposed.

Figure 1-2 Flow chart of the work
In addition to the development of a p-y criterion of rock, a method for predicting
lateral capacity of drilled shafts in rock and an elastic solution for estimation of drilled
shaft deflections in a two-layer soil/rock profile under lateral loads are proposed.
Rock Types
Rock
Classification
Rock
Strength
Side
Resistance
Failure
Modes
Ultimate
Resistance
Rock Mass
Modulus
Initial Slope of
p-y
Load Test
Data
P-y
Criterion
P-∆v Curve of
Dilatometer
Site Specific
p-y Curve
Direction I
Direction II
Lateral
Capacity
Elastic
Solution
6
Two field lateral load tests are conducted to facilitate the development and
validation of the p-y criterion for rock mass. An evaluation of various existing methods
for deriving p-y curves from the results of an instrumented lateral load test is carried out
to identify the best method.
Finally, the existing methods for deriving site specific p-y curves for soils from
pressuremeter and dilatometer tests are reviewed and examined with lateral load test
results of rock-socketed drilled shafts. The most suitable method is then recommended
for deriving p-y curves from pressuremeter and dilatometer tests.

1.4 Outlines of the Dissertation
Chapter II presents the literature review on analysis methods of laterally loaded
drilled shafts/piles in rock. Also, typical Ohio bedrock, and rock classification systems
are reviewed.
Chapter III presents the results of two full-scale lateral load tests on instrumented
drilled shafts socketed in rock. The best suitable method for driving p-y curves from the
load test results is identified and it is used to derive the site-specific p-y curves of the two
load test sites. At the end of this chapter, the existing Reese (1997) interim p-y criterion is
evaluated against the results of the two lateral load tests.
The developed method for determination of ultimate resistance of rock mass against
laterally loaded shafts is presented in Chapter IV, in which the failure modes of rock in
response to laterally loaded drilled shafts, rock strength criterion, and side shear
resistance will be elucidated. A 3D finite element modeling technique is established and
validated against a lateral load test. Simulations using 3-D FEM modeling technique
7
enable the identification of failure modes of rock mass. Finally, a method for predicting
lateral capacity of drilled shafts in rock and/or soil is proposed.
An elastic solution for predicting the deflections of laterally loaded drilled shaft in a
two-layer soil/rock system is proposed in Chapter V, in which a correlation equation for
estimating the modulus of rock mass based on modulus of intact rock and a rock
classification system is also developed. Additionally, a correlation equation for
estimating the subgrade reaction modulus of rock mass based on a 3D finite element
parametric study is established as well.
Chapter VI presents the development of a new hyperbolic p-y criterion for rock mass.
A validation of the proposed p-y criterion of rock is carried out by comparing the
predictions against actual lateral load tests results.
Chapter VII provides an evaluation of several existing methods for interpreting the
properties of rock mass and driving p-y curves from pressuremeter or dilatometer tests.
The most suitable method for deriving p-y curves for rock from dilatometer tests is
recommended based on the evaluation findings.
Chapter VIII presents summaries and conclusions of the research work.
8



CHAPTER II
LITERATURE REVIEW

2.1 Analysis Methods of Laterally Loaded Rock-Socketed Drilled Shafts
2.1.1 Introduction
To date, there are few published analysis methods for the lateral response of rock-
socketed drilled shafts. It has been a customary practice to adopt the p-y analysis with p-y
criterion developed for soils to solve the problem of rock-socketed drilled shafts (Gabr,
1993). Currently, two categories of analysis methods for laterally loaded rock-socketed
drilled shafts have been developed. One category treats rock as a continuum mass (Carter
and Kulhawy 1992; and Zhang et al 2000), the other one discretizes the rock mass into a
set of non-linear springs (Reese 1997; and Gabr et al. 2002).
Carter and Kulhawy (1992) proposed a method that treats rock mass as a
homogeneous elastic continuum. Parametric solutions for the load-displacement
relationships were generated by using the finite element technique. However, elastic
continuum model is only good for small loads. Zhang et al. (2000), therefore, developed a
nonlinear continuum approach. The approach adopts and extends the basic idea of Sun
(1994) on laterally loaded piles in soil. The elasto-plastic soil/rock response under lateral
loads and the linearly variation of deformation modulus of soil/rock along depth were
assumed. To consider the yielding, a method based on Hoek-Brown criterion (Hoek and
9
Brown 1980, 1988) was proposed to calculate the ultimate resistance of rock masses. In
practice, however, the rock masses, especially weak rock, show nonlinear and non-
homogeneous properties which can not be fully captured by an elasto-plastic model.
The second category of analysis method, such as p-y method, discretizes the rock
masses into a series of nonlinear springs. The p-y method was extended to the analysis of
single rock-socketed drilled shaft under lateral loading by Reese (1997). An interim p-y
criterion for weak rock was proposed. Thereafter, Gabr et al. (2002) proposed a p-y
criterion for weak rock based on their field test data.
In addition to the above mentioned analysis methods for solving load-deflection
relationship at the drilled shaft head, methods for estimating ultimate rock reaction have
also been proposed. Carter and Kulhawy (1992) presented a method to determine the rock
capacity by using cohesion and friction angle of rock. This method was based on a theory
of expansion of a long cylindrical cavity in an elasto-plastic, dilatant material. The
method requires input of Poisson’s ratio, shear modulus and dilation angle. By assuming
distribution of ultimate rock resistance along the depth of a shaft, the ultimate lateral
capacity of a rock-socketed drilled shaft was obtained by summing the capacity of
compressive resistance and shear resistance between shaft and rock. Cater and Kulhawy
(1992)’s method on rock resistance, treats rock mass as a homogeneous and elasto-plastic
material, without considering secondary structures of rock mass, such as cracks and
fractures.
Reese (1997) considered the secondary structure of rock mass by using a rock
strength reduction factor which can be determined from Rock Quality Designation
(RQD). However, Reese (1997)’s method for estimating ultimate rock reaction per unit
10
length ignored the contribution of shear resistance between shaft and rock. Additionally,
RQD can not fully represent all the secondary rock structures, such as spacing and
condition of discontinuities.
Zhang et al. (2000) proposed a method to estimate the ultimate reaction of rock
masses per unit shaft length using Hoek-Brown rock strength criterion (Hoek and Brown
1988), in which RQD and other secondary rock structures were included. However,
simple rock resistance distribution along the shaft circumference under lateral loads was
assumed (Carter and Kulhawy 1992). It seems that Zhang et al. (2000)’s method for
estimation of lateral capacity of rock-socketed drilled shaft considered most of
characteristics of rock mass; however, the authors did not investigate possible failure
modes of rock mass, especially possible sliding failures along pre-existing joints.
Regarding the sliding failure on joints, To et al. (2003) proposed a method to estimate the
lateral load capacity of drilled shafts in jointed rock. The block theory (Goodman and Shi
1985) was used to identify the failure block, and the static limit equilibrium was used to
obtain the ultimate capacity. The Coulomb failure criterion was utilized to model the
sliding failure on joints.
2.1.2 Brief Description of the Existing Methods
2.1.2.1 Carter and Kulhawy (1992)
Carter and Kulhawy (1992) performed a parametric study to obtain deflection u and
rotation angle θ at the shaft head. The solutions for these two variables were expressed as
a function of effective Young’s modulus E
e
and an equivalent shear modulus G
*
, by using
finite element technique. The drilled shaft is idealized as a cylindrical elastic inclusion
with effective Young’s modulus E
e
, which is defined as
11
64
D
) EI (
E
4
c
e
π
= (2-1)
in which, (EI)
c
= the actual bending rigidity of the shaft; D = diameter of the drilled shaft.
The rock mass is assumed to be a homogeneous, isotropic elastic material. The equivalent
shear modulus is given by
|
.
|

\
|
ν + =
4
3
1 G G
r
*
(2-2)
where
) 1 ( 2
E
G
r
r
ν +
= (2-3)
in which E
r
= Young’s modulus of rock, and ν = Poisson’s ratio of rock.
From the finite element analysis performed by Carter and Kulhawy (1992), it was
found that u and θ are largely dependent on the ratio of E
e
/ G
*
and the ratio of the shaft
socket length to the diameter L/D. Two categories of shafts, flexible and rigid, were
classified by the authors. A flexible pile is one in which the following condition meets:
7 / 2
*
e
G
E
D
L
|
.
|

\
|
≥ (2-4)
For a flexible drilled shaft, ground-line deflection u and rotation θ induced by the
lateral load H and the moment M at shaft top are calculated from the following equations:
7 / 3
*
e
2 *
7 / 1
*
e
*
G
E
D G
M
08 . 1
G
E
D G
H
5 . 0 u
− −
|
.
|

\
|
|
.
|

\
|
+ |
.
|

\
|
|
.
|

\
|
= (2-5)
7 / 5
*
e
3 *
7 / 3
*
e
2 *
G
E
D G
M
4 . 6
G
E
D G
H
08 . 1
− −
|
.
|

\
|
|
.
|

\
|
+ |
.
|

\
|
|
.
|

\
|
= θ (2-6)

12
A drilled shaft is considered to be rigid when
2 / 1
*
e
G
E
05 . 0
D
L
|
.
|

\
|
≤ (2-7)
For a rigid drilled shaft, ground-line deflection u and rotation θ are calculated from
the following equations:
8 / 7
2 *
3 / 1
*
D
L 2
D G
M
3 . 0
D
L 2
D G
H
4 . 0 u
− −
|
.
|

\
|
|
.
|

\
|
+ |
.
|

\
|
|
.
|

\
|
= (2-8)
3 / 5
3 *
8 / 7
2 *
D
L 2
D G
M
8 . 0
D
L 2
D G
H
3 . 0
− −
|
.
|

\
|
|
.
|

\
|
+ |
.
|

\
|
|
.
|

\
|
= θ (2-9)
For the drilled shafts having intermediate rigidity, the authors suggested that the
displacements be taken as 1.25 times the larger displacements of those calculated values
by treating the shaft as a flexible or a rigid shaft.
For ultimate capacity of rock-socketed drilled shafts, Carter and Kulhawy (1992)
proposed a solution in which they suggested that the lateral resistance were derived from
side shear τ between shaft and rock, and frontal compressive strength of rock. The
authors further suggested that the magnitude of this shear was equal to that produced in
axial loading. The assumed distribution of ultimate resistance along the shaft is shown in
Fig. 2-1, from which one can see that lateral resistance is equal to τ
max
D at the surface of
the rock and is increasing linearly with depth to a magnitude of (P
L
+ τ
max
)D at a depth of
3D. Below this depth, the ultimate resistance remains constant with depth. P
L
is the limit
stress developed in rock, which can be calculated according to the expansion theory of a
long cylindrical cavity in an elasto-plastic, cohesive-frictional, dilatant material (Carter et
13
al. 1986). The following equations were derived by Carter and Kulhawy to determine the
lateral capacity of rock-socketed drilled shafts, H
u
:
L D
6
L P
H
max
L
u
|
.
|

\
|
τ + = for L<3D (2-10)
D ) D 3 L )( P ( D 3
2
P
H
max L
2
max
L
u
− τ + + |
.
|

\
|
τ + = for L>3D (2-11)
where
max
τ = maximum unit side resistance; D = diameter of the drilled shaft; and L =
length of drilled shafts embedded in rock.

Figure 2-1 Distribution of ultimate lateral force per unit length (after Carter and Kulhawy
1992)

2.1.2.2 Reese (1997)
The p-y method for the analysis of drilled shafts in soils under lateral loading was
extended to the analysis of rock-socketed drilled shafts by Reese (1997). In order to
characterize the rock response under lateral loading, an interim p-y criterion for weak
rock was suggested. Due to the lack of adequate test data, the term “interim” was applied
Surface
of rock H
ult

max
3D
L
D D(p
u

max
)
14
to this p-y criterion. With this interim p-y criterion, COM624P or LPILE can be run to
obtain the lateral response of rock-socketed drilled shafts.
The ultimate reaction, p
u
(F/L), of rock was given by
)
D
z
4 . 1 1 ( D p
r
ci r u
+ σ α = for 0 ≤ z
r
≤ 3D (2-12)
D 2 . 5 p
ci r u
σ α = for z
r
≥ 3D (2-13)
where σ
ci
= uniaxial compressive strength of intact rock; α
r
= strength reduction factor,
which is used to account for fracturing of rock mass; D = diameter of the drilled shaft;
and z
r
= depth below rock surface. The value of α
r
is assumed to be 1/3 for RQD of 100%
and it increases linearly to unity at a RQD of zero.
The slope of initial portion of p-y curves was given by
K
ir
= k
ir
E
m
(2-13)
where K
ir
= initial tangent to p-y curve; E
m
= deformation modulus of rock masses, which
may be obtained from a pressuremeter or dilatometer test; and k
ir
= dimensionless
constant. The expressions for k
ir
, derived by correlation with experimental data, are as
follows.
)
D 3
z 400
100 ( k
r
ir
+ = for 0 ≤ z
r
≤ 3D (2-14)
k
ir
= 500 for z
r
≥ 3D (2-15)
A complete description of the interim p-y criterion may be summarized as follows.
First segment: p = K
ir
y; y ≤ y
A
(2-16)
Second segment:
25 . 0
rm
u
y
y
2
p
p
|
|
.
|

\
|
= ; y ≥ y
A
and p ≤ p
u
(2-17)
Third segment: = p
u
; p ≥ p
u
(2-18)
15
where
y
rm
= k
rm
D (2-19)
333 . 1
ir
25 . 0
rm
u
A
K ) y ( 2
p
y
(
¸
(

¸

= (2-20)
in which, k
rm
= strain at 50% of ultimate load, ranging from 0.0005 to 0.00005.
2.1.2.3 Zhang et al. (2000)
Zhang et al. (2000) proposed a nonlinear continuum method to predict the load-
displacement response of rock-socketed drilled shafts under lateral loads by treating
soil/rock as an elasto-plastic material. The approach was extended from the basic idea of
Sun (1994) on laterally loaded piles in soil.
The model of rock-socketed drilled shafts under lateral loading is shown in Fig. 2-2.
The deformation modulus of soils varies linearly from E
s1
to E
s2
. Similarly, the
deformation modulus of rock mass varies linearly from E
m1
to E
m2
at the tip of shaft and
stays constant below the shaft tip. By minimizing the energy of rock-shaft system with
respect to displacements, the following governing equations were obtained:
0 u
L
z
) 1 ( k
dz
du
L
z
) 1 (
dz
d
t 2
dz
u d
I E
s
s
s s s
s
s
s s s
4
s
4
p p
=
(
¸
(

¸

η − + η +
)
`
¹
¹
´
¦
(
¸
(

¸

η − + η − (0 ≤ z ≤ L
s
) (2-21a)
0 u
L
L z
) 1 ( k
dz
du
L
L z
) 1 (
dz
d
t 2
dz
u d
I E
m
m
s
m m m
m
m
s
m m m
4
m
4
p p
=
(
¸
(

¸

η − + η +
)
`
¹
¹
´
¦
(
¸
(

¸

η − + η −
( L
s
≤ z ≤ L) (2-21b)
where u
s
and u
m
= displacement components of the shaft in the soil and in the rock mass,
respectively; E
p
I
p
= flexural rigidity of the shaft; z = depth starting from ground line; L
s
=
shaft length embedded in soils; L
m
= shaft length embedded in rock masses; and
16
2 s
1 s
s
E
E
= η (2-22a)
2 m
1 m
m
E
E
= η (2-22b)
) 1 ( 2
m R E
t
s
1
2
2 s
s
ν +
π
= (2-22c)
) 2 1 )( 1 ( 2
m E ) 4 3 (
k
s s
2 2 s s
s
ν − ν +
ν − π
= (2-22d)
) 1 ( 2
m R E
t
m
1
2
2 m
m
ν +
π
= (2-22e)
) 2 1 )( 1 ( 2
m E ) 4 3 (
k
m m
2 2 m m
m
ν − ν +
ν − π
= (2-22f)
in which ν
s
and ν
m
= Poisson’s ratio of soils and rock masses, respectively; m
1
and m
2
=
parameters describing the behavior of the elastic foundations.

Figure 2-2 Rock-shaft model (a) Shaft and soil/rock mass system; (b) Coordinate system
and displacement components; (c) Shear force V(z) and moment M(z) acting on shaft at z
(Zhang et al. 2000)


17
The shear force V(z) acting on the shaft, shown in Fig. 2-2(c), can be obtained as
|
.
|

\
|
(
¸
(

¸

η − + η −
|
|
.
|

\
|
=
dz
du
L
z
) 1 ( t 2
dz
u d
I E ) z ( V
s
s
s s s
3
s
3
p p
(0 ≤ z ≤ L
s
) (2-23a)
|
.
|

\
|
(
¸
(

¸

η − + η −
|
|
.
|

\
|
=
dz
du
L
L z
) 1 ( t 2
dz
u d
I E ) z ( V
m
m
s
m m m
3
m
3
p p
( L
s
≤ z ≤ L) (2-23b)
and the bending moment M(z) acting on the shaft is be given by
|
|
.
|

\
|
=
2
s
2
p p
dz
u d
I E ) z ( M (0 ≤ z ≤ L
s
) (2-24a)
|
|
.
|

\
|
=
2
m
2
p p
dz
u d
I E ) z ( M (L
s
≤ z ≤ L) (2-24b)
The governing differential equations and the shear force and bending moment can be
solved using classical finite difference method and an iterative process. The above
process considers the soil/rock to be elastic. Elastic-perfectly plastic stress-strain
relationship, therefore, was proposed to consider the yielding of the soils or rock masses.
The method for considering the yielding of soil or rock mass consists of several steps
depicted in Fig. 2-3. Firstly, for the applied lateral load H and the moment M, the shaft is
analyzed by using the above elastic solutions. Secondly, the lateral reaction force p at
certain depth is computed and compared to the ultimate resistance p
u
at that depth. If p >
p
u
, take the depth z as the yielding depth z
y
. Thirdly, treat the unyielded portion of shaft
as a new shaft, and analyze it by using the elastic solution while ignoring the effect of the
yielded portion of shaft. Fourthly, repeat steps two and three until no further yielding of
soil or rock occurs. Finally, the final results can be obtained by considering the two parts
of the shaft separately. The portion of shaft in yielded soil and/or rock mass is analyzed
18
as a beam with distributed load p
u
acting on it. The other part of shaft in the unyielded
soil and/or rock mass is analyzed by using the elastic solution.
To compute the ultimate resistance p
u
of soil, Zhang et al. (2000) suggested two
existing methods for clay and sand, respectively. For clay, the equation proposed by
Matlock (1970), and Reese and Welch (1975) was suggested.
p
u
= N
p
C
u
D (2-25)
where C
u
= undrained shear strength of soil; D = diameter of drilled shafts; and
9 z
R 2
J
z
c
3 N
u
'
p
≤ +
γ
+ = (2-26)
in which
'
γ = average effective unit weight of soil above depth z; R = shaft radius; and J =
coefficient ranging from 0.25 to 0.5.

Figure 2-3 Consideration of yielding of soil and/or rock mass by decomposition of
loading (after Zhang et al. 2000)

For sand, Zhang et al. (2000) suggested the method of Fleming et al. (1992) as
follows.
zD K p
' 2
p u
γ = (2-27)
19
where K
p
= tan
2
(45
°
+
'
φ /2) = Rankine passive coefficient, in which
'
φ = effective internal
friction angle.
To compute the ultimate resistance p
u
of rock mass, Zhang et al. (2000) proposed to
utilize the assumed resistance distribution (Carter and Kulhawy 1992), shown in Fig. 2-4,
and Hoek-Brown rock strength criterion (Hoek and Brown 1988). The assumption for
resistance distribution is that the total resistance of rock mass consists of two parts: the
side resistance and the front normal resistance. The ultimate resistance p
u
can be
calculated by
p
u
= (p
L

max
)D (2-28)
where D = diameter of the drilled shaft; τ
max
= maximum shearing resistance along the
sides of the shaft; and p
L
= normal limit resistance. τ
max
was assumed to be the same as
the maximum side resistance under axial loading and can be given by
5 . 0
ci max
) ( 20 . 0 σ = τ (MPa) for smooth socket (2-29a)
5 . 0
ci max
) ( 80 . 0 σ = τ (MPa) for rough socket (2-29b)
where σ
ci
= unconfined compressive strength of the intact rock (MPa).

Figure 2-4 Components of rock mass resistance (Zhang et al. 2000)
20
The strength criterion for rock mass developed by Hoek and Brown (1980, 1988)
was adopted to determine the normal limit stress p
L
. The Hoek-Brown criterion, which is
suitable for intact rock and rock mass, can be given by
a
ci
'
3
b ci
'
3
'
1
s m
|
|
.
|

\
|
+
σ
σ
σ + σ = σ (2-30)
where σ
ci
= uniaxial compressive strength of the intact rock;
'
1
σ and
'
3
σ = the major and
minor effective principal stresses, respectively; and m
b
, s, and a = constants depending on
the characteristics of the rock.
For intact rock, m
b
= m
i
, a constant depending on rock type; s = 1; and a = 0.5. For
rock mass, the values of m
b
, s, and a can be estimated by correlations with Geological
Strength Index (GSI) (Hoek, 1994). In addition to GSI, Rock Mass Rating (RMR) of
Bieniawski can also be used to determine the constants m
b
, s, and a (Hoek and Brown
1997). With Hoek-Brown’s strength criterion, the normal limit stress p
L
, which is the
major principal effective stress
'
1
σ , can be obtained by assuming that the minor principal
effective stress is the effective overburden pressure
'
γ z.
2.1.2.4 Gabr et al. (2002)
Gabr et al. (2002) proposed a hyperbolic p-y criterion for weak rock based on field
tests on small diameter drilled shafts socketed in weak rock. The following procedure can
be used to construct a p-y curve according to Gabr et al. (2002).
Step 1: Calculation of Coefficient of Subgrade Reaction
The coefficient of subgrade reaction can be calculated as follows (Vesic, 1961):
21
12 / 1
p p
4
m
2
m
h
I E
D E
) 1 ( D
E 65 . 0
n
(
(
¸
(

¸

ν −
= (2-31)
where D is the diameter of a drilled shaft, ν is Poisson’s ratio of rock mass, and GSI is
Geological Strength Index.
Step 2: Calculation of Flexibility Factor
A flexibility factor, K
R
, is computed as follows (Poulos and Davis, 1972):
4
m
p p
R
L E
I E
K = (2-32)
where, E
p
is modulus of elasticity of shaft, I
p
is the moment of inertia of shaft, L is the
embedment length of shaft.
Step 3: Calculation of Point of Rotation
The following equation is used to define the turning point as a function of the
embedded shaft length:
L ) K log 18 . 0 1 ( T
R 0
+ = (2-33)
where, T
0
is turning point.
Step 4: Calculation of I
T
Number
I
T
= -28 - 383 log(T
0
/L) I
T
≥ 1 (2-34)
Step 5: Calculation of the Subgrade Reaction
k
h
= n
h
D (0 ≤ z ≤ T
0
) (2-35)
k
h
= I
T
n
h
D (T
0
< z ≤ L) (2-36)
Step 6: Calculation of Ultimate Resistance of Rock Mass P
u

22
The Eq. (2-28) proposed by Zhang et al. (2000) was employed to calculate the
ultimate resistance of rock as presented in Section 2.1.2.3. Smooth condition was
assumed for all the cases when the side shear resistance is concerned.
Step 7: Construction of the P-y Curve
u h
p
y
k
1
y
p
+
= (2-37)

2.1.2.5 To, Ernst, and Einstein (2003)
For the drilled shafts socketed into jointed rock, To et al. (2003) assumed a wedge
type block failure and Coulomb failure criterion to obtain the lateral capacity of drilled
shafts. Goodman and Shi (1985)’s block theory was used to determine the possible failure
block for two sets jointed rock mass with the help of AutoCAD or Excel. Due to the
complexity of the entire process to obtain the failure block, no details about the block
theory will be described here. The assumed mechanisms of sliding failure along the joint
plane and tensile failure on the rock mass, preventing the movement of the wedge, are
depicted in Fig. 2-5, where W = weight of the wedge; P = axial load of shaft; F = lateral
force; T = tensile force due to the fracture of Category II blocks, which is defined as a
block that is not removable, but becomes removable if it breaks due to the lateral force
exerted by the pier; N
1
= normal force on joint; and R
1
= tangential force on joint.
23

Figure 2-5 Typical forces on wedge
Normal force N and tangential force R can be related by the Coulomb failure
criterion as follow:
ϕ σ + = τ tan c (2-38)
where τ = shear stress; c = cohesion; σ = normal stress; φ = friction angle.
Static limit equilibrium relation between the forces on the wedge (Fig. 2-5) was used
to solve for ultimate lateral force F.
2.1.3 Comments on the Existing Analysis Methods
2.1.3.1 Carter and Kulhawy (1992)
Carter and Kulhawy (1992) provides solutions for the lateral load-deflection relation
at shaft head as well as shaft lateral capacity. For load-deflection prediction, Carter and
Kulhawy assumed rock mass as an elastic material, which implies that the solution is
only applicable to small loads. The solution for ultimate lateral capacity needs
verification. One of the drawbacks of the solution is the requirement of numerous rock
24
deformation and strength parameters, such as shear modulus, cohesion, friction angle,
and dilation angle.
DiGioia and Rojas-Gonzalez (1993) evaluated this method by using their field
lateral load test on drilled shafts socketed into rock mass. They found a reasonable
agreement between the measured and predicted displacements for these foundations at
low load levels (20-30% capacity). However, this method gave predictions that were
stiffer than observed at higher load levels. Additionally, Dykeman and Valsangkar (1996)
conducted a centrifuge test on eight model socketed shafts and used the test results to
evaluate Carter and Kulhawy’s method. The comparison showed that Carter and
Kulhawy (1992)’s method tend to overestimate the ultimate lateral capacity by a factor of
two, while it predicted smaller deflection at shaft head than measured deflection at a
given load level. In addition to their own test data, Dykeman and Valsangkar (1996)
evaluated Carter and Kulhawy’s method by using Frantzen and Stratten (1987)’s field test
data. Similar comparison results were found for the predicted deflection at shaft head.
2.1.3.2 Zhang et al. (2000)
The yielding of rock mass was considered in Zhang et al. (2000) method, however,
the elasto-perfectly plastic stress-strain relationship can not fully represent the nonlinear
behavior of rock masses. As shown in Fig. 2-6, the actual nonlinear behavior may already
appear before the stress reaches the peak. For weak rock mass, brittle post-failure may
not occur. However, for good quality rock mass, which is likely to behave as a brittle
material (Hoek and Brown 1997), the stress will drop after it reaches the peak, as shown
in Fig. 2-6. Therefore, despite that yielding of rock was considered in this method, the
25
behavior of different types of rock mass was not fully represented as an elasto-perfectly
plastic model.

Figure 2-6 The elastoplastic and brittle behavior of rock mass
2.1.3.3 To et al. (2003)
The determination of removable wedge required by this method is a very tedious
procedure which involves the use of AutoCAD and EXCEL. Furthermore, only two sets
of joints can be considered and the joints in each set should be parallel. The failure mode
is restricted to failure at the top portion of rock mass with a free surface at the ground-
line.
2.1.3.4 Existing P-y Criteria for Rock
Reese (1997)’s interim rock p-y criterion was not well calibrated due to inadequate
test data. The failure modes of rock mass were not well defined; and an estimation of
ultimate lateral capacity did not include the effect of friction between rock mass and shaft.
Determination of such parameters as constant k
rm
appears to be empirical. Cho et al.
(2001) conducted a lateral load test on rock-socketed drilled shafts. Two drilled shafts, 30
inch in diameter and 10 feet to 13 feet of socket were laterally loaded. Reese (1997)’s
interim rock p-y criterion for weak rock was used to construct p-y curves. It was found
strain
stress
Linear elastic–
perfectly plastic
Brittle
26
that the interim p-y curves underestimated the deflections of shafts when comparing with
the measured values.
Although the interim rock p-y criterion was not well established, the p-y method is
still a promising method for rock-socketed drilled shaft under lateral loads. Gabr (1993)
evaluated two field tests on rock-socketed drilled shafts under lateral loads presented by
Carter and Kulhawy (1992) by using p-y analysis and the p-y criterion for stiff clay
(Reese et al. 1975). The analysis results showed that p-y approach is analytically
attractive because it can approximately model the nonlinearity in the load-displacement
response. Gabr et al. (2002) p-y criterion is a most recently proposed p-y criterion for
weak rock. However, it has not been further validated with other load tests. The equation
for estimating modulus of subgrade reaction was based on Vesic (1961)’s equation for
beam on elastic foundation. This is not same as the case of drilled shafts embedded in
rock where shaft-rock interaction is more complicated.

2.2 Bedrocks of Ohio
2.2.1 The Distribution of Rocks in Ohio
Six of the groups or systems in the historical classification of rocks are present in the
outcrops of Ohio. The geological map, shown in Fig. 2-7, provides the distribution of
those rock systems across the state. The six systems of rock are formed in different
periods in the history of earth. They are ordered as, from the oldest to youngest,
Ordovician, Silurian, Devonian, Mississippian, Pennsylvanian, and Permian (Lamborn et
al., 1938). Usually, the older rock systems are bedded under the younger rock systems.
27
The cross section through the rocks of central Ohio from the Indiana-Ohio border to the
Ohio River is shown in Fig. 2-8.
The bedrocks exposed at the surface in Ohio are all of sedimentary types formed
from unconsolidated sediments deposited in marine, brackish, or fresh waters (Lamborn
et al. 1938). All the common varieties of the sedimentary series can be found, such as
limestone, shale, sandstone, dolomite, and conglomerate, as well as many thin beds of
coal, clay, and iron ore. In the western half of the State, the bedrock is predominantly
limestone and dolomite with minor amounts of calcareous shale. The calcareous shales
are most abundant in the southwestern part of the State, where thin limestone
interstratified with shale is the usual mode of occurrence. In the northwestern corner of
the State, non-calcareous shale is the bedrock lying immediately below the glacial drift.
In contrast to the dominantly calcareous nature of the beds exposed in the western
half of Ohio, those in the eastern half are mainly of the clastic or fragmental varieties.
Although in the Pennsylvanian and Permian systems of eastern Ohio, thick beds of shale,
shaly sandstone, and sandstone are the rule, thin beds of coal, clay, limestone, iron ore,
and conglomerate are present. Lens-like bodies of sandstone are widely distributed but
lateral gradation to shale is of common occurrence. Shales of varying character are
widely distributed both horizontally and vertically on the outcrop (Lamborn et al. 1938).
For example, the Conemaugh series of Pennsylvanian System, located in the southeastern
of Ohio, ranges from 355 feet to 545 feet and crops out across the center of southeastern
Ohio in a belt 10 to 20 miles in width. Bedded marine shales and some thin marine
limestones are presented in the lower part of this series, while the upper part includes
abundant red calcareous claystones (Fisher et al. 1968). The generalized column of
28
bedrock in Ohio is presented in Figure 2-9, which was complied by Hull (1990) and
revised by Larsen (2000).

Figure 2-7 Geological map of Ohio, showing the pattern of surface rocks across the state

Figure 2-8 Cross section through the rocks of central Ohio from the Indiana-Ohio border
to the Ohio River (taken from Feldmann et al., 1996)
29

Figure 2-9 Generalized column of bedrock units in Ohio
30
2.2.2 The Shales in Ohio
2.2.2.1 Introduction
Shale, highly distributed in eastern, southwestern, and northwestern of Ohio, is a
variety of sedimentary rock resulted from the consolidation of more or less thinly
laminated or bedded silts and clays (Lamborn et al., 1938). In general, shale is less
weather resistant than most other varieties of sedimentary rocks.
The composition of shales is variable as it is assemblage of many mineral substances
of different chemical composition. The chief mineral constituents found in shale are
hydrated aluminum silicates of varying composition, quartz, rutile, apatite, calcite,
dolomite, and iron oxides. Organic material is generally present in small amounts. The
hydrated silicates and quartz make up a large part of the average shale and in themselves
produce a rock which is white, green, or greenish gray in color. Dark bluish gray, dark
brown and black colors are usually found when finely divided organic material is
presented in the shale. Where iron oxide occurs in the absence of much organic matter,
the shale may be buff, yellowish brown, brown, or red in color. Green or greenish gray
colors are characteristic of unweathered shale rich in either iron carbonates or chloritic
material and poor in organic matter.
Shale beds are associated with sandstone, limestone, coal, and claystone. Shale may
gradually become shaly sandstone and finally sandstone as quartz increases in percentage
and size of grain. With an increase in the percentage of calcium carbonate, shale transits
to limestone. As carbonaceous material becomes greater in amount, shale is transited to
bone coal and coal; and as the fissility or shaly structure disappears, siltstone is produced
from shale.
31
2.2.2.2 Shale Classification System
Typical shales possess fissility or the property of splitting with comparative ease
along the planes of lamination or bedding. Hand specimens of typical shale are
distinguished from sandstones and conglomerates by the smaller grain size and by the
fissility; from coal, clay and mudstone by its fissility; and from limestone and dolomite
by its fissility and low carbonate content.
Various shale classification systems have been proposed by different investigators
from engineering view point, such as Wood and Deo (1975). Wood and Deo (1975)
proposed a shale classification system for embankments. Four simple tests were
suggested to characterize shales for embankment use: one cycle slaking in water, slake-
durability on an initially dry sample, slake-durability on a soaked sample, and a modified
sodium sulfate soundness test. The system classifies shales into four groups: rock-like
shales, intermediate-1 shales, intermediate-2 shales, and soil-like shales, as shown in
Figure 2-10. In Figure 2-10, (I
d
)
d
is the slake durability index of dry sample; (I
d
)
s
is the
slake durability index of soaked sample; and I
s
is the soundness index.
The classification of shales for deep foundations has not been established yet.
However, the engineering performance of shales for deep foundation design, such as
reduced shear resistance, swelling, reduced bearing capacity, could be correlated to the
durability and other properties of shales. If a series of tests on shale strength and
durability can be organized and conducted, the correlation between shale-shaft side
resistance reduction due to water intrusion and the durability index can be established
(not to be done in this study). For instance, Richardson and Wiles (1990) proposed a
correlation between the loss of shear strength and results of simple tests (e.g., sieved
32
slake durability index, natural moisture content, and change in point-load strength) for
embankment applications. The effect of slake during drilling or coring of shales should
also be taken into account. The cored shales can be used for slake-durability test.

Figure 2-10 Classification of shales for embankment construction (Wood and Deo, 1975)
2.2.2.3 Engineering Properties of Ohio Shales
A compilation of field and lab test results on Ohio shales is presented in Table 2-1,
where the unconfined compressive strength, σ
ci
, and RQD of shales are given.
Hawk and Ko (1980) examined the orthotropic nature of two shales and concluded
that the properties of both shales are represented well by a transversely isotropic model,
although an isotropic model is also acceptable. Additionally, Sargand and Hazen (1987)
Slaking test in
water in one cycle
Slake durability test
on dry samples
Slake durability test
on soaked samples
Modified
Soundness test
Slakes
completely
Does not
Slakes
completely
(I
d
)
d
<90 (I
d
)
d
>90
(I
d
)
s
< 75 90 > (I
d
)
s
> 75 (I
d
)
s
> 90
I
s
< 70 98 > I
s
> 90 I
s
> 98 90>I
s
>70
Soil-like shales Intermediate -2
shales
Intermediate -1
shales
Rock-like shales
33
conducted a series of triaxial tests on Ohio grey shales and concluded that the transverse
isotropy is an appropriate model to simulate the stress-strain relations of these shales.
Table 2-1 Engineering Properties of Rocks in Ohio
County Project
Name
Depth
(ft)
Rock Type RQD
(%)
σ
ci

(psi)
Erie ERI60
(1)
26 Medium-hard to hard shale 100 7320
Erie ERI60
(1)
15 Medium-hard to hard shale 100 6604
Meigs Pomery
Mason
(2)
29-45 Shale (50%) with
interbedded siltstone (50%)
60 905
Meigs Pomery
Mason
(2)

75-88 Siltstone, hard 56 4611
Noble Sargand and
Hazen
(3)

Grey shale 2100-
3125
Chesapeake Sargand and
Hazen
(3)

Grey shale 1400-
2800
(1): Nusairat, J., Engel, R. L., and Liang, R. Y. (2002).
(2): Data obtained from E.L. Robinson Engineering of Ohio Co. and University of Akron.
(3): Sargand and Hazen (1987). The strengths were obtained with 100 to 200 psi of
confining pressure.

2.2.2.4 Effect of Water on Shales
Water plays an important role on engineering properties of shales, such as shear
strength and durability. The shearing strength of shale is quite low, deriving from small
internal friction of microscopic platy grains as well as cohesion between particles. This
cohesion is usually dependent upon a skin of absorbed, oriented water molecules and
sometimes also upon surface tension between the water and any air that may exist from
time to time in the pores of the coarse phases of the sediment. However, excess water
under pressure may reduce the cohesion, either by altering the molecular film of water or
by destroying the surface tension as all air spaces become filled with water. Also, pore-
water pressure in the silty phases of the rock may spread the grains apart so that internal
friction is further diminished. Finally, certain clay minerals have the ability to absorb
34
water and swell, thus exerting an upward pressure that may reduce shear resistance within
a claystone. (Fisher et al., 1968)
Fisher et al. (1968) conducted slake tests, in which equal size cubic shales were
immersed in water and observed during a period of three and one-half days, on various
Ohio shales cored from southeastern Ohio. Twenty three red shale samples, eight gray
shale samples, and eleven green shales were tested. With three exceptions, the red shale
samples deteriorated completely; almost all slaked within one hour into chunks one to
two mm in diameter, and two-thirds eventually became oozes covering the bottoms of the
beakers. Three red shales became oozes within only ten minutes, and almost a dozen
flaked badly on the bedding surfaces in the same span of time.
The eight samples of gray shales, under the same treatment, were essentially
unaffected. The eleven green shales varied widely in behavior, some slaking within five
hours, a few slaking slightly, but most not deteriorating at all.
2.2.3 Limestone and Dolomite
The dolomites and limestones of western Ohio form the bedrock in approximately
20,381 square miles or in nearly one-half of the total area of the State (Stout, 1941). A
limestone is a rock composed essentially of calcium carbonate, CaCO
3
, with minor
quantities of other basic carbonates, sulphates, sulphides, phosphates, and silicates, and of
siliceous and clay-like materials. In general, they range from 80% to 98% calcium
carbonate. The limestones are deposited in shallow or comparatively shallow water and
are the result of precipitation either by direct exchange or by the agency of organic life.
Dolomite is the prominent carbonate rock throughout the stratigraphic section of
western Ohio. Such strata constitute more than 60% of the whole. Dolomite may be
35
defined as a rock composed largely of magnesium carbonate combined directly with
calcium carbonate in the molecular ratio of one to one. Its chemical composition may be
written as MgCa(CO
3
)
2
. The dolomites marine in origin, were deposited in shallow or
comparatively shallow water, and in the main are direct precipitates. Most certainly, they
were laid down in the form we now find them and are not secondary, that is, deposited as
limestone or limy sediment, then changed to dolomite (Stout, 1941).

2.3 Mechanical Characteristics of Rock and Its Classifications
2.3.1 The types of rocks
There are three basic types of rocks: igneous, sedimentary, and metamorphic. Table
2-2 presents a simplified rock classification system and common rock types.
2.3.2 Features for Rock Characterization
The difficulties in making predictions of the engineering responses of rocks and rock
mass derive largely from their discontinuous and variable nature. In fact, rock is
distinguished from other engineering materials by the presence of inherent discontinuities,
which may control its mechanical behavior. The response of intact rock material itself
may be complex and difficult to describe theoretically because the rock consists of an
aggregation of grains of material having quite different physical properties. It may
contain inter- and intra-granular micro-cracks and may have anisotropic and/or nonlinear
mechanical properties (Brown, 1993).
Qualification and quantification of in-situ rock masses are some of the most
important aspects of site characterization for design of foundations. These characteristics
generally indicate the strength, deformability, and stability of the rock masses. For
36
economical reason, it is not feasible to fully measure the characteristics of a complex
rock masses. The key factors influencing the rock mass behavior are listed in Table 2-3.
Table 2-2 Simplified Rock Classification

Discontinuities, one of the major effects, may influence the engineering response of
rock masses in a variety of ways as enumerated below:
Major division Secondary divisions Rock types
Extrusive
Volcanic explosion debris (fragmental)
Lava flows and hot siliceous clouds
Lava flows (fine-grained texture)
Tuff (lithified ash) and volcanic breccia
Obsidian (glass), pumice, and scoria
Basalt, andesite, and rhyolite
C
o
m
m
o
n

i
g
n
e
o
u
s

r
o
c
k
s

Intrusive
Dark minerals dominant
Intermediate (25-50% dark minerals)
Light color (quartz and feldspar)
Gabbro
Diorite
Granite
Major division
Texture (grain size) or
chemical composition
Rock types
Clastic rocks
Grain size large than 2 mm (pebbles,
gravel, cobbles, and boulders)
Sand-size grains, 0.0062 to 2 mm
Silt-size grain, 0.004 to 0.062 mm
Clay-size grains, less than 0.004 mm
Conglomerate (rounded cobbles) or
breccia (angular rock fragments)
Sandstone
Siltstone
Claystone and shale
C
o
m
m
o
n

s
e
d
i
m
e
n
t
a
r
y

r
o
c
k
s

Chemical
and
organic rock
Carbonate minerals (e.g., calcite)
Halite minerals
Sulfate minerals
Iron-rich minerals Siliceous minerals
Organic products
Limestone
Rock salt
Gypsum
Hematite
Chert
Coal
Major division Structure (foliated or massive) Rock types
Coarse
crystalline
Foliated
Massive
Gneiss
Metaquartzite
Medium
crystalline
Foliated
Massive
Schist
Marble, quartzite, serpentine, soapstone
C
o
m
m
o
n

m
e
t
a
m
o
r
p
h
i
c

r
o
c
k
s

Fine to
microscopic
Foliated
Massive
Phyllite, slate
Hornfels, anthracite coal
37
• The provision of planes of low shear strength along which slip might occur
• Reducing the overall shear and tensile strengths of the rock masses
• Rendering the overall mechanical response of the rock masses in the sense that
individual blocks may be free to rotate or to translate with associated slip and/or
separation at block interfaces
• Introducing a wide range of potential failure mechanisms (such as unraveling,
toppling, slip or the gravity fall of blocks or wedges)
• Influencing the stress distribution within the rock masses mainly because of their low
stiffness and strengths
• Attenuating, reflecting and refracting stress waves arising from blasting
• Controlling to a large extent the fragmentation achieved by excavation processes
• Providing major conduits for the flow of fluids through most masses due to their
permeability is orders of magnitude higher than that of intact rocks
Table 2-3 Factors influencing Rock Mass Behavior
Characteristics of
Rock
Specimen and
Environmental Conditions
State of Stress
or Strain
Method of Loading
Rock material
structure:
lithology, cracks

Rock mass
structure:
discontinuities

Properties:
mechanical,
physical
Moisture content,
temperature, and pore-
pressure

Groundwater conditions
and chemical environment

Specimen size and shape
Magnitude of
applied stress
or strain

Distribution
of stress or
strain
Type of loading:
compressive, shear

Rate of loading:
static, dynamic,
impact

Pattern of loading:
constant load,
repetitive (fatigue),
pulse



38
2.3.3 Rock Mass Classifications
In order to evaluate influence of discontinuities in rock masses for a given rock
engineering project, it is necessary that the discontinuities in rock masses should be
characterized. The International Society for Rock Mechanics selects ten parameters as
being of the primary importance for quantitative description of discontinuities in rock
masses (Brown, 1981):
1) Orientation: attitude of a discontinuity in space, described by the dip direction and dip
of the line of steepest declination in plane of discontinuity
2) Spacing: perpendicular distance between adjacent discontinuities, normally refers to
the mean or modal spacing of a set of joints
3) Persistence: discontinuity trace length as observed in an exposure which may give a
crude measure of the area extent or penetration length of a discontinuity. Termination
in solid rock or against other discontinuities reduces the persistence
4) Roughness: Inherent surface roughness and waviness relative to the mean plane of a
discontinuity. Both roughness and waviness contribute to the shear strength. Large
scale waviness may also alter the dip locally
5) Wall strength: equivalent compressive strength of adjacent rock walls of discontinuity
may be lower than rock block strength due to weathering or alteration of the walls.
6) Aperture: perpendicular distance between adjacent rock walls of a discontinuity, in
which the intervening space is air or water filled
7) Filling: material that separates the adjacent rock wall of a discontinuity and that is
usually weaker than the parent rock. Typical filling materials are sand, silt, clay,
39
breccia, gouge and mylonite. Also includes thin mineral coatings and healed
discontinuities, e.g. quartz and calcite veins
8) Seepage: water flow and free moisture visible in individual discontinuities or in the
rock mass as a whole
9) Number of sets: the number of joints comprising the intersecting joint system. The
rock mass may be further divided by individual discontinuities
10) Block size: rock block dimensions resulting from the mutual orientation of
intersection joint sets, and resulting from the spacing of individual discontinuities
may further influence the block size and shape.
To quantify the effect of discontinuities of rock masses, three different rock mass
classifications, Rock Mass Quality (Q) system of Norwegian Geotechnical Institute
(Barton et al., 1974), Rock Mass Rating (RMR) system of the Council for Scientific and
Industrial Research, South Africa (Bieniawski, 1976 and 1989), and Geological Strength
Index (GSI) system (Hoek, 1994), have been proposed in literature. All these
classification systems were developed initially for tunnel or dam applications. However,
they have been used for other engineering applications, such as slope and foundation.
2.3.3.1 Q System
Q system (Barton et al., 1974) is based on six parameters: RQD, number of joint or
discontinuity sets (J
n
), joint roughness (J
r
), joint alteration (J
a
), water flow (J
w
) and a
stress reduction factor (SRF). Because this system is mainly used for tunneling
applications in Europe, the details to quantify these parameters will not be discussed here.
The rock mass quality (Q) is defined as:
40
|
.
|

\
|
|
|
.
|

\
|
|
|
.
|

\
|
=
SRF
J
J
J
J
RQD
Q
w
a
r
n
(2-39)
2.3.3.2 RMR System
There are two versions of RMR system: one is developed by Bieniawski in 1976; the
other one is proposed by Bieniawski (1989) which is a modified version of 1976 RMR
system. In 1976 version, the major factors used to quantify the rating are:
1. Uniaxial compressive strength of intact rock
2. RQD
3. Spacing of joints
4. Strike and dip orientations of joints
5. Condition of joints
6. Ground water inflow
The RMR
76
value equals to the summation of the six ratings. Details of the rating
system are presented in Table 2-4. The 1989 version of RMR system combines
orientations rating and condition of joints rating into one rating as condition of
discontinuities. The other items are kept same; however, the values of the ratings are
changed. The RMR
89
value also equals to the summation of five ratings as presented in
detail in Table 2-5.





41
Table 2-4 Classification Parameters and Their Rating (After Bieniawski, 1976)
Uniaxial
compressive
strength of intact
rock
> 200
MPa
100-
200
MPa
50-100 MPa
25 –50
MPa
< 25 MPa
1
Rating 10 5 2 1 0
Drill core quality
RQD
90%-
100%
75%-
90%
50%-75% 25%-50%
<25% or
highly
weathered
2
Rating 20 17 14 8 3
Spacing of joints > 3 m 1-3 m 0.3-1 m
50-300
mm
< 50 mm
3
Rating 30 25 20 10 5
Strike and dip
orientations of
joints
Very
favorable
Favora
ble
Fair
Unfavorab
le
Very
unfavorable 4
Rating 15 13 10 6 3
Condition of
joints
Very tight:
separation<0.1 mm
Not continuous
Tight: <1
mm and
continuous
No gouge
Open: 1-5
mm
Continuou
s
Gouge<5
mm
Open >5
mm
Continuous
Gouge>5
mm
5
Rating 15 10 5 0
Ground water
inflow (per 10m
of tunnel length
None
<25
liters/min
25-125
liters/min
>125
liters/min 6
Rating 10 8 5 2
Rock Mass Classes and Other Rating
Class No. I II III IV V
Description
Very good
rock
Good rock Fair rock Poor rock
Very poor
rock
Total rating 100-90 90-70 70-50 50-25 <25










42
Table 2-5 Rock Mass Rating System (Bieniawski, 1989)


43
2.3.3.3 GSI System
The RMR values are difficult to estimate for very poor rock, especially for borehole
cores that contain relatively few intact core pieces longer than 4 inch (Hoek, 1994).
Therefore, GSI system was proposed by Hoek (1994) to replace the RMR rating in
developing his rock strength criterion. GSI values are based on the structure of rock mass
and the surface condition. For cored weak rock, the physical appearance of material
recovered in the core can be used to estimate a GSI value. The details for estimating GSI
value are presented in Table 2-6 to Table 2-10.
Additionally, GSI can be converted from Rock Mass Rating (RMR) (see Hoek and
Brown, 1997). If Bieniawski (1976) RMR
76
is used, the rock mass should be assumed to
be completely dry and a rating of 10 could be assigned to the groundwater, and
adjustment for discontinuity orientation value should be set to zero. If Bieniawski (1989)
RMR
89
is used, the value of 15 should be assigned to groundwater rating and adjustment
for discontinuity value is set to zero. Consequently, the following relationship can be
established.
GSI = RMR
76
(2-40)
GSI = RMR
89
- 5 (2-41)








44
Table 2-6 GSI Ranges for General Rocks (Marinos and Hoek, 2000)

45
Table 2-7 GSI Ranges for Typical Sandstones (Marinos and Hoek, 2000)

46
Table 2-8 GSI Ranges for Typical Siltstones, Claystones and Clay Shales (Marinos and
Hoek, 2000)

47
Table 2-9 GSI Ranges for Typical Limestones (Marinos and Hoek, 2000)


T
a
b
l
e

2
-
1
0

G
S
I

E
s
t
i
m
a
t
e
s

f
o
r

H
e
t
e
r
o
g
e
n
e
o
u
s

R
o
c
k

M
a
s
s
e
s

S
u
c
h

a
s

F
l
y
s
c
h

(
M
a
r
i
n
o
s

a
n
d

H
o
e
k
,

2
0
0
0
)

48
49



CHAPTER III
LATERAL LOAD TEST RESULTS AND DATA ANALYSES

This chapter presents two lateral load tests on drilled shafts socketed in rock. The
two tests were designed and conducted by E.L. Robinson Engineering of Ohio Co. The
subsurface investigation and lab test results on rock core samples are provided. The
information pertinent to test drilled shafts and instrumentation are also presented. The
lateral load test results, including deflections at shaft head at various loading levels, the
shaft deflection profiles along depth of shaft, and strain gage readings are presented.
To obtain the most accurate experimental p-y curves from strain gage readings, this
chapter also exams the accuracy of various existing methods for deriving p-y curves from
lateral load test results, particularly in terms of strain gage readings. Hypothetical cases
are used to further verify the findings from actual field load test data.
An existing weak rock p-y criterion proposed by Reese (1997) is evaluated against
the two actual lateral load tests results.

3.1 Lateral Load Test at Dayton
3.1.1 Test Site
The lateral load test was performed for assisting the design of a new two span
reinforced concrete rib arch bridge as a replacement to an existing single span steel
50
bridge where East Siebenthaler Avenue crosses the Stillwater River in Dayton, Ohio. In
general, both the Illinoian and Wisconsin glaciers covered the site. Glacial deposits in the
area are generally less than 30 feet thick. The underlying bedrock was reported to be
limestone and shale of the Richmond Formation.
Two borings, B-2 and B-4, were drilled in the approximate location of the lateral
load test, with the boring B-2 closer to the test site. As shown in Fig. 3-1, borings B-2 and
B-4 encountered fill soils to depths of 3.5 feet and 8 feet, respectively. The fill soils were
underlain by natural soils consisting of sandy silt, silt and clay, and silty clay. Borings B-
2 and B-4 encountered auger refusal at depth of 13.5 feet and 24 feet, respectively. Then,
the borings were cored to depths of 28.5 feet and 40 feet, respectively.
The bedrock encountered in the borings consists of soft to medium gray shale
interbedded with hard gray limestone. The limestone interbeds are typically less than 1
foot thick. The gray shale is slightly weathered to decomposed, weakly calcareous, and
very thinly laminated. It is broken to very broken, becoming massive near the completion
depths of the borings. The lateral load test drilled shafts were fully constructed in the gray
shale interbedded with limestone after excavation at the site.
One direct shear test was performed on a representative sample of the gray shale
obtained from boring B-4. The residual angle of internal friction of 24º was obtained by
shearing the sample along a bedding plane. Another sample of the massive gray shale
from boring B-4 was subjected to an unconfined compression test and was found to have
an unconfined compressive strength of 5668 psi. The elastic modulus of the gray shale is
590 ksi based on the unconfined compression test results. The possible lowest and largest
RMR
89
ratings of the bedrock at the test site are estimated based on the rock properties
51
presented in Fig. 3-1. These ratings are shown in Table 3-1. The GSI values in the table
were obtained by correlations with RMR
89
using the equation GSI = RMR
89
– 5 and by
setting the water rating as 15.

Figure 3-1 Soil and rock layer profiles at Dayton test site
Table 3-1 RMR Ratings and GSI Values of Rock at Dayton Test Site
RMR
89
GSI
Elevations
(ft)
(2) (3) (4) (5) (6)
Min. Max. Mean Min. Max. Mean
721-728 4 3 - 5 5 - 15 5 - 20 0 17 44 30.5 27 54 40.5
728-739 4 - 7 13 5 - 15 20 - 25 0 42 60 51 52 70 61
(2): Strength rating; (3): RQD rating; (4): Spacing rating; (5): Discontinuity rating;
(6): Ground water rating

Fill
SPT/RQD
20
Silt and
clay
15
Sandy silt 56
Stiff silty clay 85
Broken
limestone
25%
Medium hard
gray shale,
broken,
interbedded
with limestone
11%

0%
Ele. (ft)
743.6
738.6
736.1
733.6
730.4
718.6
Boring B-2
747.1
Fill
SPT/RQD
16
Loose
sandy
silt
7
Stiff silty
clay
48
8%
Soft gray shale;
weakly calcareous;
thinly laminated;
broken; interbedded
with hard gray
limestone
0%
Ele. (ft)
739.3
735.1
723.3
707.3
Boring B-4
747.3
53%
0%
12.5 feet
741.1
738.7
52
3.1.2 Test Setup and Procedure
Two 6 feet diameter drilled shafts were constructed and tested under lateral loads at
Dayton site. The shaft was reinforced with 36 #11 primary and #6 spiral bars. The test
site was excavated to the depth so that the two drilled shafts were fully constructed in
shale without any overlying soils. The elevation of the top of the drilled shafts and the
excavated ground surface was 721 feet. The embedment length of the two drilled shafts
was 18 feet. The center-to-center distance between two test shafts is 18 feet. Fig. 3-2
shows the layout of the two test shafts. The 28-day compressive strength of concrete was
around 4500 psi.
Both drilled shafts were instrumented with inclinometers for measuring the
deflection of the shaft with depth and dial gages for measuring the deflection at shaft
head. In total, three and two dial gages were installed for measuring the deflections at the
top of shaft #4 and shaft #3, respectively. The shaft #4 was instrumented with 10
vibrating wire strain gages along the depth of shaft for monitoring the rebar strains. The
strain reading will be used to determine the shaft moments and soil reactions at each gage
elevation. A CR10X Campbell Scientific Data Logger was used to collect strain gage
readings during lateral loading. The gages were placed at a depth interval of either 2 feet
or 3 feet, as shown in Fig. 3-2.
The lateral load was applied by pushing the two drilled shafts apart via a jack and
reaction beam placed between them. A load cell was installed between shaft #4 and the
jack to measure the actual applied lateral loads. The loading sequence consisted of
applying the lateral load in increments of 50 kips or 100 kips, followed by an unloading.
The maximum load was 1126 kips. Each load was held until the rate of deflection at the
53
top of shaft was less than 0.04 inch/min and the inclinometer reading of each shaft was
taken.

Figure 3-2 Instrumentation of load test at Dayton
3.1.3 Lateral Load Test Results
The measured lateral load test results include the two load-deflection curves at the
top of the two shafts, deflection versus depth profiles measured from inclinometers at
each load level, and strain gage readings at each load level.
3.1.3.1 Load-Deflection Curves at Shaft Top
Fig. 3-3 presents the load-deflection response measured at the top of each shaft
during incremental lateral loading. The deflections were averaged from the three dial
gage readings for shaft #4 and two dial gage readings for shaft #3 at each loading level.
The relatively similar response of the two shafts indicates the spatial homogeneity of the
rock mass. It can be seen that the load-deflection curve behaves nonlinearly even at small
deflections due to nonlinear response of weak rock masses. There was one cycle of lateral
0 ft
2 ft
Load
Cell
Jack
Steel
Strut
Inclinometer
Casing
3 ft
5 ft
7 ft
10 ft
13 ft
18 ft
Shaft #4 Shaft #3
Strain
Gages
Dial
Gage
Dial
Gage
36 #11
Rebar
36 #11
Rebar
Shale
6 ft
Steel Plates
Ele. 721 ft
54
load due to the loss of jacking pressure and back up at the load level of 700 kips. This
resulted in a sudden increase of deflections of both shafts. It can be concluded that the
cyclic loading would increase the lateral deflections. After unloading, both shafts
registered 0.05 inch of permanent deflection at the top of shaft.
0
200
400
600
800
1000
1200
0.000 0.050 0.100 0.150
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Shaft #3
Shaft #4

Figure 3-3 Load-deflection curves at the top of shafts
3.1.3.2 Deflection Profiles
The deflection versus depth profiles of shaft #3 and shaft #4 deduced from
inclinometer readings by assuming the deflection at the bottom of the inclinometer tube
as zero are presented in Figs. 3-4 and 3-5, respectively. Lateral deflections were
mobilized on most part of the two shafts along depth under lateral loads. Both shafts
behaved like an intermediate shaft between a rigid shaft which exhibits linear
displacement profile along the shaft length and a flexible shaft which displays fixity in
the lower portion of the shaft. For some loading levels, the inclinometer readings are
55
considered to be unreasonable. For example, the deflections at the load of 173 kips were
less than the deflections at the load of 93 kips as shown in Fig. 3-4.
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0 5 10 15 20
Depth (ft)
D
e
f
l
e
c
t
i
o
n

(
i
n
)
92 kips
173 kips
286 kips
331 kips
427 kips
582 kips
634 kips
705 kips
786 kips
846 kips
930 kips
1126 kips

Figure 3-4 Deflection-depth curves of shaft #3
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 5 10 15 20
Depth (ft)
D
e
f
l
e
c
t
i
o
n

(
i
n
)
92 kips
173 kips
286 kips
331 kips
427 kips
582 kips
634 kips
705 kips
786 kips
846 kips
930 kips
1126 kips

Figure 3-5 Deflection-depth curves of shaft #4
56
3.1.3.3 Strain Profiles
The strain readings were recorded by a data logger connected to a laptop. The
compression strain and tension strain profiles of shaft #3 and shaft #4 under various
loading levels are presented in Fig. 3-6 and 3-7, respectively. When the load was
increased from 510 kips to 582 kips, a sudden increase of tension strain at the depth of 8
feet was noticed, as shown in Fig. 3-7. This could be due to concrete cracking in the shaft.
0
2
4
6
8
10
12
14
-250 -200 -150 -100 -50 0
Strain (micro)
D
e
p
t
h

(
f
t
)
51 kips
92 kips
173 kips
286 kips
331 kips
427 kips
510 kips
582 kips
634 kips
693 kips
705 kips
786 kips
846 kips
1007 kips
1126 kips

Figure 3-6 Compression strain profiles of shaft #4
57
0
2
4
6
8
10
12
14
-150 -50 50 150 250 350 450
Strain (micro)
D
e
p
t
h

(
f
t
)
51 kips
92 kips
173 kips
286 kips
331 kips
427 kips
510 kips
582 kips
634 kips
693 kips
705 kips
786 kips
846 kips
1007 kips
1126 kips

Figure 3-7 Tension strain profiles of shaft #4

3.2 Lateral Load Test at Pomeroy-Mason
3.2.1 Test Site
The Ohio Department of Transportation (ODOT), together with the West Virginia
Department of Transportation (WVDOT) and the Federal Highway Administration
(FHWA) worked on design and construction of a replacement bridge for the U.S. Route
33 Bridge over the Ohio River between Pomeroy, Ohio and Mason, West Virginia. The
new structure will be a three-span cable-stay bridge with a reinforced concrete
superstructure and substructure. The cable stay towers are to be supported by groups of
drilled shafts. Considering the large wind load and possible ship impact on the bridge
piers, the lateral loads will be one of the key design loads for the bridge deep foundation.
58
Due to lack of competent design methods available for analyzing laterally loaded drilled
shafts in rock, a lateral load test was carried out on two drilled shafts constructed for the
West Virginia cable stay tower.
Two Borings S-9 and S-10 near the test drilled shafts were advanced from a floating
platform at the Station 29+95.5, 24 feet right and Station 30+18.7, 28.4 feet left,
respectively. During soil boring, water surface elevations of 538.0 feet and 537.9 feet
were observed, respectively. The water was 28.8 feet deep in S-9 and 28.3 feet deep in S-
10.
In Boring S-9, from the river bottom to a depth of 38.9 feet (below the water
surface), a layer of coarse and fine sand (A-3a) and fine sand (A-3) was encountered.
This layer was described as grayish brown and orangish brown, wet and loose. Water
contents varied from 20 to 24 percent and SPT N-values ranged from 4 to 8 blows per
foot. Underlying the aforementioned soil in S-9 and at the surface in S-10, gravel with
sand (A-1-b) was observed. This material extended to depths of 57.2 and 56.9 feet in
Borings S-9 and S-10, respectively. The gravel with sand was described as brown, wet
and loose to medium dense. Water contents ranged from 13 to 20 percent and SPT N-
values varied from 4 to 16 blows per foot, with an average of 9 blows per foot.
Bedrock was encountered below the soils described above at a depth of 57.2 feet
below the water surface in S-9 and at 56.9 feet in S-10. The rock core recovered was
described as shale with interbedded siltstone from bedrock surface to depths of 75.9 and
73.6 feet below the water surface in S-9 and S-10, respectively. Shale (mudstone) was
observed in S-9 from 75.9 to 102.9 feet and in S-10 from 73.6 to 103.7 feet below the
water surface. Siltstone was recovered underlying the shale to a depth of 123.3 and 118.4
59
feet below the water surface in S-9 and S-10, respectively. The borings were terminated
at these depths. The weathered rock zone was observed in the uppermost 3 feet below the
rock surface, approximately. The rock core samples were used for unconfined
compression tests. The details of Borings S-9 and S-10 are provided in Fig. 3-8 and 3-9,
respectively.
The possible highest and lowest RMR
89
ratings are estimated based on the rock
properties indicated in Fig. 3-8 and 3-9. They are summarized in Tables 3-2 and 3-3 for
S-9 and S-10, respectively. The top elevations of the rock are adjusted according to the
drilling observations during the construction of drilled shafts. The GSI values correlated
from RMR
89
using the equation GSI = RMR
89
- 5 and by setting the water rating as 15 are
also provided in Tables 3-2 and 3-3.
Table 3-2 RMR Ratings and GSI Values of Rock Based on Boring S-9
RMR
89
GSI
Elevations
(ft)
(2) (3) (4) (5) (6)
Min. Max. Mean Min. Max. Mean
490.8 – 469 4 8 5 10-20 0 27 37 32 37 47 42
469 – 461.4 7 8 5 10-20 0 30 40 35 40 50 45
461.4-453.5 0 13 5 10 0 28 28 28 38 38 38
453.5-442.8 0 3 5 10 0 18 18 18 28 28 28
442.8-434 2 17 5 10 0 34 34 34 44 44 44
(2): Strength rating; (3): RQD rating; (4): Spacing rating; (5): Discontinuity rating;
(6): Ground water rating

60

Figure 3-8 Boring S-9
509.2
Ele. (ft)
S-9
480.8
479.0
462.1
SPT/RQD
4
7
8
7
6
5
7
6
9
13
12
50+
44% 471.2
474 3797
459.1
51%
449.0
q
u
(psi)
20%
435.1
420.5
440.0 86%
19 458
430.6
76%
567
Description
Coarse and fine sand (A-3a) and fine sand
(A-3)
Top of ground
Weathered shale
Top of rock (the elevation of top of rock at
the test shaft #2 was 490.8 feet observed
during construction)
Shale (50%) with interbedded siltstone (50%).
Shale is gray to dark gray, laminated to very
thin bedded, zones silty, soft to hard.
Siltstone is gray to dark gray, thin bedded,
zones sandy, micaceous, moderately hard to
hard.
Shale (mudstone), greenish gray to
brownish red, partially clay-like, blocky
bedding, numerous slickensided fractures
with degree of 50 to 65, soft.

Cored samples for lab test were classified as
claystone.
Predominantly gray below 441.4’
Siltstone, gray to dark gray, thin bedded,
zones sandy, micaceous, hard
500.9
Gravel with sand (A-1-b) and traces of silt
and clay, brown, wet, loose to medium
dense
10
9073
464
44.3
826
1133
E
i
(ksi)
344.8
1292
1.5
81.2
γ (pcf)
164.6
165.8
147
157.9
144.2
165.6
159.2
469.0 32%
51%
Cored sample was sandstone at this elevation.
61

Figure 3-9 Boring S-10
509.6
Ele. (ft)
S-10
481.0
478.7
464.3
SPT/RQD
12
4
7
6
8
7
6
11
16
14
13
50+
60%
468.7
473.7 905
460.3
459.3
8%
52.8
449.0
q
u
(psi)
23%
434.2
421.5
441.5 38%
37%
7.0 454.3
27.8
428.2
56% 4610.6
74%
4863.6
Description
Gravel with sand and (A-1-b) traces of silt and
clay, brown, wet, loose to medium dense
Top of ground
Weathered shale
Top of rock (the elevation of top of rock at
the test shaft #1 was 485.8 feet observed
during construction)
Shale (50%) with interbedded siltstone (50%).
Shale is gray to dark gray, laminated to very
thin bedded, zones silty, soft to hard.
Siltstone is gray to dark gray, thin bedded,
zones sandy, micaceous, moderately hard to
hard. Weathered from 481’ to 478’. Clay seam
from 471.9’ to 470.7’.
Shale (mudstone), greenish gray to
brownish red, partially clay-like, blocky
bedding, numerous slickensided fractures
with degree of 50 to 65, soft.
Predominantly gray below 441.4’
Siltstone, gray to dark gray, thin bedded,
zones sandy, micaceous, hard
18
20.85
γ(pcf)
162
E
i
(ksi)
148
145.4
149.6
160.4
160.1
62
Table 3-3 RMR Ratings and GSI Values of Rock Based on Boring S-10
RMR
89
GSI
Elevations
(ft)
(2) (3) (4) (5) (6)
Min. Max. Mean Min. Max. Mean
485.8-464.3 2 13 5 10-20 0 30 40 35 40 50 45
464.3-457.3 0 3 5 10 0 18 18 18 28 28 28
457.3-447.9 0 3 5 10 0 18 18 18 28 28 28
447.9-445.8 0 8 5 10 0 23 23 23 33 33 33
(2): Strength rating; (3): RQD rating; (4): Spacing rating; (5): Discontinuity rating;
(6): Ground water rating

3.2.2 Test Set-up and Test Procedure
The test drilled shaft #1 was constructed at Station 29+00 offset 33'-1" left; and the
test drilled shaft #2 was located at Station 29+00 offset 38'-8" right. Therefore, the
Boring S-9 is closer to the test drilled shaft #2 and the Boring S-10 is closer to the test
drilled shaft #1. Due to the variability of the site, the top elevations of bedrock at the test
drilled shaft #1 and shaft #2 were observed as 485.8 feet and 490.8 feet during
construction, respectively. The elevation of ground surface of shaft #1 was 511.8 feet.
The total length of the test drilled shaft #1 and shaft #2 was 101.4 feet and 112.9 feet,
respectively. Rock socket length was 40 feet and 56.8 feet, respectively. The distance
between the top of the two test drilled shafts and the loading point during test was 3 feet.
The thickness of the soil layers at test drilled shaft #1 and shaft #2 was 26 feet and 18.8
feet.
63
The unconfined compressive strength of concrete was 5115 psi. The diameter of the
drilled shaft socketed in bedrock was 8 feet. The reinforcement of the rock-socket portion
was 28 #18 bars and the cover was 4 inch. The portion of the two drilled shafts above the
bedrock has a diameter of 8.5 feet with a 1 inch thickness of casing and 28 #18 bars. The
equivalent modulus of drilled shafts in rock is 4250 ksi. To fully mobilize the rock-shaft
interaction and isolate the overburden soils, an 11 feet diameter casing was used to form a
gap between the test drilled shaft #2 and the soils above bedrock. This means all the
lateral forces were resisted by the bedrock during the lateral load test.
Both drilled shafts were instrumented with inclinometers for measuring the
deflection along shaft length and dial gages for measuring the deflection at loading point.
Two dial gages were installed for each test drilled shaft. The test drilled shaft #1 and #2
was instrumented with 13 and 10 levels of vibrating wire strain gages along the shaft
length for monitoring the rebar strains. Two CR10X Campbell Scientific Data Loggers
were used to collect strain gage readings during test. The details of the instrumentation
are provided in Fig. 3-10.
The testing loads were applied by tensioning a tendon connecting the two drilled
shafts. The two shafts move toward each other during test. A load cell was installed
between shaft #1 and the jack to measure the actual applied lateral loads. The maximum
load applied was 275 kips. Each load increment was held until the deflection at the top of
shafts was stable and the inclinometer reading of each drilled shaft was taken.
64

Figure 3-10 Instrumentation and load test setup
3.2.3 Lateral Load Test Results
The measured data during lateral load test on the two drilled shafts include load-
deflection curves at the loading point, deflection versus depth profiles measured from
inclinometers, and strain gage readings along shaft length.
511.8
507.3
503.3
499.3
495.3
491.3
487.3
485.8
483.3
481.3
479.3
475.3
467.3
457.3
451.3
445.8
8’
8.5’
509.6
490.8
490.4
485.4
480.4
445.4
434
475.4
470.4
465.4
460.4
455.4
450.4
8’
8.5’
Top of soil
Top of rock
Strain gages
Inclinometer casing
Test Shaft #1 Test Shaft #2
8.5’-OD casing
1’’ thickness
11’-OD casing
Jack
Load cell Working Deck
Dial gages
Ele. (ft)
538
547.2 546.9
3’
Gap
28#18
4’’ cover
Ele. (ft)
65
Fig. 3-11 presents the load-deflection response measured at the loading point during
incremental lateral loading. The deflections were taken as an average of the two dial gage
readings for each shaft at each loading level. It can be seen that the deflections at the
loading point of shaft #2 is much larger than those of shaft #1 due to the use of casing in
shaft #2 for isolation of soils. The maximum deflections at the 275 kips load level of
shaft #1 and shaft #2 are 1.725 inch and 3.73 inch, respectively.
0
50
100
150
200
250
300
0 1 2 3 4
Deflection (in)
L
a
t
e
r
a
l

L
o
a
d

(
k
i
p
s
)
Shaft #1
Shaft #2

Figure 3-11 Measured load-deflection curves at loading point of Pomeroy-Mason test
The deflection profiles of test shaft #1 and shaft #2 along shaft length, deduced from
inclinometer readings by assuming that the deflection at the bottom of the inclinometer
tube was zero, are presented in Fig. 3-12 and 3-13, respectively. From Fig. 3-12, it can be
seen that the deflection at the bedrock elevation is very small. This means that most of
applied lateral load was resisted by the soils for shaft #1. However, for shaft #2, the
lateral loads were resisted by the rock due to the use of casing to isolate soil reactions.
The strain readings were recorded by a data logger connected to a laptop during test.
The compression strain and tension strain profiles of shaft #1 under various loading
66
levels are presented in Figs. 3-14 and 3-15, respectively. The depth shown in figure starts
from the top of the drilled shafts. When the load was increased from 175 kips to 200 kips,
a sudden increase of tension strain at the depth of 56 feet was noticed. The maximum
tension strain was recorded as 938 micros. This could be due to concrete cracking in the
shaft, as sound of concrete cracking was clearly audible during test at this load increment.
The tension and compression strain profiles of test shaft #2 are provided in Figs. 3-16 and
3-17, respectively. Similarly, a sudden increase of tension strain accompanying a
cracking sound was observed at the loading level of 175 kips, as shown in Fig. 3-16.
0
10
20
30
40
50
60
70
80
90
100
-0.5 0 0.5 1 1.5 2
Deflection (in.)
D
e
p
t
h

(
f
t
)
at 25 Kips
at 50 Kips
at 100 Kips
at 150 Kips
at 175 Kips
at 225 Kips
at 250 Kips
at 275 Kips
Top of Shaft, elev. 547 ft
Jacking,
elev. 544 ft
Top of rock, elev. 485.8
Top of soil, elev. 511.8 ft

Figure 3-12 Deflection-depth profiles of drilled shaft #1 at Pomeroy-Mason test
67
0
10
20
30
40
50
60
70
80
90
100
110
-1 0 1 2 3 4 5
Deflection (in.)
D
e
p
t
h

(
f
t
)
at 25 Kips
at 50 Kips
at 100 Kips
at 150 Kips
at 175 Kips
at 225 Kips
at 250 Kips
at 275 Kips
Top of Shaft, elev. 547ft
Jacking,
elev. 544ft
Top of rock, elev. 490.8 ft

Figure 3-13 Deflection-depth profiles of drilled shaft #2 at Pomeroy-Mason test
68
35
45
55
65
75
85
95
105
-100 0 100 200 300 400
Micro Strain
D
e
p
t
h

(
f
t
)
25 kips
50 kips
75 kips
100 kips
125 kips
150 kips
175 kips
200 kips
225 kips
250 kips
275 kips

Figure 3-14 Tension strain profiles of test shaft #1 of Pomeroy-Mason test
35
45
55
65
75
85
95
105
-250 -200 -150 -100 -50 0 50
Micro Strain
D
e
p
t
h

(
f
t
)
25 kips
50 kips
75 kips
100 kips
125 kips
150 kips
175 kips
200 kips
225 kips
250 kips
275 kips

Figure 3-15 Compression strain profiles of test shaft #1 of Pomeroy-Mason test
69
55
60
65
70
75
80
85
90
95
100
105
-300 0 300 600 900 1200 1500
Micro Strain
D
e
p
t
h

(
f
t
)
25 kips
50 kips
75 kips
100 kips
125 kips
150 kips
175 kips
200 kips
225 kips
250 kips
275 kips

Figure 3-16 Tension strain profiles of test shaft #2 of Pomeroy-Mason test
55
60
65
70
75
80
85
90
95
100
105
-500 -400 -300 -200 -100 0
Micro Strain
D
e
p
t
h

(
f
t
)
25 kips
50 kips
75 kips
100 kips
125 kips
150 kips
175 kips
200 kips
225 kips
250 kips
275 kips

Figure 3-17 Compression strain profiles of test shaft #2 of Pomeroy-Mason test
70
3.3 Methods for Deriving P-y Curves From Lateral Load Test Results
3.3.1 Introduction
Site specific p-y curves derived from instrumented lateral load tests are necessary
for a situation when the site geological condition may differ from those that were used to
derive existing p-y criteria.
Derivations of p-y curves using strain gage readings from fully instrumented lateral
load tests usually involve three steps: deriving deflection versus depth profile using
double integrating technique on the curvature versus depth profile of a drilled shaft from
strain data; determining the moment versus depth profile based on curvature versus depth
profile; and finally deriving p, soil resistance per unit length of a shaft, using double
differentiating technique on the moment versus depth profile. Mathematically, these three
steps can be written as follows.
( )
∫ ∫
φ = dz dz y (3-1)
φ = EI M
(3-2)
2
2
dz
M d
p − = (3-3)
where y is deflections along depth of a drilled shaft; p is net soil resistance per unit length
of the drilled shaft; M is shaft moment at depth z; EI is section modulus of the shaft, φ is
the curvature of the shaft, which can be obtained as the difference between compression
and tension strains measured at the same depth of the shaft divided by the distance
between these two strain gages.
Mathematically, double integration of discrete data points (curvature data points)
with respect to depth generally would not introduce significant numerical inaccuracy.
71
However, double differentiation of discrete curvature data points would generally result
in amplification of measurement errors and consequently inaccurate p. Therefore, several
techniques have been proposed to minimize numerical errors due to double
differentiation, such as high order global polynomial curve fitting (Reese and Welch,
1975), piecewise polynomial curve fitting (Matlock and Ripperger 1956), cubic spline
(Dou and Byrne, 1996), weighted residuals method (Wilson, 1998), and smoothed
weighted residuals method (Yang, et al. 2005). However, these methods for deriving p-y
curves from strain data have not been systematically compared and evaluated. In fact,
there is a lack of guidance for best method in deducing p-y curves based on strain data
obtained from load tests.
In this section, the existing methods for deriving p-y curves are summarized. Eight
field lateral load tests on instrumented drilled shafts in soils and/or rocks are compiled to
evaluate these methods. Comparisons of load-deflection curves of drilled shafts between
measured values and predicted values based on those derived experimental p-y curves
using various existing methods are performed. Additionally, four hypothetical cases
based on assumed p-y curves from existing p-y criteria (Matlock, 1970; Reese and Welch,
1975; et al.) and known moment profiles from COM624P (Wang and Reese, 1993) are
used to further evaluate these derivation methods by comparing deduced experimental p-
y curves with those assumed p-y curves. It is found that piecewise polynomial curve
fitting technique provides the most accurate p-y curves.
A procedure for determining an optimum strain gage spacing for planning
instrumentation of a lateral load test is suggested. Additionally, a parametric study shows
72
that the errors of those derived p-y curves are mainly due to inaccurate determination of
moment profiles from strain gage readings.
3.3.2 Method for Deriving Deflection from Strain Gage Readings
The lateral displacements of a drilled shaft along depth at each lateral load can be
deduced from strain gage readings by using Eq. (3-1). The displacements from
inclinometer readings are not directly used, because it was found from experiences that
some inclinometers tend to provide smaller deflections of a shaft under larger lateral
loads. However, the inclinometer readings could be used as an independent check on the
accuracy of the deflections deduced from strain gage readings.
The 5
th
order polynomial function given in Eq. (3-4) has been successfully used by
Wilson (1998) to fit the discrete strain points. This technique, therefore, is adopted herein
to fit discrete curvature data points along shaft depth using the least-square method.
5 4 3 5 . 2
fz ez dz cz bz a + + + + + = φ (3-4)
where a, b, c, d, e, f are curve fitting constants.
The deflections can be obtained by double integrating the above fitted curvature
versus depth profile. Two boundary conditions are required to determine two integrating
constants involved in the double integration. Based on the evaluation (to be discussed
later), it is found that the boundary combination (y
0
, y
tip
=0) for a long pile (ratio of shaft
length over diameter, L/D ≥ 10, according to Boghrat, 1990) and the combination (y
0
,
y
fixity
=0) for a short pile (L/D < 10) would provide the best deduced deflections by
comparing with inclinometer readings. Additionally, combination (y
0
, θ
0
) can be used for
a short drilled shaft when combination (y
0
, y
fixity
=0) is not available. The meanings of
these boundary conditions are as follows: y
0
is the measured deflection at the ground line;
73
θ
0
is the measured shaft tilt at the ground line; y
tip
=0 means the deflection at the tip of a
drilled shaft is assumed to be zero; y
fixity
=0 means that deflection is set to zero at a fixity
point where the deflection of the shaft is close to zero according to the inclinometer
readings.
The problems associated other combinations are observed during the evaluation as
follows. Combination (y
tip
=0, θ
tip
=0) may result in deflections at the ground line to be
different from the measured shaft deflections. Combination (y
0
, y
fixity
=0) may result in
large negative deflections at the tip of a long drilled shaft, which is deemed unreasonable.
Combination (y
0
, θ
0
) provides none zero deflections at the tip of a long drilled shaft
which may affect accuracy of shaft-head slope measurement. A similar conclusion was
reported by Dunnavant (1986). However, combination (y
0
, θ
0
) can be used for a short
drilled shaft when a fixity point is not known.
3.3.3 Determination of Moment Profiles
The nonlinearity of shaft stiffness is usually represented by M-φ relationship, which
can be inferred from field data by comparing measured values of φ to known values of M
at various loading levels, as was done by Reese and Welch (1975). When a lateral load
test is performed with a large loading eccentricity (distance between the loading point
and ground line), a pair of strain gages can be mounted at the ground line level to obtain a
representative M-φ relationship of the shaft.
Although it is preferred to use field measured moment-curvature relationship;
nevertheless, it may not always available. In this case, the relationship between M and φ
can be obtained from theory of reinforced concrete, where stress-strain curves for both
74
reinforcing steels and concrete or predefined stress-strain models are used to compute M-
φ curves (Wang and Reese, 1993).
3.3.4 Methods for Deriving P (Net Resistance)
3.3.4.1 Piecewise Polynomial Curve Fitting
Both Matlock and Ripperger (1956) and Dunnavant (1986) have used piecewise
cubic polynomial function to fit discrete moment data. The procedure described by
Dunnavant (1986) is presented in Fig. 3-18, where piecewise cubic polynomial with a
window of 5 points is employed to fit measured moment data using the least-square
technique. That is every 5 successive moment data points along the shaft length are fitted
to one cubic polynomial curve. The double differentiation of the local fitted polynomial
curve with respect to the middle point yields p (the unit reaction of soil) at that point. The
p of upper three points and bottom three points is obtained from the smoothed local cubic
polynomial moment curve using the top five points and bottom five points, respectively.
The zero moment at the loading point or a known value of moment at the ground
line should also be included in the moment profiles. The piecewise polynomial curve
fitting is a local curve fitting technique, which avoids the requirement to capture the
global trend of scattered data in global polynomial curve fitting approach. However, this
method requires minimum of five discrete data points along shaft length.
75

Figure 3-18 Procedure for reducing moment data to p using piecewise polynomial (after
Dunnavant, 1986)
3.3.4.2 High Order Global Polynomial Curve Fitting
Single high order polynomials have been used to fit moment profiles by Reese and
Welch (1975) and Wilson (1998). According to Wilson (1998), a five order polynomial
with an exponent of 2.5 rather than 2 on the quadratic term using least-square technique,
as shown in Eq. (3-5), provided the most reasonable p profiles. The non-integer form of
polynomial will yield a zero soil resistance at the ground surface assuming depth z is zero
at the ground surface. The value of p can then be obtained by double differentiating Eq.
(3-5) with respect to z.
5 4 3 5 . 2
fz ez dz cz bz a M + + + + + = (3-5)
The advantage of the method is its simplicity of application. However, the technique
is applicable only if the trend of moment profile can be captured, especially the moment
data points near the ground surface.

H
z
M
1
=A
1
+B
1
z+C
1
z
2
+D
1
z
3
M
3
=A
3
+B
3
z+C
3
z
2
+D
3
z
3
M
7
=A
7
+B
7
z+C
7
z
2
+D
7
z
3
M
2
=A
2
+B
2
z+C
2
z
2
+D
2
z
3
(1) Least-square technique is used to fit
cubic polynomials to groups of give
contiguous moment values.

(2) The first polynomial, M
1
, is
differentiated twice to evaluate p at
the 3 moment levels closest to the
surface (including the loading level).
Other polynomials, such as M
2
, are
used to evaluate p at the group center
point. The polynomial for the five
76
3.3.4.3 Weighted Residuals Method
Weighted residuals (WR) method was introduced to derive p from moment profiles
by Wilson (1998). The WR method is not a curve fitting technique; rather it is a
numerical differentiation method based on minimizing weighted residuals, as is often
used in finite element method. The main idea of WR method is to find an approximate
function a(z) to represent the actual function u(z) over some interval z=0 to z=L.
Generally, u(z)≠a(z), and the difference can be defined as a(z)-u(z)=R(z), the residual.
While R(z) may not be zero anywhere in the range of z, a(z) can be selected such that R(z)
is zero in an average sense by enforcing the following condition.
( ) ( ) 0 z d z z R
L
0
=

Ψ
(3-6)
where ψ(z) is a weighting function. This is commonly referred to as saying u(z) =a(z)
“weakly”.
WR method is used to obtain shear force profiles by differentiating the moment
profiles one time. Then, another differentiation on the derived shear force profiles using
WR method results in the soil/rock reaction profiles, p(z). As described by Wilson (1998),
the shaft can be considered as discretized finite elements with nodes at each location with
measured bending moment value. If f(z) represents the actual bending moment
distribution of the shaft as a function of depth, then f(z) is known at the nodes. Then, the
shear force g(z) as the first derivative of the bending moment distribution is defined as:
( )
( )
( ) z ' f
z d
z f d
z g = =
(3-7)
If g(z)=f’(z) “weakly”, then
77
( ) ( ) { } ( ) 0 z z z ' f z g
L
0
=

Ψ − d (3-8)
where ψ(z) is a weighting function. Both f(z) and g(z) are written as linear combinations
of shape functions of finite element type, e.g. linear “hat” functions as shown in Fig. 3-19.

Figure 3-19 Linear shape functions
For each node along the shaft, weighting function ψ(z) is taken to be the shape
function to generate a system of linear equations for the coefficients of f(z) and g(z). This
can be written as:
( ) ( )

=
Ψ =
n
i
i
z f z f
0
, and (3-9)
( ) ( )

=
Ψ =
n
i
i
z g z g
0
(3-10)
where i is the node number and ranges from 0 to n; f
i
is the measured moment; z is the
depth along the shaft; and ψ(z) is the shape function for node i. Substituting Eqs. (3-9)
and (3-10) into Eq. (3-8), Wilson (1998) developed a system of equations that can be
used to solve for the values of g(z) at each node. The solution to those equations is
provided herein.
The measured moment values inferred from strain gage readings at depth z
i
are f(z
i
).
According to the shape function expressed in Fig. 3-19, the bending moment between
adjacent nodes i and i+1 can be expressed as follows:
i-1 i i+1
1
0
General
Ψ(z)
1 2
1
0
First Element
Ψ(z)
n-2 n-1 n
1
0
Last Element
Ψ(z)
78
( ) ( ) ( )
i i
i
i
i i
i
i
z z
z z
z f
z z
z z
z f z f


+


=
+
+
+
+
1
1
1
1
(3-11)
Substituting Eqs. (3-9) and (3-10) into Eq. (3-8), one can obtain:
( )
( )
( ) 0 d
d
d
0
0
0
= Ψ
¦
¦
)
¦
¦
`
¹
¦
¦
¹
¦
¦
´
¦
Ψ
− Ψ



=
=
L
n
i
i
n
i
i
z z
z
z f
z g (3-12)
By expanding Eq. (3-12), the following equation can be obtained for the first
element.
( ) ( ) | | ( ) ( ) ( ) ( ) | | { }( )
( )
0 d
1
0
2
0 1
1 1 0 0 1 0 1
=

− − + − − −

z
z z
z z z g z z z g z z z f z f
z
z
(3-13)
For the last element, we have:
( ) ( ) | | ( ) ( ) ( ) ( ) | | { }( )
( )
0 d
1
2
1
1 1 1 1
=

− − + − − −



− − − −
z
z z
z z z g z z z g z z z f z f
n
z
n
z
n n
n n n n n n n
(3-14)
Also, for the general elements, the following equation based on Eq. (3-12) can be
obtained.
( ) ( ) | | ( ) ( ) ( ) ( ) | | { }( )
( )
( ) ( ) | | ( ) ( ) ( ) ( ) | | { }( )
( )
0 z
z z
z z z g z z z g z z z f z f
z
z z
z z z g z z z g z z z f z f
1 i
z
i
z
2
i 1 i
1 i 1 i i i 1 i i 1 i
i
z
1 i
z
2
1 i i
1 i i 1 i 1 i i 1 i i
=


− − + − − −
+


− − + − − −
+
+
+ + + +


− − − −
d
d
(3-15)
where f(z
i
) = measured value of bending moment at node i in the drilled shaft; g(z
i
) =
shear force at node i; z
i
= the depth of node i.
The Eqs. (3-13) to (3-15) can be rewritten as follows, respectively.
( ) ( )
( )
( )
( ) ( ) 0 z z
2
z z
z z
6
1
z z
g
3
z z
z z
g
z z
2
z
2
z
z z
f f
0 1
1 0 3
0
3
1
2
0 1
1
3
0 1
2
0 1
0
1 0
2
0
2
1
2
0 1
0 1
=
(
¸
(

¸

− − −





|
|
.
|

\
|
− +


(3-16)
79
( ) ( )
( ) ( )
( )
( )
0
3
z z
z z
g
z z
2
z z
z z
6
1
z z
g
z z
2
z
2
z
z z
f f
3
1 n n
2
1 n n
n
1 n n
n 1 n 3
1 n
3
n
2
1 n n
1 n
n 1 n
2
1 n
2
n
2
1 n n
1 n n
=



(
¸
(

¸

− − −


|
|
.
|

\
|
− +













(3-17)
( ) ( )
( ) ( ) −
(
¸
(

¸

− − −


|
|
.
|

\
|
− +











1 i i
i 1 i 3
1 i
3
i
2
1 i i
1 i
i 1 i
2
1 i
2
i
2
1 i i
1 i i
z z
2
z z
z z
6
1
z z
g
z z
2
z
2
z
z z
f f

( )
( )
( ) ( )
( )




|
|
.
|

\
|
− +


+


+
+
+
+
+
+ −

3
z z
z z
g
z z
2
z
2
z
z z
f f
3
z z
z z
g
3
i 1 i
2
i 1 i
i
1 i i
2
i
2
1 i
2
i 1 i
i 1 i
3
1 i i
2
1 i i
i

( )
( ) ( ) 0 z z
2
z z
z z
6
1
z z
g
i 1 i
1 i i 3
i
3
1 i
2
i 1 i
1 i
=
(
¸
(

¸

− − −

+
+
+
+
+
(3-18)
The above equation group that includes 3n equations can be expressed using
matrices as follows:
| |{ } { } B G A = (3-19)
(
(
(
(
(
(
(
(
(
(
(
(
¸
(

¸


+ −
nn n n
i i i i i i
a a
a a a
a a
1
1 1
01 00
L L L
L L L
¦
¦
¦
¦
¦
¦
)
¦
¦
¦
¦
¦
¦
`
¹
¦
¦
¦
¦
¦
¦
¹
¦
¦
¦
¦
¦
¦
´
¦

+

n
1 n
1 i
i
1 i
1
0
g
g
g
g
g
g
g
M
M
=
¦
¦
¦
¦
¦
¦
)
¦
¦
¦
¦
¦
¦
`
¹
¦
¦
¦
¦
¦
¦
¹
¦
¦
¦
¦
¦
¦
´
¦

+

n
n
i
i
i
b
b
b
b
b
b
b
1
1
1
1
0
M
M
(3-20)
where
( )
0 1 00
z z
3
1
a − = (3-21)
( )
0 1 01
z z
6
1
a − = (3-22)
( )
0 1 0
f f
2
1
b − = (3-23)
80
( )
1 i i 1 i i
z z
6
1
a
− −
− = (3-24)
( )
1 i 1 i ii
z z
3
1
a
− +
− = (3-25)
( )
i 1 i 1 i i
z z
6
1
a − =
+ +
(3-26)
( )
1 i 1 i i
f f
2
1
b
− +
− = (3-27)
( )
1 n n 1 n n
z z
6
1
a
− −
− = (3-28)
( )
1 n n n n
z z
3
1
a

− = (3-29)
( )
1 n n n
f f
2
1
b

− = (3-30)
As a result, each element in coefficient matrix [A] and vector {B} can be determined
based on the above equations, and the solution {G} is expressed as:
{ } | | { } B A G
1 −
= (3-31)
Note that the bending moment distribution f(z) is approximated as piecewise linear
function using the weighting or shape function ψ(z). The WR approximation to derivative
of f(z), however, is piecewise linear, and can be applied a second time to obtain a
piecewise linear approximation on the second derivative. Thus, a double differentiation
of moment profiles using WR method can result in the soil/rock reaction profiles, p(z).
The WR method has been coded into the software Matlab to obtain shear force profiles
and soil/rock reaction profiles based on the measured values of moment.


81
3.3.4.4 Smoothed Weighted Residuals Method
An improvement on WR method can be achieved by using the 2D negative
exponential smoothing function of commercial available software, Sigmaplot, to smooth
the data points before WR differentiation is applied. The smoothing process can provide a
smooth trend of measured moment data. Additionally, it can interpolate as many data
points as needed to make the spacing of moment data points smaller. A study by Yang, et
al. (2005) has shown that this improved technique helps generation of smooth p-y curves.
3.3.4.5 Cubic Spline Curve Fitting
A cubic spline was employed to fit the discrete moment data points to derive p by
Mezazigh and Levacher (1998). The cubic spline is perhaps the simplest interpolation
function of discrete test data that can be double differentiated. However, since a spline
fits every point exactly, it is therefore prone to errors of measurement. In this dissertation,
cubic spline curve fitting technique will not be examined.
3.3.5 Evaluation Using Field Test Data
A total of eight field lateral load tests results on fully instrumented drilled shafts in
soils and/or rocks have been compiled herein for evaluating the various methods for
deriving p-y curves. A summary of these load tests is presented in Table 3-4. Four
methods for deriving p, including piecewise polynomial, 5
th
order global polynomial,
weighted residuals (WR), and smoothed weighted residuals (SWR), are evaluated.
To deduce p-y curves, the y vs. depth profiles are first obtained by double
integrating curvature vs. depth profiles using the 5
th
order polynomial curve fitting
method. The relationship between moment and curvature is obtained by comparing
measured curvatures at the ground surface level and the known applied moments for load
82
tests PomS1 and PomS2. Extrapolation is used when curvature is greater than the
maximum curvature measured at the ground surface. For other load tests, M-φ curves are
obtained using the LPILE computer program by inputting the drilled shaft geometry,
concrete strength, and reinforcement.
Due to possible measurement errors, variability of concrete stiffness, and inaccurate
moment-curvature relationships, some moment profiles are considered as outlier. For
such load tests data (e.g., CDOTS1), these outlier moment versus depth curves were
discarded, as illustrated in Fig. 3-20.
Table 3-4 Compiled Lateral Load Test Database
No. Test Shaft Diameter
(ft)
Total
Length (ft)
Location Reference
1 PomS1 8 101 Pomeroy, OH This chapter
2 PomS2 8 112.9 Pomeroy, OH This chapter
3 CDOTC1 2.5 16.7 Denver, CO Nusairat, et al. (2004)
4 CDOTC2 2.5 16.7 Denver, CO Nusairat, et al. (2004)
5 CDOTS1 2.5 21 Denver, CO Nusairat, et al. (2004)
6 CDOTS2 2.5 21 Denver, CO Nusairat, et al. (2004)
7 DaytonS4 6 18 Dayton, OH This chapter
8 MaumeeS1 8 93 Toledo, OH Yang and Liang (2005)

3.3.5.1 Deflection versus Depth
The deflections deduced from strain gage readings are checked against the
deflections measured by inclinometers. In general, the deflections deduced from strain
readings match the deflections from inclinometer readings. A representative comparison
for test PomS1 is shown in Fig. 3-21.

83
0
5000
10000
15000
20000
25000
30000
0 50 100 150 200
Depth (in)
M
o
m
e
n
t

(
k
i
p
-
i
n
)
3 kips
8 kips
13 kips
18 kips
25 kips
35 kips
45 kips
55 kips
65 kips
Outlier Profiles

Figure 3-20 Elimination of outlier moment profiles
0
10
20
30
40
50
60
-0.05 0 0.05 0.1 0.15 0.2
Deflections (in)
D
e
p
t
h

(
f
t
)
175 kips - Inclinometer
175 kips - From strains
150 kips - Inclinometer
150 kips - From strains
50 kips - Inclinometer
50 kips - From strains


Figure 3-21 Comparison of deflections from strain readings and inclinometer for test
PomS1
84
3.3.5.2 P versus Depth Profile
The derived p values near the ground surface from WR and SWR methods are
negative for test data from MaumeeS1, CDOTS2, and PomS1, as shown in Fig. 3-22 for
MaumeeS1. Similarly, the p values near the ground surface of PomS1 are also negative
when global polynomial curve fitting method is used. Furthermore, SWR method results
in irregular spikes in p vs. depth profiles for tests data of CDOTC1 and CDOTC2, as
shown in Fig. 3-23 for CDOTC1.
Based on the above observations, it is concluded that WR and SWR methods are not
capable of producing reasonable p v. depth profiles for some of the load tests. On the
other hand, the piecewise polynomial curve fitting method seems to be able to provide a
reasonable result.
3.3.5.3 Load-Deflection Curve at Shaft Head
The predicted load-deflection curves at the ground line (or shaft head) by inputting
the deduced p-y curves into LPILE (or COM624P) program are compared with the
measured. The method proposed by Murchison and O’Neill (1984) is employed to
quantify prediction error. The prediction error is defined as the difference between the
measured and predicted deflections at the ground line divided by the measured
deflections at the same loading level, as shown in Fig. 3-24. Specifically, the errors at
four loading levels (i.e., 0.25H
max
, 0.5H
max
, 0.75H
max
, and H
max
), are computed and
summed. It should be noted that H
max
is usually taken as the maximum applied load;
however, for tests, such as CDOTS1, where elimination of outlier data has resulted in the
use of highest reasonable loads as H
max
.
85
The calculated cumulative errors for each load test data are summarized in Table 3-5.
It can be seen that piecewise polynomial curve fitting method yields the smallest
cumulative errors. If test CDOTS1 is ignored due to its inaccurate moment versus depth
profiles, the average calculated error of the four loading levels of the remaining seven
tests using the piecewise polynomial method deduced p-y curves is about 29%.
0
200
400
600
800
1000
1200
-10000 -5000 0 5000
p (lb/in)
D
e
p
t
h

(
i
n
)
Polynominal
WR
SWR
Piecewise


Figure 3-22 P vs. depth profile of test MaumeeS1
86
0
20
40
60
80
100
120
140
160
180
-10000 -5000 0 5000 10000 15000
p (lb/in)
D
e
p
t
h

(
i
n
)
Polynomial
WR
SWR
Piecewise


Figure 3-23 P vs. depth profile of test CDOTC1

Figure 3-24 Definition of deflection prediction error
Load
Deflection
Measured
Predicted
H
max
∆y
Errors= ∑∆y/y
mea
0.75H
max
0.5H
max
0.25H
max
y
mea

∆y
∆y
∆y
87
Table 3-5 Cumulative Shaft Head Deflection Errors based on Various Methods
Test Shaft Polynomial WR SWR Piecewise
PomS1 6.3 6.3 2.7 3.0
PomS2 0.9 1.0 0.8 0.9
CDOTC1 1.3 1.4 2.0 0.7
CDOTC2 1.7 2.3 1.9 1.9
CDOTS1 8.4 5.2 9.9 8.5
CDOTS2 0.5 1.4 0.4 0.5
DaytonS4 0.6 1.2 0.5 0.4
MaumeeS1 0.5 3.8 1.0 0.5
Total 20.3 22.6 19.2 16.5

3.3.5.4 Maximum Moment in Drilled Shafts
In addition to deflection prediction, the maximum moments of drilled shafts under
the four loading levels predicted using COM624P based on experimental p-y curves are
compared with the corresponding maximum moments based on strain gage readings. The
moment prediction errors, defined as the moment difference divided by the measured
moments, are summarized in Table 3-6. It can be seen that the difference among these
four methods is not significant with SWR providing the smallest error.
Table 3-6 Cumulative Moment Errors based on Various Methods
Test Shaft Polynomial WR SWR Piecewise
PomS1 0.6 1.3 0.9 0.9
PomS2 0.3 0.3 0.3 0.3
CDOTC1 0.7 0.8 0.8 0.7
CDOTC2 0.8 1.0 0.8 0.7
CDOTS1 0.3 0.2 0.2 0.2
CDOTS2 2.4 1.3 2.1 2.2
DaytonS4 0.0 0.1 0.1 0.1
MaumeeS1 1.3 1.3 0.4 1.6
Total 6.3 6.3 5.6 6.8



88
3.3.6 Evaluation Using Hypothetical Cases
3.3.6.1 Hypothetical Cases
LPILE (or COM624P) was used with the prescribed (assumed) p-y curves to
simulate results of a lateral load test, from which moment versus depth profiles under
various lateral loads were obtained. These moment profiles were then used to deduce p
versus depth profile. The p-y curves could then be obtained by using y versus depth
profiles from the simulation. By comparing the deduced p-y curves using various
derivation methods with the prescribed (assumed) p-y curves, one could then assess the
accuracy of various p derivation methods. Three p derivation methods, including
piecewise polynomial, global 5
th
order polynomial, and WR method, are included in this
evaluation study. SWR method was not evaluated, because it yields almost the same
results as WR method, if a small spacing of moment data points is chosen.
Four hypothetical cases are chosen for four types of p-y curves that could be
prescribed in COM624P program. The drilled shaft used in these four hypothetical cases
has a diameter of 3 ft and embedment length of 30 ft. The reinforcement of the shaft is 12
#14. The soil used for each case is a homogeneous soil layer. The soil parameters for the
selected four p-y criteria are summarized in Table 3-7. The spacing of moment data
points used for deducing p is prescribed as 12 inch.
Table 3-7 Soil Parameters of Hypothetical Cases
Case No. p-y criteria (COM624P) S
u
(psi) φ (º) γ'(pci) ε
50
k
s
(pci)
1 Stiff clay above water 20 - 0.072 0.004 1000
2 Stiff clay below water 20 - 0.035 0.004 1000
3 Sand - 36 0.072 - 90
4 Soft clay 2 - 0.052 0.02 20
S
u
– undrained shear strength, φ – friction angle, γ' – effective unit weight, ε
50
– strain
at 50% of maximum principle stress, k
s
– coefficient of initial slope of p-y curves.
89
3.3.6.2 Comparison of p-y Curves
The deduced p-y curves are compared with the prescribed p-y curves in Fig. 3-25. It
can be seen that the global 5
th
order polynomial derived p-y curves deviate significantly
from the prescribed p-y curves. The piecewise polynomial derived p-y curves closely
match the prescribed p-y curves except for case 2. It also shows that WR method
provides a good result.
To quantify the accuracy of these three methods investigated, the errors of the
derived p-y curves are computed. Using the same definition of error of load-deflection
curves given in Fig. 3-24, the p-y curve error is defined as ∆p/p
prescribed
, where p
prescribed
is
the p from prescribed p-y cruves and ∆p is the difference between the p from prescribed
and the derived. The errors at deflections of 0.25y
m
, 0.5y
m
, 0.75 y
m
, and y
m
, where y
m
is
the maximum y of prescribed p-y curves, are calculated for each p-y curve. The
cumulative errors of four p-y curves at four different depths (1 ft, 4 ft, 7 ft, and 10 ft) for
each case are calculated and summarized in Table 3-8. It can be seen that the piecewise
polynomial method yields the smallest error, while the global polynomial method leads to
the largest error.
Table 3-8 Cumulative Errors of p-y Curves of Hypothetical Cases
Case No. Polynomial WR Piecewise
1 4.6 1.0 0.2
2 83.5 16.4 16.5
3 4.0 1.0 0.6
4 3.5 0.8 0.4
Total 95.5 19.1 17.8

90
0
500
1000
1500
2000
2500
0 5 10 15
y (in)
p

(
l
b
/
i
n
)
COM624P
Poly.
WR
Piecewise

0
500
1000
1500
2000
2500
3000
3500
4000
0 0.5 1 1.5
y (in)
p

(
l
b
/
i
n
)
COM624P
Poly.
WR
Piecewise

(a) Case 1 – @12 in (b) Case 2 – @12 in
-100
0
100
200
300
400
500
600
0 2 4 6 8
y (in)
p

(
l
b
/
i
n
)
COM624P
Poly.
WR
Piecewise
0
50
100
150
200
250
0 2 4 6
y (in)
p

(
l
b
/
i
n
)
COM624P
Poly.
WR
Piecewise

(c) Case 3 – @12 in (d) Case 4 - @12 in
Figure 3-25 Comparison of p-y curves
3.3.7 Optimum Strain Gage Spacing
The spacing of strain gages is important for deriving accurate p-y curves from lateral
load tests. A sparsely distributed strain data points may result in inaccurate p-y curves; on
the other hand, too closely spaced strain gage distribution my be cost prohibitive. An
optimum spacing of strain gages, which balances the cost and accuracy of the deduced p-
y curves, is needed in planning of any instrumentation project for a lateral load test. The
91
following procedure is suggested for determining an optimum strain gage spacing for a
particular lateral load test so that accurate p-y curves can be derived with minimum cost
spent on strain gages. The first step is to select p-y criterion from the pool of existing
criteria in LPILE program that fit to the soil condition at test site. Next, generate the
moment vs. depth profiles of the pile at various loads using LPILE program with the
prescribed p-y curves. Then, deduce p-y curves using the piecewise polynomial curve
fitting method and LPILE generated moment vs. depth profiles. In this step, different trial
moment spacings should be used to obtain different sets of p-y curves. Finally, compare
the deduced p-y curves with the prescribed p-y curves used in step one and determine an
optimum moment (or strain gage) spacing accordingly.
To illustrate the suggested procedure, an example case with homogenous soil is
employed. Moment spacings varying from 1 ft to 7 ft are investigated. To account for the
effect of drilled shaft diameter on optimum spacing, additional two cases are introduced
with drilled shafts diameter of 6 ft and 9 ft. The ratio of shaft length over diameter is kept
as 10 for both additional cases. The soil and shaft strength parameters are kept the same
as those used for the hypothetical case 1.
The cumulative errors of deduced p-y curves at four depths based on piecewise
polynomial method for various chosen moment data point spacing is presented in Fig. 3-
26. It can be seen that 3 ft, 4 ft and 5 ft of spacing can minimize the errors of the deduced
p-y curves and be taken as the optimum spacing for a drilled shaft with diameter of 3 ft, 6
ft, and 9 ft, respectively.
92
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7
Spacing (ft)
C
u
m
u
l
a
t
i
v
e

E
r
r
o
r
s
D=3 ft
D=6 ft
D=9 ft
Optimum Spacing

Figure 3-26 Optimum spacing of strain gages
It should be noted that above recommended method for determining optimum strain
gage spacing is only valid when the moment vs. depth profiles can be accurately
determined from strain gage readings. A lateral load test usually only mobilizes the
deflection at top soil layers; therefore, it is preferred to use smaller gage spacing near
ground surface and larger spacing for the bottom portion of the drilled shafts in order to
obtain moment versus depth profile as accurately as possible.
3.3.8 Effect of Measurement Error
To study the effect of error of measured moment profiles on the deduced p-y curves,
a parametric study is performed. According to Eq. (3-2), the errors of moment-depth
profiles from strain gage readings may be due to two possible reasons: the error of strain
gage measurement and inaccurate estimate of shaft stiffness EI. The worst moment
profiles could be appearance of wavy types of profiles due to unstable strain gage
93
readings. As far as the errors due to shaft stiffness, three possible situations may occur,
including under estimating EI, over estimating EI before cracking, and over estimating EI
at positions where crack appears. A diagram of these four possible measurement errors of
moment is sketched in Fig. 3-27.

Figure 3-27 Four types of moment error profiles
The hypothetical case 1 is used to investigate the effect of moment errors on the
deduced p-y curves by modifying the moment profiles from COM624P. Two feet of
spacing of moment data points is used. The errors of moment are varied from 5% to 20%.
The p-y curves deduced from the modified moment profiles using piecewise polynomial
method are compared with the prescribed p-y curves in the initial LPILE analysis. The
cumulative p-y curve errors of four deflections (i.e., 0.25y
m
, 0.5y
m
, 0.75 y
m
, and y
m
) at
four depths (i.e., 2 ft, 6 ft, 10 ft, and 14 ft) are presented in Fig. 3-28. It can be seen that
the wavy type of moment error induces the largest error in the deduced p-y curves among
these four types of moment error. The cumulative errors caused by other three types of
error are about the same.
M
z
Type A
Actual Moment
profile
M
z
Type B
M
z
Type C
Actual Moment
profiles
M
z
Type D
94
0.0
5.0
10.0
15.0
20.0
25.0
30.0
0 5 10 15 20
Moment Error (%)
C
u
m
u
l
a
t
i
v
e

E
r
r
o
r
Type A
Type B
Type C
Type D

Figure 3-28 Cumulative p-y curve errors due to moment errors
The cumulative errors of the deduced p-y curve for the four cases are greater than
3.3 for 20% of moment error, which is much larger than the cumulative error of 0.2 when
accurate moment profiles is used as shown in Table 3-8. This implies that the errors of
deduced p-y curves are mainly due to inaccurate determination of moment profiles from
strain gage readings when an appropriate method is used to deduce the p-y curves.
3.3.9 Conclusions on Methods for Deriving p-y Curves
Based on the work presented in this section, the following conclusions can be made:
1. WR and SWR methods tend to yield unreasonable p vs. depth profiles for 5 out of 8
tests. Some of the problems are negative p values near ground surface and irregular
spikes. On the other hand, piecewise polynomial curve fitting method yields
reasonable results.
2. The deduced p-y curves using WR method leads to the largest errors on the predicted
95
load-deflection curve of drilled shafts. The p-y curves deduced from the piecewise
polynomial curve fitting method on the other hand yields the smallest error.
3. Regarding the prediction of maximum moment of a drilled shaft, the p-y curves
deduced by SWR method yields the smallest errors. However, the difference of
maximum moment prediction errors of all four methods is not significant.
4. The piecewise cubic polynomial curve fitting method provides the smallest error of
the deduced p-y curves for the 4 hypothetical cases. On the other hand, global 5
th

order polynomial curve fitting provides the largest error.
5. Based on the evaluation studies presented, piecewise cubic polynomial curve fitting
method is recommended as a method for deriving p versus depth.
6. A procedure is outlined for determining optimum strain gage spacing for developing
site-specific instrumentation plans.
7. The parametric study suggests that the error of the deduced p-y curves are mainly due
to inaccurate determination of moment profiles from strain gage readings, provided
that an appropriate method for deducing p is used.

3.4 Analyses of the Load Tests
The lateral load tests presented earlier in this chapter are used to deduce p-y curves
and to evaluate interim p-y criterion for weak rock developed by Reese (1997).
3.4.1 Evaluation of Experimental P-y Curves
By using piecewise polynomial curve fitting technique, the p-y curves for shaft #4 of
Dayton load test and those for shaft #1 and shaft #2 of Pomeroy-Mason load test at
various depths are derived and presented in Figs. 3-29, 3-30, and 3-31, respectively.
96
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
0 0.03 0.06 0.09 0.12
y (in)
p

(
l
b
/
i
n
)
36 inch
132 inch

Figure 3-29 Experimental p-y curves for shaft #4 of Dayton load test
0
1000
2000
3000
4000
5000
6000
0 0.05 0.1 0.15
y (in)
p

(
l
b
/
i
n
)
60 inch
108 inch
252 inch
300 inch

Figure 3-30 Experimental p-y curves for shaft #1 of Pomeroy-Mason load test
97
0
5000
10000
15000
20000
25000
30000
0 0.02 0.04 0.06 0.08 0.1
y (in)
p

(
l
b
/
i
n
)
6 inch
66 inch
126 inch

Figure 3-31 Experimental p-y curves for shaft #2 of Pomeroy-Mason load test
To verify the accuracy of these derived p-y curves, an analysis using LPILE with
these p-y curves was made to compare with actual measured load-deflection curves. The
lateral load was not applied at the top of the shaft for Dayton load test, while the program
LPILE requires the input load be applied at the shaft top, as shown in Fig. 3-32(a).
Therefore, to facilitate analysis using LPILE, an equivalent load combination, including a
same value of lateral load at the top of shaft and an additional moment, is used to
represent the actual applied lateral load, as shown in Fig. 3-32(b). For the load test at
Pomeroy-Mason, the deflections were measured at the loading point; therefore, the
loading point will be taken as the top of the shaft for the analysis using LPILE. The effect
of casing above rock layer at Pomeroy-Mason load test is considered in LPILE analysis.
The nonlinear pile stiffness option is chosen to account for the reduction of stiffness of
concrete shaft after the appearance of crack.
98
The load-deflection curves at the top of shaft predicted by LPILE using the derived
p-y curves are provided in Figs. 3-33, 3-34, and 3-35 for the three test drilled shafts. Even
though, the predicted deflections are less than the measured values at large load levels for
shaft #2 of Pomeroy-Mason load test. In general, the predicted load-deflections curves
match those measured curves. This provides validation that the recommended method for
deriving p-y curves can be used to obtain reasonable site specific p-y curves.

Figure 3-32 Equivalent load combinations
0
200
400
600
800
1000
1200
0 0.05 0.1 0.15
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 3-33 Prediction of load-deflection curves of shaft #4 of Dayton load test using
experimental p-y curves
H
H
M=He
e
Shale Shale
(a) Applied Load (b) Equivalent Loads
99
0
50
100
150
200
250
300
0 0.5 1 1.5 2
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 3-34 Prediction of load-deflection curves of shaft #1 of Pomeroy-Mason load test
using experimental p-y curves
0
50
100
150
200
250
300
0 1 2 3 4
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 3-35 Prediction of load-deflection curves of shaft #2 of Pomeroy-Mason load test
using experimental p-y curves

100
3.4.2 Evaluation of Existing p-y Criterion
The interim p-y criterion for weak rock proposed by Reese (1997) is evaluated by
using LPILE analysis to predict field test data. The main rock properties required by
Reese p-y criterion are γ

(effective unit weight), RQD, q
u
(unconfined compressive
strength), E
m
(modulus of rock mass), and k
rm
(strain at 50% ultimate load). A range of
0.0005 to 0.00005 of k
rm
was suggested by Reese (1997). Because of the lack of
measured k
rm
, a value of 0.0005 was used. The values of E
m
are obtained by correlations
with E
i
.

Further discussion on this correlation is provided in chapter V. The main
parameters need to represent the weak rock p-y curves using Reese criterion are tabulated
in Tables 3-9 to 3-11 for the three test shafts. The depth is from the top of rock for shaft
#4 at Dayton and shaft #2 at Pomeroy-Mason. However, the depth for shaft #1 at
Pomeroy-Mason is from the ground level.
Table 3-9 Rock Properties for LPILE Analysis
Test Shaft Depths (ft) γ

(pci) RQD E
m
(psi) q
u
(psi) k
rm
Type
0 – 7 0.038 8% 38142 5668 0.0005 Shale Shaft #4
Dayton 7 – 18 0.038 53% 98102 5668 0.0005 Shale
0 – 17.3 0.035 SPT =7.1 - φ = 30 k = 20 Sand
17.3 – 26 0.035 SPT =13.5 - φ = 32 k = 60 Sand
26 – 47.5 0.057 60% 6462 905 0.0005 Shale/Siltstone
47.5 – 54.5 0.049 8% 654 52.8 0.0005 Shale(Mudstone)
54.5 – 63.9 0.047 23% 654 7.0 0.0005 Shale(Mudstone)
Shaft #1
Pomeroy –
Mason
63.9 – 66 0.05 38% 954 27.8 0.0005 Shale(Mudstone)
0 – 21.9 0.059 44% 23885 3797 0.0005 Shale
21.9 – 29.4 0.06 32% 102807 9073 0.0005 Sandstone
29.4 – 37.3 0.049 51% 88.3 19 0.0005 Claystone
37.3 – 48 0.047 20% 55.7 44.3 0.0005 Claystone
Shaft #2
Pomeroy –
Mason
48 – 56.8 0.055 86% 6170.9 826.2 0.0005 Claystone

101
The information of test shafts used in LPILE analysis is summarized in Table 3-10.
By inputting properties of rock, soil, and shaft (summarized in Table 3-9 and Table 3.10)
into LPILE, the load-deflection curves at the loading point can be obtained. Nonlinear
shaft stiffness option is selected. The comparisons of the predicted load-deflection curves
and the measured curves are provided in Figs. 3-36 through 3-38. The predicted
deflections are larger than the measured values for shaft #1 of Pomeroy-Mason load test
according to Fig. 3-37. However, it should be noted that a 26 feet thick layer of sand
exists above the rock mass when shaft #2 was under test. This may impede full
interaction of drilled shaft and rock, as evidenced in Fig. 3-12 where the deflection at the
top rock almost is close to 0.
On the other hand, the predicted deflections are much less than the measured values
for the other two test shafts. This may be due to inaccurate determination of input
parameters or due to the unsuitability of Reese weak rock p-y criterion for the two tests.
To explore the first reason, the rock parameters are adjusted to match the measured
results. As shown in Fig. 3-36 and Fig. 3-38, good matches are obtained when the
modulus and unconfined compressive strength of top rock layer of Dayton site and those
of Pomeroy-Mason site are reduced to 2% and 1% of the original values provided in
Table 3-9, respectively. One can see that a dramatic reduction of rock properties is
needed to match the load test results.




102
Table 3-10 Test Drilled Shafts Information
Test Shaft Depths (ft)
start from
loading
point
Diameter
(ft)
Loading
Eccentricity
(ft)
Reinforcement Concrete
Strength
(psi)
Shaft #4
Dayton
0 - 18 6 0 34#11 4500
0 – 58.4 8.5 32.4 28#18 with 1
inch thick casing
5115 Shaft #1
Pomeroy –
Mason 58.4 – 98.4 8 32.4 28#18 5115
0 – 53.1 8.5 53.1 28#18 with 1
inch thick casing
5115 Shaft #2
Pomeroy –
Mason 53.1 – 109.9 8 53.1 28#18 5115

0
200
400
600
800
1000
1200
0 0.05 0.1 0.15
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted
Predicted - Adjusted

Figure 3-36 Prediction of load-deflection curve of shaft #4 of Dayton load test using
Reese weak rock p-y criterion
103
0
50
100
150
200
250
300
0 1 2 3
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 3-37 Prediction of load-deflection curve of shaft #1 of Pomeroy-Mason load test
using Reese weak rock p-y criterion
0
50
100
150
200
250
300
0 1 2 3 4
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted
Predicted - Adjusted

Figure 3-38 Prediction of load-deflection curve of shaft #2 of Pomeroy-Mason load test
using Reese weak rock p-y criterion


104
3.5 Summary and Conclusions
In this chapter, two lateral load tests results of fully instrumented drilled shafts at
Dayton site and Pomeroy-Mason site were presented. The details of instrumentation used
and load test procedures were also presented. Based on an extensive evaluation of several
existing methods, the piecewise polynomial curve fitting technique is recommended for
deriving p-y curves from strain data obtained during lateral load test. The accuracy of the
deduced p-y curves using the recommended procedure was validated against the two load
tests. Additionally, the interim weak rock p-y criterion proposed by Reese (1997) was
evaluated as well. The following conclusions can be made.
1. Due to lack of actual lateral load tests of drilled shafts in rock, the two lateral load
tests results presented here contribute to understanding and development of pertinent
p-y curves.
2. The recommended method for deriving p-y curves from strain data of load test
includes using high order polynomial curve fitting to obtain y and using piecewise
polynomial curve fitting to obtain p. The superior results of this recommended
method was supported by the smallest error on the predictions of load-deflection
curves among all other methods.
3. The deduced p-y curves from actual load test data is shown to result in reasonable
predictions of the lateral response of drilled shafts for the two load tests.
4. The evaluation of the interim weak rock p-y criterion (Reese, 1997) indicates that this
p-y criterion tends to under predict the deflections of drilled shafts in rock.

105



CHAPTER IV
LATERAL CAPACITY OF DRILLED SHAFTS IN ROCK

The ultimate capacity of laterally loaded drilled shafts in rock has received very little
attention in literature. The reason, as pointed out by Carter and Kulhawy (1992), perhaps
is that the lateral design is governed largely by displacement considerations. Nevertheless,
reliable estimate of lateral capacity is still very important, so that the margin of safety on
capacity consideration can be ascertained. Moreover, deriving p-y curve criteria requires
that theoretical values of ultimate resistance of rock.
In order to develop theoretical equations for computing the ultimate capacity, it is
essential to identify failure modes of rock surrounding the laterally loaded drilled shaft.
For this purpose, a 3-D finite element model (FEM) is developed to simulate the response
of laterally loaded drilled shafts in rock. The load test result from Dayton site is
employed to validate the FEM simulation techniques. Based on calibrated FEM analysis
techniques, stress and deformation fields around the shaft are used to develop failure
modes of rocks surrounding laterally loaded shafts. Both Hoek-Brown strength criterion
of rock mass and an empirical equation for estimating the side shear resistance of
shaft/rock interface are employed to derive the theoretical equations for compute the
ultimate resistance of rock. Finally, a method for estimating ultimate lateral capacity of a
drilled shaft in rock and/or soils is developed.
106
4.1 3-D Finite Element Modeling and Validation
Although, there are several analytical methods available for analyzing the response
of laterally loaded drilled shafts in rock, such as p-y analysis (Reese, 1997) and elasto-
plastic solution (Zhang, et al. 2000), finite element method (FEM) nevertheless provides
a powerful tool for simulating complete nonlinear and plastic behavior of rocks subjected
to laterally loaded drilled shafts. In early stage of FEM applications, Randolph (1981)
presented a 2D FEM solution by modeling soil as an elastic continuum and the pile as an
elastic beam. With the advances of FEM computing techniques, it is now possible to use
3D FEM modeling technique. For example, Wakai et al. (1999) presented a 3D FEM
analysis of laterally loaded piles in soils by modeling soil as elasto-plastic material and
pile as linear elastic material. Recently, Wallace et al (2002) presented a 3D FEM
analysis to simulate a lateral load test using ABAQUS where soils is modeled as with
Mohr-Coulomb model. However, there is a lack of finite element study on laterally
loaded drilled shafts socket in rock mass.
A 3D FEM model to simulate the response of a single drilled shaft under lateral
loads in rock using ABAQUS is presented here. The yielding of rock mass is modeled
using modified Drucker-Prager model (CAP model). To verify the FEM model, the
lateral load test result at Dayton site is employed to compare 3D FEM model predictions.
The FEM model could be used for deriving theoretical p-y curves as well as for modeling
complicated drilled shaft group.



107
4.1.1 3-D FEM Modeling
4.1.1.1 Finite Element Meshes
Fig. 4-1 shows the meshes of a single drilled shaft-rock system generated using
software package PATRAN. The drilled shaft is modeled using second order 15-node
triangular prism elements C3D15, as depicted in Fig. 4-2 (a). The second order elements
are found to be more appropriate for modeling the shaft than the first order elements,
such as 6-node triangular prism elements (C3D6) and 8-node brick elements (C3D8),
because the trial simulation with these first order elements suffered convergence problem.
The rock mass is modeled using 8-node brick elements C3D8 shown in Fig. 4-2 (b). The
bottom of rock mass is fixed. The outer boundary of rock mass are modeled using 8-node
infinite boundary elements CIN3D8, which is depicted in Fig. 4-2 (c). Because the lateral
response of a drilled shaft is mainly dependent on the rock near the top of the shaft, it is
preferred to use a fine mesh for the top layer of rock mass, as shown in Fig. 4-1.

Figure 4-1 Finite element meshes of a drilled shaft-rock system

L

0.7L

D 5D 5D

Drilled Shaft:
C3D15

Rock: C3D8
Infinite
Element:
CIN3D8

Side View Top View

Fixed Bottom

108

(a) C3D15 (b) C3D8 (c) CIN3D8
Figure 4-2 Finite elements for (a) drilled shaft, (b) surrounding rock, and (c) outside
boundary of rock

Wallace et al. (2002) found that, by studying the effect of different mesh sizes on the
shaft-soil contact pressure, 11D (D is diameter of drilled shaft) could be used to define
the dimension of the diameter of the entire mesh. The depth of rock mass beneath the
drilled shaft tip is 0.7 L, where L is embedment length of the drilled shaft. This is based
on findings from Trochanis et al. (1988).
4.1.1.2 Constitutive Models
The drilled shaft is modeled as a cylinder with linear elastic material property.
ABAQUS built-in concrete model with embedded rebars was used initially in an effort to
simulate the nonlinearity of reinforced concrete drilled shafts; however, the various
numerical convergence problems are prevalent. The modified Drucker-Prager model
(CAP model) was employed to model the rock mass. The ABAQUS built-in CAP model
is intended to model cohesive geological materials that exhibit pressure-dependent yield,
such as soils and rocks. It is based on the addition of a cap yield surface to the Drucker-
Prager plasticity model, which provides an inelastic hardening mechanism to account for
109
plastic compaction and helps control volume dilatancy when the material yields in shear
(ABAQUS, 1998).
According to ABAQUS User’s Manual (1998), the yield surface of the CAP model
is defined in two principal segments: a pressure-dependent Drucker-Prager shear failure
segment and a compression cap segment, as shown in Fig. 4-3. The Drucker-Prager
failure segment is a perfectly plastic yield surface (no hardening). A nonassociated platic
flow is used for the Drucker-Prager yield segment; and an associated plastic flow is used
for the CAP segment. The deviatoric stress measure t is defined as
] )
q
r
)(
K
1
1 (
K
1
1 [ q
2
1
t
3
− − + = (4-1)
where
trace
3
1
p − = (σ) (4-2)
q = (3/2 S : S)
0.5
(4-3)
r = (9/2 S : S · S)
1/3
(4-4)
S = σ + p I (4-5)
K is a material parameter that controls the dependence of the yield surface on the value of
the intermediate principal stress, and 0.778 ≤ K ≤ 1.0. The default value of K is 1.0.
The main input parameters for a CAP model are E
m
(elastic modulus of rock mass),
d (material cohesion in p-t plane), β (material friction angle in p-t plane), and a cap
hardening curve which is obtained from a hydrostatic compression test. The values of d
and β can be calibrated from two uniaxial or triaxial compression tests. Alternatively,
they can be calibrated from traditional Mohr-Coulomb cohesion, c, and friction angle, φ
using the following equations.
110
ϕ −
ϕ
=
sin 1
cos
c 2 d (4-6)
|
|
.
|

\
|
ϕ −
ϕ
= β
sin 3
sin 6
arctan (4-7)

Figure 4-3 Cap model: yield surface in the p-t plane (ABAQUS, 1998)
4.1.1.3 Shaft-Rock Interaction and Loading
The interaction between drilled shaft and rock mass is modeled using a surface
based contact. The shaft surface is treated as master surface, while the rock surface is
taken as slave surface. In tangential direction of the shaft-rock interface, the frictional
interaction is simulated using linear Coulomb friction theory, which needs an input of a
coefficient of friction. In normal direction, the contact surfaces transmit no contact
pressure unless the nodes of the slave surface are in contact with the master surface.
There is no limit to the magnitude of contact pressure that can be transmitted when the
surfaces are in contact (ABAQUS, 1998). The lateral loads are applied step by step after
the gravity weight of the shaft-rock system is activated.
Deviatoric
Stress
Equivalent
Pressure
111
4.1.2 Validation of the 3D FEM Model
4.1.2.1 Load Test and Input Parameters
The field lateral load test results from Dayton test is used to validate the above 3D
FEM modeling technique. The test drilled shaft is 6 feet in diameter and 18 feet in rock
socket length. The elastic modulus and Poisson’s ratio of the drilled shaft are 3800 ksi
and 0.15, respectively. The rock mass at the test site is gray shale interbedded with
limestone. The details of the test shaft dimensions and rock properties are presented in
Fig. 4-4. An unconfined compression test on an intact rock sample provided q
u

(unconfined compressive strength) of 5668 psi and E
i
(elastic modulus of intact rock) of
590 ksi. The GSI and RMR values are also summarized.

Figure 4-4 Drilled shaft dimension and rock profiles of the load test at Dayton
Based on the unconfined compression test, Mohr-Coulomb cohesion and friction
angle of the rock mass, including the effects of joints, fillings, cracks, and other
secondary structures of rock, can be correlated with Hoek-Brown (Hoek, et al., 2002)
rock strength criterion using q
u
, E
i
, and GSI. By using free software RocLab (Rocscience,
2002), Mohr-Coulomb cohesion, c, and friction angle, φ, of the two layers of rock mass
7 ft
11 ft
Rock
Soft gray shale, slightly
weathered to decomposed,
laminated with limestone
Medium gray shale,
slightly weathered,
laminated with hard
limestone
Drilled Shaft RQD RMR
89
GSI
8%-
0%
30.5 40.5
53% 51 61
2 ft
6 ft
Load
112
were determined and are summarized in Table 4-1. The Drucker-Prager cohesion, d, and
friction angle, β are estimated using Eqs. (4-6) and (4-7). The elastic modulus of rock
mass E
m
is also obtained using RocLab, which employs Hoek et al. (2002) empirical
correlation equation based on q
u
and GSI.
The Poisson’s ratio of the rock mass is assumed to be 0.3. Due to the lack of
hydrostatic tests on rock mass samples, the stress-strain curve of the unconfined
compression test is used to represent isotropic hardening curve. The input pairs of the
hydrostatic compression yield stress (p
b
, in psi) and volumetric inelastic strain (ε
in
) for the
hardening curve are as follow: (1000, 0), (1500, 0.004), and (1900, 0.01). To simplify the
contact interaction, the entire side interface between the drilled shaft and rock mass is
simulated using one contact modeling; and a second contact modeling is used to simulate
the interaction between shaft tip surface and rock mass. A coefficient of friction of 0.5 is
used based on the Mohr-Coulomb friction angle of 27º of the bottom rock layer. The
water table was above the surface of rock mass; therefore, the effective unit weight of
0.034 pci is used in analysis.
Table 4-1 Rock Mass Properties
Mohr-Coulomb
Drucker-Prager
Rock Layer
c (psi) φ (º) d (psi) β (º)
E
M
(ksi)
Top layer 154 20 440 38 241
Bottom layer 251 27 819 47 590

4.1.2.2 Simulation and the Results
The mesh sensitivity is studied by comparing FEM results based on varying the
number of elements. It is found that the mesh with 1885 elements provides almost the
same results on lateral deflections of the drilled shaft under lateral loads as the mesh with
113
2765 elements, as shown in Fig. 4-5. Therefore, it is believed that FEM simulation results
are not sensitive to mesh density.
0
200
400
600
800
1000
1200
0 0.2 0.4 0.6 0.8
Deflection (in)
L
o
a
d

(
k
i
p
s
)
1885 Elements
2765 Elements

Figure 4-5 Mesh convergence
With the constitutive model parameters described previously, the FEM predicted
load-deflection curve at the shaft head is compared with the measured one, as shown in
Fig. 4-6. It can be seen that the predicted load-deflection curve is comparable to the
measured. A good match of the deflection profiles along depth between the FEM
predicted values and the inclinometer measured values is also evident, as shown in Fig. 4-
7. Although, in general, the 3D FEM model predicted the deflection response of drilled
shafts in rock under lateral loads, the early portion of nonlinear lateral response was not
captured. This was attributed to the lack of a hydrostatic compression test to determine an
accurate hardening curve for the rock mass. A back analysis is performed to match the
entire range of lateral deflection response of the drilled shaft. With improved hardening
curves, a good match of the load-deflection curve at the head of the shaft is obtained, as
114
shown in Fig. 4-6. The hardening curve of the top rock layer is adjusted to the following
p
b
(in psi) and ε
in
pairs: (120, 0), (260, 0.01), and (400, 0.02). The cohesion of the top
rock layer is reduced to 100 psi.
0
200
400
600
800
1000
1200
0 0.05 0.1 0.15
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
ABAQUS
ABAQUS - Match

Figure 4-6 Comparison of load-deflection curves at shaft head
0
5
10
15
20
-0.05 0 0.05 0.1 0.15
Deflection (inch)
D
e
p
t
h

(
f
t
)
Measured-1126 kips
ABAQUS-1126 kips

0
5
10
15
20
-0.02 0 0.02 0.04 0.06 0.08
Deflection (inch)
D
e
p
t
h

(
f
t
)
Measured-705 kips
ABAQUS-705 kips

(a) Load at 1126 kips (b) Load at 705 kips
Figure 4-7 Comparison of deflection profiles at the load of (a) 1126 kips and (b) 705 kip
115
4.2 Failure Modes of Rock Subjected to Loading from Laterally Loaded Drilled Shafts
Very few research has been focused on the failure modes of rock mass surrounding
the drilled shafts under lateral loads. Generally, rock-socketed drilled shafts under lateral
loads may fail either due to structural failure of shaft or strength failure of rock. Attention
here is given to the later type of failure.
Carter and Kulhawy (1992) assumed that compressive failure will occur for rocks in
front of shaft and shear failure will occur for interface between shaft and rock. Zhang et
al. (2000) adopted Carter and Kulhawy (1992)’s assumption in their relevant theoretical
work. However, Reese (1997) assumed compressive failure of rock, but ignored shear
failure between shaft and rock. For jointed rock mass, To et al. (2003) proposed a wedge
type failure.
Two types of rock will be investigated in FEM simulation of rock failure modes: one
is intact rock or highly weathered rock that can be modeled as isotropic and
homogeneous rock, and the other one is jointed rock with a set of parallel weak planes.
For each type of rock, two conditions are considered: one is that the drilled shaft is fully
socketed into rock, the other one is that the drilled shaft socketed into rock with
overburden soils.
4.2.1 Failure of Isotropic and Homogeneous Rock without Overburden Soils
4.2.1.1 Finite Element Modeling
A drilled shaft with 30 in. diameter and 15 feet socket length is modeled using linear
elasticity with elastic modulus of 4000 ksi. The Modified Drucker-Prager Model (CAP
Model) is utilized to model the isotropic and homogeneous rock. The elasticity modulus
of rock is assumed to be 2000 ksi; and the Mohr-Coulomb cohesion and friction angle are
116
assumed to be 1 ksi and 27 degree, respectively. The element type C3D15 and C3D8 are
used to model drilled shafts and rock, respectively. The load applied to the top of shaft is
increased from 300 kips to 3000 kips, at which time computer analysis failed to converge.
Therefore, the analysis results at 3000 kips of loading level, are used to examine the
failure conditions in the rock.
4.2.1.2 Failure at the Top Layer of the Rock
At the failure load of 3000 kips, the rock in front of the shaft shows a forward and
upward movement according to the deformed and undeformed rock mass shown in Fig.
4-8.

Figure 4-8 The forward movement of rock mass
Apparently, it is highly likely that without overburden soil layer on top of rock, the
rock in front of the shaft will move forward under the lateral loads. Fig. 4-9 shows the
Rock
Loading
Direction
Shaft
Undeformed
Deformed
inch
117
front view of the upward movement of rock mass, in which the movement is scaled up to
150 times of the original magnitude of movement.
From Figs. 4-8 and Fig. 4-9, it can be seen that a wedge type of rock mass has the
tendency to move away from its original position. However, for this to happen, the rock
in vertical xz plane, shown in Fig. 4-8, has to reach tensile failure; and the inclined planes
of the wedge have to reach shear failure. In order to check whether the rock in xz plane
surrounding the shaft can reach tensile failure, the stress distribution in rock is shown in
Fig. 4-10. It can be seen that the tensile stress is developed. The magnitude of the
maximum tensile stress is 2763 psi which is about half of the maximum compressive
stress 5830 psi in front of the shaft. Usually, the uniaxial tensile strength of rock is about
one tenth of the uniaxial unconfined compressive strength of rock. Therefore, in this case,
the tensile failure is more likely to appear before the onset of compressive failure.

Figure 4-9 Front view of upward movement of rock mass
Rock
Rock mass in front of shaft
118

Figure 4-10 Y direction stress distribution in xz plane
Additionally, in order to confirm the proposed wedge type failure, one needs to
check if the shear stress is concentrated on shear planes. Fig. 4-11 shows the maximum
shear stress distribution on yz plane. It can be seen that the distributions of shear stresses
at three planes are similar, which means that the bottom surface of the wedge is likely to
be a plane rather than a curved surface. Although, it can not be concluded from Fig. 4-11
that the rock will be sheared out, the upward and forward movements of this portion of
rock, together with the depicted shear distribution, implies a possible failure mode shown
in Fig. 4-12.

Rock
Shaft
Tensile
Stress
119

Figure 4-11 Maximum shear stress distribution on yz plane


Figure 4-12 Proposed wedge type failure model for the top layer of rock
F
a
F
net
W

F
n
F
s
F
sb
F
nb
θ
β
H

D
σ

v0


Shaft
Rock
120
4.2.1.3 Failure of Rock at Great Depth
Generally, the wedge type of failure is unlikely to occur in rock at great depth due to
overburden pressure from top layers of rock. Possible failures of in-depth rock can be
tensile failure, compressive failure, or both. A separate FEM model was established in
order to realize material failure of rocks.
A 5 feet long drilled shaft segment with 30 inch diameter is embedded in isotropic
rock mass with the same properties as the rock used in the previous model. The lateral
load of 2000 kips, considered large enough to induce failure tensile stress in rock mass, is
applied at the top of the shaft segment. The in-situ vertical pressure, which is equivalent
to vertical pressure at 6 ft deep in rock mass, plus self weight of shaft and rock, is applied
at the top surface of the shaft and rock.
As can be seen in Fig. 4-13, the maximum tensile stress in rock surrounding the
shaft is about 42% of the maximum compressive stress. The magnitude of the maximum
tensile stress is 1610 psi, which is considered large enough to produce crack since the
tensile strength of most rocks ranges from 725 psi to 2900 psi (Afrouz, 1992). However,
the compressive strength is usually one order of magnitude higher than the tensile
strength. Therefore, the tensile failure of the rock mass will occur first for in-depth rock
layers.
Even with initiation and propagation of cracks in rock; the rock mass in front of
shaft is still able to sustain loads. In order to identify the stress redistribution after the
crack appears, the FEA model is modified to simulate a 10 inch long crack at the tensile
failure areas. Fig. 4-14 shows the predefined cracks after applying 2000 kips of lateral
loading. The stress in loading direction is redistributed as shown in Fig. 4-15, from which
121
it can be seen that the maximum tensile stress is decreased from 1610 psi to 540 psi.
Because the tensile stress of 540 psi will be less than the tensile strength of the rock, the
cracks would stop to propagate and the loads is redistributed to the rock in front of shaft.
Actually, the maximum compressive stress is increased from 3780 psi to 3850 psi and the
maximum deflection increased from 0.0553 in. to 0.0581 in.

Figure 4-13 Stress distribution at Y direction of in-depth rock layer
Other than the tensile stress and compressive stress, the friction between shaft and
surrounding rock mass is also developed during loading, as shown in Fig. 4-16. Although
the maximum relative movement between shaft and rock is at point A, shown in Fig. 4-16,
the maximum shear stress is developed at point B. Due to the applied lateral load, the
normal pressure on point A is smaller than that at other positions in the interface surface.
Shaft
Rock
Loading
Max. Compressive Stress
Max. Tensile
Stress
Deformed
Undeformed
122

Figure 4-14 The predefined cracks


Figure 4-15 The stress redistribution at Y direction of in-depth rock layer after crack
Rock
Shaft
Crack
Load Direction
Shaft
Rock
Max.
Compressive
Stress
Tensile
Stress
123

Figure 4-16 Friction distribution on the shaft-rock interface
4.2.1.4 Suggested Failure Mode for In-depth Layer of Isotropic Rock
Based on the above FEA analysis results, it is concluded that tensile failure of rock
would occur first. Then the maximum friction between shaft and rock would be reached.
Finally, the maximum compressive stress would reach the compressive strength of rock.
Therefore, the ultimate capacity of in-depth rock mass in resisting lateral loaded shaft can
be credited to compressive strength and friction between shaft and rock. This is in
conformance with the assumption made by Carter and Kulhawy (1992). Based on the
distributions of compressive stress shown in Fig. 4-13 and friction shown in Fig. 4-16,
the normal and shear stress distribution at failure is proposed as that shown in Fig. 4-17.
B
Side Wall
of Rock
Loading
A
124

Figure 4-17 Suggested stress distribution at failure at great depth
4.2.2 Failure of Jointed Rock without Overburden Soils
4.2.2.1 Jointed Material Model in ABAQUS
The jointed material model in ABAQUS is intended to provide a simple continuum
model for a material containing a high density of parallel joint surfaces where each
system of parallel joints is associated with a particular orientation, such as sedimentary
rock. The jointed material model assumes that the spacing of the joints of a particular
orientation is sufficiently close compared to characteristic dimensions in the domain of
the model such that the joints can be smeared into a continuum of slip systems.
The model is intended primarily for applications where the stresses are mainly
compressive. A joint will be open when the stress normal to the joint tries to become
tensile. Once a joint opens, the retained shear modulus on the joint is governed by the
shear retention parameter, f
sr
. If f
sr
= 0, it means that no shear stiffness is associated with
Active earth pressure

β
σ(β) = p
L
sin β
p
L
Applied load
Normal stress
Shear stress distribution
α
τ(α) = τ
max
sin 2α
τ
max

125
open joints, on the other hand, when f
sr
= 1.0, it corresponds to elastic shear stiffness in
open joints.
The failure surface for sliding on joint system a is defined by:
0 d tan p f
a a a a a
= − β − τ = (4-8)
where τ
a
= the shear stress on the joint surface; p
a
= the normal pressure across the joint;
β
a
= the friction angle for system a, and d
a
= the cohesion for system a.
In addition to the joint systems described in the above, the jointed material model
includes a bulk material failure mechanism, which is based on the Drucker-Prager failure
criterion. The friction angle, dilation angle, and cohesion for the bulk material constitute
the required input parameters.
4.2.2.2 The Modeling of Drilled Shafts in Jointed Rock Mass
The same physical model used in the previous section for drilled shaft in isotropic
rock mass is used for this study. On notable change, however, is that the rock mass model
is changed from CAP Model to Jointed Material model. For jointed material model, the
bulk failure, joint surface failure, and shear retention are the required input parameters. In
this study, one horizontal bedded joint system is assumed. For the bulk failure, the
cohesion is 2 ksi and friction angle is 27 degree. For joint surface failure, same friction
angle and 50% of cohesion of the input for bulk failure are assumed.
When the shear retention factor was assumed to be 0.8, the model can only be
converged for the applied lateral load up to 57 kips. This is a small load for a 3 ft
diameter and 15 ft long drilled shaft socketed into rock. Considering that the weak
surface of jointed rock mass has been characterized by joint surface failure criterion, the
shear retention is set to be 1.0 so that no shear stiffness is reduced when tensile stress is
126
developed across the joint surface. This change in modeling technique allows the applied
load to be up to 1000 kips.
4.2.2.3 Failure at Top Layer of Jointed Rock Mass
As it can be expected, the top layer of rock mass in front of drilled shaft experiences
the forward movement. Similar to the case of isotropic rock, the rock mass, which is
mobilized to move forward, exhibits upward movement beyond the ground level. The
mobilized maximum upward movement is 0.01 inch under 1000 kips of lateral loading.
This is much larger than the upward movement of 0.0076 inch for the shaft in isotropic
rock mass under 3000 kips loading. One could infer that the joint plane significantly
increases the displacement of rock mass under lateral loads.
Similar to the situation of isotropic case, it is reasonable to guess a wedge type of
rock mass that has the tendency to move out from its original position. In order to have a
wedge to be sheared out, the shear surface has to be formed and the backside of the
wedge has to be in tensile failure. Fig. 4-18 presents the tensile stress developed at the
backside of the wedge. It can be seen that the maximum tensile stress is around 1181 psi,
which is about 31% of the maximum compressive stress of 3858 psi developed in front of
the shaft. The tensile failure of the backside of the wedge, therefore, is more likely to
occur before the compressive failure of the rock in front of the shaft can materialize. This
is simply due to the fact that the uniaxial tensile strength of rock is usually one tenth of
the uniaxial unconfined compressive strength of rock.
It should be noted that the shear plane of the wedge has to be developed in order to
form a wedge type of failure. Fig. 4-19 shows the contours of maximum shear stress
concentration. It can be seen that the possible shear plane can be 45° to 60°. The upward
127
movement, forward movement, possible tensile failure of rock on xz plane, and the
maximum shear concentration all support the notion of the development of a wedge type
of failure.

Figure 4-18 Tensile stress on xz plane for jointed rock mass case

Figure 4-19 Maximum shear stress concentrations for jointed rock mass case
Shaft
Rock
Load
Tensile
Stress
Shaft Rock
Loa
128
4.2.2.4 Suggested Failure Mode for Jointed Rock Mass
Based on the above analysis results of drilled shaft in horizontally jointed rock mass,
the wedge type of failure is suggested for the top rock failure under lateral loads. For
jointed rock mass with non-horizontal joints, several similar 3D FEA analyses were
carried out, unfortunately, similar analysis results as the one of horizontally jointed rock
mass were obtained. It appears that the FEA modeling technique is not capable of
identifying the effects of joint direction. At this stage, the suggested failure mode for
horizontally jointed rock could be applied to rock mass with different joint directions
with a great caution.
4.2.2.5 The In-depth Failure of Jointed Rock Mass
Similar to the analysis for in-depth failure of isotropic rock mass, a shaft cylinder
with 5 feet length and 30 inch diameter embedded in jointed rock mass is created to
simulate the in-depth situation. The overburden pressure, which is equivalent to the
pressure at 6 feet deep in the jointed rock mass, is applied to the top of rock-shaft system
together with the selfweight of the rock and shaft. The stress distribution at loading
direction is similar to the case of isotropic rock, as shown in Fig. 4-13. The computed
maximum tensile stress surrounding shaft was 33% of the maximum compressive stress.
The magnitude of tensile stress of 870 psi is believed to be able to produce cracks, as the
uniaxial tensile strength of most rocks ranges from 725 psi to 2900 psi (Afrouz, 1992).
Cracks and even tensile failure, therefore, are likely to appear earlier than other types of
failure for in-depth jointed rock mass.
When the cracks start to propagate, the rock mass in front of the shaft would still be
able to sustain the applied load. In order to identify the stress redistribution after the crack
129
propagates, the FEA model is modified to simulate a 10 inch long crack at the tensile
failure areas. Similar to the case of isotropic rock, the maximum tensile stress is
decreased from 870 psi to 326 psi, which is less than the tensile strength of most rock. On
the other hand, the maximum compressive stress is increased from 2620 psi to 2790 psi.
Other than the tensile stress and compressive stress, the friction between shaft and
surrounding rock mass is also developed during loading. Similar to the situation of
isotropic rock, the maximum shear stress is developed at point B rather than at point A, as
shown in Fig. 4-16.
4.2.2.6 Suggested Failure Mode for In-depth Jointed Rock Mass
Based on the above FEA analysis results, one can conclude that failure mode for
isotropic rock shown in Fig. 4-17, could be used for horizontally jointed rock mass. For
those jointed rock with non-horizontal joints, the suggested in-depth failure mode for
horizontally jointed rock may be used with caution.
4.2.3 Failure of Isotropic and Jointed Rock with Overburden Soils
For drilled shafts socketed into isotropic or jointed rock with overburden soils, the
in-depth failure must be same as the condition of without overburden soils. The top rock
layer, however, may experience in-depth failure due to the overburden pressure.
Therefore, both top layer failure and in-depth failure, shown in Figs. 4-12 and 4-17,
should be checked for the top layer rock mass; and the smaller value should be adopted.

4.3 Rock Strength Criteria
Several rock strength criteria have been developed in the past. Table 4-2 provides a
summary of more prevalent rock strength criteria cited in literature. Among them, Hoek-
130
Brown criterion appears to be widely accepted since it has been revised many times and
correlated with field data from its inception in 1980 (Hoek and Brown, 1980).
Ramamurthy et al. (1985) criterion is particularly suitable for jointed rock mass; however,
it requires conducting well-organized lab tests on samples with weak planes.
Table 4-2 Summary of Rock Strength Criteria
Criteria Applications Advantages Limitations
Hoek-Brown
(Hoek et al.
2002)
Intact rock and
highly fractured
rock mass
Improved several times;
widely referred; calibrated
with numerous data
Not applicable for
anisotropic rocks
Ramamurthy
et al. (1985)
Intact rock, rock
mass, and
jointed rock
Applicable for anisotropic
rocks
Needs lab test on
jointed rock samples
Johnston
(1985)
Intact rock, rock
mass
Only uniaxial strength of
intact rock is required
Discontinuities are not
well considered; not
applicable for
anisotropic rocks

4.3.1 Hoek-Brown Criterion
Hoek-Brown criterion has been developed and improved several times since its first
version published in 1980. In 2002, Hoek et al. consolidated previous revisions on the
criterion and proposed the following general form of Hoek-Brown criterion.
a
ci
'
3
b ci
'
3
'
1
s m
|
|
.
|

\
|
+
σ
σ
σ + σ = σ (4-9)
where σ

1
= the major principal stress at failure; σ

3
= the minor principal stress or
confining pressure; σ
ci
= the uniaxial unconfined compressive strength of the intact rock;
m
b
, s and a = material constants that depend on the characteristics of rock mass and can
be estimated as follows (Hoek, et al., 2002).
i
r
b
m
D 14 28
100 GSI
exp m
|
|
.
|

\
|


= (4-10)
131
|
|
.
|

\
|


=
r
D 3 9
100 GSI
exp s (4-11)
) e e (
6
1
2
1
a
3 / 20 15 / GSI − −
− + = (4-12)
where D
r
is a factor depending upon the degree of disturbance to which the rock mass has
been subjected due to blast damage and stress relaxation. It varies from 0 for undisturbed
in situ rock masses to 1 for very disturbed rock masses. For deep foundation analysis,
although the excavation releases the horizontal earth pressure on rock masses, the
pouring of concrete restores the pressure. Therefore, the value of D
r
is selected as 0 for
applications in deep foundation analysis.
The material constant m
i
used in Eq. 4-10 can be determined from triaxial tests
(Hoek and Brown, 1997). When no laboratory test data are available, it can also be
estimated from Table 4-3. The values shown in Table 4-3 can be varied ±2.
Table 4-3 Values of Constant m
i
for Intact Rock (After Marinos and Hoek, 2000)
Sedimentary Rocks
Anhydrite 12 Breccias 20 Chalk 7 Claystones 4
Conglomerates 21 Dolomites 9 Greywackes 18 Gypsum 10
Marls 7 Sandstones 17 Siltstones 7 Shales 6
Crystalline
Limestones
12 Micritic
Limestones
8 Sparitic
Limestones
10
Metamorphic Rocks
Amphibolites 26 Gneiss 28 Hornfels 19 Marble 9
Metasandstone 19 Migmatite 29 Phyllites 7 Quartzites 20
Schists 10 Slates 7
Igneous Rocks
Agglomerate 19 Andesite 25 Basalt 25 Breccia 19
Dacite 25 Diabase 15 Diorite 25 Dolerite 16
Gabbro 27 Granite 32 Granodiorite 29 Norite 20
Obsidian 19 Peridotite 25 Porphyries 20 Rhyolite 25
Tuff 13

132
4.3.2 Converting Hoek-Brown Criterion to Mohr-Coulomb Criterion
Mohr-Coulomb criterion expressed in terms of shear strength and normal stress is
commonly used in geotechnical engineering disciplines. The Hoek-Brown criterion could
be converted to equivalent Mohr-Coulomb criterion, as suggested by a method proposed
by Hoek (1990).
|
|
.
|

\
|
σ′ − σ′
τ
− = φ′
3 1
2
arcsin 90 (4-13)
φ′ σ′ − τ = ′ tan c
n
(4-14)
where σ
'
1
is in-situ vertical effective stress; σ
'
1
can be obtained using Eq. (4-9); and
ci b 3 1
2
3 1
3 n
m 5 . 0 ) ( 2
) (
σ′ + σ′ − σ′
σ′ − σ′
+ σ′ = σ′ (4-15)
) ( 2
m
1 ) (
3 1
ci b
3 n
σ′ − σ′
σ
+ σ′ − σ′ = τ (4-16)

4.4 Side Shear Resistance
According to the identified failure mode for in-depth rock mass layer, the side shear
resistance between rock and shaft will contribute to some portion of the ultimate
resistance of rock. However, there are very few published methods for predicting the
horizontal side shearing resistance in the rock/shaft interface. Briaud and Smith (1983)
measured the side shearing resistance by means of mounting pressure cells in the leading
front of drilled shaft embedded in the soil. However, there was no available analysis
method to account for side shear resistance of drilled shafts in rocks in the horizontal
direction.
133
Fortunately, there are a lot of empirical methods and a few theoretical analysis
methods available for estimating the side resistance of axially loaded drilled shafts.
Johnston and Lam (1989a) proposed a theoretical procedure to estimate the side shear
strength based on energy theory. An equivalent 2D model is used represent 3 dimensional
interaction between rock and shaft, as shown in Fig. 4-20. In this model, the upper half
can move only in a vertical direction against a spring which provides the constant normal
stiffness to the system. The stiffness of the spring is selected to match the rock mass
modulus and the socket geometry for a given drilled shaft problem, that is
( ) r
E
K
m
m
1
1 υ +
= . Where r is radius of shaft;
m
E and
m
υ are Young’s modulus and
Poisson’s ratio of rock mass, respectively. Based on Mohr-Coulomb strength criterion
and dilation energy theory, a series of formulas were developed to calculate side shear
strength by Johnston and Lam (1989a).
In this method, a lot of parameters of rock mass were required as follows. This
make the application of Johnston and Lam (1989a)’s method difficult.
=
u
q Uniaxial compressive strength;
=
sl
c Cohesion from triaxial test;
= φ
p
sl
Peak angle of friction from triaxial test;
=
sh
c Cohesion from direct shear test;
= φ
r
sl
Residual angle of friction from triaxial test;
= φ
p
sh
Peak angle of friction from direct shear test;
= φ
r
sh
Residual angle of friction from direct shear test;
134
= i Asperity angle;
= K Normal stiffness of rock mass;
= σ
0 n
Initial normal stress.

Figure 4-20 Concrete and rock joint
Comparing with complicated theoretical method proposed by Johnston and Lam
(1989a), a lot of simple empirical equations have been proposed.
4.4.1 Empirical Equations for Axially Loaded Drilled Shaft in Rock
There exist several empirical relations between the ultimate vertical side shear
strength, τ
max
, and unconfined compressive strength, σ
ci
, of the intact rock. The relation
could be expressed as τ
max
= α σ
ci
, where α = 0.25, according to Toh et al. (1987).
The power-curve relationships have also been proposed. Such as, τ
max
= 0.375σ
ci
0.515

(Rosenberg and Journeaux, 1976) and τ
max
= 0.22σ
ci
0.6
(Meigh and Wolshi, 1979), where
τ
max
and σ
ci
are in MPa.
Roughness of the wall of the excavated drill hole is an important factor controlling
the development of vertical side shear resistance. Researches by Pell et al (1980) and
Johnston and Lam (1989a and 1989b) have shown that the increase in side resistance
could be significant from increase in roughness of the wall of excavated hole. From the
135
study of rock socket in mudstone and sandstone, Williams et al. (1980) found that
smooth-sided sockets exhibit a brittle type of failure, whereas sockets having an adequate
roughness exhibit ductile failure. Williams and Pells (1981) suggested that rough socket
generates a locked-in normal stress such that there is practically no distinguishing
difference between residual and peak resistance.
Pells et al. (1980) have used the size and frequency of grooves in a socket wall to
classify wall roughness as R1, R2, R3, and R4, as defined in Table 4-4.
Table 4-4 Roughness Classes (After Pells et al., 1980)
Roughness
Class
Description
R1 Straight, smooth-sided socket, grooves or indentation less 1.00 mm deep
R2
Grooves of depth 1-4 mm, width greater than 2 mm, at spacing 50 to 200
mm
R3
Grooves of depth 4-10 mm, width greater than 5 mm, at spacing 50 to 200
mm
R4
Grooves or undulations of depth greater than 10 mm, width greater than
10 mm, at spacing 50 to 200 mm

Kulhawy and Phoon (1993) developed a relatively extensive load test database for
drilled shafts in soil and rock and presented their data both for individual shaft load tests
and as site-averaged data. They concluded that the power curve relationship is closer to
the real cases. On the basis of site-averaged data, the following equations for axially
loaded drilled shafts in rock was proposed (τ
max
and σ
ci
are in MPa).
5 . 0
ci max
45 . 0 σ = τ for mean behavior (4-17)
5 . 0
ci max
67 . 0 σ = τ for upper bound (4-18)
5 . 0
ci max
23 . 0 σ = τ for lower bound (4-19)
136
Based on Kulhawy and Phoon (1993)’s database, Zhang (1997) suggested the
following relation for smooth and rough socket conditions by using individual shaft load
test data rather than the site-averaged data.
5 . 0
ci max
20 . 0 σ = τ Smooth socket (R1, R2 or R3) (4-20)
5 . 0
ci max
80 . 0 σ = τ Rough socket (R4) (4-21)
4.4.2 Suggested Empirical Equation for Side Resistance in Horizontal Direction
As described above, empirical correlations between uniaxial compressive strength of
intact rock and unit shaft resistance of rock socketed drilled shafts measured in load tests
could be expressed in a general equation as follow.
β
ασ = τ
ci max
units in MPa (4-22)
For axially loaded drilled shafts in rock, the mobilization of side shear resistance is
depicted in Fig. 4-21. On application of an axial load to the shaft, the shaft and rock mass
will deform elastically until the shear stress at the interface causes slip. Fig. 4-21 also
shows a drilled shaft section after a slip of the shaft relative to the rock takes place.
Geometrical constrains require that this relative sliding to generate dilation at interface,
and subsequent increase in socket diameter. This dilation occurs against a surrounding
rock mass that must deform to compensate for enlargement of the socket diameter;
therefore, resulting in an increase in normal stress at the shaft/rock interface (Johnston
and Lam, 1989a), which significantly enhances the vertical side shear resistance.
Additionally, to overcome the roughness of the socketed wall, the side shear resistance
increases as the wall become rougher.
137

Figure 4-21 Displacement behavior of drilled shafts in rock (Johnston and Lam, 1989a)
For laterally loaded drilled shafts in rock, the relative displacement between the
shaft and rock socket wall is in general parallel to the groove direction. There would be
no or very small dilation taking place at the interface; therefore, roughness does not
contribute much to side resistance in horizontal direction. However, the normal stress at
the leading frontal side of drilled shafts will be increased because of the applied lateral
loads, as shown schematically in Fig. 4-22. Therefore, the following empirical equation
for predicting side shear resistance, τ
max
(psi), by Kulhawy and Phoon (1993), for mean
behavior of rock is adopted.
5 . 0
ci
max
42 . 5 σ = τ where the units are in psi (4-23)

Figure 4-22 Increased normal stress due to lateral load


Shaft
Rock
Rock
Shaft
Lateral
Load
Drilled
Shaft
Normal
Stress
138
4.5 Ultimate Resistance of Rock Mass
4.5.1 Ultimate Resistance of Rock Near Surface
For highly fractured rock mass and competent rock, wedge type failure model for
top rock layer was identified earlier in Fig. 4-12. In the figure, F
net
is the total net rock
resistance; D is the diameter of the drilled shaft; H is the height of the wedge; F
a
is the
active earth force exerted on the drilled shaft; F
s
is the friction force on the sides of the
wedge; F
n
is the normal force applied to the sides of the wedge and is assumed to be
equal to the at-rest earth force; F
sb
is the friction force on the bottom face of the wedge;
F
nb
is the normal force on the bottom face of the wedge which is to be determined
through force equilibrium on vertical direction; W is the weight of the wedge; σ
'
v0
is
effective vertical soil pressure due to overburden soil on the top of rock and it is equal to
zero if no overlying soils are present. According to force equilibrium in the loading
direction, the net rock reaction can be determined as follows:
a n nb sb s net
F sin F 2 cos F sin F sin cos F 2 F − θ − β + β + β θ = (4-24)
2
0 a a
) z H ( D K
2
1
F − γ′ = (4-25)
θ β γ′ + σ′ = sec tan H K
6
1
A K F
3
0 s 0 v 0 n
(4-26)
θ β φ′ γ′ + φ′ σ′ + ′ = sec tan tan K H
6
1
A tan K A c F
0
3
s 0 v 0 s s
(4-27)
β φ′ − β
β ′ + β θ + + σ′
=
cos tan sin
cos A c cos cos F 2 W A
F
b s t 0 v
nb
(4-28)
φ′ + ′ = tan F A c F
nb b sb
(4-29)
in which γ

is the effective unit weight of the rock mass; and
139
) 2 / 45 ( tan K
2
a
φ′ − = (4-30)
φ′ − = sin 1 K
0
(4-31)
γ′
σ′

γ′

=
0 v
a
0
K
c 2
z (4-32)
θ β = sec tan H
2
1
A
2
s
(4-33)
β θ β + = sec H ) tan tan H D ( A
b
(4-34)
β θ β + = tan H ) tan tan H D ( A
t
(4-35)
) tan D H
2
1
tan tan H
3
1
( W
2 2 3
β + θ β γ′ = (4-36)
The value of β was approximated as 45+ φ'/2 by Reese et al. (1974) for their wedge
type of failure in sand. Hoek (1983) pointed out that the failure plane of rock is 45+φ
'
/2
because rock mass also follows Mohr-Coulomb failure criterion. Bowman (1958)
suggested values of θ from φ'/3 to φ'/2 for loose sand and up to φ' for dense sand for a
similar wedge type of failure in soils. However, herein the value of θ is taken as φ'/2
based on the case studies (to be discussed in detail in Chapter VI) of actual field test
results. The values of φ' and c' can be obtained from Eqs. (4-13) and (4-14) by taking the
value of σ'
3
to be the effective overburden pressure at the depth of 1/3H, since the side
surface is triangular in shape. In this study, the following equations are adopted for θ and
β.
2
φ′
= θ (4-37)
2
45
φ′
+ = β (4-38)
140
The ultimate resistance of rock mass per unit shaft length p
u
(F/L) based on the
wedge failure mode identified herein can be calculated as:
dH
dF
dH
dF
sin 2
dH
dF
cos
dH
dF
sin
dH
dF
sin cos 2
dH
dF
p
a n nb sb s net
u
− θ − β + β + β θ = =
(4-39)

where
D ) z H ( K
dH
dF
0 a
a
− γ′ = (dF
a
/dH ≥ 0) (4-40)
) H
2
1
( sec tan H K
dH
dF
0 v 0
n
γ′ + σ′ θ β = (4-41)
) tan K
2
H
tan K c ( sec tan H
dH
dF
0 0 v 0
s
φ′ γ′ + φ′ σ′ + ′ θ β = (4-42)
β φ′ − β
θ β + θ β + ′ + γ′ + σ′ θ β + γ′ + σ′ β
=
cos tan sin
dH
dF
cos cos 2 ) tan tan H 2 D ( c ) H 2 ( tan tan H ) H ( tan D
dH
dF
s
0 v
2
0 v
nb
(4-43)
) tan sec tan H 2 sec D ( c
dH
dF
tan
dH
dF
nb sb
θ β β + β ′ + φ′ = (4-44)
4.5.2 Ultimate Rock Resistance at Great Depth
For in-depth jointed rock with a set of parallel weak planes, heavily fractured rock
mass and competent rock, the failure model shown in Fig. 4-17 is adopted. It is assumed
that the ultimate resistance of rock is reached when both the maximum shear resistance
between the drilled shaft and rock mass, τ
max
, and the normal limit pressures of rock mass,
p
L
are reached. Therefore, the ultimate rock resistance per unit length, p
u
, for in-depth
rock can be computed as follows.
D p d cos ) 2 sin( 2 / D 2 d sin 2 / D p 2 p
a
2 /
0
max
2 /
0
2
L u
− α α α τ + β β =
∫ ∫
π π
(4-45)
D p D
3
2
Dp
4
p
a max L u
− τ +
π
= (4-46)
141
where D = the diameter of a drilled shaft; p
a
= the active horizontal earth pressure;
Rankine’s earth pressure theory can be used to obtain p
a
as follows.
a v a a
K c 2 K p ′ − σ′ = (4-47)
K
a
= tan
2
(45-Ф

/2) (4-48)
where σ

v
= effective overburden earth pressure including the pressure induced by
possible overlying soils. The normal limit pressure of rock mass, p
L
, is the major
principal stress at failure, σ'
1
, which can be calculated using Eq. (4-9) in which σ'
3
is
equal to σ'
v
. The ultimate shaft-rock interface friction in horizontal direction, τ
max
, can be
estimated from Eq. (4-23). The effective cohesion c' needed in Eq. (4-47) can be
calculated using Eq. (4-14) by taking σ'
3
equal to σ'
v
.
Because the ultimate capacity calculated based on the presented in-depth failure
model could be smaller than the one calculated based on wedge failure model; therefore,
for the top layer of rock mass with or without overlying soils, the ultimate resistance of
rock resistance per unit shaft length is determined as the smaller one of the values
calculated from Eqs. (4-39) and (4-46).
4.5.3 Ultimate Resistance of Jointed Rock
For jointed rock with a set of parallel weak planes which dominate the behavior of
rock mass under lateral loads, the above equations for p
u
can still be used as long as the
following modifications are made. For failure of rock mass at a great depth, the high
overburden pressure will prohibit possible sliding failure on the weak planes. Eq. (4-46)
for in-depth rock mass therefore would still work for jointed rock.
For rocks to follow the wedge type failure, the bottom plane of the failure wedge
could be either within the rock mass or on a weak plane. It is suggested to calculate p
u
for
142
both possible failure modes and choose the smaller value as the final p
u
. For failure due
to shearing through rock mass, Eq. (4-39) can be directly used. For failure shearing
through a weak plane, the same wedge failure model shown in Fig. 4-12 can be used. Eq.
(4-39) is still valid to calculate p
u
, except angle β should become the inclination of the
weak plane and the values of c' and φ' for the bottom face of the failure wedge should be
those pertaining to properties of the weak planes. Specifically, those c' and φ' needed for
Eqs. (4-43) and (4-44) should be the results of lab test on samples with a weak plane.
However, those c' and φ' in Eq. (4-42) needed for the two sides of the failure wedge can
be obtained using instantaneous cohesion and friction angle determined from correlations
with Hoek-Brown strength criterion.

4.6 New Method for Predicting Lateral Capacity of Drilled Shafts in Rock
4.6.1 Free Head Boundary
To estimate the ultimate lateral resistance of a drilled shaft in rock with free head
boundary condition, a numerical solution that is based on discretization technique is
developed. The rock mass is modeled as n horizontal layers of rock with height h
i
for
each layer, as shown in Fig. 4-23. The ultimate resistance per unit thickness of each layer
of rock p
i
(F/L) can be calculated using Eqs. (4-39) and (4-46). The total resistance of
each layer, Q
i
, is equal to p
i
h
i
. If the drilled shaft is structurally in a yield condition under
the applied lateral load, then the shaft is treated as a long or flexible shaft; otherwise, the
shaft is considered as a short or rigid one. The procedure of the proposed method is
presented as follows.
143

Figure 4-23 The Rock-Shaft Model
The first step is to assume the drilled shaft to be a short (rigid) one and find the
ultimate capacity. Based on our observations from 3D FEA analysis and the lateral load
test at Dayton, the drilled shaft rotates around some point (pivot) close to the bottom of
the shaft, as shown in Fig. 4-24 (a). To calculate the lateral capacity of drilled shafts, the
pivot point needs to be found. By taking the moment balance at the loading point, the
pivot point can be easily determined as given in Eq. (4-49). The lateral capacity H
u
can
then be determined according to force balance equation given in Eq. (4-50).
∑ ∑ = ∑ ⇒ =
+
n
1 m
i i
m
1
i i
L Q L Q 0 M (4-49)
∑ ∑ ∑
+
− = ⇒ =
n
1 m
j
m
1
i
Q Q H 0 Q (4-50)
where e is the distance between the load point and ground surface; x
r
is the depth of pivot
point; and L
i
is the distance between the loading point and concentration point of force Q
i
;
m is the layer number of rock layer where the pivot point is located; n is the total number
of rock layers.
Q
1
= q
1
h
1
Q
i
= q
i
h
i
Q
n
= q
n
h
n
e
H
Drilled Shaft
Rock
144

Figure 4-24 Lateral capacity calculation models for drilled shafts in rocks
The second step is to check whether the maximum moment of the drilled shaft
exceeds the yielding moment of the shaft. If it does not, then the assumption of a short
shaft is valid and the calculated H
u
is the final estimate of capacity. Otherwise, the
assumption of short shaft is not valid and a third step follows to obtain the capacity of the
long drilled shaft.
The maximum moment of the drilled shaft is located at the section where shear force
is zero, as shown in Fig. 4-24 (b). The depth f where shear force in the shaft is zero is
located using the following equation:

=
o
1
i u
Q H (4-51)
where H
u
is the calculated capacity in step one assuming the shaft is a rigid one; o is the
layer number of the rock layer where f is located. A trial and error process can be used to
solve Eq. (4-51). Then the maximum moment M
max
is given by:
Q
m


e
H
Drilled
Shaft
Rock
Q
m+1


Q
n


Q
i


Q
2


Q
1


L
i


x
r
f
M
max
Yielding
Hinge

(a) Short Shaft
(c) Long Shaft
(b) Moment Diagram
Q
i


S
i


145

− + =
o
1
i i u max
S Q ) f e ( H M (4-52)
where S
i
is the distance between the concentration point of Q
i
and the position of shaft
where shear force is zero.
Step three is to calculate the lateral capacity for a long drilled shaft in which a
plastic hinge appears at the location of maximum moment, as shown in Fig. 4-24 (c). An
iterative process is used to find the capacity so that the computed maximum moment in
shaft is equal to the yielding moment of the shaft, M
y
. A new value of H
u
which is less
than that calculated in first step is assumed. Then step two is used to find the maximum
moment corresponding to the new H
u
. The final H
u
for a long (flexible) drilled shaft is
obtained when convergence is achieved (i.e., the calculated maximum moment is equal to
the yielding moment of the drilled shaft).
4.6.2 Fixed Head Boundary
For the case of fixed head boundary condition, the following procedure can be used
to determine ultimate lateral capacity. As before, the rock mass is modeled as n
horizontal layers of rock with height h
i
for each layer. The ultimate resistance per unit
thickness of each layer of rock p
i
(F/L) can be calculated using Eqs. (4-39) and (4-46).
Then, the total resistance of each layer Q
i
is equal to p
i
h
i
.
The first step is to assume the drilled shaft behaves as a rigid pile, as shown in Fig.
4-25. It is assumed that the drilled shaft will move rigidly in loading direction. The
following two equations can be used to calculate the lateral capacity H
u
and the
maximum moment at the ground surface level. The value of M
max
needs to be less than
146
the yielding moment of the drilled shaft, otherwise the second step needs to be followed
to calculate H
u
.
∑ =
n
1
i u
Q H (4-53)
∑ =
n
1
i i max
L Q M (4-54)
where L
i
is the distance between the ground surface and the loading point of Q
i
.

Figure 4-25 Capacity of rigid drilled shaft at fixed head boundary
In step 2, the drilled shaft is assumed to behave as shown in Fig. 4-26, where the
drilled shaft rotates around a pivot point and the moment of shaft at the point right
beneath the cap reaches the yielding moment. By taking moment balance at the loading
point (the point beneath the cap), the rotation point can be easily determined as given by
Eq. (4-55). The lateral capacity can be determined using Eq. (4-56). Additionally, the
maximum moment M
max
shown in Fig. 4-26 needs to be less than the yielding moment
M
y
. The maximum moment of the drilled shaft is located at the section where shear force
is zero. After the location of M
max
is found using Eq. (4-51), the value of M
max
can be
H
After
Moved
Before
Moved
Q
1
Q
i

Q
n

H
u
L
i

M
max
Moment Diagram Force Balance
147
easily determined using Eq. (4-57). If the maximum moment is greater than the yielding
moment, the third step needs to be followed.
∑ = ∑ ∑ + ⇒ =
+
m
1
i i
n
1 m
i i y
L Q L Q M 0 M (4-55)
∑ − ∑ =
+
n
1 m
i
m
1
i u
Q Q H (4-56)
∑ − − ⋅ =
0
1
y i i u max
M S Q f H M (4-57)

Figure 4-26 Capacity of intermediate length drilled shaft at fixed head boundary
Step 3 assumes the drilled shaft behave as a long pile, as shown in Fig. 4-27. The
drilled shaft will have two yielding points: one is located at the top of the drilled shaft or
the loading point; the other one is located at certain depth where shear force is zero.
Assume a value of H
u
which is less than the value of H
u
calculated in Step 2. Next, find
the location where shear force is zero and the corresponding maximum moment at that
location using Eqs. (4-51) and Eq. (4-57). Repeat this procedure until the calculated
maximum moment is equal to the yielding moment. The final value of lateral capacity H
u

is then obtained after the iteration is converged.
H
After
Moved
Before
Moved
Q
1
Q
i

Q
m
H
u
L
i

M
y
Moment Diagram Force Balance
Pivot
Point
Q
m+1
Q
n

M
y
M
max
f S
i

148
The proposed method is not only suitable for drilled shafts entirely socketed in rock,
but can also be used for drilled shafts in rock layers with overlying soils. Although the
developed method is somewhat time consuming for hand calculation, a spreadsheet or
computer program can be easily developed to facilitate calculation.

Figure 4-27 Capacity of long drilled shaft at fixed head boundary

4.7 Ultimate Reaction of Soils
To make the lateral capacity prediction method presented above to be applicable to
soils, the following sections describe the methods for estimating the ultimate reaction of
clay, sand, and c-φ soils.
4.7.1 Ultimate Resistance of Clay
Baguelin et al. (1977) examined the mechanism of the lateral reaction of piles in an
elasto-plastic medium. It was found that the soil resistance is composed of the resistances
from both normal and tangential directions, as shown in Fig. 4-28.
H
After
Moved
Before
Moved
Q
1

Q
i

Q
m
H
u
M
y
Moment Diagram Force Balance
Yielding
Point
M
y
M
max L
i
149
By integrating the stress distribution as shown in Fig. 4-28, the total ultimate soil
resistance per unit shaft length, p
u
, can be calculated as:
) p ( D
4
p
max L u
τ +
π
= (4-58)
where p
L
= ultimate normal resistance of soil; τ
max
= ultimate shear resistance of soil.

Figure 4-28 Distribution of lateral reaction stresses
The theoretical equation for calculating ultimate soil normal resistance by Reese et
al. (1975) is used herein. Based on the wedge failure mode, the limit resistance of top soil
layer, p
Lt
, is given by:
D
z S 83 . 2
z S 2 p
u
u Lt
+ γ′ + =
(4-59)
where S
u
is undrained shear strength of clay; and z is the depth under consideration.
Based on the flow-around type failure for clay exists at a great depth, the limit resistance
of in-depth clay, p
Ld
, is given by:
u Ld
S 11 p =
(4-60)

β
p(β) = p
L
sin β
p
L
Applied load
α
τ(α) = τ
max
cosα
τ
max
Shear stress
Normal stress
150
The final value of p
L
is selected as the smaller value of p
Lt
and p
Ld
, using Eqs. (4-59)
and (4-60).
The ultimate side shear resistance of clay can be calculated using α method given by.
u max
S α = τ
(4-61)
where α is adhesion factor which can be found in O’Neill and Reese (1999) as follows.
α = 0.55 for S
u
/p
a
≤1.5; and
α = 0.55-0.1(S
u
/p
a
-1.5) for S
u
/p
a
>1.5 (4-62)
in which p
a
is the atmospheric pressure (14.65 psi).
4.7.2 Ultimate Resistance of Sand
The following procedure for calculating the ultimate resistance of sand was
suggested by Zhang et al. (2005). That is, the ultimate resistance per unit length of drilled
shaft can be calculated as
D ) p 8 . 0 ( p
max L u
τ + = (4-63)
The limit normal stress of sand, p
L
, according to Fleming et al. (1992), can be
calculated as
z K p
2
p L
γ′ = (4-64)
where K
p
= tan
2
(45+φ

/2) = passive earth pressure coefficient.
The ultimate τ
max
in horizontal direction is assumed to be the same as the ultimate
vertical shear resistance estimated with the following equation from API (1991)
δ γ′ = τ tan z K
max
(4-65)
151
where K = lateral earth pressure coefficient; and δ = interface friction angle between the
drilled shaft and the soil. Tables 4-5 and 4-6 taken from Kulhawy et al. (1983) and
Kulhawy (1991) can be used to estimate appropriate values for K and δ.
Table 4-5 Recommended Values of K by Kulhawy et al. (1983) and Kulhawy (1991)
Pile type and method of construction K
Pile-jetted (0.5-0.7) K
0

Pile-small displacement, driven (0.7-1.2) K
0

Pile-large displacement, driven (1.0-2.0) K
0

Drilled shaft-build using dry method with minimal sidewall
disturbance and prompt concreting
(0.9-1.0) K
0

Drilled shaft-slurry construction with good workmanship (0.9-1.0) K
0

Drilled shaft-slurry construction with poor workmanship (0.6-0.7) K
0

Drilled shaft-casing method blow water table (0.7-0.9) K
0

Note: K
0
= coefficient of lateral earth pressure at rest
Table 4-6 Recommended Values of δ by Kulhawy et al. (1983) and Kulhawy (1991)
Pile type δ
Rough concrete 1.0 φ


Smooth concrete (i.e., precast pile) (0.8-1.0) φ


Rough steel (i.e., step-taper pile) (0.7-0.9) φ


Smooth steel (i.e., pipe pile or H pile) (0.5-0.7) φ


Wood (i.e., timber pile) (0.8-0.9) φ


Drilled shaft built using dry method or with temporary casing and
good construction techniques
1.0 φ


Drilled shaft built with slurry method (higher values correspond
to more careful construction methods)
(0.8-1.0) φ



4.7.3 Ultimate Resistance of c-φ Soils
Brinch Hansen’s (1961) theory can be used to calculate p
u
of c-φ soils. An empirical
modification factor M was later suggested by Mokwa et al. (2000) to modify Brinch
Hansen’s equation as follow:
D ) K c zK ( M p
c q u
′ + γ′ = (4-66)
152
where M = an empirical modification factor = 0.85; K
q
= a coefficient for the frictional
component of net soil resistance under 3D condition; and K
c
= a coefficient for the
cohesive component of net soil resistance under 3D condition. The values of K
q
and K
c

can be calculated as follows:
D / Z 1
D / Z K K
K
q
q q
0
q
q
α +
α +
=

(4-67)
D / Z 1
D / Z K K
K
c
c c
0
c
c
α +
α +
=

(4-68)
) 5 . 0 45 sin(
sin K
K K
K
0
0
q q
0
q
q
φ′ + °
φ′

= α

(4-69)
) 5 . 0 45 sin( 2
K K
K
0
c c
0
c
c
φ′ + ° ⋅

= α

(4-70)
) 2 / 45 tan( cos e ) 2 / 45 tan( cos e K
tan ) 2 / ( tan ) 2 / ( 0
q
φ′ − ° φ′ − φ′ + ° φ′ =
φ′ φ′ − π − φ′ φ′ + π
(4-71)
φ′ − φ′ + ° φ′ =
φ′ φ′ + π
cot ] 1 ) 2 / 45 tan( cos e [ K
tan ) 2 / ( 0
c
(4-72)
φ′ =
∞ ∞
tan K d N K
0 c c q
(4-73)
∞ ∞
=
c c c
d N K (4-74)
φ′ − φ′ + ° =
φ′ π
cot ] 1 ) 2 / 45 ( tan e [ N
2 tan
c
(4-75)
φ′ + =
∞ 4
c
tan 09 . 4 58 . 1 d (4-76)
φ′ − = sin 1 K
0
(4-77)

4.8 Validation of the Derived Capacity Prediction Method
Four field lateral load tests results are employed herein to evaluate the derived
lateral capacity prediction method. Shaft #4 of Dayton load test and shaft #2 of Pomeroy-
Mason load test are selected. Additionally, the load test results reported by Hall and
Wang (2000) and the short shaft of I-85 load test reported by Gabr et al. (2002) are used.
The dimensions and strength parameters of the test drilled shafts of the three load tests
are summarized in Table 4-7, in which total length reflects the distance between the
153
loading point and the tip of the drilled shaft. The yielding moments of the drilled shafts
are obtained from LPILE computer analysis by inputting reinforcement and physical
dimensions of the drilled shafts. The soil and rock properties required for calculation are
summarized in Table 4.8. The depth is measured from the ground surface down.
Table 4-7 Test Drilled Shaft Information
Test Diameter
(ft)
Total
Length
(ft)
Loading
Eccentricity
(ft)
Reinforcement Concrete
Strength
(psi)
Yielding
Moment
(kip-ft)
Dayton 6 18 0 34#11 4500 8008
Pomeroy-
Mason
8 112.9 53.1 28#18 plus 1
inch casing
5115 21640
Hall &
Wang
(2000)
5 32 1 12#18 4000 5488
Gabr et
al. (2002)
2.5 10.2 1 12#10 plus 0.5
inch casing
5000 2819

Table 4-8 Input rock mass parameters of the load tests
Tests Depth (ft) γ' (pci) q
u
(psi) GSI m
i
Rock
0-7 0.038 5668 40 6 Shale Dayton
Shaft #4 7-18 0.038 5668 61 6 Shale
0-21.9 0.059 3797 42 6 Shale
21.9-29.4 0.060 9073 45 17 Sandstone
29.4-37.3 0.049 19 38 4 Claystone
37.3-48 0.047 44.3 28 4 Claystone
Pomeroy-
Mason
Shaft #2
48-56.8 0.055 826.2 44 4 Claystone
0-14.5 0.036 - - - Sand (φ = 34º)
14.5-18 0.057 700 34 6 Clayshale
18-20 0.057 1200 35 6 Clayshale
20-25 0.057 1750 57 6 Clayshale
Hall &
Wang
(2000)
25-31 0.059 1750 57 7 Siltstone
0-3.9 0.055 4220 59 9 Siltstone
3.9-6.2 0.055 3596 59 9 Siltstone
Gabr et
al. (2002)
6.2-9.2 0.055 6598 59 9 Siltstone

154
The four load tests did not reach either structural yielding or rock mass failure
during actual load tests. Therefore, Kulhawy and Chen (1995)’s hyperbolic curve fit
technique is used to simulate the non-linear load-deflection behavior to the ultimate load,
from which the ultimate capacity of piles (drilled shafts) is determined. The hyperbolic
equation in terms of the lateral load (H) and the lateral deflection at the loading point (δ)
can be expressed as follows:
δ +
δ
=
b a
H (4-78)
where a and b are two curve fitting constants. The ultimate lateral load capacity is defined
as the deflection δ becomes infinite large and is calculated as 1/b.
A computer program LCPILE (lateral capacity of piles) using VC++ is developed in
this study to facilitate computation of the developed method. The calculated capacities
and the measured values from hyperbolic curve fitting technique of actual load tests data
are summarized in Table 4.9. The prediction errors defined as the difference of capacities
divided by the measured capacities from load tests are given in Table 4.9 as well. The
capacity prediction error of Dayton site is relatively higher than the other two cases. This
may be due to extremely small shaft head deflection of 0.135 inch at maximum applied
lateral load of 1126 kips. Nevertheless, the developed method can yield reasonable
predictions of the lateral capacities of drilled shafts socketed in rock mass.
Table 4-9 Comparison of Lateral Capacities of Test Drilled Shafts
Test
Predicted
(kips)
Measured
(kips)
Error
(%)
Dayton 2447 1612 52
Pomeroy-Mason 405 431 -6
Hall and Wang (2000) 500 589 -15
Gabr et al. (2002) 718 677 5.7

155
4.9 Summary and Conclusions
The work presented in this chapter can be summarized as follows:
1. A 3-D FEM model for simulating a laterally loaded drilled shaft in rock is developed
and validated against the lateral load test results at Dayton site.
2. A wedge failure mode for rock mass at or near ground surface is identified using 3D
FEM analysis techniques. Strength controlled failure mode for rock at great depth is
identified as well.
3. The effect of secondary structures of rock mass, such as joints and fillings is taken
into consideration through adoption of a rock classification system GSI. Additionally,
equations by Hoek (1990) for instantaneous cohesion and friction angle of rock mass
are used in this study.
4. Empirical equation for estimating the axial side shear resistance of the rock-shaft
interface by Kulhawy and Phoon (1993) is adopted herein to compute the side shear
resistance of the interface in horizontal direction.
5. Based on the suggested failure modes, theoretical equations for calculating the
ultimate resistance of rock mass per unit shaft length, p
u
, are derived for top layer
rock mass and rock at great depth. The adoption of Hoek-Brown rock strength
criterion and the empirical equation for side shear resistance ensures that the derived
theoretical equations incorporate secondary structure effects.
6. A new method for predicting the ultimate lateral capacity of a drilled shaft socketed
in rock mass is proposed. The method is also extended for use of a drilled shaft in
various types of soils.
156
7. Finally, evaluation of the developed method for determining ultimate lateral capacity
of a rock-socketed drilled shaft against load test results validate the accuracy of the
method.
Based on above work, the following conclusions can be drawn.
1. The 3D FEM model can be used to simulate the lateral response of drilled shafts in
rock. This modeling technique can help verify a drilled shaft foundation design to
avoid an expensive lateral load test.
2. The identified rock mass failure modes due to laterally loaded drilled shaft are useful
for deriving theoretical equations for estimating the ultimate resistance of rock mass.
3. The secondary structures of rock mass were considered in the developed theoretical
equations for calculating p
u
. The accuracy of the theoretical equations for calculating
p
u
is acceptable based on evaluation against three field load test results.
4. Hyperbola extrapolation technique originally developed by Kulhawy and Chen (1995)
may under estimate the lateral capacity of drilled shafts in rock if the deflection is
small at the maximum applied load during lateral load tests.
5. The proposed method for estimating lateral capacity of drilled shafts in rock can
provide reasonable predictions according to the evaluation using field test data. The
average prediction error is about 21% which is adequate for practice considering the
measured capacity may be under estimated.
6. The proposed capacity prediction method is versatile. It can be used for estimating the
lateral capacity of drilled shafts in clay, sand, silts, and rock. The soil or rock can be
layered. Both short and long drilled shafts can be considered. The boundary condition
can free head or fixed head.
157



CHAPTER V
ELASTIC SOLUTION OF LATERALLY LOADED DRILLED SHAFTS IN ROCK

Although nonlinear p-y analysis is the most widely used method for analysis of
laterally loaded piles, an elastic subgrade reaction solution nevertheless is an important
alternative method for solving the problem in linear elastic range due to its simplicity and
easy to use. Subgrade reaction method models soils as Winkler springs and the pile as a
beam. In order to develop an elastic subgrade reaction solution for drilled shafts in rock, a
methodology for determining deformation modulus of rock mass and an empirical
equation for estimating the modulus of subgrade reaction of rock are developed in this
chapter.

5.1 Determination of Rock Mass Deformability
5.1.1 Introduction
The fact that jointed rock masses do not behave elastically has prompted the usage
of the term modulus of deformation rather than modulus of elasticity or Young’s modulus.
The Commission of Terminology of International Society for Rock Mechanics (ISRM)
published the definitions: Modulus of deformation - the ratio of stress to corresponding
strain during loading of a rock mass including elastic and inelastic behavior; Modulus of
158
elasticity – the ratio of stress to corresponding strain below the proportionality limit of a
material (Bieniawski, 1978).
There are two categories of approaches, in-situ testing and empirical equations, for
determining the deformation modulus of rock mass (E
m
). The existing in-situ testing for
determination of E
m
includes plate loading, pressuremeter/dilatometer testing, Goodman
Jack test. Although all of the in-situ testing methods have their limitations in measuring
the actual deformation modulus, they are still the most desirable approach. To reach
rocks at depth such as for deep foundation applications, pressuremeter/dilatometer test
and Goodman Jack test are most appropriate. The Goodman Jack test will be introduced
in this chapter. However, the details of pressuremeter/dilatometer test will be presented in
Chapter VII.
5.1.2 Goodman Jack Test
5.1.2.1 Description of Goodman Jack
The NX borehole jack device, also called Goodman Jack, was introduced in 1966 by
Goodman et al. (1972). The Goodman Jack test is designed to be used in 3" boreholes.
The schematic of loading mechanism of Goodman Jack is shown in Fig. 5-1 (Heuze
1984). The device consists of two steel plates forced apart by 12 racetrack-shaped pistons.
The shape of these pistons was selected to give maximum hydraulic efficiency. The jack
is attached to drill rod and inserted into the borehole. A hand pump is used to create
hydraulic pressure in the lines connected to the jack, which in turn activates the pistons
and produces a uniform and unidirectional stress field at the bearing plate. The applied
hydraulic pressure is measured with a pressure gauge. The deformation of the rock is
measured by two linear variable differential transformers (LVDT) and data are displayed
159
by the indicator at the ground surface. Two return pistons close the instrument to a
thickness of 2¾ in., affording ¼ in. clearance for positioning in an NX borehole. The
total piston travel is 0.5 inch; the LVDTs have a linear range of 0.2 inch and are adjusted
to begin their linear travel when the plates are about to contact the rock.

Figure 5-1 The schematic of loading of Goodman Jack (Heuze, 1984)
5.1.2.2 Test Data Interpretation
Goodman Jack test data was interpreted through theory of elasticity to derive the
deformation modulus of rock. Goodman et al. (1972) studied the effect of Poisson’s ratio
and the finite length of platens since the theoretical analysis of jack test assumes infinite
length of the platens. The effect of Poisson’s ratio ν is small, because a 50%
overestimation in ν, from 0.2 to 0.3, would lead to a 3.25% underestimate of modulus of
rock. The effect of finite test length was investigated by using a 3D finite element
analysis and taken into account by reducing 14% values of calculated modulus of rock.
By considering the effect of limited length and the hydraulic efficiency, the equation
for interpretation of test data is given by (Goodman et al., 1972)
d / u
Q
)] , ( K )[ 93 . 0 )( 86 . 0 ( E
d
h


β ν = (5-1)
160
where, constant 0.86 is for considering 3-D effect, constant 0.93 is for considering
hydraulic efficiency, ∆Q
h
is change of hydraulic pressure on platens, ∆u
d
is increment of
diametral displacement, d is diameter of borehole, and K(ν, β) is a constant defined in
Table 5-1.
Table 5-1 Values of Constant K(ν) for β=45°
ν 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
K(ν) 1.38 1.29 1.29 1.28 1.27 1.25 1.23 1.20 1.17 1.13 1.09

Goodman et al. (1972) also investiaged the influence of crack generated by jacking
or a nature joint located along the borehole by using finite element analysis. For a crack
extending a half radius away from the hole, the rock mass modulus would be calculated
as 13% lower than the modulus calculated without cracking.
Since introduction of the Goodman Jack, several researchers have evaluated various
data interpretation methods, such as Meyer and McVey (1974), Hustrulid (1976), Heuze
and Salem (1976), Shuri (1981) and Heuze and Amadei (1985). Meyer and McVey (1974)
conducted a jack test in a block of 5052 aluminum alloy measured 18 ×18 × 20 inch with
a 3 inch central hole. The calculated modulus using Eq. (5-1) was 2.93 × 10
6
psi, which is
much less than the known elastic modulus of aluminum alloy of 10.08 × 10
6
psi. Meyer
and McVey (1974)’s carefully conducted tests showed that the calculated modulus could
be very different from the true modulus. Subsequent to this research, Hustrulid (1976) re-
derived the K(ν, β) function in Eq. (5-1). A new expression T
*
was proposed for K. For β
= 45
°
, where the definition of β is shown in Fig. 5-1, the formula for modulus calculation
is given as:
161
d
h *
calc
u
Q
) T )( 3 )( 93 . 0 ( E


= (5-2)
It should be noted that Eq. (5-2) was based on two-dimensional analysis. Hustrulid
(1976) also discussed the effect of oversized holes on the calculation of modulus. Heuze
and Salem (1976) performed both two-dimensional and 3-dimensional analysis and found
that the ratio of plate modulus to rock modulus, the rock anisotropy, and the plate
geometry would influence the deformations during jack test. Shuri (1981) re-evaluated
the radius mismatch problem first discussed by Hustrulid (1976). Shuri analyzed both the
undersize and oversize cases relative to 3 in. borehole diameter and the modulus
reduction factors are shown in Fig. 5-2, in which E
app
is calculated modulus of rock E
calc

without considering the undersize or oversize effect; and E
act
is the modulus considering
the effect of undersize and oversize. Shuri ignored the plate bending problem. Therefore,
what Shuri refers to as the E
act
(E
actual
) is in fact the modulus without considering the plate
bending effect. It is however essential to apply the plate bending correction to yield the
true rock modulus.
Heuze and Amadei (1985) integrated all the advances made in understanding the
jack behavior prior 1985. Shuri (1981)’s work on oversize and undersize borehole was re-
evaluated. The Shuri approach for oversize holes was found to be correct. However, the
derivation by Shuri for undersize holes was found to be incorrect and was corrected by
including the new value of T
*
from Heuze and Amadei (1985). Additionally, the original
correction from E
calc
to E
true
obtained by Heuze and Salem (1976) via 3-D finite element
analysis for the plate bending, was recalculated with the T
*
of Heuze and Amadei (1985).
162
The current version of test data interpretation follows the following steps. Step 1:
calculate the E
calc
by using the following equation for full contact condition:
) T (
u
Q
) d )( 93 . 0 )( 86 . 0 ( E
*
d
h
calc


= (5-3)
The values of T
*
are provided in Table 5-2. Step 2: the E
calc
calculated from Step 1 is
corrected using Fig. 5-2 to account for the undersize and oversize effect. Finally, the
corrected modulus from Step 2 is further corrected using Fig. 5-3 to account for the
platen bending effect.

Figure 5-2 Modulus reduction vs. hydraulic pressure for various borehole diameters (a)
undersize – the pressure must be decreased by about 14% for a given E
app
/E
act
as
recalculated by Heuze et al.(1985). (b) oversize holes
163
Table 5-2 Values of T
*
(Heuze and Amadei, 1985)
ν 0.1 0.2 0.25 0.3 0.33 0.4 0.5
T
*
1.519 1.474 1.438 1.397 1.366 1.289 1.151

E
calc
(10
6
psi)
0 1 2 3 4 5
E
t
r
u
e
(
1
0
6
p
s
i
)
0
2
4
6
8
10
12
14
16
E
calc
(GPa)
0 10 20 30
E
t
r
u
e
(
G
P
a
)
0
20
40
60
80
100

Figure 5-3 Correction for platen bending of the jack (after Heuze and Amadei, 1985)
5.1.3 Selection of In-Situ Test Method
From fundamental point of view, both dilatometer and Goodman Jack involve
similar assumptions and theoretical derivations for test data interpretation. Both of them
assume that the test chamber or platen is infinite long and that the rock is isotropic,
homogeneous, and linearly elastic. The difference between the two devices is the method
to apply the pressure to the borehole walls. Dilatometer can apply uniform pressure to the
whole borehole wall; but the Goodman Jack applies unidirectional pressure to the
borehole wall.
The disadvantage of dilatometer is the limit of applied pressure, which usually is
less than 4,350 psi (30 MPa). In fact, this is a satisfactory upper limit of pressure for
164
many rock types and it exceeds the actual stress level experienced by rock in most of
civil engineering works (Goodman et al., 1972). The disadvantage of Goodman Jack is
the less precisely known pressure condition under the load comparing with dilatometer.
However, the Goodman Jack can reach 10,075 psi (69MPa) pressure limit. Further, the
ability to generate higher pressures allows the test to be carried beyond the elastic region
of many rock types.
The current practice of dilatometer and Goodman Jack in determining the
deformation modulus of rock mass has shown both successful and unsuccessful cases
(Rocha et al., 1970; Bukovansky, 1970; Georgiadis and Michalopoulos, 1986; Reese,
1997; and Littlechild, 2000). Rocha et al. (1970) pointed out that the dilatometer test
results are reliable based on their 387 dilatometer tests carried out in 9 test programs.
Georgiadis and Michalopoulos (1986) investigated the application of dilatometer test to
design grouted pile in rock. A total number of 20 dilatometer tests were performed in
seven boreholes, mainly in jointed weak mudstone. The dilatometer test results were
utilized in conjunction with laboratory test results to design laterally loaded steel-pipe
piles drilled and grouted in mudstone. Reese (1997) utilized dilatometer test results to
correlate the initial slope of his interim p-y criterion for weak rock.
However, in contrary to the preceding successfully application of dilatometer,
Bukovansky (1970) found that the deformation modulus determined from test results of
dilatometer were surprisingly low comparing that from plate-load, Flatjack, and radial
jacking tests. Littlechild et al. (2000) investigated the use of high pressure dilatometer,
self-boring pressuremeter and Goodman Jack tests in Hong Kong. It was found that the
dilatometer tests were not necessarily successful. In highly fractured rocks, such as
165
granodiorite, dilatometer membrane tended to fail (puncture) frequently. It was shown
that Goodman Jack test can be easily carried out in more uniform and competent rocks
that exhibit little fracturing or weathering. However, for weak rock, Goodman Jack tests
require careful consideration of the maximum pressure that will be applied so that the
device can be retracted from the bore hole (Littlechild et al., 2000).
Based on review provided in the above sections, it can be concluded that dilatometer
test is suitable for the determination of weak rock mass modulus, except for highly
fractured rock, while Goodman Jack test is applicable to hard rock mass. The purchasing
price for either of the test devices is around $30,000 in 2003.
5.1.4 Existing Empirical Equations
Several empirical equations are available for the determination of deformation
modulus of rock mass using empirical correlations with rock properties. Bieniawski
(1978) could be credited for the development of earlier empirical equations. Other
empirical equations found in literature are summarized in Table 5-3. The parameters used
for correlations and perceived limitations of these empirical equations are provided in
Table 5-3, in which RMR = Bieniawski (1976)’s Rock Mass Rating, E
m
= the
deformation modulus of rock mass, E
i
= the modulus of intact rock, GSI = Geological
Strength Index, UCS =σ
ci
= unconfined compressive strength of intact rock; RQD = Rock
Quality Designation; WD = weathering degree.
Gokceoglu et al. (2003) evaluated existing empirical equations by utilizing a
database consisting of 115 in-situ test data points. The database was obtained from two
dam sites with quartzdiorite and limestone, and another construction site with heavily
jointed marly rock mass. The measured deformation modulus of rock mass was obtained
166
using in-situ plate load test and dilatometer test. However, the data for weak rocks in this
database was limited.
Table 5-3 Empirical Equations for Estimating the Deformation Modulus of Rock Mass
Empirical
Equation
Required
Parameters
Limitation Equation
Bieniawski
(1978)
RMR RMR > 50 E
m
=2RMR-100 (GPa)
Serafim and
Pereira
(1983)
RMR RMR ≤ 50 E
m
=10
(RMR-10)/40
(GPa)
Nicholson&
Bieniawski
(1990)
E
i
and RMR - E
m
=E
i
[0.0028RMR
2
+0.9exp(RMR/22.82)]/
100

Hoek and
Brown
(1997) &
Hoek et al.
(2002)
GSI, σ
ci
and
D
r
which is
disturbance
factor
-
40 / ) 10 GSI ( ci
r m
10
100
) 2 / D 1 ( E

σ
− = (GPa)
for σ
ci
≤ 100MPa;
40 / ) 10 GSI (
r m
10 ) 2 / D 1 ( E

− = (GPa) for
σ
ci
>100MPa
Kayabasi et
al. (2003)
E
i
, RQD
and WD
-
1811 . 1
i
m
WD
) 100 / RQD 1 ( E
135 . 0 E
(
¸
(

¸
+
= , E
m
and
E
i
in GPa
Gokceoglu
et al. (2003)
E
i
, RQD,
UCS and
WD
-
5528 . 1
i
m
WD
) 100 / RQD 1 )( UCS / E (
001 . 0 E
(
¸
(

¸
+
=
(GPa)

The root mean square errors (RMSE) were calculated to evaluate the performance of
the existing empirical correlation equations. The evaluation results show that Hoek and
Brown (1997)’s equation gave the best results for a weak marly rock mass, having a
mean uniaxial compressive strength of 2697 psi. Kayabasi et al. (2003)’s equation
exhibited a high predictive capability with the value of RMSE equal to 0.99 for
dilatometer data including the weak marly rock mass results. However, Kayabasi et al.
(2003)’s equation was developed with half of the data points in the entire database.
167
Nicholson and Bieniawski (1990)’s equation, which was developed from the test data of
Bieniawski (1978) and Serafim and Pereira (1983), exhibited good results based on the
RMSE values calculated using all the data in Gokceoglu et al. (2003)’s database.
Contrary to these, the equations proposed by Bieniawski (1978) yielded the most
scattered results.
Based on Gokceoglu et al. (2003)’s evaluation discussed previously, it appears that
Nicholson and Bieniawski (1990)’s equation could be adopted. Although the equation
proposed by Gokceoglu et al. (2003) yields good prediction for his database, it
nevertheless needs to be further evaluated with other database. Hoek and Brown (1997)’s
equation provides good estimate on modulus for weak rock; however, Littlechild et al.
(2000) found that this equation was not appropriate for foundation design on weathered
rock according to their field test results in Hong Kong for igneous, volcanic and
metamorphosed sedimentary rocks.
5.1.5 Parameters for Determination of E
m

Based on literature review in the previous sections, in general, rock properties such
as RMR, q
u
, E
i
, and RQD have been selected to correlate with deformation modulus of
rock masses.
Kayabasi et al. (2003) employed cosine amplitude method to investigate the strength
of correlations between E
m
and the intact rock and discontinuity properties of rock
masses based on 57 test data points. The statistical analysis results are summarized in
Table 5-4. It can be seen that modulus of intact rock correlates best with E
m
. All the
parameters associated with secondary structures (such as RQD, weathering) show high
correlations with E
m
than that of the unconfined compressive strength of intact rock.
168
Table 5-4 The Strength of the Relation between E
M
and Parameters (Kayabasi et al. 2003)
Parameter Strength Value
Modulus of elasticity of intact rock 0.878
RQD 0.873
Weathering of the discontinuities 0.865
Roughness of the discontinuities 0.851
Aperture of the discontinuities 0.832
Continuity of the discontinuities 0.768
Infilling 0.761
Spacing of the discontinuities 0.750
Uniaxial compressive strength of intact rock 0.720

Although Kayabasi et al. (2003) selected the first three factors in Table 5-4 (E
i
, RQD
and weathering degree) for their correlation equation. This research adopts the modulus
of intact rock and a rating system (such as RMR and GSI) that represents the effects of all
secondary structures. The good prediction results on E
m
from Nicholson and Bieniawski
(1990)’s equation based on Gokceoglu et al. (2003)’s database and the unsatisfactory
predictions from Hoek and Brown (1997)’s equation based on on Littlechild et al.
(2000)’s test data support this conclusion.
A set of dilatometer test data in Ironton-Russell (Paul and Martin 2004) is compiled
in Table 5-5 for independent check on various correlation equations.
Table 5-5 Properties of Rock Masses in Ironton-Russell
No. RQD q
u
(psi) Ei (psi) E
M
, mea. (psi) GSI RMR WD
1 75 8406 1689.7 667.1 74 64 0.400
2 98 9544 1751.5 586.4 77 67 0.400
3 78 9559 1809.4 625.5 74 64 0.400
4 100 9005 1569.1 636.8 77 67 0.400
5 100 8002 1343.4 439.4 77 67 0.400
6 98 9137 1763.3 525.6 77 67 0.400
7 98 8578 1774.5 504.3 77 67 0.400

169
Table 5-6 provides the predictions of E
m
and the ratio of predicted over measured
deformation modulus of rock masses using Nicholson and Bieniawski (1990)’s equation,
Hoek and Brown (1997)’s equation, and Kayabasi et al. (2003)’s equation. Nicholson and
Bieniawski (1990)’s equation which employs E
i
and RMR as the parameters provides the
best estimate on E
m
, while the other two methods over-predict E
m
. Therefore, E
i
and rock
rating system (such as RMR and GSI) are the most important parameters for correlating
with modulus of rock masses.
Table 5-6 Predictions and Ratios of Predicted over Measured Modulus of Rock Masses
No. Nicholson and
Bieniawski (1990)
Hoek and Brown
(1997)
Kayabasi et al.
(2003)
Serafim and
Pereira (1983)
Prediction Ratio Prediction Ratio Prediction Ratio Prediction Ratio
1 445.0 0.67 4395.2 6.6 2033.9 3.05 2434 3.6
2 517.1 0.88 5566.1 9.5 2455.3 4.19 2893 4.9
3 476.5 0.76 4686.9 7.5 2249.9 3.60 2434 3.9
4 463.3 0.73 5406.6 8.5 2182.0 3.43 2893 4.5
5 396.7 0.90 5096.6 11.6 1816.3 4.13 2893 6.6
6 520.6 0.99 5446.1 10.4 2474.8 4.71 2893 5.5
7 523.9 1.04 5276.9 10.5 2493.4 4.94 2893 5.7

5.1.6 Proposed Empirical Equation
Although Nicholson and Bieniawski (1990)’s equation is the most promising
empirical equation for estimating deformation modulus of rock masses, the use of RMR
makes it suffer some inherent drawbacks. The RMR value is difficult to estimate for very
poor rock, especially for borehole cores in which relatively few intact core pieces longer
than 4 inch can be recovered (Hoek and Brown, 1997). In these circumstances, the
physical appearance of material recovered in the core can be used to estimate a GSI value.
Therefore, the GSI system is proposed as an alternative to the RMR system for estimating
deformation modulus of rock masses.
170
To preserve the merit of Nicholson and Bieniawski (1990)’s equation, the test data
(Bieniawski, 1978 and Serafin and Pereira, 1983) used to develop their equation is used
to develop a new empirical equation. According to Hoek and Brown (1997), GSI equals
to RMR (Bieniawski, 1976), with the ground water rating set to 10 and the adjustment for
joint orientation set to 0. Due to the lack of details of estimating RMR of these data, it
was assumed that the ground water rating was set to an average value of 5. Therefore,
GSI values of these test data are estimated by adding original RMR with 5. Additionally,
the new data from Ironton-Russell is included. A regression analysis is carried out and
the results are plotted in Fig. 5-4. The following equation is proposed to estimate
deformation modulus of rock masses using modulus of intact rock and GSI.
) e (
100
E
E
7 . 21 / GSI i
m
= (5-4)
5.1.7 Recommended Methodology for Determination of Deformation Modulus of Rock
In-situ tests usually provide more reliable values for the deformation modulus of
rock mass than empirical equations. For weak rock, dilatometer test is recommended
since this test has been widely and successfully used in geotechnical engineering
applications. However, for hard and competent rock, Goodman Jack test is recommended
since it can provide higher pressure than dilatometer. For preliminary design or where in-
situ tests are not available, the derived Eq. 5-4 is recommended to estimate the
deformation modulus of rock mass.
171
0.0
20.0
40.0
60.0
80.0
100.0
0 20 40 60 80 100
GSI
E
m
/
E
i

(
%
)
Bieniawski (1978)
Serafin and Pereira (1983)
Ironton-Russell
Regression
E
m
/E
i
=exp(GSI/21.7)/100

Figure 5-4 Proposed empirical equation using GSI

5.2 Initial Modulus of Subgrade Reaction of Rock Mass
5.2.1 Introduction
The term initial modulus of subgrade reaction refers to the initial slope of a p-y
curve. In literature, few studies and discussions are available regarding the determination
of modulus of subgrade reaction, especially for rock masses. By fitting the subgrade
reaction solution with continuum elastic solution for beam on elastic foundation, Vesic
(1961) provided an elastic solution for the modulus of subgrade reaction, K (F/L
2
) as
follows.
12 / 1
p p
4
2
I E
ED
1
E 65 . 0
K
(
(
¸
(

¸

ν −
= (5-5)
172
where E = modulus of elastic materials; ν = Poisson’s ratio; D = beam width; and E
p
I
p
=
flexural rigidity of beam.
Bowles (1988) suggested to double the value of K in Eq. (5-5) for piles under lateral
loading since the pile would have contact with soils on both sides. However, in reality,
soils are not in full contact with the piles when the lateral loads are applied. Based on
field test data, Carter (1984) modified Vesic’s equation as follows that would account for
the effect of pile diameter.
12 / 1
p p
4
ref
2
I E
ED
D ) 1 (
ED 0 . 1
K
(
(
¸
(

¸

ν −
= (5-6)
where the reference pile diameter, D
ref
= 1.0 m, E
p
I
p
= flexural rigidity of piles or drilled
shafts.
Guo (2001) proposed an expression for K based on a closed form elastic solution for
laterally loaded piles. The subgrade reaction modulus was given by:
|
|
.
|

\
|
|
|
.
|

\
|

|
|
.
|

\
|
η
η
η −
η
η
η
π
= 1
) ( K
) ( K
) ( K
) ( K
2
2
G 3
K
2
0
1 2
0
1 s
(5-7)
where G
s
= shear modulus of soils; η = a load transfer factor (Guo, 2001); and K
0
and K
1

= Bessel functions.
The above discussions were focused on piles in soils or a linear elastic media. For
drilled shafts in rock masses, Reese (1997) proposed an interim p-y curve criterion for
weak rock. In this criterion, the initial slope of p-y curves (initial modulus of subgrade
reaction), K
i
, was given by:
K
i
≈ k
i
E
m
(5-8)
173
where E
m
= modulus of rock mass; and k
i
= dimensionless constant. From experimental
data, empirical equations for determining k
i
are given as:
k
i
= (100+400z/(3D)); 0 ≤ z ≤3D (5-9)
k
i
= 500; z > 3D (5-10)
where D = the diameter of drilled shafts; and z = depth.
Vesic (1961) elastic solution for beam-on-Winkler foundation is directly applicable
to laterally loaded drilled shafts in soils or rock. Guo (2001) solution is complex for
practical use and requires further verification with actual test data. The equation proposed
by Reese (1997) is empirical in nature and only based on two load tests results. An
empirical model for initial subgrade reaction modulus of rock masses is proposed in this
study. A parametric study using the 3D FEM model presented in Chapter IV is carried
out to obtain an empirical equation for estimating initial subgrade reaction modulus.
Finally, the validation using field test data shows that the proposed empirical equation
would provide good prediction on initial subgrade reaction modulus of rock mass.
5.2.2 FEM Model and Parametric Study
5.2.2.1 FEM Modeling
The 3D FEM model developed in Chapter IV is used to simulate the response of a
laterally loaded drilled shaft in rock. The drilled shaft is modeled as a cylinder with
elastic material properties. The Modified Drucker-Prager Model (CAP Model) is utilized
to represent isotropic, homogeneous rock masses. The solid elements C3D15 and C3D8
are used to develop mesh representation for drilled shaft and rock, respectively. Surface
interface technique is employed to simulate the rock-shaft interface. Since the
174
determination of initial subgrade reaction modulus is the primary objective of this study,
only the elastic response of rock will be the concern of this parametric study.
The rock mass surrounding the shaft will be modeled as multiple horizontal layers
with 12 inch thickness for each layer. The force applied to rock mass per unit thickness
surrounding the shaft is obtained directly from the output of FEM analysis. The rock
resistance per unit shaft length p is then calculated by dividing the force with the rock
layer thickness of 12 inch for each loading level. By extracting the deflections of the shaft
under various load levels at the corresponding depth of the rock layer, p-y curves of the
rock layer can be obtained. The initial subgrade reaction modulus K
i
is obtained as the
initial tangent to the p-y curves. Hyperbolic curve fitting technique is employed to fit the
p-y curves so that a unique value of initial slope of the p-y curves is obtained. Various
factors, such as shaft diameter (D), shaft-rock relative stiffness (E
p
I
p
/(E
m
D
4
)), shaft-rock
interface properties, Poisson’s ratio (ν), and deformation modulus of rock mass (E
m
), are
studied. The diameter of shaft used for parametric study varies from 2 feet to 10 feet. A
total of 30 cases of FEM analysis were carried out.
5.2.2.2 Effect of Rock Mass Modulus E
m

The effect of deformation modulus of rock mass, E
m
, is studied. The range of
modulus was ranged from 100 to 2000 ksi for weak rock. Other pertinent parameters are
kept constant: D = 72 inch, L = 216 inch, E
p
= 4000 ksi, v = 0.3. P-y curve is generated at
a depth of 18 inch. It is found that it has a linear relationship with initial modulus of
subgrade reaction modulus K
i
, as shown in Fig. 5-5.
175
0
1000
2000
3000
4000
0 1000 2000 3000
E
m
(ksi)
K
i

(
k
s
i
)

Figure 5-5 Initial modulus of subgrade reaction for various moduli of rock
5.2.2.3 Effect of Poisson’s Ratio ν
The Poisson’s ratio for rocks is varied from 0.1 to 0.4, while maintaining other
pertinent parameters constant as follows: D = 72 inch, L = 216 inch, Ep = 4000 ksi, E
m
=
200 and 500 ksi. Again, p-y curve is generated at a depth of 18 inch. Fig. 5-6 shows an
exponential relationship between K
i
and Poisson’s ratio ν. The variation of modulus of
rock mass does not change the relationship between K
i
and ν, as shown in Fig. 5-6.
y = 2288e
-1.32x
y = 864e
-1.27x
0
500
1000
1500
2000
2500
0 0.1 0.2 0.3 0.4 0.5
Poisson's Ratio v
K
i

(
k
s
i
)
Em=500 ksi
Em=200 ksi

Figure 5-6 Initial modulus of subgrade reaction for various Poisson’s ratio
176
5.2.2.4 Effect of Depth
The relationship between initial subgrade reaction modulus and depth is also
investigated using the same set of rock properties as discussed in previous section. The
rock properties were set to be same along the depth. It is found that K
i
(the tangent to p-y
curve) decreases along depth, while the ultimate resistance of rock mass increases with
depth, as shown in Fig. 5-7. Because the rock modulus is constant along depth,
supposedly the deformability of rock should be same along depth. The difference of K
i

along depths is very small; therefore, it could be attributed to the numerical integration
for calculating p.
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
0 1 2 3
y (in)
p
(
l
b
/
i
n
)
12 in
36 in
60 in
84 in

Figure 5-7 P-y curves along with depth


177
5.2.2.5 Effect of Interface Properties
The interface between shaft and rock was modeled by a surface based interface
technique in the ABAQUS software. The coefficient of friction of the interface is the
main controlling parameter of the shaft-rock interface. The computational results shown
in Fig. 5-8 clearly indicate that the interface frictions exert no effect on the initial
modulus of subgrade.
100
150
200
0 0.5 1
Coefficient of Friction on Interface
K
i
(
k
s
i
)

Figure 5-8 Initial modulus of subgrade reaction for various rock-shaft interface frictions
5.2.2.6 Effect of Diameter
There has been a long history of debate regarding the effect of diameter on the
modulus of subgrade reaction. Terzaghi (1955) and Vesic (1961) concluded that the
diameter of piles has no effect. Additionally, Ashford and Juirnarongrit (2003) indirectly
verified that the modulus of subgrade reaction is independent of pile diameter when
examining full-scale vibration tests results on cast-in-drilled-hole piles in a sand deposit.
178
However, Carter (1984) found from his own field test data that the diameter of piles has
exerted significant effect on modulus of subgrade reaction.
In this study, the effect of diameter is investigated using the following input
parameters: D ranges from 24 inch to 120 inch, E
p
= 4000 ksi, E
m
= 500 ksi, v = 0.3. P-y
curve is generated at a depth of 18 inch. The FEM computed relationship between the
shaft diameter and initial modulus of subgrade reaction of rock mass is presented in Fig.
5-9. It can be seen that K
i
increases with diameter linearly. The shaft diameter is
normalized with a reference diameter of 1 foot.
0
500
1000
1500
2000
2500
0 3 6 9 12
D/D
ref
(D
ref
=1 ft)
K
i

(
k
s
i
)

Figure 5-9 K
i
varies with shaft diameter
5.2.2.7 Effect of Shaft-Rock Relative Stiffness
The effect of shaft-rock relative stiffness, E
p
I
p
/(E
m
D
4
), is also investigated herein.
The input parameters are: D = 72 inch, L = 216 inch, v = 0.3, E
p
varies from 3500 to
4500 ksi, E
m
varies from 100 to 2000 ksi. Again, the p-y curve is generated for a depth of
18 inch. It is found that a power law can be used to describe the relationship between K
i

179
and shaft-rock relative stiffness, as shown in Fig. 5-10. Similar results concerning the
effect of relative stiffness on K
i
were found previously by Guo (2001) in his numerical
study.
y = 491x
-0.8003
R
2
= 0.9937
0
1000
2000
3000
4000
0 0.5 1 1.5 2 2.5 3
Relative Stiffness, E
p
I
p
/(E
m
D
4
)
K
i

(
k
s
i
)

Figure 5-10 Initial modulus of subgrade reaction for various shaft-rock relative stiffness
5.2.3 Suggested Empirical Equation
Based on the parametric study presented in the previous sections, the following
conclusions may be drawn:
a). Both the modulus of rock masses and shaft diameter exhibit a linear relationship with
K
i
;
b). K
i
exponentially decreases with increasing Poisson’s ratio of rock mass;
c). Rock-shaft interface friction exerts no effect on K
i
;
d). The relative stiffness of the pile and rock has shown a power relationship with K
i
.
A regression analysis on data from the FEM parametric study is carried out. An
equation for predicting initial modulus of subgrade reaction is fitted to match K
i
values
180
obtained from FEM parametric study. As shown in Fig. 5-11, the empirical equation can
be derived as
284 . 0
4
m
p p
2
ref m i
D E
I E
e ) D / D ( E K
|
|
.
|

\
|
=
ν −
(5-11)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
0 500 1000 1500 2000 2500 3000 3500
K
i
(ksi) - Empirical Prediction
K
i

(
k
s
i
)

-

F
E
M
y=x
K
i
=E
m
(D/D
ref
)e
-2v
(E
p
I
p
/E
m
D
4
)
0.284

Figure 5-11 Comparison of FEM computed and predicted subgrade reaction modulus
5.2.4 Validation of the Empirical Equation
5.2.4.1 Field Tests
Three lateral load tests in rock (i.e., Dayton, Pomeroy-Mason, and North Carolina by
Cho, et al., 2000), are employed to verify the developed empirical equation for predicting
initial modulus of subgrade reaction of rock mass. The Dayton and Pomeroy-Mason load
tests were previously presented in Chapter III; and the load test in North Carolina is
collected from open literature. A summary of the properties of the test drilled shafts for
three cases is provided in Table 5-7.
181
Table 5-7 Summary of Lateral Load Test Drilled Shafts
Drilled Shafts Diameter (in) Length (ft) Socket
Length (ft)
Overlying
Soils (ft)
Dayton Shaft #4 72 18 18 0
Pomeroy-Mason Shaft #2 96 (socketed) 112.9 56.8 0
North Carolina Long Shaft 30 16 9.8 4.2

For the lateral load tests in North Carolina, the experimental p-y curves were derived
and presented by the authors. Although the accuracy of the experimental p-y curves was
verified by the authors, an independent verification of the experimental p-y curves is
carried out. The experimental p-y curves are used as input in COM624P analysis. The
predicted load-deflection curve at the shaft head is compared with the measured curve in
Fig. 5-12. It can be seen that a good agreement between the predicted and measured load-
deflection curves at the shaft head is achieved, thus indirectly verifying the
representativeness of the experimental p-y curves.
Load-Deflection, NC Long Shaft
0
50
100
150
200
250
0 0.5 1 1.5 2
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 5-12 Comparison of load-deflection curves of North Carolina load test
182
The experimental p-y curves of rock masses for Dayton and Pomeroy-Mason Load
tests were derived using piecewise polynomial curve fitting. The accuracy of the
experimental p-y curves of the two tests have been verified in Chapter III.
The modulus of rock masses at North Carolina test site was determined from
pressuremeter test results. The other two load tests only have information on the lab test
data of modulus of intact rock. Thus, the rock mass modulus is estimated using the
proposed empirical equation, Eq. (5-4). The mean GSI values determined in previous
sections 3.1.1 and 3.2.1 are used.
The initial linear portion of the pressuremeter test curves of the North Carolina test
was used to determine the modulus of rock masses. The calculated values of E
m
for the
two drilled shafts in the North Carolina lateral load test is provided in Table 5-8. The
values of rock mass modulus for Dayton and Pomeroy-Mason test sites are estimated
using the proposed empirical equation and they are summarized in Table 5-9.
Table 5-8 Modulus of Rock Masses Based on Pressuremeter Test
Drilled Shaft Depth (in) E
m
(psi) E
p
I
p
(lb-in
2
)
NC Long Shaft 102 25252 1.43E+11
NC Long Shaft 141 40060 1.43E+11
NC Long Shaft 180 54998 1.43E+11

Table 5-9 Modulus of Rock Masses Based on Empirical Equation
Test Depth
(in)
GSI E
i

(ksi)
E
m

(ksi)
E
p
I
p

(lb-in
2
)
Dayton Shaft #4 36 40.5 590 38.1 5.23E+12
Dayton Shaft #4 60 40.5 590 38.1 5.23E+12
Dayton Shaft #4 96 61 590 98.1 5.23E+12
Dayton Shaft #4 132 61 590 98.1 5.23E+12
Dayton Shaft #4 156 61 590 98.1 5.23E+12
Pomeroy-Mason Shaft #2 6 42 345 23.9 1.77E+13
Pomeroy-Mason Shaft #2 66 42 345 23.9 1.77E+13
183
5.2.4.2 Verification of the Empirical Equation
The above field test data are used to verify the empirical equation for predicting
initial modulus of subgrade reaction of rock mass. The Poisson’s ratio is assumed to be
0.3 for these tests. The shaft stiffness of test drilled shafts is computed based on
reinforcement and dimension and shown in Tables 5-7 and 5-8. The measured initial
subgrade reaction modulus of rock masses obtained from the experimental p-y curves for
rocks at various depths are provided in Table 5-10.
The predicted K
i
using the proposed equation and the ratio of the predicted values
over the measured values are provided in Table 5-10. The predicted K
i
and measured K
i

are plotted in Fig. 5-13. It can be seen that the predicted values are smaller than the
measured values for most of cases. In general, the empirical equation will provide
conservative predictions on the initial modulus of subgrade reaction.
Table 5-10 Measured and Predicted Initial Modulus of Subgrade Reaction
Test Depth
(in)
K
i
(psi),
Measured
K
i
(psi),
Predicted
Pred./
Mea.
Dayton Shaft #4 36 407787 199467 0.49
Dayton Shaft #4 60 355468 199467 0.56
Dayton Shaft #4 96 454901 392310 0.86
Dayton Shaft #4 132 1016462 392310 0.39
Dayton Shaft #4 156 1557558 392310 0.25
Pomeroy-Mason Shaft #2 6 350938 194066 0.55
Pomeroy-Mason Shaft #2 66 383584 194066 0.51
NC Long Shaft 102 48112 60229 1.25
NC Long Shaft 141 144581 83811 0.58
NC Long Shaft 180 216130 105159 0.49

184
0
500
1000
1500
2000
2500
3000
3500
4000
0 1000 2000 3000 4000
K
i
(ksi) - Empirical Prediction
K
i

(
k
s
i
)

-

F
E
M

&

F
i
e
l
d

D
a
t
a
K
i
=E
m
(D/D
ref
)e
-2v
(E
p
I
p
/E
m
D
4
)
0.284
y=x
FEM Data
Field Data

Figure 5-13 Validation of empirical equation using field test data

5.3 Numerical Solution for Laterally Loaded Piles in A Two Layer Soil Profile
5.3.1 Introduction
Although nonlinear p-y analysis is the most widely used method for design of
laterally loaded piles, an elastic subgrade reaction solution is a good alternative for piles
under small working loads because of its simplicity and easy for use. Subgrade reaction
method models soils as Winkler springs and pile as a beam. A closed form solution for
laterally loaded piles in one layer soil with constant subgrade reaction modulus was given
by Hetenyi (1946). A non-dimensional solution for soil stiffness linearly varying with
depth was also presented by Reese and Matlock (1956) for piles in one layer soil. In
reality, however, piles are often embedded in layered soils, such as clay overlying on
185
rock mass. The stiffness of soil could be varying with depth (such as sand and
sedimentary rock) or keeping constant along depth (such as clay).
For piles in layered soils under small working loads, although the solutions proposed
by Davisson and Gill (1963) and Pise (1982) can be used, they require the soil stiffness to
be constant along depth for each layer. This may not represent the actual soil profiles,
especially when soil stiffness is varying with depth. In this paper, a variational approach,
which has been successfully used by Shen and Teh (2004) for a laterally loaded single
pile in one layer soil, is employed to numerically solve the problem of a laterally loaded
pile in a two-layer soil system using subgrade reaction theory. The soil stiffness, which is
represented by modulus of subgrade reaction of soils, can be either constant along depth
or linearly varying with depth. The proposed solution is then able to be used for layered
soil condition with variation of stiffness, especially for piles socketed in rock with
overburden sands. Realistic soil stiffness distribution along depth can be more accurately
modelled comparing with existing solutions for a layered soil system.
The general computational software Mathematica is used to carry out the numerical
calculation, which makes the solution easy and efficient for use. The numerical solution
is validated against a non-dimensional solution (Reese and Matlock, 1956) for linearly
varying soil stiffness and against Davisson and Gill (1963) non-dimensional charts for a
two-layer soil system with constant soil stiffness. For practical use, methods for
determining modulus of subgrade reaction are provided. Two case studies using the
numerical solution for field lateral load tests are performed to further validate the
practical applicability of the numerical solution. Finally, discussions on the applications
186
of the proposed method show that the lateral deflection of a pile may be under estimated
by a factor of 2.3 if constant soil stiffness is used for the whole soil layer.
5.3.2 Definition of the Problem
It is very common for piles embedded in layered soils. The soil stiffness can be
treated as constant along depth for overconsolidated clay and igneous rocks. However, it
is reasonable to assume linearly varying soil stiffness along depth for sands and
sedimentary rocks. To make this Chapter easier to follow, the definitions of various terms
related with subgrade reaction theory are summarized in Table 5-11. Four possible soil
layer combinations are shown in Fig. 5-14, in which K
s0
, K
r0
= the initial modulus of
subgrade reaction of layer 1 and layer 2, respectively; n
hs
, n
hr
= the constant of subgrade
reaction of layer 1 and layer 2, respectively; L
s
= the thickness of layer 1; L = the total
thickness of the two layers. The four layered cases can be represented by case (a) shown
in Fig. 5-14, by choosing a zero or non-zero value for the two constants of subgrade
reaction.
Table 5-11 Summary of Definitions Related to Subgrade Reaction Theory
Description Symbol Definition Dimension
Soil resistance per unit length p Pressure times diameter F/L
Pile deflection y L
Pile diameter D L
Depth z L
Modulus of subgrade reaction K K=p/y F/L
2
Coefficient of subgrade reaction k k=K/D F/L
3

Constant of subgrade reaction n
h
n
h
= K/z; n
h
= kD/z F/L
3


187

Figure 5-14 Two-layer soil profile with four possible variations
5.3.3 Variational Solution
Shen and Teh (2002) proposed a numerical solution for pile groups under lateral
loads using a variational approach. After that, Shen and Teh (2004) presented a
variational solution for soil with stiffness increasing with depth. This study is a further
development of the variational approach to the layered soil conditions. The solution for
piles embedded in soils having the soil stiffness shown in case (a) of Fig. 5-14 will be
presented. By changing the values of the two constants of subgrade reaction, the response
of piles in the other three layered conditions can be obtained. The soil-pile system is
schematically shown in Fig. 5-15. Here H
t
and M
t
are the applied lateral load and moment
at the top of the pile; p
z
is the soil resistance per unit pile length at depth z; K
1
and K
2
are
the modulus of subgrade reaction of layer 1 and 2; and L and D are the length and
diameter of the pile, respectively.

Figure 5-15 Soil-pile system
z
K K
s0
n
hs
=0
n
hr
K
r0
L
s
L

(Case b)
z
K
K
s0
n
hs
=0
n
hr
=0
K
r0
L
s
L

(Case c)
z
K K
s0
n
hs

n
hr
=0
K
r0
L
s
L

(Case d)
Layer 1
Layer 2
z
K
n
hs

n
hr

K
r0
L
s
L

(Case a)
K
s0
K
1
=K
s0
+n
hs
z
K
2
=K
r0
+n
hr
(z-L
s
)
z
K
K
s0
n
hs

n
hr

K
r0
L
s
L

M
t
H
t
p
z
p
z
D

188
5.3.3.1 Potential Energy
The total potential energy of the pile under lateral loads, as illustrated in Fig. 5-15,
can be expressed as
t
t
t t
2
z
L
s
L
2
2
z
s
L
0
1 p p
M
z
y
y H dz y K
2
1
dz y K
2
1
U


− −

+

+ = π (5-12)
where

|
|
.
|

\
|
=
L
0
2
2
z
2
p p p
dz
dz
y d
I E
2
1
U (5-13)
U
p
is the elastic strain energy of the pile, where E
p
is the pile Young’s modulus; I
p
is the
moment of the inertia of the pile section; and y
z
is the deflection of the pile at depth z.
The second and third terms on the right side of Eq. (5-12) are the energy consumed by the
soil resistance of the first and second soil layers, respectively. The fourth and fifth terms
are the input work done by the lateral load H
t
and the moment M
t
acting at the pile head,
where y
t
is the deflection of the pile at the pile head.
5.3.3.2 Deflection Series
Shen and Teh (2002) employed two finite series to model the deflections of a pile
subjected to lateral loads and moments at the pile head respectively. In this study, the two
finite series is also employed to closely match the pile deflections under lateral loads and
moments. The deflections of a pile subjected to a lateral load at the pile head can be
approximately described in matrix form by
} { } Z { y
h
T
h zh
δ = (5-14)
where
T
h
}
L
z n
sin ...
L
z 2
sin ,
L
z
sin ,
L
z
, 1 { } Z {
π π π
= (5-15)
189
T
hn 2 h 1 h h h h
} ... , , b , a { } { β β β = δ (5-16)
The deflections of a pile subjected to a moment at the pile head can be
approximately represented in matrix form by
} { } Z { y
m
T
m zm
δ = (5-17)
where
T
m
}
L 2
z ) 1 n 2 (
cos ...
L 2
z 3
cos ,
L 2
z
cos ,
L
z
, 1 { } Z {
π − π π
= (5-18)
T
mn 2 m 1 m m m m
} ... , , b , a { } { β β β = δ (5-19)
In above equations, a
h
and a
m
are the unknown constants to represent the rigid body
movement of the pile under lateral loads and moments, respectively; b
h
and b
m
are the
unknown constants to capture the deflections due to the rotation of the pile; and β
hi
and
β
mi
are the unknown constants to capture the nonlinear mode of the deflected pile; and n
is the number of terms used in the trigonometric function. The above expressions for pile
deflections are validated in the subsequent section by the comparisons with the non-
dimensional solutions by Reese and Matlock (1956) and Davisson and Gill (1963). The
total deflection (y
z
) of the pile subjected to lateral loads and moments is the summation of
y
zh
and y
zm
.
5.3.3.3 Minimization of Potential Energy
The principle of minimum potential energy requires that π
p
be an extremum with
respect to the admissible deflection field characterized by the undetermined coefficients
in Eqs. (5-14) and (5-17). This leads to:
) 2 n ... 2 , 1 i ( 0
i
p
+ = =
δ ∂
π ∂
(5-20)
190
where δ
i
denotes the undetermined coefficients in Eq. (5-16) for the lateral load case and
in Eq. (5-19) for the moment case, respectively. By using Eq. (5-16), the potential energy
can then be reduced to:
i
t
t
i
t
t
i
z
z
L
s
L
2
i
z
z
s
L
0
1
i
p
/
z
y
M
y
H dz
y
y K dz
y
y K
U
δ ∂ |
.
|

\
|


∂ +
δ

=
δ ∂


+
δ ∂


+
δ ∂

(5-21)
Eq. (5-21) is the governing variational formulation that a single pile in two layers
soil/rock needs to follow.
5.3.3.4 Numerical Solution
The Eq. (5-21) will be solved by separating the effects of lateral loads and moments
on pile deflections. For the case of piles under a lateral load, by substituting Eqs. (5-13)
and (5-14) into Eq. (5-21), we have:
hi
ht
t h
T
h
hi
zh L
s
L 2
T
h
hi
zh
s
L
0
1
2
T
h
2
hi
2
zh
2
L
0 p p
y
H } ]{ dz } Z {
y
K dz } Z {
y
K dz
z
} Z {
)
z
y
(
I E [
δ

= δ
δ ∂


+
δ ∂


+


δ ∂




(5-22)
For the case of piles under moment, by substituting Eqs. (5-13) and (5-17) into Eq.
(5-21), we have:
mi
mt
t m
T
m
mi
zm L
s
L 2
T
m
mi
zm
s
L
0
1
2
T
m
2
mi
2
zm
2
L
0 p p
)
z
y
(
M } ]{ dz } Z {
y
K dz } Z {
y
K dz
z
} Z {
)
z
y
(
I E [
δ ∂



= δ
δ ∂


+
δ ∂


+


δ ∂




(5-23)
where δ
hi
and δ
mi
are the constants in the vectors {δ
h
} and {δ
m
} respectively; y
ht
and y
mt

are the deflections of the pile at the pile head under the applied lateral load and moment
respectively.
The above two equations can be reduced to matrix forms as:
} H { } ]{ K [
h H
= δ (5-24)
} M { } ]{ K [
m M
= δ (5-25)
191
where [K
H
] and [K
M
] are the matrices reflecting the pile-soil stiffness under a lateral load
and a moment, respectively; {H} and {M} are the vectors reflecting the lateral load and
moment applied at the pile head, respectively.
After the two matrices [K
H
] and [K
M
] are obtained according to Eqs. (5-22) and (5-
23), then the vectors {δ
h
} and {δ
m
} can be calculated. Correspondingly, the deflections of
a pile under a lateral load and a moment can be calculated using Eq. (5-14) and (5-17),
respectively. By superposing y
zh
and y
zm
, the deflection of a pile, y
z
, can be obtained. The
moment of the pile at any depth can be calculated by double differentiating y
z
.
The above described procedure is for free-head pile condition. For a fixed-head pile
under a lateral load, the moment developed at the pile head due to the fixity can be
obtained based on the zero-rotation condition at the pile head. A try-and-error procedure
can be used to find out the value of the generated moment by ensuring the zero rotation
of the pile at the pile head.
5.3.4 Validations
The above calculation procedure is easily coded into Mathematica. It is found that a
size of 10×10 for the stiffness matrices [K
H
] and [K
M
] provides sufficiently accurate
results. The accuracy of the present solution is verified with a non-dimensional solution
by Reese and Matlock (1956) for a single soil layer with stiffness increasing with depth.
Four cases with a constant of subgrade reaction, n
h
, varying from 1.6 to 152 MN/m
3
are
evaluated. The pile used for comparison has a diameter of 0.76 m and a length of 6 m.
The pile head deflections at a lateral load of 44.5 kN and a moment of 27.1 kN-m
predicted by Reese and Matlock (1956) and the proposed solution for various n
h
values
are presented in Fig. 5-16. Apparently, a very good match is obtained.
192
0
100
200
300
400
500
600
0.000 0.200 0.400 0.600 0.800
Deflection (in)
n
h

(
p
c
i
)
Reese & Matlock
Proposed

Figure 5-16 Comparison with Reese and Matlock solutions– varying soil stiffness
For layered soil condition, Davisson and Gill (1963) non-dimensional charts are
employed to validate the proposed solution. Constant modulus of subgrade reaction for
each layer is required by Davisson and Gill (1963) method. A hypothetical case of a
0.305 m diameter concrete pile embedded 7.32 m in a uniform soft clay. A subgrade
reaction modulus, K
r0
, of 1723.8 kPa is selected for the soft clay according to Terzaghi
(1955) recommendations on K for stiff clay. The top portion of the soft clay is then
replaced with 1D (diameter of pile) to 4D thick of medium and stiff clay with K
s0
values
of 3447.5 kPa and 6895 kPa, respectively. The pile has modulus of elasticity of 24.8 GPa.
Both free head with a lateral load of 89 kN and a moment of 135.6 kN-m and fixed head
with a lateral load of 89 kN are considered. Comparisons of deflections at the pile head
and the maximum moments in the pile for cases with varying surface layer thickness L
s

and varying modulus of subgrade reaction K
s0
by Davisson and Gill non-dimensional
193
charts and the proposed solution are presented in Figs. 5-17 and 5-18. It can be seen that
a good match for both deflection and maximum moment predictions is achieved.
0
0.5
1
1.5
2
2.5
3
3.5
0 1 2 3 4 5
L
s
/D
D
e
f
l
e
c
t
i
o
n

(
i
n
c
h
)
Davisson
and Gill
Proposed
K
s0
=250 psi
K
r0
=250 psi, D = 1 ft, Free Head
K
s0
=500 psi
K
s0
=1000 psi

(a) Deflection
500
600
700
800
900
1000
1100
1200
1300
0 1 2 3 4 5
L
s
/D
M
a
x
i
m
u
m

m
o
m
e
n
t

(
k
i
p
-
i
n
)
Davisson
and Gill
Proposed
K
s0
=250 psi
K
r0
=250 psi, D = 1 ft, Free Head
K
s0
=500 psi
K
s0
=1000 psi

(b) Maximum moment
Figure 5-17 Comparisons with Davisson and Gill method for free head condition
194
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 1 2 3 4 5
L
s
/D
D
e
f
l
e
c
t
i
o
n

(
i
n
c
h
)
Davisson
and Gill
Proposed
K
s0
=250 psi
K
r0
=250 psi, D = 1 ft, Fixed Head
K
s0
=500 psi
K
s0
=1000 psi

(a) Deflection
-900
-800
-700
-600
-500
-400
-300
0 1 2 3 4 5
L
s
/D
M
a
x
i
m
u
m

m
o
m
e
n
t

(
k
i
p
-
i
n
)
Davisson
and Gill
Proposed
K
s0
=250 psi
K
r0
=250 psi,D = 1 ft, Fixed Head
K
s0
=500 psi
K
s0
=1000 psi

(b) Maximum moment
Figure 5-18 Comparison with Davisson and Gill solution for fixed head condition
195
5.3.5 Methods for Determining Input Parameters
For soils, modulus of subgrade reaction can be determined using Carter (1984)
equation as previously provided in Eq. (5-6). The value of n
h
of soils can be determined
by fitting a straight line through calculated K values along depth.
If Carter (1984) equation can not be used due to lack of soil input parameters, the
following alternatives can be used. For sand, modulus of subgrade reaction is usually
assumed to be linearly varying with depth. The constant of subgrade reaction n
h
can be
determined from Table 5-12 proposed by Liang (2002). This SPT correlation table was
developed based on a sensitivity study on an extensive lateral load test database. For clay,
modulus of subgrade reaction K is generally assumed to be constant with depth except
the surface portion of the soil. Terzaghi (1955) recommendations on K of 7.16 MPa, 14.3
MPa, and 28.6 MPa for stiff, very stiff, and hard clay can be used for rough estimation.
For rock, the empirical equation Eq. (5-11) presented in previous section can be used
to estimate modulus of subgrade reaction. The value of n
h
of rock can be determined by
fitting a straight line through calculated K values along depth. The following case study
will demonstrate the use of these soil or rock parameter determination tables and
equations.
5.3.6 Case Study
The applicability of the proposed solution for piles in layered soils, especially rock
with an overlying soil layer, is validated against the results of Shaft #1 of Pomeroy-
Mason lateral load test. The lateral load test was performed on a free-head drilled shaft
with a diameter of 102 inch and a total embedment length of 792 inch. The shaft was
embedded in rock with 312 inch of overlying sand.
196
The shaft and soil information are depicted in Fig. 5-19. The average SPT N value of
the sand was 9.5. The modulus of rock mass for various rock depths obtained from Eq.
(5-4) using modulus of intact rock core and mean GSI values are also provided in Fig. 5-
19. It is assumed that the stiffness of the sand and the rock layers increasing with depth.
The constant of subgrade reaction for the sand layer, n
hs
, is taken as 38.3 pci according to
the correlation table by Liang (2002).

Figure 5-19 Shaft and soil profiles of the case study
Two sets values of modulus of subgrade reaction for various depths of weak rock
mass are used for comparison. One set of K
rz
is directly obtained from the experimental
p-y curves derived from strain data in Chapter III, as shown in Fig. 5-20. The depth is
measured from the top of rock. The trendline, which fits the data points meanwhile
ensuring a positive value for K
r0
, gives a value of 7191 psi for K
r0
and a value of 273.5
pci for n
hr
. The other set of K
rz
is correlated from E
m
using the proposed equation (Eq. 5-
11). Based on the values of K
rz
from the proposed empirical equation, K
r0
and n
hr
are
obtained to be 7452.9 psi and 21.8 pci, respectively.
z (inch)
K (psi)
K
s0
=0
n
hs
=20 pci

n
hr

K
r0
L
s
=312
L=792

M
t
= H
t
× 388.8 inch
H
t
Sand

96 inch

z (inch)

Shale
Sandstone
Claystone
SPT
N = 9.5
312

E
m
(psi)
6462
457

618

690

792

654
654
954
102 inch
197
Other required input parameters for the code programmed in Mathematica are given
as: K
s0
=0; L
s
= 312 inch; L = 792 inch; E
p
= 6522 ksi. The comparison between the
measured deflections from inclinometer readings and the predicted values using the
proposed solution at the soil surface is shown in Fig. 5-21. It can be seen that the
proposed method over predicts the deflections. This is desirable from view point of safety
margin. Additionally, it can be seen that the predicted deflection values do not change
much when the n
hr
of the bottom rock layer is varied from 21.8 pci to 273.5 pci. This
conforms to the previous conclusion that the lateral response mainly depends on
properties of the surface layer. From this case study, it can be concluded that the
proposed solution can provide reasonable prediction on deflections of laterally loaded
drilled shafts in a two-layer soil/rock system.
0
50
100
150
200
250
300
350
0 200 400 600
Depth (in)
K
i

(
k
s
i
)
K
i
from
p-y curves
K
i
from proposed
equation

Figure 5-20 Interpretation of subgrade reaction modulus of rock
198
0
50
100
150
200
250
300
0 0.2 0.4 0.6 0.8
Deflection at Surface (inch)
L
o
a
d

(
k
i
p
)
Measured
Predicted - based
on K from proposed
equation
Predicted -
based on K
from p-y
c r es

Figure 5-21 Comparison of shaft head deflections
5.4 Summary and Conclusions
In this chapter, a methodology for determining the deformation modulus of rock
mass is proposed. Next, an empirical equation for estimating modulus of subgrade
reaction modulus of rock mass is developed based on a parametric study using 3D FEM
analysis. Finally, an elastic numerical solution based on subgrade reaction theory is
proposed for a two layer soil/rock profile, especially rock mass overlying with a layer soil,
using variational approach.
Based on above work, the following conclusions are drawn:
1. Dilatometer test is good for weak rock and the Goodman Jack test is good for hard
and competent rock;
199
2. The proposed empirical equation for estimating the deformation modulus of rock
mass can provide reasonable estimation on E
m
according to the field data from
Ironton-Russell;
3. The validation using field test data on the proposed empirical equation for
determining the modulus of subgrade reaction K shows that it can yield reasonable
and conservative results;
4. The proposed elastic numerical solution can match the results from rigorous subgrade
reaction solution for an assumed one layer soil problem;
5. The validation using field lateral load test data shows that the proposed elastic
numerical solution can provide reasonable results for small working loads. Therefore,
the elastic solution provides an efficient way to preliminarily estimate the deflections
of a drilled shaft in layered soil/rock under small working lateral loads.
200



CHAPTER VI
P-Y CRITERION FOR ROCK MASS

In this chapter, a hyperbolic p-y criterion is developed based on considerations of
both theoretical derivations and numerical (finite element) analysis results. The methods
for determining pertinent rock parameters needed for constructing the p-y curves are
presented. Two full-scale lateral load tests introduced in Chapter III are used to validate
the applicability of the developed hyperbolic p-y curves for rock. The comparisons
between the computed shaft responses and the actual measured are fairly good for the
load-deflection curves at the load point as well as the maximum bending moment in the
shaft. Although some materials have been presented in previous chapters, essential
information is duplicated here to make this chapter easy to read.

6.1 General Shape of P-y Curve in Rock
Two full-scale lateral load tests results of fully instrumented drilled shafts socketed
into rock mass have been presented in Chapter III. Piecewise polynomial curve fitting
technique was employed to deduce p-y curves from strain gage readings obtained during
lateral load tests. As can be seen in Figs. 6-1 and 6-2, the general shape of these p-y
curves for rocks could be fit mathematically by a hyperbolic equation given in Eq. (6-1)
and illustrated schematically in Fig. 6-3.
201
u i
p
y
K
1
y
p
+
= (6-1)
where p
u
is the ultimate resistance of rock mass per unit shaft length, and K
i
is the initial
tangent slope to the p-y curves.
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
0 0.05 0.1
y (in)
p

(
l
b
/
i
n
)
36 in-Experimental
132 in-Experimental
36 in-Hyperbola Fit
132 in-Hyperbola Fit
K
i
- Initial Slope
P
u
- asymptote
Hyperbolic Curve

Figure 6-1 P-y curves deduced from shaft #4 of load test at Dayton
0
5000
10000
15000
20000
25000
30000
0 0.02 0.04 0.06 0.08 0.1
y (in)
p

(
l
b
/
i
n
)
6 in - Experimental
6 in - Hyperbola Fit
Hyperbolic Curve

Figure 6-2 P-y curves deduced from Shaft #2 of load test at Pomeroy-Mason
202

Figure 6-3 Schematics of a hyperbolic p-y curve
6.2 Determination of p
u

6.2.1 Failure Modes
A 3D finite element model (FEM) using ABAQUS was used to investigate the
mobilization mechanisms of resistance of rock mass near ground surface and at great
depth. Detailed investigation results have been presented previously in Chapter IV.
For rock mass existing near ground surface, the FEM simulation results show that
the lateral loading of the shaft tends to fail the rock mass as a wedge as shown in Fig. 6-4,
where, F
net
is the total net rock resistance, D is the diameter of the drilled shaft, H is the
height of the wedge, F
a
is the active earth force exerted on the drilled shaft, F
s
is the
friction force on the sides of the wedge, F
n
is the normal force applied to the side faces
and is assumed to be equal to at-rest earth force, F
sb
is the friction force on the bottom
surface, F
nb
is the normal force on the bottom surface determined through force
equilibrium in vertical direction, W is the weight of the wedge, σ
'
v0
is effective
overburden earth pressure on the top of rock and equals to zero if no overlying soil is
present.
K
i
p
u

y
p
Hyperbola
203

Figure 6-4 Failure mode for rock near ground surface
For rock mass at great depth, considering the possible reduction of strength of brittle
rock mass after the maximum stress reaches the peak strength of rock, the ultimate state
(or failure) is defined when the rock has experienced maximum compressive stress equal
to compressive strength and the rock-shaft side shear strength has reached, as shown in
Fig. 6-5, where τ
max
is the maximum side shear resistance between a drilled shaft and
rock mass, and p
L
is the normal limit pressure of rock mass. The patterns of normal and
shear stress distribution at the ultimate state are obtained from FEM simulation results
and could be described as sine functions. In particular, FEM study shows that the normal
stresses are very small at points A and D in Fig. 6-5; whereas, the maximum shear
stresses are observed at points C and E and negligible shear stresses are observed at
points B and D.
F
a
F
net
W

F
n
F
s
F
sb
F
nb
θ
β
H

D
σ

v0
204

Figure 6-5 Failure mode of rock at great depth
6.2.2 P
u
Near Surface
For wedge type of failure, the ultimate rock resistance per unit length, p
u
(in the unit
of Force/Length) of the drilled shaft at depth H is given as follows (See Section 4.5.1 for
more detailed derivation).
5 4 3 2 1 u
C sin C 2 cos C sin C sin cos C 2 p − θ − β + β + β θ =
(6-2)
where
) tan K
2
H
tan K c ( sec tan H C
0 0 v 0 1
φ′ γ′ + φ′ σ′ + ′ θ β = (6-3)
) tan sec tan H 2 sec D ( c tan C C
3 2
θ β β + β ′ + φ′ = (6-4)
β φ′ − β
θ β + θ β + ′ + γ′ + σ′ θ β + γ′ + σ′ β
=
cos tan sin
cos cos C 2 ) tan tan H 2 D ( c ) H 2 ( tan tan H ) H ( tan D
C
1 0 v
2
0 v
3
(6-5)
) H
2
1
( sec tan H K C
0 v 0 4
γ′ + σ′ θ β = (6-6)
D ) z H ( K C
0 a 5
− γ′ = (C
5
≥ 0) (6-7)
) 2 / 45 ( tan K
2
a
φ′ − = (6-8)
Active earth pressure p
a

δ
σ(δ) = p
L
sin δ
p
L
p
u
Normal stress
Shear stress distribution
α
τ(α) = τ
max
sin 2α
τ
max

A
B
C
D
E
205
φ′ − = sin 1 K
0
(6-9)
γ′
σ′

γ′

=
0 v
a
0
K
c 2
z
(6-10)
β = 45+ φ'/2 (6-11)
θ = φ'/2 (6-12)
in which c' is the effective cohesion of the rock mass, φ' is the effective friction angle of
the rock mass, γ' is the effective unit weight of rock mass. Discussions of methods for
determining pertinent rock mass properties will be discussed later in this Chapter.
6.2.3 P
u
at Great Depth
When the existing vertical soil pressure is sufficiently high, for example, as in the
case of shaft/rock interaction at great depth, then FEM simulation results reveal that the
ultimate rock resistance is reached when both the maximum shear and normal pressure of
rock mass have reached their limiting values τ
max
and p
L
, respectively. The distributions
of the shear and normal stresses based on FEM analysis are depicted in Fig. 6-5. As
previously shown in detail in Section 4.5.2 of Chapter IV, the ultimate rock resistance, p
u
,
at great depth can be calculated as follows.
D ) p
3
2
p
4
( p
a max L u
− τ +
π
= (6-13)
where p
a
= active horizontal earth pressure and is given by
a v a a
K c 2 K p ′ − σ′ = (p
a
≥ 0) (6-14)
σ'
v
= effective overburden pressure at the depth under consideration including the
pressure from overburden soils.

206
6.2.4 P
u
of Jointed Rock
For jointed rock with a set of parallel weak planes which dominates the behavior of
rock mass under lateral loads, the above equations for p
u
can still be used after making
the following adjustments. For rock mass failure at a great depth, the high overburden
pressure will prohibit the possible sliding failure on the weak planes. Therefore, Eq. (6-13)
for in-depth rock mass still works for jointed rock.
For rocks that follow the wedge type failure, the bottom plane of the failure wedge
can either be within the rock mass or on a weak plane. It is therefore necessary to
calculate p
u
for both failure modes and that the smaller value is used as the final p
u
. For
failure mode due to shearing failure of rock mass, Eq. (6-2) can be directly used. For
failure mode due to shearing failure through a weak plane, the same wedge failure model
shown in Fig. 6-4 can be used. Eq. (6-2) is therefore valid for calculating p
u
, except that
the angle β is the inclination of the weak plane and the values of c' and φ' for the bottom
face of the failure wedge should be obtained from laboratory tests by shearing samples
along a weak plane. Specifically, the c' and φ' needed for Eqs. (6-4) and (6-5) should
come from the results of lab tests on samples with a weak plane. However, the c' and φ' in
Eq. (6-3) needed for the two sides of the failure wedge can still be obtained using
instantaneous cohesion and friction angle correlated from Hoek-Brown strength criterion.

6.3 Initial Tangent to P-y Curve K
i

The initial tangent to a p-y curve, which has a unit of F/L
2
, was recommended
previously by Reese (1997) as follows.
K
i
≈ k
i
E
m
(6-15)
207
where k
i
= a dimensionless constant. Reese (1997) further suggested the following
empirical correlations:
k
i
= 100 + 400 z/(3D) for 0 ≤ z ≤3D (6-16)
k
i
= 500 for z > 3D (6-17)
Although Reese’s equation is simple to use, it nevertheless does not take into
account other potential influencing factors, such as relative stiffness between shaft and
rock, Poisson’s ratio of rock, and diameter of shaft.
The parametric study presented preveously in Chapter V has led to the findings that
the following semi-empirical equation may work better.
284 . 0
4
m
p p
2
ref m i
D E
I E
e ) D / D ( E K
|
|
.
|

\
|
=
ν −
(6-18)
where E
m
= modulus of rock mass; D
ref
= reference shaft diameter equal to one foot.

6.4 Rock Mass Properties
6.4.1 Strength Parameters
Hoek-Brown (H-B) strength criterion for rock mass (Hoek et al., 2002) possesses
distinct advantages in that it considers the effect of secondary structures of rock (e.g.,
joint, filling, and spacing) on the rock strength. It should be noted that this strength
criterion is not suitable for rock with one or two sets of dominant joints. The H-B
strength criterion can be expressed as
a
s
ci
'
3
b
m
ci
'
3
'
1
|
|
|
.
|

\
|
+
σ
σ
σ + σ = σ
(6-19)
208
where σ
'
1
= the major principal stress at failure, σ
'
3
= the minor principal stress at failure,
σ
ci
= uniaxial unconfined compressive strength of intact rock, GSI = Geological Strength
Index (Marinos and Hoek, 2000), m
b
, s and a = constants that depend on the
characteristics of rock mass and can be estimated as follows (Hoek, et al., 2002).
i
r
b
m
D 14 28
100 GSI
exp m
|
|
.
|

\
|


= (6-20)
|
|
.
|

\
|


=
r
D 3 9
100 GSI
exp s (6-21)
) e e (
6
1
2
1
a
3 / 20 15 / GSI − −
− + = (6-22)
where m
i
is a material constant that can be obtained from Table 4-3 in Chapter IV, D
r
is
a factor that depends upon the degree of disturbance to which the rock mass has been
subjected to due to blast damage and stress relaxation. For deep foundation excavation,
D
r
can be assumed to be zero.
Hoek (1990) further provides the method for estimating Mohr-Coulomb friction
angle φ' and cohesion c' from Hoek-Brown strength criterion as follows.
|
|
.
|

\
|
σ′ − σ′
τ
− = φ′
3 1
2
arcsin 90 (6-23)
φ′ σ′ − τ = ′ tan c
n
(6-24)
where σ
'
1
can be obtained using Eq. (6-19); and
ci b 3 1
2
3 1
3 n
m 5 . 0 ) ( 2
) (
σ′ + σ′ − σ′
σ′ − σ′
+ σ′ = σ′ (6-25)
) ( 2
m
1 ) (
3 1
ci b
3 n
σ′ − σ′
σ
+ σ′ − σ′ = τ (6-26)
209
σ
'
3
is computed as vertical effective stress.
6.4.2 Rock-Shaft Interface Strength
The empirical equation for estimating side shear resistance, τ
max
(MPa), proposed by
Kulhawy and Phoon (1993) is adopted here.
5 . 0
ci
max
42 . 5 σ = τ where the units are in psi (6-27)
6.4.3 Rock Mass Modulus E
m

The empirical equation (6-18) requires input of E
m
in order to compute K
i
. Ideally, it
is preferred to perform dilatometer/pressuremeter tests to obtain in-situ modulus of rock
mass. However, if dilatometer tests could not be performed at the site, then the following
empirical equation derived in Chapter V could be used.
) e (
100
E
E
7 . 21 / GSI i
m
= (6-28)
where E
i
is the modulus of intact rock core that could be determined from unconfined
compression test on rock core samples.

6.5 Construction of P-y curves for Rock Mass
To calculate p
u
according to wedge failure mode (Eq. 6-2), it should be noted that
the value of σ'
3
can be taken as effective overburden pressure at the depth of 1/3H for
estimating the values of φ' and c' using Eqs. (6-23) and (6-24), as the side surface is
triangular. For in-depth failure mode, p
L
is taken as the major principal stress at failure σ'
1
,
which is calculated using Eq. (6-19) by equating σ'
3
to σ'
v
. The effective cohesion c'
needed in Eq. (6-14) can be calculated using Eq. (6-24) by equating σ'
3
to σ'
v
.
210
The value of p
u
is the smaller of the values calculated using Eqs. (6-2) and (6-13).
The main parameters required for calculating p
u
, other than the properties of a drilled
shaft, are: σ
ci
(unconfined compressive strength of intact rock core), GSI, and m
i
. The
value of m
i
can be obtained according to Marinos and Hoek (2000). GSI values are based
on the structure and the surface condition of rock mass. For cored rock, the physical
appearance of the material recovered in the core can be used to estimate GSI value using
the charts presented by Marinos and Hoek (2000).
Additionally, GSI can be obtained from Rock Mass Rating (RMR) (see Hoek and
Brown, 1997). If Bieniawski (1976) RMR
76
is used, the rock mass should be assumed to
be completely dry and a rating of 10 could be assigned to the groundwater, and
adjustment for discontinuity orientation value should be set to zero. If Bieniawski (1989)
RMR
89
is used, the value of 15 should be assigned to groundwater rating and adjustment
for discontinuity value is set to zero. Consequently, the following relationship can be
established.
GSI = RMR
76
(6-29)
GSI = RMR
89
-5 (6-30)

6.6 Comparison of the Proposed P-y Criterion with That of Gabr et al. (2002)
Although the proposed p-y criterion has same mathematic expression as that of Gabr et al.
(2002), the equations for calculating the modulus of rock mass and ultimate resistance of
rock mass are different. Gabr et al. (2002)’s p-y criterion is mainly based on the
equations of Vesic (1961) for calculating K
i
and Zhang et al. (2000)’s equation for
calculating p
u
. As pointed out previously in Chapter II, Vesic (1961)’s equation was
211
developed for the case of beam on elastic foundation, while Zhang et al. (2000)’s
equation was based on a pure assumption of stress distribution of shaft-rock interaction.
However, the proposed p-y criterion employs 3D FEM studies to carefully identify the
failure modes of rock mass under lateral loads and establish an empirical equation for
estimating modulus of rock mass. The main differences of the two p-y criteria are
summarized in Table 6-1.

6.7 Case Studies
The two lateral load tests reported in Chapter III and another two load tests by Gabr
et al. (2002) are used for validating the developed p-y criterion. The soils above the rock
at the test sites were either excavated or isolated using a casing with its diameter larger
than the diameter of test drilled shafts. This was done to eliminate the influence of soils
on the soil/shaft lateral interaction. The diameter of test drilled shafts ranges from 6 ft to
8 ft; and the rock-socket length ranges from 18 ft to 57 ft. The types of rock at these sites
are shale, limestone, claystone, siltstone, sandstone, and mudstone. The unconfined
compressive strength of rock mass ranges from 19 psi to 9073 psi. The modulus of intact
rock mass ranges from 1.5 ksi to 1292 ksi.






212
Table 6-1 Comparison of P-y Criteria
Proposed p-y criterion P-y criterion by Gabr et al. (2002)
P-y
expression
u i
p
y
K
1
y
p
+
=

u i
p
y
K
1
y
p
+
=

K
i
,
modulus of
subgrade
reaction
284 . 0
4
m
p p
2
ref m i
D E
I E
e ) D / D ( E K
|
|
.
|

\
|
=
ν −

This empirical equation is
developed based on a
parametric study using 3D
FEM analysis which is
validated against the Dayton
lateral load test results. Effects
of diameter of drilled shaft,
modulus of rock mass,
Poisson’s ratio of rock, and
shaft-rock relative stiffness on
K
i
are considered. It is
validated against field test data
at Dayton, Pomeroy-Mason,
and North Carolina reported by
Cho et al. (2001).
When dilatometer tests are performed
to obtain E
m
, the following procedure
is used:
For z < T
0
,
12 / 1
p p
4
2
i
I E
ED
1
E 65 . 0
K
(
(
¸
(

¸

ν −
=
This
equation was originally proposed by
Vesic (1961) based on a solution for
beam on elastic foundation.
For z>T
0
,
12 / 1
p p
4
2
T i
I E
ED
1
E 65 . 0
I K
(
(
¸
(

¸

ν −
=

where T
0
is depth of point of rotation,
and T
0
= [1+0.18log(E
p
I
p
/E
m
/L
4
)]L.
The value of K
i
at a depth greater than
T
0
is increased by an empirical factor
I
T
which is equal to
-28 – 383 log (T
0
/L).
E
m
,
modulus of
rock mass
Empirical equation
) e (
100
E
E
7 . 21 / GSI i
m
= can be
used. This equation was
developed based on Ironton-
Russell dilatometer test data
and historical data by
Bieniawski (1978) and Serafin
and Pereira (1983).
Empirical equation proposed by Hoek
and Brown (1997) is used as follows.
) GPa ( 10
100
E
40 / ) 10 GSI ( ci
m

σ
= where
σ
ci
is unconfined compressive strength
of rock core sample. This equation
predicted 6.6 to 10 times of the
measured E
m
according to the
dilatometer test data from Ironton-
Russell project.
P
u
, ultimate
resistance
of rock per
unit length
of shaft
Eq. (6-2) for rock layer near
surface

D ) p
3
2
p
4
( p
a max L u
− τ +
π
= for
in-depth layer
p
L
is calculated using Hoek-
Brown strength criterion and
τ
max
is based on Kulhawy and
Phoon (1993) equation:
) MPa ( 45 . 0
5 . 0
ci max
σ = τ .
p
u
= (p
L
+ τ
max
) D. This was proposed
by Zhang et al. (2000).

p
L
is calculated using Hoek-Brown
strength criterion and
) MPa ( 2 . 0
5 . 0
ci max
σ = τ .

213
6.7.1 Dayton Load Test
The drilled shafts tested at Dayton site are 6 feet in diameter with 18 feet rock socket
in shale. The test set up and the instrumentation installed in the drilled shafts were
depicted previously in Fig. 3-2. A total of 36 #11 primary rebars and #6 spirals are used
to form a reinforcement cage. Details of this full-scale lateral load test were reported in
Section 3.1 of Chapter III. The rock consists of soft to medium gray shale interbedded
with hard gray limestone. The limestone interbeds are typically less than 1 ft thick. The
gray shale is slightly weathered to decomposed, weakly calcareous, and very thinly
laminated. The parameters used for generating the proposed p-y curves for rock are
summarized in Table 6-2. The determination of the values of σ
ci
, GSI and E
i
has been
discussed previously in Section 3.1.1. The values of m
i
are obtained from Table 4-3 based
on rock types previously described in section 3.1.1.
Table 6-2 Input Rock Mass Parameters of Dayton Load Test
Depth (in) γ' (pci) σ
ci
(psi) GSI m
i
E
i
(ksi)
36 0.038 5668 40.5 6 590
60 0.038 5668 40.5 6 590
96 0.038 5668 61 6 590
132 0.038 5668 61 6 590
156 0.038 5668 61 6 590

The p-y curves generated from hyperbolic criterion and tabulated rock properties are
presented in Fig. 6-6. They are fed into LPILE computer program (Reese, et. al., 2004) to
compute the response of the test shaft #4 under the applied lateral loads. The pertinent
properties for drilled shafts are as follows: D = 6 ft; E
p
= 3961000 psi; Poisson’ ratio =
0.3; I
p
= 1319167 in
4
. Non-linear shaft stiffness option in LPILE is used to take into
214
account of the evolution of stiffness due to concrete cracking. The modulus of rock mass
is obtained by correlation using the empirical equation (6-28).
The predicted load-deflection curve at the shaft head is compared with the measured
in Fig. 6-7. In general, a good agreement between the measured and the predicted can be
observed. Although there is some discrepancy between the two curves at small load
levels, the predicted deflections are larger than the measured values (on safe side). As
shown in Fig. 6-8, a good match of the maximum moments in the drilled shaft at various
lateral load levels applied at shaft head is achieved between the measured and computed.
The comparison between the measured deflection-depth curves from inclinometer
readings and those predicted from LPILE analysis is provided in Fig. 6-9. Fig. 6-10
shows the comparison of the measured moment-depth curves from strain gage readings
and those predicted form LPILE analysis. From Figs. 6-9 and 6-10, it can be seen that the
predictions are comparable with those measured values.
0
5000
10000
15000
20000
25000
0 0.05 0.1 0.15
y (in)
p

(
l
b
/
i
n
)
36 in
132 in

Figure 6-6 Hyperbolic p-y curves of Dayton site
215
0
200
400
600
800
1000
1200
0 0.05 0.1 0.15
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 6-7 Comparison of load-deflection of test shaft #4 at Dayton load test
0
200
400
600
800
1000
1200
0 10000 20000 30000 40000
Max. Moment (Kip-in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 6-8 Comparison of load-Maximum moment of test shaft #4 at Dayton load test
216
0
2
4
6
8
10
12
14
16
18
20
-0.05 0 0.05 0.1 0.15
Deflection (in)
D
e
p
t
h

(
f
t
)
1126 kip - Measured
1126 kip - Predicted
634 kip - Measured
634 kip - Predicted
331 kip - Measured
331 kip - Predicted

Figure 6-9 Comparisons of deflection-depth curves of shaft #4 at Dayton test
0
2
4
6
8
10
12
14
16
18
20
-15000 -5000 5000 15000 25000
Moment (kip-in)
D
e
p
t
h

(
f
t
)
786 kip - Measured
786 kip - Predicted
331 kip - Measured
331 kip - Predicted
510 kip -Measured
510 kip - Predicted

Figure 6-10 Comparisons of moment-depth curves of shaft #4 at Dayton test
217
6.7.2 Pomeroy-Mason Load Test
The results of a lateral load test on two drilled shafts socket in shale near Ohio River
between Pomeroy, Ohio and Mason, West Virginia were reported in Section 3.2 of
Chapter III. The test drilled shaft was 8.5 feet in diameter and 112.9 feet long, in which
rock socket length was 56.8 feet. The diameter of the rock socket was 8 feet. A total of
28#18 bars and #6 ties at 12 inch spacing are used to form the reinforcement cage. The
28-day unconfined compressive strength of concrete was 5115 psi. To eliminate soil and
shaft interaction, a 1 inch thick casing with 8.5 ft diameter and a second casing with 11 ft
diameter were installed in the top 56 feet of overburden soil. The details of the load test
set up and instrumentation were previously shown in Fig. 3-10.
The bedrock core recovered was described as shale with interbedded siltstone,
sandstone and mudstone. The properties of rock used to generate p-y curves are
summarized in Table 6-2. The depth shown in Table 6-3 is measured from the top of the
bedrock. Poisson’s ratio of the rock mass is assumed to be 0.3. The modulus of rock mass
is determined by the empirical correlation equation (6-28). The determination of the
values of γ', σ
ci
, GSI and E
i
has been discussed previously in Section 3.1.1. The values of
m
i
are obtained from Table 4-3 based on rock types previously described in section 3.2.1.
Table 6-3 Input Rock Mass Parameters of Pomeroy-Mason Load Test
Depth (in) γ' (pci) σ
ci
(psi) GSI m
i
E
i
(ksi)
6 0.059 3797 42 6 344.8
66 0.059 3797 42 6 344.8
126 0.059 3797 42 6 344.8
210 0.059 3797 42 6 344.8
314 0.060 9073 45 17 1292
391 0.049 19 38 4 1.5
504 0.047 44.3 28 4 1.5
648 0.055 826.2 44 4 81
218
The p-y curves generated from hyperbolic criterion and tabulated rock properties are
presented in Fig. 6-11. The computer program LPILE with the generated p-y curves is
used to compute the response of the test shafts under the applied lateral loads. The effect
of 1 inch thick casing for the drilled shaft above rock is included in the computer analysis.
The computed load-deflection curve at the loading point of the test Shaft #2 is
compared with the actual measured in Figs. 6-12. It can be seen that a good match
between the predicted curve and the measured curve is achieved for the initial portion of
the curve. Although LPILE with the input p-y curves under-predicts the deflections at
large loading levels, the error defined as the ratio of the deflection difference divided by
the measured deflection, is less than 34%. A good agreement between the measured
maximum moment from strain gage readings and the LPILE computed maximum
moment in the shaft is achieved, as shown in Fig. 6-13. The comparison between the
measured deflection-depth curves from inclinometer readings and those predicted from
LPILE analysis is provided in Fig. 6-14. Fig. 6-15 shows the comparison of the measured
moment-depth curves from strain gage readings and those predicted form LPILE analysis.
From Figs. 6-14 and 6-15, it can be seen that the predicted values match those measured.
219
0
5000
10000
15000
20000
25000
30000
0 0.1 0.2 0.3 0.4 0.5
y (in)
p

(
l
b
/
i
n
)
6 in
66 in
391 in

Figure 6-11 Hyperbolic p-y curves of Pomeroy-Mason site
0
50
100
150
200
250
300
0 1 2 3 4
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 6-12 Comparison of load-deflection at the loading point for Pomeroy-Mason test
220
0
50
100
150
200
250
300
0 50000 100000 150000 200000
Max. Moment (Kip-in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted

Figure 6-13 Comparison of load-Maximum moment of Pomeroy-Mason load test
0
20
40
60
80
100
120
-1 0 1 2 3 4 5
Deflection (in)
D
e
p
t
h

(
f
t
)
275 kip - Measured
275 kip - Predicted
150 kip - Measured
150 kip - Predicted
50 kip - Measured
50 kip - Predicted

Figure 6-14 Comparisons of deflection-depth curves of shaft #2 at Pomeroy-Mason test
221
56
66
76
86
96
106
116
-50000 0 50000 100000 150000 200000
Deflection (in)
D
e
p
t
h

(
f
t
)
275 kip - Measured
275 kip - Predicted
150 kip - Measured
150 kip - Predicted
50 kip - Measured
50 kip - Predicted

Figure 6-15 Comparisons of moment-depth curves of shaft #2 at Pomeroy-Mason test
6.7.3 Load Tests at North Carolina
Two lateral load tests at North Carolina were reported by Gabr et al. (2002). The
drilled shafts are 2.5 feet in diameter with 9.15 feet to 13.8 feet rock socket in siltstone
and sandstone. The test set up is depicted in Figs. 6-16 and 6-17 for the load test at I-40
and I-85, respectively. The reinforcement was 12 #10 primary rebars and a 0.5 inch thick
casing. The rock consists of soft to medium hard siltstone and sandstone. The parameters
used for generating the proposed p-y curves for rock are reported by Gabr et al. (2002)
and summarized in Table 6-4 and Table 6-5 for load test at I-40 and I-85, respectively.
222

Figure 6-16 Layout of I-40 load test

Figure 6-17 Layout of I-85 load test
Table 6-4 Input Rock Mass Parameters of I-40 Load Test
Depth (in) γ' (pci) σ
ci
(psi) GSI m
i
E
m
(psi)
Short Shaft
35 0.092 1639 87 9 23345
90 0.092 1769 74 9 28362
121 0.092 5061 76 19 63351
Long Shaft
24 0.092 1769 57 9 25230
75 0.092 4002 66 14 57203
122 0.046 3756 76 14 54201
150 0.046 3538 74 14 50620
3.3 ft
Jack
9 ft
Short Shaft
Long Shaft
Dial Gage
12 #10
Rebar
Rock
2.5 ft
3.3 ft
13.8 ft
0.5 inch
casing
3.3 ft
Jack
11 ft
Short Shaft
Long Shaft
Dial Gage
12 #10
Rebar
Rock
2.5 ft
3.3 ft
13.3 ft
0.5 inch
casing
223
Table 6-5 Input Rock Mass Parameters of I-85 Load Test
Depth (in) γ' (pci) σ
ci
(psi) GSI m
i
E
m
(psi)
Short Shaft
23.6 0.055 4220 59 9 15646
61 0.055 3596 59 9 13340
90.2 0.055 6598 59 9 48749
Long Shaft
29.5 0.055 3625 38 9 32509
78.7 0.055 4162 38 9 15428
112 0.055 4785 38 9 167026
145 0.055 4785 59 9 87667

The p-y curves generated from the hyperbolic criterion and tabulated rock properties
are fed into LPILE computer program (Reese, et. al., 2004) to compute the response of
the test shafts under the applied lateral loads. Non-linear shaft stiffness option in LPILE
is used to take into account of the evolution of stiffness due to concrete cracking.
The predicted load-deflection curve at the shaft head is compared with the measured
in Figs. 6-18 and 6-19 for I-40 short and long shaft, respectively. In general, a good
agreement between the measured and the predicted can be observed. The prediction
results of I-85 short and long shaft are presented in Figs. 6-20 and 6-21, respectively.
Smaller deflections are predicted for this load test. However, the discrepancy is not very
significant, especially for the case of long shaft.
224
0
50000
100000
150000
200000
250000
300000
350000
400000
0 0.2 0.4 0.6
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted

Figure 6-18 Comparison of load-deflection curves of I-40 short shaft
0
50000
100000
150000
200000
250000
300000
350000
400000
0 0.2 0.4 0.6 0.8
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted

Figure 6-19 Comparison of load-deflection curves of I-40 long shaft
225
0
50000
100000
150000
200000
250000
300000
350000
0 0.5 1 1.5 2
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted

Figure 6-20 Comparison of load-deflection curves of I-85 short shaft

0
50000
100000
150000
200000
250000
300000
350000
0 0.2 0.4 0.6 0.8
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted

Figure 6-21 Comparison of load-deflection curves of I-85 long shaft

226
6.8 Conclusions
A hyperbolic p-y criterion for rock is developed in this study that can be used in
conjunction with either COM624P or LIPLE computer analysis program to predict the
deflection, moment, and shear responses of the shaft under the applied lateral loads. 3-D
finite element simulation results have provided basic understanding of the mobilization
mechanisms of lateral resistance of rock mass to the drilled shafts, from which analytical
equations are derived for computing p
u
. Hoek-Brown strength criterion for the rock mass
as well as pertinent empirical equations for the rock/shaft interface strength have been
utilized in deriving the analytical equations. Proper considerations for the effects of joints
and discontinuities on the rock mass modulus and strength are discussed. Evaluations
based on comparisons between the predicted and measured responses of full-scale lateral
load tests on fully instrumented drilled shafts have shown the practical uses of the
proposed p-y criterion and the associated methods for determining the corresponding
input of rock properties.
227



CHAPTER VII
DERIVING P-Y CURVE FROM DILATOMETER TESTS

7.1 Pressuremeter and Dilatometer
The term pressuremeter was first used by Menard to describe his testing equipment
developed in 1955. The widely recognized definition of pressuremeter is as follows: a
cylindrical probe that has an expandable flexible membrane designed to apply a uniform
pressure to the walls of a borehole. Pressuremeter is usually used for the testing in soils
and weak rock due to the limit pressure that can be applied (10 MPa). For hard rock, the
high limit pressure (30 MPa) testing device dilatometer, which shares the same testing
principle as the pressuremeter, can be used. Figure 7-1 and 7-2 show respectively the
pressuremeter and dilatometer manufactured by RocTest.
Usually, the raw data collected during the pressuremeter/dilatometer test consists of
the pressure read on the gauge of the control unit P
R
and the volume read on the volume
measuring device of the control unit V
R
. Two calibrations are needed to correct the raw
data. The pressure calibration of the probe is performed at ground level, beside the
control unit, unconfined, to establish the pressure-volume relationship of the probe itself.
This is a measure of the probe inertia. The second calibration is the volume calibration of
the complete system, including the probe, coaxial tubing and control unit circuitry. The
probe is confined by placing it in a steel casing. The pressure-volume relationship of the
228
system is determined. This calibration is a measure of the intrinsic volumetric expansion
of the components under pressure. The calibration of volume and pressure losses is
schematically presented in Fig. 7-3.

Figure 7-1 Menard G-Am pressuremeter from RocTest

Figure 7-2 PROBEX-1 Dilatometer from RocTest

Control Unit
Probe

ACCULOG-X
Readout
Manual
Pump
Probe
229
The measured pressure and volume can be corrected using the following equations
per ASTM D 4719.
c R
P P P P − + =
δ
(7-1)
) P P ( a V V
R R δ
+ − = (7-2)
where P is the corrected water pressure exerted by the probe on the soil,
P
R
is the pressure reading on the control unit,
P
δ
is the hydrostatic pressure between control unit and the probe,
P
c
is the pressure correction due to stiffness of instrument at corresponding
volume, determined in Fig. 7-3,
V is the corrected increase in volume of the measuring portion of the probe,
V
R
is the volume reading on readout device, and
a is the slope shown in Fig. 7-3.

Figure 7-3 Calibration for volume and pressure losses
For dilatometer test data, the two calibration procedures are the same as those of
pressuremeter. However, the equations used for correction is slightly different. Due to the
Slope = a
P
c
V
o
l
u
m
e

I
n
j
e
c
t
e
d

i
n

P
r
o
b
e
,

V
R

V
R

Pressure in Probe, P
R
+P
δ
Pressure calibration curve
Volume calibration curve
230
nature of dual-action oil-type hydraulic pump system, the pressure readings can be
corrected using the following equation.
h 82 . 8 P 955 . 0 P
R
∆ + = (7-3)
where ∆h is the difference in elevation between the manual pump and the center of the
dilatometer probe in meters.
Two deformation components contribute to the value of “a” shown in Fig. 7-3. They
are the intrinsic volumetric expansion of the dilatometer system known as “c” and the
small expansion undergone by the calibration tube during pressurization. The expansion
of the thick wall metallic tube is determined theoretically and is expressed by the “b”
parameter. Because the volume calibration of dilatometer test reaches high pressure,
therefore the volume change of the steel casing needs to be considered. The following
equation can be used to correct the volume readings.
cP V V
R
− = (7-4)
where c = a – b, and
e E
)] m 1 ( e r [ V 2
b
tube
tube
+ +
= (7-5)
where V
tube
is the volume taken by the dilatable membrane of the probe of the dilatometer
when in contact with the metallic calibration tube, r is the internal radius of the
calibration tube, e is the wall thickness of the calibration tube, m is the Poisson’s ratio of
the calibration tube material, and E
tube
is the modulus of elasticity of the calibration tube
material. The PROBEX-1 dilatometer from RocTest is supplied with a standard, steel
calibration tube. Assuming a dilatable length of the PROBEX-1 dilatometer membrane
equal to 18 inch (457 mm), one obtains a “b” value of 84.57 × 10
-6
cm
3
/kPa.
231
A typical corrected pressure-volume curve is presented in Fig. 7-4, where P
i
is the
corrected pressure when the probe made contact with the borehole, V
i
is the corrected
volume reading at the pressure of P
i
, ∆P is the corrected pressure increase in the center
part of the straight line portion of the pressure-volume curve, ∆V is the corrected volume
increase in the center part of the straight line portion of the pressure-volume curve,
corresponding to ∆P pressure increase, P
y
is the yield pressure at the end of the straight
line portion of the pressure-volume curve, the creep curve is pressure versus the volume
difference between the volume measured at 60 seconds and 30 seconds, and V
m
is the
corrected volume in the center portion of the ∆V volume increase. The values of P
i
and
P
y
can be determined from the intersections of the creep curve. With the corrected P-V
curve, the soil/rock shear modulus, Young’s modulus, shear strength, and p-y curve can
be obtained approximately by assuming that the expansion of the pressuremeter probe is
considered to be the expansion of an infinitely long cylinder in an elastic infinite mass of
soil/rock.

Figure 7-4 Typical pressuremeter/dilatometer test curve
C
o
r
r
e
c
t
e
d

V
o
l
u
m
e
,

V

Corrected Pressure Applied to Borehole Wall, P

Creep curve
Test curve
V
m
∆V

∆P
P
i
P
y
V
i
232
7.1.1 Modulus of Rock Mass
By assuming the rock mass is an elastic media and the cavity is expended in
infinitely long, the pressuremeter/dilatometer modulus E of rock mass or soils can be
determined as follows (ASTM D 4719):
V
P
) V V )( 1 ( 2 E
m 0


+ υ + = (7-6)
where υ is Poisson’s ratio, and V
0
is the initial volume of the measuring portion of the
uninflated probe at the ground surface.
7.1.2 Limit Pressure
According to ASTM D4719, the limit pressure (P
l
) is defined as the pressure where
the probe volume reaches twice the original soil/rock cavity volume, defined as the
volume V
0
+V
i
where V
i
is defined in Fig. 7-4. The volume reading at the twice of the
original soil cavity volume is (V
0
+ 2V
i
).
The limit pressure is usually not obtained by direct measurements during the test due
to the limit of the probe expansion or the need for excessively high pressure. If the test
was conducted to read sufficient plastic deformation, the limit pressure can be determined
by a 1/V versus P plot, as shown in Fig. 7-5.
7.1.3 Undrained Shear Strength
The undrained shear strength S
u
of cohesive soils or soft rock can be determined in
various ways, including the limit pressure method (Bishop et al., 1954), empirical
correlations (Menard, 1970), the yield pressure method, and the Gibson-Anderson
method.
233
0.0
100.0
200.0
300.0
400.0
500.0
600.0
700.0
800.0
900.0
1000.0
0 0.001 0.002 0.003 0.004
1/V (1/cc)
C
o
r
r
e
c
t
e
d

P
r
e
s
s
u
r
e

(
k
P
a
)
1/(V
0
+2V
i
)
P
l
Linear Extropolation

Figure 7-5 Determination of limit pressure from inverse of volume versus pressure
Bishop et al. (1945) derived a theoretical expression of the limit pressure using
plasticity theory with Tresca criterion:
|
|
.
|

\
|
ν +
+ + =
u
u i l
S ) 1 ( 2
E
ln 1 S P P (7-7)
where P
l
is the limit pressure determined from Fig. 7-5, and P
i
is the horizontal in-situ
earth pressure at rest determined from Fig. 7-4. Equation (7-7) can be rewritten as:
β
=
*
l
u
P
S (7-8)
where the net limit pressure P
l
*
is equal to P
l
-P
i
. The value of β can range from 5.2 to 7.5
for typical ratio of E/S
u
varies from 200 to 2000 for clay. Menard (1970) proposed a
value of 5.5 for computing the residual strength of clay.
The yield pressure method was based on the theoretical expression of the yield
pressure P
y
:
234
i y u
P P S − = (7-9)
Both P
y
and P
i
are obtained from the pressuremeter curve. This method is not
recommended as it typically overestimates the undrained shear strength (Briaud, 1992).
The Gibson-Anderson (1961) method is based on the theoretical expression of the
pressuremeter curve after the yield pressure:
|
|
.
|

\
| ∆
+ + =
V
V
ln
S
G
ln S P P
u
u y
(7-10)
where P is the corrected pressuremeter pressure, G is the shear modulus, ∆V is the
increase in cavity volume, and V is the current volume of the cavity. A plot of P versus
ln(∆V/V) for the pressuremeter test data points past the yield (or creep) pressure leads to
a straight line. The slope of the straight line is S
u
.
Baguelin et al. (1978) presented an extensive comparison of undrained shear
strength S
u
and P
l
*
based on pressuremeter tests on clay. Based on Baguelin et al. (1978)
database and Briaud’s own data, Briaud (1989) developed a nonlinear relationship
between S
u
and P
l
*
by performing a regression analysis.
75 . 0
i l u
) P P ( 25 . 0 S − = with units of ksf (7-11)
This equation was adopted by FHWA (Briaud, 1989) and is suggested for future use
as it is based on an extensive database and is simple for use. However, it should be noted
that this equation was based on test data on clay.
7.1.4 Friction Angle
There are several ways of obtaining the friction angle φ of a cohesionless soil from a
pressuremeter test; however, none of them is very satisfactory (Briaud, 1992). These
235
methods include the yield pressure method (Briaud, 1992), the limit pressure method
(Briaud, 1992), and the Hughes-Wroth-Windle method (Hughes, et al. 1977).
The yield pressure method employs the theoretical expression of the effective stress
yield pressure P
y
’:
) sin 1 ( P P
i y
ϕ +

=

(7-12)
where P
i
’ is the effective horizontal pressure at rest. This method is not used because it is
too difficult to determine P
y
’ with enough precision.
The limit pressure P
l
’ can also be used to theoretically obtain the friction angle.
)
a
K 1 ( 5 . 0
i
i l
sin P
G
) sin 1 ( P P

|
|
.
|

\
|
ϕ

ϕ +

=

(7-13)
where K
a
is the coefficient of active earth pressure, G is shear modulus of the soil. This
method is not used because of a few shortcomings as pointed out by Briaud (1992). For
instances, the method assumes no volume change of soil. It is based on theory of
expansion of infinitely long cavity while actual pressuremeter probe has a finite length.
Additionally, the value of G needs to be evaluated.
The Hughes-Wroth-Windle method (Hughes, et al. 1977) takes into consideration
the dilatancy of cohesionless soils. The method is based on the theoretical expression of
the pressuremeter curve past the yield pressure. This expression is:
t tan cons ) u p log(
sin 2
sin 1 ) sin 1 ( K
2
C
R
R
log
0
c
c
+ −
|
|
.
|

\
|
ϕ′
ϕ′ + + ϕ′ −
=
|
|
.
|

\
|
+

(7-14)
where R
c
is the initial radius of the cavity, ∆R
c
is the change in radius of the cavity, C is
the intercept of the volumetric strain versus engineering shear strain plot, P is the total
236
pressuremeter pressure, u
0
is the pore water pressure, φ

is the peak friction angle and K
is:
|
.
|

\
| ϕ′
+ =
°
2
45 tan K
CV 2
(7-15)
where φ

CV
is the friction angle at constant volume. For dense sand, the value of C is zero;
for medium and loose sands, the value of C can be approximately assumed to be zero.
Recommended values of φ

CV
are presented in Table 7-1. The results of the pressuremeter
test are plotted as log(P-u
0
) versus log(∆R
c
/ R
c
), and the slope s of the line at large
deformation is a function of φ

only, thus allowing determination of φ

.
ϕ′ + + ϕ′ −
ϕ′
=
sin 1 ) sin 1 ( K
sin 2
s (7-16)
It should be noted that this technique was developed for self-boring pressuremeter test
results and unproven for preboring pressuremeter tests.
Table 7-1 Preliminary estimates of φ

CV
(Robertson and Hughes, 1986)
Soil Type φ

CV
(˚) Soil Type φ

CV
(˚)
Well-graded gravel-sand-silt 40 Uniform medium sand 34
Uniform coarse sand 37 Well-graded fine sand 34
Well-graded medium sand 37 Uniform fine sand 30
Assign lower values for well-rounded particles.
Assign higher values for angular particles.

For weak rock, Haberfield and Johnston (1993) presented a curve fitting technique
to estimate the strength parameters. The method is based on the theory of expansion of an
infinitely long cylindrical cavity in an elastic, perfectly plastic, homogeneous and
isotropic Mohr-Coulomb material with cohesion c and internal friction angle φ.
237
The drained conditions are assumed and all parameters are expressed in effective
stress parameters. Before yield, the expansion of the cavity is governed by:
G 2
P P
i
c

= ε (7-17)
where ε
c
is the cavity strain, P is the probe pressure, P
i
is the in situ horizontal stress and
G is the shear modulus of the weak rock.
After yielding, the cavity expansion for pressure in excess of the yield pressure P
y
is
given by:
(
(
¸
(

¸

+
|
|
.
|

\
|
+
|
|
.
|

\
|
= ε
− −
3
n / ) 1 n (
0
y
2
m / ) 1 m (
0
y
1 c
b
r
r
b
r
r
b
G 2
1
(7-18)
where r
y
is the radius of the yield zone around the cavity, r
0
is the original radius of the
cavity, and
) P P (
) n m (
) mn 1 (
) 1 (
1 m
m 2
b
i y 1

(
¸
(

¸

ν −
+
+
ν −


= (7-19)
) P P (
) n m (
) 1 m (
) 1 ( n 2 b
i y 2

+
+
ν − = (7-20)
) P P (
) 1 m (
) 1 m (
) 2 1 ( b
i y 3


+
ν − = (7-21)
) 1 m /( m
y 0
y
ˆ ) 1 m ( P
ˆ ) 1 m ( P
r
r

(
(
¸
(

¸

σ + −
σ + −
= (7-22)
ϕ −
ϕ +
=
sin 1
sin 1
m (7-23)
ψ −
ψ +
=
sin 1
sin 1
n (7-24)
238
ϕ −
ϕ
= σ
sin 1
cos c 2
ˆ (7-25)
where ψ is the dilation angle of the weak rock, and ν is the Poisson’s ratio.
This method requires adjusting six parameters G, ν, c, φ, ψ, and P
i
to fit
pressuremeter curves. Given values of G, ν, ψ, and P
i
, the values of c and φ can be
obtained by fitting the pressuremeter test curve. However, this method is difficult to use
without the aid of a computer program.

7.2 Deriving p-y Curves from Pressuremeter/Dilatometer Test Results
There are several methods available for deriving p-y curves from pressuremeter test
results. However, all these methods were developed based on pressuremeter tests in soils.
The applications of these methods for pressuremeter tests in rock have not been verified
yet.
Menard et al. (1969) developed a method for deriving p-y curves from preboring
pressuremeter test. This method was presented by Baguelin et al. (1978) in English. It
considers that the p-y curve to be bilinear elastic and perfect plastic. Based on Menard’s
analysis on settlement of a strip footing, the value of first slope of p-y curve k was
proposed as follows.
D
E 6 D
D
65 . 2 D
E 9
2
k
1
0
0
α
+
|
|
.
|

\
|
=
α
(D>0.6m or 2 feet) (7-26)
18
3 ) 65 . 2 ( 4
E
D
k
1 α +
=
α
(D<0.6m or 2 feet) (7-27)
239
where D = the diameter of piles; the reference diameter D
0
= 0.6 m (2 feet); E = the
modulus of soils from pressuremeter test; α = rheological factor, which is dependent on
the soil type and the ratio E/P
l
*
(P
l
*
is the net limit pressure), as given in Table 7-2.
Near the ground surface, the soil reaction is less than at great depth because of
possibility for soil heave. For soils above the critical depth z
c
, which is in the order of 2D
for cohesive soils and 4D for granular soils, the subgrade reaction will be reduced by a
factor λ
z
.
2
) z / z ( 1
c
z
+
= λ (7-28)
where z = depth.
Table 7-2 The Rheological Factor α for Various Soils (Baguelin et al., 1978)
Peat Clay Silt Sand
Sand and
gravel Soil Type
E/P
l
*
α E/P
l
*
α E/P
l
*
α E/P
l
*
α E/P
l
*
α
Over-
consolidated
>16 1 >14 2/3 >12 1/2 >10 1/3
Normally
consolidated
1 9-16 2/3 8-14 1/2 7-12 1/3 6-10 1/4
Weathered
and/or
remolded
7-9 1/2 1/2 1/3 1/4

Rock Extremely fractured Other
Slightly fractured or
extremely weathered
α = 1/3 α = 1/2 α = 2/3

The typical p-y curve by Menard et al. (1969) is shown in Fig. 7-6. When the soil
pressure p/D (p is soil resistance per unit length of pile; and D is the shaft diameter)
reaches the yield pressure P
y
, the first linear line stops at this point. The slope value k
p
of
the second linear part is half of the k of the first part. The ultimate unit soil reaction p
u
/D
is equal to the limit pressure p
l
of pressuremeter test for the depth greater than the critical
depth. For soils above the critical depth, the value of k, k
p
, and p
u
should be reduced by
240
the factor λ
z
. This method tends to give deflections larger than the measured deflection
often by a factor of two (Baguelin et al., 1978).

Figure 7-6 P-y curves from pressuremeter (Baguelin et al., 1978)
Baguelin, et al. (1978) method was incorporated in a more complete form in a
design manual by the French Petroleum Institute in 1983. It uses the results of selfboring
pressuremeter tests. The p-y curve at a depth z for the pile is obtained from the
pressuremeter expansion curve at same depth z as follow:
D P p
*
η = (7-29)
R
V
V
2
1
y
0

= (7-30)
where p is the soil resistance on the pile expressed as a force per unit length of pile, y is
the pile horizontal displacement, η is the lateral resistance factor varying from 0.33 to 3,
P
*
is the net pressure (P-P
i
) in the pressuremeter curve, R is the pile radius, V
0
is the
initial volume of the probe, ∆V is the volume injected into the probe, and P
i
is the in-situ
at-rest horizontal stress.
p/D (F/L
2
)
P
l

P
y

0
y
k
1
k
p
=0.5k
1
p
u
/D = P
l

z ≥ z
c
p
u
/D = P
l
/2
z = 0

241
Robertson et al. (1983) refined a method which was previously proposed by Hughes
et al. (1979) to derive p-y curves directly from pressuremeter tests. The similarity of
pressuremeter test and lateral loading of pile was the basis of the method. The effect of
pile driving on soils was simulated with a self-driving pressuremeter.
For piles under lateral loading, the limiting lateral resistance from clay is
approximately 9 S
u
, where S
u
is the undrained shear strength. Whereas in the case of the
pressuremeter, the limiting pressure (P
l
– P
i
) is approximately 5S
u
. Based on these
assumptions on the relationship between limit resistance and undrained shear strength of
clay, the following procedure was proposed by Robertson et al. (1983) and illustrated in
Fig. 7-7.
D P p
*
α = (7-31)
D
V
V
4
1
y
0

= (7-32)
where D is the diameter of piles, α is 2 for clays and 1.5 for sands (Robertson et al.,
1986). This method is similar to Baguelin et al. (1978) method except the value of
coefficient α or η is different.

Figure 7-7 Steps for constructing p-y curve from pressuremeter test
242
Briaud, et al (1983) developed a method for deriving p-y curves from the results of
pressuremeter test. This method considers that a p-y curve is made of a front resistance
Q-y curve and a friction resistance F-y curve. This method is also applicable to test
results of dilatometer.
The Q-y and F-y curves can be obtained point by point from the pressuremeter curve
as follows:
) D )( P )( SQ ( Q
*
= (7-32)
where Q = the frontal soil resistance on the pile,
D = pile diameter or width,
P
*
= the net pressure which is equal to (P-P
i
),
SQ = shape factor for pressure reaction,
=
4
π
for circular piles,
=1.0 for square piles.
X
P
) X 1 )( X )( D )( SF ( F
*


+ = (7-33)
in which, F = the frictional soil resistance on the pile,
SF = shape factor for shear reaction,
= 0.79 for circular piles,
= 1.76 for square piles,
∆P
*
= the increase of net pressure,
X = ∆V’/V
0

where V
0

is equal to V
0
+V
i
and ∆V

is the volume injected in the
probe from V
0

point on.
243
0
V
V
4
D
y

′ ∆
= (7-34)
where y = the horizontal displacement of the pile,
After Q, F, and y values corresponding to each data point of pressure-volume curve
are obtained, the p-y curve can be constructed from the summation of Q-y curve and F-y
curve using the equation p = Q + F.
At shallow depth, the lack of vertical confinement influences the results of
pressuremeters as well as the soil resistance to laterally loaded drilled shafts. The effects
due to lack of vertical confinement was introduced by Smith (1983) as follow: no
influence is considered for the F-y curve, while for the Q-y curve the critical depth D
c
is
determined from Fig. 7-8 and within that depth Q is multiplied by a reduction factor RF
from Fig. 7-9. This reduction was developed for soils, it will not be considered for rock
since for most cases a thick layer of soil exists above the rock mass.

Figure 7-8 Determination of the critical depth (Smith, 1983)

244

Figure 7-9 Reduction factor for depth within critical depth (Smith, 1983)
The above reviewed methods for deriving p-y curves from pressuremeter test results
were developed for applications in soils. Briaud et al. (1983) method is considered as the
most suitable method for applications in rock, as it does not require empirical coefficients
and avoids the difficulty of determining the yield and limit pressure. Additionally, it
considers both shear stress and normal stress as a result of soil or rock-shaft interaction.
Therefore, it is decided to evaluate Briaud et al. (1983) method for applications to drilled
shafts socketed in rock.

7.3 Evaluation
Briaud et al. (1983) method for deriving p-y curves from dilatometer test results is
evaluated against field tests reported by Cho et al. (2001) and Gabr et al. (2002). The
lateral load test by Cho et al. (2001) was performed on a free-head drilled shaft with a
diameter of 30 inch and an embedment length of 168 inch. The shaft was embedded in
soft weathered meta-argillite rock with 39 inch of overlying sandy silt and a 12 inch thick
of dense sand.
245
The pertinent shaft and soil information are shown in Fig. 7-10. The average SPT N
values of the sandy silt layer and dense sand layer were reported as 14 and greater than
100, respectively. Dilatometer (DM) tests were performed at depths of 8.5, 11.8, and 15
feet. The results of the dilatometer tests are provided in Fig. 7-11. P-y curves of the test
site interpreted from DM test results using Briaud et al. (1983) method are shown in Fig.
7-12.
These interpreted p-y curves are input into LPILE program to predict the lateral
response of the test drilled shaft. The sandy silt and dense sand is modeled using LPILE
internal sand p-y curves. The friction angles of the sandy silt and dense sand layers were
interpreted as 32˚ and 45˚ according to the SPT correlation table by Liang (2002).
Nonlinear option of shaft stiffness is chosen for analysis. The predicted deflections at the
top of the shaft are compared with the measured values, as shown in Fig. 7-13. It can be
seen that the predicted deflections are little bitter smaller than the measured values.

Figure 7-10 Shaft and soil profiles of the case study
M
t
= H
t
× 12 inch H
t
Sandy silt

30 in.
z (inch)

Soft
weathered
meta-
argillite
rock

SPT
14
39

74

94

168
114
133
153
24 in.
51

Dense sand
100+
8.5 ft
DM

11.8 ft
15 ft

246
0
5000
10000
15000
20000
25000
30000
0 200 400 600 800
Volume (cc)
P
r
e
s
s
u
r
e

(
k
P
a
)
Depth 8.5 ft
Depth 11.8 ft
Depth 15 ft

Figure 7-11 Dilatometer test results of the case study (after Cho et al., 2001)
0
20000
40000
60000
80000
100000
120000
140000
160000
0 0.5 1 1.5
y (in)
p

(
l
b
/
i
n
)
Depth 8.5 ft
Depth 11.8 ft
Depth 15 ft

Figure 7-12 P-y curves from dilatometer tests of Cho et al. (2001)
247
0
50
100
150
200
250
0 0.5 1 1.5 2
Deflection (in)
L
o
a
d

(
k
i
p
s
)
Measured
Predicted-PM
Predicted-PM-
Adjusted

Figure 7-13 Comparison of the measured and predicted deflections at shaft top
The two lateral load tests, i.e., load tests at I-40 and I-85, reported by Gabr et al
(2002) are also employed to evaluate Briaud et al. (1983) method. The test shaft and rock
information have been described in Chapter VI. Dilatometer tests were performed at the
two test sites. The derived p-y curves from dilatometer test results are provided in Figs.
7-14 to 7-17 for the four test shafts. The performance of these derived p-y curves are
evaluated by inputting them into LPILE and comparing with the measured load-
deflection curves, as shown in Figs. 7-18 to 7-21. It can be seen that the predicted
deflections are smaller than the measured values. In order to match the measured
deflections, a factor of 0.5 is multiplied to the calculated p values from dilatometer test
results. The adjusted p-y curves are then input into LPILE to predict the deflections, as
shown in Fig. 7-13 and Figs. 7-18 to 21. It can be seen that such adjustment on Briaud et
al. (1983) method can make it works much better for most of the cases. Therefore, it is
248
concluded that the modified Briaud et al (1983) method (i.e., reducing the calculated
value of p by 50%) can provide reasonable site-specific p-y curves.
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
0 0.2 0.4 0.6 0.8 1
y (in)
p

(
l
b
/
i
n
)
10.7 ft
7.4 ft
4.13 ft

Figure 7-14 Derived p-y curves from dilatometers for I-40 short shaft
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
0 0.5 1 1.5
y (in)
p

(
l
b
/
i
n
)
13.2 ft
6.8 ft

Figure 7-15 Derived p-y curves from dilatometers for I-40 long shaft
249
0
10000
20000
30000
40000
50000
60000
70000
0 0.5 1 1.5
y (in)
p

(
l
b
/
i
n
)
9.2 ft
3 ft

Figure 7-16 Derived p-y curves from dilatometers for I-85 short shaft
0
10000
20000
30000
40000
50000
60000
70000
80000
0 0.2 0.4 0.6 0.8 1
y (in)
p

(
l
b
/
i
n
)
11.5 ft
8.2 ft
4.9 ft
1.6 ft

Figure 7-17 Derived p-y curves from dilatometers for I-85 long shaft
250
0
50000
100000
150000
200000
250000
300000
350000
400000
0 0.2 0.4 0.6 0.8
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted-PM
Predicted-PM-
Adjusted

Figure 7-18 Comparison of load-deflection curves of I-40 short shaft
0
50000
100000
150000
200000
250000
300000
350000
400000
0 0.2 0.4 0.6 0.8
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted-PM
Predicted-PM-Adjusted

Figure 7-19 Comparison of load-deflection curves of I-40 long shaft

251
0
50000
100000
150000
200000
250000
300000
350000
0 0.5 1 1.5 2
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted-PM
Predicted-PM-Adjusted

Figure 7-20 Comparison of load-deflection curves of I-85 short shaft
0
50000
100000
150000
200000
250000
300000
350000
0 0.2 0.4 0.6 0.8
Deflection (in)
L
o
a
d

(
l
b
)
Measured
Predicted-PM
Predicted-PM-Adjusted

Figure 7-21 Comparison of load-deflection curves of I-85 long shaft

252



CHAPTER VIII
SUMMARIES AND CONCLUSIONS

8.1 Summaries
Towards the objective of developing a new p-y criterion of rock mass, a series of
field work and theoretical work have been carried out. A detailed literature review was
performed to study existing design and analysis methods of laterally loaded drilled shafts
and piles in rock. A new hyperbolic p-y criterion of rock was proposed based on the field
test data and extensive theoretical work.
A 3D finite element model simulating the response of laterally loaded drilled shafts
in rock using ABAQUS was established to develop an empirical correlation equation for
estimating the initial slope of a p-y curve of rock. Additionally, theoretical equations for
determining the ultimate resistance of rock mass were derived based on failure modes of
rock mass, a rock strength criterion, and an existing empirical equation for estimating the
side shear resistance between rock and drilled shafts. The failure modes of rock mass
were identified through a series of 3D FEM study.
In addition to the development of a p-y criterion of rock, a method for predicting
lateral capacity of drilled shafts in rock and/or soils was developed. This estimate can be
used to ensure adequate safety margin of a design of drilled shafts in rock. An elastic
solution for predicting the lateral deflections of piles/drilled shafts embedded in a two-
253
layer soil/rock system was proposed. This solution allows a quick estimation of drilled
shaft deflections under lateral loads. Furthermore, an evaluation of various existing
methods for deriving p-y curves from the results of an instrumented lateral load test was
carried out and the most suitable p-y curve derivation method was identified.
Finally, the existing methods for deriving site specific p-y curves of soils from
pressuremeter and dilatometer tests were reviewed and examined with the lateral load test
results of rock-socketed drilled shafts. Briaud et al. (1983) method, with a modification of
reducing p value by 50%, was recommended for deriving p-y curves of rock from the
pressuremeter and dilatometer tests.

8.2 Conclusions
Based on the research work performed in this study, the following conclusions can
be drawn:
1. The evaluation on the interim p-y criterion (Reese, 1997) for weak rock using the two
lateral load tests reveals that this p-y criterion tends to under-predict the deflections
which may result in an unsafe design.
2. The suggested method for deriving experimental p-y curves from load test results, i.e.
using high order polynomial curve fitting technique to obtain y and using piecewise
polynomial curve fitting technique to obtain p, demonstrated its superiority against
other methods by providing the smallest error on the predictions of the load-
deflection curves.
3. The validation of the 3D FEM model using a field test suggests that it can be used to
simulate the lateral response of drilled shafts in rock. This simulation has provided
254
basic understanding of the mobilization mechanisms of lateral resistance of rock mass
to the drilled shafts, from which analytical equations are derived for computing
ultimate resistance of rock p
u
.
4. The evaluation of the proposed method for estimating lateral capacity of drilled shafts
in rock using field test results showed that this method can provide reasonable
predictions of the lateral capacity of drilled shafts in rock. The average prediction
error was around 21% which is acceptable for practice, considering the measured
capacity may be under estimated due to a small deflection at the maximum applied
lateral load. Additionally, the proposed capacity prediction method is versatile. It can
be used for estimating the lateral capacity of drilled shafts in layered clay, sand, silts,
and rock. Both short and long drilled shafts under free or fixed head condition can be
considered.
5. The comparisons with rigorous solutions and the validation against field lateral load
test data have shown that the proposed elastic numerical solution can provide
reasonable predictions of shaft deflection under small working loads. This solution
provides an efficient way to estimate the deflections of a drilled shaft in a layered
soil/rock profile under small working lateral loads. Additionally, the proposed
empirical equation for estimating the deformation modulus of rock mass and the
empirical equation for determining the modulus of subgrade reaction of rock can
provide reasonable prediction results according to the validations against field test
data.
6. The hyperbolic p-y criterion for rock developed in this study can be used in
conjunction with computer analysis programs, such as COM624P, LPLE, or FBPIER,
255
to predict the deflection, moment, and shear responses of a drilled shaft under the
applied lateral loads. Considerations of the effects of joints and discontinuities on the
rock mass modulus and strength are included in the p-y criterion. Evaluations based
on comparisons between the predicted and measured responses of full-scale lateral
load tests on fully instrumented drilled shafts have shown the practical uses of the
proposed p-y criterion and the associated methods for determining the required input
of rock parameters.
7. The Briaud et al. (1983) method with a modification of reducing the calculated p
values by 50% for deriving the p-y curves from dilatometer tests is found to provide
reasonable predictions on the lateral deflections of a drilled shaft in rock according to
the evaluations against full-scale lateral field tests. The modified method is suggested
for design of laterally loaded drilled shafts in rock using dilatometer test results.
8. The research findings of this study provide a complete solution for design of drilled
shafts socketed in rock under lateral loads. The proposed elastic solution can be used
for a preliminary design of shafts under service loads. The computer program
LCPILE can be used for limit state design to guarantee adequate safety against design
loads. The proposed hyperbolic p-y curves can be used in conjunction of COM624P
or LPILE for the final design. The modified Briaud et al. (1983) method can be used
to validate the final design if dilatometer tests data are available.

8.3 Future Studies
More lateral load tests on drilled shafts socketed in various types of rock with the
accompanying dilatometer tests at the load test sites should be performed to further
256
validate the developed p-y criterion, the capacity prediction method, and the validity of
the modified Briaud et al. (1983) method for rock.
A preliminary design chart for selecting drilled shaft diameter and rock socketed
length can be developed based on LCPILE and the elastic solutions to ensure both
capacity and service limits criteria are met.
A future study on determining p-multipliers for a drilled shaft group in rock mass
can be very helpful. This objective can be accomplished through a well-planned field test
or a centrifuge test on drilled shafts groups socketed in rock mass. Additionally, a 3D
finite element study may also be employed to determine p-multipliers for a drilled shaft
group socketed in rock mass. Furthermore, the determination of group effect of lateral
capacity of group drilled shafts in rock is of practical interest since drilled shafts are often
constructed in a group.
A study on drilled shaft in rock under cyclic lateral loads will help seismic design of
deep foundations socketed in rock mass. A field test or centrifuge test using displacement
controlled loading will be very helpful for developing p-y curves of rock under cyclic
loading.
Although the proposed hyperbolic p-y curves is suitable for weak and hard rock by
considering the effects of joints and discontinuities of rock mass, it is not ready to be
applied to intermediate geomaterials. Therefore, it is necessary to study the lateral
response of drilled shafts socketed intermediate geomaterials, as these materials conform
a significant portion of the supporting medium for drilled shafts.

257



REFERENCES
ABAQUS (1998). ABAQUS Standard User’s Manual, Version 5.8, Hibbitt, Karlsson &
Sorensen, Inc.

Afrouz, A.A. (1992). Practical handbook of rock mass classification systems and modes
of ground failure, Florida, CRC Press, Inc.

American Petroleum Institute (API). (1991). “Recommended practice for planning,
designing and constructing fixed offshore platforms.” API recommended practice 2A
(RP2A), 19
th
Ed., Washington, D. C.

ASTM D4719-00, (2000). “Standard test method for prebored pressuremeter testing in
soils.” ASTM International.

Baguelin, F., Frank, R., and Said, Y. H. (1977). “Theoretical study of lateral reaction
mechanism of piles.” Geotechnique, 27, No. 3, p. 405-434.

Baguelin, F., Jezequel, F. F., and Shields, D. H. (1978). The Pressuremeter and
Foundation Engineering, Trans Tech Publications, Clausthal, Germany.

Barton, N., Lien, R., and Lunde, J. (1974). “Engineering classification of rock masses for
the design of tunnel support.” Rock Mechanics, Vol. 6, No. 4, Vienna, Austria, p. 189-
236.

Bieniawski, Z. T. (1976). “Rock mass classifications in rock engineering.” Proceedings
of the Symposium on Exploration for Rock Engineering, Z. T. Bieniawski, ed., Vol. 1, A.
A. Balkema, Rotterdam, Holland, p. 97-106.

Bieniawski, Z.T., (1978). “Determining rock mass deformability: experience from case
histories.” Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 15, p. 237-248.

258
Bieniawski, Z. T. (1989). Engineering Rock Mass Classification, p. 251. Wiley, New
York.

Bishop, R. F., Hill, R., and Mott, N. F., (1945). “The theory of indentation and hardness
tests.” The Proceedings of the Physical Society, London, Vol. 57, No. 321, Part 3, May, p.
147-159.

Boghrat, A., 1990, “Sound Barrier Wall Foundations in Granular Material,”
Transportation Research Record, Issue: 1288, p. 158-167

Bowles, J. E. (1988). Foundation analysis and design, 4
th
Edition, McGraw Hill, New
York, 1004.

Bowman, E. R. (1958). “Investigation of the lateral resistance to movement of a plate in
cohesionless soil.” Unpublished thesis, College of Engineering, The University of Texas
at Austin.

Briaud, (1989). “The pressuremeter test for highway applications.” FHWA Publication
No. FHWA-IP-89-008. Department of Transportation, Federal Highway Administration,
McLean, VA 22101-2296, USA.

Briaud, J.L., (1992) The Pressuremeter, published by A.A.Balkema, Rotterdam,
Netherlands.

Briaud, J. L., Smith, T. D., and Meyer, B. J. (1983), “Using the Pressuremeter Curve to
Design Laterally Loaded Piles.” Proceedings of Annual Offshore Technology Conference,
Vol. 8, p. 495-502.

Brinch Hansen, J. (1961). “The ultimate resistance of rigid piles against transversal
forces.” Geoteknisk Institut. Bull., No.12, Copenhagen.

Brown, E. T. (1981). Rock Characterization, Testing and Monitoring: ISRM Suggested
Methods, Pergamon Press, Oxford.

Brown, E.T. (1993). “The nature and fundamentals of rock engineering.” Comprehensive
Rock Engineering – Principle, Practice and Projects. Ed; J.A. Hudson, Pergamon Press,
p.1-23.
259
Bukovansky, M. (1970). “Determination of elastic properties of rocks using various on-
site and laboratory methods.” International Soc. Rock Mechanics, Proc., Vol. 1, No 2-9, p.
329-332.

Carter, D. P. (1984). “A non-linear soil model for predicting lateral pile response.” Rep.
No. 359, Civil Engineering Dept., Univ. of Auckland, New Zealand.

Carter, J. P., Booker, J. R., and Yeung, S. K. (1986). “Cavity expansion in cohesive
frictional soils.” Geotechnique, 36(3), p. 349-358.

Carter, J. P., and Kulhawy, F. H. (1992). “Analysis of laterally loaded shafts in rock.”
Journal of Geotechnical Engineering, ASCE, 118(6), p. 839-855.

Cho, K. H., Clark, S. C., Keaney, B. D., Gabr, M. A., and Borden, R. H. (2001).
“Laterally loaded drilled shafts embedded in soft rock.” Transportation Research Record
1772, p. 3-11.

DiGioia, A. M., Jr., and Rojas-Gonzalez, L. F. (1993). “Discussion on ‘Analysis of
laterally loaded shafts in rock.’” Journal of Geotechnical Engineering, ASCE, 119(12), p.
2014-2015.

Dou, H. and Byrne, P. M., (1996). “Dynamic Response of Single Piles and Soil-pile
Interaction,” Canadian Geotechnical Journal, Vol. 33, p. 80-96.

Dunnavant, T. W., (1986). Experimental and Analytical Investigation of the Behavior of
Single Piles in Overconsolidated Clay Subjected to Cyclic Lateral Loads, Ph.D.
Dissertation of University of Houston, Texas.

Dykeman, P., and Valsangkar, A. J. (1996). “Model studies of socketed caissons in soft
rock.” Canadian Geotechnical Journal, v. 33 issue 5, p. 747-759.

Feldmann, R.M., and Hackathorn, M. (1996). Fossils of Ohio, Bulletin 70, Ohio Division
of Geological Survery.

Fisher, S. P., Fanaff, A. S., and Picking, L. W. (1968). “Landslide of southeastern Ohio.”
Ohio Journal of Science, V. 68, p. 65-80.
260
Fleming, W. G. K., Weltman, A.J., Randolph, M. F., and Elson, W. K. (1992). Piling
Engineering. Blackie, Glasgow, and London.

Frantzen, J., and Stratten, W. F. (1987). “Final report: p-y curve data for laterally loaded
piles in shale and sandstone.” Rep. No. FHWA-KS-87-2, Kansas Department of
Transportation, Topeka, Kansas.

Gabr, M. A. (1993). “Discussion on ‘Analysis of laterally loaded shafts in rock.’” Journal
of Geotechnical Engineering, ASCE, 119(12), p. 2015-2018.

Gabr, M.A., Borden, R.H., Cho, K.H., Clark, S.C., and Nixon, J.B. (2002). “P-y curves
for laterally loaded drilled shafts embedded in weathered rock.” FHWA/NC/2002-08,
North Carolina Department of Transportation.

Georgiadis, M., and Michalopoulos, A. P., (1986). “Dilatometer tests for the design of
grouted piles in rock.” Use of In Situ Tests in Geotechnical Engineering, Proceedings of
In Situ '86, Blacksburg, VA, USA, p. 560-568.

Gibson, R. E., and Anderson, W. F. (1961). “In situ measurement of soil properties with
the pressuremeter.” Civil Engineering and Public Works Review, London, 56 No. 658.

Gokceoglu, C., Sonmez, H., and Kayabasi, A. (2003). ‘Predicting the deformation moduli
of rock masses’ International Journal of Rock Mechanics and Mining Sciences, V. 40,
p.701-710.

Goodman, R., and Shi, G. (1985). Block theory and its application to rock engineering,
Prentice-Hall, Englewood Cliffs, N.J.

Guo, W. D. (2001). “Subgrade modulus for laterally loaded piles.” Proceedings of the
Eighth International Conference on Civil and Structural Engineering Computing, Vienna,
Austria.

Haberfield, C. M., and Johnston, I. W. (1993). “Factors influencing the interpretation of
pressuremeter tests in soft rock.” Proceedings of the international conference on
geotechnical engineering of hard soils-soft rocks, Athens, p. 525-531.

261
Hall, C. G., and M.C. Wang. (2000). Behavior of laterally loaded caissons in weak rock.
ASCE Geotechnical Special Publication, Issue 100, New Technological and Design
Developments in Deep Foundations, p. 240-253.

Hawk, D. J., and Ho, H. Y., (1980) “A study of the orthotropy of coal and other rock
materials,” Final Report to U.S. Dept. of Interior and Bureau of Mines, University of
Colorado.

Hetenyi, M. (1946). Beams on elastic foundations. Ann Arbor: University of Michigan
Press.

Heuze, F.E., (1984) “Suggested method for estimating the in-situ modulus of deformation
of rock using the NX-borehole jack.” Geotechnical Testing Journal, Vol.7, No. 4, p.205-
210.

Heuze, F.E. and Amadei, B. (1985) “The NX-borehole jack: A lesson in trial and error”
International Journal of Rock Mechanics and Mining Science, Vol. 22, No.2, p.105-112.

Heuze, F.E. and Salem A. (1976) “Plate bearing and borehole jack tests in rock- a finite
element analysis.” Proc. 17
th
U.S. Symp. On Rock Mechanics, Snowbird, Utah, p. 4B8-1
– 4B8-6.

Hoek, E. (1983). “Strenth of jointed rock massses” Geotechnique, v. 33 issue 3, p. 187-
223.

Hoek, E., (1990). “Estimating Mohr-Coulomb friction and cohesion values from the
Hoek-Brown failure criterion.” Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., Vol. 27,
No. 3, p. 227-229.

Hoek, E. (1994). “Strength of rock and rock masses,” ISRM News Journal, 2 (2), p. 4-16.

Hoek, E., and Brown, E. T. (1980). “Empirical strength criterion fro rock masses.”
American Society of Civil Engineers, Journal of the Geotechnical Engineering Division,
v. 106 issue 9, p. 1013-1035.

262
Hoek, E., and Brown, E.T. (1988). “The Hoek-Brown criterion – a 1988 update.” 15
th

Canadian Rock Mechanics Symposium: rock engineering for underground excavations,
University of Toronto, p. 31-38.

Hoek, E., and Brown , E.T. (1997). “Practical estimates of rock mass strength.” Int.J.
Rock Mech. Min. Sci., 34(8), p. 1165-1186.

Hoek, E., Carranza-Torres, C., and Corkum, B. (2002) “Hoek-Brown failure criterion –
2002 edition.” Proc. North American Rock Mechanics Society Meeting in Toronto.

Hughes, J. M. O., Wroth, G. P., and Windle, D. (1977). “Pressuremeter tests in sand.”
Geotechnique, Vol. 27, No. 4.

Hull, D.N. (1990) and Larsen, G.E. (2000) www.dnr.state.oh.us/geosurvey/pdf/
stratcol.pdf, accessed on Aug. 2004.

Hustrulid, W. A. (1976) “An analysis of the Goodman jack.” Proc. 17
th
U.S. Symp. On
Rock Mechanics, Snowbird, Utah, p. 4B10-1 – 4B10-8.

Johnston, I. W. (1985). “Strength of intact geomechanical materials.” J.Geotech. Engrg.,
ASCE, vol. 111, p. 730-749.

Johnston, J. W. and Lam, T. S. K., (1989a), “Shear Behavior of Regular Triangular
Concrete/Rock Joints÷Analysis,” Journal of Geotechnical Engineering, Vol. 115, No. 5,
p. 711-727.

Johnston, J. W. and Lam, T. S. K., (1989b), “Shear Behavior of Regular Triangular
Concrete/Rock Joints÷Evaluation,” Journal of Geotechnical Engineering, Vol. 115, No.
5, p. 728-740.

Kayabasi, A., Gokceoglu, C., and Ercanoglu, M. (2003). “Estimating the deformation
modulus of rock masses: a comparative study.” International Journal of Rock Mechanics
& Mining Science, Vol. 40, p. 55-63.

Kulhawy, F. H. et al. (1983). “Transmission line structure foundations for uplift-
compression loading.” Rep. No. EL-2870, Electric Power Research Institute, Palo Alto,
Calif.
263
Kulhawy, F. H. (1991). “Drilled shaft foundations.” Foundation Engineering Handbook,
2
nd
Ed., Chap. 14, H. –Y. Fang ed., Van Nostrand Reinhold, New York.

Kulhawy, F. H., and Y. -J. Chen. (1995). “A thirty year perspective of Broms’ lateral
loading models, as applied to drilled shafts.” Bengt B. Broms Symposium on
Geotechnical Engineering, Singapore, p. 225-240.

Kulhaway, F. H. and Phoon, K. K. (1993), “Drilled Shaft Side Resistance in Clay Soil to
Rock,” Proc. on Conf. On Design and Performance of Deep Foundation: Piles and Piers
in Soil and Soft Rock. ASEC, Geotechnical Special Publication 38, p. 172-183.

Lamborn, R.E., Austin, C. R., and Schaaf, D. (1938). Shales and surface clays of Ohio,
Geological Survey of Ohio, fourth series, bulletin 39.

Liang, R. Y. (2002). “Drilled shaft foundations for noise barrier walls and slope
stabilization.” Final Rep. FHWA/OH-2002/038, Ohio Department of Transportation.

Littlechild, Brian D., et al.(2000) “Determination of rock mass modulus for foundation
design.” Geotechnical Special Publication, issue 97, p. 213-228.

Marinos, P., and Hoek, E. (2000). “GSI: A geologically friendly tool for rock mass
strength estimation.” Proc. of the International Conference on Geotechnical and
Geological Engineering (GeoEng 2000), Melbourne, Australia, p. 1422-1440.

Matlock, H. (1970). “Correlations for design of laterally loaded in soft clay.” Proc. 2
nd

Offshore Technology Conference, Offshore Technology Conference, Houston, Vol.1, p.
577-594.

Matlock, H. and Ripperger, E. A., (1956). “Procedures and Instrumentation for Tests on a
Laterally Loaded Pile,” Proceedings, Eighth Texas Conference on Soil Mechanics and
Foundation Engineering, Austin, Tex.

Meigh, A. C. (1979), “Design Parameters for Weak Rocks,” Proceedings, 7
th
Europe
Conference on Soil Mechanics and Foundation, p. 59-79.

264
Menard, L., Bourdon, G., and Gambin, M. (1969). “Methode Generale de Calcul d’un
Rideau ou Pieu Sollicite Horizontalement en Fonction des Resultats Pressiometriques,”
Sols-Soils, No. 22/23.

Meyer, T. O. and McVey, J. R. (1974) “NX borehole jack modulus determinations in
homogeneous, isotropic, elastic materials.” USBM RI 7855.

Mezazigh, S. and Levacher, D., (1998). “Laterally Loaded Piles in Sand: Slope Effect on
p-y Reaction Curves.” Canadian Geotechnical Journal, Vol. 35, No. 3, p. 433-441.

Mokwa, R. L., Duncan, J. M., and Helmers, M. J. (2000). “Development of p-y curves
for partly saturated silts and clays.” Proceedings of Geo-Denver 2000: New technological
and design developments in deep foundations, ASCE, p. 224-237.

Murchison, J. M. and O’Neill, M. W., 1984. “Evaluation of P-y Relationships in
Cohesionless Soils,” Analysis and Design of Pile Foundations, ASCE, Ed. J. R. Meyer,
p.174-191.

Nicholson, G.A., Bieniawski, Z.T.(1990). “A nonlinear deformation modulus based on
rock mass classification.” International Journal of Min Geol Eng. Vol. 8, p. 181-202.

Nusairat, J., Engel, R. L., and Liang, R. Y. (2002). “Instrumentation, monitoring and
testing at the ERI-60-3100 bridge project.” Rep. Prepared for Ohio Department of
Transportation, Columbus, OH.

Nusairat, J., Liang, R., Engel, R., Hanneman, D., Abu-Hejleh, N., and Yang, K. (2004).
Drilled Shaft Design for Sound Barrier Walls, Signs, and Signals. Final Report, CDOT-
DTD-R-2004-8, Colorado Department of Transportation.

O’Neill, M. W., and Reese, L. C. (1999). Drilled Shafts: Construction Procedure and
Design Methods, FHWA-IF-99-025.

Pells, P. J. N. et al. (1980), “An Experimental Investigation into the Side Shear of
Socketed in Sandstone,” Proceedings, International Conference on Structural
Foundations in Rock, Sydney, p. 291-302.

265
Pise, P. J. (1982). “Laterally loaded piles in a two-layer soil system.” Journal of the
Geotechnical Engineering Division, ASCE, Vol. 108, No. GT9, p. 1177-1181.

Poulos, H. G. (1971). “Behavior of laterally loaded piles. I: Single piles.” J. Soil Mech.
Found. Div., ASCE, 97(5), p. 711-731.

Ramamurthy, T., Rao, G. V., and Rao, K.S. (1985). “A strength criterion for rocks.” Proc.
Indian Geotech. Conf., Roorkee, I, p. 59-64.

Randolph, M. F. (1981). “The response of flexible piles to lateral loading.” Geotechnique,
31(2), p. 247-259.

Reese, L. C. (1997). “Analysis of laterally loaded piles in weak rock.” J. Geotech. Envir.
Engrg., ASCE, 123(11), p. 1010-1017.

Reese, L. C., Cox, W. R. and Koop, F. D. (1974). “Analysis of laterally loaded piles in
sand.” Proc. Of 6th offshore Tech. Conf., Houston, TX, Vol. 2, p. 473-483.

Reese, L. C., Cox, W. R., and Koop, F. D. (1975). “Field testing and analysis of laterally
loaded piles in stiff clay.” 7
th
Offshore Technology Conference, p. 671-690.

Reese, L. C., and Matlock, H. (1956). “Non-dimensional solutions for laterally loaded
piles with soil modulus assumed proportional to depth.” Proceedings, Eight Texas
Conference on Soil Mechanics and Foundation Engineering, Austin, Texas, p. 1-23.

Reese, L. C., and Welch, R. C. (1975). “Lateral loadings of deep foundations in stiff
clay.” Journal of Geotechnical Engineering, ASCE, 101(7), p. 633-649.

Richardson, D. N., and Wiles, T. T., (1990) “Shale durability rating system based on loss
of shear strength,” Journal of Geotechnical Engineering, Vol. 116, No. 12, p. 1864-1880.

Robertson, P. K., and Hughes, J. M. O. (1986). “Determination of properties of sand from
selfboring pressuremeter tests.” The pressuremeter and its Marine Applications: Second
International Symp., ASTM STP 950, Texas A&M University.

266
Robertson, P. K., Hughes, J. M. O., Campanella, R. G., and Sy, A. (1983). “Design of
laterally loaded displacement piles using a driven pressuremeter.” ASTM SPT Symposium:
Design & Performance of Laterally Loaded Piles and Pile Groups.

Robertson, P. K., Hughes, J. M. O., Campanella, R. G., Brown, P. and McKeown, S.
(1986). “Design of laterally loaded piles using the pressuremeter”. The pressuremeter
and its Marine Applications: Second International Symp., ASTM STP 950, Texas A&M
University.

Rocha, M., Silveira, A., Rodrigues, F. P., Silverio, A., and Ferreira, A. (1970).
“Characterization of the deformability of rock masses by dilatometer tests.” International
Soc Rock Mechanics, Proc., Vol. 1, No. 2-32, p. 509-516.

Rocscience (2002), RocLab, a free software from www.rocscience.com, Rocscience, Inc.,
Toronto, Canada.

Rosenberg, P. and Journeaux, N. L. (1976), “Friction and End Bearing Tests on Bedrock
for High Capacity Socket Design,” Canadian Geotechnical Journal, Vol. 13, p. 324-333.

Sargand, S. M, and Hazen, G. A. (1987). “Deformation behavior of shales” Int. J. Rock
Mech. Min. Sci. & Geomech. Abstr., Vol. 24, No. 6, p. 365-370.

Seidel, J. P. and Collingwood, B. (2001), “A new Socket Roughness Factor for Prediction
of Rock Socket Shaft Resistance,” Can. Geotech. J., Vol. 38, p. 138-153.

Serafim J.L. and Pereira J.P. (1983). “Consideration of the geomechanical classification
of Bieniawski.” Proceeding of International Symposium on Engineering Geology and
Underground Construction, Lisbon 1(II), p. 33-44.

Shen, W. Y., and Teh, C. I. (2002). “Analysis of laterally loaded pile groups using a
variational approach.” Geotechnique, 52(3), p. 201-208.

Shen, W. Y., and Teh, C. I. (2004). “Analysis of laterally loaded piles in soil with
stiffness increasing with depth.” Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 130, No. 8, p. 878- 881.

267
Shuri, F.S. (1981) “Borehole diameter as a factor in borehole jack results.” Proc. 22
nd
U.S.
Symp. On Rock Mechanics, Cambridge, Massachusetts, pp. 392-397, MIT Press,
Cambridge.

Smith, T. D. (1983). “Pressuremeter Design Method for Signal Piles Subjected to static
Lateral Load.” Ph. D. Thesis Texas A&M University.

Stout, Wilber (1941). Dolomites and limestones of western Ohio, Geological Survey of
Ohio, fourth series, bulletin 42.

Sun, K. (1994). “Laterally loaded piles in elastic media.” Journal of Geotechnical
Engineering, ASCE, 120(8), p. 1324-1344.

Terzaghi, K. (1955). “Evaluation of coefficient of subgrade reaction.” Geotechnique, 5(4),
p. 297-326.

To, A. C., Ernst, H., and Einstein, H. H. (2003). “Lateral load capacity of drilled shafts in
jointed rock.” Journal of Geotechnical and Geoenvironmental Engineering, Vol. 129, No.
8, p. 711-726.

Toh, C. T. et. al. (1989), “Design Parameters for Bored piles in a Weathered Sedimentary
Formation,” Procs. of the 12
th
International Conference on Soil Mechanics and
Foundation Engineering, Vol. 2, p. 1073-1078.

Trochianis, A., Bielak, J., and Christiano, P. (1988). “A three-dimensional nonlinear
study of piles leading to the development of a simplified model.” Report R-88-176, Dept.
of Civil Engr., Carnegie Inst. of Technology.

Vesic, A. S. (1961). “Beam on elastic subgrade and the Winkler hypothesis.” Proc. 5
th
Int.
Conf. Soil Mechanics and Foundation Engineering, Paris, Vol. 1, p. 845-850.

Wakai, A., Gose, S., and Ugai, K. (1999). “3-D elasto-plastic finite element analyses of
pile foundations subjected to lateral loading.” Soils and Foundations, Vol. 39, No. 1, p.
97-111.

Wallace, J. W. et al. (2002). “Cyclic large deflection testing of shaft bridges part II:
analytical studies.” Report from California Dept. of Transportation.
268
Wang, S. T., and L. C. Reese. (1993). COM624P – Laterally Loaded Pile Analysis
Program for the Microcomputer. Version 2.0, Federal Highway Administration
Publication No. FHWA-SA-91-048.

Williams, A. F., Johnston, I. W., and Donald, I. B. (1980). “The Design of Socketed piles
in weak rock,” Proceedings of International Conference on Structural Foundations on
Rock, Sydney, p. 327-347.

Williams, A. F., and Pells P. J. N. (1981). “Side Resistance Rock Sockets in Sandstone,
Mudstone, and Shale,” Canadian Geotechnical Journal, Vol. 18, p. 502-523.

Wilson, D. 1998. Soil-pile-superstructure interaction in liquefying sand and soft clay. Ph.
D. Dissertation, University of California at Davis.

Wood, L. E., and Deo, P. (1975). “A suggested system for classifying shale materials for
embankments,” Bulletin of the Association of Engineering Geologists, Vol. XII, No. 1, p.
39-55.

Yang, K., and Liang, R. (2005). “Lateral Response of Large Diameter Drilled Shafts in
Clay.” Proceedings of DFI 30
th
Annual Conference on Deep Foundations, Chicago, IL.

Yang, K., Liang, R., and Liu, S. (2005). “Analysis and test of rock-socketed drilled shafts
under lateral loads.” The 40
th
U.S. Symposium on Rock Mechanics, Anchorage, Alaska.

Zhang, L. (1997). Analysis and design of axially loaded drilled shafts socketed into rock.
MS thesis, Massachusetts Institute of Technology, Cambridge, MA.

Zhang, L., Ernst, H., and Einstein, H. H. (2000). “Nonlinear analysis of laterally loaded
rock-socketed shafts”, Journal of Geotechnical and Geoenvironmental Engineering.
v. 126 issue 11, p. 955-968.

Zhang, L., Silva, F., and Grismala, R. (2005). “Ultimate lateral resistance to piles in
cohesionless soils.” Journal of Geotechnical and Geoenvironmental Engineering. v.131
No. 1, p. 78-83.

ANALYSIS OF LATERALLY LOADED DRILLED SHAFTS IN ROCK

A Dissertation Presented to The Graduate Faculty of The University of Akron

In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

Ke Yang May, 2006

ANALYSIS OF LATERALLY LOADED DRILLED SHAFTS IN ROCK

Ke Yang

Dissertation

Approved: Advisor Dr. Robert Liang Committee Member Dr. Wieslaw Binienda Committee Member Dr. Ernian Pan Committee Member Dr. Yueh-Jaw Lin Committee Member Dr. Chien-Chung Chan

Accepted: Department Chair Dr. Wieslaw Binienda Dean of the College Dr. George Haritos Dean of the Graduate School Dr. George Newkome Date

ii

ABSTRACT

Drilled shafts socketed into rock are widely used as foundations for bridges and other important structures. Rock-socketed drilled shafts are also used to stabilize a landslide. The main loads applied on the drilled shafts are axial compressive or uplift loads as well as lateral loads with accompanying moments. Although there exist several analysis and design methods especially for rock-socketed drilled shafts under lateral loading, these methods were developed with assumptions without actual validations with field load test results. Some of the methods have been found to provide unsafe designs when compared to recently available field test data. Therefore, there is a need to develop a more rational design approach for laterally loaded drilled shafts socketed in rock. A hyperbolic non-linear p-y criterion for rock is developed in this study that can be used in conjunction with existing computer programs, such as COM624P, LPILE, and FBPIER, to predict the deflection, moment, and shear responses of a shaft under the applied lateral loads. Considerations for the effects of joints and discontinuities on the rock mass modulus and strength are included in the p-y criterion. Evaluations based on comparisons between the predicted and measured responses of full-scale lateral load tests on fully instrumented drilled shafts have shown the applicability of the proposed p-y

iii

criterion and the associated methods for determining the required input of rock parameters. In addition to the development of a hyperbolic p-y criterion for rock, a method for predicting lateral capacities of drilled shafts in rock and/or soils is developed for assessing the safety margin of the designed shafts against the design loads. A computer program LCPILE is developed using VC++ to facilitate computations. An elastic solution based on a variational approach is also developed for determining drilled shaft elastic deflection due to applied lateral loads in a two-layer soil layer system. The computational algorithm was coded in a Mathematica file for easy application. Finally, Briaud’s method for deriving p-y curves of rock from pressuremeter or dilatometer test results is evaluated using available field test data. A modification to the Briaud’s method is recommended for applications in rocks.

iv

who made all of this possible. for their endless encouragement and patience.DEDICATION To my wife. v . my father and mother.

has been a constant source of inspiration. What I learned from him is not only the knowledge. I am also grateful to my wonderful friends in Akron. and gratitude to a number of people whose support has been significant in completing this dissertation. Ernian Pan. and his amazing energy. Dr. Jamal Nusairat is highly appreciated. Yueh-Jaw Lin. the great help from Dr. I am left with profound feelings of accomplishment. ChienChung Chan. excitement. motivation and guidance throughout the past three and a half years. Dr. vi . Ohio. I have learned so much from his keen insight.ACKNOWLEDGEMENTS As I approach the end of my graduate tenure. My advisor Professor Robert Y. Their aid has been invaluable during the period of this work. My wife. and have always been encouraging and understanding. Liang. but also the way to be a positive and energetic person. parents. Additionally. Wieslaw Binienda. Thanks in large part to the kindness and mentoring provided by the committee members: Dr. and brother have been extremely supportive of my studies. his research and problem solving abilities. and Dr.

......1 Introduction ............................. 26 2.......................................... 8 2......................................2 The Shales in Ohio ............. 1 1...........................................1 Statement of the Problem ...................3 Comments on the Existing Analysis Methods .............................................................2 Bedrocks of Ohio ...........................................4 Outlines of the Dissertation......................TABLE OF CONTENTS Page LIST OF TABLES ……………………………………………………………………............. 8 2..............................................1 The Distribution of Rocks in Ohio...................1........................ 34 vii ... 23 2........... LITERATURE REVIEW ....................................2 Objectives................................................ 6 II......................... 1 1.................. INTRODUCTION................................... 30 2....................................................... 8 2...............................xvi CHAPTER I........................................... 26 2......................2........................................2 Brief Description of the Existing Methods .1 Analysis Methods of Laterally Loaded Rock-Socketed Drilled Shafts ............................... 10 2............1............................. 3 1...................3 Limestone and Dolomite ................................2...1........2..............xiii LIST OF FIGURES …………………………………………………………………….......................................... 4 1........3 Scope of the Work................................................................

............3............ 52 3.............3............3............................................. 70 3............2............... 35 2.............4 Methods for Deriving P (Net Resistance) ........3.................................................................1 The types of rocks .........3................2...................................... 81 3....3...................................................................... 62 3............................................................................ 64 3.........................................8 Effect of Measurement Error ................1.....................2 Method for Deriving Deflection from Strain Gage Readings.....................................................3..1... 53 3. 74 3............... 38 III................................ 49 3...........................3. 94 viii ...................3...2.......................................................... 70 3.................... 57 3..............3.........3 Lateral Load Test Results..2 Lateral Load Test at Pomeroy-Mason............ 90 3...... 57 3......... 72 3.................... 73 3.......3 Lateral Load Test Results....... 49 3....3 Determination of Moment Profiles ........3............................................ 88 3................................................2 Test Setup and Procedure. LATERAL LOAD TEST RESULTS AND DATA ANALYSES ..6 Evaluation Using Hypothetical Cases................................................................9 Conclusions on Methods for Deriving p-y Curves ....................1................2 Test Set-up and Test Procedure.........1 Introduction ..5 Evaluation Using Field Test Data .................1 Test Site......................................... 92 3.....1 Test Site.............3......................3 Rock Mass Classifications ....................... 35 2..........7 Optimum Strain Gage Spacing............................3 Mechanical Characteristics of Rock and Its Classifications ..............................................................2 Features for Rock Characterization......1 Lateral Load Test at Dayton. 35 2... 49 3..............2.............3 Methods for Deriving P-y Curves From Lateral Load Test Results ........................................

4.............2. 115 4..................... 130 4................................ 138 4................. 111 4.... 105 4............ 136 4.........1 Hoek-Brown Criterion ...........2..2 Failure Modes of Rock Subjected to Loading from Laterally Loaded Drilled Shafts .........................4 Analyses of the Load Tests ........................................1 Empirical Equations for Axially Loaded Drilled Shaft in Rock ...........................................4...2 Validation of the 3D FEM Model ............................................................................... 124 4..........1 Evaluation of Experimental P-y Curves.......................2 Converting Hoek-Brown Criterion to Mohr-Coulomb Criterion....................... 138 4.... 115 4.........................1 Ultimate Resistance of Rock Near Surface........... 140 ix .....................................................3................................................................................... 129 4........3 Failure of Isotropic and Jointed Rock with Overburden Soils....1 3-D FEM Modeling. 132 4.....2 Ultimate Rock Resistance at Great Depth.......................2 Suggested Empirical Equation for Side Resistance in Horizontal Direction....1 Failure of Isotropic and Homogeneous Rock without Overburden Soils................4........................................ 106 4......... 107 4........ 134 4.................................4...........1 3-D Finite Element Modeling and Validation.... 104 IV...............................................3.........5 Ultimate Resistance of Rock Mass .2 Evaluation of Reese Interim Rock p-y Criterion...... 129 4......................1..............................................5 Summary and Conclusions.........................3 Rock Strength Criteria................... 95 3................ LATERAL CAPACITY OF DRILLED SHAFTS IN ROCK .. 100 3.......................5....5...4 Side Shear Resistance ...............3................ 132 4............2 Failure of Jointed Rock without Overburden Soils.............. 95 3............................1.................2.............................

........................6 Proposed Empirical Equation........ 157 5......................................................................................... 170 5........4 Existing Empirical Equations..1.......1 Ultimate Resistance of Clay.. 158 5........................ 171 5...................... 142 4...........7...................................................................1............................ 142 4..................2 Fixed Head Boundary ... 148 4..........................................................................6 New Method for Predicting Lateral Capacity of Drilled Shafts in Rock.....................3 Selection of In-Situ Test Method ...................3 Ultimate Resistance of Jointed Rock ... 145 4....................... 141 4.............7 Ultimate Reaction of Soils .7.2 Initial Modulus of Subgrade Reaction of Rock Mass ..........................6.6......... 157 5.... 152 4............ 148 4................... 155 V....2 Goodman Jack Test ......... 163 5........................................ 150 4.......................2 Ultimate Resistance of Sand ..........................1 Free Head Boundary ............................... ELASTIC SOLUTION OF LATERALLY LOADED DRILLED SHAFTS IN ROCK....................................7 Recommended Methodology for Determination of Deformation Modulus of Rock ..........2..........................9 Summary and Conclusions....................................................4.............1 Determination of Rock Mass Deformability..................8 Validation of the Derived Capacity Prediction Method............................................................................................ 173 x ........7............1 Introduction ..... 169 5............. 171 5............................................... 157 5...............5 Parameters for Determination of Em .................. 151 4.................................5....1............................... 167 5....................2 FEM Model and Parametric Study.................1..1............2.........................................................1 Introduction ...............1........... 165 5........................3 Ultimate Resistance of c-φ Soils .............................1...............................

....................................................1 Introduction ...2... 195 5.................3..... 200 6.........................................1 Strength Parameters .......................................................................................... 202 6..................................................................3................ 186 5.........................3.................... 207 6..............................4 Validation of the Empirical Equation ....................................2.5 Methods for Determining Input Parameters.............................................2....2................... 187 5....................2.......................................................................... 191 5...................................................................2....................1 General Shape of P-y Curve in Rock .... 204 6........ 202 6..................................................3....................4............................................................................4 Pu of Jointed Rock....3 Suggested Empirical Equation .....................................................3 Rock Mass Modulus Em .....2 Definition of the Problem .......4 Validations .............3 Pu at Great Depth..........................................2 Pu Near Surface .. 184 5...........3.................... 209 xi ...2 Rock-Shaft Interface Strength....... 207 6...................3 Numerical Solution for Laterally Loaded Piles in A Two Layer Soil Profile............................... 184 5.......................... 206 6.............. 205 6............................................................6 Case Study.5.................. 179 5........4 Rock Mass Properties.......... 206 6................................................ P-Y CRITERION FOR ROCK MASS ...........3 Initial Tangent to P-y Curve Ki .................................... 209 6...............................................4 Summary and Conclusions........... 200 6.......................................................... 180 5.......................3 Variational Solution ...2 Determination of pu .........................1 Failure Modes ..............4....................................4....................................................................3................................. 198 VI................ 195 5...............

..3 Load Tests at North Carolina ..........................................1.......... 255 REFERENCES ................................ 232 7................................................... 210 6..1 Dayton Load Test............................................................................................................................. 257 xii ......................7......................................... 227 7.......2 Limit Pressure .7 Case Studies ...................................6 Comparison of the Proposed P-y Criterion with That of Gabr et al.........3 Future Studies.............. 244 VIII.................................. 234 7............................................................................ 213 6............. 238 7.......................................... 227 7..............7..2 Pomeroy-Mason Load Test.................................................. 232 7.......4 Friction Angle ................... 232 7.... 211 6................................................................................................. 252 8.............5 Construction of P-y curves for Rock Mass ....................1 Summaries.................................................................................... 253 8.......................................................................................7..... DERIVING P-Y CURVE FROM DILATOMETER TESTS........1............................................................... 252 8................................1 Pressuremeter and Dilatometer ....................................................... 226 VII.................3 Undrained Shear Strength .......3 Evaluation ................................. 221 6......................................... (2002).........................................................................................2 Deriving p-y Curves from Pressuremeter/Dilatometer Test Results............. 209 6.........................................................1........2 Conclusions .............. 217 6......1 Modulus of Rock Mass ...................1............................8 Conclusions .......................................6................................. SUMMARIES AND CONCLUSIONS........

LIST OF TABLES Table 2-1 2-2 2-3 2-4 2-5 2-6 2-7 2-8 2-9 2-10 3-1 3-2 3-3 3-4 3-5 Page Engineering Properties of Rocks in Ohio……………………………… 33 Simplified Rock Classification………………………………………… 36 Factors influencing Rock Mass Behavior……………………………… 37 Classification Parameters and Their Rating (After Bieniawski. 46 GSI Ranges for Typical Limestones (Marinos and Hoek. Claystones and Clay Shales (Marinos and Hoek. 87 xiii . 2000) ……………44 GSI Ranges for Typical Sandstones (Marinos and Hoek. 48 RMR Ratings and GSI Values of Rock at Dayton Test Site……………51 RMR Ratings and GSI Values of Rock Based on Boring S-9…………. 41 Rock Mass Rating System (Bieniawski. 2000)………... 1989) ……………………….. 2000)………. 2000) ………………………….………... 59 RMR Ratings and GSI Values of Rock Based on Boring S-10……….82 Cumulative Shaft Head Deflection Errors based on Various Methods……………………………………………. 1976)………………………………………………..... 62 Compiled Lateral Load Test Database…………………………………..47 GSI Estimates for Heterogeneous Rock Masses Such as Flysch (Marinos and Hoek.42 GSI Ranges for General Rocks (Marinos and Hoek. 45 GSI Ranges for Typical Siltstones.. 2000) ………………….

.. 153 Comparison of Lateral Capacities of Test Drilled Shafts……………… 154 Values of Constant K(ν) for β=45°……………………………………. 153 Input rock mass parameters of the load tests………………………….. 1980) ………………………….. 168 Properties of Rock Masses in Ironton-Russell………………………….. 2003) ………………………………………………….. (1983) and Kulhawy (1991) …………………………………………………... 87 Soil Parameters of Hypothetical Cases………………………………… 88 Cumulative Errors of p-y Curves of Hypothetical Cases………………. (1983) and Kulhawy (1991) …………………………………………………. 100 Test Drilled Shafts Information………………………………………. 112 Summary of Rock Strength Criteria…………………………………… 130 Values of Constant mi for Intact Rock (After Marinos and Hoek.... 102 Rock Mass Properties…………………………………………………. 151 Recommended Values of δ by Kulhawy et al. 168 Predictions and Ratios of Predicted over Measured Modulus of Rock Masses…………………………………… 169 xiv ... 135 Recommended Values of K by Kulhawy et al.. 1985) ……………………………... 2000)……………………………………….131 Roughness Classes (After Pells et al. 163 Empirical Equations for Estimating the Deformation Modulus of Rock Mass…………………………………. 160 Values of T* (Heuze and Amadei.89 Rock Properties for LPILE Analysis…………………………………. 151 Test Drilled Shaft Information…………………………………………....3-6 3-7 3-8 3-9 3-10 4-1 4-2 4-3 4-4 4-5 4-6 4-7 4-8 4-9 5-1 5-2 5-3 5-4 5-5 5-6 Cumulative Moment Errors based on Various Methods……………….166 The Strength of the Relation between EM and Parameters (Kayabasi et al.

. 223 Preliminary estimates of φ’CV (Robertson and Hughes. 239 xv . 212 Input Rock Mass Parameters of Dayton Load Test……………………. 182 Modulus of Rock Masses Based on Empirical Equation………………. 222 Input Rock Mass Parameters of I-85 Load Test………………………. 213 Input Rock Mass Parameters of Pomeroy-Mason Load Test…………. 1978) …………………………………..... 181 Modulus of Rock Masses Based on Pressuremeter Test……………….. 186 Comparison of P-y Criteria……………………………………………. 1986)………… 236 The Rheological Factor α for Various Soils (Baguelin et al.5-7 5-8 5-9 5-10 5-11 6-1 6-2 6-3 6-4 6-5 7-1 7-2 Summary of Lateral Load Test Drilled Shafts…………………………. 183 Summary of Definitions Related to Subgrade Reaction Theory………. 217 Input Rock Mass Parameters of I-40 Load Test………………………. 182 Measured and Predicted Initial Modulus of Subgrade Reaction……….

32 Soil and rock layer profiles at Dayton test site ……………………….. 29 Classification of shales for embankment construction (Wood and Deo.... 13 Rock-shaft model (a) Shaft and soil/rock mass system.... 28 Generalized column of bedrock units in Ohio…………………………...LIST OF FIGURES Figure 1-1 1-2 2-1 2-2 Page Drilled shaft and soil models of p-y analysis…………………………. 3 Flow chart of the work………………………………………………… 5 Distribution of ultimate lateral force per unit length (after Carter and Kulhawy 1992) ……….... 2000) ……………....... 16 Consideration of yielding of soil and/or rock mass by decomposition of loading (after Zhang et al. 1996) ………………………………………………..... (b) Coordinate system and displacement components. 28 Cross section through the rocks of central Ohio from the Indiana-Ohio border to the Ohio River (taken from Feldmann et al. 51 xvi 2-3 2-4 2-5 2-6 2-7 2-8 2-9 2-10 3-1 .... (c) Shear force V(z) and moment M(z) acting on shaft at z (Zhang et al.. 2000) ………………………………………………...... 18 Components of rock mass resistance (Zhang et al.......... 2000) ……………….. showing the pattern of surface rocks across the state ………………………………………. 1975) ……………………………………………….. 23 The elastoplastic and brittle behavior of rock mass …………………… 25 Geological map of Ohio. 19 Typical forces on wedge ……………………………………………….

.. 68 Tension strain profiles of test shaft #2 of Pomeroy-Mason test………. 77 Elimination of outlier moment profiles…………………………………83 xvii .. 61 Instrumentation and load test setup…………………………………….. 56 Tension strain profiles of shaft #4……………………………………...64 Measured load-deflection curves at loading point of Pomeroy-Mason test…………………………………………………… 65 Deflection-depth profiles of drilled shaft #1 at Pomeroy-Mason test………………………………………………… 66 Deflection-depth profiles of drilled shaft #2 at Pomeroy-Mason test………………………………………………… 67 Tension strain profiles of test shaft #1 of Pomeroy-Mason test………... 55 Deflection-depth curves of shaft #4……………………………………. 57 Boring S-9 ……………………………………………………………. 1986) …………………………………..3-2 3-3 3-4 3-5 3-6 3-7 3-8 3-9 3-10 3-11 3-12 3-13 3-14 3-15 3-16 3-17 3-18 3-19 3-20 Instrumentation of load test at Dayton ………………………………… 53 Load-deflection curves at the top of shafts……………………………. 75 Linear shape functions…………………………………………………. 60 Boring S-10……………………………………………………………. 54 Deflection-depth curves of shaft #3……………………………………. 55 Compression strain profiles of shaft #4………………………………... 69 Compression strain profiles of test shaft #2 of Pomeroy-Mason test……………………………………69 Procedure for reducing moment data to p using piecewise polynomial (after Dunnavant. 68 Compression strain profiles of test shaft #1 of Pomeroy-Mason test………………………………………...

99 Prediction of load-deflection curves of shaft #2 of Pomeroy-Mason load test using experimental p-y curves……………. 98 Prediction of load-deflection curves of shaft #1 of Pomeroy-Mason load test using experimental p-y curves……………. 103 xviii . 103 Prediction of load-deflection curve of shaft #2 of Pomeroy-Mason load test using Reese weak rock p-y criterion………. 99 Prediction of load-deflection curve of shaft #4 of Dayton load test using Reese weak rock p-y criterion………………….. depth profile of test MaumeeS1…………………………………. 96 Experimental p-y curves for shaft #1 of Pomeroy-Mason load test……………………………………………… 96 Experimental p-y curves for shaft #2 of Pomeroy-Mason load test……………………………………………… 97 Equivalent load combinations…………………………………………....3-21 3-22 3-23 3-24 3-25 3-26 3-27 3-28 3-29 3-30 3-31 3-32 3-33 3-34 3-35 3-36 3-37 3-38 Comparison of deflections from strain readings and inclinometer for test PomS1………………………………………. 83 P vs... 86 Definition of deflection prediction error………………………………..92 Four types of moment error profiles…………………………………… 93 Cumulative p-y curve errors due to moment errors…………………….... 85 P vs. 98 Prediction of load-deflection curves of shaft #4 of Dayton load test using experimental p-y curves………………………. 94 Experimental p-y curves for shaft #4 of Dayton load test……………. depth profile of test CDOTC1……………………………………. 86 Comparison of p-y curves……………………………………………… 90 Optimum spacing of strain gages……………………………………….. 102 Prediction of load-deflection curve of shaft #1 of Pomeroy-Mason load test using Reese weak rock p-y criterion………...

121 The predefined cracks………………………………………………….124 Tensile stress on xz plane for jointed rock mass case…………………. 110 Drilled shaft dimension and rock profiles of the load test at Dayton………………………………………………. and (c) outside boundary of rock………………. 127 Concrete and rock joint………………………………………………… 134 xix . 107 Finite elements for (a) drilled shaft... 116 Front view of upward movement of rock mass………………………. 119 Stress distribution at Y direction of in-depth rock layer………………. 123 Suggested stress distribution at failure at great depth ………………….. 111 Mesh convergence……………………………………………………. (b) surrounding rock. 108 Cap model: yield surface in the p-t plane (ABAQUS. 122 Friction distribution on the shaft-rock interface………………………. 119 Proposed wedge type failure model for the top layer of rock…………. 113 Comparison of load-deflection curves at shaft head…………………… 114 Comparison of deflection profiles at the load of (a) 1126 kips and (b) 705 kip ……………………………… 114 The forward movement of rock mass…………………………………....... 127 Maximum shear stress concentrations for jointed rock mass case………………………………………………….. 122 The stress redistribution at Y direction of in-depth rock layer after crack………………………………………….. 1998)…………..4-1 4-2 4-3 4-4 4-5 4-6 4-7 4-8 4-9 4-10 4-11 4-12 4-13 4-14 4-15 4-16 4-17 4-18 4-19 4-20 Finite element meshes of a drilled shaft-rock system …………………. 117 Y direction stress distribution in xz plane………………………………118 Maximum shear stress distribution on yz plane……………………….

. 163 Proposed empirical equation using GSI………………………………...….. 137 Increased normal stress due to lateral load…………………………….143 Lateral capacity calculation models for drilled shafts in rocks…………144 Capacity of rigid drilled shaft at fixed head boundary………………… 146 Capacity of intermediate length drilled shaft at fixed head boundary………………………………………………….(1985). 1985) ……………………………………. 171 Initial modulus of subgrade reaction for various moduli of rock……… 175 Initial modulus of subgrade reaction for various Poisson’s ratio……… 175 P-y curves along with depth ……………………………………………176 Initial modulus of subgrade reaction for various rock-shaft interface frictions…………………………. 1989a)………………………… ………………….... 137 The Rock-Shaft Model…………………………………………………. 162 Correction for platen bending of the jack (after Heuze and Amadei.. (b) oversize holes…………………………………………………. 147 Capacity of long drilled shaft at fixed head boundary…………………. hydraulic pressure for various borehole diameters (a) undersize – the pressure must be decreased by about 14% for a given Eapp/Eact as recalculated by Heuze et al... 1984) …………. 159 Modulus reduction vs...……… 177 Ki varies with shaft diameter…………………………………………...… 179 5-3 5-4 5-5 5-6 5-7 5-8 5-9 5-10 xx .. 148 Distribution of lateral reaction stresses………………………………… 149 The schematic of loading of Goodman Jack (Heuze. 178 Initial modulus of subgrade reaction for various shaft-rock relative stiffness………………………………….4-21 4-22 4-23 4-24 4-25 4-26 4-27 4-28 5-1 5-2 Displacement behavior of drilled shafts in rock (Johnston and Lam.

215 Comparison of load-Maximum moment of test shaft #4 at Dayton load test………………………………………..196 Interpretation of subgrade reaction modulus of rock………………….. 202 Failure mode for rock near ground surface……………………………. 194 Shaft and soil profiles of the case study………………………………. 181 Validation of empirical equation using field test data…………………. 198 P-y curves deduced from shaft #4 of load test at Dayton……………… 201 P-y curves deduced from Shaft #2 of load test at Pomeroy-Mason…………………………………………….. 192 Comparisons with Davisson and Gill method for free head condition ………………………………………………… 193 Comparison with Davisson and Gill solution for fixed head condition ……………………………………………….201 Schematics of a hyperbolic p-y curve………………………………….. 214 Comparison of load-deflection of test shaft #4 at Dayton load test………………………………………........ 184 Two-layer soil profile with four possible variations…………………… 187 Soil-pile system…………………………………………………………187 Comparison with Reese and Matlock solutions– varying soil stiffness ……………………………………….. 204 Hyperbolic p-y curves of Dayton site…………………………………...5-11 5-12 5-13 5-14 5-15 5-16 5-17 5-18 5-19 5-20 5-21 6-1 6-2 6-3 6-4 6-5 6-6 6-7 6-8 Comparison of FEM computed and predicted subgrade reaction modulus…………………………………………….. 180 Comparison of load-deflection curves of North Carolina load test……………………………………………….. 197 Comparison of shaft head deflections…………………………………. 215 xxi . 203 Failure mode of rock at great depth……………………………………..

219 Comparison of load-deflection at the loading point for Pomeroy-Mason test………………………………… 219 Comparison of load-Maximum moment of Pomeroy-Mason load test…………………………………………… 220 Comparisons of deflection-depth curves of shaft #2 at Pomeroy-Mason test……………………………………. 220 Comparisons of moment-depth curves of shaft #2 at Pomeroy-Mason test……………………………………..... 231 Determination of limit pressure from inverse of volume versus pressure………………. 1978) ……………….... 216 Hyperbolic p-y curves of Pomeroy-Mason site………………………..6-9 6-10 6-11 6-12 6-13 6-14 6-15 6-16 6-17 6-18 6-19 6-20 6-21 7-1 7-2 7-3 7-4 7-5 7-6 Comparisons of deflection-depth curves of shaft #4 at Dayton test………………………………………. 221 Layout of I-40 load test………………………………………………… 222 Layout of I-85 load test………………………………………………… 222 Comparison of load-deflection curves of I-40 short shaft………... 228 PROBEX-1 Dilatometer from RocTest……………………………….…………………… 233 P-y curves from pressuremeter (Baguelin et al. 229 Typical pressuremeter/dilatometer test curve………………………….... 216 Comparisons of moment-depth curves of shaft #4 at Dayton test………………………………………. 228 Calibration for volume and pressure losses…………………………….. 240 xxii .…… 224 Comparison of load-deflection curves of I-40 long shaft……………… 224 Comparison of load-deflection curves of I-85 short shaft……………... 225 Comparison of load-deflection curves of I-85 long shaft……………… 225 Menard G-Am pressuremeter from RocTest…………………………....

………. 241 Determination of the critical depth (Smith. 249 Derived p-y curves from dilatometers for I-85 long shaft……….….. (2001)………………… 246 Comparison of the measured and predicted deflections at shaft top………………………………………... 251 Comparison of load-deflection curves of I-85 long shaft…….…….……. 250 Comparison of load-deflection curves of I-85 short shaft………….. 251 xxiii . 1983) …………………….7-7 7-8 7-9 7-10 7-11 7-12 7-13 7-14 7-15 7-16 7-17 7-18 7-19 7-20 7-21 Steps for constructing p-y curve from pressuremeter test…………….. 246 P-y curves from dilatometer tests of Cho et al... 1983)…………………………………….. 247 Derived p-y curves from dilatometers for I-40 short shaft……………. 250 Comparison of load-deflection curves of I-40 long shaft………….. 249 Comparison of load-deflection curves of I-40 short shaft…………...…. 243 Reduction factor for depth within critical depth (Smith. 244 Shaft and soil profiles of the case study………………………………..….. 248 Derived p-y curves from dilatometers for I-85 short shaft……………..... 2001)………………………………………………….245 Dilatometer test results of the case study (after Cho et al. 248 Derived p-y curves from dilatometers for I-40 long shaft……….

Reese (1997). including those by Carter and Kulhawy (1992). Rock-socketed drilled shafts are also used to stabilize slopes. for laterally loaded drilled shafts. For axially loaded drilled shafts socketed into rock.CHAPTER I INTRODUCTION 1. Although there exist several analysis and design methods specially for rock-socketed drilled shafts under lateral loading. there is a lack of validated. and Gabr et al. Zhang et al.1 Statement of the Problem Drilled shafts socketed into rock are widely used as foundations for bridges and other important structures. 1 . However. 1993). however. numerous research efforts have been conducted in the past. This practice has created erroneous designs and often leads to excessive socket length. The main loads applied to the drilled shafts are axial compressive or uplift loads as well as lateral loads with accompanying moments. especially for the determinations of side shear resistance. It has been a customary practice to adopt the techniques developed for laterally loaded piles in soil to solve the problem of rock-socketed drilled shafts under lateral loading (Gabr. (2000). these methods were developed with limited validations against field lateral load test data. rational analysis and design methods. (2002).

The lack of validated design methods stimulates the need to develop a more rational design approach for laterally loaded drilled shafts in rock. Zhang et al. However. and the method of To et al. 1996. several other researchers. Dykeman and Valsangkar. shown in Fig. and To et al. The 2 . there is also a need for the development of methods for estimating lateral capacity of shafts. In addition to the need to develop analysis methods for predicting lateral shaft deflections. Additionally. The p-y method has been widely and successfully used for the design of laterally loaded drilled shafts in soils for decades. 2003) have been developed for estimating lateral capacity of piles in rock. This method is based on a numerical solution of a physical model based on a beam on Winkler foundation. (2000)’s analysis method has not been evaluated by others due to the complexity of the method and the needs of a computer program. and Cho et al. 1-1. They concluded that the two methods provided very unconservative results. (2000).Several researchers (such as DiGioia and Rojas-Gonzalez... 1993. Carter and Kulhawy’s (1992) method does not consider the effect of secondary structures of rock mass. especially for weak rock masses which can not be exactly characterized by elastic-perfectly plastic model. Additionally. 1992. (2003) is only suitable for two sets of parallel joints and rigid piles. a method for predicting ultimate capacity of drilled shafts in rock mass needs to be developed. such as Reese (1997) and Zhang et al. Very few methods (Carter and Kulhawy. The assumption of an elastic-perfectly plastic model for rock masses by Zhang et al. have proposed methods to predict the ultimate lateral resistance of rock per unit shaft length. (2000) prohibited its wide application. 2001) have evaluated the methods of Carter and Kulhawy (1992) and Reese (1997) using their lateral load test data.

2) Develop a simple elastic solution for predicting deflections of laterally loaded drilled shafts in rock in a two-layer soil/rock profile. if a rational and appropriate p-y criterion for rock masses can be developed. H M y p p p p z y y y y Figure 1-1 Drilled shaft and soil models of p-y analysis It is believed that the non-linear p-y approach is the best way to analyze and design laterally loaded drilled shafts in rock. non-linear springs. this study will focus on the development of p-y curves for rock masses. such as pressuremeter or dilatometer tests. Therefore. and the second is to derive the p-y curves using in-situ tests.2 Objectives The objectives of this research are to: 1) Develop a method to predict the lateral capacity of drilled shafts in rock masses. 3 . while the soil-shaft interaction is represented by discrete.structural behavior of the drilled shafts is modeled as a beam. 1. There are two ways that one can generate p-y criterion: one is to use fundamental rock properties along with basic rock-shaft interaction mechanics.

4) Identify the best field or laboratory test methods for determining rock mass properties used in the developed p-y criterion.3) Develop a p-y criterion for rock mass. Additionally. 5) Perform full-scale field lateral load tests on fully instrumented drilled shafts to obtain reliable and comprehensive field test data for validating the p-y criterion. it is necessary to obtain the initial slope and ultimate rock reaction.3 Scope of the Work The work involved in this study mainly consists of two parts: one is the theoretical work to develop a p-y criterion for rock and evaluation of an existing approach for deriving site specific p-y curves using dilatometer tests. The types of rock of interest and rock classification systems are identified. Specifically. A literature review on the design and analysis methods of laterally loaded drilled shafts and piles in rock is performed. To construct a hyperbolic p-y curve. the other is to conduct full-scale field lateral load tests. the work to be done are outlined as follows and depicted in Figure 1-2. 1. The determination of ultimate rock reaction involves a 3D finite element analysis to identify the possible failure modes of rocks 4 . Necessary correlations between rock properties and the p-y curve parameters will also be established. an evaluation is carried out to identify the most suitable method for deriving p-y curves based on test results of a fully instrumented lateral load test. 6) Review and recommend a best approach for deriving site specific p-y curves of rock mass from pressuremeter or dilatometer test data. A lateral capacity prediction method and elastic solution for estimating shaft deflection under lateral loads are developed as well.

5 . Direction I Rock Types Direction II P-∆v Curve of Dilatometer Rock Classification Side Resistance Rock Strength Failure Modes Rock Mass Modulus Ultimate Resistance Load Test Data Initial Slope of p-y Lateral Capacity P-y Criterion Elastic Solution Site Specific p-y Curve Figure 1-2 Flow chart of the work In addition to the development of a p-y criterion of rock. a method for predicting lateral capacity of drilled shafts in rock and an elastic solution for estimation of drilled shaft deflections in a two-layer soil/rock profile under lateral loads are proposed.subjected to movement of laterally loaded drilled shafts. a correlation between the slope and deformation modulus of rock masses is established using a parametric study based on a 3D FEM model. a hyperbolic p-y criterion for rock mass is proposed. Rock strength criteria and side shear resistance between rock and shaft are carefully identified. To obtain the initial slope of p-y curves. Finally. based on the ultimate rock reaction and initial slope of p-y curve.

An evaluation of various existing methods for deriving p-y curves from the results of an instrumented lateral load test is carried out to identify the best method. Also.4 Outlines of the Dissertation Chapter II presents the literature review on analysis methods of laterally loaded drilled shafts/piles in rock. A 3D finite element modeling technique is established and validated against a lateral load test. and rock classification systems are reviewed. the existing methods for deriving site specific p-y curves for soils from pressuremeter and dilatometer tests are reviewed and examined with lateral load test results of rock-socketed drilled shafts. At the end of this chapter. Chapter III presents the results of two full-scale lateral load tests on instrumented drilled shafts socketed in rock. Simulations using 3-D FEM modeling technique 6 . 1. rock strength criterion. and side shear resistance will be elucidated. The developed method for determination of ultimate resistance of rock mass against laterally loaded shafts is presented in Chapter IV. the existing Reese (1997) interim p-y criterion is evaluated against the results of the two lateral load tests. typical Ohio bedrock.Two field lateral load tests are conducted to facilitate the development and validation of the p-y criterion for rock mass. The most suitable method is then recommended for deriving p-y curves from pressuremeter and dilatometer tests. in which the failure modes of rock in response to laterally loaded drilled shafts. The best suitable method for driving p-y curves from the load test results is identified and it is used to derive the site-specific p-y curves of the two load test sites. Finally.

An elastic solution for predicting the deflections of laterally loaded drilled shaft in a two-layer soil/rock system is proposed in Chapter V. Finally. Additionally.enable the identification of failure modes of rock mass. Chapter VII provides an evaluation of several existing methods for interpreting the properties of rock mass and driving p-y curves from pressuremeter or dilatometer tests. Chapter VI presents the development of a new hyperbolic p-y criterion for rock mass. in which a correlation equation for estimating the modulus of rock mass based on modulus of intact rock and a rock classification system is also developed. a correlation equation for estimating the subgrade reaction modulus of rock mass based on a 3D finite element parametric study is established as well. a method for predicting lateral capacity of drilled shafts in rock and/or soil is proposed. The most suitable method for deriving p-y curves for rock from dilatometer tests is recommended based on the evaluation findings. Chapter VIII presents summaries and conclusions of the research work. A validation of the proposed p-y criterion of rock is carried out by comparing the predictions against actual lateral load tests results. 7 .

1 Analysis Methods of Laterally Loaded Rock-Socketed Drilled Shafts 2. and Gabr et al. two categories of analysis methods for laterally loaded rock-socketed drilled shafts have been developed. Carter and Kulhawy (1992) proposed a method that treats rock mass as a homogeneous elastic continuum.CHAPTER II LITERATURE REVIEW 2. To consider the yielding. The elasto-plastic soil/rock response under lateral loads and the linearly variation of deformation modulus of soil/rock along depth were assumed. However. Currently. Parametric solutions for the load-displacement relationships were generated by using the finite element technique. the other one discretizes the rock mass into a set of non-linear springs (Reese 1997. It has been a customary practice to adopt the p-y analysis with p-y criterion developed for soils to solve the problem of rock-socketed drilled shafts (Gabr. a method based on Hoek-Brown criterion (Hoek and 8 . The approach adopts and extends the basic idea of Sun (1994) on laterally loaded piles in soil. 1993). One category treats rock as a continuum mass (Carter and Kulhawy 1992.1. (2000).1 Introduction To date. therefore. elastic continuum model is only good for small loads. 2002). Zhang et al. and Zhang et al 2000). developed a nonlinear continuum approach. there are few published analysis methods for the lateral response of rocksocketed drilled shafts.

The second category of analysis method.Brown 1980. Carter and Kulhawy (1992) presented a method to determine the rock capacity by using cohesion and friction angle of rock. such as p-y method. Reese (1997)’s method for estimating ultimate rock reaction per unit 9 . discretizes the rock masses into a series of nonlinear springs. (2002) proposed a p-y criterion for weak rock based on their field test data. By assuming distribution of ultimate rock resistance along the depth of a shaft. especially weak rock. In practice. However. without considering secondary structures of rock mass. In addition to the above mentioned analysis methods for solving load-deflection relationship at the drilled shaft head. Cater and Kulhawy (1992)’s method on rock resistance. however. the ultimate lateral capacity of a rock-socketed drilled shaft was obtained by summing the capacity of compressive resistance and shear resistance between shaft and rock. Gabr et al. Reese (1997) considered the secondary structure of rock mass by using a rock strength reduction factor which can be determined from Rock Quality Designation (RQD). treats rock mass as a homogeneous and elasto-plastic material. such as cracks and fractures. the rock masses. show nonlinear and nonhomogeneous properties which can not be fully captured by an elasto-plastic model. This method was based on a theory of expansion of a long cylindrical cavity in an elasto-plastic. dilatant material. 1988) was proposed to calculate the ultimate resistance of rock masses. The p-y method was extended to the analysis of single rock-socketed drilled shaft under lateral loading by Reese (1997). methods for estimating ultimate rock reaction have also been proposed. The method requires input of Poisson’s ratio. shear modulus and dilation angle. An interim p-y criterion for weak rock was proposed. Thereafter.

by using finite element technique. especially possible sliding failures along pre-existing joints. The solutions for these two variables were expressed as a function of effective Young’s modulus Ee and an equivalent shear modulus G*. Regarding the sliding failure on joints. The drilled shaft is idealized as a cylindrical elastic inclusion with effective Young’s modulus Ee. However. the authors did not investigate possible failure modes of rock mass. The Coulomb failure criterion was utilized to model the sliding failure on joints.2 Brief Description of the Existing Methods 2.2.1 Carter and Kulhawy (1992) Carter and Kulhawy (1992) performed a parametric study to obtain deflection u and rotation angle θ at the shaft head. RQD can not fully represent all the secondary rock structures. and the static limit equilibrium was used to obtain the ultimate capacity. which is defined as 10 . however. To et al. (2000)’s method for estimation of lateral capacity of rock-socketed drilled shaft considered most of characteristics of rock mass. 2. in which RQD and other secondary rock structures were included. It seems that Zhang et al. Additionally. simple rock resistance distribution along the shaft circumference under lateral loads was assumed (Carter and Kulhawy 1992). (2000) proposed a method to estimate the ultimate reaction of rock masses per unit shaft length using Hoek-Brown rock strength criterion (Hoek and Brown 1988).length ignored the contribution of shear resistance between shaft and rock. The block theory (Goodman and Shi 1985) was used to identify the failure block.1. (2003) proposed a method to estimate the lateral load capacity of drilled shafts in jointed rock. such as spacing and condition of discontinuities.1. Zhang et al.

08 * 2  e  *  G D  G  −3 / 7 (2-5) −5 / 7  H  E  θ = 1.5 *  e  *  G D  G  −1 / 7  M  E  + 1. flexible and rigid. and ν = Poisson’s ratio of rock. D = diameter of the drilled shaft. ground-line deflection u and rotation θ induced by the lateral load H and the moment M at shaft top are calculated from the following equations:  H  E  u = 0. were classified by the authors. isotropic elastic material. The rock mass is assumed to be a homogeneous.08 * 2  e  *  G D  G  −3 / 7  M  E  + 6. (EI)c = the actual bending rigidity of the shaft. A flexible pile is one in which the following condition meets: L  Ee  ≥  D  G*  2/7 (2-4) For a flexible drilled shaft.4 * 3  e  *  G D  G  (2-6) 11 . From the finite element analysis performed by Carter and Kulhawy (1992). The equivalent shear modulus is given by  3  G * = G r 1 + ν   4  where (2-2) Gr = Er 2(1 + ν) (2-3) in which Er = Young’s modulus of rock.Ee = (EI) c πD 4 64 (2-1) in which. Two categories of shafts. it was found that u and θ are largely dependent on the ratio of Ee / G* and the ratio of the shaft socket length to the diameter L/D.

which can be calculated according to the expansion theory of a long cylindrical cavity in an elasto-plastic. and frontal compressive strength of rock.25 times the larger displacements of those calculated values by treating the shaft as a flexible or a rigid shaft. Carter and Kulhawy (1992) proposed a solution in which they suggested that the lateral resistance were derived from side shear τ between shaft and rock. the authors suggested that the displacements be taken as 1. the ultimate resistance remains constant with depth. ground-line deflection u and rotation θ are calculated from the following equations:  H  2L  u = 0.A drilled shaft is considered to be rigid when L E  ≤ 0. cohesive-frictional. The assumed distribution of ultimate resistance along the shaft is shown in Fig. Below this depth. from which one can see that lateral resistance is equal to τmaxD at the surface of the rock and is increasing linearly with depth to a magnitude of (PL + τmax)D at a depth of 3D.4 *    G D  D  −1 / 3  M  2L  + 0. 2-1. dilatant material (Carter et 12 . The authors further suggested that the magnitude of this shear was equal to that produced in axial loading.3 * 2    G D  D  −7 / 8  M  2L  + 0.3 * 2    G D  D  −7 / 8 (2-8) −5 / 3  H  2L  θ = 0.8 * 3    G D  D  (2-9) For the drilled shafts having intermediate rigidity.05 e  * D G  1/ 2 (2-7) For a rigid drilled shaft. PL is the limit stress developed in rock. For ultimate capacity of rock-socketed drilled shafts.

al. an interim p-y criterion for weak rock was suggested. D = diameter of the drilled shaft. Hult Surface of rock 3D L Dτmax D D(pu+τmax) Figure 2-1 Distribution of ultimate lateral force per unit length (after Carter and Kulhawy 1992) 2. the term “interim” was applied 13 . In order to characterize the rock response under lateral loading.1. 1986).2. Hu: P L  H u =  L + τ max D L  6  for L<3D (2-10) P  H u =  L + τ max 3D 2 + ( PL + τ max )(L − 3D)D for L>3D 2   (2-11) where τ max = maximum unit side resistance. and L = length of drilled shafts embedded in rock.2 Reese (1997) The p-y method for the analysis of drilled shafts in soils under lateral loading was extended to the analysis of rock-socketed drilled shafts by Reese (1997). The following equations were derived by Carter and Kulhawy to determine the lateral capacity of rock-socketed drilled shafts. Due to the lack of adequate test data.

The value of αr is assumed to be 1/3 for RQD of 100% and it increases linearly to unity at a RQD of zero. derived by correlation with experimental data.to this p-y criterion. k ir = (100 + kir = 500 400z r ) 3D for zr ≥ 3D for 0 ≤ zr ≤ 3D (2-14) (2-15) A complete description of the interim p-y criterion may be summarized as follows. y ≥ yA and p ≤ pu (2-17) (2-18) p ≥ pu 14 . The expressions for kir.  y  y  rm     0. The slope of initial portion of p-y curves was given by Kir = kir Em (2-13) where Kir = initial tangent to p-y curve.25 . y ≤ yA (2-16) p Second segment: p = u 2 Third segment: = pu. which is used to account for fracturing of rock mass. Em = deformation modulus of rock masses. With this interim p-y criterion.2α r σ ci D for zr ≥ 3D (2-13) where σci = uniaxial compressive strength of intact rock. D = diameter of the drilled shaft. of rock was given by p u = α r σ ci D(1 + 1. are as follows. The ultimate reaction.4 zr ) D for 0 ≤ zr ≤ 3D (2-12) p u = 5. COM624P or LPILE can be run to obtain the lateral response of rock-socketed drilled shafts. which may be obtained from a pressuremeter or dilatometer test. pu (F/L). and zr = depth below rock surface. First segment: p = Kir y. and kir = dimensionless constant. αr = strength reduction factor.

3 Zhang et al. EpIp = flexural rigidity of the shaft. Similarly. krm = strain at 50% of ultimate load.2. The approach was extended from the basic idea of Sun (1994) on laterally loaded piles in soil. (2000) Zhang et al. The model of rock-socketed drilled shafts under lateral loading is shown in Fig. The deformation modulus of soils varies linearly from Es1 to Es2.1.333 (2-19) (2-20) in which.where yrm = krm D   pu yA =   0. (2000) proposed a nonlinear continuum method to predict the loaddisplacement response of rock-socketed drilled shafts under lateral loads by treating soil/rock as an elasto-plastic material. the deformation modulus of rock mass varies linearly from Em1 to Em2 at the tip of shaft and stays constant below the shaft tip. 2. and 15 . the following governing equations were obtained: EpIp  d4us z  du s  z  d  − 2t s  + k s ηs + (1 − ηs )  u s = 0 (0 ≤ z ≤ Ls) (2-21a) ηs + (1 − ηs )  4 dz  L s  dz  Ls  dz   z − L s  du m  z − Ls  d4u m d  EpIp − 2t m η m + (1 − η m )  + k m η m + (1 − η m )  u m = 0 4 dz  L m  dz  Lm  dz  ( Ls ≤ z ≤ L) (2-21b) where us and um = displacement components of the shaft in the soil and in the rock mass. ranging from 0.25  2( y rm ) K ir  1.0005 to 0. respectively. By minimizing the energy of rock-shaft system with respect to displacements. Lm = shaft length embedded in rock masses. 2-2. z = depth starting from ground line. Ls = shaft length embedded in soils.00005.

(c) Shear force V(z) and moment M(z) acting on shaft at z (Zhang et al. m1 and m2 = parameters describing the behavior of the elastic foundations. (b) Coordinate system and displacement components. respectively. Figure 2-2 Rock-shaft model (a) Shaft and soil/rock mass system. 2000) 16 .ηs = E s1 E s2 E m1 E m2 (2-22a) ηm = (2-22b) ts = πE s 2 R 2 m1 2(1 + ν s ) π(3 − 4ν s )E s 2 m 2 2(1 + ν s )(1 − 2ν s ) πE m 2 R 2 m1 2(1 + ν m ) π(3 − 4ν m )E m 2 m 2 2(1 + ν m )(1 − 2ν m ) (2-22c) ks = (2-22d) tm = (2-22e) km = (2-22f) in which νs and νm = Poisson’s ratio of soils and rock masses.

If p > pu. repeat steps two and three until no further yielding of soil or rock occurs. therefore. Elastic-perfectly plastic stress-strain relationship. take the depth z as the yielding depth zy. can be obtained as  d3u V (z) = E p I p  3s  dz   d 3u m  V(z) = E p I p  3  dz   z  du   − 2 t s ηs + (1 − η s )  s   L s  dz      z − L s  du m   − 2 t m η m + (1 − η m )    L m  dz    (0 ≤ z ≤ Ls) (2-23a) ( Ls ≤ z ≤ L) (2-23b) and the bending moment M(z) acting on the shaft is be given by  d2u M ( z ) = E p I p  2s  dz   d2u m M (z ) = E p I p   dz 2          (0 ≤ z ≤ Ls) (2-24a) (Ls ≤ z ≤ L) (2-24b) The governing differential equations and the shear force and bending moment can be solved using classical finite difference method and an iterative process. Thirdly. Fourthly. and analyze it by using the elastic solution while ignoring the effect of the yielded portion of shaft. Finally. for the applied lateral load H and the moment M. the final results can be obtained by considering the two parts of the shaft separately. treat the unyielded portion of shaft as a new shaft. The method for considering the yielding of soil or rock mass consists of several steps depicted in Fig. The portion of shaft in yielded soil and/or rock mass is analyzed 17 . Secondly. the lateral reaction force p at certain depth is computed and compared to the ultimate resistance pu at that depth.The shear force V(z) acting on the shaft. 2-3. Firstly. the shaft is analyzed by using the above elastic solutions. The above process considers the soil/rock to be elastic. was proposed to consider the yielding of the soils or rock masses. 2-2(c). shown in Fig.

pu = NpCuD where Cu = undrained shear strength of soil. For clay. (2000) suggested two existing methods for clay and sand. R = shaft radius. The other part of shaft in the unyielded soil and/or rock mass is analyzed by using the elastic solution.25 to 0. (1992) as follows.as a beam with distributed load pu acting on it. To compute the ultimate resistance pu of soil. Zhang et al. Figure 2-3 Consideration of yielding of soil and/or rock mass by decomposition of loading (after Zhang et al. and Reese and Welch (1975) was suggested. p u = K 2 γ ' zD p (2-27) 18 . and (2-25) Np = 3 + γ' J z+ z≤9 cu 2R (2-26) in which γ ' = average effective unit weight of soil above depth z. Zhang et al. 2000) For sand. D = diameter of drilled shafts. and J = coefficient ranging from 0.5. respectively. the equation proposed by Matlock (1970). (2000) suggested the method of Fleming et al.

shown in Fig. Figure 2-4 Components of rock mass resistance (Zhang et al. The assumption for resistance distribution is that the total resistance of rock mass consists of two parts: the side resistance and the front normal resistance. The ultimate resistance pu can be calculated by pu = (pL +τmax)D (2-28) where D = diameter of the drilled shaft. 2000) 19 .80(σ ci ) 0.5 where σci = unconfined compressive strength of the intact rock (MPa).5 (MPa) (MPa) for smooth socket for rough socket (2-29a) (2-29b) τ max = 0. and Hoek-Brown rock strength criterion (Hoek and Brown 1988). and pL = normal limit resistance. To compute the ultimate resistance pu of rock mass. τmax = maximum shearing resistance along the sides of the shaft. 2-4. τmax was assumed to be the same as the maximum side resistance under axial loading and can be given by τ max = 0. (2000) proposed to utilize the assumed resistance distribution (Carter and Kulhawy 1992).where Kp = tan2(45°+ φ ' /2) = Rankine passive coefficient.20(σ ci ) 0. in which φ ' = effective internal friction angle. Zhang et al.

the normal limit stress pL. 1988) was adopted to determine the normal limit stress pL. and a (Hoek and Brown 1997). σ1 and σ 3 = the major and minor effective principal stresses.5. can be given by ' σ1 = ' σ3   σ' + σ ci  m b 3 + s    σ ci   a (2-30) ' ' where σci = uniaxial compressive strength of the intact rock. (2002). 1961): 20 .2.The strength criterion for rock mass developed by Hoek and Brown (1980. With Hoek-Brown’s strength criterion. The following procedure can be used to construct a p-y curve according to Gabr et al. can be obtained by assuming that the minor principal effective stress is the effective overburden pressure γ ' z.1. mb = mi. and a = 0. Rock Mass Rating (RMR) of Bieniawski can also be used to determine the constants mb. and a can be estimated by correlations with Geological Strength Index (GSI) (Hoek. which is the ' major principal effective stress σ1 . Step 1: Calculation of Coefficient of Subgrade Reaction The coefficient of subgrade reaction can be calculated as follows (Vesic. In addition to GSI. (2002) proposed a hyperbolic p-y criterion for weak rock based on field tests on small diameter drilled shafts socketed in weak rock. For rock mass. s. The Hoek-Brown criterion. and mb. s.4 Gabr et al. s = 1. 1994). and a = constants depending on the characteristics of the rock. which is suitable for intact rock and rock mass. 2. s. respectively. (2002) Gabr et al. a constant depending on rock type. For intact rock. the values of mb.

T0 is turning point. Step 3: Calculation of Point of Rotation The following equation is used to define the turning point as a function of the embedded shaft length: T0 = (1 + 0. is computed as follows (Poulos and Davis. Ip is the moment of inertia of shaft. Ep is modulus of elasticity of shaft. KR.65E m  E m D 4  nh =   D(1 − ν 2 )  E p I p    1 / 12 (2-31) where D is the diameter of a drilled shaft. Step 4: Calculation of IT Number IT = -28 .383 log(T0/L) IT ≥ 1 (2-34) (2-33) Step 5: Calculation of the Subgrade Reaction kh = nh D (0 ≤ z ≤ T0) kh = IT nh D (T0 < z ≤ L) Step 6: Calculation of Ultimate Resistance of Rock Mass Pu (2-35) (2-36) 21 . 1972): KR = EpIp E m L4 (2-32) where. Step 2: Calculation of Flexibility Factor A flexibility factor. ν is Poisson’s ratio of rock mass.0. L is the embedment length of shaft.18 log K R )L where. and GSI is Geological Strength Index.

but becomes removable if it breaks due to the lateral force exerted by the pier. preventing the movement of the wedge.1. and Einstein (2003) For the drilled shafts socketed into jointed rock. Step 7: Construction of the P-y Curve p= y 1 y + kh pu (2-37) 2. are depicted in Fig. Due to the complexity of the entire process to obtain the failure block. and R1 = tangential force on joint. T = tensile force due to the fracture of Category II blocks. which is defined as a block that is not removable. (2000) was employed to calculate the ultimate resistance of rock as presented in Section 2. To et al.2.The Eq. Goodman and Shi (1985)’s block theory was used to determine the possible failure block for two sets jointed rock mass with the help of AutoCAD or Excel. no details about the block theory will be described here. F = lateral force. The assumed mechanisms of sliding failure along the joint plane and tensile failure on the rock mass.5 To. (2003) assumed a wedge type block failure and Coulomb failure criterion to obtain the lateral capacity of drilled shafts.3. N1 = normal force on joint. Smooth condition was assumed for all the cases when the side shear resistance is concerned. (2-28) proposed by Zhang et al. where W = weight of the wedge. 2-5. P = axial load of shaft.2.1. 22 . Ernst.

2.Figure 2-5 Typical forces on wedge Normal force N and tangential force R can be related by the Coulomb failure criterion as follow: τ = c + σ tan ϕ where τ = shear stress.3.3 Comments on the Existing Analysis Methods (2-38) 2. which implies that the solution is only applicable to small loads. 2-5) was used to solve for ultimate lateral force F. Static limit equilibrium relation between the forces on the wedge (Fig.1. c = cohesion.1 Carter and Kulhawy (1992) Carter and Kulhawy (1992) provides solutions for the lateral load-deflection relation at shaft head as well as shaft lateral capacity. The solution for ultimate lateral capacity needs verification.1. For load-deflection prediction. σ = normal stress. φ = friction angle. One of the drawbacks of the solution is the requirement of numerous rock 23 . Carter and Kulhawy assumed rock mass as an elastic material.

the stress will drop after it reaches the peak. the 24 . However. brittle post-failure may not occur. 2. The comparison showed that Carter and Kulhawy (1992)’s method tend to overestimate the ultimate lateral capacity by a factor of two. such as shear modulus. cohesion. which is likely to behave as a brittle material (Hoek and Brown 1997). Dykeman and Valsangkar (1996) conducted a centrifuge test on eight model socketed shafts and used the test results to evaluate Carter and Kulhawy’s method. however. 2-6. the actual nonlinear behavior may already appear before the stress reaches the peak. They found a reasonable agreement between the measured and predicted displacements for these foundations at low load levels (20-30% capacity). However.2 Zhang et al. despite that yielding of rock was considered in this method. Similar comparison results were found for the predicted deflection at shaft head. friction angle. 2-6. In addition to their own test data. (2000) method. as shown in Fig. the elasto-perfectly plastic stress-strain relationship can not fully represent the nonlinear behavior of rock masses.deformation and strength parameters. while it predicted smaller deflection at shaft head than measured deflection at a given load level. Dykeman and Valsangkar (1996) evaluated Carter and Kulhawy’s method by using Frantzen and Stratten (1987)’s field test data. and dilation angle.3. As shown in Fig.1. (2000) The yielding of rock mass was considered in Zhang et al. DiGioia and Rojas-Gonzalez (1993) evaluated this method by using their field lateral load test on drilled shafts socketed into rock mass. for good quality rock mass. this method gave predictions that were stiffer than observed at higher load levels. Therefore. For weak rock mass. Additionally.

Reese (1997)’s interim rock p-y criterion for weak rock was used to construct p-y curves. stress Linear elastic– perfectly plastic Brittle strain Figure 2-6 The elastoplastic and brittle behavior of rock mass 2.4 Existing P-y Criteria for Rock Reese (1997)’s interim rock p-y criterion was not well calibrated due to inadequate test data. and an estimation of ultimate lateral capacity did not include the effect of friction between rock mass and shaft. (2003) The determination of removable wedge required by this method is a very tedious procedure which involves the use of AutoCAD and EXCEL. The failure mode is restricted to failure at the top portion of rock mass with a free surface at the groundline.3.3 To et al.1. only two sets of joints can be considered and the joints in each set should be parallel.1. It was found 25 .behavior of different types of rock mass was not fully represented as an elasto-perfectly plastic model. The failure modes of rock mass were not well defined. (2001) conducted a lateral load test on rock-socketed drilled shafts. 2.3. Determination of such parameters as constant krm appears to be empirical. Furthermore. Cho et al. Two drilled shafts. 30 inch in diameter and 10 feet to 13 feet of socket were laterally loaded.

2-7. Devonian. 1938). Gabr (1993) evaluated two field tests on rock-socketed drilled shafts under lateral loads presented by Carter and Kulhawy (1992) by using p-y analysis and the p-y criterion for stiff clay (Reese et al. Gabr et al. and Permian (Lamborn et al. They are ordered as. This is not same as the case of drilled shafts embedded in rock where shaft-rock interaction is more complicated. Pennsylvanian. Although the interim rock p-y criterion was not well established. from the oldest to youngest.that the interim p-y curves underestimated the deflections of shafts when comparing with the measured values.2. Mississippian..1 The Distribution of Rocks in Ohio Six of the groups or systems in the historical classification of rocks are present in the outcrops of Ohio. However. provides the distribution of those rock systems across the state. Usually. shown in Fig. 26 . the older rock systems are bedded under the younger rock systems. (2002) p-y criterion is a most recently proposed p-y criterion for weak rock. Silurian. The six systems of rock are formed in different periods in the history of earth.2 Bedrocks of Ohio 2. the p-y method is still a promising method for rock-socketed drilled shaft under lateral loads. 1975). 2. it has not been further validated with other load tests. The equation for estimating modulus of subgrade reaction was based on Vesic (1961)’s equation for beam on elastic foundation. Ordovician. The analysis results showed that p-y approach is analytically attractive because it can approximately model the nonlinearity in the load-displacement response. The geological map.

1938). thin beds of coal. limestone. For example. 2-8. 1938). clay. non-calcareous shale is the bedrock lying immediately below the glacial drift. 1968). as well as many thin beds of coal. brackish. ranges from 355 feet to 545 feet and crops out across the center of southeastern Ohio in a belt 10 to 20 miles in width. sandstone. the bedrock is predominantly limestone and dolomite with minor amounts of calcareous shale. and conglomerate. located in the southeastern of Ohio. iron ore. or fresh waters (Lamborn et al. and iron ore. shaly sandstone. All the common varieties of the sedimentary series can be found. Bedded marine shales and some thin marine limestones are presented in the lower part of this series. those in the eastern half are mainly of the clastic or fragmental varieties. shale. Shales of varying character are widely distributed both horizontally and vertically on the outcrop (Lamborn et al. dolomite. Lens-like bodies of sandstone are widely distributed but lateral gradation to shale is of common occurrence. In contrast to the dominantly calcareous nature of the beds exposed in the western half of Ohio. The bedrocks exposed at the surface in Ohio are all of sedimentary types formed from unconsolidated sediments deposited in marine. In the western half of the State. Although in the Pennsylvanian and Permian systems of eastern Ohio.The cross section through the rocks of central Ohio from the Indiana-Ohio border to the Ohio River is shown in Fig. clay. The generalized column of 27 . In the northwestern corner of the State. where thin limestone interstratified with shale is the usual mode of occurrence. the Conemaugh series of Pennsylvanian System. and conglomerate are present. thick beds of shale. while the upper part includes abundant red calcareous claystones (Fisher et al. and sandstone are the rule. such as limestone. The calcareous shales are most abundant in the southwestern part of the State.

showing the pattern of surface rocks across the state Figure 2-8 Cross section through the rocks of central Ohio from the Indiana-Ohio border to the Ohio River (taken from Feldmann et al. Figure 2-7 Geological map of Ohio. 1996) 28 .. which was complied by Hull (1990) and revised by Larsen (2000).bedrock in Ohio is presented in Figure 2-9.

Figure 2-9 Generalized column of bedrock units in Ohio 29 .

green. is a variety of sedimentary rock resulted from the consolidation of more or less thinly laminated or bedded silts and clays (Lamborn et al.2. quartz. shale is less weather resistant than most other varieties of sedimentary rocks. southwestern. rutile. apatite. Dark bluish gray.2.. calcite. brown. 30 .2. In general. siltstone is produced from shale. With an increase in the percentage of calcium carbonate. The chief mineral constituents found in shale are hydrated aluminum silicates of varying composition. The composition of shales is variable as it is assemblage of many mineral substances of different chemical composition. yellowish brown. dolomite. highly distributed in eastern. The hydrated silicates and quartz make up a large part of the average shale and in themselves produce a rock which is white. shale is transited to bone coal and coal. or red in color. Where iron oxide occurs in the absence of much organic matter.2. and northwestern of Ohio. and iron oxides. and claystone. coal. As carbonaceous material becomes greater in amount. Organic material is generally present in small amounts. Shale beds are associated with sandstone. Shale may gradually become shaly sandstone and finally sandstone as quartz increases in percentage and size of grain. or greenish gray in color. dark brown and black colors are usually found when finely divided organic material is presented in the shale. Green or greenish gray colors are characteristic of unweathered shale rich in either iron carbonates or chloritic material and poor in organic matter.1 Introduction Shale. 1938). and as the fissility or shaly structure disappears.2 The Shales in Ohio 2. limestone. shale transits to limestone. the shale may be buff.

and soil-like shales.. and from limestone and dolomite by its fissility and low carbonate content. as shown in Figure 2-10.g. Four simple tests were suggested to characterize shales for embankment use: one cycle slaking in water. In Figure 2-10. from coal. The system classifies shales into four groups: rock-like shales. such as reduced shear resistance. Wood and Deo (1975) proposed a shale classification system for embankments. clay and mudstone by its fissility. slakedurability on an initially dry sample. and a modified sodium sulfate soundness test.2. (Id)d is the slake durability index of dry sample. intermediate-1 shales. and Is is the soundness index. Various shale classification systems have been proposed by different investigators from engineering view point. the correlation between shale-shaft side resistance reduction due to water intrusion and the durability index can be established (not to be done in this study). If a series of tests on shale strength and durability can be organized and conducted.2. sieved 31 . Hand specimens of typical shale are distinguished from sandstones and conglomerates by the smaller grain size and by the fissility. swelling.2 Shale Classification System Typical shales possess fissility or the property of splitting with comparative ease along the planes of lamination or bedding. the engineering performance of shales for deep foundation design. intermediate-2 shales. For instance. reduced bearing capacity. could be correlated to the durability and other properties of shales. The classification of shales for deep foundations has not been established yet. Richardson and Wiles (1990) proposed a correlation between the loss of shear strength and results of simple tests (e. However.2. slake-durability on a soaked sample. (Id)s is the slake durability index of soaked sample. such as Wood and Deo (1975).

Hawk and Ko (1980) examined the orthotropic nature of two shales and concluded that the properties of both shales are represented well by a transversely isotropic model.2. and change in point-load strength) for embankment applications. Sargand and Hazen (1987) 32 . and RQD of shales are given.3 Engineering Properties of Ohio Shales A compilation of field and lab test results on Ohio shales is presented in Table 2-1. 1975) 2. Additionally. natural moisture content. where the unconfined compressive strength. The effect of slake during drilling or coring of shales should also be taken into account.2. σci. The cored shales can be used for slake-durability test.slake durability index. Slaking test in water in one cycle Slake durability test on dry samples Slake durability test on soaked samples Modified Soundness test Slakes completely Does not Slakes completely (Id)d<90 (Id)d>90 (Id)s < 75 90 > (Id)s > 75 (Id)s > 90 Is < 70 90>Is >70 98 > Is > 90 Is > 98 Soil-like shales Intermediate -2 shales Intermediate -1 shales Rock-like shales Figure 2-10 Classification of shales for embankment construction (Wood and Deo. although an isotropic model is also acceptable.

Y. Also. excess water under pressure may reduce the cohesion. This cohesion is usually dependent upon a skin of absorbed. However. (2): Data obtained from E. R. (3): Sargand and Hazen (1987). oriented water molecules and sometimes also upon surface tension between the water and any air that may exist from time to time in the pores of the coarse phases of the sediment. The strengths were obtained with 100 to 200 psi of confining pressure. Finally. Engel. Table 2-1 Engineering Properties of Rocks in Ohio Project Depth Rock Type RQD σci Name (ft) (%) (psi) (1) Erie ERI60 26 Medium-hard to hard shale 100 7320 (1) Erie ERI60 15 Medium-hard to hard shale 100 6604 Meigs Pomery 29-45 Shale (50%) with 60 905 (2) Mason interbedded siltstone (50%) 75-88 Siltstone.L. (2002). L.. either by altering the molecular film of water or by destroying the surface tension as all air spaces become filled with water. and Liang. certain clay minerals have the ability to absorb 33 County . Robinson Engineering of Ohio Co..2. J. 2. porewater pressure in the silty phases of the rock may spread the grains apart so that internal friction is further diminished.2. The shearing strength of shale is quite low. such as shear strength and durability.conducted a series of triaxial tests on Ohio grey shales and concluded that the transverse isotropy is an appropriate model to simulate the stress-strain relations of these shales. deriving from small internal friction of microscopic platy grains as well as cohesion between particles. R. hard 56 4611 Meigs Pomery (2) Mason Grey shale 2100Noble Sargand and 3125 Hazen(3) Chesapeake Sargand and Grey shale 1400Hazen(3) 2800 (1): Nusairat. and University of Akron.4 Effect of Water on Shales Water plays an important role on engineering properties of shales.

they range from 80% to 98% calcium carbonate. 1941). with minor quantities of other basic carbonates. under the same treatment. and almost a dozen flaked badly on the bedding surfaces in the same span of time. A limestone is a rock composed essentially of calcium carbonate. 1968) Fisher et al. Dolomite may be 34 .3 Limestone and Dolomite The dolomites and limestones of western Ohio form the bedrock in approximately 20. Three red shales became oozes within only ten minutes. With three exceptions.2. and eleven green shales were tested. some slaking within five hours. were essentially unaffected. and two-thirds eventually became oozes covering the bottoms of the beakers. but most not deteriorating at all. CaCO3.water and swell. and of siliceous and clay-like materials. sulphides. (1968) conducted slake tests. phosphates. Dolomite is the prominent carbonate rock throughout the stratigraphic section of western Ohio. eight gray shale samples. on various Ohio shales cored from southeastern Ohio. the red shale samples deteriorated completely. (Fisher et al. in which equal size cubic shales were immersed in water and observed during a period of three and one-half days.. almost all slaked within one hour into chunks one to two mm in diameter. The limestones are deposited in shallow or comparatively shallow water and are the result of precipitation either by direct exchange or by the agency of organic life. Twenty three red shale samples. and silicates.381 square miles or in nearly one-half of the total area of the State (Stout. The eight samples of gray shales. 2. a few slaking slightly. In general. Such strata constitute more than 60% of the whole. thus exerting an upward pressure that may reduce shear resistance within a claystone. sulphates. The eleven green shales varied widely in behavior.

Its chemical composition may be written as MgCa(CO3)2. 2. and metamorphic. they were laid down in the form we now find them and are not secondary. It may contain inter. For 35 . that is. The dolomites marine in origin. were deposited in shallow or comparatively shallow water. deposited as limestone or limy sediment. These characteristics generally indicate the strength. 2. deformability. 1993).2 Features for Rock Characterization The difficulties in making predictions of the engineering responses of rocks and rock mass derive largely from their discontinuous and variable nature. rock is distinguished from other engineering materials by the presence of inherent discontinuities.defined as a rock composed largely of magnesium carbonate combined directly with calcium carbonate in the molecular ratio of one to one. sedimentary. Most certainly. The response of intact rock material itself may be complex and difficult to describe theoretically because the rock consists of an aggregation of grains of material having quite different physical properties. 1941).3. In fact. Qualification and quantification of in-situ rock masses are some of the most important aspects of site characterization for design of foundations. Table 2-2 presents a simplified rock classification system and common rock types.and intra-granular micro-cracks and may have anisotropic and/or nonlinear mechanical properties (Brown. and stability of the rock masses. and in the main are direct precipitates. which may control its mechanical behavior.3 Mechanical Characteristics of Rock and Its Classifications 2.1 The types of rocks There are three basic types of rocks: igneous. then changed to dolomite (Stout.3.

quartzite. slate Hornfels. pumice. soapstone Phyllite.004 to 0. calcite) Halite minerals Sulfate minerals Iron-rich minerals Siliceous minerals Organic products Structure (foliated or massive) Foliated Massive Foliated Massive Foliated Massive Gneiss Metaquartzite Rock types Tuff (lithified ash) and volcanic breccia Obsidian (glass). it is not feasible to fully measure the characteristics of a complex rock masses. andesite. may influence the engineering response of rock masses in a variety of ways as enumerated below: 36 . anthracite coal Discontinuities.062 mm Clay-size grains.. one of the major effects. and boulders) Sand-size grains.0062 to 2 mm Silt-size grain. cobbles. 0.004 mm Carbonate minerals (e.g.economical reason. less than 0. The key factors influencing the rock mass behavior are listed in Table 2-3. gravel. serpentine. Table 2-2 Simplified Rock Classification Common igneous rocks Major division Secondary divisions Volcanic explosion debris (fragmental) Lava flows and hot siliceous clouds Lava flows (fine-grained texture) Dark minerals dominant Intermediate (25-50% dark minerals) Light color (quartz and feldspar) Texture (grain size) or chemical composition Grain size large than 2 mm (pebbles. and rhyolite Gabbro Diorite Granite Rock types Conglomerate (rounded cobbles) or breccia (angular rock fragments) Sandstone Siltstone Claystone and shale Limestone Rock salt Gypsum Hematite Chert Coal Rock types Extrusive Intrusive Major division Common sedimentary rocks Common metamorphic rocks Clastic rocks Chemical and organic rock Major division Coarse crystalline Medium crystalline Fine to microscopic Schist Marble. and scoria Basalt. 0.

repetitive (fatigue). and chemical environment of stress or impact strain Specimen size and shape Pattern of loading: constant load. slip or the gravity fall of blocks or wedges) • Influencing the stress distribution within the rock masses mainly because of their low stiffness and strengths • • • Attenuating. physical Specimen and State of Stress Method of Loading Environmental Conditions or Strain Moisture content. and poreapplied stress compressive. Magnitude of Type of loading: temperature. pulse 37 .• • • The provision of planes of low shear strength along which slip might occur Reducing the overall shear and tensile strengths of the rock masses Rendering the overall mechanical response of the rock masses in the sense that individual blocks may be free to rotate or to translate with associated slip and/or separation at block interfaces • Introducing a wide range of potential failure mechanisms (such as unraveling. reflecting and refracting stress waves arising from blasting Controlling to a large extent the fragmentation achieved by excavation processes Providing major conduits for the flow of fluids through most masses due to their permeability is orders of magnitude higher than that of intact rocks Table 2-3 Factors influencing Rock Mass Behavior Characteristics of Rock Rock material structure: lithology. toppling. cracks Rock mass structure: discontinuities Properties: mechanical. shear pressure or strain Rate of loading: Groundwater conditions Distribution static. dynamic.

described by the dip direction and dip of the line of steepest declination in plane of discontinuity 2) Spacing: perpendicular distance between adjacent discontinuities. Termination in solid rock or against other discontinuities reduces the persistence 4) Roughness: Inherent surface roughness and waviness relative to the mean plane of a discontinuity. normally refers to the mean or modal spacing of a set of joints 3) Persistence: discontinuity trace length as observed in an exposure which may give a crude measure of the area extent or penetration length of a discontinuity. clay. 38 . 1981): 1) Orientation: attitude of a discontinuity in space. Both roughness and waviness contribute to the shear strength. in which the intervening space is air or water filled 7) Filling: material that separates the adjacent rock wall of a discontinuity and that is usually weaker than the parent rock. silt. Typical filling materials are sand. Large scale waviness may also alter the dip locally 5) Wall strength: equivalent compressive strength of adjacent rock walls of discontinuity may be lower than rock block strength due to weathering or alteration of the walls. it is necessary that the discontinuities in rock masses should be characterized.2. The International Society for Rock Mechanics selects ten parameters as being of the primary importance for quantitative description of discontinuities in rock masses (Brown.3 Rock Mass Classifications In order to evaluate influence of discontinuities in rock masses for a given rock engineering project. 6) Aperture: perpendicular distance between adjacent rock walls of a discontinuity.3.

South Africa (Bieniawski. 1974). joint alteration (Ja). Rock Mass Quality (Q) system of Norwegian Geotechnical Institute (Barton et al. such as slope and foundation. quartz and calcite veins 8) Seepage: water flow and free moisture visible in individual discontinuities or in the rock mass as a whole 9) Number of sets: the number of joints comprising the intersecting joint system. gouge and mylonite. they have been used for other engineering applications. All these classification systems were developed initially for tunnel or dam applications. Also includes thin mineral coatings and healed discontinuities. 2.3. The rock mass may be further divided by individual discontinuities 10) Block size: rock block dimensions resulting from the mutual orientation of intersection joint sets. and Geological Strength Index (GSI) system (Hoek. 1994). the details to quantify these parameters will not be discussed here. The rock mass quality (Q) is defined as: 39 . number of joint or discontinuity sets (Jn). three different rock mass classifications.3.. 1976 and 1989). have been proposed in literature.1 Q System Q system (Barton et al. and resulting from the spacing of individual discontinuities may further influence the block size and shape. joint roughness (Jr). e. water flow (Jw) and a stress reduction factor (SRF). 1974) is based on six parameters: RQD. Rock Mass Rating (RMR) system of the Council for Scientific and Industrial Research.breccia. Because this system is mainly used for tunneling applications in Europe.. To quantify the effect of discontinuities of rock masses.g. However.

Ground water inflow The RMR76 value equals to the summation of the six ratings.3. The other items are kept same. Uniaxial compressive strength of intact rock 2. The RMR89 value also equals to the summation of five ratings as presented in detail in Table 2-5. however.3. 40 . RQD  J r Q=  J  J   n  a  J w    SRF    (2-39) 2. In 1976 version. Strike and dip orientations of joints 5. the major factors used to quantify the rating are: 1. RQD 3. Spacing of joints 4.2 RMR System There are two versions of RMR system: one is developed by Bieniawski in 1976. Condition of joints 6. Details of the rating system are presented in Table 2-4. The 1989 version of RMR system combines orientations rating and condition of joints rating into one rating as condition of discontinuities. the values of the ratings are changed. the other one is proposed by Bieniawski (1989) which is a modified version of 1976 RMR system.

Table 2-4 Classification Parameters and Their Rating (After Bieniawski.1 mm Not continuous 15 None 6 10 8 5 Rock Mass Classes and Other Rating I II III IV Very good Good rock Fair rock Poor rock rock 100-90 90-70 70-50 50-25 41 . 1976) Uniaxial compressive 1 strength of intact rock Rating Drill core quality RQD 2 Rating 3 Spacing of joints Rating Strike and dip orientations of joints Rating Condition of joints Rating Ground water inflow (per 10m of tunnel length Rating Class No.3-1 m 20 Fair 10 Tight: <1 mm and continuous No gouge 10 <25 liters/min < 25 MPa 0 <25% or highly weathered 3 < 50 mm 5 Very unfavorable 3 Open >5 mm Continuous Gouge>5 mm 0 >125 liters/min 2 V Very poor rock <25 4 5 Very tight: separation<0. Description Total rating > 200 MPa 10 90%100% 20 >3m 30 Very favorable 15 100200 MPa 5 75%90% 17 1-3 m 25 Favora ble 13 25 –50 MPa 1 25%-50% 8 50-300 mm 10 Unfavorab le 6 Open: 1-5 mm Continuou s Gouge<5 mm 5 25-125 liters/min 50-100 MPa 2 50%-75% 14 0.

Table 2-5 Rock Mass Rating System (Bieniawski. 1989) 42 .

If Bieniawski (1976) RMR76 is used. GSI = RMR76 GSI = RMR89 . If Bieniawski (1989) RMR89 is used. GSI can be converted from Rock Mass Rating (RMR) (see Hoek and Brown. 1997). The details for estimating GSI value are presented in Table 2-6 to Table 2-10. Therefore. GSI system was proposed by Hoek (1994) to replace the RMR rating in developing his rock strength criterion. Additionally. the physical appearance of material recovered in the core can be used to estimate a GSI value. Consequently. the value of 15 should be assigned to groundwater rating and adjustment for discontinuity value is set to zero. GSI values are based on the structure of rock mass and the surface condition.2.5 (2-40) (2-41) 43 . and adjustment for discontinuity orientation value should be set to zero. the following relationship can be established.3. the rock mass should be assumed to be completely dry and a rating of 10 could be assigned to the groundwater.3 GSI System The RMR values are difficult to estimate for very poor rock. especially for borehole cores that contain relatively few intact core pieces longer than 4 inch (Hoek. 1994).3. For cored weak rock.

Table 2-6 GSI Ranges for General Rocks (Marinos and Hoek, 2000)

44

Table 2-7 GSI Ranges for Typical Sandstones (Marinos and Hoek, 2000)

45

Table 2-8 GSI Ranges for Typical Siltstones, Claystones and Clay Shales (Marinos and Hoek, 2000)

46

Table 2-9 GSI Ranges for Typical Limestones (Marinos and Hoek, 2000)

47

Table 2-10 GSI Estimates for Heterogeneous Rock Masses Such as Flysch (Marinos and Hoek, 2000)

48

1. Robinson Engineering of Ohio Co. An existing weak rock p-y criterion proposed by Reese (1997) is evaluated against the two actual lateral load tests results. this chapter also exams the accuracy of various existing methods for deriving p-y curves from lateral load test results. The lateral load test results. The two tests were designed and conducted by E.CHAPTER III LATERAL LOAD TEST RESULTS AND DATA ANALYSES This chapter presents two lateral load tests on drilled shafts socketed in rock. Hypothetical cases are used to further verify the findings from actual field load test data. The subsurface investigation and lab test results on rock core samples are provided. The information pertinent to test drilled shafts and instrumentation are also presented. including deflections at shaft head at various loading levels. To obtain the most accurate experimental p-y curves from strain gage readings.1 Lateral Load Test at Dayton 3.1 Test Site The lateral load test was performed for assisting the design of a new two span reinforced concrete rib arch bridge as a replacement to an existing single span steel 49 . the shaft deflection profiles along depth of shaft.L. and strain gage readings are presented. particularly in terms of strain gage readings. 3.

silt and clay. weakly calcareous. Ohio. It is broken to very broken.5 feet and 8 feet. respectively. were drilled in the approximate location of the lateral load test. Borings B2 and B-4 encountered auger refusal at depth of 13. Two borings. The lateral load test drilled shafts were fully constructed in the gray shale interbedded with limestone after excavation at the site. The fill soils were underlain by natural soils consisting of sandy silt. In general. The underlying bedrock was reported to be limestone and shale of the Richmond Formation. The limestone interbeds are typically less than 1 foot thick. B-2 and B-4. 3-1. with the boring B-2 closer to the test site.5 feet and 40 feet. The gray shale is slightly weathered to decomposed. both the Illinoian and Wisconsin glaciers covered the site. respectively. and very thinly laminated.5 feet and 24 feet. Then. The possible lowest and largest RMR89 ratings of the bedrock at the test site are estimated based on the rock properties 50 . The bedrock encountered in the borings consists of soft to medium gray shale interbedded with hard gray limestone. The residual angle of internal friction of 24º was obtained by shearing the sample along a bedding plane. Another sample of the massive gray shale from boring B-4 was subjected to an unconfined compression test and was found to have an unconfined compressive strength of 5668 psi. As shown in Fig. Glacial deposits in the area are generally less than 30 feet thick. The elastic modulus of the gray shale is 590 ksi based on the unconfined compression test results. becoming massive near the completion depths of the borings. borings B-2 and B-4 encountered fill soils to depths of 3. the borings were cored to depths of 28. One direct shear test was performed on a representative sample of the gray shale obtained from boring B-4. respectively.bridge where East Siebenthaler Avenue crosses the Stillwater River in Dayton. and silty clay.

(4): Spacing rating.3 738. Boring B-2 SPT/RQD Ele.6 741. interbedded with limestone 12. 27 52 Max.3 48 730.3 Figure 3-1 Soil and rock layer profiles at Dayton test site Table 3-1 RMR Ratings and GSI Values of Rock at Dayton Test Site Elevations (2) (ft) 721-728 4 3 . (5): Discontinuity rating. These ratings are shown in Table 3-1.1 Fill 20 743.5 5 . broken. The GSI values in the table were obtained by correlations with RMR89 using the equation GSI = RMR89 – 5 and by setting the water rating as 15.presented in Fig.20 13 0 (3) (4) (5) (6) Min.7 5 .25 0 (2): Strength rating.6 Sandy silt 56 Boring B-4 Ele.4 Medium hard 11% gray shale. 0% broken.3 Fill Loose sandy silt SPT/RQD 16 7 739. (ft) 747.1 733.5 feet Soft gray shale. Mean 54 70 40. Mean 44 60 30.1 Stiff silty clay 85 Broken limestone 25% Stiff silty clay 723.5 51 Min.5 61 RMR89 GSI 728-739 4 . (6): Ground water rating 51 . (3): RQD rating.6 707. weakly calcareous. 17 42 Max. interbedded with hard gray limestone 0% 8% 0% 53% 718.1 Silt and 15 clay 738. 3-1. thinly laminated. (ft) 747.15 20 .7 735.6 736.15 5 .

In total. 3-2 shows the layout of the two test shafts. The shaft #4 was instrumented with 10 vibrating wire strain gages along the depth of shaft for monitoring the rebar strains. as shown in Fig. Fig. followed by an unloading. The 28-day compressive strength of concrete was around 4500 psi. three and two dial gages were installed for measuring the deflections at the top of shaft #4 and shaft #3. The loading sequence consisted of applying the lateral load in increments of 50 kips or 100 kips. The embedment length of the two drilled shafts was 18 feet. The shaft was reinforced with 36 #11 primary and #6 spiral bars. Each load was held until the rate of deflection at the 52 . A load cell was installed between shaft #4 and the jack to measure the actual applied lateral loads. The strain reading will be used to determine the shaft moments and soil reactions at each gage elevation. The maximum load was 1126 kips. 3-2. The elevation of the top of the drilled shafts and the excavated ground surface was 721 feet. The test site was excavated to the depth so that the two drilled shafts were fully constructed in shale without any overlying soils. The gages were placed at a depth interval of either 2 feet or 3 feet. Both drilled shafts were instrumented with inclinometers for measuring the deflection of the shaft with depth and dial gages for measuring the deflection at shaft head. A CR10X Campbell Scientific Data Logger was used to collect strain gage readings during lateral loading.1.2 Test Setup and Procedure Two 6 feet diameter drilled shafts were constructed and tested under lateral loads at Dayton site. respectively. The center-to-center distance between two test shafts is 18 feet. The lateral load was applied by pushing the two drilled shafts apart via a jack and reaction beam placed between them.3.

1. The relatively similar response of the two shafts indicates the spatial homogeneity of the rock mass.1 Load-Deflection Curves at Shaft Top Fig.04 inch/min and the inclinometer reading of each shaft was taken. and strain gage readings at each load level. There was one cycle of lateral 53 .1. Dial Gage Load Jack Cell 0 ft 2 ft Steel Plates Strain Gages 6 ft Steel Strut Dial Gage Inclinometer Casing Ele.3 Lateral Load Test Results The measured lateral load test results include the two load-deflection curves at the top of the two shafts. The deflections were averaged from the three dial gage readings for shaft #4 and two dial gage readings for shaft #3 at each loading level.top of shaft was less than 0. 3.3. deflection versus depth profiles measured from inclinometers at each load level. 721 ft 3 ft 5 ft 7 ft 10 ft 13 ft 36 #11 Rebar Shaft #4 Shale 36 #11 Rebar 18 ft Shaft #3 Figure 3-2 Instrumentation of load test at Dayton 3. It can be seen that the load-deflection curve behaves nonlinearly even at small deflections due to nonlinear response of weak rock masses. 3-3 presents the load-deflection response measured at the top of each shaft during incremental lateral loading.

For some loading levels.1.3. respectively.000 Shaft #3 Shaft #4 0.2 Deflection Profiles The deflection versus depth profiles of shaft #3 and shaft #4 deduced from inclinometer readings by assuming the deflection at the bottom of the inclinometer tube as zero are presented in Figs. the inclinometer readings are 54 .150 Deflection (in) Figure 3-3 Load-deflection curves at the top of shafts 3.100 0. It can be concluded that the cyclic loading would increase the lateral deflections.05 inch of permanent deflection at the top of shaft. 3-4 and 3-5. 1200 1000 800 Load (kips) 600 400 200 0 0. After unloading. both shafts registered 0.050 0. Lateral deflections were mobilized on most part of the two shafts along depth under lateral loads. Both shafts behaved like an intermediate shaft between a rigid shaft which exhibits linear displacement profile along the shaft length and a flexible shaft which displays fixity in the lower portion of the shaft. This resulted in a sudden increase of deflections of both shafts.load due to the loss of jacking pressure and back up at the load level of 700 kips.

06 0. 0.06 0.08 Deflection (in) 0.02 0 0 -0.08 0.1 Deflection (in) 0.14 0.04 0.12 0.1 0.04 0.12 0.02 0 0 -0. the deflections at the load of 173 kips were less than the deflections at the load of 93 kips as shown in Fig.02 Depth (ft) 5 10 15 20 92 kips 173 kips 286 kips 331 kips 427 kips 582 kips 634 kips 705 kips 786 kips 846 kips 930 kips 1126 kips Figure 3-4 Deflection-depth curves of shaft #3 0.considered to be unreasonable. For example. 3-4.02 Depth (ft) 5 10 15 20 92 kips 173 kips 286 kips 331 kips 427 kips 582 kips 634 kips 705 kips 786 kips 846 kips 930 kips 1126 kips Figure 3-5 Deflection-depth curves of shaft #4 55 .

3.1.3.3 Strain Profiles The strain readings were recorded by a data logger connected to a laptop. The compression strain and tension strain profiles of shaft #3 and shaft #4 under various loading levels are presented in Fig. 3-6 and 3-7, respectively. When the load was increased from 510 kips to 582 kips, a sudden increase of tension strain at the depth of 8 feet was noticed, as shown in Fig. 3-7. This could be due to concrete cracking in the shaft.

-250

-200 51 kips 92 kips 173 kips 286 kips 331 kips 427 kips 510 kips 582 kips 634 kips 693 kips 705 kips 786 kips 846 kips 1007 kips 1126 kips

Strain (micro) -150 -100

-50

0 0 2 Depth (ft) 4 6 8 10 12 14

Figure 3-6 Compression strain profiles of shaft #4

56

-150

-50 51 kips 92 kips 173 kips 286 kips 331 kips 427 kips 510 kips 582 kips 634 kips 693 kips 705 kips 786 kips 846 kips 1007 kips 1126 kips

50

Strain (micro) 150

250

350

450 0 2 Depth (ft) 4 6 8 10 12 14

Figure 3-7 Tension strain profiles of shaft #4

3.2 Lateral Load Test at Pomeroy-Mason 3.2.1 Test Site

The Ohio Department of Transportation (ODOT), together with the West Virginia Department of Transportation (WVDOT) and the Federal Highway Administration (FHWA) worked on design and construction of a replacement bridge for the U.S. Route 33 Bridge over the Ohio River between Pomeroy, Ohio and Mason, West Virginia. The new structure will be a three-span cable-stay bridge with a reinforced concrete superstructure and substructure. The cable stay towers are to be supported by groups of drilled shafts. Considering the large wind load and possible ship impact on the bridge piers, the lateral loads will be one of the key design loads for the bridge deep foundation. 57

Due to lack of competent design methods available for analyzing laterally loaded drilled shafts in rock, a lateral load test was carried out on two drilled shafts constructed for the West Virginia cable stay tower. Two Borings S-9 and S-10 near the test drilled shafts were advanced from a floating platform at the Station 29+95.5, 24 feet right and Station 30+18.7, 28.4 feet left, respectively. During soil boring, water surface elevations of 538.0 feet and 537.9 feet were observed, respectively. The water was 28.8 feet deep in S-9 and 28.3 feet deep in S10. In Boring S-9, from the river bottom to a depth of 38.9 feet (below the water surface), a layer of coarse and fine sand (A-3a) and fine sand (A-3) was encountered. This layer was described as grayish brown and orangish brown, wet and loose. Water contents varied from 20 to 24 percent and SPT N-values ranged from 4 to 8 blows per foot. Underlying the aforementioned soil in S-9 and at the surface in S-10, gravel with sand (A-1-b) was observed. This material extended to depths of 57.2 and 56.9 feet in Borings S-9 and S-10, respectively. The gravel with sand was described as brown, wet and loose to medium dense. Water contents ranged from 13 to 20 percent and SPT Nvalues varied from 4 to 16 blows per foot, with an average of 9 blows per foot. Bedrock was encountered below the soils described above at a depth of 57.2 feet below the water surface in S-9 and at 56.9 feet in S-10. The rock core recovered was described as shale with interbedded siltstone from bedrock surface to depths of 75.9 and 73.6 feet below the water surface in S-9 and S-10, respectively. Shale (mudstone) was observed in S-9 from 75.9 to 102.9 feet and in S-10 from 73.6 to 103.7 feet below the water surface. Siltstone was recovered underlying the shale to a depth of 123.3 and 118.4 58

feet below the water surface in S-9 and S-10, respectively. The borings were terminated at these depths. The weathered rock zone was observed in the uppermost 3 feet below the rock surface, approximately. The rock core samples were used for unconfined compression tests. The details of Borings S-9 and S-10 are provided in Fig. 3-8 and 3-9, respectively. The possible highest and lowest RMR89 ratings are estimated based on the rock properties indicated in Fig. 3-8 and 3-9. They are summarized in Tables 3-2 and 3-3 for S-9 and S-10, respectively. The top elevations of the rock are adjusted according to the drilling observations during the construction of drilled shafts. The GSI values correlated from RMR89 using the equation GSI = RMR89 - 5 and by setting the water rating as 15 are also provided in Tables 3-2 and 3-3. Table 3-2 RMR Ratings and GSI Values of Rock Based on Boring S-9 Elevations (2) (ft) 490.8 – 469 469 – 461.4 461.4-453.5 453.5-442.8 442.8-434 4 7 0 0 2 8 8 13 3 17 5 5 5 5 5 10-20 10-20 10 10 10 0 0 0 0 0 (3) (4) (5) (6) Min. Max. Mean Min. 27 30 28 18 34 37 40 28 18 34 32 35 28 18 34 37 40 38 28 44 Max. Mean 47 50 38 28 44 42 45 38 28 44 RMR89 GSI

(2): Strength rating; (3): RQD rating; (4): Spacing rating; (5): Discontinuity rating; (6): Ground water rating

59

Ele. (ft) 509.2

S-9 SPT/RQD 4 7 8 7 6 5 7 6 9 13 12 10 50+

Description Top of ground Coarse and fine sand (A-3a) and fine sand (A-3)

500.9

Gravel with sand (A-1-b) and traces of silt and clay, brown, wet, loose to medium dense Top of rock (the elevation of top of rock at the test shaft #2 was 490.8 feet observed during construction)

480.8 479.0 474 471.2 469.0 464 462.1 459.1 458 449.0 440.0

Weathered shale qu (psi) Ei (ksi) γ (pcf) Shale (50%) with interbedded siltstone (50%). Shale is gray to dark gray, laminated to very 3797 344.8 164.6 thin bedded, zones silty, soft to hard. Siltstone is gray to dark gray, thin bedded, zones sandy, micaceous, moderately hard to 44% 32% hard. 9073 1292 165.8 Cored sample was sandstone at this elevation. Shale (mudstone), greenish gray to brownish red, partially clay-like, blocky bedding, numerous slickensided fractures with degree of 50 to 65, soft. Cored samples for lab test were classified as claystone. Predominantly gray below 441.4’

51% 19 20% 44.3 86% 1.5 147 144.2

435.1 430.6 420.5

826 51% 1133 76% 567

81.2 157.9 Siltstone, gray to dark gray, thin bedded, 165.6 zones sandy, micaceous, hard 159.2 Figure 3-8 Boring S-9 60

Ele. gray to dark gray. thin bedded. brown. loose to medium dense 481. wet.4’ 434. thin bedded.6 S-10 SPT/RQD 12 4 7 6 8 7 6 11 16 14 13 50+ Description Top of ground Gravel with sand and (A-1-b) traces of silt and clay. zones sandy. zones silty. micaceous.7 468.3 449.9’ to 470.6 160. partially clay-like. micaceous. (ft) 509.85 149.4 160. Weathered from 481’ to 478’. hard .1 Figure 3-9 Boring S-10 61 Siltstone. zones sandy.7 464. greenish gray to brownish red. soft to hard. moderately hard to 60% hard.0 478.5 Top of rock (the elevation of top of rock at the test shaft #1 was 485.2 428.2 421.0 23% 18 148 145.7’. soft.3 459. Shale (mudstone). blocky bedding. 905 Siltstone is gray to dark gray.3 454.8 8% 7. numerous slickensided fractures with degree of 50 to 65.5 56% 4610.8 feet observed during construction) Weathered shale qu (psi) Ei (ksi) γ(pcf) Shale (50%) with interbedded siltstone (50%).0 441.8 20.4 38% 27.6 37% Predominantly gray below 441. laminated to very 162 thin bedded. Clay seam from 471. Shale is gray to dark gray.3 460.7 473.6 74% 4863. 52.

3 464.Table 3-3 RMR Ratings and GSI Values of Rock Based on Boring S-10 Elevations (2) (ft) 485. 62 .3-447.2 Test Set-up and Test Procedure The test drilled shaft #1 was constructed at Station 29+00 offset 33'-1" left. Max. respectively.8 feet and 490. The elevation of ground surface of shaft #1 was 511.8 feet. The distance between the top of the two test drilled shafts and the loading point during test was 3 feet. (4): Spacing rating. 30 18 18 23 40 18 18 23 35 18 18 23 40 28 28 33 Max.8 feet. The thickness of the soil layers at test drilled shaft #1 and shaft #2 was 26 feet and 18. the top elevations of bedrock at the test drilled shaft #1 and shaft #2 were observed as 485. respectively. Rock socket length was 40 feet and 56.8 2 0 0 0 13 3 3 8 5 5 5 5 10-20 10 10 10 0 0 0 0 (3) (4) (5) (6) Min. (5): Discontinuity rating.4 feet and 112.8 feet during construction. Mean Min.9 447. Therefore.9-445. respectively.8-464.3-457.8 feet. and the test drilled shaft #2 was located at Station 29+00 offset 38'-8" right. (6): Ground water rating 3. Mean 50 28 28 33 45 28 28 33 RMR89 GSI (2): Strength rating. (3): RQD rating. The total length of the test drilled shaft #1 and shaft #2 was 101.9 feet. Due to the variability of the site.2.3 457. the Boring S-9 is closer to the test drilled shaft #2 and the Boring S-10 is closer to the test drilled shaft #1.

The unconfined compressive strength of concrete was 5115 psi. The details of the instrumentation are provided in Fig. Two dial gages were installed for each test drilled shaft. The test drilled shaft #1 and #2 was instrumented with 13 and 10 levels of vibrating wire strain gages along the shaft length for monitoring the rebar strains. The maximum load applied was 275 kips. 3-10. The portion of the two drilled shafts above the bedrock has a diameter of 8. The reinforcement of the rock-socket portion was 28 #18 bars and the cover was 4 inch. Each load increment was held until the deflection at the top of shafts was stable and the inclinometer reading of each drilled shaft was taken. A load cell was installed between shaft #1 and the jack to measure the actual applied lateral loads. 63 . The equivalent modulus of drilled shafts in rock is 4250 ksi.5 feet with a 1 inch thickness of casing and 28 #18 bars. The diameter of the drilled shaft socketed in bedrock was 8 feet. an 11 feet diameter casing was used to form a gap between the test drilled shaft #2 and the soils above bedrock. This means all the lateral forces were resisted by the bedrock during the lateral load test. To fully mobilize the rock-shaft interaction and isolate the overburden soils. Both drilled shafts were instrumented with inclinometers for measuring the deflection along shaft length and dial gages for measuring the deflection at loading point. Two CR10X Campbell Scientific Data Loggers were used to collect strain gage readings during test. The testing loads were applied by tensioning a tendon connecting the two drilled shafts. The two shafts move toward each other during test.

8 470.3 445.4 475.4 460.3 475.8 487.Ele.5’ Ele.8 509.4 455.2.3 467.6 Top of rock 28#18 4’’ cover Strain gages 490. 64 .5’-OD casing 1’’ thickness Gap Top of soil 8.3 503.2 Jack Load cell 538 8.4 434 Inclinometer casing Test Shaft #1 8’ Test Shaft #2 Figure 3-10 Instrumentation and load test setup 3.3 485. deflection versus depth profiles measured from inclinometers.4 485.4 465.4 480.3 451.4 490.4 450.3 481.5’ Dial gages Working Deck 11’-OD casing 8.3 479.3 Lateral Load Test Results The measured data during lateral load test on the two drilled shafts include loaddeflection curves at the loading point.3 491. (ft) 546.4 445.3 483.3 499.3 457. (ft) 547. and strain gage readings along shaft length.3 495.9 3’ 507.8 8’ 511.

3-12 and 3-13.Fig. The strain readings were recorded by a data logger connected to a laptop during test. From Fig.73 inch. The compression strain and tension strain profiles of shaft #1 under various loading 65 . 3-12. This means that most of applied lateral load was resisted by the soils for shaft #1. 3-11 presents the load-deflection response measured at the loading point during incremental lateral loading. deduced from inclinometer readings by assuming that the deflection at the bottom of the inclinometer tube was zero. respectively.725 inch and 3. for shaft #2. The deflections were taken as an average of the two dial gage readings for each shaft at each loading level. it can be seen that the deflection at the bedrock elevation is very small. respectively. the lateral loads were resisted by the rock due to the use of casing to isolate soil reactions. are presented in Fig. 300 250 Lateral Load (kips) 200 150 100 50 0 0 1 2 Deflection (in) 3 4 Shaft #1 Shaft #2 Figure 3-11 Measured load-deflection curves at loading point of Pomeroy-Mason test The deflection profiles of test shaft #1 and shaft #2 along shaft length. It can be seen that the deflections at the loading point of shaft #2 is much larger than those of shaft #1 due to the use of casing in shaft #2 for isolation of soils. The maximum deflections at the 275 kips load level of shaft #1 and shaft #2 are 1. However.

elev.levels are presented in Figs. elev. elev. a sudden increase of tension strain accompanying a cracking sound was observed at the loading level of 175 kips. This could be due to concrete cracking in the shaft.5 2 Top of soil. respectively.5 1 Top of Shaft.) -0. When the load was increased from 175 kips to 200 kips. 485. 3-16. 544 ft 0 0. The maximum tension strain was recorded as 938 micros. as shown in Fig. respectively. elev. as sound of concrete cracking was clearly audible during test at this load increment. 3-14 and 3-15. 3-16 and 3-17. 511. 547 ft 1. Deflection (in. The tension and compression strain profiles of test shaft #2 are provided in Figs.8 ft Figure 3-12 Deflection-depth profiles of drilled shaft #1 at Pomeroy-Mason test 66 . a sudden increase of tension strain at the depth of 56 feet was noticed.8 at 25 Kips at 50 Kips at 100 Kips at 150 Kips at 175 Kips at 225 Kips at 250 Kips at 275 Kips Jacking.5 0 10 20 30 40 Depth (ft) 50 60 70 80 90 100 Top of rock. Similarly. The depth shown in figure starts from the top of the drilled shafts.

544ft at 25 Kips at 50 Kips at 100 Kips at 150 Kips at 175 Kips at 225 Kips at 250 Kips at 275 Kips Figure 3-13 Deflection-depth profiles of drilled shaft #2 at Pomeroy-Mason test 67 . elev. elev.8 ft 60 70 80 90 100 110 0 1 2 Top of Shaft. elev.) -1 0 10 20 30 40 Depth (ft) 50 Top of rock. 547ft 3 4 5 Jacking. 490.Deflection (in.

-100 35 45 55 Depth (ft) 65 75 85 95 105 0 Micro Strain 100 200 300 400 25 kips 50 kips 75 kips 100 kips 125 kips 150 kips 175 kips 200 kips 225 kips 250 kips 275 kips Figure 3-14 Tension strain profiles of test shaft #1 of Pomeroy-Mason test -250 -200 Micro Strain -150 -100 -50 35 45 55 0 50 25 kips 50 kips 75 kips 100 kips 125 kips 150 kips 175 kips 200 kips 225 kips 250 kips 275 kips Depth (ft) 65 75 85 95 105 Figure 3-15 Compression strain profiles of test shaft #1 of Pomeroy-Mason test 68 .

-300 55 60 65 70 Depth (ft) 75 80 85 90 95 100 105 0 300 Micro Strain 600 900 1200 1500 25 kips 50 kips 75 kips 100 kips 125 kips 150 kips 175 kips 200 kips 225 kips 250 kips 275 kips Figure 3-16 Tension strain profiles of test shaft #2 of Pomeroy-Mason test Micro Strain -300 -200 -500 -400 -100 55 60 65 0 25 kips 50 kips 75 kips 100 kips 125 kips 150 kips 175 kips 200 kips 225 kips 250 kips 275 kips 70 75 80 85 90 95 100 105 Figure 3-17 Compression strain profiles of test shaft #2 of Pomeroy-Mason test 69 Depth (ft) .

φ is the curvature of the shaft. 70 . y = ∫ (∫ φdz )dz (3-1) (3-2) (3-3) M = EIφ p=− d 2M dz 2 where y is deflections along depth of a drilled shaft. EI is section modulus of the shaft. and finally deriving p. Mathematically. determining the moment versus depth profile based on curvature versus depth profile.3 Methods for Deriving P-y Curves From Lateral Load Test Results 3. using double differentiating technique on the moment versus depth profile. soil resistance per unit length of a shaft. these three steps can be written as follows.3. M is shaft moment at depth z. Derivations of p-y curves using strain gage readings from fully instrumented lateral load tests usually involve three steps: deriving deflection versus depth profile using double integrating technique on the curvature versus depth profile of a drilled shaft from strain data. double integration of discrete data points (curvature data points) with respect to depth generally would not introduce significant numerical inaccuracy.3. Mathematically. which can be obtained as the difference between compression and tension strains measured at the same depth of the shaft divided by the distance between these two strain gages.1 Introduction Site specific p-y curves derived from instrumented lateral load tests are necessary for a situation when the site geological condition may differ from those that were used to derive existing p-y criteria. p is net soil resistance per unit length of the drilled shaft.

Additionally. such as high order global polynomial curve fitting (Reese and Welch. the existing methods for deriving p-y curves are summarized. these methods for deriving p-y curves from strain data have not been systematically compared and evaluated. Therefore. It is found that piecewise polynomial curve fitting technique provides the most accurate p-y curves. 1975. piecewise polynomial curve fitting (Matlock and Ripperger 1956). A procedure for determining an optimum strain gage spacing for planning instrumentation of a lateral load test is suggested. 1996). Reese and Welch. and smoothed weighted residuals method (Yang. 2005). 1970.However. Additionally. 1993) are used to further evaluate these derivation methods by comparing deduced experimental py curves with those assumed p-y curves. cubic spline (Dou and Byrne. weighted residuals method (Wilson. Comparisons of load-deflection curves of drilled shafts between measured values and predicted values based on those derived experimental p-y curves using various existing methods are performed. In fact.) and known moment profiles from COM624P (Wang and Reese. In this section. et al. several techniques have been proposed to minimize numerical errors due to double differentiation. there is a lack of guidance for best method in deducing p-y curves based on strain data obtained from load tests. Eight field lateral load tests on instrumented drilled shafts in soils and/or rocks are compiled to evaluate these methods. et al. four hypothetical cases based on assumed p-y curves from existing p-y criteria (Matlock. 1975). double differentiation of discrete curvature data points would generally result in amplification of measurement errors and consequently inaccurate p. a parametric study shows 71 . However. 1998).

1990) and the combination (y0. The 5th order polynomial function given in Eq. φ = a + bz + cz 2.5 + dz 3 + ez 4 + fz 5 (3-4) where a. Additionally. ytip=0) for a long pile (ratio of shaft length over diameter. is adopted herein to fit discrete curvature data points along shaft depth using the least-square method. it is found that the boundary combination (y0. The deflections can be obtained by double integrating the above fitted curvature versus depth profile. However. e. f are curve fitting constants. (3-4) has been successfully used by Wilson (1998) to fit the discrete strain points. yfixity=0) is not available. d. This technique. Two boundary conditions are required to determine two integrating constants involved in the double integration. because it was found from experiences that some inclinometers tend to provide smaller deflections of a shaft under larger lateral loads.3. θ0) can be used for a short drilled shaft when combination (y0. 3. Based on the evaluation (to be discussed later). the inclinometer readings could be used as an independent check on the accuracy of the deflections deduced from strain gage readings. The displacements from inclinometer readings are not directly used.that the errors of those derived p-y curves are mainly due to inaccurate determination of moment profiles from strain gage readings. yfixity=0) for a short pile (L/D < 10) would provide the best deduced deflections by comparing with inclinometer readings.2 Method for Deriving Deflection from Strain Gage Readings The lateral displacements of a drilled shaft along depth at each lateral load can be deduced from strain gage readings by using Eq. (3-1). c. according to Boghrat. b. therefore. L/D ≥ 10. combination (y0. 72 . The meanings of these boundary conditions are as follows: y0 is the measured deflection at the ground line.

3.3 Determination of Moment Profiles The nonlinearity of shaft stiffness is usually represented by M-φ relationship. the relationship between M and φ can be obtained from theory of reinforced concrete. In this case. The problems associated other combinations are observed during the evaluation as follows. as was done by Reese and Welch (1975). combination (y0. θ0) can be used for a short drilled shaft when a fixity point is not known. θtip=0) may result in deflections at the ground line to be different from the measured shaft deflections. yfixity=0 means that deflection is set to zero at a fixity point where the deflection of the shaft is close to zero according to the inclinometer readings. Combination (y0. When a lateral load test is performed with a large loading eccentricity (distance between the loading point and ground line). which is deemed unreasonable. a pair of strain gages can be mounted at the ground line level to obtain a representative M-φ relationship of the shaft. ytip=0 means the deflection at the tip of a drilled shaft is assumed to be zero. However.θ0 is the measured shaft tilt at the ground line.3. yfixity=0) may result in large negative deflections at the tip of a long drilled shaft. Combination (y0. it may not always available. Combination (ytip=0. θ0) provides none zero deflections at the tip of a long drilled shaft which may affect accuracy of shaft-head slope measurement. A similar conclusion was reported by Dunnavant (1986). Although it is preferred to use field measured moment-curvature relationship. which can be inferred from field data by comparing measured values of φ to known values of M at various loading levels. nevertheless. where stress-strain curves for both 73 .

4.3. where piecewise cubic polynomial with a window of 5 points is employed to fit measured moment data using the least-square technique.1 Piecewise Polynomial Curve Fitting Both Matlock and Ripperger (1956) and Dunnavant (1986) have used piecewise cubic polynomial function to fit discrete moment data. which avoids the requirement to capture the global trend of scattered data in global polynomial curve fitting approach. this method requires minimum of five discrete data points along shaft length. 1993). The piecewise polynomial curve fitting is a local curve fitting technique. However. The double differentiation of the local fitted polynomial curve with respect to the middle point yields p (the unit reaction of soil) at that point. The procedure described by Dunnavant (1986) is presented in Fig.3. The zero moment at the loading point or a known value of moment at the ground line should also be included in the moment profiles. respectively.reinforcing steels and concrete or predefined stress-strain models are used to compute Mφ curves (Wang and Reese. 3. 3-18. 74 .4 Methods for Deriving P (Net Resistance) 3. That is every 5 successive moment data points along the shaft length are fitted to one cubic polynomial curve. The p of upper three points and bottom three points is obtained from the smoothed local cubic polynomial moment curve using the top five points and bottom five points.

(3-5).5 rather than 2 on the quadratic term using least-square technique. provided the most reasonable p profiles. The value of p can then be obtained by double differentiating Eq. a five order polynomial with an exponent of 2.4.5 + dz 3 + ez 4 + fz 5 (3-5) The advantage of the method is its simplicity of application. However. M1 . The polynomial for the five M7=A7+B7z+C7z2+D7z3 Figure 3-18 Procedure for reducing moment data to p using piecewise polynomial (after Dunnavant. as shown in Eq. are used to evaluate p at the group center point.2 High Order Global Polynomial Curve Fitting Single high order polynomials have been used to fit moment profiles by Reese and Welch (1975) and Wilson (1998). such as M2.3.H M1=A1+B1z+C1z2+D1z3 M2=A2+B2z+C2z2+D2z3 (1) Least-square technique is used to fit cubic polynomials to groups of give contiguous moment values. 1986) 3. The non-integer form of polynomial will yield a zero soil resistance at the ground surface assuming depth z is zero at the ground surface. (3-5) with respect to z. According to Wilson (1998). (2) The first polynomial. Other polynomials. 75 . is z M3=A3+B3z+C3z2+D3z3 differentiated twice to evaluate p at the 3 moment levels closest to the surface (including the loading level). M = a + bz + cz 2. especially the moment data points near the ground surface. the technique is applicable only if the trend of moment profile can be captured.

This is commonly referred to as saying u(z) =a(z) “weakly”.3 Weighted Residuals Method Weighted residuals (WR) method was introduced to derive p from moment profiles by Wilson (1998). Then. rather it is a numerical differentiation method based on minimizing weighted residuals. p(z). As described by Wilson (1998). another differentiation on the derived shear force profiles using WR method results in the soil/rock reaction profiles.4. The main idea of WR method is to find an approximate function a(z) to represent the actual function u(z) over some interval z=0 to z=L. Generally. If f(z) represents the actual bending moment distribution of the shaft as a function of depth. and the difference can be defined as a(z)-u(z)=R(z). as is often used in finite element method. the shear force g(z) as the first derivative of the bending moment distribution is defined as: g(z ) = df (z ) = f ' (z ) dz (3-7) If g(z)=f’(z) “weakly”. WR method is used to obtain shear force profiles by differentiating the moment profiles one time. The WR method is not a curve fitting technique. While R(z) may not be zero anywhere in the range of z. L 0 ∫ R (z )Ψ (z )dz = 0 (3-6) where ψ(z) is a weighting function. Then. then f(z) is known at the nodes.3.3. u(z)≠a(z). the residual. the shaft can be considered as discretized finite elements with nodes at each location with measured bending moment value. a(z) can be selected such that R(z) is zero in an average sense by enforcing the following condition. then 76 .

L 0 ∫ {g(z ) − f ' (z )}Ψ (z )dz = 0 (3-8) where ψ(z) is a weighting function. linear “hat” functions as shown in Fig. The measured moment values inferred from strain gage readings at depth zi are f(zi). Substituting Eqs. Wilson (1998) developed a system of equations that can be used to solve for the values of g(z) at each node. The solution to those equations is provided herein. and ψ(z) is the shape function for node i. Both f(z) and g(z) are written as linear combinations of shape functions of finite element type. (3-8). This can be written as: f (z ) = ∑ f i Ψ (z ) i =0 n n . the bending moment between adjacent nodes i and i+1 can be expressed as follows: 77 . fi is the measured moment. According to the shape function expressed in Fig. and (3-9) g (z ) = ∑ g i Ψ (z ) i =0 (3-10) where i is the node number and ranges from 0 to n. 3-19. (3-9) and (3-10) into Eq. z is the depth along the shaft. 1 Ψ(z) 0 i-1 i General i+1 1 Ψ(z) 0 1 2 1 Ψ(z) 0 First Element n-2 n-1 n Last Element Figure 3-19 Linear shape functions For each node along the shaft. e. 3-19. weighting function ψ(z) is taken to be the shape function to generate a system of linear equations for the coefficients of f(z) and g(z).g.

zi = the depth of node i. g(zi) = shear force at node i. for the general elements. (3-12) can be obtained. The Eqs. the following equation can be obtained for the first element. (3-8). (3-9) and (3-10) into Eq.f (z ) = f (z i ) z i +1 − z z − zi + f ( z i +1 ) z i +1 − z i z i +1 − z i (3-11) Substituting Eqs. 2  z2 z0  (z1 − z 0 )3 g0 g1  1 + − z 0 z1  − − 2  2  (z − z )2 2 3 (z1 − z 0 )  (z1 − z 0 )2 1 0  f1 − f 0 z 0 z1 1 3  3  6 z1 − z 0 − 2 (z1 − z 0 ) = 0   ( ) (3-16) 78 . we have: {[ f (z n ) − f (z n−1 )] − [(z n − z )g (z n−1 ) + (z − z n−1 )g (z n )]}(z − z n−1 ) dz = 0 (z n − z n−1 )2 z n −1 z ∫ n (3-14) Also. (3-13) to (3-15) can be rewritten as follows. (3-12). one can obtain: L n  n  d ∑ f i Ψ (z )  i =0 Ψ ( z )dz = 0 ∫  ∑ g i Ψ (z ) − dz 0 i = 0       (3-12) By expanding Eq. the following equation based on Eq. {[ f (z1 ) − f (z 0 )] − [(z1 − z )g (z 0 ) + (z − z 0 )g (z1 )]}(z1 − z ) dz = 0 (z1 − z 0 )2 z0 z ∫ 1 (3-13) For the last element. {[f (z i ) − f (z i−1 )] − [(z i − z )g(z i−1 ) + (z − z i−1 )g(z i )]}(z − z i−1 ) dz + (z i − z i−1 )2 z i −1 z ∫ i z i +1 z ∫ i {[f (z i+1 ) − f (z i )] − [(z i+1 − z )g(z i ) + (z − z i )g(z i+1 )]}(z i+1 − z ) dz = 0 (z i+1 − z i )2 (3-15) where f(zi) = measured value of bending moment at node i in the drilled shaft. respectively.

2  (z n − z n −1 )3 = 0 f n − f n −1  z n z 2 −1 g n −1 z z gn  1 3  z − z 3 −1 − n −1 n (z n − z n −1 ) − + n − z n −1z n  − n 2 2 2  (z − z )2  6 n 2 2 3  (z n − z n −1 ) (z n − z n −1 )  n n −1   ( ) (3-17) f i − f i −1 (z i − z i−1 )2 gi   z2 z2 g i −1  i + i −1 − z z  − i −1 i   2 2 2  (z i − z i −1 )  z i −1z i 1 3  3  6 z i − z i −1 − 2 (z i − z i −1 ) −   ( ) (z i − z i−1 )2 (z i+1 − z i ) g i +1 2 (z i − z i−1 )3 3 + 2   z2 (z i+1 − z i )3 gi  i +1 + z i − z z  − − i i +1  2 3 (z i+1 − z i )2  2 (z i+1 − z i )2   f i +1 − f i z i z i +1 1 3  3  6 z i +1 − z i − 2 (z i +1 − z i ) = 0   ( ) (3-18) The above equation group that includes 3n equations can be expressed using matrices as follows: [A]{G} = {B}  a 00              a 01 L L a i i −1 L ai i L a i i +1 L L a n n −1             a nn    g 0   b0   g  b   1   1   M   M       g i−1   bi −1       g i  =  bi  g     i+1   bi +1   M   M    g n −1  bn −1    gn     b  (3-19) (3-20)  n  where a 00 = 1 (z1 − z 0 ) 3 (3-21) (3-22) (3-23) a 01 = b0 = 1 (z1 − z 0 ) 6 1 (f1 − f 0 ) 2 79 .

The WR approximation to derivative of f(z). however. The WR method has been coded into the software Matlab to obtain shear force profiles and soil/rock reaction profiles based on the measured values of moment. and can be applied a second time to obtain a piecewise linear approximation on the second derivative. and the solution {G} is expressed as: {G} = [A]−1 {B} (3-31) Note that the bending moment distribution f(z) is approximated as piecewise linear function using the weighting or shape function ψ(z). p(z). each element in coefficient matrix [A] and vector {B} can be determined based on the above equations. 80 . a double differentiation of moment profiles using WR method can result in the soil/rock reaction profiles. is piecewise linear. Thus.a i i −1 = 1 (z i − z i −1 ) 6 (3-24) (3-25) (3-26) (3-27) (3-28) (3-29) (3-30) a ii = 1 (z i+1 − z i−1 ) 3 a i i +1 = bi = 1 (z i+1 − z i ) 6 1 (f i+1 − f i−1 ) 2 1 (z n − z n −1 ) 6 a n n −1 = an n = bn = 1 (z n − z n −1 ) 3 1 (f n − f n −1 ) 2 As a result.

To deduce p-y curves.3. cubic spline curve fitting technique will not be examined. However. 3. weighted residuals (WR). A study by Yang. The cubic spline is perhaps the simplest interpolation function of discrete test data that can be double differentiated. depth profiles are first obtained by double integrating curvature vs. depth profiles using the 5th order polynomial curve fitting method.4 Smoothed Weighted Residuals Method An improvement on WR method can be achieved by using the 2D negative exponential smoothing function of commercial available software. In this dissertation. Sigmaplot. et al. it can interpolate as many data points as needed to make the spacing of moment data points smaller.3.5 Cubic Spline Curve Fitting A cubic spline was employed to fit the discrete moment data points to derive p by Mezazigh and Levacher (1998). and smoothed weighted residuals (SWR).3.3. Four methods for deriving p. the y vs. since a spline fits every point exactly. 3. 5th order global polynomial.5 Evaluation Using Field Test Data A total of eight field lateral load tests results on fully instrumented drilled shafts in soils and/or rocks have been compiled herein for evaluating the various methods for deriving p-y curves. The smoothing process can provide a smooth trend of measured moment data.4. including piecewise polynomial. (2005) has shown that this improved technique helps generation of smooth p-y curves.4. it is therefore prone to errors of measurement. Additionally. are evaluated. The relationship between moment and curvature is obtained by comparing measured curvatures at the ground surface level and the known applied moments for load 81 . to smooth the data points before WR differentiation is applied. A summary of these load tests is presented in Table 3-4.

Table 3-4 Compiled Lateral Load Test Database No. For other load tests. the deflections deduced from strain readings match the deflections from inclinometer readings. variability of concrete stiffness. CO Denver. and reinforcement. et al. (2004) Nusairat. (2004) This chapter Yang and Liang (2005) 3. OH Reference This chapter This chapter Nusairat.5 CDOTS1 2. CDOTS1). Extrapolation is used when curvature is greater than the maximum curvature measured at the ground surface. A representative comparison for test PomS1 is shown in Fig. For such load tests data (e.1 Deflection versus Depth The deflections deduced from strain gage readings are checked against the deflections measured by inclinometers. as illustrated in Fig.5. (2004) Nusairat. In general.7 16. 82 . 3-20.9 16. 3-21.3. CO Dayton. CO Denver.tests PomS1 and PomS2. some moment profiles are considered as outlier. OH Pomeroy.7 21 21 18 93 Location Pomeroy. concrete strength. CO Denver. Test Shaft 1 2 3 4 5 6 7 8 Diameter (ft) PomS1 8 PomS2 8 CDOTC1 2.. et al. et al.g. (2004) Nusairat.5 CDOTS2 2. and inaccurate moment-curvature relationships. M-φ curves are obtained using the LPILE computer program by inputting the drilled shaft geometry. Due to possible measurement errors. OH Denver.5 CDOTC2 2. OH Toledo. these outlier moment versus depth curves were discarded.5 DaytonS4 6 MaumeeS1 8 Total Length (ft) 101 112. et al.

30000 3 kips 25000 Moment (kip-in) 20000 15000 Outlier Profiles 10000 5000 0 0 50 100 Depth (in) 150 200 8 kips 13 kips 18 kips 25 kips 35 kips 45 kips 55 kips 65 kips

Figure 3-20 Elimination of outlier moment profiles
Deflections (in) -0.05 0 0 0.05 0.1 0.15 0.2

10

20 Depth (ft)

30 175 kips - Inclinometer 175 kips - From strains 150 kips - Inclinometer 150 kips - From strains 50 kips - Inclinometer 50 kips - From strains

40

50

60

Figure 3-21 Comparison of deflections from strain readings and inclinometer for test PomS1 83

3.3.5.2 P versus Depth Profile The derived p values near the ground surface from WR and SWR methods are negative for test data from MaumeeS1, CDOTS2, and PomS1, as shown in Fig. 3-22 for MaumeeS1. Similarly, the p values near the ground surface of PomS1 are also negative when global polynomial curve fitting method is used. Furthermore, SWR method results in irregular spikes in p vs. depth profiles for tests data of CDOTC1 and CDOTC2, as shown in Fig. 3-23 for CDOTC1. Based on the above observations, it is concluded that WR and SWR methods are not capable of producing reasonable p v. depth profiles for some of the load tests. On the other hand, the piecewise polynomial curve fitting method seems to be able to provide a reasonable result. 3.3.5.3 Load-Deflection Curve at Shaft Head The predicted load-deflection curves at the ground line (or shaft head) by inputting the deduced p-y curves into LPILE (or COM624P) program are compared with the measured. The method proposed by Murchison and O’Neill (1984) is employed to quantify prediction error. The prediction error is defined as the difference between the measured and predicted deflections at the ground line divided by the measured deflections at the same loading level, as shown in Fig. 3-24. Specifically, the errors at four loading levels (i.e., 0.25Hmax, 0.5Hmax, 0.75Hmax, and Hmax), are computed and summed. It should be noted that Hmax is usually taken as the maximum applied load; however, for tests, such as CDOTS1, where elimination of outlier data has resulted in the use of highest reasonable loads as Hmax.

84

The calculated cumulative errors for each load test data are summarized in Table 3-5. It can be seen that piecewise polynomial curve fitting method yields the smallest cumulative errors. If test CDOTS1 is ignored due to its inaccurate moment versus depth profiles, the average calculated error of the four loading levels of the remaining seven tests using the piecewise polynomial method deduced p-y curves is about 29%.

p (lb/in) -10000 0 -5000 0 5000

200

400 Depth (in)

600

800

1000

Polynominal WR SWR Piecewise

1200
Figure 3-22 P vs. depth profile of test MaumeeS1

85

p (lb/in) -10000 0 20 40 60 Depth (in) 80 100 120 140 160 180 Polynomial WR SWR Piecewise -5000 0 5000 10000 15000

Figure 3-23 P vs. depth profile of test CDOTC1 Load Hmax 0.75Hmax 0.5Hmax 0.25Hmax ∆y ∆y Errors= ∑∆y/ymea Measured ∆y

∆y

Predicted

ymea

Deflection

Figure 3-24 Definition of deflection prediction error 86

Table 3-5 Cumulative Shaft Head Deflection Errors based on Various Methods Test Shaft Polynomial PomS1 6.3 PomS2 0.9 CDOTC1 1.3 CDOTC2 1.7 CDOTS1 8.4 CDOTS2 0.5 DaytonS4 0.6 MaumeeS1 0.5 Total 20.3 3.3.5.4 Maximum Moment in Drilled Shafts In addition to deflection prediction, the maximum moments of drilled shafts under the four loading levels predicted using COM624P based on experimental p-y curves are compared with the corresponding maximum moments based on strain gage readings. The moment prediction errors, defined as the moment difference divided by the measured moments, are summarized in Table 3-6. It can be seen that the difference among these four methods is not significant with SWR providing the smallest error. Table 3-6 Cumulative Moment Errors based on Various Methods Test Shaft Polynomial PomS1 0.6 PomS2 0.3 CDOTC1 0.7 CDOTC2 0.8 CDOTS1 0.3 CDOTS2 2.4 DaytonS4 0.0 MaumeeS1 1.3 Total 6.3 WR 1.3 0.3 0.8 1.0 0.2 1.3 0.1 1.3 6.3 SWR 0.9 0.3 0.8 0.8 0.2 2.1 0.1 0.4 5.6 Piecewise 0.9 0.3 0.7 0.7 0.2 2.2 0.1 1.6 6.8 WR 6.3 1.0 1.4 2.3 5.2 1.4 1.2 3.8 22.6 SWR 2.7 0.8 2.0 1.9 9.9 0.4 0.5 1.0 19.2 Piecewise 3.0 0.9 0.7 1.9 8.5 0.5 0.4 0.5 16.5

87

02 20 Su – undrained shear strength. ε50 – strain at 50% of maximum principle stress. Table 3-7 Soil Parameters of Hypothetical Cases ε50 ks (pci) Case No.6 Evaluation Using Hypothetical Cases 3.052 0. The spacing of moment data points used for deducing p is prescribed as 12 inch.3. Three p derivation methods. one could then assess the accuracy of various p derivation methods. are included in this evaluation study. ks – coefficient of initial slope of p-y curves. if a small spacing of moment data points is chosen. because it yields almost the same results as WR method.072 0.3. φ – friction angle. 88 . By comparing the deduced p-y curves using various derivation methods with the prescribed (assumed) p-y curves. from which moment versus depth profiles under various lateral loads were obtained.072 90 4 Soft clay 2 0. and WR method. global 5th order polynomial.6. The soil used for each case is a homogeneous soil layer. The p-y curves could then be obtained by using y versus depth profiles from the simulation. γ' – effective unit weight. including piecewise polynomial. p-y criteria (COM624P) Su (psi) φ (º) γ'(pci) 1 Stiff clay above water 20 0. Four hypothetical cases are chosen for four types of p-y curves that could be prescribed in COM624P program. SWR method was not evaluated.1 Hypothetical Cases LPILE (or COM624P) was used with the prescribed (assumed) p-y curves to simulate results of a lateral load test. The reinforcement of the shaft is 12 #14.004 1000 2 Stiff clay below water 20 0. These moment profiles were then used to deduce p versus depth profile.035 0. The soil parameters for the selected four p-y criteria are summarized in Table 3-7. The drilled shaft used in these four hypothetical cases has a diameter of 3 ft and embedment length of 30 ft.3.004 1000 3 Sand 36 0.

8 89 . the p-y curve error is defined as ∆p/pprescribed.1 Piecewise 0.2 16.0 0.2 Comparison of p-y Curves The deduced p-y curves are compared with the prescribed p-y curves in Fig.3. It can be seen that the piecewise polynomial method yields the smallest error. 4 ft. It also shows that WR method provides a good result. It can be seen that the global 5th order polynomial derived p-y curves deviate significantly from the prescribed p-y curves. 1 2 3 4 Total Polynomial 4.5 95.3. where pprescribed is the p from prescribed p-y cruves and ∆p is the difference between the p from prescribed and the derived. and 10 ft) for each case are calculated and summarized in Table 3-8. The piecewise polynomial derived p-y curves closely match the prescribed p-y curves except for case 2. 3-25.4 17.5ym.0 3.4 1. To quantify the accuracy of these three methods investigated. where ym is the maximum y of prescribed p-y curves.5 4.75 ym. 7 ft.0 16.25ym.8 19.5 WR 1. the errors of the derived p-y curves are computed. The errors at deflections of 0. The cumulative errors of four p-y curves at four different depths (1 ft.6 83. Table 3-8 Cumulative Errors of p-y Curves of Hypothetical Cases Case No.5 0. are calculated for each p-y curve. 3-24. 0.6.6 0. while the global polynomial method leads to the largest error. 0. and ym. Using the same definition of error of load-deflection curves given in Fig.

WR Piecewise 1. on the other hand. which balances the cost and accuracy of the deduced py curves. An optimum spacing of strain gages.@12 in Figure 3-25 Comparison of p-y curves 3.2500 2000 p (lb/in) 1500 1000 500 0 0 5 y (in) 10 15 p (lb/in) COM624P Poly. A sparsely distributed strain data points may result in inaccurate p-y curves. WR Piecewise 4000 3500 3000 2500 2000 1500 1000 500 0 0 0.5 (a) Case 1 – @12 in 600 500 400 p (lb/in) 300 200 100 0 -100 0 2 4 y (in) 6 8 COM624P Poly. WR Piecewise (b) Case 2 – @12 in 250 200 p (lb/in) 150 100 50 0 0 2 y (in) 4 6 COM624P Poly.3. too closely spaced strain gage distribution my be cost prohibitive.7 Optimum Strain Gage Spacing The spacing of strain gages is important for deriving accurate p-y curves from lateral load tests. WR Piecewise (c) Case 3 – @12 in (d) Case 4 .5 y (in) 1 COM624P Poly. The 90 . is needed in planning of any instrumentation project for a lateral load test.

Finally. 6 ft. In this step. and 9 ft. Moment spacings varying from 1 ft to 7 ft are investigated. The first step is to select p-y criterion from the pool of existing criteria in LPILE program that fit to the soil condition at test site. The soil and shaft strength parameters are kept the same as those used for the hypothetical case 1. The cumulative errors of deduced p-y curves at four depths based on piecewise polynomial method for various chosen moment data point spacing is presented in Fig.following procedure is suggested for determining an optimum strain gage spacing for a particular lateral load test so that accurate p-y curves can be derived with minimum cost spent on strain gages. an example case with homogenous soil is employed. To account for the effect of drilled shaft diameter on optimum spacing. respectively. depth profiles of the pile at various loads using LPILE program with the prescribed p-y curves. 326. deduce p-y curves using the piecewise polynomial curve fitting method and LPILE generated moment vs. Next. 91 . 4 ft and 5 ft of spacing can minimize the errors of the deduced p-y curves and be taken as the optimum spacing for a drilled shaft with diameter of 3 ft. It can be seen that 3 ft. generate the moment vs. Then. The ratio of shaft length over diameter is kept as 10 for both additional cases. compare the deduced p-y curves with the prescribed p-y curves used in step one and determine an optimum moment (or strain gage) spacing accordingly. additional two cases are introduced with drilled shafts diameter of 6 ft and 9 ft. To illustrate the suggested procedure. different trial moment spacings should be used to obtain different sets of p-y curves. depth profiles.

3. a parametric study is performed. (3-2).8 Effect of Measurement Error D=3 ft D=6 ft D=9 ft Optimum Spacing To study the effect of error of measured moment profiles on the deduced p-y curves. the errors of moment-depth profiles from strain gage readings may be due to two possible reasons: the error of strain gage measurement and inaccurate estimate of shaft stiffness EI. The worst moment profiles could be appearance of wavy types of profiles due to unstable strain gage 92 .7 6 Cumulative Errors 5 4 3 2 1 0 0 1 2 3 4 5 6 7 Spacing (ft) Figure 3-26 Optimum spacing of strain gages It should be noted that above recommended method for determining optimum strain gage spacing is only valid when the moment vs. According to Eq. A lateral load test usually only mobilizes the deflection at top soil layers. depth profiles can be accurately determined from strain gage readings. it is preferred to use smaller gage spacing near ground surface and larger spacing for the bottom portion of the drilled shafts in order to obtain moment versus depth profile as accurately as possible. therefore. 3.

75 ym. M M M M Actual Moment profile z Type A z Type B z Actual Moment profiles z Type D Type C Figure 3-27 Four types of moment error profiles The hypothetical case 1 is used to investigate the effect of moment errors on the deduced p-y curves by modifying the moment profiles from COM624P. 10 ft.. 0. A diagram of these four possible measurement errors of moment is sketched in Fig. 3-28. over estimating EI before cracking. and 14 ft) are presented in Fig. 93 . It can be seen that the wavy type of moment error induces the largest error in the deduced p-y curves among these four types of moment error. As far as the errors due to shaft stiffness. three possible situations may occur. 0. The errors of moment are varied from 5% to 20%. The cumulative p-y curve errors of four deflections (i. The cumulative errors caused by other three types of error are about the same. and ym) at four depths (i. 2 ft. 6 ft. Two feet of spacing of moment data points is used.25ym. 0.. and over estimating EI at positions where crack appears. The p-y curves deduced from the modified moment profiles using piecewise polynomial method are compared with the prescribed p-y curves in the initial LPILE analysis.e. including under estimating EI.readings.5ym. 3-27.e.

On the other hand.0 25.0 0.2 when accurate moment profiles is used as shown in Table 3-8. 2. 3.9 Conclusions on Methods for Deriving p-y Curves Based on the work presented in this section.0 5. WR and SWR methods tend to yield unreasonable p vs.3 for 20% of moment error. the following conclusions can be made: 1.30. piecewise polynomial curve fitting method yields reasonable results.0 10.0 Cumulative Error 20.0 15. This implies that the errors of deduced p-y curves are mainly due to inaccurate determination of moment profiles from strain gage readings when an appropriate method is used to deduce the p-y curves. depth profiles for 5 out of 8 tests. The deduced p-y curves using WR method leads to the largest errors on the predicted 94 .0 0 5 10 Moment Error (%) 15 20 Type A Type B Type C Type D Figure 3-28 Cumulative p-y curve errors due to moment errors The cumulative errors of the deduced p-y curve for the four cases are greater than 3. Some of the problems are negative p values near ground surface and irregular spikes. which is much larger than the cumulative error of 0.3.

1 Evaluation of Experimental P-y Curves By using piecewise polynomial curve fitting technique. A procedure is outlined for determining optimum strain gage spacing for developing site-specific instrumentation plans. the p-y curves deduced by SWR method yields the smallest errors. 3-30. the p-y curves for shaft #4 of Dayton load test and those for shaft #1 and shaft #2 of Pomeroy-Mason load test at various depths are derived and presented in Figs. However.4. The piecewise cubic polynomial curve fitting method provides the smallest error of the deduced p-y curves for the 4 hypothetical cases. provided that an appropriate method for deducing p is used. and 3-31. 5. 3. On the other hand. 6. 7. The p-y curves deduced from the piecewise polynomial curve fitting method on the other hand yields the smallest error. Based on the evaluation studies presented. the difference of maximum moment prediction errors of all four methods is not significant.load-deflection curve of drilled shafts. respectively. piecewise cubic polynomial curve fitting method is recommended as a method for deriving p versus depth. 3. The parametric study suggests that the error of the deduced p-y curves are mainly due to inaccurate determination of moment profiles from strain gage readings. 3-29. 95 . Regarding the prediction of maximum moment of a drilled shaft. 4. 3. global 5th order polynomial curve fitting provides the largest error.4 Analyses of the Load Tests The lateral load tests presented earlier in this chapter are used to deduce p-y curves and to evaluate interim p-y criterion for weak rock developed by Reese (1997).

1 0.12 36 inch 132 inch Figure 3-29 Experimental p-y curves for shaft #4 of Dayton load test 6000 60 inch 5000 4000 p (lb/in) 3000 2000 1000 0 0 0.05 y (in) 0.03 0.06 y (in) 0.15 108 inch 252 inch 300 inch Figure 3-30 Experimental p-y curves for shaft #1 of Pomeroy-Mason load test 96 .09 0.18000 16000 14000 12000 p (lb/in) 10000 8000 6000 4000 2000 0 0 0.

3-32(a). the deflections were measured at the loading point. as shown in Fig. 97 . while the program LPILE requires the input load be applied at the shaft top.02 0.1 y (in) 6 inch 66 inch 126 inch Figure 3-31 Experimental p-y curves for shaft #2 of Pomeroy-Mason load test To verify the accuracy of these derived p-y curves. 3-32(b).06 0. For the load test at Pomeroy-Mason. is used to represent the actual applied lateral load. an equivalent load combination. to facilitate analysis using LPILE. as shown in Fig.04 0. including a same value of lateral load at the top of shaft and an additional moment. The effect of casing above rock layer at Pomeroy-Mason load test is considered in LPILE analysis. the loading point will be taken as the top of the shaft for the analysis using LPILE.08 0.30000 25000 20000 p (lb/in) 15000 10000 5000 0 0 0. therefore. Therefore. The nonlinear pile stiffness option is chosen to account for the reduction of stiffness of concrete shaft after the appearance of crack. an analysis using LPILE with these p-y curves was made to compare with actual measured load-deflection curves. The lateral load was not applied at the top of the shaft for Dayton load test.

the predicted deflections are less than the measured values at large load levels for shaft #2 of Pomeroy-Mason load test.15 Deflection (in) Measured Predicted Figure 3-33 Prediction of load-deflection curves of shaft #4 of Dayton load test using experimental p-y curves 98 . 3-34.The load-deflection curves at the top of shaft predicted by LPILE using the derived p-y curves are provided in Figs. Even though. In general. the predicted load-deflections curves match those measured curves. This provides validation that the recommended method for deriving p-y curves can be used to obtain reasonable site specific p-y curves. M=He H e Shale H Shale (a) Applied Load (b) Equivalent Loads Figure 3-32 Equivalent load combinations 1200 1000 Load (kips) 800 600 400 200 0 0 0. and 3-35 for the three test drilled shafts. 3-33.05 0.1 0.

5 1 Deflection (in) Figure 3-34 Prediction of load-deflection curves of shaft #1 of Pomeroy-Mason load test using experimental p-y curves 300 250 Load (kips) 200 150 100 50 0 0 1 2 Deflection (in) 3 4 Measured Predicted Measured Predicted 1.5 2 Figure 3-35 Prediction of load-deflection curves of shaft #2 of Pomeroy-Mason load test using experimental p-y curves 99 .300 250 Load (kips) 200 150 100 50 0 0 0.

0005 Claystone 0.0005 Shale(Mudstone) 0. qu (unconfined compressive strength).055 86% 0.0005 Claystone 100 .3 – 48 48 – 56.3. However. The values of Em are obtained by correlations with Ei.9 – 66 0 – 21.0 27.3 Mason 37. and krm (strain at 50% ultimate load). RQD.7 6170.0005 Shale/Siltstone 0.4 – 37.00005 of krm was suggested by Reese (1997).0005 Shale 0.05 38% 0.5 Mason 54.5 0. Em (modulus of rock mass).3 55.0005 Shale(Mudstone) 0.047 23% 0.049 8% 0. a value of 0.057 60% 0.8 3797 9073 19 44.9 63. A range of 0.047 20% 0.0005 to 0.3 17.5 – 63.1 0. Table 3-9 Rock Properties for LPILE Analysis Test Shaft Depths (ft) γ’ (pci) 0–7 7 – 18 0 – 17. the depth for shaft #1 at Pomeroy-Mason is from the ground level.2 Evaluation of Existing p-y Criterion The interim p-y criterion for weak rock proposed by Reese (1997) is evaluated by using LPILE analysis to predict field test data.059 44% 0. Because of the lack of measured krm.035 SPT =13.0005 Sandstone 0.5 Pomeroy – 47.9 – 29.035 SPT =7.2 krm Type 0. The depth is from the top of rock for shaft #4 at Dayton and shaft #2 at Pomeroy-Mason.0005 Shale k = 20 Sand k = 60 Sand 0.3 – 26 Shaft #1 26 – 47.4.049 51% 0.9 Shaft #2 21.0005 Shale 0.8 Shaft #4 Dayton RQD Em (psi) qu (psi) 38142 98102 6462 654 654 954 23885 102807 88.9 5668 5668 φ = 30 φ = 32 905 52.038 8% 0.5 – 54.0005 Claystone 0.3 826.8 7. Further discussion on this correlation is provided in chapter V. The main rock properties required by Reese p-y criterion are γ’ (effective unit weight).06 32% 0.4 Pomeroy – 29.0005 Shale(Mudstone) 0.0005 was used.038 53% 0. The main parameters need to represent the weak rock p-y curves using Reese criterion are tabulated in Tables 3-9 to 3-11 for the three test shafts.

respectively. 3-37. The comparisons of the predicted load-deflection curves and the measured curves are provided in Figs. This may impede full interaction of drilled shaft and rock. As shown in Fig. the load-deflection curves at the loading point can be obtained. good matches are obtained when the modulus and unconfined compressive strength of top rock layer of Dayton site and those of Pomeroy-Mason site are reduced to 2% and 1% of the original values provided in Table 3-9. 101 . However. One can see that a dramatic reduction of rock properties is needed to match the load test results. as evidenced in Fig. 3-36 and Fig. By inputting properties of rock.The information of test shafts used in LPILE analysis is summarized in Table 3-10. soil. the predicted deflections are much less than the measured values for the other two test shafts. the rock parameters are adjusted to match the measured results. 3-36 through 3-38. 3-38.10) into LPILE. it should be noted that a 26 feet thick layer of sand exists above the rock mass when shaft #2 was under test. Nonlinear shaft stiffness option is selected. and shaft (summarized in Table 3-9 and Table 3. The predicted deflections are larger than the measured values for shaft #1 of Pomeroy-Mason load test according to Fig. To explore the first reason. This may be due to inaccurate determination of input parameters or due to the unsuitability of Reese weak rock p-y criterion for the two tests. 3-12 where the deflection at the top rock almost is close to 0. On the other hand.

05 0.1 0.1 – 53.4 – 58.Adjusted Figure 3-36 Prediction of load-deflection curve of shaft #4 of Dayton load test using Reese weak rock p-y criterion 102 .1 34#11 28#18 with 1 inch thick casing 28#18 28#18 with 1 inch thick casing 28#18 Concrete Strength (psi) 4500 5115 5115 5115 5115 Shaft Dayton Shaft Pomeroy Mason Shaft Pomeroy Mason 1200 1000 Load (kips) 800 600 400 200 0 0 0.4 53.1 53.1 – 109.4 #2 0 – 53.9 8.Table 3-10 Test Drilled Shafts Information Test Shaft Depths (ft) Diameter start from (ft) loading point #4 0 .5 8 Loading Reinforcement Eccentricity (ft) 0 32.18 6 #1 0 – 58.4 32.15 Deflection (in) Measured Predicted Predicted .4 – 98.5 8 8.

Adjusted Figure 3-38 Prediction of load-deflection curve of shaft #2 of Pomeroy-Mason load test using Reese weak rock p-y criterion 103 .300 250 Load (kips) 200 150 100 50 0 0 1 2 3 Deflection (in) Measured Predicted Figure 3-37 Prediction of load-deflection curve of shaft #1 of Pomeroy-Mason load test using Reese weak rock p-y criterion 300 250 Load (kips) 200 150 100 50 0 0 1 2 Deflection (in) 3 4 Measured Predicted Predicted .

the interim weak rock p-y criterion proposed by Reese (1997) was evaluated as well. Based on an extensive evaluation of several existing methods. 104 . Due to lack of actual lateral load tests of drilled shafts in rock. the piecewise polynomial curve fitting technique is recommended for deriving p-y curves from strain data obtained during lateral load test. The following conclusions can be made. The accuracy of the deduced p-y curves using the recommended procedure was validated against the two load tests. The superior results of this recommended method was supported by the smallest error on the predictions of load-deflection curves among all other methods. The details of instrumentation used and load test procedures were also presented.5 Summary and Conclusions In this chapter. The recommended method for deriving p-y curves from strain data of load test includes using high order polynomial curve fitting to obtain y and using piecewise polynomial curve fitting to obtain p. 3. 4. 1.3. 2. The evaluation of the interim weak rock p-y criterion (Reese. Additionally. the two lateral load tests results presented here contribute to understanding and development of pertinent p-y curves. two lateral load tests results of fully instrumented drilled shafts at Dayton site and Pomeroy-Mason site were presented. The deduced p-y curves from actual load test data is shown to result in reasonable predictions of the lateral response of drilled shafts for the two load tests. 1997) indicates that this p-y criterion tends to under predict the deflections of drilled shafts in rock.

Finally. it is essential to identify failure modes of rock surrounding the laterally loaded drilled shaft. as pointed out by Carter and Kulhawy (1992). a 3-D finite element model (FEM) is developed to simulate the response of laterally loaded drilled shafts in rock. The load test result from Dayton site is employed to validate the FEM simulation techniques. For this purpose. deriving p-y curve criteria requires that theoretical values of ultimate resistance of rock. perhaps is that the lateral design is governed largely by displacement considerations. Both Hoek-Brown strength criterion of rock mass and an empirical equation for estimating the side shear resistance of shaft/rock interface are employed to derive the theoretical equations for compute the ultimate resistance of rock. reliable estimate of lateral capacity is still very important. stress and deformation fields around the shaft are used to develop failure modes of rocks surrounding laterally loaded shafts. Moreover. so that the margin of safety on capacity consideration can be ascertained.CHAPTER IV LATERAL CAPACITY OF DRILLED SHAFTS IN ROCK The ultimate capacity of laterally loaded drilled shafts in rock has received very little attention in literature. In order to develop theoretical equations for computing the ultimate capacity. 105 . Nevertheless. Based on calibrated FEM analysis techniques. The reason. a method for estimating ultimate lateral capacity of a drilled shaft in rock and/or soils is developed.

it is now possible to use 3D FEM modeling technique. In early stage of FEM applications. (1999) presented a 3D FEM analysis of laterally loaded piles in soils by modeling soil as elasto-plastic material and pile as linear elastic material. such as p-y analysis (Reese. 2000). the lateral load test result at Dayton site is employed to compare 3D FEM model predictions.1 3-D Finite Element Modeling and Validation Although.4. A 3D FEM model to simulate the response of a single drilled shaft under lateral loads in rock using ABAQUS is presented here. For example. To verify the FEM model. However. The FEM model could be used for deriving theoretical p-y curves as well as for modeling complicated drilled shaft group. Recently. The yielding of rock mass is modeled using modified Drucker-Prager model (CAP model). With the advances of FEM computing techniques. Wallace et al (2002) presented a 3D FEM analysis to simulate a lateral load test using ABAQUS where soils is modeled as with Mohr-Coulomb model. Randolph (1981) presented a 2D FEM solution by modeling soil as an elastic continuum and the pile as an elastic beam. 106 . there is a lack of finite element study on laterally loaded drilled shafts socket in rock mass. finite element method (FEM) nevertheless provides a powerful tool for simulating complete nonlinear and plastic behavior of rocks subjected to laterally loaded drilled shafts. Wakai et al. et al. 1997) and elastoplastic solution (Zhang. there are several analytical methods available for analyzing the response of laterally loaded drilled shafts in rock.

7L Figure 4-1 Finite element meshes of a drilled shaft-rock system 107 . which is depicted in Fig. such as 6-node triangular prism elements (C3D6) and 8-node brick elements (C3D8). 4-2 (b). because the trial simulation with these first order elements suffered convergence problem.1 Finite Element Meshes Fig.4. The drilled shaft is modeled using second order 15-node triangular prism elements C3D15. 4-1 shows the meshes of a single drilled shaft-rock system generated using software package PATRAN. The bottom of rock mass is fixed. 4-1. The rock mass is modeled using 8-node brick elements C3D8 shown in Fig. 4-2 (c). 4-2 (a).1. 5D L D 5D Drilled Shaft: C3D15 Infinite Element: CIN3D8 Rock: C3D8 Side View Fixed Bottom Top View 0. it is preferred to use a fine mesh for the top layer of rock mass. The second order elements are found to be more appropriate for modeling the shaft than the first order elements.1.1.1 3-D FEM Modeling 4. as depicted in Fig. The outer boundary of rock mass are modeled using 8-node infinite boundary elements CIN3D8. as shown in Fig. Because the lateral response of a drilled shaft is mainly dependent on the rock near the top of the shaft.

This is based on findings from Trochanis et al. The modified Drucker-Prager model (CAP model) was employed to model the rock mass.7 L. the various numerical convergence problems are prevalent. where L is embedment length of the drilled shaft. 11D (D is diameter of drilled shaft) could be used to define the dimension of the diameter of the entire mesh. The ABAQUS built-in CAP model is intended to model cohesive geological materials that exhibit pressure-dependent yield. The depth of rock mass beneath the drilled shaft tip is 0.1. such as soils and rocks. by studying the effect of different mesh sizes on the shaft-soil contact pressure. (2002) found that. 4.(a) C3D15 (b) C3D8 (c) CIN3D8 Figure 4-2 Finite elements for (a) drilled shaft.2 Constitutive Models The drilled shaft is modeled as a cylinder with linear elastic material property. however.1. ABAQUS built-in concrete model with embedded rebars was used initially in an effort to simulate the nonlinearity of reinforced concrete drilled shafts. It is based on the addition of a cap yield surface to the DruckerPrager plasticity model. (b) surrounding rock. which provides an inelastic hardening mechanism to account for 108 . and (c) outside boundary of rock Wallace et al. (1988).

the yield surface of the CAP model is defined in two principal segments: a pressure-dependent Drucker-Prager shear failure segment and a compression cap segment. The values of d and β can be calibrated from two uniaxial or triaxial compression tests. The default value of K is 1. and friction angle.778 ≤ K ≤ 1. β (material friction angle in p-t plane).0. The deviatoric stress measure t is defined as t= 1 1 1 r q[1 + − (1 − )( ) 3 ] 2 K K q (4-1) where 1 p = − trace (σ) 3 (4-2) (4-3) (4-4) (4-5) q = (3/2 S : S)0. as shown in Fig. 4-3. A nonassociated platic flow is used for the Drucker-Prager yield segment. The main input parameters for a CAP model are Em (elastic modulus of rock mass). φ using the following equations. they can be calibrated from traditional Mohr-Coulomb cohesion. According to ABAQUS User’s Manual (1998). and an associated plastic flow is used for the CAP segment.0. The Drucker-Prager failure segment is a perfectly plastic yield surface (no hardening). Alternatively. 1998). c. 109 . and 0.5 r = (9/2 S : S · S)1/3 S=σ+pI K is a material parameter that controls the dependence of the yield surface on the value of the intermediate principal stress. d (material cohesion in p-t plane). and a cap hardening curve which is obtained from a hydrostatic compression test.plastic compaction and helps control volume dilatancy when the material yields in shear (ABAQUS.

1998) 4. 1998). which needs an input of a coefficient of friction. In normal direction. The lateral loads are applied step by step after the gravity weight of the shaft-rock system is activated.3 Shaft-Rock Interaction and Loading The interaction between drilled shaft and rock mass is modeled using a surface based contact.d = 2c cos ϕ 1 − sin ϕ (4-6)  6 sin ϕ  β = arctan  3 − sin ϕ     (4-7) Deviatoric Stress Equivalent Pressure Figure 4-3 Cap model: yield surface in the p-t plane (ABAQUS. The shaft surface is treated as master surface.1. In tangential direction of the shaft-rock interface. 110 . There is no limit to the magnitude of contact pressure that can be transmitted when the surfaces are in contact (ABAQUS.1. the contact surfaces transmit no contact pressure unless the nodes of the slave surface are in contact with the master surface. the frictional interaction is simulated using linear Coulomb friction theory. while the rock surface is taken as slave surface.

The rock mass at the test site is gray shale interbedded with limestone. The details of the test shaft dimensions and rock properties are presented in Fig. φ.2 Validation of the 3D FEM Model 4. Ei. 2002) rock strength criterion using qu. slightly weathered to decomposed. The GSI and RMR values are also summarized. Mohr-Coulomb cohesion and friction angle of the rock mass. An unconfined compression test on an intact rock sample provided qu (unconfined compressive strength) of 5668 psi and Ei (elastic modulus of intact rock) of 590 ksi.1. and other secondary structures of rock..15. The elastic modulus and Poisson’s ratio of the drilled shaft are 3800 ksi and 0. and GSI. Drilled Shaft Load 7 ft Rock Soft gray shale. of the two layers of rock mass 111 .5 40.1 Load Test and Input Parameters The field lateral load test results from Dayton test is used to validate the above 3D FEM modeling technique.5 2 ft 11 ft 53% 51 61 6 ft Figure 4-4 Drilled shaft dimension and rock profiles of the load test at Dayton Based on the unconfined compression test. slightly weathered. 4-4. including the effects of joints. cracks.4. 2002). By using free software RocLab (Rocscience. c. Mohr-Coulomb cohesion. and friction angle. respectively. laminated with hard limestone RQD RMR89 GSI 8%0% 30. The test drilled shaft is 6 feet in diameter and 18 feet in rock socket length. laminated with limestone Medium gray shale. et al.2. fillings. can be correlated with Hoek-Brown (Hoek.1.

the entire side interface between the drilled shaft and rock mass is simulated using one contact modeling.01). The input pairs of the hydrostatic compression yield stress (pb. The Poisson’s ratio of the rock mass is assumed to be 0.5 is used based on the Mohr-Coulomb friction angle of 27º of the bottom rock layer. and friction angle. The elastic modulus of rock mass Em is also obtained using RocLab. 0). A coefficient of friction of 0. The Drucker-Prager cohesion. the stress-strain curve of the unconfined compression test is used to represent isotropic hardening curve.were determined and are summarized in Table 4-1.2. Due to the lack of hydrostatic tests on rock mass samples. Table 4-1 Rock Mass Properties Rock Layer Top layer Bottom layer Mohr-Coulomb c (psi) 154 251 φ (º) 20 27 Drucker-Prager d (psi) β (º) 440 38 819 47 EM (ksi) 241 590 4. To simplify the contact interaction. therefore. 0. the effective unit weight of 0. It is found that the mesh with 1885 elements provides almost the same results on lateral deflections of the drilled shaft under lateral loads as the mesh with 112 .1. and (1900.2 Simulation and the Results The mesh sensitivity is studied by comparing FEM results based on varying the number of elements. (1500. 0.004). which employs Hoek et al. in psi) and volumetric inelastic strain (εin) for the hardening curve are as follow: (1000. d. β are estimated using Eqs. and a second contact modeling is used to simulate the interaction between shaft tip surface and rock mass.034 pci is used in analysis. (2002) empirical correlation equation based on qu and GSI.3. (4-6) and (4-7). The water table was above the surface of rock mass.

6 0. as shown in Fig. 4-5. as 113 . in general. a good match of the load-deflection curve at the head of the shaft is obtained. With improved hardening curves.4 Deflection (in) 0. Although. the 3D FEM model predicted the deflection response of drilled shafts in rock under lateral loads. A back analysis is performed to match the entire range of lateral deflection response of the drilled shaft. 4-6. It can be seen that the predicted load-deflection curve is comparable to the measured. the FEM predicted load-deflection curve at the shaft head is compared with the measured one.8 1885 Elements 2765 Elements Figure 4-5 Mesh convergence With the constitutive model parameters described previously. it is believed that FEM simulation results are not sensitive to mesh density. as shown in Fig. This was attributed to the lack of a hydrostatic compression test to determine an accurate hardening curve for the rock mass. A good match of the deflection profiles along depth between the FEM predicted values and the inclinometer measured values is also evident.2 0. Therefore. 1200 1000 Load (kips) 800 600 400 200 0 0 0.2765 elements. as shown in Fig. 47. the early portion of nonlinear lateral response was not captured.

Match Figure 4-6 Comparison of load-deflection curves at shaft head Deflection (inch) 0 0.15 -0. (260.05 0.05 0. The cohesion of the top rock layer is reduced to 100 psi.02 0. 0. 0. 0).01). The hardening curve of the top rock layer is adjusted to the following pb (in psi) and εin pairs: (120.05 0.1 0.1 0 5 -0.15 Deflection (in) Measured ABAQUS ABAQUS .02 0 5 Depth (ft) 10 15 20 0 Deflection (inch) 0.08 Depth (ft) 10 15 20 Measured-1126 kips ABAQUS-1126 kips Measured-705 kips ABAQUS-705 kips (a) Load at 1126 kips (b) Load at 705 kips Figure 4-7 Comparison of deflection profiles at the load of (a) 1126 kips and (b) 705 kip 114 .04 0. 1200 1000 Load (kips) 800 600 400 200 0 0 0. 4-6.06 0.shown in Fig.02). and (400.

However. Generally.1. For each type of rock.1 Finite Element Modeling A drilled shaft with 30 in. two conditions are considered: one is that the drilled shaft is fully socketed into rock. diameter and 15 feet socket length is modeled using linear elasticity with elastic modulus of 4000 ksi. Attention here is given to the later type of failure. The elasticity modulus of rock is assumed to be 2000 ksi. and the other one is jointed rock with a set of parallel weak planes. the other one is that the drilled shaft socketed into rock with overburden soils.4. The Modified Drucker-Prager Model (CAP Model) is utilized to model the isotropic and homogeneous rock. 4. (2000) adopted Carter and Kulhawy (1992)’s assumption in their relevant theoretical work. Two types of rock will be investigated in FEM simulation of rock failure modes: one is intact rock or highly weathered rock that can be modeled as isotropic and homogeneous rock.2.2. and the Mohr-Coulomb cohesion and friction angle are 115 .1 Failure of Isotropic and Homogeneous Rock without Overburden Soils 4. Zhang et al. Reese (1997) assumed compressive failure of rock. Carter and Kulhawy (1992) assumed that compressive failure will occur for rocks in front of shaft and shear failure will occur for interface between shaft and rock. For jointed rock mass. rock-socketed drilled shafts under lateral loads may fail either due to structural failure of shaft or strength failure of rock. To et al. (2003) proposed a wedge type failure.2 Failure Modes of Rock Subjected to Loading from Laterally Loaded Drilled Shafts Very few research has been focused on the failure modes of rock mass surrounding the drilled shafts under lateral loads. but ignored shear failure between shaft and rock.

respectively.2. the rock in front of the shaft will move forward under the lateral loads. The element type C3D15 and C3D8 are used to model drilled shafts and rock. Fig. the rock in front of the shaft shows a forward and upward movement according to the deformed and undeformed rock mass shown in Fig. respectively. Therefore. The load applied to the top of shaft is increased from 300 kips to 3000 kips.1. the analysis results at 3000 kips of loading level. Deformed Undeformed Shaft Loading Direction Rock inch Figure 4-8 The forward movement of rock mass Apparently. at which time computer analysis failed to converge. 4-9 shows the 116 .assumed to be 1 ksi and 27 degree. 4-8. it is highly likely that without overburden soil layer on top of rock.2 Failure at the Top Layer of the Rock At the failure load of 3000 kips. 4. are used to examine the failure conditions in the rock.

4-8. the rock in vertical xz plane. Therefore. Rock mass in front of shaft Rock Figure 4-9 Front view of upward movement of rock mass 117 . the uniaxial tensile strength of rock is about one tenth of the uniaxial unconfined compressive strength of rock. has to reach tensile failure. In order to check whether the rock in xz plane surrounding the shaft can reach tensile failure. Usually. 4-8 and Fig. the stress distribution in rock is shown in Fig.front view of the upward movement of rock mass. From Figs. However. The magnitude of the maximum tensile stress is 2763 psi which is about half of the maximum compressive stress 5830 psi in front of the shaft. shown in Fig. in this case. It can be seen that the tensile stress is developed. and the inclined planes of the wedge have to reach shear failure. it can be seen that a wedge type of rock mass has the tendency to move away from its original position. for this to happen. the tensile failure is more likely to appear before the onset of compressive failure. in which the movement is scaled up to 150 times of the original magnitude of movement. 4-9. 4-10.

implies a possible failure mode shown in Fig. It can be seen that the distributions of shear stresses at three planes are similar. it can not be concluded from Fig. together with the depicted shear distribution. 4-11 that the rock will be sheared out. which means that the bottom surface of the wedge is likely to be a plane rather than a curved surface.Shaft Tensile Stress Rock Figure 4-10 Y direction stress distribution in xz plane Additionally. the upward and forward movements of this portion of rock. 118 . one needs to check if the shear stress is concentrated on shear planes. Fig. 4-11 shows the maximum shear stress distribution on yz plane. in order to confirm the proposed wedge type failure. 4-12. Although.

Shaft Rock Figure 4-11 Maximum shear stress distribution on yz plane σ’v0 θ Fsb Fnb Fs W β H Fnet Fa D Fn Figure 4-12 Proposed wedge type failure model for the top layer of rock 119 .

which is considered large enough to produce crack since the tensile strength of most rocks ranges from 725 psi to 2900 psi (Afrouz. Possible failures of in-depth rock can be tensile failure. is applied at the top surface of the shaft and rock. the FEA model is modified to simulate a 10 inch long crack at the tensile failure areas. Fig. The stress in loading direction is redistributed as shown in Fig. the tensile failure of the rock mass will occur first for in-depth rock layers. 1992). the maximum tensile stress in rock surrounding the shaft is about 42% of the maximum compressive stress. Even with initiation and propagation of cracks in rock.2. from which 120 .1. A separate FEM model was established in order to realize material failure of rocks. which is equivalent to vertical pressure at 6 ft deep in rock mass. As can be seen in Fig. The in-situ vertical pressure. 4-14 shows the predefined cracks after applying 2000 kips of lateral loading. 4-15. Therefore. is applied at the top of the shaft segment.4. compressive failure. However. The lateral load of 2000 kips. A 5 feet long drilled shaft segment with 30 inch diameter is embedded in isotropic rock mass with the same properties as the rock used in the previous model. The magnitude of the maximum tensile stress is 1610 psi. 4-13.3 Failure of Rock at Great Depth Generally. considered large enough to induce failure tensile stress in rock mass. or both. the rock mass in front of shaft is still able to sustain loads. the wedge type of failure is unlikely to occur in rock at great depth due to overburden pressure from top layers of rock. In order to identify the stress redistribution after the crack appears. plus self weight of shaft and rock. the compressive strength is usually one order of magnitude higher than the tensile strength.

to 0. Tensile Stress Shaft Rock Loading Figure 4-13 Stress distribution at Y direction of in-depth rock layer Other than the tensile stress and compressive stress. the normal pressure on point A is smaller than that at other positions in the interface surface. Due to the applied lateral load. the maximum shear stress is developed at point B. the cracks would stop to propagate and the loads is redistributed to the rock in front of shaft. 121 . 4-16.0581 in. the maximum compressive stress is increased from 3780 psi to 3850 psi and the maximum deflection increased from 0. Deformed Max. Although the maximum relative movement between shaft and rock is at point A.it can be seen that the maximum tensile stress is decreased from 1610 psi to 540 psi. Because the tensile stress of 540 psi will be less than the tensile strength of the rock. the friction between shaft and surrounding rock mass is also developed during loading. shown in Fig. 4-16.0553 in. Actually. Compressive Stress Undeformed Max. as shown in Fig.

Shaft Crack Rock Load Direction Figure 4-14 The predefined cracks Max. Compressive Stress Rock Shaft Tensile Stress Figure 4-15 The stress redistribution at Y direction of in-depth rock layer after crack 122 .

2. 123 . the normal and shear stress distribution at failure is proposed as that shown in Fig. Finally. the maximum compressive stress would reach the compressive strength of rock. This is in conformance with the assumption made by Carter and Kulhawy (1992). Based on the distributions of compressive stress shown in Fig.1. 4-16. the ultimate capacity of in-depth rock mass in resisting lateral loaded shaft can be credited to compressive strength and friction between shaft and rock. it is concluded that tensile failure of rock would occur first. Then the maximum friction between shaft and rock would be reached.Side Wall of Rock B A Loading Figure 4-16 Friction distribution on the shaft-rock interface 4. Therefore. 4-17. 4-13 and friction shown in Fig.4 Suggested Failure Mode for In-depth Layer of Isotropic Rock Based on the above FEA analysis results.

The model is intended primarily for applications where the stresses are mainly compressive. fsr.2 Failure of Jointed Rock without Overburden Soils 4.2. If fsr = 0.2. A joint will be open when the stress normal to the joint tries to become tensile.Normal stress pL Shear stress distribution τ max β σ(β) = pL sin β α τ(α) = τ maxsin 2α Active earth pressure Applied load Figure 4-17 Suggested stress distribution at failure at great depth 4. such as sedimentary rock.1 Jointed Material Model in ABAQUS The jointed material model in ABAQUS is intended to provide a simple continuum model for a material containing a high density of parallel joint surfaces where each system of parallel joints is associated with a particular orientation. the retained shear modulus on the joint is governed by the shear retention parameter. Once a joint opens.2. it means that no shear stiffness is associated with 124 . The jointed material model assumes that the spacing of the joints of a particular orientation is sufficiently close compared to characteristic dimensions in the domain of the model such that the joints can be smeared into a continuum of slip systems.

joint surface failure.8. 4. In addition to the joint systems described in the above. one horizontal bedded joint system is assumed. the cohesion is 2 ksi and friction angle is 27 degree.0. and shear retention are the required input parameters. βa = the friction angle for system a. pa = the normal pressure across the joint. On notable change. dilation angle. on the other hand. the bulk failure. and da = the cohesion for system a.2. the jointed material model includes a bulk material failure mechanism.2 The Modeling of Drilled Shafts in Jointed Rock Mass The same physical model used in the previous section for drilled shaft in isotropic rock mass is used for this study. it corresponds to elastic shear stiffness in open joints.open joints. the shear retention is set to be 1. when fsr = 1. which is based on the Drucker-Prager failure criterion. In this study. Considering that the weak surface of jointed rock mass has been characterized by joint surface failure criterion. This is a small load for a 3 ft diameter and 15 ft long drilled shaft socketed into rock. and cohesion for the bulk material constitute the required input parameters. The friction angle. For the bulk failure. is that the rock mass model is changed from CAP Model to Jointed Material model. For joint surface failure. For jointed material model. The failure surface for sliding on joint system a is defined by: f a = τ a − p a tan β a − d a = 0 (4-8) where τa = the shear stress on the joint surface. the model can only be converged for the applied lateral load up to 57 kips.2. however.0 so that no shear stiffness is reduced when tensile stress is 125 . When the shear retention factor was assumed to be 0. same friction angle and 50% of cohesion of the input for bulk failure are assumed.

2. This is simply due to the fact that the uniaxial tensile strength of rock is usually one tenth of the uniaxial unconfined compressive strength of rock. The upward 126 . The mobilized maximum upward movement is 0. In order to have a wedge to be sheared out. which is about 31% of the maximum compressive stress of 3858 psi developed in front of the shaft. Fig. The tensile failure of the backside of the wedge. it is reasonable to guess a wedge type of rock mass that has the tendency to move out from its original position.developed across the joint surface.3 Failure at Top Layer of Jointed Rock Mass As it can be expected. It should be noted that the shear plane of the wedge has to be developed in order to form a wedge type of failure. This is much larger than the upward movement of 0. the rock mass. is more likely to occur before the compressive failure of the rock in front of the shaft can materialize. the shear surface has to be formed and the backside of the wedge has to be in tensile failure. which is mobilized to move forward. 4.01 inch under 1000 kips of lateral loading. 4-19 shows the contours of maximum shear stress concentration. 4-18 presents the tensile stress developed at the backside of the wedge. Fig.0076 inch for the shaft in isotropic rock mass under 3000 kips loading. the top layer of rock mass in front of drilled shaft experiences the forward movement. therefore. exhibits upward movement beyond the ground level. It can be seen that the possible shear plane can be 45° to 60°. Similar to the situation of isotropic case. It can be seen that the maximum tensile stress is around 1181 psi.2. One could infer that the joint plane significantly increases the displacement of rock mass under lateral loads. Similar to the case of isotropic rock. This change in modeling technique allows the applied load to be up to 1000 kips.

forward movement.movement. and the maximum shear concentration all support the notion of the development of a wedge type of failure. Shaft Rock Load Tensile Stress Figure 4-18 Tensile stress on xz plane for jointed rock mass case Shaft Rock Loa Figure 4-19 Maximum shear stress concentrations for jointed rock mass case 127 . possible tensile failure of rock on xz plane.

When the cracks start to propagate. the wedge type of failure is suggested for the top rock failure under lateral loads.2. as the uniaxial tensile strength of most rocks ranges from 725 psi to 2900 psi (Afrouz.4 Suggested Failure Mode for Jointed Rock Mass Based on the above analysis results of drilled shaft in horizontally jointed rock mass. Cracks and even tensile failure. a shaft cylinder with 5 feet length and 30 inch diameter embedded in jointed rock mass is created to simulate the in-depth situation. is applied to the top of rock-shaft system together with the selfweight of the rock and shaft. therefore. The magnitude of tensile stress of 870 psi is believed to be able to produce cracks.2.2. the rock mass in front of the shaft would still be able to sustain the applied load. 4. For jointed rock mass with non-horizontal joints.5 The In-depth Failure of Jointed Rock Mass Similar to the analysis for in-depth failure of isotropic rock mass. It appears that the FEA modeling technique is not capable of identifying the effects of joint direction. In order to identify the stress redistribution after the crack 128 . several similar 3D FEA analyses were carried out. The stress distribution at loading direction is similar to the case of isotropic rock. the suggested failure mode for horizontally jointed rock could be applied to rock mass with different joint directions with a great caution. The overburden pressure. as shown in Fig. 1992). 4-13.4. At this stage. The computed maximum tensile stress surrounding shaft was 33% of the maximum compressive stress. unfortunately. similar analysis results as the one of horizontally jointed rock mass were obtained. are likely to appear earlier than other types of failure for in-depth jointed rock mass.2. which is equivalent to the pressure at 6 feet deep in the jointed rock mass.

as shown in Fig. shown in Figs. may experience in-depth failure due to the overburden pressure. the maximum tensile stress is decreased from 870 psi to 326 psi. For those jointed rock with non-horizontal joints. the FEA model is modified to simulate a 10 inch long crack at the tensile failure areas. however.3 Failure of Isotropic and Jointed Rock with Overburden Soils For drilled shafts socketed into isotropic or jointed rock with overburden soils. the maximum compressive stress is increased from 2620 psi to 2790 psi.2. 4-17. Other than the tensile stress and compressive stress.3 Rock Strength Criteria Several rock strength criteria have been developed in the past. one can conclude that failure mode for isotropic rock shown in Fig. 4-12 and 4-17. the suggested in-depth failure mode for horizontally jointed rock may be used with caution. 4-16.2. Table 4-2 provides a summary of more prevalent rock strength criteria cited in literature. Similar to the situation of isotropic rock. and the smaller value should be adopted. Therefore.2. 4. the friction between shaft and surrounding rock mass is also developed during loading. On the other hand. Among them. The top rock layer.propagates.6 Suggested Failure Mode for In-depth Jointed Rock Mass Based on the above FEA analysis results. Hoek129 . the maximum shear stress is developed at point B rather than at point A. 4. both top layer failure and in-depth failure. Similar to the case of isotropic rock. which is less than the tensile strength of most rock. the in-depth failure must be same as the condition of without overburden soils. 4. should be checked for the top layer rock mass. could be used for horizontally jointed rock mass.

rock mass Advantages Improved several times. mb. widely referred. Table 4-2 Summary of Rock Strength Criteria Criteria Hoek-Brown (Hoek et al. Ramamurthy et al. however. consolidated previous revisions on the criterion and proposed the following general form of Hoek-Brown criterion. s and a = material constants that depend on the characteristics of rock mass and can be estimated as follows (Hoek. ' σ1 ' = σ3   σ' + σ ci  m b 3 + s    σ ci   a (4-9) where σ’1 = the major principal stress at failure. 2002) Ramamurthy et al. it requires conducting well-organized lab tests on samples with weak planes. In 2002. (1985) Johnston (1985) Applications Intact rock and highly fractured rock mass Intact rock.  GSI − 100  m b = exp  28 − 14D m i  r   (4-10) 130 . 2002). Hoek et al. σci = the uniaxial unconfined compressive strength of the intact rock.3. (1985) criterion is particularly suitable for jointed rock mass. 1980). rock mass.1 Hoek-Brown Criterion Hoek-Brown criterion has been developed and improved several times since its first version published in 1980. not applicable for anisotropic rocks 4.. and jointed rock Intact rock. calibrated with numerous data Applicable for anisotropic rocks Only uniaxial strength of intact rock is required Limitations Not applicable for anisotropic rocks Needs lab test on jointed rock samples Discontinuities are not well considered. σ’3 = the minor principal stress or confining pressure. et al.Brown criterion appears to be widely accepted since it has been revised many times and correlated with field data from its inception in 1980 (Hoek and Brown.

For deep foundation analysis. When no laboratory test data are available. Therefore. 1997). 4-10 can be determined from triaxial tests (Hoek and Brown. although the excavation releases the horizontal earth pressure on rock masses. the pouring of concrete restores the pressure. it can also be estimated from Table 4-3. The values shown in Table 4-3 can be varied ±2. the value of Dr is selected as 0 for applications in deep foundation analysis. Table 4-3 Values of Constant mi for Intact Rock (After Marinos and Hoek. 2000) Anhydrite Conglomerates Marls Crystalline Limestones 12 21 7 12 Sedimentary Rocks Breccias 20 Chalk Dolomites 9 Greywackes Sandstones 17 Siltstones Micritic 8 Sparitic Limestones Limestones Metamorphic Rocks Gneiss 28 Hornfels Migmatite 29 Phyllites Slates 7 Igneous Rocks Andesite 25 Basalt Diabase 15 Diorite Granite 32 Granodiorite Peridotite 25 Porphyries 7 18 7 10 19 7 25 25 29 20 Claystones Gypsum Shales 4 10 6 Amphibolites 26 Metasandstone 19 Schists 10 Agglomerate Dacite Gabbro Obsidian Tuff 19 25 27 19 13 Marble Quartzites Breccia Dolerite Norite Rhyolite 9 20 19 16 20 25 131 . The material constant mi used in Eq. GSI − 100  s = exp   9 − 3D  r   a= 1 1 −GSI / 15 + (e − e − 20 / 3 ) 2 6 (4-11) (4-12) where Dr is a factor depending upon the degree of disturbance to which the rock mass has been subjected due to blast damage and stress relaxation. It varies from 0 for undisturbed in situ rock masses to 1 for very disturbed rock masses.

there are very few published methods for predicting the horizontal side shearing resistance in the rock/shaft interface. and σ′n = σ ′ + 3 ′ (σ1 − σ ′ ) 2 3 ′ 2(σ1 − σ ′ ) + 0.4. However.5m b σ ′ 3 ci m b σ ci ′ 2(σ1 − σ′ ) 3 (4-15) τ = (σ ′n − σ′ ) 1 + 3 (4-16) 4.4 Side Shear Resistance According to the identified failure mode for in-depth rock mass layer. (4-9). σ'1 can be obtained using Eq. there was no available analysis method to account for side shear resistance of drilled shafts in rocks in the horizontal direction. 132 . the side shear resistance between rock and shaft will contribute to some portion of the ultimate resistance of rock. as suggested by a method proposed by Hoek (1990). However.  2τ  φ′ = 90 − arcsin  σ′ − σ′   3  1 (4-13) (4-14) c′ = τ − σ′n tan φ′ where σ'1 is in-situ vertical effective stress.3. Briaud and Smith (1983) measured the side shearing resistance by means of mounting pressure cells in the leading front of drilled shaft embedded in the soil.2 Converting Hoek-Brown Criterion to Mohr-Coulomb Criterion Mohr-Coulomb criterion expressed in terms of shear strength and normal stress is commonly used in geotechnical engineering disciplines. The Hoek-Brown criterion could be converted to equivalent Mohr-Coulomb criterion.

φ r = Residual angle of friction from triaxial test. respectively. sh 133 . 4-20. Johnston and Lam (1989a) proposed a theoretical procedure to estimate the side shear strength based on energy theory. φ r = Residual angle of friction from direct shear test.Fortunately. c sh = Cohesion from direct shear test. The stiffness of the spring is selected to match the rock mass modulus and the socket geometry for a given drilled shaft problem. sl p φ sh = Peak angle of friction from direct shear test. qu = Uniaxial compressive strength. An equivalent 2D model is used represent 3 dimensional interaction between rock and shaft. This make the application of Johnston and Lam (1989a)’s method difficult. a lot of parameters of rock mass were required as follows. In this method. In this model. Based on Mohr-Coulomb strength criterion and dilation energy theory. E m and υ m are Young’s modulus and (1 + υ m ) r Poisson’s ratio of rock mass. the upper half can move only in a vertical direction against a spring which provides the constant normal stiffness to the system. p φ sl = Peak angle of friction from triaxial test. that is K= Em 1 . as shown in Fig. Where r is radius of shaft. c sl = Cohesion from triaxial test. a series of formulas were developed to calculate side shear strength by Johnston and Lam (1989a). there are a lot of empirical methods and a few theoretical analysis methods available for estimating the side resistance of axially loaded drilled shafts.

1 Empirical Equations for Axially Loaded Drilled Shaft in Rock There exist several empirical relations between the ultimate vertical side shear strength. where τmax and σci are in MPa. K = Normal stiffness of rock mass. From the 134 . Researches by Pell et al (1980) and Johnston and Lam (1989a and 1989b) have shown that the increase in side resistance could be significant from increase in roughness of the wall of excavated hole. where α = 0. The power-curve relationships have also been proposed. 4.6 (Meigh and Wolshi. and unconfined compressive strength. 1979). Roughness of the wall of the excavated drill hole is an important factor controlling the development of vertical side shear resistance. Figure 4-20 Concrete and rock joint Comparing with complicated theoretical method proposed by Johnston and Lam (1989a). 1976) and τmax= 0. according to Toh et al. Such as. σ n 0 = Initial normal stress. τmax = 0. (1987).375σci0. τmax. σci. The relation could be expressed as τmax = α σci.22σci0. a lot of simple empirical equations have been proposed.i = Asperity angle.515 (Rosenberg and Journeaux. of the intact rock.25.4.

smooth-sided socket. They concluded that the power curve relationship is closer to the real cases. and R4. (1980) found that smooth-sided sockets exhibit a brittle type of failure.5 for lower bound ci (4-17) (4-18) (4-19) 135 . width greater than 10 mm. (1980) have used the size and frequency of grooves in a socket wall to classify wall roughness as R1.5 for mean behavior ci τ max = 0. at spacing 50 to 200 mm Kulhawy and Phoon (1993) developed a relatively extensive load test database for drilled shafts in soil and rock and presented their data both for individual shaft load tests and as site-averaged data. width greater than 5 mm. On the basis of site-averaged data. the following equations for axially loaded drilled shafts in rock was proposed (τmax and σci are in MPa).67σ 0.5 for upper bound ci τ max = 0. grooves or indentation less 1. 1980) Roughness Class R1 R2 R3 R4 Description Straight. Pells et al. τ max = 0.23σ 0. as defined in Table 4-4.. whereas sockets having an adequate roughness exhibit ductile failure. R2. R3. Williams et al. at spacing 50 to 200 mm Grooves of depth 4-10 mm.00 mm deep Grooves of depth 1-4 mm.45σ 0. at spacing 50 to 200 mm Grooves or undulations of depth greater than 10 mm.study of rock socket in mudstone and sandstone. Table 4-4 Roughness Classes (After Pells et al. width greater than 2 mm. Williams and Pells (1981) suggested that rough socket generates a locked-in normal stress such that there is practically no distinguishing difference between residual and peak resistance.

the mobilization of side shear resistance is depicted in Fig.20σ 0. Fig. τ max = 0. Additionally. empirical correlations between uniaxial compressive strength of intact rock and unit shaft resistance of rock socketed drilled shafts measured in load tests could be expressed in a general equation as follow.4. which significantly enhances the vertical side shear resistance.80σ 0. 136 . resulting in an increase in normal stress at the shaft/rock interface (Johnston and Lam. Geometrical constrains require that this relative sliding to generate dilation at interface. On application of an axial load to the shaft. τ max = ασ β ci units in MPa (4-22) For axially loaded drilled shafts in rock. 1989a). 4-21 also shows a drilled shaft section after a slip of the shaft relative to the rock takes place. to overcome the roughness of the socketed wall.5 Smooth socket (R1. 4-21.2 Suggested Empirical Equation for Side Resistance in Horizontal Direction As described above.Based on Kulhawy and Phoon (1993)’s database. R2 or R3) ci τ max = 0. the side shear resistance increases as the wall become rougher. therefore.5 Rough socket (R4) ci (4-20) (4-21) 4. and subsequent increase in socket diameter. Zhang (1997) suggested the following relation for smooth and rough socket conditions by using individual shaft load test data rather than the site-averaged data. the shaft and rock mass will deform elastically until the shear stress at the interface causes slip. This dilation occurs against a surrounding rock mass that must deform to compensate for enlargement of the socket diameter.

Therefore. the following empirical equation for predicting side shear resistance. therefore.42σ 0. by Kulhawy and Phoon (1993).Rock Shaft Rock Shaft Figure 4-21 Displacement behavior of drilled shafts in rock (Johnston and Lam. the normal stress at the leading frontal side of drilled shafts will be increased because of the applied lateral loads. roughness does not contribute much to side resistance in horizontal direction. There would be no or very small dilation taking place at the interface. However. τmax (psi). for mean behavior of rock is adopted. 1989a) For laterally loaded drilled shafts in rock. 4-22. as shown schematically in Fig. the relative displacement between the shaft and rock socket wall is in general parallel to the groove direction. τ max = 5.5 where the units are in psi ci (4-23) Lateral Load Drilled Shaft Normal Stress Figure 4-22 Increased normal stress due to lateral load 137 .

Fn is the normal force applied to the sides of the wedge and is assumed to be equal to the at-rest earth force. and 138 . Fsb is the friction force on the bottom face of the wedge.1 Ultimate Resistance of Rock Near Surface For highly fractured rock mass and competent rock. D is the diameter of the drilled shaft. In the figure.5 Ultimate Resistance of Rock Mass 4.5. the net rock reaction can be determined as follows: Fnet = 2Fs cos θ sin β + Fsb sin β + Fnb cos β − 2Fn sin θ − Fa Fa = 1 K a γ ′D(H − z 0 ) 2 2 (4-24) (4-25) 1 Fn = K 0 σ′v 0 A s + K 0 γ ′H 3 tan β sec θ 6 (4-26) 1 Fs = c′A s + K 0 σ′v 0 tan φ′A s + H 3 K 0 γ ′ tan φ′ tan β sec θ 6 Fnb = σ′v 0 A t + W + 2Fs cos θ cos β + c′A b cos β sin β − tan φ′ cos β (4-27) (4-28) (4-29) Fsb = c′A b + Fnb tan φ′ in which γ’ is the effective unit weight of the rock mass. H is the height of the wedge. Fa is the active earth force exerted on the drilled shaft. 4-12. According to force equilibrium in the loading direction. wedge type failure model for top rock layer was identified earlier in Fig. W is the weight of the wedge. Fnb is the normal force on the bottom face of the wedge which is to be determined through force equilibrium on vertical direction. Fs is the friction force on the sides of the wedge. Fnet is the total net rock resistance.4. σ'v0 is effective vertical soil pressure due to overburden soil on the top of rock and it is equal to zero if no overlying soils are present.

(4-13) and (4-14) by taking the value of σ'3 to be the effective overburden pressure at the depth of 1/3H. However. Hoek (1983) pointed out that the failure plane of rock is 45+φ'/2 because rock mass also follows Mohr-Coulomb failure criterion. since the side surface is triangular in shape. The values of φ' and c' can be obtained from Eqs. the following equations are adopted for θ and β. θ= φ′ 2 φ′ 2 (4-37) (4-38) β = 45 + 139 . Bowman (1958) suggested values of θ from φ'/3 to φ'/2 for loose sand and up to φ' for dense sand for a similar wedge type of failure in soils. (1974) for their wedge type of failure in sand. In this study. herein the value of θ is taken as φ'/2 based on the case studies (to be discussed in detail in Chapter VI) of actual field test results.K a = tan 2 (45 − φ′ / 2) K 0 = 1 − sin φ′ (4-30) (4-31) z0 = 2c′ γ′ K a − σ′v 0 γ′ (4-32) As = 1 2 H tan β sec θ 2 (4-33) (4-34) (4-35) (4-36) A b = (D + H tan β tan θ)H sec β A t = (D + H tan β tan θ)H tan β 1 1 W = γ ′( H 3 tan 2 β tan θ + H 2 D tan β) 3 2 The value of β was approximated as 45+ φ'/2 by Reese et al.

p u = 2∫ π/2 0 p L D / 2 sin 2 βdβ + 2 ∫ π/2 0 τ max D / 2 sin( 2α) cos αdα − p a D (4-45) pu = π 2 Dp L + Dτ max − p a D 4 3 140 (4-46) .2 Ultimate Rock Resistance at Great Depth For in-depth jointed rock with a set of parallel weak planes. 4-17 is adopted.The ultimate resistance of rock mass per unit shaft length pu (F/L) based on the wedge failure mode identified herein can be calculated as: pu = dF dF dFnet dF dF dF = 2 cos θ sin β s + sin β sb + cos β nb − 2 sin θ n − a dH dH dH dH dH dH (4-39) where dFa = γ ′K a (H − z 0 )D (dFa/dH ≥ 0) dH dFn 1 = K 0 H tan β sec θ(σ ′v 0 + γ ′H) dH 2 dFs H = H tan β sec θ(c ′ + K 0 σ ′v 0 tan φ′ + K 0 γ ′ tan φ′) 2 dH dFnb = dH D tan β(σ′v 0 + Hγ ′) + H tan 2 β tan θ(2σ′v 0 + Hγ ′) + c ′(D + 2H tan β tan θ) + 2 cos β cos θ sin β − tan φ′ cos β (4-40) (4-41) (4-42) dFs dH (4-43) dFsb dF = tan φ′ nb + c ′(D sec β + 2H tan β sec β tan θ) dH dH (4-44) 4. pu. τmax. heavily fractured rock mass and competent rock.5. the ultimate rock resistance per unit length. Therefore. pL are reached. the failure model shown in Fig. It is assumed that the ultimate resistance of rock is reached when both the maximum shear resistance between the drilled shaft and rock mass. and the normal limit pressures of rock mass. for in-depth rock can be computed as follows.

σ'1. The effective cohesion c' needed in Eq. The normal limit pressure of rock mass. Rankine’s earth pressure theory can be used to obtain pa as follows. (4-47) can be calculated using Eq. (4-14) by taking σ'3 equal to σ'v. which can be calculated using Eq. 4. for the top layer of rock mass with or without overlying soils. pL. the bottom plane of the failure wedge could be either within the rock mass or on a weak plane. therefore. τmax.5. For rocks to follow the wedge type failure. (4-46) for in-depth rock mass therefore would still work for jointed rock. Eq. For failure of rock mass at a great depth. Because the ultimate capacity calculated based on the presented in-depth failure model could be smaller than the one calculated based on wedge failure model. can be estimated from Eq.where D = the diameter of a drilled shaft.3 Ultimate Resistance of Jointed Rock For jointed rock with a set of parallel weak planes which dominate the behavior of rock mass under lateral loads. (4-23). (4-9) in which σ'3 is equal to σ'v. (4-39) and (4-46). the above equations for pu can still be used as long as the following modifications are made. is the major principal stress at failure. The ultimate shaft-rock interface friction in horizontal direction. It is suggested to calculate pu for 141 . pa = the active horizontal earth pressure. the high overburden pressure will prohibit possible sliding failure on the weak planes. the ultimate resistance of rock resistance per unit shaft length is determined as the smaller one of the values calculated from Eqs. p a = K a σ ′v − 2c ′ K a (4-47) (4-48) Ka = tan2 (45-Ф’/2) where σ’v = effective overburden earth pressure including the pressure induced by possible overlying soils.

The procedure of the proposed method is presented as follows. 4-12 can be used. (4-39) is still valid to calculate pu. (4-39) and (4-46). (4-42) needed for the two sides of the failure wedge can be obtained using instantaneous cohesion and friction angle determined from correlations with Hoek-Brown strength criterion. a numerical solution that is based on discretization technique is developed. the same wedge failure model shown in Fig. If the drilled shaft is structurally in a yield condition under the applied lateral load. 4-23.1 Free Head Boundary To estimate the ultimate lateral resistance of a drilled shaft in rock with free head boundary condition. as shown in Fig. However.both possible failure modes and choose the smaller value as the final pu. then the shaft is treated as a long or flexible shaft. The rock mass is modeled as n horizontal layers of rock with height hi for each layer. For failure shearing through a weak plane. Eq. except angle β should become the inclination of the weak plane and the values of c' and φ' for the bottom face of the failure wedge should be those pertaining to properties of the weak planes. (4-43) and (4-44) should be the results of lab test on samples with a weak plane. those c' and φ' in Eq. Eq.6 New Method for Predicting Lateral Capacity of Drilled Shafts in Rock 4. otherwise. Qi. 4.6. the shaft is considered as a short or rigid one. those c' and φ' needed for Eqs. The ultimate resistance per unit thickness of each layer of rock pi (F/L) can be calculated using Eqs. 142 . For failure due to shearing through rock mass. The total resistance of each layer. Specifically. (4-39) can be directly used. is equal to pihi.

the pivot point can be easily determined as given in Eq. (4-50). the pivot point needs to be found.H e Q1 = q1 h1 Qi = qi hi Rock Qn = qn hn Drilled Shaft Figure 4-23 The Rock-Shaft Model The first step is to assume the drilled shaft to be a short (rigid) one and find the ultimate capacity. Based on our observations from 3D FEA analysis and the lateral load test at Dayton. n is the total number of rock layers. (4-49). 143 . ∑ M = 0 ⇒ ∑ Qi Li = ∑ Qi Li 1 m +1 m n (4-49) ∑Q = 0 ⇒ H = ∑Q − ∑Q i 1 m +1 m n j (4-50) where e is the distance between the load point and ground surface. The lateral capacity Hu can then be determined according to force balance equation given in Eq. the drilled shaft rotates around some point (pivot) close to the bottom of the shaft. 4-24 (a). To calculate the lateral capacity of drilled shafts. xr is the depth of pivot point. as shown in Fig. By taking the moment balance at the loading point. and Li is the distance between the loading point and concentration point of force Qi. m is the layer number of rock layer where the pivot point is located.

the assumption of short shaft is not valid and a third step follows to obtain the capacity of the long drilled shaft. as shown in Fig. A trial and error process can be used to solve Eq. 4-24 (b). Then the maximum moment Mmax is given by: 144 . The depth f where shear force in the shaft is zero is located using the following equation: Hu = ∑Q 1 o i (4-51) where Hu is the calculated capacity in step one assuming the shaft is a rigid one. (4-51). Otherwise.H e Drilled Shaft Q1 Li Rock Qi Si Mmax Yielding Hinge f xr Q2 Qi Qm Qm+1 Qn (a) Short Shaft (b) Moment Diagram (c) Long Shaft Figure 4-24 Lateral capacity calculation models for drilled shafts in rocks The second step is to check whether the maximum moment of the drilled shaft exceeds the yielding moment of the shaft. o is the layer number of the rock layer where f is located. If it does not. The maximum moment of the drilled shaft is located at the section where shear force is zero. then the assumption of a short shaft is valid and the calculated Hu is the final estimate of capacity.

. the calculated maximum moment is equal to the yielding moment of the drilled shaft). The value of Mmax needs to be less than 145 . (4-39) and (4-46). The final Hu for a long (flexible) drilled shaft is obtained when convergence is achieved (i. as shown in Fig. A new value of Hu which is less than that calculated in first step is assumed. the rock mass is modeled as n horizontal layers of rock with height hi for each layer.e. the following procedure can be used to determine ultimate lateral capacity. The following two equations can be used to calculate the lateral capacity Hu and the maximum moment at the ground surface level. Then. 4. It is assumed that the drilled shaft will move rigidly in loading direction.M max = H u (e + f ) − ∑Q S i 1 o i (4-52) where Si is the distance between the concentration point of Qi and the position of shaft where shear force is zero. An iterative process is used to find the capacity so that the computed maximum moment in shaft is equal to the yielding moment of the shaft. The ultimate resistance per unit thickness of each layer of rock pi (F/L) can be calculated using Eqs. As before. 4-24 (c).6. The first step is to assume the drilled shaft behaves as a rigid pile. Step three is to calculate the lateral capacity for a long drilled shaft in which a plastic hinge appears at the location of maximum moment. Then step two is used to find the maximum moment corresponding to the new Hu. the total resistance of each layer Qi is equal to pihi. as shown in Fig. 4-25.2 Fixed Head Boundary For the case of fixed head boundary condition. My.

H Hu Q1 Li Qi Qn Before Moved After Moved Force Balance Moment Diagram Mmax Figure 4-25 Capacity of rigid drilled shaft at fixed head boundary In step 2. otherwise the second step needs to be followed to calculate Hu. H u = ∑ Qi 1 n n (4-53) M max = ∑ Q i L i 1 (4-54) where Li is the distance between the ground surface and the loading point of Qi.the yielding moment of the drilled shaft. After the location of Mmax is found using Eq. (4-56). where the drilled shaft rotates around a pivot point and the moment of shaft at the point right beneath the cap reaches the yielding moment. The maximum moment of the drilled shaft is located at the section where shear force is zero. Additionally. 4-26. 4-26 needs to be less than the yielding moment My. the value of Mmax can be 146 . By taking moment balance at the loading point (the point beneath the cap). (4-51). the maximum moment Mmax shown in Fig. the rotation point can be easily determined as given by Eq. The lateral capacity can be determined using Eq. (4-55). the drilled shaft is assumed to behave as shown in Fig.

find the location where shear force is zero and the corresponding maximum moment at that location using Eqs. 147 . as shown in Fig. the other one is located at certain depth where shear force is zero. The drilled shaft will have two yielding points: one is located at the top of the drilled shaft or the loading point. (4-57). Repeat this procedure until the calculated maximum moment is equal to the yielding moment. The final value of lateral capacity Hu is then obtained after the iteration is converged. If the maximum moment is greater than the yielding moment. ∑ M = 0 ⇒ M y + ∑ Qi Li = ∑ Qi Li m +1 1 m 1 n n m (4-55) H u = ∑ Qi − ∑ Qi m +1 0 (4-56) M max = H u ⋅ f − ∑ Q i Si − M y 1 (4-57) H Hu Li Pivot Point Before Moved After Moved Qm+1 Qn Force Balance Moment Diagram My Q1 Qm Si Qi My f Mmax Figure 4-26 Capacity of intermediate length drilled shaft at fixed head boundary Step 3 assumes the drilled shaft behave as a long pile. Next. (4-57). 4-27. (4-51) and Eq. the third step needs to be followed.easily determined using Eq. Assume a value of Hu which is less than the value of Hu calculated in Step 2.

4. as shown in Fig.7. and c-φ soils. 4-28. the following sections describe the methods for estimating the ultimate reaction of clay.7 Ultimate Reaction of Soils To make the lateral capacity prediction method presented above to be applicable to soils.The proposed method is not only suitable for drilled shafts entirely socketed in rock. (1977) examined the mechanism of the lateral reaction of piles in an elasto-plastic medium. 148 . but can also be used for drilled shafts in rock layers with overlying soils. a spreadsheet or computer program can be easily developed to facilitate calculation. sand.1 Ultimate Resistance of Clay Baguelin et al. H Hu My Q1 Yielding Point My Qi Mmax Q m Li Before Moved After Moved Force Balance Moment Diagram Figure 4-27 Capacity of long drilled shaft at fixed head boundary 4. It was found that the soil resistance is composed of the resistances from both normal and tangential directions. Although the developed method is somewhat time consuming for hand calculation.

is given by: p Ld = 11S u (4-60) 149 . 4-28. the limit resistance of top soil layer. τmax = ultimate shear resistance of soil. pLt. Based on the flow-around type failure for clay exists at a great depth. and z is the depth under consideration.83S u z D (4-59) where Su is undrained shear strength of clay. is given by: p Lt = 2S u + γ ′z + 2. (1975) is used herein. can be calculated as: pu = π D(p L + τ max ) 4 (4-58) where pL = ultimate normal resistance of soil. Based on the wedge failure mode. pu.By integrating the stress distribution as shown in Fig. pLd. the total ultimate soil resistance per unit shaft length. the limit resistance of in-depth clay. Normal stress pL Shear stress β α τ max p(β) = pL sin β τ(α) = τ max cosα Applied load Figure 4-28 Distribution of lateral reaction stresses The theoretical equation for calculating ultimate soil normal resistance by Reese et al.

(4-59) and (4-60).65 psi).The final value of pL is selected as the smaller value of pLt and pLd.5) for Su/pa >1.8p L + τ max )D (4-63) The limit normal stress of sand.5 in which pa is the atmospheric pressure (14. τ max = αS u (4-61) where α is adhesion factor which can be found in O’Neill and Reese (1999) as follows.55 for Su/pa ≤1. using Eqs. The ultimate τmax in horizontal direction is assumed to be the same as the ultimate vertical shear resistance estimated with the following equation from API (1991) τ max = Kγ ′z tan δ (4-65) 150 .7. (2005). The ultimate side shear resistance of clay can be calculated using α method given by. and α = 0. α = 0. pL.1(Su/pa-1. the ultimate resistance per unit length of drilled shaft can be calculated as p u = (0. according to Fleming et al.55-0. 4. (1992). can be calculated as p L = K 2 γ ′z p (4-64) where Kp = tan2(45+φ’/2) = passive earth pressure coefficient.5.2 Ultimate Resistance of Sand (4-62) The following procedure for calculating the ultimate resistance of sand was suggested by Zhang et al. That is.

pipe pile or H pile) Wood (i. Tables 4-5 and 4-6 taken from Kulhawy et al.3 Ultimate Resistance of c-φ Soils δ 1.5-0.. driven Pile-large displacement.. driven Drilled shaft-build using dry method with minimal sidewall disturbance and prompt concreting Drilled shaft-slurry construction with good workmanship Drilled shaft-slurry construction with poor workmanship Drilled shaft-casing method blow water table Note: K0 = coefficient of lateral earth pressure at rest K (0. step-taper pile) Smooth steel (i.7-1.9) φ’ (0.6-0.7) φ’ (0.7-0.9) K0 Table 4-6 Recommended Values of δ by Kulhawy et al.0) φ’ (0.8-0.8-1. timber pile) Drilled shaft built using dry method or with temporary casing and good construction techniques Drilled shaft built with slurry method (higher values correspond to more careful construction methods) 4.e.where K = lateral earth pressure coefficient..9) φ’ 1.0) φ’ Brinch Hansen’s (1961) theory can be used to calculate pu of c-φ soils.0 φ’ (0. (1983) and Kulhawy (1991) Pile type and method of construction Pile-jetted Pile-small displacement.e.7) K0 (0.7-0.7.7) K0 (0.0) K0 (0.9-1.e. (1983) and Kulhawy (1991) can be used to estimate appropriate values for K and δ.5-0. (1983) and Kulhawy (1991) Pile type Rough concrete Smooth concrete (i. and δ = interface friction angle between the drilled shaft and the soil. (2000) to modify Brinch Hansen’s equation as follow: p u = M ( γ ′zK q + c ′K c ) D (4-66) 151 .0) K0 (0..0) K0 (0.8-1. An empirical modification factor M was later suggested by Mokwa et al.e. precast pile) Rough steel (i.0 φ’ (0.0-2.2) K0 (1.9-1. Table 4-5 Recommended Values of K by Kulhawy et al.

where M = an empirical modification factor = 0. The values of Kq and Kc can be calculated as follows: Kq = Kc = αq = K0 + K∞αq Z / D q q 1+ αq Z / D K0 + K∞αcZ / D c c 1+ αc Z / D K 0 sin φ ′ ∞ K q − K 0 sin(45° + 0.8 Validation of the Derived Capacity Prediction Method Four field lateral load tests results are employed herein to evaluate the derived lateral capacity prediction method. Shaft #4 of Dayton load test and shaft #2 of PomeroyMason load test are selected.5φ ′) q K0 q (4-67) (4-68) (4-69) (4-70) (4-71) (4-72) (4-73) (4-74) (4-75) (4-76) (4-77) αc = K0 c K∞ − K0 c c ⋅ 2 sin(45° + 0.5φ ′) K 0 = e ( π / 2+φ′) tan φ′ cos φ′ tan(45° + φ′ / 2) − e −( π / 2−φ′) tan φ′ cos φ′ tan(45° − φ′ / 2) q K 0 = [e ( π / 2+ φ′) tan φ′ cos φ ′ tan(45° + φ ′ / 2) − 1] cot φ ′ c ∞ ∞ K q = N cd c K 0 tan φ′ K ∞ = N cd ∞ c c N c = [e π tan φ′ tan 2 ( 45° + φ ′ / 2) − 1] cot φ ′ 4 d∞ c = 1. The dimensions and strength parameters of the test drilled shafts of the three load tests are summarized in Table 4-7. (2002) are used. Kq = a coefficient for the frictional component of net soil resistance under 3D condition.85.09 tan φ ′ K 0 = 1 − sin φ ′ 4.58 + 4. and Kc = a coefficient for the cohesive component of net soil resistance under 3D condition. in which total length reflects the distance between the 152 . Additionally. the load test results reported by Hall and Wang (2000) and the short shaft of I-85 load test reported by Gabr et al.

The yielding moments of the drilled shafts are obtained from LPILE computer analysis by inputting reinforcement and physical dimensions of the drilled shafts. The depth is measured from the ground surface down.038 0.036 0.055 0.9 3.2 al.055 0.5 10.059 0.5 5000 inch casing Reinforcement Yielding Moment (kip-ft) 8008 21640 5488 2819 Table 4-8 Input rock mass parameters of the load tests Tests Dayton Shaft #4 PomeroyMason Shaft #2 Depth (ft) 0-7 7-18 0-21.2 6.057 0.4 29. Table 4-7 Test Drilled Shaft Information Test Diameter Total (ft) Length (ft) Dayton 6 18 Pomeroy.9 Mason Hall & 5 32 Wang (2000) Gabr et 2.3-48 48-56. The soil and rock properties required for calculation are summarized in Table 4.5-18 18-20 20-25 25-31 0-3.2 γ' (pci) 0.2 44 700 34 1200 35 1750 57 1750 57 4220 59 3596 59 6598 59 mi 6 6 6 17 4 4 4 6 6 6 7 9 9 9 Rock Shale Shale Shale Sandstone Claystone Claystone Claystone Sand (φ = 34º) Clayshale Clayshale Clayshale Siltstone Siltstone Siltstone Siltstone Hall & Wang (2000) Gabr et al.059 0.1 1 1 Concrete Strength (psi) 34#11 4500 28#18 plus 1 5115 inch casing 12#18 4000 12#10 plus 0.038 0.5 14.047 0.loading point and the tip of the drilled shaft.2-9.9-29.055 qu (psi) GSI 5668 40 5668 61 3797 42 9073 45 19 38 44.8 0-14.9 21.049 0.3 37.8 112.4-37.057 0. (2002) Loading Eccentricity (ft) 0 53.8. (2002) 153 .3 28 826.055 0.9-6.057 0.060 0.

The four load tests did not reach either structural yielding or rock mass failure during actual load tests. Therefore, Kulhawy and Chen (1995)’s hyperbolic curve fit technique is used to simulate the non-linear load-deflection behavior to the ultimate load, from which the ultimate capacity of piles (drilled shafts) is determined. The hyperbolic equation in terms of the lateral load (H) and the lateral deflection at the loading point (δ) can be expressed as follows:
H= δ a + bδ

(4-78)

where a and b are two curve fitting constants. The ultimate lateral load capacity is defined as the deflection δ becomes infinite large and is calculated as 1/b. A computer program LCPILE (lateral capacity of piles) using VC++ is developed in this study to facilitate computation of the developed method. The calculated capacities and the measured values from hyperbolic curve fitting technique of actual load tests data are summarized in Table 4.9. The prediction errors defined as the difference of capacities divided by the measured capacities from load tests are given in Table 4.9 as well. The capacity prediction error of Dayton site is relatively higher than the other two cases. This may be due to extremely small shaft head deflection of 0.135 inch at maximum applied lateral load of 1126 kips. Nevertheless, the developed method can yield reasonable predictions of the lateral capacities of drilled shafts socketed in rock mass. Table 4-9 Comparison of Lateral Capacities of Test Drilled Shafts Test Dayton Pomeroy-Mason Hall and Wang (2000) Gabr et al. (2002) Predicted Measured Error (kips) (kips) (%) 2447 1612 52 405 431 -6 500 589 -15 718 677 5.7 154

4.9 Summary and Conclusions The work presented in this chapter can be summarized as follows: 1. A 3-D FEM model for simulating a laterally loaded drilled shaft in rock is developed and validated against the lateral load test results at Dayton site. 2. A wedge failure mode for rock mass at or near ground surface is identified using 3D FEM analysis techniques. Strength controlled failure mode for rock at great depth is identified as well. 3. The effect of secondary structures of rock mass, such as joints and fillings is taken into consideration through adoption of a rock classification system GSI. Additionally, equations by Hoek (1990) for instantaneous cohesion and friction angle of rock mass are used in this study. 4. Empirical equation for estimating the axial side shear resistance of the rock-shaft interface by Kulhawy and Phoon (1993) is adopted herein to compute the side shear resistance of the interface in horizontal direction. 5. Based on the suggested failure modes, theoretical equations for calculating the ultimate resistance of rock mass per unit shaft length, pu, are derived for top layer rock mass and rock at great depth. The adoption of Hoek-Brown rock strength criterion and the empirical equation for side shear resistance ensures that the derived theoretical equations incorporate secondary structure effects. 6. A new method for predicting the ultimate lateral capacity of a drilled shaft socketed in rock mass is proposed. The method is also extended for use of a drilled shaft in various types of soils.

155

7. Finally, evaluation of the developed method for determining ultimate lateral capacity of a rock-socketed drilled shaft against load test results validate the accuracy of the method. Based on above work, the following conclusions can be drawn. 1. The 3D FEM model can be used to simulate the lateral response of drilled shafts in rock. This modeling technique can help verify a drilled shaft foundation design to avoid an expensive lateral load test. 2. The identified rock mass failure modes due to laterally loaded drilled shaft are useful for deriving theoretical equations for estimating the ultimate resistance of rock mass. 3. The secondary structures of rock mass were considered in the developed theoretical equations for calculating pu. The accuracy of the theoretical equations for calculating pu is acceptable based on evaluation against three field load test results. 4. Hyperbola extrapolation technique originally developed by Kulhawy and Chen (1995) may under estimate the lateral capacity of drilled shafts in rock if the deflection is small at the maximum applied load during lateral load tests. 5. The proposed method for estimating lateral capacity of drilled shafts in rock can provide reasonable predictions according to the evaluation using field test data. The average prediction error is about 21% which is adequate for practice considering the measured capacity may be under estimated. 6. The proposed capacity prediction method is versatile. It can be used for estimating the lateral capacity of drilled shafts in clay, sand, silts, and rock. The soil or rock can be layered. Both short and long drilled shafts can be considered. The boundary condition can free head or fixed head. 156

CHAPTER V ELASTIC SOLUTION OF LATERALLY LOADED DRILLED SHAFTS IN ROCK Although nonlinear p-y analysis is the most widely used method for analysis of laterally loaded piles, an elastic subgrade reaction solution nevertheless is an important alternative method for solving the problem in linear elastic range due to its simplicity and easy to use. Subgrade reaction method models soils as Winkler springs and the pile as a beam. In order to develop an elastic subgrade reaction solution for drilled shafts in rock, a methodology for determining deformation modulus of rock mass and an empirical equation for estimating the modulus of subgrade reaction of rock are developed in this chapter.

5.1 Determination of Rock Mass Deformability 5.1.1 Introduction

The fact that jointed rock masses do not behave elastically has prompted the usage of the term modulus of deformation rather than modulus of elasticity or Young’s modulus. The Commission of Terminology of International Society for Rock Mechanics (ISRM) published the definitions: Modulus of deformation - the ratio of stress to corresponding strain during loading of a rock mass including elastic and inelastic behavior; Modulus of

157

elasticity – the ratio of stress to corresponding strain below the proportionality limit of a material (Bieniawski, 1978). There are two categories of approaches, in-situ testing and empirical equations, for determining the deformation modulus of rock mass (Em). The existing in-situ testing for determination of Em includes plate loading, pressuremeter/dilatometer testing, Goodman Jack test. Although all of the in-situ testing methods have their limitations in measuring the actual deformation modulus, they are still the most desirable approach. To reach rocks at depth such as for deep foundation applications, pressuremeter/dilatometer test and Goodman Jack test are most appropriate. The Goodman Jack test will be introduced in this chapter. However, the details of pressuremeter/dilatometer test will be presented in Chapter VII. 5.1.2 Goodman Jack Test

5.1.2.1 Description of Goodman Jack The NX borehole jack device, also called Goodman Jack, was introduced in 1966 by Goodman et al. (1972). The Goodman Jack test is designed to be used in 3" boreholes. The schematic of loading mechanism of Goodman Jack is shown in Fig. 5-1 (Heuze 1984). The device consists of two steel plates forced apart by 12 racetrack-shaped pistons. The shape of these pistons was selected to give maximum hydraulic efficiency. The jack is attached to drill rod and inserted into the borehole. A hand pump is used to create hydraulic pressure in the lines connected to the jack, which in turn activates the pistons and produces a uniform and unidirectional stress field at the bearing plate. The applied hydraulic pressure is measured with a pressure gauge. The deformation of the rock is measured by two linear variable differential transformers (LVDT) and data are displayed 158

25% underestimate of modulus of rock. Figure 5-1 The schematic of loading of Goodman Jack (Heuze. the LVDTs have a linear range of 0.1. The effect of finite test length was investigated by using a 3D finite element analysis and taken into account by reducing 14% values of calculated modulus of rock. The total piston travel is 0. would lead to a 3.2 Test Data Interpretation Goodman Jack test data was interpreted through theory of elasticity to derive the deformation modulus of rock. β)] ∆Q h ∆u d / d 159 (5-1) . 1972) E = (0. Two return pistons close the instrument to a thickness of 2¾ in. 1984) 5. from 0.3.2.. Goodman et al. affording ¼ in. The effect of Poisson’s ratio ν is small. the equation for interpretation of test data is given by (Goodman et al.86)(0.93)[K (ν..2 inch and are adjusted to begin their linear travel when the plates are about to contact the rock. By considering the effect of limited length and the hydraulic efficiency.2 to 0. because a 50% overestimation in ν.5 inch. clearance for positioning in an NX borehole. (1972) studied the effect of Poisson’s ratio and the finite length of platens since the theoretical analysis of jack test assumes infinite length of the platens.by the indicator at the ground surface.

the rock mass modulus would be calculated as 13% lower than the modulus calculated without cracking. Subsequent to this research.3 0. (5-1). Hustrulid (1976). β) is a constant defined in Table 5-1. Meyer and McVey (1974)’s carefully conducted tests showed that the calculated modulus could be very different from the true modulus.05 1. which is much less than the known elastic modulus of aluminum alloy of 10.25 1. the formula for modulus calculation is given as: 160 .27 1.1 1.86 is for considering 3-D effect. d is diameter of borehole.08 × 106 psi. (1972) also investiaged the influence of crack generated by jacking or a nature joint located along the borehole by using finite element analysis.15 0. Shuri (1981) and Heuze and Amadei (1985).29 0. several researchers have evaluated various data interpretation methods.17 0. Hustrulid (1976) rederived the K(ν. A new expression T* was proposed for K. constant 0. ∆ud is increment of diametral displacement.93 is for considering hydraulic efficiency. ∆Qh is change of hydraulic pressure on platens. (5-1) was 2. β) function in Eq.38 0.09 1.20 0.35 1.2 0. constant 0.93 × 106 psi. For β = 45°.5 1. Since introduction of the Goodman Jack.28 1. 5-1. where the definition of β is shown in Fig.25 0. such as Meyer and McVey (1974). The calculated modulus using Eq.29 0. For a crack extending a half radius away from the hole.45 1.where.4 1. Heuze and Salem (1976). Table 5-1 Values of Constant K(ν) for β=45° ν K(ν) 0 1. and K(ν. Meyer and McVey (1974) conducted a jack test in a block of 5052 aluminum alloy measured 18 ×18 × 20 inch with a 3 inch central hole.13 0.23 Goodman et al.

borehole diameter and the modulus reduction factors are shown in Fig. The Shuri approach for oversize holes was found to be correct. 161 . 5-2. Heuze and Amadei (1985) integrated all the advances made in understanding the jack behavior prior 1985. Shuri (1981)’s work on oversize and undersize borehole was reevaluated. was recalculated with the T* of Heuze and Amadei (1985). Hustrulid (1976) also discussed the effect of oversized holes on the calculation of modulus. However. Shuri (1981) re-evaluated the radius mismatch problem first discussed by Hustrulid (1976). (5-2) was based on two-dimensional analysis. what Shuri refers to as the Eact (Eactual) is in fact the modulus without considering the plate bending effect. Heuze and Salem (1976) performed both two-dimensional and 3-dimensional analysis and found that the ratio of plate modulus to rock modulus.E calc = (0. and the plate geometry would influence the deformations during jack test. and Eact is the modulus considering the effect of undersize and oversize. the rock anisotropy. the derivation by Shuri for undersize holes was found to be incorrect and was corrected by including the new value of T* from Heuze and Amadei (1985). It is however essential to apply the plate bending correction to yield the true rock modulus.93)(3)(T * ) ∆Q h ∆u d (5-2) It should be noted that Eq. Shuri analyzed both the undersize and oversize cases relative to 3 in. the original correction from Ecalc to Etrue obtained by Heuze and Salem (1976) via 3-D finite element analysis for the plate bending. Therefore. Shuri ignored the plate bending problem. in which Eapp is calculated modulus of rock Ecalc without considering the undersize or oversize effect. Additionally.

93)(d ) ∆Q h * (T ) ∆u d (5-3) The values of T* are provided in Table 5-2.(1985). Finally. Figure 5-2 Modulus reduction vs. hydraulic pressure for various borehole diameters (a) undersize – the pressure must be decreased by about 14% for a given Eapp/Eact as recalculated by Heuze et al. 5-3 to account for the platen bending effect.The current version of test data interpretation follows the following steps. the corrected modulus from Step 2 is further corrected using Fig. Step 1: calculate the Ecalc by using the following equation for full contact condition: E calc = (0. (b) oversize holes 162 .86)(0. Step 2: the Ecalc calculated from Step 1 is corrected using Fig. 5-2 to account for the undersize and oversize effect.

1985) 5.3 0. this is a satisfactory upper limit of pressure for 163 .25 0.3 Selection of In-Situ Test Method From fundamental point of view.4 0.33 0.1 0.519 1. and linearly elastic. 1985) ν 0.151 Ecalc(GPa) 0 16 14 12 10 8 6 4 20 2 0 0 1 2 3 4 5 0 60 Etrue(GPa) 100 10 20 30 80 Etrue(10 psi) 6 40 Ecalc(106psi) Figure 5-3 Correction for platen bending of the jack (after Heuze and Amadei. In fact. homogeneous.289 1. which usually is less than 4. Both of them assume that the test chamber or platen is infinite long and that the rock is isotropic. The difference between the two devices is the method to apply the pressure to the borehole walls.1. but the Goodman Jack applies unidirectional pressure to the borehole wall.Table 5-2 Values of T* (Heuze and Amadei.5 T* 1.2 0.366 1. Dilatometer can apply uniform pressure to the whole borehole wall.474 1.438 1.397 1. The disadvantage of dilatometer is the limit of applied pressure. both dilatometer and Goodman Jack involve similar assumptions and theoretical derivations for test data interpretation.350 psi (30 MPa).

Bukovansky (1970) found that the deformation modulus determined from test results of dilatometer were surprisingly low comparing that from plate-load. 1970. (1970) pointed out that the dilatometer test results are reliable based on their 387 dilatometer tests carried out in 9 test programs. 1986. 1997.075 psi (69MPa) pressure limit.. Bukovansky. mainly in jointed weak mudstone. self-boring pressuremeter and Goodman Jack tests in Hong Kong. Reese (1997) utilized dilatometer test results to correlate the initial slope of his interim p-y criterion for weak rock. Georgiadis and Michalopoulos. such as 164 . the ability to generate higher pressures allows the test to be carried beyond the elastic region of many rock types. 1972). However.many rock types and it exceeds the actual stress level experienced by rock in most of civil engineering works (Goodman et al. and Littlechild. Flatjack.. A total number of 20 dilatometer tests were performed in seven boreholes. (2000) investigated the use of high pressure dilatometer. and radial jacking tests. the Goodman Jack can reach 10. Further. It was found that the dilatometer tests were not necessarily successful. However. Georgiadis and Michalopoulos (1986) investigated the application of dilatometer test to design grouted pile in rock. 1970. Littlechild et al. Reese. in contrary to the preceding successfully application of dilatometer. 2000). The current practice of dilatometer and Goodman Jack in determining the deformation modulus of rock mass has shown both successful and unsuccessful cases (Rocha et al. Rocha et al. The dilatometer test results were utilized in conjunction with laboratory test results to design laterally loaded steel-pipe piles drilled and grouted in mudstone. In highly fractured rocks. The disadvantage of Goodman Jack is the less precisely known pressure condition under the load comparing with dilatometer.

GSI = Geological Strength Index. it can be concluded that dilatometer test is suitable for the determination of weak rock mass modulus. Bieniawski (1978) could be credited for the development of earlier empirical equations.1. and another construction site with heavily jointed marly rock mass. RQD = Rock Quality Designation. for weak rock. Other empirical equations found in literature are summarized in Table 5-3.4 Existing Empirical Equations Several empirical equations are available for the determination of deformation modulus of rock mass using empirical correlations with rock properties. The parameters used for correlations and perceived limitations of these empirical equations are provided in Table 5-3. except for highly fractured rock. 5. 2000). Goodman Jack tests require careful consideration of the maximum pressure that will be applied so that the device can be retracted from the bore hole (Littlechild et al. It was shown that Goodman Jack test can be easily carried out in more uniform and competent rocks that exhibit little fracturing or weathering. WD = weathering degree.000 in 2003.. Em = the deformation modulus of rock mass. Ei = the modulus of intact rock. in which RMR = Bieniawski (1976)’s Rock Mass Rating. The purchasing price for either of the test devices is around $30. Gokceoglu et al. However.granodiorite. The database was obtained from two dam sites with quartzdiorite and limestone. (2003) evaluated existing empirical equations by utilizing a database consisting of 115 in-situ test data points. The measured deformation modulus of rock mass was obtained 165 . UCS =σci= unconfined compressive strength of intact rock. while Goodman Jack test is applicable to hard rock mass. dilatometer membrane tended to fail (puncture) frequently. Based on review provided in the above sections.

RQD. RQD and WD Ei. However. (2003) Required Parameters RMR RMR Ei and RMR Limitation RMR > 50 RMR ≤ 50 Equation Em=2RMR-100 (GPa) Em=10(RMR-10)/40 (GPa) Em=Ei[0.using in-situ plate load test and dilatometer test. 166 .1811 GSI.9exp(RMR/22.5528 Ei in GPa  (E / UCS)(1 + RQD / 100)  E m = 0. the data for weak rocks in this database was limited. However. σci >100MPa Ei. (2003)’s equation was developed with half of the data points in the entire database.0028RMR2+0.99 for dilatometer data including the weak marly rock mass results.135 i  WD   1. The evaluation results show that Hoek and Brown (1997)’s equation gave the best results for a weak marly rock mass. Kayabasi et al. (2003) Gokceoglu et al.82)]/ 100 E m = (1 − D r / 2) for σci ≤ 100MPa. Em and 1. Table 5-3 Empirical Equations for Estimating the Deformation Modulus of Rock Mass Empirical Equation Bieniawski (1978) Serafim and Pereira (1983) Nicholson& Bieniawski (1990) Hoek and Brown (1997) & Hoek et al. UCS and WD  E (1 + RQD / 100)  E m = 0. (2002) Kayabasi et al. σci and Dr which is disturbance factor σ ci ( GSI−10) / 40 10 100 (GPa) (GPa) for E m = (1 − D r / 2)10 (GSI−10) / 40 . Kayabasi et al. having a mean uniaxial compressive strength of 2697 psi.001 i  WD   (GPa) The root mean square errors (RMSE) were calculated to evaluate the performance of the existing empirical correlation equations. (2003)’s equation exhibited a high predictive capability with the value of RMSE equal to 0.

weathering) show high correlations with Em than that of the unconfined compressive strength of intact rock. the equations proposed by Bieniawski (1978) yielded the most scattered results. (2003) employed cosine amplitude method to investigate the strength of correlations between Em and the intact rock and discontinuity properties of rock masses based on 57 test data points. Contrary to these.1. which was developed from the test data of Bieniawski (1978) and Serafim and Pereira (1983). Hoek and Brown (1997)’s equation provides good estimate on modulus for weak rock. volcanic and metamorphosed sedimentary rocks. (2003) yields good prediction for his database. The statistical analysis results are summarized in Table 5-4. 5. qu. (2003)’s evaluation discussed previously. exhibited good results based on the RMSE values calculated using all the data in Gokceoglu et al. All the parameters associated with secondary structures (such as RQD. it appears that Nicholson and Bieniawski (1990)’s equation could be adopted. in general. 167 . however.5 Parameters for Determination of Em Based on literature review in the previous sections. Kayabasi et al. it nevertheless needs to be further evaluated with other database. and RQD have been selected to correlate with deformation modulus of rock masses. It can be seen that modulus of intact rock correlates best with Em. Ei. rock properties such as RMR. Although the equation proposed by Gokceoglu et al.Nicholson and Bieniawski (1990)’s equation. Littlechild et al. (2003)’s database. (2000) found that this equation was not appropriate for foundation design on weathered rock according to their field test results in Hong Kong for igneous. Based on Gokceoglu et al.

6 504.878 0. A set of dilatometer test data in Ironton-Russell (Paul and Martin 2004) is compiled in Table 5-5 for independent check on various correlation equations. This research adopts the modulus of intact rock and a rating system (such as RMR and GSI) that represents the effects of all secondary structures.400 0.Table 5-4 The Strength of the Relation between EM and Parameters (Kayabasi et al. RQD and weathering degree) for their correlation equation.1 586. mea.400 0.4 625.400 0. RQD 1 75 2 98 3 78 4 100 5 100 6 98 7 98 qu (psi) 8406 9544 9559 9005 8002 9137 8578 Ei (psi) 1689.8 439. 2003) Parameter Modulus of elasticity of intact rock RQD Weathering of the discontinuities Roughness of the discontinuities Aperture of the discontinuities Continuity of the discontinuities Infilling Spacing of the discontinuities Uniaxial compressive strength of intact rock Strength Value 0. (2003)’s database and the unsatisfactory predictions from Hoek and Brown (1997)’s equation based on on Littlechild et al.873 0.768 0.1 1343.7 1751.750 0. The good prediction results on Em from Nicholson and Bieniawski (1990)’s equation based on Gokceoglu et al. (2000)’s test data support this conclusion.400 0.3 1774.5 636.851 0. Table 5-5 Properties of Rock Masses in Ironton-Russell No.4 525.5 EM. (psi) 667.400 168 .400 0.832 0.400 0. (2003) selected the first three factors in Table 5-4 (Ei.3 GSI 74 77 74 77 77 77 77 RMR 64 67 64 67 67 67 67 WD 0.5 1809.761 0.720 Although Kayabasi et al.865 0.4 1763.4 1569.

1 0.9 10.3 0.9 463. Ei and rock rating system (such as RMR and GSI) are the most important parameters for correlating with modulus of rock masses.04 5276.9 7. Therefore. Nicholson and Bieniawski (1990)’s equation which employs Ei and RMR as the parameters provides the best estimate on Em.5 2182. especially for borehole cores in which relatively few intact core pieces longer than 4 inch can be recovered (Hoek and Brown.2 6.5 2493. Hoek and Brown (1997)’s equation.71 2893 5.0 0.9 476. and Kayabasi et al. 169 .6 2033.19 2893 4.5 396.88 5566.4 4.6 11.76 4686.6 517.4 2474.6 1816. while the other two methods over-predict Em. 1997).Table 5-6 provides the predictions of Em and the ratio of predicted over measured deformation modulus of rock masses using Nicholson and Bieniawski (1990)’s equation.99 5446.1 10. (2003)’s equation.7 Proposed Empirical Equation Although Nicholson and Bieniawski (1990)’s equation is the most promising empirical equation for estimating deformation modulus of rock masses.8 4.90 5096.1 9.43 2893 4. the use of RMR makes it suffer some inherent drawbacks. the GSI system is proposed as an alternative to the RMR system for estimating deformation modulus of rock masses.13 2893 6.67 4395.9 1.5 2249.1.3 4.5 2455.7 0.73 5406.6 0. 1 2 3 4 5 6 7 5.5 523.05 2434 3. Serafim and Bieniawski (1990) (1997) (2003) Pereira (1983) Prediction Ratio Prediction Ratio Prediction Ratio Prediction Ratio 445. the physical appearance of material recovered in the core can be used to estimate a GSI value.9 3.60 2434 3.6 8. The RMR value is difficult to estimate for very poor rock.94 2893 5. Therefore.5 0. In these circumstances.6 520.0 3.9 3.6 Nicholson and Hoek and Brown Kayabasi et al.3 4. Table 5-6 Predictions and Ratios of Predicted over Measured Modulus of Rock Masses No.

1983) used to develop their equation is used to develop a new empirical equation. GSI values of these test data are estimated by adding original RMR with 5. it was assumed that the ground water rating was set to an average value of 5. 170 .7 Recommended Methodology for Determination of Deformation Modulus of Rock In-situ tests usually provide more reliable values for the deformation modulus of rock mass than empirical equations. Therefore. A regression analysis is carried out and the results are plotted in Fig. However. the new data from Ironton-Russell is included.7 (e ) 100 (5-4) 5. Due to the lack of details of estimating RMR of these data. 1978 and Serafin and Pereira. For weak rock. Additionally. 5-4 is recommended to estimate the deformation modulus of rock mass.To preserve the merit of Nicholson and Bieniawski (1990)’s equation. Em = E i GSI / 21. 5-4. According to Hoek and Brown (1997). The following equation is proposed to estimate deformation modulus of rock masses using modulus of intact rock and GSI. 1976). for hard and competent rock. For preliminary design or where insitu tests are not available. Goodman Jack test is recommended since it can provide higher pressure than dilatometer. GSI equals to RMR (Bieniawski. with the ground water rating set to 10 and the adjustment for joint orientation set to 0. the test data (Bieniawski. the derived Eq.1. dilatometer test is recommended since this test has been widely and successfully used in geotechnical engineering applications.

7)/100 Em/Ei (%) 20 40 GSI 60 80 100 Figure 5-4 Proposed empirical equation using GSI 5.65E  ED 4  K=   1 − ν2 EpIp    1 / 12 (5-5) 171 .100.0 80.2.0 20. K (F/L2) as follows.0 40. few studies and discussions are available regarding the determination of modulus of subgrade reaction. By fitting the subgrade reaction solution with continuum elastic solution for beam on elastic foundation.2 Initial Modulus of Subgrade Reaction of Rock Mass 5. 0.0 60.1 Introduction The term initial modulus of subgrade reaction refers to the initial slope of a p-y curve. especially for rock masses.0 0 Bieniawski (1978) Serafin and Pereira (1983) Ironton-Russell Regression Em/Ei=exp(GSI/21.0 0. Vesic (1961) provided an elastic solution for the modulus of subgrade reaction. In literature.

the initial slope of p-y curves (initial modulus of subgrade reaction). was given by: K i ≈ k i Em (5-8) 172 . Bowles (1988) suggested to double the value of K in Eq.0 m. in reality. In this criterion. and EpIp = flexural rigidity of beam. η = a load transfer factor (Guo. EpIp = flexural rigidity of piles or drilled shafts. D = beam width. ν = Poisson’s ratio.where E = modulus of elastic materials. Guo (2001) proposed an expression for K based on a closed form elastic solution for laterally loaded piles. 1. 2001). Reese (1997) proposed an interim p-y curve criterion for weak rock. For drilled shafts in rock masses. Dref = 1. Carter (1984) modified Vesic’s equation as follows that would account for the effect of pile diameter. Based on field test data. The subgrade reaction modulus was given by: 3πG s K= 2 2  K (η)   2   K 1 (η)   2η 1  − 1  −η   K 0 (η)   K 0 (η)       (5-7) where Gs = shear modulus of soils. soils are not in full contact with the piles when the lateral loads are applied. The above discussions were focused on piles in soils or a linear elastic media.0ED K= (1 − ν 2 )D ref  ED 4     EpIp    1 / 12 (5-6) where the reference pile diameter. Ki. (5-5) for piles under lateral loading since the pile would have contact with soils on both sides. and K0 and K1 = Bessel functions. However.

The drilled shaft is modeled as a cylinder with elastic material properties.where Em = modulus of rock mass.2. Vesic (1961) elastic solution for beam-on-Winkler foundation is directly applicable to laterally loaded drilled shafts in soils or rock.2. Guo (2001) solution is complex for practical use and requires further verification with actual test data. respectively. the validation using field test data shows that the proposed empirical equation would provide good prediction on initial subgrade reaction modulus of rock mass. An empirical model for initial subgrade reaction modulus of rock masses is proposed in this study. A parametric study using the 3D FEM model presented in Chapter IV is carried out to obtain an empirical equation for estimating initial subgrade reaction modulus. From experimental data. Since the 173 . and z = depth. Finally. empirical equations for determining ki are given as: ki = (100+400z/(3D)). and ki = dimensionless constant. The solid elements C3D15 and C3D8 are used to develop mesh representation for drilled shaft and rock.2. The equation proposed by Reese (1997) is empirical in nature and only based on two load tests results. The Modified Drucker-Prager Model (CAP Model) is utilized to represent isotropic. homogeneous rock masses. Surface interface technique is employed to simulate the rock-shaft interface. 5.2 FEM Model and Parametric Study (5-9) (5-10) 5.1 FEM Modeling The 3D FEM model developed in Chapter IV is used to simulate the response of a laterally loaded drilled shaft in rock. 0 ≤ z ≤3D ki = 500. z > 3D where D = the diameter of drilled shafts.

and deformation modulus of rock mass (Em). The force applied to rock mass per unit thickness surrounding the shaft is obtained directly from the output of FEM analysis. as shown in Fig. shaft-rock relative stiffness (EpIp/(EmD4)). v = 0. 174 . The rock mass surrounding the shaft will be modeled as multiple horizontal layers with 12 inch thickness for each layer.2. Various factors. p-y curves of the rock layer can be obtained. Other pertinent parameters are kept constant: D = 72 inch. Poisson’s ratio (ν). The rock resistance per unit shaft length p is then calculated by dividing the force with the rock layer thickness of 12 inch for each loading level. is studied. 5. A total of 30 cases of FEM analysis were carried out. are studied. By extracting the deflections of the shaft under various load levels at the corresponding depth of the rock layer. It is found that it has a linear relationship with initial modulus of subgrade reaction modulus Ki.determination of initial subgrade reaction modulus is the primary objective of this study. such as shaft diameter (D). only the elastic response of rock will be the concern of this parametric study.2.3. The range of modulus was ranged from 100 to 2000 ksi for weak rock. shaft-rock interface properties. 5-5.2 Effect of Rock Mass Modulus Em The effect of deformation modulus of rock mass. The diameter of shaft used for parametric study varies from 2 feet to 10 feet. P-y curve is generated at a depth of 18 inch. L = 216 inch. Hyperbolic curve fitting technique is employed to fit the p-y curves so that a unique value of initial slope of the p-y curves is obtained. Ep = 4000 ksi. The initial subgrade reaction modulus Ki is obtained as the initial tangent to the p-y curves. Em.

Em = 200 and 500 ksi. The variation of modulus of rock mass does not change the relationship between Ki and ν.1 0.3 Effect of Poisson’s Ratio ν The Poisson’s ratio for rocks is varied from 0.4.5 Poisson's Ratio v Em=200 ksi -1.32x -1. Fig.4000 3000 Ki (ksi) 2000 1000 0 0 1000 Em (ksi) 2000 3000 Figure 5-5 Initial modulus of subgrade reaction for various moduli of rock 5.2 0. Ep = 4000 ksi. while maintaining other pertinent parameters constant as follows: D = 72 inch.1 to 0. 5-6.27x Figure 5-6 Initial modulus of subgrade reaction for various Poisson’s ratio 175 .2. 5-6 shows an exponential relationship between Ki and Poisson’s ratio ν.4 0.2. 2500 Em=500 ksi 2000 Ki (ksi) 1500 y = 2288e 1000 500 y = 864e 0 0 0. Again.3 0. L = 216 inch. p-y curve is generated at a depth of 18 inch. as shown in Fig.

it could be attributed to the numerical integration for calculating p.5. therefore.2.4 Effect of Depth The relationship between initial subgrade reaction modulus and depth is also investigated using the same set of rock properties as discussed in previous section. 50000 45000 40000 35000 30000 25000 20000 15000 10000 5000 0 0 1 y (in) Figure 5-7 P-y curves along with depth p(lb/in) 12 in 36 in 60 in 84 in 2 3 176 . while the ultimate resistance of rock mass increases with depth.2. 5-7. as shown in Fig. The difference of Ki along depths is very small. The rock properties were set to be same along the depth. Because the rock modulus is constant along depth. supposedly the deformability of rock should be same along depth. It is found that Ki (the tangent to p-y curve) decreases along depth.

200 Ki(ksi) 150 100 0 0. The coefficient of friction of the interface is the main controlling parameter of the shaft-rock interface.6 Effect of Diameter There has been a long history of debate regarding the effect of diameter on the modulus of subgrade reaction.2. Ashford and Juirnarongrit (2003) indirectly verified that the modulus of subgrade reaction is independent of pile diameter when examining full-scale vibration tests results on cast-in-drilled-hole piles in a sand deposit. Additionally.5. 177 .5 Coefficient of Friction on Interface 1 Figure 5-8 Initial modulus of subgrade reaction for various rock-shaft interface frictions 5.5 Effect of Interface Properties The interface between shaft and rock was modeled by a surface based interface technique in the ABAQUS software.2. 5-8 clearly indicate that the interface frictions exert no effect on the initial modulus of subgrade. The computational results shown in Fig. Terzaghi (1955) and Vesic (1961) concluded that the diameter of piles has no effect.2.2.

Ep varies from 3500 to 4500 ksi. v = 0.3. Again.7 Effect of Shaft-Rock Relative Stiffness The effect of shaft-rock relative stiffness.2. v = 0. It can be seen that Ki increases with diameter linearly.However.3. is also investigated herein. Ep = 4000 ksi. EpIp/(EmD4). The input parameters are: D = 72 inch. 2500 2000 Ki (ksi) 1500 1000 500 0 0 3 6 D/Dref (Dref =1 ft) 9 12 Figure 5-9 Ki varies with shaft diameter 5. Em = 500 ksi.2. The FEM computed relationship between the shaft diameter and initial modulus of subgrade reaction of rock mass is presented in Fig. the effect of diameter is investigated using the following input parameters: D ranges from 24 inch to 120 inch. P-y curve is generated at a depth of 18 inch. In this study. Carter (1984) found from his own field test data that the diameter of piles has exerted significant effect on modulus of subgrade reaction. The shaft diameter is normalized with a reference diameter of 1 foot. Em varies from 100 to 2000 ksi. L = 216 inch. 5-9. It is found that a power law can be used to describe the relationship between Ki 178 . the p-y curve is generated for a depth of 18 inch.

4000 3000 Ki (ksi) y = 491x 2000 1000 0 0 0. d). A regression analysis on data from the FEM parametric study is carried out. c). 5-10. Both the modulus of rock masses and shaft diameter exhibit a linear relationship with K i.and shaft-rock relative stiffness. as shown in Fig. b). An equation for predicting initial modulus of subgrade reaction is fitted to match Ki values 179 . Ki exponentially decreases with increasing Poisson’s ratio of rock mass. EpIp/(EmD ) Figure 5-10 Initial modulus of subgrade reaction for various shaft-rock relative stiffness 5.5 4 2 -0.3 Suggested Empirical Equation Based on the parametric study presented in the previous sections.5 2 2. Similar results concerning the effect of relative stiffness on Ki were found previously by Guo (2001) in his numerical study.9937 3 Relative Stiffness.5 1 1.8003 R = 0. The relative stiffness of the pile and rock has shown a power relationship with Ki.2. Rock-shaft interface friction exerts no effect on Ki. the following conclusions may be drawn: a).

the empirical equation can be derived as K i = E m (D / D ref )e − 2ν   EpIp  E D4  m     0.Empirical Prediction y=x Ki=Em (D/Dref)e (Ep Ip /EmD ) -2v 4 0. Pomeroy-Mason. 5-11.2..2.obtained from FEM parametric study.FEM 3000 2500 2000 1500 1000 500 0 0 500 1000 1500 2000 2500 3000 3500 Ki (ksi) .4 Validation of the Empirical Equation 5. and North Carolina by Cho. are employed to verify the developed empirical equation for predicting initial modulus of subgrade reaction of rock mass.284 (5-11) 4500 4000 3500 Ki (ksi) .4. The Dayton and Pomeroy-Mason load tests were previously presented in Chapter III.e. and the load test in North Carolina is collected from open literature. 2000). et al. As shown in Fig. 180 . Dayton.284 Figure 5-11 Comparison of FEM computed and predicted subgrade reaction modulus 5.1 Field Tests Three lateral load tests in rock (i.. A summary of the properties of the test drilled shafts for three cases is provided in Table 5-7.

9 16 Socket Length (ft) 18 56. an independent verification of the experimental p-y curves is carried out. It can be seen that a good agreement between the predicted and measured loaddeflection curves at the shaft head is achieved. The experimental p-y curves are used as input in COM624P analysis. the experimental p-y curves were derived and presented by the authors. Although the accuracy of the experimental p-y curves was verified by the authors.Table 5-7 Summary of Lateral Load Test Drilled Shafts Drilled Shafts Dayton Shaft #4 Pomeroy-Mason Shaft #2 North Carolina Long Shaft Diameter (in) 72 96 (socketed) 30 Length (ft) 18 112.8 9. Load-Deflection.2 For the lateral load tests in North Carolina. NC Long Shaft 250 200 Load (kips) 150 100 50 0 0 0.8 Overlying Soils (ft) 0 0 4. 5-12. The predicted load-deflection curve at the shaft head is compared with the measured curve in Fig.5 1 Measured Predicted 1.5 2 Deflection (in) Figure 5-12 Comparison of load-deflection curves of North Carolina load test 181 . thus indirectly verifying the representativeness of the experimental p-y curves.

Thus. The initial linear portion of the pressuremeter test curves of the North Carolina test was used to determine the modulus of rock masses.23E+12 5.77E+13 1.1 98.23E+12 1.The experimental p-y curves of rock masses for Dayton and Pomeroy-Mason Load tests were derived using piecewise polynomial curve fitting. Table 5-8 Modulus of Rock Masses Based on Pressuremeter Test Drilled Shaft NC Long Shaft NC Long Shaft NC Long Shaft Depth (in) 102 141 180 Em (psi) 25252 40060 54998 EpIp (lb-in2) 1.1 38.1 23.1 are used.2.43E+11 Table 5-9 Modulus of Rock Masses Based on Empirical Equation Test Dayton Shaft #4 Dayton Shaft #4 Dayton Shaft #4 Dayton Shaft #4 Dayton Shaft #4 Pomeroy-Mason Shaft #2 Pomeroy-Mason Shaft #2 Depth (in) 36 60 96 132 156 6 66 182 GSI 40.9 EpIp (lb-in2) 5. the rock mass modulus is estimated using the proposed empirical equation.23E+12 5.1 and 3.5 61 61 61 42 42 Ei (ksi) 590 590 590 590 590 345 345 Em (ksi) 38. (5-4).43E+11 1.43E+11 1. The values of rock mass modulus for Dayton and Pomeroy-Mason test sites are estimated using the proposed empirical equation and they are summarized in Table 5-9.1.1 98. Eq. The calculated values of Em for the two drilled shafts in the North Carolina lateral load test is provided in Table 5-8.77E+13 .9 23. The other two load tests only have information on the lab test data of modulus of intact rock. The accuracy of the experimental p-y curves of the two tests have been verified in Chapter III.5 40. The mean GSI values determined in previous sections 3.23E+12 5. The modulus of rock masses at North Carolina test site was determined from pressuremeter test results.23E+12 5.1 98.

Ki Measured 407787 355468 454901 1016462 1557558 350938 383584 48112 144581 216130 Ki (psi). In general. The Poisson’s ratio is assumed to be 0. The shaft stiffness of test drilled shafts is computed based on reinforcement and dimension and shown in Tables 5-7 and 5-8.39 0. It can be seen that the predicted values are smaller than the measured values for most of cases./ Mea. The predicted Ki and measured Ki are plotted in Fig.2. 5-13.51 1. the empirical equation will provide conservative predictions on the initial modulus of subgrade reaction.86 0. The measured initial subgrade reaction modulus of rock masses obtained from the experimental p-y curves for rocks at various depths are provided in Table 5-10.55 0. 0.25 0.25 0.49 0.56 0. Table 5-10 Measured and Predicted Initial Modulus of Subgrade Reaction Test Dayton Shaft #4 Dayton Shaft #4 Dayton Shaft #4 Dayton Shaft #4 Dayton Shaft #4 Pomeroy-Mason Shaft #2 Pomeroy-Mason Shaft #2 NC Long Shaft NC Long Shaft NC Long Shaft Depth (in) 36 60 96 132 156 6 66 102 141 180 (psi).49 183 .3 for these tests.5.58 0.4. Predicted 199467 199467 392310 392310 392310 194066 194066 60229 83811 105159 Pred.2 Verification of the Empirical Equation The above field test data are used to verify the empirical equation for predicting initial modulus of subgrade reaction of rock mass. The predicted Ki using the proposed equation and the ratio of the predicted values over the measured values are provided in Table 5-10.

3 Numerical Solution for Laterally Loaded Piles in A Two Layer Soil Profile 5. piles are often embedded in layered soils. an elastic subgrade reaction solution is a good alternative for piles under small working loads because of its simplicity and easy for use. A closed form solution for laterally loaded piles in one layer soil with constant subgrade reaction modulus was given by Hetenyi (1946). A non-dimensional solution for soil stiffness linearly varying with depth was also presented by Reese and Matlock (1956) for piles in one layer soil.3.FEM & Field Data 3500 3000 2500 2000 1500 1000 500 0 0 1000 Field Data FEM Data Ki=Em(D/Dref)e (Ep Ip /Em D ) -2v 4 0. such as clay overlying on 184 .284 y=x 2000 3000 4000 Ki (ksi) .Empirical Prediction Figure 5-13 Validation of empirical equation using field test data 5.1 Introduction Although nonlinear p-y analysis is the most widely used method for design of laterally loaded piles.4000 Ki (ksi) . In reality. Subgrade reaction method models soils as Winkler springs and pile as a beam. however.

This may not represent the actual soil profiles. The general computational software Mathematica is used to carry out the numerical calculation. In this paper. although the solutions proposed by Davisson and Gill (1963) and Pise (1982) can be used. they require the soil stiffness to be constant along depth for each layer. For piles in layered soils under small working loads. For practical use. which makes the solution easy and efficient for use. especially for piles socketed in rock with overburden sands. which has been successfully used by Shen and Teh (2004) for a laterally loaded single pile in one layer soil.rock mass. is employed to numerically solve the problem of a laterally loaded pile in a two-layer soil system using subgrade reaction theory. can be either constant along depth or linearly varying with depth. Two case studies using the numerical solution for field lateral load tests are performed to further validate the practical applicability of the numerical solution. The proposed solution is then able to be used for layered soil condition with variation of stiffness. especially when soil stiffness is varying with depth. The stiffness of soil could be varying with depth (such as sand and sedimentary rock) or keeping constant along depth (such as clay). a variational approach. Finally. The numerical solution is validated against a non-dimensional solution (Reese and Matlock. 1956) for linearly varying soil stiffness and against Davisson and Gill (1963) non-dimensional charts for a two-layer soil system with constant soil stiffness. Realistic soil stiffness distribution along depth can be more accurately modelled comparing with existing solutions for a layered soil system. The soil stiffness. methods for determining modulus of subgrade reaction are provided. which is represented by modulus of subgrade reaction of soils. discussions on the applications 185 .

Table 5-11 Summary of Definitions Related to Subgrade Reaction Theory Description Soil resistance per unit length Pile deflection Pile diameter Depth Modulus of subgrade reaction Coefficient of subgrade reaction Constant of subgrade reaction Symbol p y D z K k nh Definition Pressure times diameter Dimension F/L L L L F/L2 F/L3 F/L3 K=p/y k=K/D nh = K/z. 5. 5-14.2 Definition of the Problem It is very common for piles embedded in layered soils. by choosing a zero or non-zero value for the two constants of subgrade reaction. respectively. 5-14. The soil stiffness can be treated as constant along depth for overconsolidated clay and igneous rocks. To make this Chapter easier to follow. it is reasonable to assume linearly varying soil stiffness along depth for sands and sedimentary rocks. Ls = the thickness of layer 1. in which Ks0. The four layered cases can be represented by case (a) shown in Fig. nhr = the constant of subgrade reaction of layer 1 and layer 2. respectively.3 if constant soil stiffness is used for the whole soil layer. nhs. Kr0 = the initial modulus of subgrade reaction of layer 1 and layer 2. the definitions of various terms related with subgrade reaction theory are summarized in Table 5-11. Four possible soil layer combinations are shown in Fig. However.of the proposed method show that the lateral deflection of a pile may be under estimated by a factor of 2. L = the total thickness of the two layers.3. nh = kD/z 186 .

pz is the soil resistance per unit pile length at depth z.3 Variational Solution Shen and Teh (2002) proposed a numerical solution for pile groups under lateral loads using a variational approach. This study is a further development of the variational approach to the layered soil conditions. Ht Mt pz Ls pz D L z Ks0 K nhs K1=Ks0+nhsz Kr0 nhr K2=Kr0+nhr(z-Ls) Figure 5-15 Soil-pile system 187 .3. the response of piles in the other three layered conditions can be obtained.Ks0 Ls L z K nhs Layer 1 Kr0 nhr Layer 2 (Case a) Ls L z Ks0 nhs=0 Kr0 nhr K Ks0 Ls L z K nhs=0 Kr0 Ks0 nhs Ls L z K Kr0 nhr =0 (Case c) nhr =0 (Case d) (Case b) Figure 5-14 Two-layer soil profile with four possible variations 5. 5-15. Here Ht and Mt are the applied lateral load and moment at the top of the pile. 5-14 will be presented. After that. respectively. The soil-pile system is schematically shown in Fig. Shen and Teh (2004) presented a variational solution for soil with stiffness increasing with depth. The solution for piles embedded in soils having the soil stiffness shown in case (a) of Fig. K1 and K2 are the modulus of subgrade reaction of layer 1 and 2. By changing the values of the two constants of subgrade reaction. and L and D are the length and diameter of the pile.

respectively. sin . where Ep is the pile Young’s modulus.. sin L L L L (5-15) 188 .3. In this study.. 5-15. the two finite series is also employed to closely match the pile deflections under lateral loads and moments. 5.3.5.2 Deflection Series Shen and Teh (2002) employed two finite series to model the deflections of a pile subjected to lateral loads and moments at the pile head respectively. and yz is the deflection of the pile at depth z.1 Potential Energy The total potential energy of the pile under lateral loads. (5-12) are the energy consumed by the soil resistance of the first and second soil layers. where yt is the deflection of the pile at the pile head.3. πz nπz T 2πz z } . The second and third terms on the right side of Eq. The deflections of a pile subjected to a lateral load at the pile head can be approximately described in matrix form by y zh = {Z h }T {δ h } (5-14) where {Z h } = {1. sin . The fourth and fifth terms are the input work done by the lateral load Ht and the moment Mt acting at the pile head. Ip is the moment of the inertia of the pile section. as illustrated in Fig. can be expressed as πp = U p + ∂y t 1 Ls 1 L 2 2 Mt ∫0 K 1 y z dz + ∫Ls K 2 y z dz − H t y t − 2 2 ∂z (5-12)  d2yz 1 L where U p = ∫0 E p I p   dz 2 2    dz   2 (5-13) Up is the elastic strain energy of the pile.3.

(5-14) and (5-17). b h .3 Minimization of Potential Energy The principle of minimum potential energy requires that πp be an extremum with respect to the admissible deflection field characterized by the undetermined coefficients in Eqs. cos 2L 2L 2L L (5-18) (5-19) {δ m } = {a m . respectively. 5..3.. β h 2 . and βhi and βmi are the unknown constants to capture the nonlinear mode of the deflected pile. β m 2 . (2n − 1) πz T πz 3πz z } . bh and bm are the unknown constants to capture the deflections due to the rotation of the pile. ah and am are the unknown constants to represent the rigid body movement of the pile under lateral loads and moments.. cos ..3. β h1 . cos . This leads to: ∂π p ∂δ i =0 (i = 1. The above expressions for pile deflections are validated in the subsequent section by the comparisons with the nondimensional solutions by Reese and Matlock (1956) and Davisson and Gill (1963).β hn }T (5-16) The deflections of a pile subjected to a moment at the pile head can be approximately represented in matrix form by y zm = {Z m }T {δ m } (5-17) where {Z m } = {1. The total deflection (yz) of the pile subjected to lateral loads and moments is the summation of yzh and yzm.n + 2) (5-20) 189 .... β m1 ..β mn }T In above equations. b m .2.{δ h } = {a h . and n is the number of terms used in the trigonometric function.

we have: ∂( L [ ∫0 E p I p ∂ 2 y zm ∂z 2 ∂δ mi ) 2 ∂ {Z m }T ∂z 2 L dz + ∫0 s ∂y ∂y L K 1 zm {Z m }T dz + ∫L K 2 zm {Z m }T dz]{δ m } = M t s ∂δ mi ∂δ mi ∂( ∂y mt ) ∂z ∂δ mi (5-23) where δhi and δmi are the constants in the vectors {δh} and {δm} respectively. (5-21).where δi denotes the undetermined coefficients in Eq. (5-21) is the governing variational formulation that a single pile in two layers soil/rock needs to follow. the potential energy can then be reduced to: ∂U p ∂δ i L + ∫0 s K1 y z ∂y ∂y z ∂y  ∂y  L dz + ∫L K 2 y z z dz = H t t + M t ∂ t  / ∂δ i s ∂δ i ∂δ i δi  ∂z  (5-21) Eq. (5-16) for the lateral load case and in Eq. (5-13) and (5-17) into Eq. we have: ∂( L [ ∫0 E p I p ∂ 2 y zh ∂z 2 ∂δ hi ) 2 ∂y ∂ {Z h }T ∂y ∂y L L dz + ∫0 s K 1 zh {Z h }T dz + ∫L K 2 zh {Z h }T dz]{δ h } = H t ht 2 s δ hi ∂δ hi ∂δ hi ∂z (5-22) For the case of piles under moment. (5-21) will be solved by separating the effects of lateral loads and moments on pile deflections. For the case of piles under a lateral load.4 Numerical Solution The Eq. 5. (5-21). The above two equations can be reduced to matrix forms as: [K H ]{δ h } = {H} [K M ]{δ m } = {M} (5-24) (5-25) 190 . (5-19) for the moment case. yht and ymt are the deflections of the pile at the pile head under the applied lateral load and moment respectively. respectively.3. By using Eq. (5-13) and (5-14) into Eq. by substituting Eqs.3. (5-16). by substituting Eqs.

where [KH] and [KM] are the matrices reflecting the pile-soil stiffness under a lateral load and a moment. respectively. the deflection of a pile. The moment of the pile at any depth can be calculated by double differentiating yz.3. After the two matrices [KH] and [KM] are obtained according to Eqs. respectively. respectively. yz. The accuracy of the present solution is verified with a non-dimensional solution by Reese and Matlock (1956) for a single soil layer with stiffness increasing with depth. The pile used for comparison has a diameter of 0. A try-and-error procedure can be used to find out the value of the generated moment by ensuring the zero rotation of the pile at the pile head. It is found that a size of 10×10 for the stiffness matrices [KH] and [KM] provides sufficiently accurate results. The above described procedure is for free-head pile condition. the moment developed at the pile head due to the fixity can be obtained based on the zero-rotation condition at the pile head. Correspondingly. {H} and {M} are the vectors reflecting the lateral load and moment applied at the pile head. For a fixed-head pile under a lateral load.1 kN-m predicted by Reese and Matlock (1956) and the proposed solution for various nh values are presented in Fig.4 Validations The above calculation procedure is easily coded into Mathematica. 5-16.6 to 152 MN/m3 are evaluated. nh. the deflections of a pile under a lateral load and a moment can be calculated using Eq. varying from 1. Four cases with a constant of subgrade reaction. then the vectors {δh} and {δm} can be calculated. (5-22) and (523). By superposing yzh and yzm.5 kN and a moment of 27. (5-14) and (5-17). a very good match is obtained. can be obtained. 5. The pile head deflections at a lateral load of 44.76 m and a length of 6 m. 191 . Apparently.

6 kN-m and fixed head with a lateral load of 89 kN are considered.5 kPa and 6895 kPa.32 m in a uniform soft clay. A hypothetical case of a 0.600 0. The pile has modulus of elasticity of 24.8 GPa. A subgrade reaction modulus.8 kPa is selected for the soft clay according to Terzaghi (1955) recommendations on K for stiff clay. respectively.305 m diameter concrete pile embedded 7.200 0. The top portion of the soft clay is then replaced with 1D (diameter of pile) to 4D thick of medium and stiff clay with Ks0 values of 3447.000 Reese & Matlock Proposed 0.800 Deflection (in) Figure 5-16 Comparison with Reese and Matlock solutions– varying soil stiffness For layered soil condition. Kr0. Davisson and Gill (1963) non-dimensional charts are employed to validate the proposed solution.600 500 400 nh (pci) 300 200 100 0 0. Both free head with a lateral load of 89 kN and a moment of 135. of 1723. Comparisons of deflections at the pile head and the maximum moments in the pile for cases with varying surface layer thickness Ls and varying modulus of subgrade reaction Ks0 by Davisson and Gill non-dimensional 192 .400 0. Constant modulus of subgrade reaction for each layer is required by Davisson and Gill (1963) method.

It can be seen that a good match for both deflection and maximum moment predictions is achieved.charts and the proposed solution are presented in Figs.5 1 0. Free Head 3. D = 1 ft. D = 1 ft.5 2 1.5 3 Deflection (inch) 2.5 0 0 Ks0=250 psi Ks0=500 psi Ks0=1000 psi Davisson and Gill Proposed 1 2 Ls /D 3 4 5 (a) Deflection 1300 Maximum moment (kip-in) 1200 1100 1000 900 800 700 600 500 0 1 2 Ls /D 3 4 5 Davisson and Gill Proposed Ks0=1000 psi Ks0=500 psi Ks0=250 psi Kr0=250 psi. Free Head (b) Maximum moment Figure 5-17 Comparisons with Davisson and Gill method for free head condition 193 . Kr0=250 psi. 5-17 and 5-18.

6 0. Fixed Head -900 Maximum moment (kip-in) -800 -700 -600 -500 -400 -300 0 Ks0=250 psi Ks0=500 psi Ks0=1000 psi Davisson and Gill Proposed 1 2 Ls/D 3 4 5 (b) Maximum moment Figure 5-18 Comparison with Davisson and Gill solution for fixed head condition 194 .8 Deflection (inch) 0.2 0.1 0 0 Kr0=250 psi.7 0.4 0.1 0. D = 1 ft.9 0.3 0.5 0.D = 1 ft. Fixed Head Ks0=250 psi Ks0=500 psi Davisson and Gill Proposed Ks0=1000 psi 1 2 Ls /D 3 4 5 (a) Deflection Kr0=250 psi.

the following alternatives can be used. especially rock with an overlying soil layer. 5. For clay. The constant of subgrade reaction nh can be determined from Table 5-12 proposed by Liang (2002). and 28.5.16 MPa. modulus of subgrade reaction can be determined using Carter (1984) equation as previously provided in Eq. If Carter (1984) equation can not be used due to lack of soil input parameters. (5-11) presented in previous section can be used to estimate modulus of subgrade reaction.3. modulus of subgrade reaction is usually assumed to be linearly varying with depth. This SPT correlation table was developed based on a sensitivity study on an extensive lateral load test database. is validated against the results of Shaft #1 of PomeroyMason lateral load test. modulus of subgrade reaction K is generally assumed to be constant with depth except the surface portion of the soil.3 MPa. 195 .5 Methods for Determining Input Parameters For soils. (5-6). The lateral load test was performed on a free-head drilled shaft with a diameter of 102 inch and a total embedment length of 792 inch.6 MPa for stiff. Terzaghi (1955) recommendations on K of 7. For sand. the empirical equation Eq.6 Case Study The applicability of the proposed solution for piles in layered soils. and hard clay can be used for rough estimation. 14.3. The value of nh of rock can be determined by fitting a straight line through calculated K values along depth. very stiff. The shaft was embedded in rock with 312 inch of overlying sand. For rock. The following case study will demonstrate the use of these soil or rock parameter determination tables and equations. The value of nh of soils can be determined by fitting a straight line through calculated K values along depth.

The shaft and soil information are depicted in Fig. The average SPT N value of the sand was 9. which fits the data points meanwhile ensuring a positive value for Kr0. Based on the values of Krz from the proposed empirical equation.9 psi and 21. as shown in Fig.5. (5-4) using modulus of intact rock core and mean GSI values are also provided in Fig. One set of Krz is directly obtained from the experimental p-y curves derived from strain data in Chapter III.5 pci for nhr. nhs. is taken as 38. The trendline.3 pci according to the correlation table by Liang (2002). The constant of subgrade reaction for the sand layer.8 inch Ht 102 inch z (inch) 312 457 618 690 792 SPT Sand N = 9. Kr0 and nhr are obtained to be 7452. 5-20.5 Em (psi) 6462 654 654 954 96 inch Ls=312 Shale Sandstone Claystone L=792 z (inch) Kr0 nhr Ks0=0 K (psi) nhs =20 pci Figure 5-19 Shaft and soil profiles of the case study Two sets values of modulus of subgrade reaction for various depths of weak rock mass are used for comparison. 519. It is assumed that the stiffness of the sand and the rock layers increasing with depth. Mt = Ht × 388. 5-19. The depth is measured from the top of rock.8 pci. The other set of Krz is correlated from Em using the proposed equation (Eq. respectively. 511). 196 . The modulus of rock mass for various rock depths obtained from Eq. gives a value of 7191 psi for Kr0 and a value of 273.

it can be concluded that the proposed solution can provide reasonable prediction on deflections of laterally loaded drilled shafts in a two-layer soil/rock system.Other required input parameters for the code programmed in Mathematica are given as: Ks0 =0. This conforms to the previous conclusion that the lateral response mainly depends on properties of the surface layer. From this case study.5 pci. 0 0 50 100 Ki (ksi) 150 200 250 300 350 Depth (in) 200 400 600 Ki from p-y curves Ki from proposed equation Figure 5-20 Interpretation of subgrade reaction modulus of rock 197 .8 pci to 273. Ep = 6522 ksi. it can be seen that the predicted deflection values do not change much when the nhr of the bottom rock layer is varied from 21. L = 792 inch. Additionally. It can be seen that the proposed method over predicts the deflections. 5-21. Ls = 312 inch. The comparison between the measured deflections from inclinometer readings and the predicted values using the proposed solution at the soil surface is shown in Fig. This is desirable from view point of safety margin.

Based on above work.based on K from proposed equation Figure 5-21 Comparison of shaft head deflections 5. Finally. the following conclusions are drawn: 1.6 Deflection at Surface (inch) 0. especially rock mass overlying with a layer soil. using variational approach. an empirical equation for estimating modulus of subgrade reaction modulus of rock mass is developed based on a parametric study using 3D FEM analysis. Next.300 Measured 250 200 Load (kip) 150 100 50 0 0 0. Dilatometer test is good for weak rock and the Goodman Jack test is good for hard and competent rock. an elastic numerical solution based on subgrade reaction theory is proposed for a two layer soil/rock profile. 198 .2 0.4 Summary and Conclusions In this chapter.4 0.8 Predicted based on K from p-y c r es Predicted . a methodology for determining the deformation modulus of rock mass is proposed.

Therefore. 3. 4. The validation using field test data on the proposed empirical equation for determining the modulus of subgrade reaction K shows that it can yield reasonable and conservative results. The validation using field lateral load test data shows that the proposed elastic numerical solution can provide reasonable results for small working loads. The proposed empirical equation for estimating the deformation modulus of rock mass can provide reasonable estimation on Em according to the field data from Ironton-Russell. 199 . 5.2. The proposed elastic numerical solution can match the results from rigorous subgrade reaction solution for an assumed one layer soil problem. the elastic solution provides an efficient way to preliminarily estimate the deflections of a drilled shaft in layered soil/rock under small working lateral loads.

Piecewise polynomial curve fitting technique was employed to deduce p-y curves from strain gage readings obtained during lateral load tests. The methods for determining pertinent rock parameters needed for constructing the p-y curves are presented. a hyperbolic p-y criterion is developed based on considerations of both theoretical derivations and numerical (finite element) analysis results. Although some materials have been presented in previous chapters. essential information is duplicated here to make this chapter easy to read. (6-1) and illustrated schematically in Fig. 6-1 and 6-2. The comparisons between the computed shaft responses and the actual measured are fairly good for the load-deflection curves at the load point as well as the maximum bending moment in the shaft. Two full-scale lateral load tests introduced in Chapter III are used to validate the applicability of the developed hyperbolic p-y curves for rock.1 General Shape of P-y Curve in Rock Two full-scale lateral load tests results of fully instrumented drilled shafts socketed into rock mass have been presented in Chapter III. 200 .CHAPTER VI P-Y CRITERION FOR ROCK MASS In this chapter. the general shape of these p-y curves for rocks could be fit mathematically by a hyperbolic equation given in Eq. 6. 6-3. As can be seen in Figs.

06 0.04 0.05 0.02 0.1 y (in) 6 in .p= y y 1 + Ki pu (6-1) where pu is the ultimate resistance of rock mass per unit shaft length.asymptote p (lb/in) 0.Initial Slope 36 in-Experimental 132 in-Experimental 36 in-Hyperbola Fit 132 in-Hyperbola Fit Hyperbolic Curve Pu .Hyperbola Fit Figure 6-2 P-y curves deduced from Shaft #2 of load test at Pomeroy-Mason 201 .Experimental 6 in . and Ki is the initial tangent slope to the p-y curves.08 0. 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 0 Ki .1 y (in) Figure 6-1 P-y curves deduced from shaft #4 of load test at Dayton 30000 25000 20000 p (lb/in) Hyperbolic Curve 15000 10000 5000 0 0 0.

where. W is the weight of the wedge. σ'v0 is effective overburden earth pressure on the top of rock and equals to zero if no overlying soil is present. the FEM simulation results show that the lateral loading of the shaft tends to fail the rock mass as a wedge as shown in Fig. Detailed investigation results have been presented previously in Chapter IV. 6-4. 202 . D is the diameter of the drilled shaft. Fnet is the total net rock resistance. Fa is the active earth force exerted on the drilled shaft. Fn is the normal force applied to the side faces and is assumed to be equal to at-rest earth force. H is the height of the wedge.2 Determination of pu 6.2.p pu Hyperbola Ki y Figure 6-3 Schematics of a hyperbolic p-y curve 6. Fsb is the friction force on the bottom surface. Fs is the friction force on the sides of the wedge. Fnb is the normal force on the bottom surface determined through force equilibrium in vertical direction.1 Failure Modes A 3D finite element model (FEM) using ABAQUS was used to investigate the mobilization mechanisms of resistance of rock mass near ground surface and at great depth. For rock mass existing near ground surface.

as shown in Fig. where τmax is the maximum side shear resistance between a drilled shaft and rock mass. and pL is the normal limit pressure of rock mass. 203 . whereas. FEM study shows that the normal stresses are very small at points A and D in Fig. The patterns of normal and shear stress distribution at the ultimate state are obtained from FEM simulation results and could be described as sine functions. the maximum shear stresses are observed at points C and E and negligible shear stresses are observed at points B and D. 6-5. the ultimate state (or failure) is defined when the rock has experienced maximum compressive stress equal to compressive strength and the rock-shaft side shear strength has reached. In particular. 6-5. considering the possible reduction of strength of brittle rock mass after the maximum stress reaches the peak strength of rock.σ’v0 θ Fsb Fnb Fs W β H Fnet Fa D Fn Figure 6-4 Failure mode for rock near ground surface For rock mass at great depth.

2. p u = 2C1 cos θ sin β + C 2 sin β + C3 cos β − 2C 4 sin θ − C 5 (6-2) where C1 = H tan β sec θ(c′ + K 0σ′v0 tan φ′ + H K 0 γ′ tan φ′) 2 (6-3) (6-4) (6-5) (6-6) (6-7) (6-8) C 2 = C3 tan φ′ + c′( D sec β + 2H tan β sec β tan θ) C3 = D tan β(σ′v0 + Hγ′) + H tan 2 β tan θ(2σ′v0 + Hγ′) + c′(D + 2H tan β tan θ) + 2C1 cos β cos θ sin β − tan φ′ cos β C 4 = K 0 H tan β sec θ(σ′v0 + 1 γ′H) 2 C5 = γ′K a ( H − z 0 )D (C5 ≥ 0) K a = tan 2 (45 − φ′ / 2) 204 .5.Normal stress pL Shear stress distribution B τ max C E δ A σ(δ) = pL sin δ α τ(α) = τ maxsin 2α D Active earth pressure pa pu Figure 6-5 Failure mode of rock at great depth 6. pu (in the unit of Force/Length) of the drilled shaft at depth H is given as follows (See Section 4.1 for more detailed derivation).2 Pu Near Surface For wedge type of failure. the ultimate rock resistance per unit length.

for example. pu. The distributions of the shear and normal stresses based on FEM analysis are depicted in Fig. γ' is the effective unit weight of rock mass. respectively. as in the case of shaft/rock interaction at great depth.K 0 = 1 − sin φ′ z0 = σ′ − v0 γ′ γ′ K a 2c′ (6-9) (6-10) (6-11) (6-12) β = 45+ φ'/2 θ = φ'/2 in which c' is the effective cohesion of the rock mass. Discussions of methods for determining pertinent rock mass properties will be discussed later in this Chapter. 6. the ultimate rock resistance. π 2 p u = ( p L + τ max − p a )D 3 4 (6-13) where pa = active horizontal earth pressure and is given by p a = K a σ′v − 2c′ K a (pa ≥ 0) (6-14) σ'v = effective overburden pressure at the depth under consideration including the pressure from overburden soils. at great depth can be calculated as follows.2. 205 . 6-5.2 of Chapter IV. then FEM simulation results reveal that the ultimate rock resistance is reached when both the maximum shear and normal pressure of rock mass have reached their limiting values τmax and pL. As previously shown in detail in Section 4.3 Pu at Great Depth When the existing vertical soil pressure is sufficiently high. φ' is the effective friction angle of the rock mass.5.

6. (6-2) is therefore valid for calculating pu. (6-2) can be directly used. the c' and φ' in Eq. the c' and φ' needed for Eqs. For rock mass failure at a great depth. Eq. Therefore. the above equations for pu can still be used after making the following adjustments. which has a unit of F/L2. For failure mode due to shearing failure through a weak plane. Eq. Eq. (6-3) needed for the two sides of the failure wedge can still be obtained using instantaneous cohesion and friction angle correlated from Hoek-Brown strength criterion.3 Initial Tangent to P-y Curve Ki The initial tangent to a p-y curve.4 Pu of Jointed Rock For jointed rock with a set of parallel weak planes which dominates the behavior of rock mass under lateral loads. 6. the bottom plane of the failure wedge can either be within the rock mass or on a weak plane. was recommended previously by Reese (1997) as follows. (6-13) for in-depth rock mass still works for jointed rock. Specifically. However. (6-4) and (6-5) should come from the results of lab tests on samples with a weak plane. It is therefore necessary to calculate pu for both failure modes and that the smaller value is used as the final pu. 6-4 can be used. the high overburden pressure will prohibit the possible sliding failure on the weak planes. For failure mode due to shearing failure of rock mass. the same wedge failure model shown in Fig. For rocks that follow the wedge type failure.2. K i ≈ k i Em 206 (6-15) . except that the angle β is the inclination of the weak plane and the values of c' and φ' for the bottom face of the failure wedge should be obtained from laboratory tests by shearing samples along a weak plane.

Dref = reference shaft diameter equal to one foot.1 Strength Parameters Hoek-Brown (H-B) strength criterion for rock mass (Hoek et al. The H-B strength criterion can be expressed as   σ'   ' ' σ1 = σ3 + σci  m b 3 + s  σci     a (6-19) 207 .where ki = a dimensionless constant. and diameter of shaft.. Reese (1997) further suggested the following empirical correlations: ki = 100 + 400 z/(3D) for 0 ≤ z ≤3D ki = 500 for z > 3D (6-16) (6-17) Although Reese’s equation is simple to use. 2002) possesses distinct advantages in that it considers the effect of secondary structures of rock (e.4 Rock Mass Properties 6. 6.. filling.g. Poisson’s ratio of rock. it nevertheless does not take into account other potential influencing factors. joint.284 (6-18) where Em = modulus of rock mass. such as relative stiffness between shaft and rock. The parametric study presented preveously in Chapter V has led to the findings that the following semi-empirical equation may work better. and spacing) on the rock strength. It should be noted that this strength criterion is not suitable for rock with one or two sets of dominant joints. K i = E m (D / D ref )e − 2ν   EpIp  E D4  m     0.4.

σci = uniaxial unconfined compressive strength of intact rock. Hoek (1990) further provides the method for estimating Mohr-Coulomb friction angle φ' and cohesion c' from Hoek-Brown strength criterion as follows.  2τ  φ′ = 90 − arcsin   σ′ − σ′  3  1 c ′ = τ − σ ′n tan φ′ (6-23) (6-24) where σ'1 can be obtained using Eq. 2000). et al.5m b σ ′ 2(σ1 3 ci m b σ ci ′ 2(σ1 − σ ′ ) 3 (6-25) τ = (σ ′n − σ ′ ) 1 + 3 (6-26) 208 . s and a = constants that depend on the characteristics of rock mass and can be estimated as follows (Hoek. For deep foundation excavation. 2002). (6-19). and σ ′n = σ ′ + 3 ′ (σ1 − σ ′ ) 2 3 ′ − σ ′ ) + 0.  GSI − 100  m b = exp  28 − 14D m i  r    GSI − 100  s = exp  9 − 3D   r   a= 1 1 −GSI / 15 + (e − e − 20 / 3 ) 2 6 (6-20) (6-21) (6-22) where mi is a material constant that can be obtained from Table 4-3 in Chapter IV. Dr is a factor that depends upon the degree of disturbance to which the rock mass has been subjected to due to blast damage and stress relaxation. σ'3 = the minor principal stress at failure. GSI = Geological Strength Index (Marinos and Hoek. Dr can be assumed to be zero. mb..where σ'1 = the major principal stress at failure.

(6-19) by equating σ'3 to σ'v.3 Rock Mass Modulus Em The empirical equation (6-18) requires input of Em in order to compute Ki. τ max = 5. τmax (MPa). it should be noted that the value of σ'3 can be taken as effective overburden pressure at the depth of 1/3H for estimating the values of φ' and c' using Eqs. then the following empirical equation derived in Chapter V could be used. Em = E i GSI / 21. (6-14) can be calculated using Eq. which is calculated using Eq. 6-2). proposed by Kulhawy and Phoon (1993) is adopted here. For in-depth failure mode.7 (e ) 100 (6-28) where Ei is the modulus of intact rock core that could be determined from unconfined compression test on rock core samples.4. 6. However.σ'3 is computed as vertical effective stress.5 where the units are in psi ci (6-27) 6.2 Rock-Shaft Interface Strength The empirical equation for estimating side shear resistance. The effective cohesion c' needed in Eq. if dilatometer tests could not be performed at the site. (6-23) and (6-24). as the side surface is triangular. Ideally.4.42σ 0. it is preferred to perform dilatometer/pressuremeter tests to obtain in-situ modulus of rock mass. pL is taken as the major principal stress at failure σ'1.5 Construction of P-y curves for Rock Mass To calculate pu according to wedge failure mode (Eq. (6-24) by equating σ'3 to σ'v. 209 . 6.

Vesic (1961)’s equation was 210 . As pointed out previously in Chapter II. (6-2) and (6-13). the physical appearance of the material recovered in the core can be used to estimate GSI value using the charts presented by Marinos and Hoek (2000). (2002)’s p-y criterion is mainly based on the equations of Vesic (1961) for calculating Ki and Zhang et al. GSI can be obtained from Rock Mass Rating (RMR) (see Hoek and Brown. Additionally. GSI values are based on the structure and the surface condition of rock mass. and mi. (2002). (2000)’s equation for calculating pu. 1997). The main parameters required for calculating pu. the following relationship can be established. are: σci (unconfined compressive strength of intact rock core).The value of pu is the smaller of the values calculated using Eqs. The value of mi can be obtained according to Marinos and Hoek (2000). other than the properties of a drilled shaft. Consequently. GSI = RMR76 GSI = RMR89-5 (6-29) (6-30) 6. If Bieniawski (1989) RMR89 is used. For cored rock. If Bieniawski (1976) RMR76 is used. Gabr et al. the value of 15 should be assigned to groundwater rating and adjustment for discontinuity value is set to zero. GSI. the equations for calculating the modulus of rock mass and ultimate resistance of rock mass are different. the rock mass should be assumed to be completely dry and a rating of 10 could be assigned to the groundwater. (2002) Although the proposed p-y criterion has same mathematic expression as that of Gabr et al.6 Comparison of the Proposed P-y Criterion with That of Gabr et al. and adjustment for discontinuity orientation value should be set to zero.

(2000)’s equation was based on a pure assumption of stress distribution of shaft-rock interaction. The main differences of the two p-y criteria are summarized in Table 6-1. limestone. The unconfined compressive strength of rock mass ranges from 19 psi to 9073 psi. (2002) are used for validating the developed p-y criterion.7 Case Studies The two lateral load tests reported in Chapter III and another two load tests by Gabr et al. However.5 ksi to 1292 ksi. The modulus of intact rock mass ranges from 1. the proposed p-y criterion employs 3D FEM studies to carefully identify the failure modes of rock mass under lateral loads and establish an empirical equation for estimating modulus of rock mass. and mudstone. The types of rock at these sites are shale. The soils above the rock at the test sites were either excavated or isolated using a casing with its diameter larger than the diameter of test drilled shafts. The diameter of test drilled shafts ranges from 6 ft to 8 ft.developed for the case of beam on elastic foundation. This was done to eliminate the influence of soils on the soil/shaft lateral interaction. while Zhang et al. claystone. sandstone. siltstone. and the rock-socket length ranges from 18 ft to 57 ft. 211 . 6.

(2000). Poisson’s ratio of rock.  EpIp   K i = E m ( D / D ref )e − 2ν  modulus of  E D4   m  subgrade This empirical equation is reaction developed based on a parametric study using 3D FEM analysis which is validated against the Dayton lateral load test results. the following procedure is used: 4   For z < T0.65E  ED    2 1 / 12 1 − ν  EpIp    where T0 is depth of point of rotation. of rock per π 2 p u = ( p L + τ max − p a )D for unit length pL is calculated using Hoek-Brown 4 3 of shaft strength criterion and in-depth layer pL is calculated using Hoekτ max = 0. The value of Ki at a depth greater than T0 is increased by an empirical factor IT which is equal to -28 – 383 log (T0/L).Table 6-1 Comparison of P-y Criteria Proposed p-y criterion P-y expression p= y y 1 + Ki pu P-y criterion by Gabr et al. Em.2σ 0. This was proposed surface resistance by Zhang et al. (2002) p= y y 1 + Ki pu 0. and T0 = [1+0. modulus of rock mass. It is validated against field test data at Dayton. K i = 0. This equation was developed based on IrontonRussell dilatometer test data and historical data by Bieniawski (1978) and Serafin and Pereira (1983). and shaft-rock relative stiffness on Ki are considered.45σ 0. Effects of diameter of drilled shaft. Pomeroy-Mason. This equation predicted 6.6 to 10 times of the measured Em according to the dilatometer test data from IrontonRussell project. ultimate Eq.5 (MPa ) . K i = I T 0. Empirical equation proposed by Hoek and Brown (1997) is used as follows. (2001).7 ) can be rock mass 100 used.18log(EpIp/Em/L4)]L. ci Brown strength criterion and τmax is based on Kulhawy and Phoon (1993) equation: τ max = 0.284 Ki. 4 For z>T0. Pu. Empirical equation modulus of E E m = i (e GSI / 21. ci 212 . (6-2) for rock layer near pu = (pL + τmax) D.5 (MPa ) . σ ci (GSI−10) / 40 Em = 10 (GPa ) where 100 σci is unconfined compressive strength of rock core sample.65E  ED  2 1 / 12 1 − ν  EpIp    This equation was originally proposed by Vesic (1961) based on a solution for beam on elastic foundation. and North Carolina reported by Cho et al. When dilatometer tests are performed to obtain Em.

1. The gray shale is slightly weathered to decomposed.5 40. 6-6.038 0.038 0. Non-linear shaft stiffness option in LPILE is used to take into 213 .1 of Chapter III.1..1 Dayton Load Test The drilled shafts tested at Dayton site are 6 feet in diameter with 18 feet rock socket in shale. et. Details of this full-scale lateral load test were reported in Section 3.038 0. and very thinly laminated. A total of 36 #11 primary rebars and #6 spirals are used to form a reinforcement cage.1. Poisson’ ratio = 0.038 0. The determination of the values of σci. weakly calcareous.7. Ep = 3961000 psi. The test set up and the instrumentation installed in the drilled shafts were depicted previously in Fig. The parameters used for generating the proposed p-y curves for rock are summarized in Table 6-2. Table 6-2 Input Rock Mass Parameters of Dayton Load Test Depth (in) 36 60 96 132 156 γ' (pci) 0.5 61 61 61 mi 6 6 6 6 6 Ei (ksi) 590 590 590 590 590 The p-y curves generated from hyperbolic criterion and tabulated rock properties are presented in Fig. al. The limestone interbeds are typically less than 1 ft thick.038 σci (psi) 5668 5668 5668 5668 5668 GSI 40. 2004) to compute the response of the test shaft #4 under the applied lateral loads. Ip = 1319167 in4.3.1. GSI and Ei has been discussed previously in Section 3.6. 3-2. The values of mi are obtained from Table 4-3 based on rock types previously described in section 3. They are fed into LPILE computer program (Reese. The pertinent properties for drilled shafts are as follows: D = 6 ft. The rock consists of soft to medium gray shale interbedded with hard gray limestone.

it can be seen that the predictions are comparable with those measured values. Fig. 6-8. a good match of the maximum moments in the drilled shaft at various lateral load levels applied at shaft head is achieved between the measured and computed. 25000 36 in 20000 p (lb/in) 15000 10000 5000 0 0 0. The predicted load-deflection curve at the shaft head is compared with the measured in Fig.1 0. In general.15 132 in Figure 6-6 Hyperbolic p-y curves of Dayton site 214 . The comparison between the measured deflection-depth curves from inclinometer readings and those predicted from LPILE analysis is provided in Fig. the predicted deflections are larger than the measured values (on safe side). 6-10 shows the comparison of the measured moment-depth curves from strain gage readings and those predicted form LPILE analysis. a good agreement between the measured and the predicted can be observed. 6-9.05 y (in) 0.account of the evolution of stiffness due to concrete cracking. From Figs. 6-7. As shown in Fig. Although there is some discrepancy between the two curves at small load levels. 6-9 and 6-10. The modulus of rock mass is obtained by correlation using the empirical equation (6-28).

1 0. Moment (Kip-in) 40000 Measured Predicted Figure 6-8 Comparison of load-Maximum moment of test shaft #4 at Dayton load test 215 .1200 1000 Load (kips) 800 600 400 200 0 0 0.15 Deflection (in) Figure 6-7 Comparison of load-deflection of test shaft #4 at Dayton load test Measured Predicted 1200 1000 Load (kips) 800 600 400 200 0 0 10000 20000 30000 Max.05 0.

15 Figure 6-9 Comparisons of deflection-depth curves of shaft #4 at Dayton test Moment (kip-in) -15000 0 2 4 6 Depth (ft) 8 10 12 14 16 18 20 786 kip .Predicted -5000 5000 15000 25000 Figure 6-10 Comparisons of moment-depth curves of shaft #4 at Dayton test 216 .1 0.Predicted 0 0.Predicted 331 kip .05 0.Deflection (in) -0.Measured 634 kip .05 0 2 4 6 Depth (ft) 8 10 12 14 16 18 20 1126 kip .Predicted 331 kip .Measured 786 kip .Predicted 634 kip .Measured 1126 kip .Measured 331 kip .Measured 331 kip .Predicted 510 kip -Measured 510 kip .

Table 6-3 Input Rock Mass Parameters of Pomeroy-Mason Load Test Depth (in) 6 66 126 210 314 391 504 648 γ' (pci) 0.059 0.3 826. GSI and Ei has been discussed previously in Section 3.059 0.059 0. The details of the load test set up and instrumentation were previously shown in Fig.1.2.8 344.049 0. 3-10.5 1. The values of mi are obtained from Table 4-3 based on rock types previously described in section 3. The determination of the values of γ'. The bedrock core recovered was described as shale with interbedded siltstone.2 of Chapter III.059 0. The modulus of rock mass is determined by the empirical correlation equation (6-28).9 feet long.5 81 .1.060 0.1.2 217 GSI 42 42 42 42 45 38 28 44 mi 6 6 6 6 17 4 4 4 Ei (ksi) 344. Poisson’s ratio of the rock mass is assumed to be 0.7.2 Pomeroy-Mason Load Test The results of a lateral load test on two drilled shafts socket in shale near Ohio River between Pomeroy.055 σci (psi) 3797 3797 3797 3797 9073 19 44.8 344.8 1292 1. σci. in which rock socket length was 56. To eliminate soil and shaft interaction. The test drilled shaft was 8. The 28-day unconfined compressive strength of concrete was 5115 psi. West Virginia were reported in Section 3. The depth shown in Table 6-3 is measured from the top of the bedrock. The diameter of the rock socket was 8 feet.8 344.5 feet in diameter and 112. a 1 inch thick casing with 8.8 feet. The properties of rock used to generate p-y curves are summarized in Table 6-2. Ohio and Mason.047 0.6. A total of 28#18 bars and #6 ties at 12 inch spacing are used to form the reinforcement cage. sandstone and mudstone.3.5 ft diameter and a second casing with 11 ft diameter were installed in the top 56 feet of overburden soil.

is less than 34%. 6-12. 6-14. The effect of 1 inch thick casing for the drilled shaft above rock is included in the computer analysis.The p-y curves generated from hyperbolic criterion and tabulated rock properties are presented in Fig. A good agreement between the measured maximum moment from strain gage readings and the LPILE computed maximum moment in the shaft is achieved. The computed load-deflection curve at the loading point of the test Shaft #2 is compared with the actual measured in Figs. it can be seen that the predicted values match those measured. 6-11. 6-14 and 6-15. 218 . as shown in Fig. The comparison between the measured deflection-depth curves from inclinometer readings and those predicted from LPILE analysis is provided in Fig. 6-15 shows the comparison of the measured moment-depth curves from strain gage readings and those predicted form LPILE analysis. 6-13. Fig. Although LPILE with the input p-y curves under-predicts the deflections at large loading levels. The computer program LPILE with the generated p-y curves is used to compute the response of the test shafts under the applied lateral loads. the error defined as the ratio of the deflection difference divided by the measured deflection. From Figs. It can be seen that a good match between the predicted curve and the measured curve is achieved for the initial portion of the curve.

4 0.1 0.3 0.2 y (in) Figure 6-11 Hyperbolic p-y curves of Pomeroy-Mason site 6 in 66 in 391 in 0.30000 25000 20000 p (lb/in) 15000 10000 5000 0 0 0.5 300 250 Load (kips) 200 150 100 Measured 50 0 0 1 2 Deflection (in) 3 4 Predicted Figure 6-12 Comparison of load-deflection at the loading point for Pomeroy-Mason test 219 .

Measured 50 kip .300 250 Load (kips) 200 150 100 50 0 0 50000 100000 150000 Max.Predicted 150 kip . Moment (Kip-in) 200000 Measured Predicted Figure 6-13 Comparison of load-Maximum moment of Pomeroy-Mason load test Deflection (in) -1 0 20 40 Depth (ft) 60 80 100 120 275 kip .Measured 150 kip .Predicted 0 1 2 3 4 5 Figure 6-14 Comparisons of deflection-depth curves of shaft #2 at Pomeroy-Mason test 220 .Measured 275 kip .Predicted 50 kip .

The reinforcement was 12 #10 primary rebars and a 0. The drilled shafts are 2.3 Load Tests at North Carolina Two lateral load tests at North Carolina were reported by Gabr et al. The parameters used for generating the proposed p-y curves for rock are reported by Gabr et al. 221 .Predicted 150 kip .8 feet rock socket in siltstone and sandstone.7. (2002).Measured 50 kip .Deflection (in) -50000 56 66 76 Depth (ft) 86 96 106 116 275 kip .Predicted 0 50000 100000 150000 200000 Figure 6-15 Comparisons of moment-depth curves of shaft #2 at Pomeroy-Mason test 6.5 feet in diameter with 9.Measured 275 kip .Measured 150 kip .15 feet to 13. The rock consists of soft to medium hard siltstone and sandstone.Predicted 50 kip .5 inch thick casing. 6-16 and 6-17 for the load test at I-40 and I-85. The test set up is depicted in Figs. (2002) and summarized in Table 6-4 and Table 6-5 for load test at I-40 and I-85. respectively. respectively.

046 0.3 ft Dial Gage Jack 0.5 ft 3.5 inch casing 12 #10 Rebar Short Shaft 2.092 0.046 σci (psi) GSI Short Shaft 1639 87 1769 74 5061 76 Long Shaft 1769 57 4002 66 3756 76 3538 74 222 mi 9 9 19 9 14 14 14 Em (psi) 23345 28362 63351 25230 57203 54201 50620 .8 ft 3.092 0.3 ft Rock 11 ft 13.3 ft Long Shaft Figure 6-16 Layout of I-40 load test 3.092 0.5 inch casing 12 #10 Rebar Short Shaft 2.5 ft 13.3 ft Rock 9 ft Long Shaft Figure 6-17 Layout of I-85 load test Table 6-4 Input Rock Mass Parameters of I-40 Load Test Depth (in) 35 90 121 24 75 122 150 γ' (pci) 0.092 0.3 ft Dial Gage Jack 0.3.092 0.

a good agreement between the measured and the predicted can be observed.Table 6-5 Input Rock Mass Parameters of I-85 Load Test Depth (in) 23.6 61 90. 6-18 and 6-19 for I-40 short and long shaft.2 29. Non-linear shaft stiffness option in LPILE is used to take into account of the evolution of stiffness due to concrete cracking. respectively.055 0.055 0. respectively.. 2004) to compute the response of the test shafts under the applied lateral loads.7 112 145 γ' (pci) 0. 6-20 and 6-21. et. especially for the case of long shaft.055 σci (psi) GSI Short Shaft 4220 59 3596 59 6598 59 Long Shaft 3625 38 4162 38 4785 38 4785 59 mi 9 9 9 9 9 9 9 Em (psi) 15646 13340 48749 32509 15428 167026 87667 The p-y curves generated from the hyperbolic criterion and tabulated rock properties are fed into LPILE computer program (Reese. 223 .055 0. the discrepancy is not very significant.055 0. The predicted load-deflection curve at the shaft head is compared with the measured in Figs. The prediction results of I-85 short and long shaft are presented in Figs. al.055 0. However. Smaller deflections are predicted for this load test.055 0. In general.5 78.

6 Deflection (in) Measured Predicted Figure 6-18 Comparison of load-deflection curves of I-40 short shaft 400000 350000 300000 Load (lb) 250000 200000 150000 100000 50000 0 0 0.4 0.2 0.6 0.4 Deflection (in) 0.2 0.400000 350000 300000 Load (lb) 250000 200000 150000 100000 50000 0 0 0.8 Measured Predicted Figure 6-19 Comparison of load-deflection curves of I-40 long shaft 224 .

5 1 Deflection (in) 1.4 Deflection (in) 0.5 2 Measured Predicted Figure 6-20 Comparison of load-deflection curves of I-85 short shaft 350000 300000 250000 Load (lb) 200000 150000 100000 50000 0 0 0.350000 300000 250000 Load (lb) 200000 150000 100000 50000 0 0 0.6 0.8 Measured Predicted Figure 6-21 Comparison of load-deflection curves of I-85 long shaft 225 .2 0.

226 . Hoek-Brown strength criterion for the rock mass as well as pertinent empirical equations for the rock/shaft interface strength have been utilized in deriving the analytical equations. 3-D finite element simulation results have provided basic understanding of the mobilization mechanisms of lateral resistance of rock mass to the drilled shafts. moment. Evaluations based on comparisons between the predicted and measured responses of full-scale lateral load tests on fully instrumented drilled shafts have shown the practical uses of the proposed p-y criterion and the associated methods for determining the corresponding input of rock properties. and shear responses of the shaft under the applied lateral loads.6.8 Conclusions A hyperbolic p-y criterion for rock is developed in this study that can be used in conjunction with either COM624P or LIPLE computer analysis program to predict the deflection. Proper considerations for the effects of joints and discontinuities on the rock mass modulus and strength are discussed. from which analytical equations are derived for computing pu.

For hard rock. beside the control unit. The widely recognized definition of pressuremeter is as follows: a cylindrical probe that has an expandable flexible membrane designed to apply a uniform pressure to the walls of a borehole. Two calibrations are needed to correct the raw data. The pressure calibration of the probe is performed at ground level. The probe is confined by placing it in a steel casing. to establish the pressure-volume relationship of the probe itself. Usually. including the probe. The pressure-volume relationship of the 227 . Figure 7-1 and 7-2 show respectively the pressuremeter and dilatometer manufactured by RocTest. unconfined. which shares the same testing principle as the pressuremeter. can be used. coaxial tubing and control unit circuitry.1 Pressuremeter and Dilatometer The term pressuremeter was first used by Menard to describe his testing equipment developed in 1955.CHAPTER VII DERIVING P-Y CURVE FROM DILATOMETER TESTS 7. the raw data collected during the pressuremeter/dilatometer test consists of the pressure read on the gauge of the control unit PR and the volume read on the volume measuring device of the control unit VR. Pressuremeter is usually used for the testing in soils and weak rock due to the limit pressure that can be applied (10 MPa). The second calibration is the volume calibration of the complete system. the high limit pressure (30 MPa) testing device dilatometer. This is a measure of the probe inertia.

7-3. The calibration of volume and pressure losses is schematically presented in Fig. Probe Control Unit Figure 7-1 Menard G-Am pressuremeter from RocTest Manual Pump Probe ACCULOG-X Readout Figure 7-2 PROBEX-1 Dilatometer from RocTest 228 .system is determined. This calibration is a measure of the intrinsic volumetric expansion of the components under pressure.

the equations used for correction is slightly different.The measured pressure and volume can be corrected using the following equations per ASTM D 4719. Pc is the pressure correction due to stiffness of instrument at corresponding volume. However. PR is the pressure reading on the control unit. the two calibration procedures are the same as those of pressuremeter. P = PR + P δ − Pc V = VR − a (PR + Pδ ) where P is the corrected water pressure exerted by the probe on the soil. VR (7-1) (7-2) Pressure calibration curve VR Pc Volume calibration curve Slope = a Pressure in Probe. PR+Pδ Figure 7-3 Calibration for volume and pressure losses For dilatometer test data. Pδ is the hydrostatic pressure between control unit and the probe. determined in Fig. Volume Injected in Probe. Due to the 229 . V is the corrected increase in volume of the measuring portion of the probe. 7-3. VR is the volume reading on readout device. 7-3. and a is the slope shown in Fig.

the pressure readings can be corrected using the following equation. The expansion of the thick wall metallic tube is determined theoretically and is expressed by the “b” parameter. The following equation can be used to correct the volume readings. and Etube is the modulus of elasticity of the calibration tube material. m is the Poisson’s ratio of the calibration tube material. e is the wall thickness of the calibration tube.nature of dual-action oil-type hydraulic pump system. The PROBEX-1 dilatometer from RocTest is supplied with a standard. They are the intrinsic volumetric expansion of the dilatometer system known as “c” and the small expansion undergone by the calibration tube during pressurization. therefore the volume change of the steel casing needs to be considered. 7-3.955PR + 8.82∆h (7-3) where ∆h is the difference in elevation between the manual pump and the center of the dilatometer probe in meters.57 × 10-6 cm3/kPa. Assuming a dilatable length of the PROBEX-1 dilatometer membrane equal to 18 inch (457 mm). and b= 2Vtube [r + e(1 + m)] E tube e (7-5) where Vtube is the volume taken by the dilatable membrane of the probe of the dilatometer when in contact with the metallic calibration tube. one obtains a “b” value of 84. Because the volume calibration of dilatometer test reaches high pressure. steel calibration tube. 230 . P = 0. r is the internal radius of the calibration tube. V = VR − cP (7-4) where c = a – b. Two deformation components contribute to the value of “a” shown in Fig.

and p-y curve can be obtained approximately by assuming that the expansion of the pressuremeter probe is considered to be the expansion of an infinitely long cylinder in an elastic infinite mass of soil/rock.A typical corrected pressure-volume curve is presented in Fig. and Vm is the corrected volume in the center portion of the ∆V volume increase. shear strength. ∆V is the corrected volume increase in the center part of the straight line portion of the pressure-volume curve. Vi is the corrected volume reading at the pressure of Pi. V Test curve Vm Vi ∆V Creep curve ∆P Py Pi Corrected Pressure Applied to Borehole Wall. 7-4. P Figure 7-4 Typical pressuremeter/dilatometer test curve 231 . With the corrected P-V curve. Corrected Volume. the creep curve is pressure versus the volume difference between the volume measured at 60 seconds and 30 seconds. the soil/rock shear modulus. corresponding to ∆P pressure increase. Young’s modulus. ∆P is the corrected pressure increase in the center part of the straight line portion of the pressure-volume curve. The values of Pi and Py can be determined from the intersections of the creep curve. where Pi is the corrected pressure when the probe made contact with the borehole. Py is the yield pressure at the end of the straight line portion of the pressure-volume curve.

the pressuremeter/dilatometer modulus E of rock mass or soils can be determined as follows (ASTM D 4719): E = 2(1 + υ)(V0 + Vm ) ∆P ∆V (7-6) where υ is Poisson’s ratio. 7. the limit pressure (Pl) is defined as the pressure where the probe volume reaches twice the original soil/rock cavity volume. empirical correlations (Menard. The limit pressure is usually not obtained by direct measurements during the test due to the limit of the probe expansion or the need for excessively high pressure. 7. the yield pressure method. the limit pressure can be determined by a 1/V versus P plot.1. 232 .7.1 Modulus of Rock Mass By assuming the rock mass is an elastic media and the cavity is expended in infinitely long. and the Gibson-Anderson method. as shown in Fig.1.2 Limit Pressure According to ASTM D4719. If the test was conducted to read sufficient plastic deformation. 7-5. 1970). defined as the volume V0+Vi where Vi is defined in Fig.1.3 Undrained Shear Strength The undrained shear strength Su of cohesive soils or soft rock can be determined in various ways. 1954). 7-4.. and V0 is the initial volume of the measuring portion of the uninflated probe at the ground surface. The volume reading at the twice of the original soil cavity volume is (V0 + 2Vi). including the limit pressure method (Bishop et al.

5 for typical ratio of E/Su varies from 200 to 2000 for clay.004 Figure 7-5 Determination of limit pressure from inverse of volume versus pressure Bishop et al.0 200.002 1/V (1/cc) 0. The yield pressure method was based on the theoretical expression of the yield pressure Py: 233 . Menard (1970) proposed a value of 5.5 for computing the residual strength of clay.0 0 Pl Corrected Pressure (kPa) Linear Extropolation 1/(V0 +2Vi ) 0.1000.0 100.2 to 7. Equation (7-7) can be rewritten as: Su = Pl * β (7-8) where the net limit pressure Pl* is equal to Pl-Pi.001 0.0 600.003 0. 7-5.0 900.0 800.0 500. The value of β can range from 5.0 300. and Pi is the horizontal in-situ earth pressure at rest determined from Fig. 7-4.0 400.0 0.0 700. (1945) derived a theoretical expression of the limit pressure using plasticity theory with Tresca criterion:  E Pl = Pi + S u 1 + ln  2(1 + ν)S u      (7-7) where Pl is the limit pressure determined from Fig.

7. However. and V is the current volume of the cavity. Based on Baguelin et al. none of them is very satisfactory (Briaud. The slope of the straight line is Su. ∆V is the increase in cavity volume. The Gibson-Anderson (1961) method is based on the theoretical expression of the pressuremeter curve after the yield pressure:  G ∆V  P = Py + S u  ln  S + ln V   u   (7-10) where P is the corrected pressuremeter pressure.25(Pl − Pi ) 0. G is the shear modulus. These 234 . however. This method is not recommended as it typically overestimates the undrained shear strength (Briaud.4 Friction Angle There are several ways of obtaining the friction angle φ of a cohesionless soil from a pressuremeter test. 1992). Baguelin et al. it should be noted that this equation was based on test data on clay. S u = 0. (1978) presented an extensive comparison of undrained shear strength Su and Pl* based on pressuremeter tests on clay. 1989) and is suggested for future use as it is based on an extensive database and is simple for use. A plot of P versus ln(∆V/V) for the pressuremeter test data points past the yield (or creep) pressure leads to a straight line. 1992).S u = Py − Pi (7-9) Both Py and Pi are obtained from the pressuremeter curve. (1978) database and Briaud’s own data.1.75 with units of ksf (7-11) This equation was adopted by FHWA (Briaud. Briaud (1989) developed a nonlinear relationship between Su and Pl* by performing a regression analysis.

This expression is:  ∆R c C   K (1 − sin ϕ′) + 1 + sin ϕ′  log  log(p − u 0 ) + cons tan t   R + 2=   2 sin ϕ′   c   (7-14) where Rc is the initial radius of the cavity.5(1− K a ) (7-13) where Ka is the coefficient of active earth pressure. 1977). The yield pressure method employs the theoretical expression of the effective stress yield pressure Py’: ′ ′ Py = Pi (1 + sin ϕ) (7-12) where Pi’ is the effective horizontal pressure at rest. 1992). the limit pressure method (Briaud. This method is not used because of a few shortcomings as pointed out by Briaud (1992). et al.methods include the yield pressure method (Briaud. G is shear modulus of the soil. The limit pressure Pl’ can also be used to theoretically obtain the friction angle. C is the intercept of the volumetric strain versus engineering shear strain plot. the value of G needs to be evaluated. For instances. and the Hughes-Wroth-Windle method (Hughes. ∆Rc is the change in radius of the cavity. The Hughes-Wroth-Windle method (Hughes. This method is not used because it is too difficult to determine Py’ with enough precision. P is the total 235 . et al. 1992).  G   Pl ′ = Pi ′ (1 + sin ϕ)  P ′ sin ϕ   i  0. 1977) takes into consideration the dilatancy of cohesionless soils. the method assumes no volume change of soil. Additionally. It is based on theory of expansion of infinitely long cavity while actual pressuremeter probe has a finite length. The method is based on the theoretical expression of the pressuremeter curve past the yield pressure.

perfectly plastic. s= 2 sin ϕ′ K (1 − sin ϕ′) + 1 + sin ϕ′ (7-16) It should be noted that this technique was developed for self-boring pressuremeter test results and unproven for preboring pressuremeter tests. 236 . and the slope s of the line at large deformation is a function of φ’ only.pressuremeter pressure. Haberfield and Johnston (1993) presented a curve fitting technique to estimate the strength parameters. 1986) Soil Type φ’CV (˚) Soil Type Well-graded gravel-sand-silt 40 Uniform medium sand Uniform coarse sand 37 Well-graded fine sand Well-graded medium sand 37 Uniform fine sand Assign lower values for well-rounded particles. Recommended values of φ’CV are presented in Table 7-1. the value of C can be approximately assumed to be zero. the value of C is zero. thus allowing determination of φ’. Assign higher values for angular particles. φ’ is the peak friction angle and K is: ϕ′   K = tan 2  45 ° + CV  2   (7-15) where φ’CV is the friction angle at constant volume. for medium and loose sands. Table 7-1 Preliminary estimates of φ’CV (Robertson and Hughes. The results of the pressuremeter test are plotted as log(P-u0) versus log(∆Rc/ Rc). φ’CV (˚) 34 34 30 For weak rock. The method is based on the theory of expansion of an infinitely long cylindrical cavity in an elastic. homogeneous and isotropic Mohr-Coulomb material with cohesion c and internal friction angle φ. For dense sand. u0 is the pore water pressure.

the expansion of the cavity is governed by: εc = P − Pi 2G (7-17) where εc is the cavity strain. Pi is the in situ horizontal stress and G is the shear modulus of the weak rock. r0 is the original radius of the cavity.The drained conditions are assumed and all parameters are expressed in effective stress parameters. After yielding. Before yield. P is the probe pressure. and b1 =  − 2m  (1 + mn ) (1 − ν) (m + n ) − ν  (Py − Pi ) m −1   (m + 1) (Py − Pi ) (m + n ) (7-19) b 2 = 2n (1 − ν) b 3 = (1 − 2ν) ry (7-20) (m + 1) (Py − Pi ) (m − 1) m /( m −1) (7-21)  P(m − 1) + σ  ˆ =  ˆ r0  Py (m − 1) + σ   1 + sin ϕ 1 − sin ϕ (7-22) m= (7-23) n= 1 + sin ψ 1 − sin ψ 237 (7-24) . the cavity expansion for pressure in excess of the yield pressure Py is given by: 1   ry  b1  εc = 2G   r0       ( m −1) / m  ry + b2  r  0     ( n −1) / n  + b3    (7-18) where ry is the radius of the yield zone around the cavity.

φ. (1978) in English.65) α + 3α = k E 18 (D<0.ˆ σ= 2c cos ϕ 1 − sin ϕ (7-25) where ψ is the dilation angle of the weak rock.  1 2 D D 0  2. ψ. Given values of G. Menard et al. ν. the value of first slope of p-y curve k was proposed as follows.65 =  k 9E D0   α  + D (D>0. this method is difficult to use without the aid of a computer program. This method was presented by Baguelin et al. ψ. and Pi. This method requires adjusting six parameters G.2 Deriving p-y Curves from Pressuremeter/Dilatometer Test Results There are several methods available for deriving p-y curves from pressuremeter test results. all these methods were developed based on pressuremeter tests in soils.6m or 2 feet) (7-27) 238 . The applications of these methods for pressuremeter tests in rock have not been verified yet. the values of c and φ can be obtained by fitting the pressuremeter test curve. 7. (1969) developed a method for deriving p-y curves from preboring pressuremeter test. and ν is the Poisson’s ratio. However.6m or 2 feet)  6E  α (7-26) 1 D 4(2. However. and Pi to fit pressuremeter curves. c. It considers that the p-y curve to be bilinear elastic and perfect plastic. Based on Menard’s analysis on settlement of a strip footing. ν.

the value of k.. E = the modulus of soils from pressuremeter test. the first linear line stops at this point. α = rheological factor. kp. For soils above the critical depth. and pu should be reduced by 239 . For soils above the critical depth zc. Table 7-2 The Rheological Factor α for Various Soils (Baguelin et al. 1978) Soil Type Overconsolidated Normally consolidated Weathered and/or remolded Rock Peat E/Pl* α Clay E/Pl* >16 1 9-16 7-9 α 1 2/3 1/2 >14 8-14 Silt E/Pl* α 2/3 1/2 1/2 Sand E/Pl* >12 7-12 α 1/2 1/3 1/3 Slightly fractured or extremely weathered α = 2/3 Sand and gravel E/Pl* α >10 6-10 1/3 1/4 1/4 Extremely fractured α = 1/3 Other α = 1/2 The typical p-y curve by Menard et al.where D = the diameter of piles. the soil reaction is less than at great depth because of possibility for soil heave. When the soil pressure p/D (p is soil resistance per unit length of pile. the subgrade reaction will be reduced by a factor λz. the reference diameter D0 = 0. λz = 1 + (z / z c ) 2 (7-28) where z = depth. 7-6. which is dependent on the soil type and the ratio E/Pl* (Pl* is the net limit pressure). (1969) is shown in Fig. which is in the order of 2D for cohesive soils and 4D for granular soils. as given in Table 7-2. The ultimate unit soil reaction pu/D is equal to the limit pressure pl of pressuremeter test for the depth greater than the critical depth. and D is the shaft diameter) reaches the yield pressure Py. The slope value kp of the second linear part is half of the k of the first part. Near the ground surface.6 m (2 feet).

5k pu/D = Pl/2 z=0 pu/D = Pl z ≥ zc 0 y Figure 7-6 P-y curves from pressuremeter (Baguelin et al. This method tends to give deflections larger than the measured deflection often by a factor of two (Baguelin et al. et al.. and Pi is the in-situ at-rest horizontal stress. (1978) method was incorporated in a more complete form in a design manual by the French Petroleum Institute in 1983.. It uses the results of selfboring pressuremeter tests. 240 .the factor λz. 1978). η is the lateral resistance factor varying from 0.33 to 3. R is the pile radius. ∆V is the volume injected into the probe. 1978) Baguelin. y is the pile horizontal displacement. The p-y curve at a depth z for the pile is obtained from the pressuremeter expansion curve at same depth z as follow: p = ηP * D y= 1 ∆V R 2 V0 (7-29) (7-30) where p is the soil resistance on the pile expressed as a force per unit length of pile. p/D (F/L2) Pl 1 Py k 1 kp=0. V0 is the initial volume of the probe. P* is the net pressure (P-Pi) in the pressuremeter curve.

The similarity of pressuremeter test and lateral loading of pile was the basis of the method. where Su is the undrained shear strength. p = αP * D y= 1 ∆V D 4 V0 (7-31) (7-32) where D is the diameter of piles.Robertson et al. α is 2 for clays and 1. (1983) refined a method which was previously proposed by Hughes et al. Figure 7-7 Steps for constructing p-y curve from pressuremeter test 241 . (1979) to derive p-y curves directly from pressuremeter tests. 7-7. Based on these assumptions on the relationship between limit resistance and undrained shear strength of clay. the limiting pressure (Pl – Pi) is approximately 5Su. The effect of pile driving on soils was simulated with a self-driving pressuremeter. This method is similar to Baguelin et al.5 for sands (Robertson et al. the following procedure was proposed by Robertson et al. the limiting lateral resistance from clay is approximately 9 Su. (1978) method except the value of coefficient α or η is different. 1986). (1983) and illustrated in Fig.. For piles under lateral loading. Whereas in the case of the pressuremeter.

This method considers that a p-y curve is made of a front resistance Q-y curve and a friction resistance F-y curve. F = (SF)(D)(X)(1 + X) ∆P * ∆X (7-33) in which. = 0. et al (1983) developed a method for deriving p-y curves from the results of pressuremeter test. F = the frictional soil resistance on the pile. ∆P* = the increase of net pressure. SQ = shape factor for pressure reaction. = (7-32) π for circular piles. 242 . = 1. P* = the net pressure which is equal to (P-Pi). X = ∆V’/V0’ where V0’ is equal to V0+Vi and ∆V’ is the volume injected in the probe from V0’ point on. This method is also applicable to test results of dilatometer. D = pile diameter or width. 4 =1.Briaud. SF = shape factor for shear reaction.79 for circular piles. The Q-y and F-y curves can be obtained point by point from the pressuremeter curve as follows: Q = (SQ)(P * )(D) where Q = the frontal soil resistance on the pile.0 for square piles.76 for square piles.

1983) 243 . Figure 7-8 Determination of the critical depth (Smith. it will not be considered for rock since for most cases a thick layer of soil exists above the rock mass. the p-y curve can be constructed from the summation of Q-y curve and F-y curve using the equation p = Q + F. while for the Q-y curve the critical depth Dc is determined from Fig.y= D ∆V ′ ′ 4 V0 (7-34) where y = the horizontal displacement of the pile. 7-8 and within that depth Q is multiplied by a reduction factor RF from Fig. After Q. the lack of vertical confinement influences the results of pressuremeters as well as the soil resistance to laterally loaded drilled shafts. This reduction was developed for soils. The effects due to lack of vertical confinement was introduced by Smith (1983) as follow: no influence is considered for the F-y curve. F. At shallow depth. and y values corresponding to each data point of pressure-volume curve are obtained. 7-9.

it is decided to evaluate Briaud et al. (1983) method for applications to drilled shafts socketed in rock. (2002). 244 . (1983) method is considered as the most suitable method for applications in rock. as it does not require empirical coefficients and avoids the difficulty of determining the yield and limit pressure.3 Evaluation Briaud et al.Figure 7-9 Reduction factor for depth within critical depth (Smith. (1983) method for deriving p-y curves from dilatometer test results is evaluated against field tests reported by Cho et al. Therefore. Additionally. 1983) The above reviewed methods for deriving p-y curves from pressuremeter test results were developed for applications in soils. The lateral load test by Cho et al. it considers both shear stress and normal stress as a result of soil or rock-shaft interaction. The shaft was embedded in soft weathered meta-argillite rock with 39 inch of overlying sandy silt and a 12 inch thick of dense sand. 7. (2001) and Gabr et al. Briaud et al. (2001) was performed on a free-head drilled shaft with a diameter of 30 inch and an embedment length of 168 inch.

5. The friction angles of the sandy silt and dense sand layers were interpreted as 32˚ and 45˚ according to the SPT correlation table by Liang (2002). P-y curves of the test site interpreted from DM test results using Briaud et al. 11. and 15 feet. The results of the dilatometer tests are provided in Fig. Ht 24 in. It can be seen that the predicted deflections are little bitter smaller than the measured values. The sandy silt and dense sand is modeled using LPILE internal sand p-y curves. 7-13. Mt = Ht × 12 inch SPT 14 100+ DM 8. respectively. as shown in Fig. The predicted deflections at the top of the shaft are compared with the measured values. 7-11.The pertinent shaft and soil information are shown in Fig. Dilatometer (DM) tests were performed at depths of 8. Nonlinear option of shaft stiffness is chosen for analysis. The average SPT N values of the sandy silt layer and dense sand layer were reported as 14 and greater than 100. These interpreted p-y curves are input into LPILE program to predict the lateral response of the test drilled shaft. 7-10. (1983) method are shown in Fig. 39 51 z (inch) 74 94 114 133 153 168 30 in.8.8 ft 15 ft Sandy silt Dense sand Soft weathered metaargillite rock Figure 7-10 Shaft and soil profiles of the case study 245 . 7-12.5 ft 11.

8 ft Depth 15 ft Figure 7-11 Dilatometer test results of the case study (after Cho et al.5 ft Depth 11.5 Depth 8.30000 25000 Pressure (kPa) 20000 15000 10000 5000 0 0 200 400 600 Volume (cc) 800 Depth 8..5 ft Depth 11. (2001) 246 .5 y (in) 1 1.8 ft Depth 15 ft Figure 7-12 P-y curves from dilatometer tests of Cho et al. 2001) 160000 140000 120000 p (lb/in) 100000 80000 60000 40000 20000 0 0 0.

The derived p-y curves from dilatometer test results are provided in Figs. reported by Gabr et al (2002) are also employed to evaluate Briaud et al.250 200 Load (kips) 150 100 50 0 0 0.. In order to match the measured deflections. a factor of 0. 7-18 to 7-21. Dilatometer tests were performed at the two test sites. The adjusted p-y curves are then input into LPILE to predict the deflections.5 is multiplied to the calculated p values from dilatometer test results. 7-18 to 21. (1983) method can make it works much better for most of the cases. 7-13 and Figs. (1983) method.e.5 Deflection (in) 2 Figure 7-13 Comparison of the measured and predicted deflections at shaft top The two lateral load tests. The performance of these derived p-y curves are evaluated by inputting them into LPILE and comparing with the measured loaddeflection curves. Therefore. as shown in Fig. It can be seen that such adjustment on Briaud et al. 7-14 to 7-17 for the four test shafts. load tests at I-40 and I-85.5 Measured Predicted-PM Predicted-PMAdjusted 1 1. i. It can be seen that the predicted deflections are smaller than the measured values. as shown in Figs. The test shaft and rock information have been described in Chapter VI. it is 247 .

reducing the calculated value of p by 50%) can provide reasonable site-specific p-y curves.concluded that the modified Briaud et al (1983) method (i.13 ft p (lb/in) 0.e.8 ft 1.8 1 Figure 7-14 Derived p-y curves from dilatometers for I-40 short shaft 100000 90000 80000 70000 60000 50000 40000 30000 20000 10000 0 0 0.4 0.4 ft 4.2 0. 100000 90000 80000 70000 60000 50000 40000 30000 20000 10000 0 0 0.6 y (in) 10.5 Figure 7-15 Derived p-y curves from dilatometers for I-40 long shaft 248 ..2 ft 6.7 ft 7.5 y (in) 1 p (lb/in) 13.

2 0.8 1 11.5 9.6 0.4 y (in) 0.70000 60000 50000 p (lb/in) 40000 30000 20000 10000 0 0 0.2 ft 3 ft Figure 7-16 Derived p-y curves from dilatometers for I-85 short shaft 80000 70000 60000 p (lb/in) 50000 40000 30000 20000 10000 0 0 0.6 ft Figure 7-17 Derived p-y curves from dilatometers for I-85 long shaft 249 .5 ft 8.9 ft 1.5 y (in) 1 1.2 ft 4.

8 Measured Predicted-PM Predicted-PM-Adjusted Figure 7-19 Comparison of load-deflection curves of I-40 long shaft 250 .2 0.4 Deflection (in) 0.2 0.400000 350000 300000 Load (lb) 250000 200000 150000 100000 50000 0 0 0.6 0.4 Deflection (in) 0.8 Measured Predicted-PM Predicted-PMAdjusted Figure 7-18 Comparison of load-deflection curves of I-40 short shaft 400000 350000 300000 Load (lb) 250000 200000 150000 100000 50000 0 0 0.6 0.

6 0.4 Deflection (in) 0.2 Measured Predicted-PM Predicted-PM-Adjusted 0.5 2 Figure 7-20 Comparison of load-deflection curves of I-85 short shaft 350000 300000 250000 Load (lb) 200000 150000 100000 50000 0 0 0.5 Measured Predicted-PM Predicted-PM-Adjusted 1 Deflection (in) 1.350000 300000 250000 Load (lb) 200000 150000 100000 50000 0 0 0.8 Figure 7-21 Comparison of load-deflection curves of I-85 long shaft 251 .

1 Summaries Towards the objective of developing a new p-y criterion of rock mass. A detailed literature review was performed to study existing design and analysis methods of laterally loaded drilled shafts and piles in rock. a series of field work and theoretical work have been carried out. A new hyperbolic p-y criterion of rock was proposed based on the field test data and extensive theoretical work. Additionally. A 3D finite element model simulating the response of laterally loaded drilled shafts in rock using ABAQUS was established to develop an empirical correlation equation for estimating the initial slope of a p-y curve of rock. In addition to the development of a p-y criterion of rock.CHAPTER VIII SUMMARIES AND CONCLUSIONS 8. theoretical equations for determining the ultimate resistance of rock mass were derived based on failure modes of rock mass. a method for predicting lateral capacity of drilled shafts in rock and/or soils was developed. The failure modes of rock mass were identified through a series of 3D FEM study. An elastic solution for predicting the lateral deflections of piles/drilled shafts embedded in a two252 . a rock strength criterion. and an existing empirical equation for estimating the side shear resistance between rock and drilled shafts. This estimate can be used to ensure adequate safety margin of a design of drilled shafts in rock.

was recommended for deriving p-y curves of rock from the pressuremeter and dilatometer tests. demonstrated its superiority against other methods by providing the smallest error on the predictions of the loaddeflection curves. The validation of the 3D FEM model using a field test suggests that it can be used to simulate the lateral response of drilled shafts in rock. (1983) method.layer soil/rock system was proposed. This simulation has provided 253 . the existing methods for deriving site specific p-y curves of soils from pressuremeter and dilatometer tests were reviewed and examined with the lateral load test results of rock-socketed drilled shafts. The suggested method for deriving experimental p-y curves from load test results. 2. The evaluation on the interim p-y criterion (Reese.2 Conclusions Based on the research work performed in this study. 3.e. Finally. the following conclusions can be drawn: 1. Furthermore. with a modification of reducing p value by 50%. Briaud et al. 8. using high order polynomial curve fitting technique to obtain y and using piecewise polynomial curve fitting technique to obtain p. This solution allows a quick estimation of drilled shaft deflections under lateral loads. i. an evaluation of various existing methods for deriving p-y curves from the results of an instrumented lateral load test was carried out and the most suitable p-y curve derivation method was identified. 1997) for weak rock using the two lateral load tests reveals that this p-y criterion tends to under-predict the deflections which may result in an unsafe design.

254 . This solution provides an efficient way to estimate the deflections of a drilled shaft in a layered soil/rock profile under small working lateral loads. sand. and rock. such as COM624P. Both short and long drilled shafts under free or fixed head condition can be considered. silts. The hyperbolic p-y criterion for rock developed in this study can be used in conjunction with computer analysis programs. It can be used for estimating the lateral capacity of drilled shafts in layered clay. The evaluation of the proposed method for estimating lateral capacity of drilled shafts in rock using field test results showed that this method can provide reasonable predictions of the lateral capacity of drilled shafts in rock. considering the measured capacity may be under estimated due to a small deflection at the maximum applied lateral load. LPLE. Additionally. 4. the proposed capacity prediction method is versatile. or FBPIER. The average prediction error was around 21% which is acceptable for practice. 6. the proposed empirical equation for estimating the deformation modulus of rock mass and the empirical equation for determining the modulus of subgrade reaction of rock can provide reasonable prediction results according to the validations against field test data. 5. from which analytical equations are derived for computing ultimate resistance of rock pu.basic understanding of the mobilization mechanisms of lateral resistance of rock mass to the drilled shafts. The comparisons with rigorous solutions and the validation against field lateral load test data have shown that the proposed elastic numerical solution can provide reasonable predictions of shaft deflection under small working loads. Additionally.

8. The modified method is suggested for design of laterally loaded drilled shafts in rock using dilatometer test results. 8. The computer program LCPILE can be used for limit state design to guarantee adequate safety against design loads. The research findings of this study provide a complete solution for design of drilled shafts socketed in rock under lateral loads. (1983) method can be used to validate the final design if dilatometer tests data are available. The Briaud et al. moment. The proposed elastic solution can be used for a preliminary design of shafts under service loads. Evaluations based on comparisons between the predicted and measured responses of full-scale lateral load tests on fully instrumented drilled shafts have shown the practical uses of the proposed p-y criterion and the associated methods for determining the required input of rock parameters. The proposed hyperbolic p-y curves can be used in conjunction of COM624P or LPILE for the final design. (1983) method with a modification of reducing the calculated p values by 50% for deriving the p-y curves from dilatometer tests is found to provide reasonable predictions on the lateral deflections of a drilled shaft in rock according to the evaluations against full-scale lateral field tests.to predict the deflection. Considerations of the effects of joints and discontinuities on the rock mass modulus and strength are included in the p-y criterion. 7.3 Future Studies More lateral load tests on drilled shafts socketed in various types of rock with the accompanying dilatometer tests at the load test sites should be performed to further 255 . and shear responses of a drilled shaft under the applied lateral loads. The modified Briaud et al.

and the validity of the modified Briaud et al. the capacity prediction method. A preliminary design chart for selecting drilled shaft diameter and rock socketed length can be developed based on LCPILE and the elastic solutions to ensure both capacity and service limits criteria are met. it is necessary to study the lateral response of drilled shafts socketed intermediate geomaterials. Therefore. the determination of group effect of lateral capacity of group drilled shafts in rock is of practical interest since drilled shafts are often constructed in a group. A future study on determining p-multipliers for a drilled shaft group in rock mass can be very helpful. Additionally. This objective can be accomplished through a well-planned field test or a centrifuge test on drilled shafts groups socketed in rock mass. as these materials conform a significant portion of the supporting medium for drilled shafts. a 3D finite element study may also be employed to determine p-multipliers for a drilled shaft group socketed in rock mass. A field test or centrifuge test using displacement controlled loading will be very helpful for developing p-y curves of rock under cyclic loading. Furthermore. A study on drilled shaft in rock under cyclic lateral loads will help seismic design of deep foundations socketed in rock mass. Although the proposed hyperbolic p-y curves is suitable for weak and hard rock by considering the effects of joints and discontinuities of rock mass. 256 . it is not ready to be applied to intermediate geomaterials. (1983) method for rock.validate the developed p-y criterion.

and Said. Z. A.T. 405-434. J. No. Vol. Inc. Vol. Practical handbook of rock mass classification systems and modes of ground failure. (1977).A. Y.. Holland.” Int. Baguelin. “Theoretical study of lateral reaction mechanism of piles.” Rock Mechanics.” API recommended practice 2A (RP2A). T. Bieniawski. D.. Balkema. Trans Tech Publications. Germany. p. No. 19th Ed. “Engineering classification of rock masses for the design of tunnel support. H..8. 237-248. ASTM D4719-00. Rock Mech. A. H. N. 6. Washington. 189236. (1976). F.” ASTM International.REFERENCES ABAQUS (1998). Inc. designing and constructing fixed offshore platforms. R. Florida.” Geotechnique. F. Baguelin. & Geomech. R. ABAQUS Standard User’s Manual. 97-106. Hibbitt. ed. Austria. Vienna. and Lunde. 257 . Z. Clausthal. p. Min.. Rotterdam. Afrouz. Bieniawski. D. Frank. Sci. “Rock mass classifications in rock engineering. “Determining rock mass deformability: experience from case histories. Bieniawski. (1992). (1978). (1991).” Proceedings of the Symposium on Exploration for Rock Engineering. Z. and Shields. Version 5. The Pressuremeter and Foundation Engineering. American Petroleum Institute (API). (1974). Karlsson & Sorensen. (2000). p. 4.. p.. F. (1978). 15... A. “Standard test method for prebored pressuremeter testing in soils. Abstr. F. C. “Recommended practice for planning. Lien. J. Barton.. CRC Press. Jezequel.. 3. T. 1. 27.

College of Engineering. Smith. No. Wiley. Pergamon Press. “The pressuremeter test for highway applications. 147-159. 4th Edition. T. “The nature and fundamentals of rock engineering. R. Bull. E.. (1989). FHWA-IP-89-008.” Geoteknisk Institut. 57. Rotterdam.. (1983). Testing and Monitoring: ISRM Suggested Methods. (1958).” FHWA Publication No. J. R. Hill. No. (1981). Briaud. F. Pergamon Press. 1004. N. Hudson. 321.Bieniawski. London. (1945). Rock Characterization. 1990.. J. published by A.” The Proceedings of the Physical Society. McGraw Hill. May. Briaud. Part 3. Practice and Projects. Brinch Hansen.Balkema. T. USA. “The theory of indentation and hardness tests. (1961). Brown. Briaud. L. and Meyer. Bishop. F. p. p. 8. Issue: 1288. Ed. 258 . The University of Texas at Austin.T.1-23. p.. “Using the Pressuremeter Curve to Design Laterally Loaded Piles. McLean..A. Engineering Rock Mass Classification. Department of Transportation. New York. J. New York. B.. p. (1992) The Pressuremeter. Oxford. Foundation analysis and design. E. VA 22101-2296. 158-167 Bowles.” Transportation Research Record.A. Federal Highway Administration. (1993). Netherlands. Bowman.L. Vol. E. J. p. Brown. T. “Sound Barrier Wall Foundations in Granular Material. and Mott. (1988). E. Copenhagen. Boghrat. Vol.” Unpublished thesis. (1989).” Comprehensive Rock Engineering – Principle..12. R. D.” Proceedings of Annual Offshore Technology Conference. J. “The ultimate resistance of rigid piles against transversal forces.. A. “Investigation of the lateral resistance to movement of a plate in cohesionless soil. 251. J. 495-502. Z.

Carter. Vol.” Rep. (1968). “Laterally loaded drilled shafts embedded in soft rock. and Borden. K. (1970). DiGioia. A. No. A. P. (1984). A. Keaney.” International Soc. H. W.. 359. T. P.. Fanaff. 65-80. and Rojas-Gonzalez. 33. Fisher.” Canadian Geotechnical Journal. 3-11. P. 68. p.. F. Gabr. Ohio Division of Geological Survery. B.. R.M. Dunnavant. 329-332. S. and Valsangkar. M. 118(6). Dou. Carter. Fossils of Ohio. and Yeung. A. Dissertation of University of Houston.. No 2-9.” Ohio Journal of Science.D. p.” Geotechnique. Ph. p. S. “Discussion on ‘Analysis of laterally loaded shafts in rock. P. P. 1. Rock Mechanics. K. Vol. (2001). C. New Zealand. J. Experimental and Analytical Investigation of the Behavior of Single Piles in Overconsolidated Clay Subjected to Cyclic Lateral Loads. v. Carter. p. 33 issue 5. M. “Dynamic Response of Single Piles and Soil-pile Interaction. ASCE.. Cho. “Determination of elastic properties of rocks using various onsite and laboratory methods. 80-96.. S. S. (1993). “Model studies of socketed caissons in soft rock. Feldmann. Univ. 747-759. p. H. 349-358. D. V. J.. Bulletin 70. L. M. “Landslide of southeastern Ohio. J.. (1996). 119(12). 839-855.. 259 .” Transportation Research Record 1772. (1986).” Journal of Geotechnical Engineering. (1996)... P. Clark. Civil Engineering Dept. H. p. Jr. L. and Picking. Booker. p. p. “Cavity expansion in cohesive frictional soils. “A non-linear soil model for predicting lateral pile response. and Byrne.. H. R. F. and Kulhawy. 36(3).. (1996). Texas.Bukovansky.. of Auckland. W.’” Journal of Geotechnical Engineering.” Canadian Geotechnical Journal. M. R. Proc. Dykeman. “Analysis of laterally loaded shafts in rock. (1986). M. D. J. 2014-2015. (1992).. ASCE.. and Hackathorn.

London.J. (2001). and Shi.. 119(12).J. FHWA-KS-87-2. W. (1993). “Dilatometer tests for the design of grouted piles in rock. Proceedings of In Situ '86. S. M. Weltman. J. G. 560-568.Fleming.. and Johnston. J. ‘Predicting the deformation moduli of rock masses’ International Journal of Rock Mechanics and Mining Sciences. Englewood Cliffs. “In situ measurement of soil properties with the pressuremeter. Blacksburg. I. M.. p. Athens.. Austria. VA. Cho.” Rep. V. K. 260 .. Block theory and its application to rock engineering. North Carolina Department of Transportation. A. F. F. 658.. (1992). C. M. 525-531. Goodman. Guo.. and London. p. Piling Engineering. A. H. Glasgow. R. W. (1993). Gabr.” Proceedings of the international conference on geotechnical engineering of hard soils-soft rocks. P. D.. (1987). 40. and Kayabasi. ASCE.” Use of In Situ Tests in Geotechnical Engineering. N.H. W.. (1986). Sonmez. M. Kansas Department of Transportation. K.” Civil Engineering and Public Works Review. and Anderson. Gabr..’” Journal of Geotechnical Engineering. Frantzen. Prentice-Hall.. Borden. Georgiadis..B. 2015-2018. “Final report: p-y curve data for laterally loaded piles in shale and sandstone. p.” FHWA/NC/2002-08. (1985). Randolph. and Stratten. and Elson. K.. Haberfield. (2002). W.701-710. G. W. and Michalopoulos. A. R. Kansas. Vienna. USA. (1961). C. Topeka. “Factors influencing the interpretation of pressuremeter tests in soft rock. 56 No. R.A.C.H. F. “P-y curves for laterally loaded drilled shafts embedded in weathered rock. p.. A. “Discussion on ‘Analysis of laterally loaded shafts in rock.” Proceedings of the Eighth International Conference on Civil and Structural Engineering Computing. and Nixon. Gokceoglu. “Subgrade modulus for laterally loaded piles.. M. Blackie. W. Gibson. No. E. Clark. (2003).

Heuze. Wang. “Strenth of jointed rock massses” Geotechnique. No. J. (1976) “Plate bearing and borehole jack tests in rock.a finite element analysis..Hall. E. Hoek. “Estimating Mohr-Coulomb friction and cohesion values from the Hoek-Brown failure criterion. and Amadei. p. Hoek. (1984) “Suggested method for estimating the in-situ modulus of deformation of rock using the NX-borehole jack. and M.E. 261 .. 4B8-1 – 4B8-6. J. Y. p. p. “Strength of rock and rock masses. 27. Utah.205210. 1013-1035.S. 2 (2). p. 3. p.S. F. and Brown. (1994).C. 22. New Technological and Design Developments in Deep Foundations.. T.7. Hoek.. F. (1980). On Rock Mechanics.” Int.2. Rock Mech.. (1990). 187223.” Final Report to U.” Geotechnical Testing Journal. E. University of Colorado. M. E. 4. No. (1985) “The NX-borehole jack: A lesson in trial and error” International Journal of Rock Mechanics and Mining Science.. (1980) “A study of the orthotropy of coal and other rock materials. p. Sci. 240-253. p. D. No. v.105-112. C. (2000). 4-16. 17th U. B.E. Vol. Hetenyi. Journal of the Geotechnical Engineering Division. F. Behavior of laterally loaded caissons in weak rock. 106 issue 9.” Proc. and Ho. E. of Interior and Bureau of Mines. H.. “Empirical strength criterion fro rock masses. E.” ISRM News Journal. Ann Arbor: University of Michigan Press. p. Snowbird. (1946).” American Society of Civil Engineers. Issue 100. Min. Dept. 227-229. ASCE Geotechnical Special Publication. & Geomech.E. 33 issue 3. (1983). Heuze. Symp. Hawk. Hoek. v. Vol. Beams on elastic foundations. and Salem A. G. Heuze. Abstr. Vol.

. Vol.Geotech...oh. Symp. E.” Journal of Geotechnical Engineering. (1997). (2000) www. 262 . 2004. Wroth. W.J. K. ASCE. Vol. D. Johnston.us/geosurvey/pdf/ stratcol. Hughes. (2002) “Hoek-Brown failure criterion – 2002 edition.T. Hoek. “Estimating the deformation modulus of rock masses: a comparative study. D. “Transmission line structure foundations for upliftcompression loading. E. (1985). G.” International Journal of Rock Mechanics & Mining Science. (1977)..” Proc. (1989a). E. p. and Ercanoglu. 4. Snowbird. and Lam. 115. Vol. J. 115. Kulhawy. 728-740. On Rock Mechanics. Sci. Johnston.” J.E. (2003).N..state. 27. (1983). 40. J. 4B10-1 – 4B10-8. p. Gokceoglu. and Windle. Utah. p. 711-727. P.” Journal of Geotechnical Engineering. vol. p. 5. K. J. No.T. No. Hustrulid. Kayabasi.S. No. (1988). (1989b). Hull. 5.pdf. 31-38. 34(8). 111. S.. T. et al. Min. G. EL-2870. accessed on Aug. p. (1976) “An analysis of the Goodman jack. W. Palo Alto.dnr. and Lam. “Strength of intact geomechanical materials. H. O. W. E. 1165-1186... “Shear Behavior of Regular Triangular Concrete/Rock JointsAnalysis. C. “Pressuremeter tests in sand. Johnston. p. A.Hoek. North American Rock Mechanics Society Meeting in Toronto. A. and Brown . M. and Brown. 55-63. p. Carranza-Torres. “The Hoek-Brown criterion – a 1988 update.” Geotechnique. 17th U.” Int. and Corkum. E. Rock Mech. I. Engrg. No. W..” Rep. Electric Power Research Institute. Calif. (1990) and Larsen. M. C. F.. University of Toronto. B. S. “Shear Behavior of Regular Triangular Concrete/Rock JointsEvaluation.” Proc. 730-749..” 15th Canadian Rock Mechanics Symposium: rock engineering for underground excavations. T. Vol. Hoek.. “Practical estimates of rock mass strength.

H. “Drilled Shaft Side Resistance in Clay Soil to Rock. 7th Europe Conference on Soil Mechanics and Foundation. ASEC. and Schaaf.” Final Rep. F. et al. –Y. FHWA/OH-2002/038. issue 97. H.. Houston. Van Nostrand Reinhold. K. Shales and surface clays of Ohio. Liang. R. New York. E. Matlock. Tex. and Y. fourth series. Chap. Offshore Technology Conference. -J. Ohio Department of Transportation. p. Vol. Austin. P. p. and Hoek. 577-594. Fang ed. (1995). (2000). of the International Conference on Geotechnical and Geological Engineering (GeoEng 2000).. and Phoon. Austin. “GSI: A geologically friendly tool for rock mass strength estimation.” Proceedings. H. H. 172-183. Brian D. Singapore. p. “Drilled shaft foundations for noise barrier walls and slope stabilization. p.” Proc. “Procedures and Instrumentation for Tests on a Laterally Loaded Pile. Broms Symposium on Geotechnical Engineering. “Drilled shaft foundations. 213-228. E. F. (1938). Geotechnical Special Publication 38.(2000) “Determination of rock mass modulus for foundation design. 59-79. 263 . p. H. On Design and Performance of Deep Foundation: Piles and Piers in Soil and Soft Rock. A.. F. Meigh. Lamborn.E. (1979).” Bengt B.. 2nd Offshore Technology Conference. (1991). and Ripperger. R. K. (1956). Littlechild. Kulhaway. Y. (2002). C. p. Melbourne. as applied to drilled shafts..” Proc. Kulhawy. “Correlations for design of laterally loaded in soft clay.. R.. (1970). Australia. “Design Parameters for Weak Rocks. Chen. D.” Foundation Engineering Handbook.” Proc. Matlock. Geological Survey of Ohio. C.” Geotechnical Special Publication. on Conf. 1422-1440. 225-240. 2nd Ed. H. bulletin 39. “A thirty year perspective of Broms’ lateral loading models.Kulhawy. Marinos.1. 14.” Proceedings. (1993).. A. Eighth Texas Conference on Soil Mechanics and Foundation Engineering.

p. No. “Development of p-y curves for partly saturated silts and clays.” Rep. C. and O’Neill. Prepared for Ohio Department of Transportation.. M. Colorado Department of Transportation. “Instrumentation. Mezazigh. R. (1998). 3. and Reese. O’Neill. Meyer.” Analysis and Design of Pile Foundations.. p. J.. “Evaluation of P-y Relationships in Cohesionless Soils. (2000). “Laterally Loaded Piles in Sand: Slope Effect on p-y Reaction Curves. monitoring and testing at the ERI-60-3100 bridge project. 433-441. Duncan.” Canadian Geotechnical Journal. and Yang. Vol. J. (1969).. L. Y. Pells.. K.. 1984. (2002). D. Murchison. Ed. G. p.” Proceedings. W. and Signals. 35. S. R. and Gambin. R. Columbus. et al. Mokwa. “A nonlinear deformation modulus based on rock mass classification. Sydney. isotropic. P. Nusairat. OH. CDOTDTD-R-2004-8.T. J. elastic materials. M. R.” Proceedings of Geo-Denver 2000: New technological and design developments in deep foundations. 22/23. R. Final Report. Drilled Shafts: Construction Procedure and Design Methods.174-191. T.. J. and McVey.” Sols-Soils. 264 . M.. 181-202. M.” USBM RI 7855. (1974) “NX borehole jack modulus determinations in homogeneous. and Helmers. L. Signs.. W. ASCE. Nusairat. “An Experimental Investigation into the Side Shear of Socketed in Sandstone..A. ASCE. (2004).. Bieniawski..Menard. FHWA-IF-99-025. Meyer. G. Engel. Z. Vol. p.” International Journal of Min Geol Eng. Liang. M. R. Abu-Hejleh. J. “Methode Generale de Calcul d’un Rideau ou Pieu Sollicite Horizontalement en Fonction des Resultats Pressiometriques. and Liang. N. 8. 291-302. Drilled Shaft Design for Sound Barrier Walls. J. (1999). p. R. Bourdon. Hanneman. 224-237. L. International Conference on Structural Foundations in Rock. J. O. No. L. D.(1990). Engel. J. (1980)... Nicholson.. M. and Levacher. N.

(1990) “Shale durability rating system based on loss of shear strength.” J. p. and Rao. Reese. 59-64.. Austin. 1-23. GT9. 2. “Laterally loaded piles in a two-layer soil system.. G. R. Richardson. Randolph. F. Engrg. 1864-1880. J. (1971).” Proc.” 7th Offshore Technology Conference. ASCE. Texas A&M University. (1986). 1177-1181. T. R. Vol. (1975).” The pressuremeter and its Marine Applications: Second International Symp. C. p. (1981).. “Behavior of laterally loaded piles... O. and Matlock.. F. C. P. Houston. L. P. and Wiles. T. D. p. 633-649. C. (1985). and Koop. and Koop. (1956). D. (1975). M.” Journal of the Geotechnical Engineering Division. 671-690. (1997).” Journal of Geotechnical Engineering. 123(11). 101(7). G. Reese. 473-483. and Hughes. T... p. 1010-1017. “A strength criterion for rocks. 711-731. L. Poulos. J. “Field testing and analysis of laterally loaded piles in stiff clay. No. 116. “Determination of properties of sand from selfboring pressuremeter tests. K. 247-259. M. Vol. and Welch. K. ASCE. “Lateral loadings of deep foundations in stiff clay. Of 6th offshore Tech. R. p. Found... L. Cox.” Proc. p. Div. p.” J. (1974). 31(2). Geotech. (1982). Robertson. Eight Texas Conference on Soil Mechanics and Foundation Engineering.. L. W.” Journal of Geotechnical Engineering. Reese. F. “Non-dimensional solutions for laterally loaded piles with soil modulus assumed proportional to depth. Rao. Soil Mech. N. Conf. C.. “Analysis of laterally loaded piles in weak rock. I: Single piles. ASCE.. Cox.Pise. Reese. Indian Geotech..S. No. p. “The response of flexible piles to lateral loading. C. H. p. “Analysis of laterally loaded piles in sand. C. H. 108. Texas. ASTM STP 950.” Geotechnique.” Proceedings. p. 12. Roorkee. I. TX. Conf. 97(5). V. Vol. D. 265 . W. ASCE. Ramamurthy. Envir.. L. Reese.

No. B. (1976).881.rocscience. “Characterization of the deformability of rock masses by dilatometer tests.” Geotechnique. P. and Hazen. S. A. 130. p. J. 33-44. Texas A&M University.” International Soc Rock Mechanics. (2002). M. Serafim J. P. R. 509-516. 138-153. P. Sargand. 38. 13. N. G.. I. W. (1983). & Geomech. “Friction and End Bearing Tests on Bedrock for High Capacity Socket Design. 24. RocLab. 878. 365-370. Vol. No. Canada.P. Hughes. L.. Y. and Collingwood. I. Campanella. Vol. No. W.. Sci. Abstr. Rocscience. Proc. Shen.. Campanella. P. A. 1. “Analysis of laterally loaded pile groups using a variational approach. and Pereira J.. (2001).. p. Brown. P. A. (1970). and Teh. and Teh. Vol. p. “Consideration of the geomechanical classification of Bieniawski. S.Robertson. 6. Shen. (2004). M. C. 52(3). a free software from www. “Deformation behavior of shales” Int. Inc.. F. “Design of laterally loaded piles using the pressuremeter”. J. Rocscience (2002).. J. 8. M. p.L. M.. and Ferreira.. G. and Sy. p. Vol. Robertson. P. A. R. Vol. C.. p. K.. A. J. Rodrigues. Rocha. Hughes. (1983).” Journal of Geotechnical and Geoenvironmental Engineering.. Silveira. O. ASTM STP 950.. Toronto.. 324-333. O. and Journeaux. p. K. Silverio.com. Geotech. 266 . (1987).” Can. G. 201-208. Seidel. (1986). “A new Socket Roughness Factor for Prediction of Rock Socket Shaft Resistance. Y. Rosenberg.” Canadian Geotechnical Journal...” Proceeding of International Symposium on Engineering Geology and Underground Construction. “Design of laterally loaded displacement piles using a driven pressuremeter. Rock Mech. Lisbon 1(II). J. and McKeown. Min. The pressuremeter and its Marine Applications: Second International Symp.” ASTM SPT Symposium: Design & Performance of Laterally Loaded Piles and Pile Groups. 2-32. “Analysis of laterally loaded piles in soil with stiffness increasing with depth.

(1988).. (1981) “Borehole diameter as a factor in borehole jack results. C.” Report from California Dept. of the 12th International Conference on Soil Mechanics and Foundation Engineering. (2002). of Transportation. et al.. C. Carnegie Inst. J. 2. To. H. Stout. “Evaluation of coefficient of subgrade reaction. Sun.. “Design Parameters for Bored piles in a Weathered Sedimentary Formation.. and Einstein. K. J.. 97-111. Vol. p. p. (1961). of Technology. Paris. Gose. 1073-1078.” Procs.S. Smith. 711-726. T. 5th Int. (1955). 120(8). Terzaghi. (1989). (2003). 39. K. Wallace. Dolomites and limestones of western Ohio. No. 129.S. “Laterally loaded piles in elastic media. A. Thesis Texas A&M University. 267 . 22nd U. 1324-1344. Vol. “Pressuremeter Design Method for Signal Piles Subjected to static Lateral Load. “Cyclic large deflection testing of shaft bridges part II: analytical studies. 5(4).” Ph. D. T. Cambridge. bulletin 42. (1999).” Proc. p. et. 845-850.” Proc. Dept. p. p. “A three-dimensional nonlinear study of piles leading to the development of a simplified model. Massachusetts. S. 1. H. Toh. Ernst. P. ASCE. F. No. Conf. K. Vesic. and Ugai. Bielak.” Journal of Geotechnical Engineering. pp. Wilber (1941). al. H. (1983). “Lateral load capacity of drilled shafts in jointed rock. Wakai. and Christiano.” Soils and Foundations. A.” Journal of Geotechnical and Geoenvironmental Engineering.Shuri. Trochianis. A. W. fourth series. “3-D elasto-plastic finite element analyses of pile foundations subjected to lateral loading. of Civil Engr. On Rock Mechanics.” Geotechnique. Vol. A. 8. Cambridge. 297-326. Soil Mechanics and Foundation Engineering. S. MIT Press. 392-397. D.. “Beam on elastic subgrade and the Winkler hypothesis. 1. Symp.. (1994). Geological Survey of Ohio. p.” Report R-88-176. Vol.

(1981).” Proceedings of International Conference on Structural Foundations on Rock. p.Wang... Williams. Version 2. H. 18.. Federal Highway Administration Publication No. and Donald. (1993). A. Liang. D... Alaska. Dissertation. E. No. Analysis and design of axially loaded drilled shafts socketed into rock. 1998. FHWA-SA-91-048. p. 268 . Zhang.” Canadian Geotechnical Journal.” Proceedings of DFI 30th Annual Conference on Deep Foundations. C. (1980). 327-347.” Journal of Geotechnical and Geoenvironmental Engineering. “Nonlinear analysis of laterally loaded rock-socketed shafts”. (1997).. and L. MA.0. MS thesis. F. H. Sydney. R. Wilson. (2005). R. Yang. R. v.. (2005). Chicago. 502-523.. v. 1. and Grismala. XII. Silva. B.131 No. University of California at Davis.. L. 1. “The Design of Socketed piles in weak rock.” The 40th U. L. “Side Resistance Rock Sockets in Sandstone. W. S. “Ultimate lateral resistance to piles in cohesionless soils. Williams. and Liu. L. Yang. F. “A suggested system for classifying shale materials for embankments. S. D. A. Journal of Geotechnical and Geoenvironmental Engineering. Vol. and Pells P. T. F. p. Soil-pile-superstructure interaction in liquefying sand and soft clay. I. COM624P – Laterally Loaded Pile Analysis Program for the Microcomputer. 955-968. 78-83. Reese. J. (1975). Ph. Zhang. L.. Johnston. Cambridge. p. K. Ernst. Massachusetts Institute of Technology. H. Vol. IL. and Shale. I. and Liang. p. K. 126 issue 11. “Lateral Response of Large Diameter Drilled Shafts in Clay. “Analysis and test of rock-socketed drilled shafts under lateral loads. Wood. Symposium on Rock Mechanics. P. Anchorage.. Zhang. (2000).S. and Einstein. (2005).” Bulletin of the Association of Engineering Geologists. N.. and Deo. 39-55. Mudstone.