You are on page 1of 274

THE EFFECTS OF SLAKING ON THE

ENGINEERING BEHAVIOR
OF CLAY SHALES

by

Michael Edward Botts

B.S., Auburn University, 1976


M.A., Washington University
in St. Louis, 1979

A thesis submitted to the


Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Department of Civil, Environmental, and
Architectural Engineering
1986
Botts, Michael Edward (Ph.D., Civil, Environmental, and Architectural
Engineering)
The Effects of Slaking on the Engineering Behavior of Clay Shales
Thesis directed by Associate Professor Stein Sture

Numerous foundation and slope stability problems worldwide have been contributed to
the presence of clay shale deposits. Present experimental and theoretical methods used
in geotechnical engineering practice are inadequate for assessing the stability of clay
shales. The major difficulties with clay shales are attributed to two properties: They are
intermediate in behavior between rock and soil, and they tend to transgress from rock-
like to soil-like materials within relatively short time-frames.

Evidence suggests that softening along fissures is important to the rapid loss of strength
in clay shales. Yet, geotechnical literature is surprisingly devoid of studies concerning
theoretical or experimental aspects of fissure deterioration in clay shales. Additionally,
clay shales invariably exhibit a strong tendency to slake, or disperse, during rewetting, a
property which certainly contributes to the softening of clays shales. However, few
reported studies have investigated the slaking resistance of geological materials under
conditions of confinement found in the field, while no reports were found which
investigate the effects of slaking on strength or strain behavior.

The research presented in this dissertation provides initial theoretical and experimental
assessments of the effects of fissure deterioration on the engineering behavior of clay
shales. In particular, the drastic decreases in strength resulting from slaking are examined
in detail.
The author introduces the possibility incorporating the effects of slaking into the critical
state soil mechanics model. The role of slaking is seen in this context as simply another
path for altering the water content of the clay shale, while the strength envelop remains
constant.

Laboratory experiments involving triaxial compression tests were performed on Pierre


shale samples which had undergone various degrees of slaking under confinement of 10
psi. The results from these tests show very significant reductions in strength (up to
80%), resulting from a single slaking cycle. The shear strength data from both unaltered
and softened samples were successfully normalized using critical state concepts, and
displayed a well-defined two-segment failure envelope. This suggests that drastic lateral
and temporal variations in the strength of clay shale deposits, which result from slaking,
might be accounted for using the critical state approach.
ACKNOWLEDGEMENTS

Seldom is anything accomplished without the assistance or encouragement of others.


This dissertation is no exception. Like a rebellious teenager, I felt a need to pursue this
research in my own way. Yet, without the advice and encouragement of Dr. Stein Sture,
there were times when this research might have gone no further. I thank him for his
seemingly infinite patience, and for always being there when I needed him the most.
Likewise, I owe much to the fatherly advice of Dr. Hon-Yim Ko; his deep understanding
of both research, and people, helped me through some difficult times.

The ideas for this research were inspired during brief studies of the clay shale problems
in Italy. I thank Dr. Robert Schiffman at the University of Colorado, and Professors
Arturo Pellegrino, Arrigo Croce, and Carlo Viggiani at the Universitá di Napoli, for
providing this opportunity. Special appreciation is expressed for my dear friend and
colleague, Professor Luciano Picarelli, whose enthusiasm for this research has been a
source of much of my own enthusiasm and inspiration. In addition, the encouragement
and advice of Dr. Bob Fleming of the U.S.G.S., during the initial stages of this research,
were invaluable, as has been the advice of my committee members, Drs. Bernard
Amadei, Dobroslav Znidarcic, and William Braddock, during the later stages. I would
also like to thank Dr. Nicolas Costes, of NASA Marshall Space Flight Center, for having
the patience and enthusiasm required to introduce a geologist to the strange world of soil
mechanics.

Of course, no one could survive the perils of graduate school without the friendship of
the students who share in the misery and joys. Thanks to Steve Ketcham, who eased my
transition to Boulder and forced me to have fun in spite of myself, and particularly to
Ates and Canan Ontuna, Roberto Azevedo, Izabel Duarte, Vincenzo Pane, Sebastiano
Perriello-Zampelli, Paolo Croce, and Dan Egging, all of whom gave me something I will
keep the rest of my life. No less appreciation is reserved for Bob Scavuzzo, Emir
Macari, Trond Mageli, and all the others who gave me a smile from day to day.

Much appreciation goes to Mom and Dad, and the rest of my family; their support and
belief in me have always inspired me in my endeavors. Last, but certainly not least, I
thank my wife, Mary Lynn, whose love gave me the strength and encouragement to
follow the path that I felt was right, and whose patience and unselfishness allowed me
the freedom to complete the task. This dissertation is lovingly dedicated to her.
CONTENTS

CHAPTER

I. INTRODUCTION 1

II ENGINEERING PROBLEMS ASSOCIATED WITH CLAY SHALE


DEPOSITS 5
Characteristic Problems Related to Clay Shale Deposits
Around the World 6
British clay shales 6
Clay shales of North America 9
Clay shales along the Panama Canal 15
Italian clay shales 18
Residual Versus Softened Strength 24
Comparative Summary and Discussion 26

III. SOFTENING MECHANISMS IN CLAY SHALE DEPOSITS 35


Softening Mechanisms 36
The importance of fissures 36
Equilibration of negative pore water pressures 36
Fissure deterioration model 38
Progressive failure mechanism 41
Interactions between softening mechanisms 44
Slaking in Clay Shales 47
The slaking process . 47
Testing for susceptibility to slaking 49
Summary of slaking in clay shales 50
IV. THEORETICAL ASPECTS OF FISSURE SOFTENING IN CLAY
SHALES 53
Model for Fissure Softening 53
Implications of the Fissure Softening Model 56
The shear strength of jointed masses 56
The shear strength of a single, clean, filled, or softened joint 56
The strength of a clay mass consisting of a soft matrix surrounding
stiff, intact cores 61
The effects of progressive fissure deterioration on the engineering
behavior of clay shales 61

V. THE ROLE OF SLAKING WITHIN THE CRITICAL STATE MODEL 68


Review of the Principles of Critical State Soil Mechanics . 68
The Role of "Aging" Within the Critical State Concept 77
Slaking within the Critical State Concept 80
Possible Complications Resulting from the Fissured Nature of Clay
Shales 85
Testing the Critical State Model with Regard to Slaking 86
Summary on the Role of Slaking within the Critical State Model 88

VI. EXPERIMENTAL METHODOLOGY 90


Testing Apparatus 91
Hoek cell 91
Conventional triaxial cell 92
Sample Preparation 97
Block sample retrieval 97
Sample coring and trimming 97
Sample drying and the inducement of fissures 100
Test Procedure 101
General considerations 101
Initial conventional triaxial tests 102
Improved conventional triaxial tests 105
Water content measurements 108

VII. TEST RESULTS AND ANALYSIS . 110


Volumetric Changes During Drying and Wetting 110
Drying 110
Rewetting 113
Summary of drying and wetting data 117
Stress-Strain Response During Shear 119
Axial stress-strain response 120
Volumetric strain during shear 130
The Shear Strength of Progressively Softening Pierre Shale 132
The Effect of Sample Orientation on the Strength of Pierre Shale 136
Summary of Test Results 139

VIII. EVALUATION OF TESTS RESULTS WITHIN THE CONCEPT


OF CRITICAL STATE MECHANICS 143
Assessment of the Effects of Slaking Within the Critical State
Concept 143
Obtaining the normalization parameters 145
Normalization of data 148
Normalized strength for "high-angle" tests 153
Discussion on Slaking and the Shape of Failure Surfaces for
Anisotropic Material 155
The effects of fissuration and anisotropy on the shape of failure
surfaces 155
Slaking and the intensity of natural remolding 160
Slaking and the Critical State Concept:
Recommendations for Further Studies and Practical Applications 163
Practical applications 163
Recommendations for future studies 167

VII. CONCLUSIONS 173

VIII. BIBLIOGRAPHY 178

APPENDIX
A. CLASSIFICATION SCHEMES 189
B. PROCESSES ACTING DURING THE FORMATION OF CLAY
SHALES 204
C. THE MECHANICS OF SINGLE FISSURES 217
D. THE STRENGTH OF A CLAY MASS CONSISTING OF
INTACT CORES SURROUNDED BY A MUD MATRIX 228
E. PRESENTATION OF TEST RESULTS 232
FIGURES

2.1. Tentative relationship between average shear strength along slip


surfaces and time, for cuttings & retaining walls in London clay. 07
2.2. Relationship between cohesion intercept and time between con-
struction and failure. 07
2.3. Failure envelope for London clay based on back-analysis of first-
time slides. 09
2.4. Location map for major dams built in clay shales in the North
Central United States and Canada. 12
2.5. Proposed clay shale design strength parameters for Bearpaw
shale at Gardner Dam. 14
2.6. Summary of the shear strength data for the Cucaracha clay shale. 17
2.7. Map of the landslide areas and their relationship to underlying
rock type in Italy. 19
2.8 Relationship between unconfined compression strength and water
content for different degrees of weathering in the Laga formation, Italy. 22
2.9 Variation of shear strength in an Italian clay shale as a function
of time of storage in a humidity room. 23
2.10. Shear strength envelopes for undisturbed core samples of firm and
weathered Lugagno clay shale. 25
3.1. Definition of Bishop's average pore pressure ratio. 37
3.2. Contours of shear strength in a slope and its relation to corresponding
stress states in an overconsolidated clay. 42
3.3. Schematic illustrating complex action reaction paths possible in clay
shale materials. 45
3.4. Modes of swelling associated with simple rebound and slaking in an
overconsolidated clay or clay shale. 51
4.1. Schematic of the proposed model for progressive deterioration of
clay shales. 54
4.2. Photo showing joint deterioration in the Pierre shale of
South Dakota. 55
4.3. Schematic illustrating various forms for joints. 58
4.4 Schematic illustrating the progression of softening in undulating and
Planar joints. 59
4.5 Schematic of a highly altered clay shale consisting of stiff lumps
of intact clay within a matrix of highly weathering clay. 62
4.6 Schematic illustrating possible changes in the failure surface
resulting from progressive softening of a clay shale. 62
4.7 Schematic illustrating changes in the stress-strain behavior of a
clay shale undergoing progressive softening as illustrated in
Fig. 4.6. 65
4.8 Schematic illustrating changes in the strength envelope resulting
from progressive softening of a clay shale as illustrated in Fig. 4.6. 66
5.1. Ultimate failure points for drained and undrained tests on normally
Consolidated specimens of Weald clay. 70
5.2. Stress paths in (a) Q':P' and (b) v:P' space for undrained tests on
normally consolidated samples. 71
5.3. Stress paths in (a) Q':P' and (b) v:P' space for drained triaxial tests
on normally consolidated samples. 72
5.4. The critical state line in v:ln P' space. 73
5.5. Stress path followed in a drained triaxial compression test on an
overconsolidated clay 73
5.6. Normalized failure surface for drained and undrained tests on
overconsolidated samples of Weald clay. 74
5.7. Schematic of expected undrained test paths for samples at different
overconsolidation ratios. 75
5.8. The complete boundary surface in three dimensions ; Q':P':v space 76
5.9. The drained path in Q':P':v space. 77
5.10 Failure states of drained tests on samples at different overconsolidation
Ratios. 78
5.11 Predicted failure points for overconsolidated and normally consolidated
Clays. 79
5.12 Geological history and compressibility of normally consolidated
clays, showing effects of "aging". 80
5.13 Changes in the undrained shear strength ratio and the consolidation
pressure ratio versus the plasticity index for aged" and "young"
normally consolidated clays. 81
5.14 Normalization of the undrained shear strength for "young" and "aged"
Clays. 81
5.15 Possible compression and swelling paths for a clay undergoing
aging and slaking. 82
5.16 Possible loading/unloading paths (a) before and (b) during drained tests. 84
5.17. The Hvorslev surface in normalized P':Q' space. 87
6.1. Schematic of the triaxial test cell and measuring apparatus. 93
6.2. Schematic of the test sample assembly. 95
6.3. Photograph of dried cores of Pierre shale showing different
orientations of fissure pattern. 99
6.4. Irregular volumetric data from tests on two poorly saturated samples 104
7.1. Drying curve showing the rate of decrease in water content with
increased time of open-air drying of Pierre shale samples. 111
7.2. Drying and wetting curves as a function of drying time for the
Pierre shale. 116
7.3. Representative stress-strain plots of samples having undergone
no drying and half-an-hour prior to wetting and shear testing. 121
7.4. Representative stress-strain plots of samples having undergone
similar wetting drying cycles, but tested at different confining pressures. 122
7.5. Comparative stress-strain plots showing drastic softening of Pierre
shale after slaking. 124
7.6. Representative stress-strain plot showing three distinct segments in
the prefailure curve. 125
7.7. Two- and three-segment loading curves for other stiff, or cemented clays. 127
7.8. Schematic explaining two-segment loading curves, resulting from the
combining of frictional and bonding resistance forces. 128
7.9. Plot of Youngs modulus and specific volume for loading curve
segment I and II. 129
7.10. Prefailure P'-v paths for all Pierre shale samples tested at 30 psi
Confining pressure. 129
7.11. Peak and ultimate values of P',Q', and v for unaltered samples of
Pierre shale. 133
7.12. Peak and ultimate values for P',Q', and v for Pierre shale samples
dried for 4 days and wet for 1 day. 134
7.13. Final and ultimate values for P',Q', and v for all Pierre sample,
unaltered and softened. 135
7.14. Comparison of loading curves for unaltered Pierre shale samples
cored perpendicular and at a highly oblique angle to the major
plane of fissuration.
137
7.15. Comparison of loading curves for softened Pierre shale samples
cored perpendicular and at a highly oblique angle to the major
plane of fissuration. 138
8.1. Peak and ultimate values for P',Q', and v for all samples of
Pierre shale, unaltered and softened (reproduced from Fig. 7.13. 144
8.2. Relationship between the specific volume and ln P' at peak
strength for Pierre shale samples at confining pressures of 10, 30,
and 50 psi. 145
8.3. Values of v and ln P' at the ultimate (final) strength of all Pierre
Shale samples tested at 10, 30, and 50 psi. 147
8.4. Values of v and ln P' at the ultimate (final) strength of
unaltered Pierre shale samples tested at 10, 30, and 50 psi. 147
8.5. Normalized peak values of P' and Q' for all samples of Pierre
shale, unaltered and softened, showing well defined
failure surface. 149
8.6. Normalized peak values of P' and Q' on log-log scale for all
samples of Pierre shale, unaltered and softened, showing
extension of failure surface. 151
8.7. Normalized pre-failure loading paths for all samples of Pierre
shale, unaltered and softened. 154
8.8. Schematic comparing (a) the Patton model for a sawtooth
joint, to (b) normalized failure envelopes for Pierre shale. 157
8.9. Normalized strength envelope showing low stress
stress strengths, overconsolidated strengths, and critical
state strengths for heavily-overconsolidated clays in
Canada. 158
8.10. Normal consolidation and critical state lines in P':v space for
natural and remolded clays of Fig. 8.9. 159
8.11. Failure envelope for natural and remolded samples, with data
normalized according to respective normal consolidation lines. 160
8.12. Complex P':v history followed by Pierre shale sample 30/28-3
prior to testing. 162
A.1. Classification scheme of Underwood (1967). 192
A.2. Classification scheme of Gamble (1971), based on the relationship
between slaking durability and plastic index. 196
A.3. Classification scheme of Deo (1972), based entirely on resistance
to slaking. 197
A.4. Two part classification scheme of Morgen-Stern and
Eigenbrod (1974). 199
A.5. Modified classification scheme of the present author showing clay
shale as a unique entity of argillaceous materials. 200
B.1. Relative abundance of major groups of clay minerals in Phanerozoic
Mudrocks. 206
B.2. Probable stress history of the Bearpaw sediments. 207
B.3. Energies of repulsion, attraction, and net curve of interaction
forparallel flat plates. 209
B.4. Schematic of the stress history of an over-consolidated clay over
geological time. 211
B.5. Schematic showing void ratio response to reloading after rebound. 212
B.6. Aging affects observed in the laboratory for a normally consolidated
Clay. 213
B.7. Schematic illustrating aging effects resulting from sustained loading
Over geological time. 214
C.1. Schematic illustrating various shapes and roughnesses that are
possible in joints. 218
C.2. Schematic of joint models used by Patton. 219
C.3. Comparison of normalized shear strength of a joint as predicted by
equations of Ladanyi and Archambault, and the model of Patton 221
C.4. Roughness profiles and corresponding ranges of JRC values
associated with each. 222
C.5. Plots showing dependency of shear strength on scale and JRC. 223
C.6. Critical state model of Roberds and Einstein for the behavior of
rock joints. 224
C.7. Shear strength of a rough joint as a function of the joint-fill
Thickness. 226
D.1. Two-dimensional friction model for granular materials. 229
TABLES

2.1. Landslide susceptible clay shales in United States 11

2.2. Shear strength characteristics of the Crete Nere in the Sinnin Valley,
Italy 25

2.3. Mineralogy and plasticity of various clay shale materials 28

2.4 Various strength parameters for unaltered and softened clay shale
units worldwide 30

2.5. Mohr-Coulomb strength parameters for unaltered and softened clay


shale units 31

2.6. Summary of clay shale data presented in Tables 2.3, 2.4, and 2.5 33

3.1. Relative susceptibility of various clay types, based on ratios defined by


Bjerrum (1967) 43

6.1. Calculated mineralogy mode of Pierre shale using microscopic and


x-ray data 98

6.2. Procedure check-list for triaxial tests on Pierre shale samples 106

7.1. Values for void ratio, saturation, and shrinkage of Pierre shale,
measured at the end of the drying period 112

7.2. Values for axial and radial strains, void ratio, water content, and
degree of saturation resulting from rewetting of Pierre shale
samples 116

7.3. Testing program on Pierre shale 119


7.3. Values of Youngs modulus and Poissons ratio for loading
segments II and III for Pierre samples undergoing softening by
slaking 126

7.4. The change of specific volume, v, with stress, P', for loading
segments II and III in Pierre shale 131

8.1. Normalized peak strengths for all samples of Pierre shale,


unaltered and softened 150

8.2. Cohesion and internal friction angle values unaltered Pierre shale,
as well as for samples which have been rewetted after various periods
of drying 150

A.1. Geological classification of mudrocks by Ingram (1953) 153

A.2. Geological classification of mudrocks by Folk (1968) 191

A.3. Classification scheme of Skempton and Hutchinson (1969) 191

B.1. Estimated maximum preconsolidation loads on North American


clay shale units 207
1

CHAPTER I

INTRODUCTION

Clay shale deposits throughout the world have become notorious as a result of the
numerous foundation and slope stability problems with which they are often associated.
Past engineering failures have demonstrated that experimental and theoretical methods
typically used in geotechnical engineering practice are not adequate for determining the
stability of these unique materials. Present engineering design in clay shales relies primarily
on experience obtained from past failures, and results in the liberal use of highly
conservative and costly factors of safety.

The major difficulties in assessing and predicting the engineering behavior of clay shales
can be attributed to two unique properties of these materials: (a) clay shales are
intermediate in behavior between rock and soil, and (b) clay shales tend to transgress from
rock-like to soil-like materials within a relatively short time period. Changes in the
strength of clay shales can be very drastic, commonly exhibiting 40% to 80% reductions in
shear strength over periods ranging from 2 to 70 years. Internal friction angles of 20o to
30o in unaltered clay shales are often reduced to extremely low values of 2o to 6o after
softening.

The factors that control the magnitude and time frame of these changes have not been well
understood nor have they been seriously investigated. Evidence suggests that softening
along fissures may play a very important role in the rapid loss of strength in clay shale
deposits. However, the geotechnical and geological literature is surprisingly devoid of
systematic studies concerned with theoretical or experimental aspects of fissure
deterioration in clay shales.

In addition, it is widely recognized that clay shales which have been totally or partially
dried, exhibit a strong tendency to slake (i.e. disperse) during rewetting. This susceptibility
of clay shales to slaking is certainly a major factor, if not the major factor, involved in the
softening of clay shale deposits in the field. Several techniques have been developed to
assess the slake resistance of geological materials under unconfined conditions. However,
slaking in the field typically occurs in the presence of confining stresses. Yet, surprisingly
2

few reported studies have investigated the slaking resistance of geological materials under
conditions of confinement. Furthermore, the author has found no published systematic
investigations into the effects of slaking on the strength and stress-strain behavior of any
geological materials.

The research presented in this dissertation was undertaken in order to provide initial
theoretical and experimental assessments of the effects of fissure deterioration on the
strength and stress-strain behavior of clay shales. In particular, the drastic changes that can
occur in response to wetting and drying cycles (i.e. slaking) are examined in detail. A
primary contribution of this research is the incorporation of the slaking process into the
critical state soil mechanics model.

Organization of Dissertation
Primarily due to the transitional nature of clay shales, a somewhat multidisciplinary
knowledge base is necessary before one can fully understand clay shale behavior. The
author therefore conducted an extensive literary review, concentrating on field and
laboratory observations of clay shale behavior in various parts of the world, in addition to
such subject matter as the mechanics of joints and jointed rock, the theory and
measurement of slaking potential, and the critical state soil mechanics model. The author
augmented these literary studies with personal fields studies of clay shales in southern Italy
and in South Dakota, USA.

One of the intentions of the author in writing this dissertation has been to present an
insightful review of many of the problems associated with clay shale deposits. However, in
order to further preserve the coherency and continuity of this dissertation, reviews and
discussions on tangential subject matter have been placed within separate appendices, and
will be referred to in the appropriate sections of the main body. These include reviews and
discussions regarding the classification of clay shales (Appendix A), the formation of clay
shales and the nature of clay shale bonds (Appendix B), the mechanics of clean and filled
joints (Appendix C), and the mechanics of a clay mass consisting of intact cores
surrounded by a mud matrix (Appendix D). For the enthusiastic reader, these appendices
provide concise reviews into the appropriate subject, and hopefully some insight into the
complexities of clay shale behavior.
3

The primary background material for this dissertation is presented in Chapter II and III.
Chapter II includes a review of the major difficulties encountered by engineers dealing
with clay shales, and discusses the inadequacies of present geotechnical techniques to
accurately assess the long-term stability of clay shale deposits. As discussed previously,
most of these difficulties have resulted from our inability to adequately account for
softening in clay shales. In Chapter III, the author reviews and discusses the various
mechanisms proposed to account for the dramatic softening in clay shales.

The main purpose of this dissertation is to investigate the potential changes in strength and
stress- strain behavior that might occur as a result of fissure deterioration, particularly in
response to wetting and drying cycles. The author considers the theoretical aspects of the
softening problem from two particular viewpoints. First, in Chapter IV, a model for the
progressive deterioration of a fissured clay shale is presented. This model considers an
initially unaltered clay shale consisting of fissures along which softening agents are
introduced. Softening of the clay shale mass is initiated along the fissure walls and
advances into the mass by progressively increasing the depth of softening perpendicular to
the fissure wall. From a mechanical viewpoint, the clay shale can therefore be considered
as passing through four stages: (1) initially considered as a stiff clay with a network of
clean fissures, (2) then as a stiff clay consisting of fissures filled with soft clay, (3) next as
a matrix of soft mud surrounding intact cores of stiff clay, and (4) finally as a thoroughly
softened clay mass. The author considers, in theory, the effects that this progressive
softening may have on the strength and stress-strain response of the clay shale mass.

Second, in Chapter V, the author introduces the possibility of incorporating the effects of
slaking into the 'critical state soil mechanics' model. The role of slaking is seen in this
context as simply another path for altering the water content of the clay shale, while the
strength envelope remains constant or changes predictably. The importance of this
possibility cannot be overstressed. If the process of slaking can be constrained by the
critical state model, then the engineering behavior of a clay shale which has undergone, or
will undergo, a complex history of wetting and drying, is dependent only on the final
specific volume and stress state, and not on the wetting-drying history. The critical state
model could therefore greatly simplify the monitoring and prediction of changes in the
strength of clay shales.

Finally, the author has performed laboratory experiments involving triaxial compression
4

tests on Pierre shale samples that have undergone various cycles of wetting and drying
while under confinement. The purpose of these experiments were two-fold: (1) to provide
initial assessment of the effects of slaking on the strength and stress-strain response of a
clay shale, and (2) to test the feasibility of incorporating slaking into the critical state soil
mechanics model. The methodology and results of these experiments are presented in
Chapters VI and VII, while the critical state model for slaking is evaluated and discussed
in Chapter VIII. Finally, the conclusions in Chapter IX present a concise summary of the
major points of the entire dissertation.
5

CHAPTER II

ENGINEERING PROBLEMS ASSOCIATED


WITH CLAY SHALE DEPOSITS

Clay shales are extensively exposed throughout the world and are invariably responsible
for numerous slope stability and foundation problems in these areas. Where clay shales
outcrop at the surface, the terrain is generally characterized by low-lying and gently rolling
hills. Where clay shales underlie more resistant materials, numerous slope failures can be
found in valleys and road cuts.

Clay shales have consistencies that range from stiff clay to shale. As discussed in detail in
Appendix A, the term "clay shale" is being increasingly used to define a stiff, fissured clay,
or shale, which is highly susceptible to significant deterioration as a result of interaction
with water. Besides resulting in very low strength, this transitional nature of clay shales
creates some special problems with regard to the analysis of slopes and foundations.

Geotechnical engineers are generally accustom to viewing geological materials as either a


rock, with engineering behavior primarily controlled by fissures and joints, or as a soils
whose behavior is highly susceptible to the fabric and water content of the intact material.
However, clay shales are intermediate between rock and soil, and typically exhibit
properties of both. Furthermore, the behavior of clay shales tends to transgress from rock-
like to more soil-like within time frames of a few months to 70 years or more.

These peculiar properties of clay shales create difficulties in analyzing these deposits by
means of standard laboratory and mathematical means. In this chapter, the author reviews
the engineering problems associated with typical clay shale deposits in four areas of the
world: North America, England, Italy, and Panama.

Evidence presented in the following sections suggests the importance of fissure


deterioration as a viable model for explaining much of the long-term reduction of strength
exhibited by clay shales. As indicated below and in the following chapters, there is a great
6

need for systematic research into the effects of fissure deterioration, and particularly the
effects of softening due to slaking.

Characteristic Problems Related to Clay Shale Deposits Around the World


British clay shales. The clay shales of England include the Lias, Oxford, and London
clays, and range in age from Lower Jurassic (Lias) to Eocene (London clay). Although
these clays vary significantly in mineralogy, depositional history, grain size distribution,
color, and even consistency, they are all characterized by overconsolidation, the presence
of fissures, and the tendency to exhibit significant reductions in strength over time periods
from a few months to 70 years or more.

In his paper on the Lias clay, Cassell (1948) discussed slides that occurred 27 to 70 years
after the slopes were cut at angles of 20 to 26 degrees. Shear tests indicated that the
strength within the slip planes was reduced to between 1/5 and 1/26 of the compressive
strength of the nearby undisturbed material. Cassel attributed these long term failures to
progressive deterioration of the Lias clay within the zone of fluctuating ground water
tables, and stated that factors of safety derived from circular arc theory and laboratory
tests on undisturbed materials were not adequate for assessing the stability of these clays.

Of all the clay shales of England, the stiff fissured London clay has become the most
notorious. Present deposits consist of a 5 to 15 meter (16 to 49 ft) mantle of brown,
oxidized London clay, underlain by less altered, yet fissured, blue-grey London clay. Some
of the reported failure surfaces have been forced into the blue clay by the presence of
retaining walls (Skempton, 1977). However, most of the slips occur predominantly within
the brown London clay, while some of the slip surfaces appear even to be controlled by
the contact of the brown and blue clays (Gregory, 1844; Skempton, 1942; Henkel, 1957).

The London clay is fairly uniform with regard to its geotechnical properties. The Liquid
Limit ranges from 70 to 90, and the Plastic Limit, while generally about 28, varies
between 24 and 32 (Henkel, 1957). The natural water content is typically slightly above
the Plastic Limit. The clay fraction of the brown London clay is about 55%, with the
mineralogy of the clay fraction being 47% illite, 35% montmorillonite, 15% kaolinite, and
3% chlorite (Burnett and Fookes, 1974). The relative abundance of montmorillonite
7

Figure 2.1. Tentative relation between average shear strength along slip surfaces
and time, for cuttings & retaining walls in London clay (Skempton, 1948).

Figure 2.2. Relationship between cohesion intercept and time between


construction and failure (Henkel, 1957).
8

in the London clay is in contrast to the clay mineralogy of the other British clay shales, in
which expandable clays are reported to be rare (Attewell and Taylor, 1973).

As in the Lias clay, slope failures have occurred in the London clay tens of years after the
slopes were constructed. In slides involving cuts and retaining walls constructed in the
early 1900's, Skempton (1948) calculated a reduction of the original shear strength from
2600 psf to only 700 psf after periods of 7 to 30 years. Tentative relations between the
shear strength and the time after excavation are presented in Figs. 2.1 and 2.2. These plots
indicate that the London clay exhibits a 50% loss of strength after 10 to 15 years, and a
loss of about 70% to 80% of its strength after some 70 years.

The loss of strength in the London clay appears to be related to a loss of cohesion. As
reviewed in Skempton, 1977), standard triaxial and shear box tests indicate strength
parameters of
o
c' = 14 kN/m2 (2 psi) φ' = 20 ,

while triaxial tests on large diameter specimens (250mm) give values of

o
c' = 7 kN/m2 (1 psi) φ' = 20 .

As illustrated in Fig. 2.3 from Chandler and Skempton (1974), the results from back
analysis of first-time slides indicate that the most realistic values for the strength in the
field are given by

o
c' = 1 kN/m2 φ' = 20 ,

with a lower limit of

o
c' = 0 kN/m2 φ' = 20

(Fig. 2.3). It is important to note that the effective angle of friction calculated for first-time
o
slides is significantly above the residual friction angle of 13 , and does not exhibit a
reduction with time.
9

Figure 2.3. Failure envelope for London clay based on back-analysis of first time
slides (Skempton, 1977, after Chandler and Skempton, 1974).

The mechanism by which the cohesion of the London clay is reduced toward zero is not
fully understood. Several authors have noted the presence of very moist, softened zones
along fissures both within and outside of the failure zones, suggesting that the influx of
water into the clay may result in softening with time (Gregory, 1844; Delabache, 1844;
Terzaghi, 1936; Cassel, 1948; Skempton, 1942; Henkel, 1957). However, based on
piezometric measurements within "old" slopes, Skempton (1977) has proposed that the
long-term reduction of strength in the London clay may be related to the very slow
equilibration of negative pore water pressures. These two models will be discussed in
more detail in the following chapter.

Clay Shales of North America. Troublesome clay shales are widespread within the U.S.
and Canada, and have been responsible for costly problems in many major construction
projects and along many highways. Many clays and shales in the U.S. have not been
recognized in the literature as "clay shales", although they characteristically exhibit a
tendency to lose shear strength due to interaction with water. A list of some of the
10

landslide-prone clay shales in the U.S. was compiled by Fleming et al (1970) and is
presented in Table 2.1. Most of these deposits are Cretaceous or younger (Pliocene,
Eocene, and Miocene), although a few are much older Paleozoic deposits.

The most notorious clay shales in North America are located in the upper Missouri and
South Saskatchewan River basins. This notoriety results partially from problems
encountered during four major construction projects at Fort Peck, Gardiner, Oahe, and
Garrison Dams, and partially from the enormous areal extent of these materials (see Fig.
2.4). In the U.S., the Pierre shale and its stratigraphically equivalent deposits alone
underlie an area of approximately 600,000 square miles and outcrop in an area of 230,000
square miles (Tourtelot, 1962). These deposits extent into large areas of Canada as well.

The most troublesome clay shales in this region are the Pierre, Bearpaw, and Claggett
shales. Portions of the Fort Union group and Judith River formation have also caused
engineering troubles to a lesser degree. These clay shales can in general be characterized
by overconsolidation, by the presence of slickensides and fissures, by high swelling and
high plasticity resulting from the presence of montmorillonite, by extremely high slaking
potential, and by very low cohesion and friction angles as exhibited under long-term
conditions in the field.

Although natural landslides occur throughout the clay shales of north-central U.S. and
central Canada, these slides were not considered a major engineering problem until the
construction of large dams and major highways began in 1933. In 1938, before the
completion of the Fort Peck Dam, a major landslide in the Bearpaw shale foundation
resulted in re-calculation of strength parameters, as well as a year delay in the completion
of the project (Fleming et al, 1970). Subsequently, zones of weathered clay shale were
identified at the Fort Peck Dam site, extending to depths of 30 to 50 feet. Although the
water content of the unaltered Bearpaw shale varies between 11 to 18%, the water
content in the weathered zones were as high as 40%. The landslide at Fort Peck Dam
apparently resulted from failure to account for overstressing of this degraded clay shale
(Middlebrook, 1942). Slopes at the dam were redesigned using shear strength parameters
o
of c = 2.8 psi and φ= 10.5 .
11

Table 2.1. Landslide susceptible clay shales in United States (Fleming et al,
1970).
12

Figure 2.4. Location map for major dams built in clay shales in the North
Central United States and Canada (Fleming et al, 1970).

Similarly, the Bearpaw shale at Gardiner Dam has been arbitrarily divided into three zones
based on consistency: soft, medium, and hard. The upper soft zone has been significantly
disturbed and softened by swelling and weathering, with fissures and slickensides
frequently showing signs of past desiccation. The intact material between these fissures
can be readily remolded with the fingers (Peterson et al, 1960). In contrast, the lower hard
zone is less disturbed, more uniform, harder, more dull in appearance, and shows fewer
slickensides. The medium zone is transitional between these two zones. The water content
varied from only 20 to 27% in the hard zone to 29 to 36% in the soft zone. Similarly, the
unconfined compressive strength varied from 400 psi in the hard zone, to a remarkable
low of only 7 psi in the soft zone.

Although laboratory tests at Gardiner Dam gave very consistent strength values for the
hard zone, extreme variation of water content and consistency, and the presence of
slickensides, made it virtually impossible to obtain useful strength parameters by
laboratory testing of the soft shale. Final design criteria were based primarily on analysis of
slopes that had failed in other areas or during construction at Gardiner Dam. Failure zones
13

at Gardiner Dam were invariably within the soft shale zone, and often near the bottom of
the zone. The design strength parameters which were established based on these criteria
o
are presented in Fig. 2.5, and indicate values of c' = 0 and φ' = 6.5 for the softened
material.

Similar difficulties were encountered during the construction of Oahe Dam in South
Dakota. Direct shear tests on firm samples of the underlying Pierre shale gave values of

o
c = 8.3 to 12.5 psi and φ′= 20 to 22

for residual strength. Direct shear tests on the under lying weathered Pierre shale indicated
much lower shear strength values of

o
c = 3.6 psi and φ′
= 11.9 ,

which were adopted as the design criteria. However, slope failures in the softened shale
during construction required recalculation of the strength parameters to even lower values
of

o
c = 2.1 psi and φ′
= 8.5 .

Later slides indicated that these values also slightly overestimated the shear strength of the
slope abutments. The decision to completely remove the weathered shale from one of the
slide-prone abutments resulted in an extra 6.5 million cu yd of evacuated material (Fleming
et al, 1970). In addition to slope failures, the Pierre shale at Oahe Dam was responsible for
excessive rebound of the outlet-works stilling basin and resulted in redesign of the basin
and anchoring of the underlying floor.

The creation of these dams and their reservoirs has resulted in relocation of many roads
and bridges. In addition, modern highway design criteria for grade and alignment required
numerous road cuts and fills. Since these changes, landsliding has become a major problem
along highways in the north central U.S. and Canada.
14

Figure 2.5. Proposed clay shale design strength parameters for Bearpaw shale at
Gardiner Dam (Fleming et al, 1970).
15

Bruce and Bump (1967) reported on major difficulties encountered during the
construction of a 12-mile extension to Highway 44 in South Dakota. Within this short
distance, major failures of the Pierre shale occurred in four natural slopes and under three
fill deposits while still in the construction stage. Excavations of the failed slopes uncovered
springs and concentrations of discontinuities. Failure zones under the 40 and 65 ft high
fills extended 10 to 40 ft below the natural ground surface. At completion of the project in
1966, instability of the Pierre shale was responsible for a cost increase of $409,800 over
the original contract, or an additional $25,234 per mile of finished highway. Although the
slopes were flattened and all active material was removed, Scully (1973) reported that
movement began again in 1969. In 1984, the present author similarly noted many very
recent slope failures along this highway section. Construction and maintenance of such
highways in areas of underlying clay shales has been very costly.

Within the clay shales of North America, failures have occurred in both natural and man-
o
made slopes with inclinations as low as 3 . Most major failures in these materials have
been related to the presence of very low-strength zones of softened and weathered clay
shale. Laboratory testing has proved virtually useless for establishing design criteria in
construction projects, and invariably overestimates the in-situ strength of the foundations
and slopes. Slope design in the clay shales of north central U.S. and central Canada has, by
necessity, relied primarily on the analysis of local failures in natural and man-made slopes.
This has resulted in costly and dangerously unreliable engineering design.

Clay Shales Along the Panama Canal. The excavation of the Panama Canal was greatly
hindered by landsliding in the clay shale phases of the Cucaracha formation. The most
troublesome slides were undoubtedly the East and West Culebra slides which alone added
more than 50 million cu yd of material that needed to be excavated. These slides, which
began in 1907, converged in massive failures in August and September 1915, accompanied
by upheaval of the canal bottom and blockage of the canal. Since 1916, movement of
these slides has continued in the form of slow, sporadic flows into the canal.

Binger (1948) describes the clay shale of the Cucaracha formation as

...the clay shales all appear to be disturbed, and they contain many
degrees of slickensides and fractures or joints .... Badly crushed,
gougelike zones of varying thicknesses have been encountered in virtually
16

every exploratory drill hole reaching these shales. A zone of such material
having a thickness of more than five feet ... was found in a testpit
excavation. The material in this zone was so soft that it flowed into the
excavation even when the wall bracing’s were carried within a foot of the
bottom of the shaft.

The water content of the solid Cucaracha clay shale is about 17 to 18% (Binger, 1948),
and is generally 7 to 15 below the Plastic Limit (Banks, 1971). In slickensided zones at the
probable depth of sliding, within badly crushed or gouged zones, and within slide debris,
the water content can range as high as 30 to 35%. Piezometric measurements indicate a
reduced pore water pressure (i.e. less than the canal level) in the East and West Culebra
slopes, probably reflecting rebound and swelling occurring as a result of canal excavation
(Banks, 1971).

Analysis of slopes at the Panama Canal indicates drastic reductions in strength have
occurred within the Cucaracha clay shales. Within five months in 1912, the effective shear
strength dropped by 20%. From 1912 to 1915, this had been reduced 22% on the east
bank and 35% on the west bank. By March 1947, strength along the east and west banks
was about 20% of the original strength as measured in 1912 (Binger, 1948).

In recent studies, Banks (1978) observed three modes of failure in the Cucaracha clay
shale slopes:

(1) first time slides under short-term conditions,


(2) first time slides under long-term conditions,
(3) slides along pre-existing slip surfaces.

He also performed a wide variety of shear strength tests on four groups of samples:
(1) those apparently void of slickensides,
(2) samples with obvious slickensides,
(3) samples repeatedly sheared until residual strength was reached, and finally,
(4) remolded samples which were consolidated from a slurry (‘fully softened” samples).
17

Figure 2.6. Summary of the shear strength data for the Cucaracha clay shale
(Banks, 1978).

The results from these tests are plotted in Fig. 2.6, and indicate a wide range of values for
"intact" specimens. In addition, slurry consolidated samples exhibit a strength that is
slightly higher than residual strength.

After analyzing failed slopes in the Cucaracha clay shale, Banks concluded that:

(1) the strength mobilized in first time slides which occur immediately after
excavation is best represented by the peak strength of unaltered, slickensided
samples,
(2) the strength mobilized in first time slides which occur after long periods have
passed is equivalent to the "fully softened strength" obtained from slurry
consolidated samples, and
18

(3) the strength activated along pre-existing slip surfaces is of course equal to the
residual strength.

This significant observation implies that some natural "remolding" mechanism is acting on
the deposits of Cucaracha clay shale and reducing the strength toward that of a normally-
consolidated clay. Other observations discussed above, further suggest that the softening
of the Cucaracha clay shale is accompanied by an increase in the water content.

Italian Clay Shales. The Italian Peninsula has had a long history of landslide problems.
The vast majority of landslides in Italy occur within clay shales and rock/clay shale
melanges. As seen in Fig. 2.7, the presence of landslides in any area is strongly influenced
by the presence of these deposits (Esu, 1977). In southern and central Italy, as well as in
Sicily, these deposits generally belong to either the Liguride Complex, an Eocene flysch
consisting of alternating layers of black clay shales with various marine clastics, or the
Sicilide Complex, a Cretaceous to Oligocene deposit consisting of red and green clays or
clay shales, tuffite, and an arenaceous flysch unit (D'Argenio et al, 1975).

Although the clay shales of the Sicilide Complex cover only 10% of the entire Apennine
area in southern Italy, Belviso et al (1977) have estimated that this unit alone accounts for
up to 90% of the landslide events in this area. The terrain underlain by thick clay shale
deposits in Italy is characterized by numerous old landslides, and gentle rolling hills that
contrast sharply with nearby steeper peaks composed of other geological materials. The
o
slopes formed in the clay shales generally have inclinations no greater than 10 to 20
(Evangelista et al, 1977).

Villages in the countryside of southern Italy have historically been built on slopes or at the
summit of hills. Because deposits of clay shale underlie many of these villages, their
foundations have become unstable resulting in cracking, shifting, and even sliding of
buildings and streets. A consortium of geologist and geotechnical engineers was
established in the 1970's to study the foundation problems that threaten the village of
Bisaccia. These problems appear to be directly related to the instability of the variegated
clay shales that underlie the foundation material of the village. Similarly, enormous,
reactivated slides and flows which threaten the villages of Calitri and Senerchia are at least
partially affected by underlying layer of variegated clay shales (Cotecchia, 1982; Maugeri
et al, 1982).
19

Figure 2.7. Map of the landslide areas and their relationship to underlying rock
type in Italy (Esu, 1977).

The Italian clay shales can, in general, be described as an overconsolidated clay of medium
to firm consistency. The clay mineral fraction is predominantly kaolinite and smectite
(montmorillonite) with lesser amounts of illite. These clay shales generally exhibit two size
orders of fabric complexity, which are probably indicative of tectonic shearing. Samples
can be easily separated into relatively coherent rhombohedral lenses, on a scale of 1 to 10
centimeters, bounded by curved or planar shear discontinuities that are polished and often
20

striated. With significantly more effort, these lenses can be further broken down to expose
a more scaly fabric inside.

In the Sicilide Units, this complex fabric is accompanied by highly distorted reddish-brown
and olive green color banding of thickness ranging from centimeters to several meters.
Therefore, the clay shales are often referred to as "argille scaliose (scaly clays)", "argille
varicolori (varicolored or variegated clays), or "complesso caotico (chaotic melange)",
depending on the preferences of local geologists and geotechnical engineers and on the
dominant characteristics of the clay shale.

The geotechnical behavior of the Italian clay shales is characterized by high plasticity,
medium to high slaking potential, high swelling potential, and a tendency to soften under
environmental conditions found in the field. This softening, as well as the presence of the
complex fabric of the clay shales, greatly complicates slope stability analysis in areas
where these materials are involved. The geotechnical properties of the Italian clay shales
have been shown to vary significantly with composition (Belviso et al, 1977), and
direction of shear (Picarelli, 1981; Fenelli et al, 1981).

However, much of the scatter of shear strength values may also result from varying
degrees of alteration within the samples. In the field, the clay shales of Italy have been
divided into three zones based on the amount of weathering that has occurred (Fenelli et
al, 1982). They are indicated as Zone I for deeper, unaltered material, Zone II for partially
altered clay shale, and Zone III for overlying totally altered mudstone. In Zone I, the clay
shale is characterized by the complex rhombohedral structure described earlier. In
contrast, the altered material in Zone III consists of fragments smaller than one centimeter,
forming a mass with higher porosity in which the original structures have been obliterated
(A.G.I., 1977).

As presented in Fig. 2.8, the water content significantly increases as weathering


progresses, while the unconfined shear strength is drastically reduced. The unaltered clay
shale in zone I exhibits an unconfined compressive strength as high as 12 MN/m2
(83 kips), while the altered material and the material sampled from landslides both show
very low strengths of 0 to only .5 MN/m2 (3.4 psi). The plot also illustrates the rather
drastic decrease in strength which can occur with very little change in water content. In
21

fact, an increase in the water content results in a 90 to 95% reduction of the unconfined
strength.

Several researchers have alluded to the role of softening in controlling the in-situ behavior
of the Italian clay shales. D'Elia (1980) performed direct hear tests on three groups of
specimens taken from a slide area in Italy: (1) fresh material which had not undergone any
significant alteration, (2) "partially softened" clays which had been stored in a humidity
room for about four months and (3) "softened" material in which remained in the humidity
room for eight months. As presented in the Mohr-Coulomb plots of Fig. 2.9, the strength
o
parameters measured were reduced from c' = 35 KPa (5.1 psi) and φ' = 28 for the
o
unsoftened material to c' = 12 KPa (1.7 psi) and φ' = 18 for softened material. Back-
analysis of slope failures show that strength mobilized in deeper slides is similar to the
strength parameters measured for "partially softened" material and is much greater than
the residual strength. Furthermore, some translational slides of weathered clays exhibited
strengths as low as the strength of the "fully softened" material.

Similarly, Manfredini et al (1981) investigated the influence of softening within the black
clay shales (crete nere) of the Liguride Complex, and reported that the slaking potential of
these montmorillonite-poor clay shales was greatly increased by the presence of a scaley
fabric. Slake durability tests indicated that for non- scaley specimens, 60 to 90% of the
original material remained after two cycles, whereas only 15-25% remained in specimens
where a scaley fabric was present. In the field, weathering tends to obliterate original
structure of the material, resulting in a homogeneous, soft, and relatively plastic clay
matrix with remnants of scales and plates. The results of triaxial and direct shear tests are
presented in Table 2.2, and indicate that the strength parameters of the in-situ and earth
flow materials, as measured in the laboratory, are much higher than those for completely
remolded samples. However, back- analysis of the strength mobilized along the slip
surface of the slides indicates strength parameters which are close to the peak strength of
remolded samples. Again these values are well above the residual strength of the clay
shales.

Cancelli (1981) investigated the effects of softening in the Lugagnano clay shale which is
responsible for numerous slope stability problems in northern Italy. While the Lugagnano
clay shale appears to be void of the complex scaley fabric of the southern Italian clay
22

Figure 2.8. Relationship between unconfined compression strength and water content
for the different degrees of weathering in the Laga formation, Italy (A.G.I., 1977).
23

Figure 2.9. Variation of shear strength in an Italian clay shale as a function of time of storage in a
humidity room (D'Elia, 1980).
24

and shales, it is highly fissured and undergoes significant softening in natural slopes.
Results from triaxial and direct shear tests on undisturbed core samples are presented in
Fig. 2.10, indicate that the softening of in-situ Lugagnano clay shale results in a reduction
of the effective cohesion from 55 kPa (8.0 psi) to zero, while the effective friction angle
remains the same, or is only slightly reduced. The residual strength is represented by an
o
effective cohesion values of zero and a significantly lower effective friction angle of 11 to
o
13 . Values obtained from back-analysis of several periodic slides indicate that the
mobilized strength is very close to the residual strength values. However, assuming c' = 0
o
for back-analysis of first-time slides, indicates an effective friction angle of 24 . This value
is slightly lower than that measured for the softened material, but is much higher than the
residual friction angle.

Residual Versus Softened Strength


In order for the strength of a clay shale to be equal to its residual value, it is generally
assumed that the material must experience a shear deformation in excess of that required
to mobilize the peak strength. It has long been recognized that the resistance mobilized by
reactivated landslides is equal to the residual strength of the material within the slip zone.

In addition, some first-time slope failures have occurred under states of stress which
would have mobilized only residual strengths (Krahn et al, 1979; Palladino and Peck,
1972). Such slope failures can generally be contributed to slip along surfaces which have
been pre- sheared by tectonic deformations, or by deformations resulting from unloading.
These pre-sheared surfaces can result from passive failure during the erosion of overlying
sediment (Nichols, 1980) or the cutting of river valleys (Matheson and Thompson, 1973),
from ice thrusting during glaciation (Krahn et al, 1979), or from differential unloading
during progressive deglaciation of valleys (Palladino and Peck, 1972).

In his presentation at the Seminar on the Geotechnics of Clay Shales in Denver on March
15, 1984, Ralph Peck presented numerous examples of slope failures in various clay shales
of the United States in which slip apparently mobilized residual strength along sub-
horizontal, pre-sheared bedding planes. He further advocated that the presence of pre-
sheared horizontal surfaces are so prevalent in some clay shale deposits, that their
25

Figure 2.10. Shear strength envelopes for undisturbed core samples of firm (φp')
and weathered (φs) Lugagnano clay shale. Residual strength is indicated by φr
(Cancelli, 1981).

existence should be assumed in projects where a potential slide would be intolerable.


Unfortunately, there is a growing tendency for researchers and practicing engineers to
consider the assumption of residual strength as a "catch-all" cure for the complications
encountered in clay shale deposits.

However, there is a vast amount of evidence indicating that most first-time slides within
clay shales occur at strengths significantly greater than residual strength. Furthermore,
there is no evidence suggesting that the strength of all slopes in clay shales will be reduced
to residual values within the engineering lifetime of these projects. This is certainly the
case for the London clay, in which changes in the stability of man-made slopes have been
measured for up to 70 years after construction.

As reported by Skempton (1977):


26

It appears that the displacements preceding a first-time slide are sufficient


to cause some progressive failure, reducing the strength toward the fully
softened or the lower limit of fissure strength; but the displacements are
not so large as to reduce the strength to the residual value.

Evidence presented earlier in this chapter indicates that first-time slides within the clay
shales of the upper Missouri and South Saskatchewan River Basins, as well as the
Cuacarcha clay shale in Panama and the clay shales of Italy, mobilize softened strengths
which are significantly greater than the residual values.

In cases where old landslides are being reactivated, or where pre-sheared surfaces are
known to exist, residual strength must be assumed along these surfaces. If the location and
orientation of pre-sheared surfaces cannot be accurately determined, then it is wise to
assume that the residual strength will be mobilized along the entire slip surface. However,
in areas where pre- sheared surfaces have not been recognized, and where a landslide,
though troublesome, is tolerable, assuming residual strength will probably result in
overconservative design and can drastically increase the construction expense of
excavation or of building slope and foundation supports.

Comparative Summary and Discussion


The above review briefly illustrates some of the extreme challenges faced by the engineer
dealing with clay shales. Our present understanding of the engineering behavior of these
materials is primitive relative to the current state-of-the-art for other geological materials.
Engineering designs based on traditional methods for analyzing geological materials have
been dangerously inadequate. In fact, most successful designs in clay shales have been
based on past failures, rather than on adequate test results.

The author feels that most of our past inadequacies in dealing with these materials have
resulted from our lack of appreciation of their transitional nature. This has resulted in
dangerous and costly slope and foundation designs. The transitional nature of clay shales
is both temporal and physical. Physically, clay shales are transitional between rock and
soil, and therefore exhibit properties of both. This has been a source of problem for
geotechnical engineers, who traditionally view geological materials in terms of rock
mechanics or soil mechanics, both rarely in terms of both. In addition, clay shales are
27

transitional in time, and tend to transgress from rock-like behavior to soil-like behavior
within a relatively short time period. Such rapid changes in material properties create
challenges in classification and engineering design, both of which are traditionally based on
material properties as they exist at the present and not on possible future properties.

Tables 2.3 to 2.6 list reported values for several geotechnical properties of these clay
shales. The mineralogy and plasticity values presented in Table 2.3, illustrate that the
characteristically high to very high plasticity of clay shales. Tables 2.4 and 2.5 list various
strength parameters which have been measured for unaltered and softened clay shales, as
well as those strength values which have been calculated based on back calculation of
failed slopes. Table 2.6 summarizes all of this information in a more concise format.

The strength data presented in these tables, as well as the above review, indicate that some
first-time slides occurring immediately after excavation in clay shale have mobilized the
peak strength of jointed or slickensided samples. However, the peak strength of intact
material is never mobilized by clay shale deposits in the field, indicating the importance of
considering the jointed, and therefore rock-like, nature of clay shales. In cases when
failure has occurred along pre- sheared surfaces, the mobilized resistance is characterized
by the residual strength. However, the growing tendency for practicing engineers to
always assume that clay shales will mobilize the residual strength within the life-time of an
engineering project is not supported by the data presented, and can lead to costly
overconservatism. Indeed, much of the evidence presented here indicates that the strength
typically mobilized in the field during first-time, long-term slides, is characterized by
values which are well below peak strength and well above residual strength.

The difference between the strength of unaltered clay shale and that of the fully softened
material can be quite large. Thus, the amount of strength reduction that takes place over
relatively short periods of time can be quite drastic. For example, the Cucaracha shale in
Panama lost 80% of its original strength within 35 years. Furthermore, this can result in
extreme scatter in strength values as measured within a single clay shale deposit at any
28

Table 2.3. Mineralogy and plasticity of various clay shale materials


29

Table 2.3 cont.


30

Table 2.4 Various strength parameters for unaltered and softened clay shale units worldwide
31

Table 2.5. Mohr-Coulomb strength parameters for unaltred and softened clay shale units worldwide.
32

Table 2.5. Cont.


33

Table .2.6 Summary of clay shale data presented in Tables 2.3, 2.4, and 2.5.
34

time. As an example of the latter case, consider that the unconfined compressive strength
of the Pierre shale has been reported as ranging from zero to over 17,000 kPa!

The drastic, progressive softening of clay shales with time, and the extreme variability of
test results create very difficult challenges for the geotechnical engineer dealing with these
materials. How does one account for such large changes in strength and compressibility
with time? What material and environmental factors control the amount and rate of change
that might occur over the lifetime of an engineering project, and how does one monitor
and predict these changes? Furthermore, even without considering the time factor, is it
possible for the engineer to account for the extreme variability of strength that can occur
within a given deposit, without relying on a prohibitive amount of site investigation?

All of these questions must be answered before the softening of clay shales can be
adequately accounted for. In this dissertation, the author investigates these questions both
theoretically and experimentally. The following chapter reviews the present state of
knowledge regarding the mechanisms by which clay shale softening might be occurring,
and in particular, discusses the lack of systematic studies on the effects of slaking in the
field.
35

CHAPTER III

SOFTENING MECHANISMS IN
CLAY SHALE DEPOSITS

The review and discussions of the previous chapter emphasized the importance of
material softening in controlling the strength and compressibility of clay shale deposits.
The softening of clay shale was shown to result in up to 80% loss of strength in some
deposits after 30 to 70 years, and was probably responsible for the extreme variation of
material strength observed at several engineering sites. In order to be able to account
for such changes and variations in the engineering behavior of clay shales, it is
important to understand the mechanisms by which the material properties of clay shales
are altered. While many researchers have speculated on the causes of softening in clay
shales, few have carried out extensive investigations concerning these possible
softening mechanisms. Therefore, there is very little understanding of the factors which
control the amount and rate of softening in clay shale deposits. Certainly, the lack of
such studies is one of the major reasons that clay shales have remained one of the most
difficult engineering materials.

Two primary mechanisms have been proposed as playing important roles in the
softening of clay shales. These include the equilibration of negative pore pressures and
the deterioration of fissures by means of chemical alteration or slaking. As discussed
below, these two mechanisms are not entirely independent of one another, and it is
probable that both act to some degree in all clay shale deposits. In addition, these
mechanisms surely act in conjunction with the progressive failure mechanism in such a
way that they are enhanced by, as well enhance, progressive failure. It may be difficult
to uncouple the effects of one mechanism from the other.

Still, it is essential that we recognize which material and environmental factors control
the extent and rate of softening at any given site. In addition, the engineer needs to
know the effect that a given amount of deterioration will have on the strength and
stress-strain behavior of the clay shale deposit. This chapter will briefly review the
present state of knowledge regarding the proposed softening mechanisms.
36

Softening Mechanisms
The importance of fissures. There are three important effects of fissuration in
controlling the behavior of clay shale deposits. First, they provide inherent planes of
weakness along which shear can occur. It is for this reason that clay shales must also be
considered from a rock mechanics point of view. Second, the fissures greatly increase
the permeability of clay shale deposits. Without the conduits provided by fissures, most
clay shales would be virtually impervious. Third, fissures significantly increase the
surface area exposed to weathering agents.

Thus, in addition to significantly weakening a clay shale mass, fissures greatly enhance
the process of deterioration in clay shales, by first allowing a greater influx of water and
other weathering agents into the interior of the mass, and by then exposing more
surface area of the rock mass on which these agents can act. No matter which softening
mechanism is acting on a clay shale deposit, the rate and extent of softening is highly
dependent on the presence of fissures.

In addition, whether softening occurs in response to the equilibration of negative pore


pressures, or by chemical alteration or slaking, softening surely weakens fissure wall
material first before progressing further into the clay shale mass. For this reason, the
author proposes a simple fissure deterioration model in Chapter IV and discusses the
implications of this model with regard to the possible changes in engineering behavior
that can occur in response to progressive softening of clay shales.

Equilibration of negative pore water pressures. As unloading occurs in response to


excavation or natural erosion, negative pressures can be created within the pore fluid of
a clayey deposit. If the permeability of the clayey material is low, these negative pore
pressures may persist for extended periods of time, and may act to initially strengthen
the deposit. However, as these negative pore pressures are equilibrated by the influx of
fluid, the apparent strength resulting from the tensile pressures is destroyed.

Base on piezometric measurements within the blue and brown zones of the London
clay, Skempton (1977) proposed that long-term reduction in the strength of the
London clay might be related to extremely slow dissipation of negative pore pressures
37

Figure 3.1. Definition of Bishop's average pore pressure ratio, ru (Skempton,


1977).
which had been generated in response to the cutting of the slope. Previous
measurements in the London clay had indicated that the value of Bishop's average pore
pressure ratio, ru, as defined in Fig. 3.1, was in the range between 0.25 and 0.35 for
conditions of equilibrium. This value had been reached in slope of brown London clay
which had been cut 125 years previously. However, in the facing slope which had been
recut 19 years before the piezometric measurements were taken, the pore pressures
were only one-half of the equilibrium value. Skempton thus concluded that the
equilibration of negative pore pressures in the London clay required about 40 to 50
years, and could thus account for the loss of strength in slopes of London clay.

As been determined previously by several researchers, Skempton confirmed that the


strength mobilized at failure was equal to the "fully-softened strength. The very slow
equilibration of negative pore pressure was therefore not proposed as a mechanism by
38

which peak strength was exceeded or by-passed. The proposed mechanism does,
however, suggest that the delays in slope failures of the London clay might be
controlled by the time required to dissipate negative pore pressures, rather than the
time necessary for deterioration to occur. It is important to note that the slides
evaluated by Skempton in his 1977 report, deliberately excluded shallow slips and slips
in zones of seasonal variation. Unlike the slides evaluated, these slips typically occur
after exceptionally heavy rainfall, especially following prolonged dry periods. For these
cases, as well as for the slides investigated by Skempton, the mechanism of softening
must still be assessed.

That negative pore pressures still exist in the London clay 20 to 40 years after
excavations is both surprising and significant. It had been assumed that the presence of
fissures would allow rapid dissipation of negative pore pressures. However, it is not
necessary to implicate extremely low permeability to account for the presence of
negative pore pressures many years after excavation.

It is important to assess whether the negative pore pressures result entirely from the
initial slope cutting, or whether they might in addition result from soil suction
associated with material deterioration and swelling. Furthermore, it should be
determined whether failure of the London clay occurs primarily along fissures. If this is
the case, one must be concerned about measuring the equilibrium of pore pressure
along fissures, and not the equilibrium within the clay mass itself. Finally, equilibration
of negative pore pressure is certainly associated with the influx of more water into the
material, and it is important to consider how the generation and dissipation of the
pressures might occur in conjunction with the deterioration and swelling discussed in
the next section.

Fissure deterioration model. As early as 1844, an engineer, Gregory, and a geologist,


Sir Henry Delabache, observed the softening of fissures within the London clay and
suggested that progressive softening might account for slips which occurred several
years after construction. Soon after, Collins (1886) suggested that clay slopes along
canals would eventually be reduced to inclinations similar to those exhibited in the
natural terrain. These observations were not expounded on further until 1936, and
according to Skempton (1948):
39

one of the reasons for the long delay in the development of soil
mechanics can be found in the apparently 'treacherous' and
unpredictable behavior of stiff-fissured clays: especially since these are
wide-spread in south east England and in France.

Terzaghi (1936) reintroduced the idea that the strength of the stiff, fissured clays, such
as the London clay, depended on the spacing of the fissures and on the degree of
softening adjacent to the fissures. He also emphasized the behavioral difference
between the relatively stable non-fissured clays and the troublesome, fissured clays. The
process, as described by Terzaghi and Peck (1948), is as follows:

Almost every stiff clay is weakened by a network of hair cracks or


slickensides. If the surfaces of weakness subdivide the clay into small
fragments 1 in. or less in size, a slope may become unstable during
construction or shortly thereafter. ... If the spacing of the joints in the
clay is greater than several inches, slopes may remain stable for many
years or even decades after the cut is made. The lapse of time between
the excavation of the cut and the failure of the slope indicates a
gradual loss of the strength of the soil. Before excavation, the clay is
very rigid, and the fissures are completely closed. The reduction of
stress during excavation causes an expansion of the clay, and some of
the fissures open. Water then enters and softens the clay adjoining
these fissures. Unequal swelling produces new fissures until the larger
chunks disintegrate, and the mass is transformed into a soft matrix
containing hard cores. ... The water seems to cause only deterioration
of the clay structure; seepage pressures appear to be of no
consequence.

Since Terzaghi's statements, numerous researchers have observed softening and


increased water content of the London clay along failure zones and within unfailed
fissures (Skempton, 1942; Cassel, 1948; Henkel, 1957). It is of course important in
future studies to distinguish between softening which occurs prior to failure and that
occurring as a result of failure. Further evidence for the deterioration of fissures prior
to slope failure, comes from Hutchinson (1970) who noted that mudflows in the
London clay often consisted of a mud matrix surrounding hard clay fragments.
40

As discussed in the previous chapter, many slides have been occurred within the upper
partially weathered layers of the clay shales of North America and Italy. In addition,
numerous authors have reported localized softening occurring along failure planes and
irregularly into fissures in lower zones (Widger and Frelund, 1979; Scully, 1973;
Cancelli, 1981; Manfredini et al, 1981). Scully discussed in detail the very localized
existence of "seeps" consisting of highly weathered Pierre shale with water contents
greater than 130%. Even at depths of over 100', softened zones were found such that
the undrained compressive strength varied from the softened strength of 3 TSF to
85 TSF within a 2' distance. Certainly, the occurrence of these softened zones must be
joint controlled.

Deterioration of the strength along fissures can result from either (1) chemical
alteration, or from (2) increasing the amount of water adsorbed by the clay material.
Chemical alteration of a clay shale deposit can involve either the precipitation or
deposition of new minerals, or the chemical transformation of existing minerals.
Furthermore, chemical alteration can act to either decrease strength by breaking bonds,
by increasing porosity, or it can actually increase strength by creating new bonds,
particularly in the presence of cementing agents. The mineralogy can be altered such
that properties such as the swelling potential, or the residual strength, are either more
favorable, or less favorable to stability of the slope or foundation.

Although the effects of chemical alteration are probably of extreme importance in


controlling the long- term strength of clay shale deposits, this dissertation will not
discuss this process or its effects in any detail. Unfortunately, extensive investigations
regarding chemical alteration and its potentially important effects in clay shales are
surprisingly few. Our understanding of clay shale behavior would be greatly advanced
by more numerous and more extensive studies on chemical alteration in clay shales,
particularly if the results are incorporated into the fissure deterioration model to be
discussed in the next chapter.

However, the primary emphasis of the present investigation is on the processes of


slaking and swelling, during which bonds are destroyed and the moisture content
increased without altering the mineralogy of the clay shale. Evidence presented above
suggests that slaking probably play a major role in the deterioration of clay shales.
41

Therefore, the slaking process will be reviewed and discussed in more detail toward the
end of this chapter.

Progressive failure mechanism. Progressive failure is a mechanism by which the


strength of clay shale along a potential slip zone can be progressively reduced from
peak values to residual values. If the stresses at any position along this potential slip
surface exceed the peak strength of the material, and if displacement is allowed, then
localized failure will occur and the strength of the material at that position will decrease
toward the residual value (Skempton, 1964). In the example of a slope, illustrated in
the schematics of Fig. 3.2, the stresses at the toe of the potential slip surface were
sufficient to displace the material beyond the peak strength in the zone D-F. This will
cause a redistribution of stresses in the local area such that the peak strength might
likewise be exceeded in the adjacent material, as in zone F-G. This process will
continue until the average shear strength along the potential slip plane is no longer
sufficient to resist complete failure along the plane, or until the redistributed stress state
is such that the peak strength is no longer exceeded at any point within the slip plane.

Several factors determine the susceptibility of a material to progressive failure. Bjerrum


(1967) discussed several of these factors as they relate to a uniform slope. In order for
progressive failure to be initiated, stresses must locally exceed the peak strength of the
material. Thus, the danger of initiating progressive failure increases with an increase of
the ratio, ph/sp, where ph is the horizontal effective stress and sp is the peak effective
shear strength. Furthermore, the local differential strains in the advancing failure
surface must be sufficient to strain the clay beyond peak resistance. The ratio, eh/ep, is
a measure of the amount that the horizontal strain, eh, resulting from the removal of
lateral load, will exceed the peak failure strain, ep, and an increase in this ratio also
increases the susceptibility of the slope to progressive failure. The final ratio, sp/sr, of
peak strength to residual strength, indicates that progressive failure is also favored in
soils which undergo rapid and drastic reductions in strength after the peak strength is
mobilized.

Table 3.1 illustrates the relative danger of progressive failure for various materials. As
discussed in Appendix B, clay shales can be described as overconsolidated plastic clays
with strong bonds, and thus correspond to the third and fourth columns in the
42

Figure 3.2. Contours of shear strain in a slope, and its relation to


corresponding stress states in an overconsolidated clay (Atkinson and
Bransby, 1978).
43

Table 3.1 Relative susceptibility of various clay types, based on several ratios defined by Bjerrum.
44

table. As measured by the factors discussed above, unweathered clay shales exhibit low
susceptibility to progressive failure. However, the weathering of clay shales breaks
down bonds and creates conditions of very high horizontal stresses relative to
the vertical stresses. Duncan and Dunlop (1968) have shown that such high horizontal
stresses can cause a ten-fold increase of the maximum shear strength at the toe of a
slope where progressive failure would most likely be initiated. Therefore, clay shales
that have been weathered, or are in the process of weathering, are very susceptible to
progressive failure, and it appears that the mechanisms of progressive failure and clay
shale deterioration can act in combination to greatly increase the risk of slope failure.

Interactions between softening mechanisms. In the natural environment, the


softening mechanisms discussed above probably all act to some degree in all clay shale
deposits. In addition, the softening of clay shales may involve complex interactions
between these softening processes, and it may be difficult, if not impossible, to
uncouple the effects of one mechanism from those of another.

As an example, we will look at two fictitious cases, whereby the processes discussed
above are activated or accelerated by different "triggers". In the first case, the stress
state on a slope is altered by either excavation, or by the natural process of erosion.
This action results in two responses: fissures are propagated or opened, and negative
pore pressures are generated. The generation of negative pore pressures actually acts to
initially stabilize the slope, so that the eventual loss of this pore water tension can alone
result in slope failure some time after excavation or erosion. However, other processes
can in turn act to greatly complicate this simple explanation. For instance, the opening
of fissures and the presence of negative pore pressures will result in increased water
influx if a ready source is available.
As illustrated in the schematic of Fig. 3.3, this influx of water can then result in one, or
more, of the following actions:
(1) the negative pore pressure is reduced, thereby decreasing the effective shear
strength,
(2) bonds are broken, resulting in swelling and slaking, an increase in the negative pore
pressure, an increase in water content, and a reduction in the effective shear
strength,
(3) the influx of chemical agents is increased, alteration or deposition of minerals
occurs, and the effective shear strength is either increased or decreased.
45

Figure 3.3 Schematic complex action-reaction paths possible in clay shale materials.
46

The decrease in effective shear strength by any of these methods can thereby result in
viscous or "immediate" local strains, which in turn cause either (1) total failure, (2) a
reduction of strength toward residual values, or (3) dilation. If dilation occurs, then
negative pore pressures are again increased resulting in a "positive feedback"
mechanism by which the stability of the slope depends more and more on the presence
of negative pore pressures, and less on the progressively decreasing material strength.
As the above cycle continues, failure can occur if the effective stresses exceed the
mobilized strength, regardless of whether this occurs as a result of the dissipation of
negative pore pressure, or the reduction of material strength.

Several important observations can thus be made regarding negative pore pressures in
clay shale deposits. First, negative pore pressures, as measured by piezometers in the
field, can result from material slaking and swelling, or from dilation during shear, as
well as from those stresses generated by an excavation or erosion. Second, the
excessive length of time required for "dissipation" of negative pore pressures, as
reported for instance by Skempton (1977), may indicate the superposition of negative
pore pressures generated by swelling and dilation onto those resulting directly from the
excavation or erosion. Third, the discussion above suggests that it is not necessary to
assume very low permeability in order to account for the long times required for the
dissipation of negative pore pressures.

In the second general case to be discussed, the triggering action results from wetting
and drying cycles occurring in the absence of exterior stress changes. A common
characteristic of true clay shales is their tendency to swell and slake in response to
cycles of wetting and drying. As discussed above, this results in an increase in water
content, a decrease in material strength, an increase in negative pore pressure, and
possibly an increase in permeability as fissure walls are softened. In this case, negative
pore pressures can be attributed to capillary tension in the partially saturated material,
as well as the soil suction acting in the previous example. In addition to decreasing the
material strength, swelling tends to increase horizontal stresses, creating dangerous
conditions that favor progressive failure. Reduction of strength in this general case can
essentially follow the same cycle illustrated in Fig. 3.3, although the cycle is initiated by
a different process.
47

The two typical cases discussed above illustrate that the progressive deterioration of
clay shales in the field involves a rather complex interplay of the effects of dissipation
of negative pore pressures, chemical alteration, slaking and swelling, and the dilation
and fabric reorientation associated with shear strain. Dissipation of negative pore
pressures involves rather well- refined principles of fluid flow through porous media. In
contrast, the changes in strength and stress-strain response associated with mineral
alteration, swelling, or slaking may be more difficult to assess or predict.

In addition, the history of slaking and chemical alteration in clay shales can be very
complex. These processes have been investigated very little, or in some cases, not at
all. Any significant advances in our ability to analyze the behavior of clay shale
materials in the field, will probably result directly from investigations into the effects of
mineral alteration, swelling, and slaking.

Slaking in Clay Shales


As discussed in Appendix A, clay shales are characteristically highly susceptible to
slaking. It is highly probable that the process of slaking plays an important role in the
softening of clay shales. Numerous researchers have shown that the strength of clays
and clay shales is closely related to the water content, and that fissure walls in failed
clay shales often exhibit higher water contents than the adjacent clay material. In the
absence of mineralogical changes, water content can be increased by (a) dilation during
shear, (b) simple swelling related to elastic rebound following unloading, and (c)
swelling and slaking related to the breaking of interparticle bonds in response to
wetting, or wetting and drying cycles in the absence of external load changes.

The slaking process. The term "slaking" usually implies the phenomena of material
disruption or dispersion observed when dried or undisturbed chunks of clay or shale are
immersed in water. Moriwaki (1974) noted four modes of slaking, including

(a) swelling, described as an increase in bulk volume without visible cracking or


significant loss of material,
(b) body slaking, which appears to originate from internal processes and which rapidly
traverses large portions of mass with no apparent deterioration between cracks,
48

(c) surface slaking, characterized by loss of mass due to "sloughing" of tiny flakes of
grains from the entire surface with no apparent cracks in the underlying material,
and
(d) dispersion, characterized by loss of mass resulting from the separation of clay-sized
grains which go into spontaneous suspension, rather than settling.

As discussed by Bjerrum (1967), most slaking is assumed to result from the disruption
of diagenetic bonds and the release of stored strain energy. However, it has also been
shown that slaking can result from the compression of trapped air within the clay or
shale mass, particularly in soils containing highly-expansive clay minerals. As water is
pulled into the clay mass by capillary forces, or suction pressures generated by
expansive clay minerals, air pockets can become trapped and compressed. Wetting of
the mass stops when the pressure on the air pockets equals the suction or capillary
pressure of the water. However, the rock or clay mass may fail in tension before this
equilibrium is reached, resulting in body slaking, or possibly surface slaking, as
described above. The importance of this mechanism in some clays and shales can be
demonstrated by the absence of slaking when slake testing is performed under a
vacuum.

Some clays and shales at natural water content slake when immersed in water. Others,
when immersed, will remain stable with regard to slaking even though stresses are
released. However, if these materials are first dried, and then rewetted, slaking may
occur. In this case the process is referred to as "drying-induced" slaking. Nakano
(1970) presented evidence which supports the hypothesis that some materials will not
slake as long as the water content remains above a certain threshold, but if the water
content is lowered below this threshold, slaking will occur during either drying or
rewetting.

Moriwaki (1974) recognized three mechanisms of swelling and slaking, including (a)
swelling by osmotic forces, (b) swelling by hydration of ions and clay surfaces, and (c)
compressed-air slaking, and concluded from his investigations on artificial shales that
the dominant slaking mechanism is controlled by the mineralogy:
(a) sodium clays - osmotic swelling,
(b) calcium montmorillonite - osmotic swelling and hydration,
(c) calcium illite - hydration and compressed-air,
49

(d) calcium kaolinite - compressed-air only.

Moriwaki further concluded that the susceptibility of any material to slaking will
depend not only on the mineralogy, but also on the "physico-chemical characteristics",
such as bonding, and the chemistry of the slaking fluid.

McClure (1980) performed an extensive investigations on the slaking behavior of many


natural clays and shales, including many of the clay shales of North America. He
determined that, of all the natural materials he tested, (a) 40% behaved as expected
based on physico- chemical characteristics and Moriwaki's observations on artificial
shales, (b) 16% showed minor variation from expected behavior, (c) 20% deviated
significantly due to cementation, and (d) 20% more deviated significantly due to
unknown causes. McClure therefore concluded that the forces which bind shales and
clays together (i.e. diagenetic bonds) need to be further identified and understood.

Testing for susceptibility to slaking. The slake tests employed in the present
classification schemes of argillaceous materials include (1) the modified jar slake test,
(2) the slake durability test, (3) the "one- dimensional free swell test", and (4) the
sulfate soundness test. In the modified jar slaking test (Moriwaki, 1974), an
undisturbed or dried sample is placed on a wire mesh, which is then lowered into a jar
of water. The slaking potential is indicated by the loss of weight as the material
crumples and falls through the mesh. The slake durability test of Franklin and Chandra
(1972) is performed by rotating six 40 to 60 gram samples in a wire mesh drum, which
is partly immersed in water. The weight percentage of material remaining in the drum
after one cycle (200 revolutions in ten minutes) defines the "slake durability index". The
sulfate soundness test is similar to the above tests, except that the previously dried
samples are wetted in a sodium or magnesium sulfate solution. The percentage of
material retained on a 3/8" sieve indicates the sulfate soundness index.

The slaking factor of Morgenstern and Eigenbrod (1974) is unique in that it is based on
the one-dimensional free swell of a laterally-confined sample. In these tests, the change
of height, and therefore the change of water content, were measured as a function of
wetting and drying cycles. Increased swelling is assumed to indicate progressive slaking
within the specimen.
50

The modified jar slaking, slake durability, and sulfate soundness tests described above,
all provide very useful information on the relative "weatherability" of various
argillaceous materials. However, the extent to which these tests significantly model
conditions existing in the field is limited. All of these tests measure the amount of
slaking under conditions of zero confining pressure. The previous review of slaking
processes discusses several different mechanisms of slaking, each caused by various
magnitudes of disruptive stresses. It is conceivable that slaking forces exhibited by
some materials may be balanced by very low confining pressures. In such materials,
slaking would only affect the very top surface of deposits in the field. In addition, the
slake durability test employs a tumbling factor, which generally does not exist in the
field.

More important, however, is the fact that none of these tests measure the relative
reduction of shear strength that occurs as these materials undergo deterioration in the
field. Although these tests do measure the important tendency of a material to degrade,
they do not necessarily measure the relative reduction of strength due to degradation.

Summary of slaking in clay shales. The present methods of measuring slaking


susceptibility have been reviewed, and their deficiencies discussed. Evidence presented
in Chapters II and III indicates that slaking and swelling may be important factors in the
reduction of the strength of clay shales in the field. It is clear that the methods
commonly employed for investigating the process of slaking do not provide adequate
information regarding the effects of slaking on the strength and stress-strain behavior of
clay shales in the field.

First, the phenomenon of slaking is generally studied only under zero confining
pressures. The potential influence of both hydrostatic and deviatoric stresses on
inhibiting or enhancing the slaking process is not fully appreciated. Second, the slaking
in clay shales is primarily confined to the walls of fissures and joints, and it is not well
understood whether the process will enhance or inhibit further slaking by increasing or
decreasing the effective permeability of the clay shale mass.

Furthermore, the relationship between the processes of slaking and swelling is not often
fully appreciated. Swelling occurs as additional water is absorbed either within the
interlayers of the clay mineral structure or within the viscous water layer
51

Figure 3.4. Modes of swelling associated with simple rebound and slaking in
an overconsolidated clay or clay shale.

surrounding individual clay particles or aggregates. Swelling generally occurs in one of


two modes, both involving the disruption of interlayer or interparticle bonds. The first
mode involves a decrease in external stress, which results in the breakage of bonds, and
the subsequent pulling in of water in response to repulsion between particles. This is
the mode of swelling associated with simple rebound during the formation of
overconsolidated clay, and is represented by the path A-B in Fig. 3.4.

The second mode of swelling can occur under constant external load, and results from
the disruption of bonds due to slaking processes. With unconfined clay chunks, like
52

those used in slake durability tests, the disruption of bonds due to the immersion, or
wetting and drying cycles, can act to break apart the specimen. This is the phenomenon
that is typically referred to as "slaking". However, in a confined clay mass, "breaking-
up" of the material is inhibited and the disruption of bonds is instead accompanied by
swelling and softening of the clay mass. Thus, as illustrated by the path B-C in Fig. 3.4,
slaking under constant external load can result in swelling of the mass, accompanied by
an increase in the overall water content.

The literature is essentially void of reported systematic investigations into the changes
in strength and stress-strain that result from slaking. Since the slaking history of a clay
shale deposit can be very complex, determining a relationship between the material
strength and the degree of slaking may be very difficult. However, there are unique
relationships between the strength of clay materials and their water contents. In
particular, development of the critical state theory of soil mechanics has greatly
enhanced our understanding of the complex interaction between the stress state and the
water content of a saturated clay mass. In Chapter V, the author introduces the idea
that the effects of slaking might be able to be incorporated into the critical state
concept. If successful, the critical state model would provide very important constraints
on the slaking phenomena, and would greatly simplify the monitoring and predicting of
the effects of slaking.
53

CHAPTER IV

THEORETICAL ASPECTS OF FISSURE SOFTENING


IN CLAY SHALES

In the natural environment, clay shales are typically jointed. The highest strength
mobilized by clay shales in the field is never greater than the strength of jointed or
slickensided samples. In addition, much evidence has been presented which indicates
that the softening of clay shale deposits is initiated along fissures, thereby further
concentrating the zones of weakness along fissure planes. Therefore, it is generally
important to consider the jointed nature of clay shales, and not threat these materials
strictly as classical soils.

Since the first suggestions by Gregory (1844) and Delabache (1844) that the strength
of the London clay was being reduced by a softening of the clay material adjacent to
fissures, several other researchers have alluded to this possible mechanism to account
for the deterioration of strength in many clay shales. However, the author has found no
in-depth discussions as to the mode by which deterioration progresses in clay shales,
nor have detailed theoretical or experimental investigations been carried out with
regard to the progressive changes in strength and stress-strain behavior that occur in
response to fissure deterioration.

In this chapter, the author proposes a simple model for fissure deterioration, and
investigates the implications of this model. With the aid of reviews presented in
Appendices C and D, possible changes in the strength and stress-strain behavior of a
progressively softening fissured mass will be discussed. Although subsequent chapters
will concentrate primarily on the slaking mechanism in clay shales, the concepts
presented in this chapter should be relevent to any material undergoing fissure
deterioration by any softening mechanism.

Model for Fissure Softening


The model for progressive deterioration of clay shales, as envisioned by the author, is
illustrated by the schematic in Fig. 4.1. In the schematic are four cross sections of a
theoretical clay shale mass, each representing a different stage of deterioration. Stage 1
54

Figure 4.1. Schematic of the proposed model for progressive deterioration of


clay shales.
represents a fissured clay shale in which no alteration has occurred. At stage 2,
alteration has been initiated along the fissures, and progresses in stages 3 and 4, until
the mass is entirely deteriorated. This model is further illustrated by the photograph in
Fig. 4.2, which shows joint controlled deterioration in a portion of the Pierre shale that
is in an intermediate stage of weathering.

At stage 1, the strength and stress-strain response is probably controlled primarily by


the frequency, strength, and orientation of the fissures, and thus behaves similar to a
jointed rock. At stage 4, however, the material is in theory behaving much like a
homogeneous, naturally-remolded clay. During the intermediate stage 2, the clay shale
mass can be considered as a rock-like material with the strength controlled by the
55

Figure 4.2. Photo showing joint deterioration in the Pierre shale of South
Dakota (photo by author).

orientation and shear resistance of soft, filled joints. Similarly, at stage 3, the mass
might be best considered as a soft matrix surrounding large "grains" of stiff, intact
material, much like a "boulder clay". The major difficulty in modeling the behavior of
clay shales is that the failure of a clay shale slope or foundation can occur at any stage
of the deterioration as depicted, and requires only that the shear strength is reduced to
a level such that it can no longer resist the shear stresses acting on the potential slip
surface.

From this discussion of the fissure deterioration model, it is apparent that the analysis
of the strength and stress-strain response of a progressively deteriorating clay shale
should consider four aspects of soil and rock mechanics: (a) the effects of remolding on
the behavior of a homogeneous, but not necessarily isotropic, clay mass, (b) the
behavior of a material consisting of stiff cores of intact clay within a soft mud matrix,
(c) the strength along joints with a filling of soft mud, and (d) the behavior of a fissured
mass.
56

Implications of the Fissure Softening Model


The shear strength of jointed masses. The shear strength of a jointed mass is
dependent not only on the strength along the individual joints, but also on such factors
as the orientation and spacing of the joint, or joint network, relative to the orientation
of the stress field. Methods to analyze the engineering behavior of jointed masses are
numerous and often quite complex, and it is not within the framework of this
dissertation to discuss these models in any detail.

In their very simplest form, all of these models consider joints as planes of weakness
along which shear failure is preferred. However, as discussed above, the ability to
mobilize this low resistance depends on the spatial relationship of the joint system to
the stress field, as well as on the magnitude of stresses. Failure in a jointed mass can
occur entirely along joints, entirely through intact material, or both along joints and
through intact material. The strength mobilized by a jointed mass generally lies between
that of the intact material and that along a single joint.

The shear strength of a single clean, filled, or softened joint. A detailed review on
the mechanics of single joints is presented in Appendix C. The present section will
discuss how the knowledge presented in that review might be incorporated into the
fissure softening model.

The shear behavior of a single clean joint is controlled affected by several factors: (a)
the shear and tensile strengths acting between or within the joint walls, (b) the
roughness of the joint surface, and (c) the orientation of the joint relative to the stress
state. A joint may refer to either a simple, well defined crack, or a complex shear zone
consisting of interwoven cracks or an infilling of softened material or granulated
"gouge".

As illustrated in Fig. 4.3, joint surfaces may be undulated or stepped, as well as planar,
and on a smaller scale, these surfaces may be rough, smooth, or very polished. The
effect of non-planar, or rough, joint surfaces is to create a certain degree of
interlocking between the two joint surfaces, as they try to slip past each other.
Therefore, in order for slip to occur, these rough peaks or "asperites" must either slide
over one another, resulting in dilation normal to the joint, or the asperities must
57

themselves break by shear or tensile failure. The interlocking of asperities can


contribute to the measured cohesion observed in fissured clays.

A filling of softer material within a fissure or joint tends to, of course, separate the
walls of the fissure, and thereby prevents, or minimizes, the interlocking of smaller
asperities. For most filled joints, the failure envelope is located between that for the
fillings and that for a similar clean joint. As the thickness of the filling increases relative
to the amplitude of the asperites, the strength and stiffness of filled joints decrease.

The thickness of filling required to significantly reduce the strength of a joint depends
on the roughness of the joint walls. For joints that are smooth and planar, the strength
of the joint can be rapidly reduced to the strength of the fill material. In addition, if the
strength of the fill material is much lower than the shear strength along the clean joint,
the reduction of strength resulting from the presence of a filling can be rather drastic.
Finally, a filling of swelling clay is particularly troublesome due to the loss of strength
and high pressures associated with swelling.

The softening of wall material along a fissure is essentially equivalent to the progressive
thickening of a fill within the fissure. However, softening along fissures also results in
the weakening or destruction of asperities along the fissure walls. The detrimental
effects of joint softening are therefore three-fold. First, softening creates a low-friction
coating, or weak fill, and therefore reduces the strength and stiffness in accordance
with the above discussion. Second, the process of softening tends to greatly reduce the
roughness, or JRC, of the joint walls; this reduces the amount of interlocking between
joint walls, and also decreases the thickness of filling required to reduce the strength of
the joint to that of the soft fill material. Finally, softening of the joint walls, of course,
reduces the shear resistance, or JCS, of the wall material, and facilitates the breaking of
any remaining asperities.

The transgression of softening in a rough, undulating joint, and a rough, planar joint,
are illustrated by the schematic in Fig. 4.4. Softening acts initially to remove the higher
order, or small-scale, irregularities along the joint walls. However, if the degree of
softening is excessive, or the initial roughness coefficient is small, then the large-scale
irregularities may likewise be removed. In such cases, the strength would be
58

Figure 4.3. Schematic illustrating various forms for joints.


59

Figure 4.4. Schematic illustrating the progression of softening in undulating and planar joints.
60

reduced to that of the fill material. In addition, the progressive reduction in the strength
of the joint wall material, may allow increased breakage through the irregularities
before they have been fully softened, as in case 1 (t=3) of Fig. 4.4. For joints that are
initially planar and relatively smooth, as in case 2, only a very small thickness of
softened fill is required to totally reduce the joint strength to the fully softened value.

Much more research is essential before the effects of progressive softening of joint
walls are fully appreciated. Several factors may influence the nature by which softening
affects the strength of a joint:

(1) Softening typically reduces the roughness of a joint wall; however, if the material
along the joint wall is variable in its susceptibility to weathering, then softening
may be confined to local areas, leaving other areas to remain as intact
irregularities. In such cases, the joint may actually become rougher, resulting in
an increase in the degree of interlocking once shear displacement has been
initiated. This condition might be further enhanced if the rate of softening is
intense.
(2) If the penetration of softening agents, such as water, or dissolved ions, into the
joint walls is slow, then the boundary between the intact wall material and the
softened “fill” is likely to be well defined. However, if the rate of penetration is
rapid, or if the agents are able to act over long periods of time, then there may
not be a distinct boundary between softened and intact material.

(3) joint walls is slow, then the boundary between the intact wall material and the
softened "fill" is likely to be well defined. However, if the rate of penetration is
rapid, or if the agents are able to act over long periods of time, then there may
not be a distinct boundary between softened and intact material.

(4) For the case of joint softening in response to drying-induced slaking, the nature
of the boundary between intact and softened material will be determined by the
extent to which drying progresses into the joint walls, and by the number of
wetting and drying cycles. If drying is limited to a thin layer of material adjacent
to the joint, and particularly if numerous cycles of wetting and drying are
allowed to occur, then a well defined boundary will be formed between intact
material and a much softer fill material. Therefore, the thickness of the softened
61

zone, as well as the nature of the gradation from intact to softened material,
depends on the rates of drying and wetting relative to the cyclic periods over
which wetting and drying occur.

(5) There is considerable uncertainty as to the effect of softening and swelling on


joint permeability. Swelling in a joint may decrease the joint permeability by
filling voids between the walls. However, swelling might also increase the
permeability of the jointed mass by increasing the width of conduits through
which flow occurs. In the former case, swelling produces a negative feedback
mechanism by which further softening is inhibited, whereas the latter case is
accompanied by a positive feedback mechanism which enhances further
softening. The relative influence of these two factors may depend on whether
swelling is prohibited by external forces, as well as on the extent to which the
joint Figure
width is4.5. Schematic
increased of deterioration.
by wall a highly altered clay shale consisting
of stiff lumps of intact clay within a matrix of highly weathered
In conclusion, for joints that are initiallyclay.
planar and smooth, the amount of fissure
softening required to significantly reduce the shear resistance along the fissure may be
quite small. Slaking and swelling along fissure walls reduces the degree of wall
roughness, enhances shear through asperities, and creates a fill that reduces the amount
of interlocking between joint walls. All of these affects act to significantly reduce the
strength along fissures.

The strength of a clay mass consisting of a soft matrix surrounding stiff, intact
cores. If extensive alteration of a clay shale is allowed to continue, the material at
failure may consist of an assemblage of stiff, intact clay cores surrounded by a soft,
remolded clay or mud matrix. An example of an extreme case is presented in Fig. 4.5.
Such materials can typically fail at very low slope angles in the form of mudflows

Initially, the degree of softening within the clay mass may be such that there still exist
some interlocking between some stiff clay cores. In this case, some significant cohesion
might exist as a result of this interlocking. With further degradation, the stiff
cores are no longer close enough to interlock, but shear through the clay mass may
require the rotation or translation of some of these cores. The strength of such a clay
mass is very near, if not equal to, the strength of the softened clay. Ultimately, with

Figure 4.6. Schematic illustrating possible changes in the failure surface


resulting from progressive softening in a clay shale.
62

continued softening, stiff lumps of clay may continue to exist within the soft clay
matrix, but their presence would not significantly affect the shear strength of the mass.

Therefore it is important to note that it is not necessary to soften the entire mass of a
clay shale before it will exhibit the "fully softened", or remolded, strength. It is only
necessary that the concentration of intact clay cores is sufficiently low as to inhibit
significant interaction between intact cores. The interlocking component of strength,
which is certainly the major compenent of strength in fissured clay shales, could in
theory be eliminated if only 5% of the clay shale material is softened. In reality, the
percentage of softening required to eliminate interlocking is probably somewhat higher,
but may still be surprisingly low.

The Effects of Progressive Fissure Deterioration on the Engineering


Behavior of Clay Shales
The previous Fig. 4.1 schematically illustrated four stages of alteration in clay shales.
This figure has been modified in Fig. 4.6, in order to illustrate how fissure deterioration
might alter the failure plane, and thus the engineering behavior, within a clay shale.

The unaltered clay shale, should be considered as a highly jointed mass, and not as a
homogeneous soil in the classical sense. If the strength of the fissures is much less than
that of the intact material, and if the applied stresses are low, then the failure plane may
be controlled entirely by the location of the fissures. This condition is indicated by the
solid line in stage 1 of Fig. 4.6. Like hard, jointed rocks, the strength and stress-strain
response is therefore controlled by the shear resistance along the individual fissures,
and on the geometric relationship between these fissures and the acting stress field. In
contrast, if the resistance of the intact material is not much higher than that of the
fissures, and particularly if high stresses are applied, then failure may occur along the
plane of maximum shear stress, indicated by the dashed line, and would therefore pass
entirely through intact material. Field evidence indicates that this is rarely the case in
unaltered clay shales.

Shear along the failure plane of the unaltered clay shale mass in stage 1, could be
considered analogous to shear along a single "joint" given by the dark jagged line. This
"joint" is characterized as having a high first-order roughness with saw-tooth asperities,
63

while the higher-order roughness is controlled by the roughness along the true fissures.
The stress-strain response at low stresses would be similar to that illustrated in
Fig. 4.7a, which indicates low compressibility, high dilation perpendicular to the failure
plane, high peak strength, and a significant post-peak drop in strength. As in Fig. 4.8a,
the typical strength envelope is given by the Fairhurst equation for rock joints. The
strength parameters are characterized at low confining stresses by a high apparent
cohesion, resulting from interlocking of the large "saw-teeth" and by a high friction
angle which results from the added "i" component discussed in Appendix C. At much
higher stresses, failure occurs through asperities, so that the strength parameters are
essentially equal to the cohesion and friction angle values for intact material.

For the partially softened clay shale represented by stage 2 in Fig. 4.6, deterioration has
been confined to the walls adjacent to the fissures. The "analogous joint" along which
failure is preferred is again given by the thick, dark line, and might be characterized as
an undulating filled joint. As illustrated by Fig. 4.7b, the typical stress-strain behavior at
low stresses might initially exhibit low stiffness and high compressibility, as the layer of
soft material within the joints is compressed. However, with further shear displacement,
the intact walls of the fissures again make contact, resulting in increased stiffness. The
peak resistance would be lower than that for the unaltered clay shale due to a decrease
in the first-order roughness of the failure plane, as well as a significant decrease in the
strength along the true fissures. As discussed previously, the decrease in fissure
strength, in response to softening, is a result of smoothing of the joint walls, separation
of the joint walls by softened fill material, and softening of the asperities along the
fissure. Therefore, the strength envelope, represented in Fig. 4.8b, indicates a lower
cohesion and a lower friction angle at low confining stresses, as a result of the
smoothing of both first and second-order roughness. In addition, the apparent cohesion
at higher stresses may also be reduced if softening has enhanced the breaking of the
larger asperities within the analogous joint.
64

Figure 4.7. Schematic illustrating changes in the stress-strain behavior of a


clay shale undergoing progressive softening as illustrated in Fig. 4.6.
65

Figure 4.8. Schematic illustrating changes in the strength envelope resulting


from progressive softening of a clay shale as illustrated in Fig. 4.6.
66

Once fissure softening has progressed to the third stage the plane of failure may be
almost entirely within softened material, as shown in Fig. 4.6. Although there may be
some interference from some intact cores within the failure plane, shear displacement
even at high stresses, will probably occur by rotation of these cores, rather than failing
through them. As can be deduced by comparing the curves in Figs. 4.7c and d with
those of Figs. 4.8c and d, the engineering behavior at this stage is very similar to that
for the fully deteriorated clay shale. In such cases, the strength and stress-strain
response is controlled entirely by the softened material.

In summary, analysis of the strength of a clay shale undergoing progressive softening


along fissures is complex, and generally requires some understanding of principles from
both soil and rock mechanics. However, as will be discussed in the following chapter,
the concept of critical state soil mechanics may provide an important tool with which
the analysis of softening clay shales might be greatly simplified.
68

CHAPTER V

THE ROLE OF SLAKING WITHIN THE


CRITICAL STATE MODEL

Much research in the past three decades has been oriented toward defining the limiting
stress states within which a soil can exist, independent of stress path history. Sufficient
evidence exists to support the concept that the critical state of a homogeneous, non-
fissured, saturated soil can be defined by two stress invariants and the water content (or
specific volume). This concept, generally referred to as Critical State Soil Mechanics,
provides a concise framework within which the seemingly complex behavior of soil can
be understood. In this chapter, the author proposes that the effects of slaking might
similarly be considered within the framework of the critical state model. If so, this
would greatly simplify the otherwise complex analysis of a clay or clay shale which is
undergoing softening in response to slaking.

Review of the Principles of Critical State Soil Mechanics


In stress analysis, it is convenient to use stress parameters that are independent of the
choice of reference axes. One set of these, so-called, stress invariants are the octahedral
normal effective stress, σ'oct, and the octahedral shear effective stress, τ'oct, which can
be defined in terms of effective principal stresses as

σ'oct = (σ'1+σ'2+σ'3)/3, (5.1)

τ'oct = [(σ'1-σ'2)2+(σ'2-σ'3)2+(σ'3-σ'1)2]1/2. (5.2)


3
For the special case of standard triaxial shear states, where σ'2 = σ'3,

σ'oct = (σ'1+2σ'3)/3, (5.3)

τ'oct = √2(σ'1-σ'3)/3. (5.4)


To avoid the recurring √2/3 term, new invariants, P' and Q', are defined for the triaxial
case, as
69

P' = (σ'1 + 2σ'3)/3 = σ'oct, (5.5)

Q' = (σ'1 -σ'3) = 3τ'oct/ √2 . (5.6)

The stress invariants are defined such that, for an isotropic, homogeneous material, P'
defines the hydrostatic component responsible for isotropic volumetric changes, and Q'
incorporates all of the shear stresses which can cause distortion.

The basic principles of critical state soil mechanics have been discussed in some detail
in the text by Atkinson and Bransby (1978). Much of the review below is a summary of
that text and will therefore not reference the numerous, individual contributions which
have led to the development of the critical state concept.

For normally-consolidated clays which have been isotropically compressed and then
loaded to failure in drained and undrained triaxial tests, data indicate that the
relationship between P', Q', and the specific volume, v, at failure can be uniquely
defined by a single line, as illustrated in Fig. 5.1. This line is referred to as the "critical
state line". The stress paths followed during drained and undrained tests are
respectively shown in Q':P' and v:P' spaces in Figs. 5.2 and 5.3. An important property
of the critical state line is that it uniquely defines the residual strength, regardless of the
stress path followed.

The projection of the critical state line in Q':P' space is defined by a line through the
origin, and with a slope designated as M. In v:P' space the projected critical state line is
curved and similar in shape to the normal consolidation line. In fact, when plotted in
v:ln P' space the critical state line can be approximated by a straight line with a slope
nearly equal to the normal consolidation line, as in Fig. 5.4.

Although somewhat more complicated than normally consolidated clays, the behavior
of overconsolidated clays can still be accounted for within the critical state concept.
The typical stress path for an heavily overconsolidated clay in Q':P' space during a
drained triaxial test is illustrated in Fig. 5.5. Since σ3' remains constant for a typically
drained triaxial test,
∆Q' = ∆σ 1' , (5.8)
∆P' = ∆σ1'/3 , (5.9)
70

Figure 5.1. Ultimate failure points for drained and undrained tests on normally
consolidated specimens of Weald clay (Atkinson and Bransby, 1978; after Parry,
1960).
71

Figure 5.2. Stress paths in (a) Q':P' and (b) v:P' space for undrained tests on
normally consolidated samples (Atkinson and Bransby, 1978).
72

Figure 5.3. Stress paths in (a) Q':P' and (b) v:P' space for drained triaxial
tests normally consolidated samples (Atkinson and Bransby, 1978).
73

Figure 5.4. The critical state line in v:ln P' space (Atkinson and Bransby,
1978; after Parry, 1960).

Figure 5.5. Stress path followed in a drained triaxial compression test on an


overconsolidated clay (Atkinson and Bransby, 1978).
74

and
∆Q'/ ∆P' = 3 . (5.10)

The stress path for heavily overconsolidated clays is characterized by an increase of P'
and Q' until peak strength is mobilized at stresses beyond those at the critical state line,
followed by a reduction of P' and Q' toward the critical state line. Thus, for
overconsolidated clays, the critical state line represents the "ultimate" or "residual"
strength.

The peak strength of overconsolidated clays is dependent on the specific volume, as


well as P' and Q'. However, it has been found that the differences in specific volume can
be accounted for if P' and Q' are normalized using the value of pe, which is the stress on
the normal consolidation line at that particular specific volume. Data shown in Fig. 5.6
indicates that the peak failure surface for heavily overconsolidated clays can be defined
by a straight line on a plot of Q'/Pe' versus P'/Pe'. After reaching this surface, called the
Hvorslev surface, the stresses tend toward the critical state line, as shown in Fig. 5.7
for undrained triaxial tests. In contrast, lightly overconsolidated clays exhibit an
increase in the stresses until they intersect the Roscoe surface, which is defined by the
stress paths of normally consolidated clays. These surfaces are better
visualized in a three-dimensional plot of Q':P':v space, as in Fig. 5.8.

Figure 5.6. Normalized failure surface for drained and undrained tests on
overconsolidated samples of Weald clay (Atkinson and Bransby, 1978; after
Parry, 1960).
75

Figure 5.7. Schematic of expected undrained tests paths for samples at


different overconsolidation ratios (Atkinson and Bransby, 1978).

The Roscoe, Hvorslev, and tensile failure surface define the boundaries which confine
the possible stress states for a given soil. Similarly, the critical state line defines the
state of stress at which further changes of P', Q', or v cannot occur. In addition to being
limited by these surfaces, the possible stress path is restricted by the testing procedure.
For example, in Fig. 5.9, is shown the plane along which the stresses can move during a
drained test.

In summary, the behavior of normally and overconsolidated clays can be illustrated in


the Q':P' and v:P' plots of Fig. 5.10. The plot shows the stress paths and failure loci for
samples that have all been consolidated to the same preconsolidation stress, Pc, and
then allowed to swell under different confining pressures. Samples which lie on or to
the right of the critical state line in v:P' space, will exhibit a decrease in specific volume
and will fail at the critical state line. In contrast, samples which lie to the left of the
critical state line will exhibit an initial decrease in specific volume, followed by a large
increase after failure, and will tend to fail at peak strengths which are higher than those
given by the critical state line. After peak failure, however, the stresses will drop until
they are compatible with the critical state line.
76

Figure 5.8. The complete state boundary surface in three dimensions ; Q':P':v
space (Atkinson and Bransby, 1978).

of the critical state line will exhibit an initial decrease in specific volume, followed by a
large increase after failure, and will tend to fail at peak strengths which are higher than
those given by the critical state line. After peak failure, however, the stresses will drop
until they are compatible with the critical state line.

Atkinson and Bransby (1978) have suggested based on theoretical arguments, that the
locus of peak failure for heavily overconsolidated clays might lie along the swelling line.
However, data presented by Henkel (1959) and shown in Fig. 5.11, indicates that the
peak failure surface for artificially overconsolidated Weald clay is does not line along
the swelling line.
77

Figure 5.9. The drained plane in Q':P':v space (Atkinson and Bransby, 1978).

The Role of "Aging" Within the Critical State Concept


The process of "aging", which occurs in response to sustained consolidation loads over
long periods of time, has been discussed in Appendix B which discusses the formation
of clay shales. For the sake of convenience, Fig. B.7 is repeated here as Fig. 5.12.
There is some question as to effects of aging with regard to the critical state concept. In
particular, the question arises as to whether the process of aging is accompanied by a
change in the intrinsic material properties and therefore a displacement of the material
yield surfaces, or whether the effect of aging is simply to alter the "current" physical
state without affecting the validity of the critical state concept or the position of the
existing yield surfaces.

The illustration in Fig. 5.12 shows that if a normally consolidated clay is allowed to age,
and is then consolidated in the laboratory, the e-log P' curve will show an abrupt
78

Figure 5.10. Failure states of drained tests on samples at different


overconsolidation ratios (Atkinson and Bransby, 1978).
change in compressibility at a pressure, Pc', which is greater than the actual load,
Po'.This gives the false impression that the clay has been subjected to a past load
greater than the load that existed. The clay is essentially behaving as if it had been
previously consolidated to Pc' and then allowed to swell under a reduced load,
suggesting that the normal consolidation line is still uniquely defined and not dependent
on the path of loading.
79

Figure 5.11. Predicted failure points for overconsolidated and normally


consolidated clays (Henkel, 1959).

Bjerrum (1973) has shown, as in Fig. 5.13, that the ratio of the undrained shear
strength obtained from vane tests over the true consolidation pressure, Su/Po',varies
with the plasticity, Ip, and is greater for aged clays. He has further shown that the ratio
of Pc'/Po',increases with plasticity, but is constant with depth of the deposit. Tavenas
and Leroueil (1977) modified the plots of Bjerrum by combining the Su/Po' and Pc/Po'
curves in order to normalize the shear strength with regard to Pc', and determined that
the ratio of Su/Pc for the aged clay is nearly identical to Su/Po' for the young clay (see
Fig. 5.14). This behavior is identical to that for overconsolidated clays considered
80

Figure 5.12. Geological history and compressibility of normally consolidated


clays, showing effects of "aging" (Fleming et al, 1970; after Bjerrum, 1972).

within the concept of critical state soil mechanics, and suggests that the material yield
surfaces defined by the critical state model remain undisturbed during the aging
process. Still, more extensive research is required to verify these findings.

Slaking within the Critical State Concept


In discussions regarding the critical state concept, it is common to consider the act of
swelling as it occurs in response to either a decrease in P' while Q' remains constant, or
a decrease in both P' and Q'. However, the potentially important role of slaking in
increasing the water content of a clay, independent of stress changes, has not been fully
appreciated. The present author proposes that swelling associated with the slaking
processes should be considered within the critical state model. The strength exhibited
by a clay shale which has undergone slaking, might be dependent only on the final water
content and not on the slaking history.
81

Figure 5.13. Changes in the undrained shear strength ratio and the consolidation
pressure ratio versus the plasticity index for "aged" and "young" normally
consolidated clays (Bjerrum, 1973).

Figure 5.14. Normalization of the undrained shear strength for "young" and "aged"
clays (after Bjerrum. 1973).
82

As illustrated in Fig. 5.15, an overconsolidated clay or clay shale which has undergone
consolidation, and possibly aging, followed by swelling due to rebound after unloading,
may undergo further swelling in response to slaking near the surface. Whereas the aging
process can cause a normally consolidated clay to behave as if it were overconsolidated,
the process of slaking might, likewise, cause a clay to reverse these effects, by reducing
the degree of overconsolidation. Therefore, in much the same way that a sustained load
can act as an "aging" mechanism, slaking serves as a mechanism by which a clay might
regain some of its youth, or be "rejuvenated".

This rejuvenation process can be demonstrated by the plots in Figs. 5.16. As discussed
in the previous section on aging, a clay can become overconsolidated along at least two
paths: (a) by a reduction of pressure after consolidation, resulting in swell due to
rebound in accordance to the "swelling lines" in v:P' space, or (b) by the "aging"
process, whereby the specific volume is reduced under sustained constant loads.
Regardless of the path, we'll assume that a clay has reached the stage of
overconsolidation represented by point 4 in the v:P' plot of Fig. 5.16a. If sheared to

Figure 5.15. Possible compression and swelling paths for a clay undergoing
aging and slaking.
83

failure in a drained triaxial test, this clay would exhibit a relatively high peak strength
greater than the critical state strength, and a volume change during shear characterized
by an initial decrease followed by a large increase. This behavior is illustrated by the
curve 4 in the plot of shear stress versus specific volume in Fig. 5.16b.

Suppose however, that instead of being tested, the sample were allowed to slake and
swell under a constant P', until the state represented by point 3 on the v:P' plot were
obtained. The sample would then exhibit a reduced strength which is closer to the
critical state strength, and the amount of dilation following peak strength would be
much less. The clay is therefore behaving as though it were less overconsolidated than
it was at point 4.

The major question to pose is whether the clay which has arrived at the state
represented by point 3 by the process of slaking, will behave exactly the same as the
clay that arrived at point 3 by aging, or simple rebound following unloading. If not,
then it must be assumed that the boundaries which define the critical state model
change in response to the process of slaking. However, if the behavior of these clays is
the same, then the critical state model can truly be considered as stress-path
independent with regard to slaking.

As illustrated in Fig. 5.16, it seems possible that an overconsolidated clay can be


naturally remolded by the slaking process to such an extent that the clay is completely
rejuvenated and approaches a normally consolidated state. This may account for
observations that the strength of many clay shales tends toward the "fully softened"
strength, which is essentially the strength of a remolded, normally-consolidated clay
(Skempton, 1970). Two factors may prevent a clay shale from fully softening to the
strength of a normally consolidated clay in the field. Due to the increase in the
horizontal stress component that accompanies swelling in the field, the clay may fail in
passive failure before obtaining the normally consolidated state by means of slaking.

Furthermore, an important question is whether there is a limit to the extent to which a


clay shale can slake and swell. It may be possible that the amount of available energy
released during slaking is limited to only that amount of strain energy gained during
aging. If that proves to be the case, then a clay can only regain the "youth" lost during
the "aging" process, and not lost as a result of normal consolidation.
84

Figure 5.16. Possible loading/unloading paths (a) before and (b) during
drained tests (modified from Atkinson and Bransby, 1978).

It has been previously stated that the critical state line represents a stress condition at
which no further changes in the stress state can occur. It is therefore interesting to
consider the potential effects of slaking on a clay which has reached the critical state.
One of three responses is possible: (a) further slaking cannot occur and the material
yield surfaces are not altered, (b) slaking does not alter the yield surfaces, but instead
alters the stress state of the soil away from the critical state line, but within the confines
85

of the yield surfaces, resulting in a hardening of the clay, or (c) slaking alters the
position of the yield surface defined by the previous critical state model.

Possible Complications Resulting From the Fissured Nature of Clay


Shales
The presence of fissures in clay shales may complicate the incorporation of the slaking
process into the critical state model. These complications result from three aspects of
fissuration. First, unless a clay shale deposit is intensely intersected by randomly
oriented fissures, the fissuration results in material anisotropy. The anisotropy resulting
from fissures may be superimposed on any anisotropic components that might exist
within the intact clay shale itself. Still, most geological materials in nature are
anisotropic. Therefore, except for possible differences in magnitude, the complications
resulting from anisotropy in clay shales are no different from those resulting from
anisotropy in all natural argillaceous materials.

The other two possible complications result from the probability that the softening
within clay shales is not homogeneous, but is at least initially concentrated along fissure
walls. In the discussion in the previous chapter, it was suggested that the softening
along fissures could actually alter the shape and roughness of the potential failure plane,
as well as reduce the shear strength between fissure walls. Therefore, in a fissured mass
undergoing slaking, the peak strength of the mass might be reduced to a greater extent
than that predicted by the critical state model.

In addition, when applying the critical state model, the water content is generally
assumed to be homogeneous throughout the sample. In a progressively softening clay
shale, the water content varies from its highest values near the fissures to lower values
within the interior cores of intact clay. Therefore, the overall water content measured
for the clay shale sample may be different from the water content within the failure
zone, particularly if failure occurs primarily along fissures. However, as will be
discussed in the next section, the change of water content is more important than the
absolute value. During the initial stages of softening of a clay shale, the change in total
water content is probably equal to that for the failure zone. In later stages of softening,
however, the failure plane may already be located entirely within the softened portion of
the clay shale. Any additional increases in the total water content would probably occur
86

within scattered intact cores and therefore might not significantly alter the strength of
the mass.

Testing the Critical State Model With Regard to Slaking


As discussed previously, the critical state concept states that the strength of any ideal,
isotropic, overconsolidated clay is dependent only on two stress invariants, P' and Q',
and the specific volume, v. As has been illustrated in Fig. 5.9, the stress path of a lightly
overconsolidated clay is constrained by the Roscoe surface, while the stress path of a
heavily overconsolidated clay is similarly constrained by the Hvorslev surface. A major
test of the applicability of the critical state model to the process of slaking is whether
the Hvorslev surface can be defined for a clay shale which is undergoing softening by
slaking.

The parameter, Pe', the equivalent pressure at any specific volume, has been discussed
previously, and can be obtained from the equation for the normal consolidation line, as

Pe' = exp[(N-v)/λ], (5.11)


where v is the current specific volume, the slope of the normal consolidation line in
v:ln P space, and N is typically defined as the specific volume of the soil at P' equal to
1 kN/m2 on the normal consolidation line. By normalizing the stress invariants with Pe,
as in Fig. 5.17, the complete Hvorslev surface can be defined by a line whose equation
is

Q'/Pe' = g + h(P'/Pe'). (5.12)


In addition, the equation for the Hvorslev surface is given by

Q' = (M-h)exp[(Γ-v)/λ] + hP', (5.13)


where M is the slope of the critical state line in P':Q' space, and Γ is arbitrarily defined
as the value of v corresponding to P' equal to 1 kN/m2 on the critical state line.

The value for λ is generally obtained directly from consolidation data. However, as was
illustrated in Fig. 5.4, the slope of the critical state line in v:ln P space is
87

Figure 5.17. The Hvorslev surface in normalized P':Q' space (Atkinson and
Bransby, 1978).
approximately equal to the slope of the normal consolidation line, λ. Therefore, the
value for λ can be obtained from either consolidation data or data defining the critical
state line. In the analysis of experimental data in Chapter VII, the value of λ will be
obtained using the critical state line.

It should also be noted that the use of the normal consolidation line for defining the
equivalent pressure, Pe' is convenient, but arbitrary. The value of Pe' could just as
successfully be defined using the critical state line, or in fact any equivalent line with a
slope of λ. By rearranging equation 5.13, we obtain

Q' = (M-h)[ exp(Γ/λ) / exp(v/λ) ] + hP' (5.14)

and

Q'/exp(-v/λ) = B + hP'/exp(-v/λ) (5.15)

where B is equal to the constant


88

B = (M-h) exp(Γ/λ). (5.16)

Thus, to prove the existence of the Hvorslev surface, and to define the slope, h,
requires only one material constant, λ. Furthermore, the value of h depends on the
changes in the specific volume, v, relative to λ rather than on its absolute values. Thus,
even though the total water content (or specific volume) may be different than that
within the failure plane, as discussed previously, the slope of the Hvorslev surface can
still be obtained as long as the changes in total and local water contents are the same.

Summary on the Role of Slaking Within the Critical State Model


The concept of critical state soil mechanics has proved to be a very concise and useful
method for analyzing the strength of a homogeneous, non-fissured, saturated clay. The
author proposes that it may also be equally useful for assessing the changes in strength
resulting from slaking in clay shale.

The importance of considering the process of slaking within the concept of critical state
soil mechanics cannot be overemphasized. The slaking history of a clay could be
extreme complex, and nearly impossible to determine or predict. Furthermore, to study
the effects of slaking under laboratory conditions, the researcher must investigate
several variables, including, but not limited to, the number of wetting and drying cycles,
the length of each cycle, and the extent of drying during each cycles. Clearly, in
accounting for the effects of slaking within the field or laboratory, is would be
extremely useful, if not vital, for the results to be dependent only on the final specific
volume, or water content, and not on the path along which that specific volume was
obtained.

The major question regarding the incorporation of both aging and slaking into the
critical state concept is whether these processes simply provide alternate paths along
which the specific volume of a clay can be altered within the confines of the defined
critical state boundaries, or whether these boundaries are themselves altered by the
processes of aging and slaking. The critical state model has been shown to be path
independent with regard to P' and Q'. It is important to likewise determine the extent of
path independence with regard to the specific volume.
89

The author has carried out triaxial testing on clay shale samples which have undergone
various degrees of softening by slaking. These will be compared to similar test on
unaltered clay shale in order to assess the feasibility of incorporating the slaking process
into the critical state soil mechanics concept. The test procedures are described in the
following chapter, while analysis of the data is presented in Chapter VII. In addition to
investigating other aspects of the slaking process, the analysis will in particular
concentrate on the question of whether the peak strengths of the intact and softened
clay shale define a common Hvorslev surface.
90

CHAPTER VI

EXPERIMENTAL METHODOLOGY

In Chapters II and III, much evidence was presented which emphasized the important
role of fissure softening in affecting the engineering behavior of clay shales. The
theoretical effects of fissure softening on the strength and stress-strain response of clay
shales were discussed in detail in the Chapter IV. It has been shown that a primary
mechanism by which softening may occur involves the process of slaking.

However, most of the present studies have been limited to the phenomena of slaking at
zero confining pressures. Very few, if any, systematic studies have measured slake
durability in the presence of confining pressures, and this author knows of no
published data which directly relates the process of slaking to changes in the
engineering behavior of clays or clay shales.

The previous chapter introduced the possibility of accounting for the effects of slaking
within the concept of critical state mechanics, and emphasized the usefulness of
studying slaking within this framework. If the effects of slaking can be considered
within the framework of the critical state concept, the analysis of progressively
softening clay shale would be greatly simplified. Outside of this dissertation, the
present author has found no experimental nor theoretical investigations concerning the
effects of slaking within the critical state model.

For these reasons, an experimental program was developed by the author, with the
primary objectives of (1) providing initial investigation into the effects of drying-
induced slaking on the strength and stress-strain response of a highly fissured clay
shale and (2) assessing the feasibility of incorporating the slaking process into the
critical state concept. These experiments primarily involved triaxial compression tests
on samples of Pierre shale, which had undergone various degrees of slaking while
under a low confining pressures. Particular care was taken to record changes in the
water content occurring within the samples before, during, and after the tests. The
apparatus, and the sampling and testing procedures will be discussed in detail below.
91

Testing Apparatus
Hoek cell. The triaxial testing was initially intended to be performed within Hoek cells
commonly used for the testing of rock. This apparatus had been chosen because of the
rather high strength of undisturbed Pierre shale, and because the Hoek cells at the
University of Colorado are capable of accurately measuring the radial displacement in
three directions. The Hoek cell consists of a stiff, metal outer cylinder and a flexible,
semi-permanent, cylindrical membrane which surrounds the rock specimen. The
confining pressure, which acts in the radial direction only, is applied by pressurizing oil
within the cavity between the membrane and the outer frame. Six cantilever strain
gages rest against the membrane surrounding the sample and measure the radial
displacement of the specimen. The radial strain gages are connected in a wheatstone
bridge configuration, such that the bridge resistance can be measured and analyzed by
a computerized data aquisition system outside of the cell.

Unfortunately, after an extensive period of rebuilding and calibration of the Hoek cell,
the apparatus was found to be inadequate for the test program. The reasons for this
were as follows:
(1) Although the thinnest membranes possible were used, these still proved to be
too stiff for testing at the rather low confining pressures used. This at times
resulted in a significant amount of undesired membrane support on the sample,
and at other times, resulted in gaps between the membrane and sample.
(2) Samples often "jammed" after fracturing or swelling, making it difficult to
remove the specimen after the test. Often the membrane was distorted or
destroyed during this process, requiring messy, time-consuming repairs.
(3) Samples were always destroyed during removal, thereby prohibiting any post-
test examination of the sample. In addition, the metal cylinder prohibited any
visual examination once the sample was placed into the cell.
(4) Because of the membrane problems, and because of physical and electrical
problems with the strain gages, measurement of the radial displacements was
often unreliable.
The Hoek cell was therefore abandoned for the conventional triaxial cell commonly in
use in soil mechanics.
92

Conventional triaxial cell. The conventional triaxial test cell was chosen instead of
other test cells for several reasons:
a) the triaxial tests provides complete information regarding the stress state acting on
the sample and is flexible with regard to stress paths;
b) unlike the direct shear test, the triaxial test accounts for the interaction between
fissure systems rather than strictly measuring the strength along a single joint;
direct shear tests would therefore not be able to account for the significant
reduction in interlocking which results from the rounding of the sharp corners at
the intersection of two fissures;
c) similarly, in the triaxial test the plane of failure is not predetermined, as is done
with the direct shear test; as has been shown, alteration of the failure plane during
fissure softening can act to greatly reduce the strength of the jointed mass;
d) the triaxial test results are not significantly altered by boundary effects, as are those
of the direct shear test;
e) the triaxial test allows complete control over changes in the water content of the
sample;
f) visual contact can be maintained with the triaxial cell, allowing assessment of the
various modes of failure;
g) the classical research and discussions regarding critical state soil mechanics are
closely associated with the conventional triaxial test.

A schematic of the triaxial cell is shown in Fig. 6.1. The triaxial cell consists of a
plexiglass outer cylinder and close-fitting, metal top and bottom. These are held
tightly in place by vertical bolts which resist the hydrostatic pressures created by
pressurized water inside the cell. Within the triaxial cell, the sample is supported by
upper and lower stainless steel pistons, which are also used to apply axial loads on the
sample.
93

Figure 6.1. Schematic of the triaxial test cell and measuring apparatus.
94

The "blow-up" of the sample in Fig. 6.2, illustrates that the clay shale sample is
enclosed in two very thin flexible membranes. Between the two membranes is a layer
of silicone grease to prevent permeation of water through the membranes during long
tests. Sample drainage and wetting takes place through the porous, ceramic stones at
the top and bottom of the sample. In later tests, water flow within the sample was
enhanced by four 3/4" strips of filter paper (VWR grade 615) running vertically
between the sample and membrane. The plexiglass loading caps at the top and bottom
of the sample, hold the porous stones and allow drainage to tubes at the side of the
heads. The greased membranes are held tightly to the sides of the plexiglass loading
heads by rubber sealing rings, preventing flow of water out of the membranes. The
steel loading heads, which fit into the plexiglass heads, are designed to evenly spread
the vertical load applied by the upper and lower stainless steel pistons. A steel ball
between the upper piston and metal head minimizes eccentric loads on the sample.

As shown in Fig. 6.1, the vertical load is measured by a calibrated loading ring on the
upper piston, while the vertical displacement is recorded from a dial gauge between
the upper piston and cell body. A constant all-around pressure is applied to the sample
by means of pressurized water within the cell. This pressure is supplied by an air pump
and is held constant by a pressure regulator.

Any changes in the volume of water in the cell can be measured from the pipette,
which also acts as an air-water interface. At constant cell pressure, this change in cell
pressure results primarily from changes in sample volume, and from changes in the
displacement of water by the piston as it moves in or out of the cell. A constant "back
pressure" is similarly applied to the water within the sample, such that the volume of
water flowing in or out of the sample can be recorded from a second pipette. This dual
pipette system provided a means of "cross checking" the changes in sample volume,
and as will be discussed later, proved very useful for recognizing problems in the
testing procedure. The glass, graduated pipettes were enclosed on both ends using
coupled brass pipe joints, with a rubber "O" ring between. Tightening of the brass
joints onto the "O"
95

Figure 6.2. Schematic of the test sample assembly.


96

The process of slaking was carried out under conditions of constant hydrostatic stress.
During these periods, the upper piston was loaded with weights in order to
compensate the upward force generated by the hydrostatic pressure of the cell water.
Therefore, the sample was able to deform both radially and vertically in response to
wetting, while the stresses remained constant. However, after slaking, the clay shale
was then loaded vertically to failure using a displacement- controlled Wykeham-
Farrance loading frame. Using displacement-controlled rather than stress-controlled
loading allowed the assessment of post peak stress- strain behavior.

With this load frame, various sized gears can be set to provide for different rates of
loading. For the first series of tests in the triaxial cell, the loading gears were set to
provide a loading rate of 0.0018"/min. For later tests, the gears were adjusted so that
the loading rate would initially be 0.0018"/min and then adjusted to 0.0008"/min prior
to sample failure. Primarily because of the flexibility of the load ring, the loading
system was not very stiff, resulting in some strain energy being stored by the loading
system during loading.

One effect of this lack of stiffness was that the actual initial rate of loading experienced
by the sample was greatly reduced to approximately half of that predicted by the gear
ratios. The actual rate of loading gradually increased at higher stresses, but never quite
reached the rates given above. Although this slower rate of loading was not
particularly undesirable, the presence of strain energy stored by the machine did create
the undesired effect of rapidly loading the sample after peak strength was reached.
This primarily occurred with unaltered samples which were able to resist high stresses
before failing, and was somewhat minimized by turning the loading machine off until
equilibrium was again maintained.

For the present tests, the post-peak strength was more important than the stress-strain
curve after failure. Therefore, this lack of machine stiffness did not significantly affect
the desired results. However, for those test programs where the post-peak stress-
strain behavior is important, it is strongly beneficial to employ either an
electronic pressure transducer in place of load rings, or a variable stiffness loading
system, such as that presented by Sture (1976) and Sture and Ko (1978).
97

Sample Preparation
The Pierre shale is well-suited for the desired test program for several reasons: (a) in
its undisturbed state, the material has a high peak strength, and exhibits a high ratio of
peak to residual strengths, (b) the difference in strength between the unaltered and
softened states is very drastic, (c) the material is highly susceptible to slaking, rapidly
slaking from a rock consistency to mud after only one cycle of drying and wetting, (d)
the shale exhibits high swelling pressures in response to rewetting after drying, and (e)
the shale readily forms horizontal and vertical fissures during drying.

Block sample retrieval. Several blocks of Pierre shale were dislodged by backhoe
from an open pit quarry of the Ideal Cement Company located about 15 miles south of
Boulder, Colorado along state highway 93, in sec 5, T 1S, R 70W. The composition of
the Pierre shale was determined by Braddock and Machette (1976) and is given in
Table 6.1. The excavated blocks ranged in size from one to a few feet per side. In
order to prevent deterioration due to drying, these blocks were immediately placed in
several ten-gallon drums filled with water. The drums were then placed in a wet room
at the University of Colorado where they remained for some ten years before being
cored for the present test program. The blocks of Pierre shale do not appear to have
suffered any significant alteration as a result of stress unloading or soaking in water.
However, if any of the blocks or fragments of the Pierre shale is allowed to even
partially dry, and is then rewetted, the shale will rapidly break down into small chunks
and then to soft mud in a matter of 15 to 20 minutes.

Sample coring and trimming. In order to perform the desired triaxial tests, it was
necessary to obtain NX sized cores (2-1/8in or 56mm) from the large blocks. Because
the Pierre shale is both hard and highly susceptible to softening, this proved to be very
difficult. The hardness of the Pierre shale required the use of a diamond-
studded, rotary coring drill. The cores used in this study were obtained using a coring
drill mounted in the laboratory.
98

Table 6.1. Calculated mineralogy mode of Pierre shale using microscopic and
x-ray data (Braddock and Machette, 1976).

During the initial attempts at coring, the downward driving force was manually applied
by the machine operator. However, it was difficult to manually maintain a constant
moderate force, and this method proved fruitless for the Pierre shale. During even brief
periods when excessive force was unavoidably applied, two undesired effects
occurred: (a) the core experienced high torque stresses and twisted apart, and (b) large
amounts of heat was generated by friction, resulting in drying of the clay shale, and
subsequent softening to mud with rewetting. Some success of obtaining cores of Pierre
shale in the field has been achieved using pressurized air as the drilling "fluid".
However, a crude attempt at air drilling in the laboratory resulted in drying of the
material and "locking up" of the sample inside the core tube.
99

Figure 6.3. Photograph and sketches of dried cores of Pierre shale showing
different orientations of fissure pattern (photo by author).

Finally, successful core samples were obtained in the laboratory, using moderate
amounts of water, and a constant, moderate downward force supplied by weights
attached to the coring drill. This configuration resulted in a coring rate of about one
inch of core length for every three minutes of drilling time. After washing off the mud
100

along the sides, the core samples exhibited very smooth and straight sides, with no
further sign of alteration.

In order to eliminate any anistropic effects during the primary test program, all of
these samples were cored perpendicular to the dominant plane of fissuration. This
plane corresponds to either the original bedding plane, the plane perpendicular to
vertical unloading, or probably both. The dominant plane of fissuration is not obvious
in wet, undisturbed samples, but can be readily distinguished by the pattern of
fissuration resulting from drying. As illustrated by the photos in Fig. 6.3, dried Pierre
shale samples exhibit one well- defined, primary direction of uninterrupted fissures
accompanied by one or more sets of secondary fissures. Although the orientation of
the secondary set is varied within the same sample, it is always roughly perpendicular
to the primary set. The samples used for the test program were all cored perpendicular
to the primary plane of fissuration, with the exception of two samples which were
cored about 70o from horizontal in order to assess the effects of sample orientation.

After coring, the samples were trimmed on the top and bottom in two stages: the first
using a rotary diamond rock saw to make initial cuts, and the second involving leveling
and smoothing of the top and bottom, using a lapidary polishing table. In order to
maintain a length to diameter ratio of about two, the samples were trimmed to lengths
of about 4.2 inches. Due to overpolishing to remove chipped corners and to level the
tops and bottoms, most samples were slightly shorter than 4.2 inches, although none
were less than 4.02 inches.

Great care was taken to prevent any drying of the clay shale at all times during and
after the sample preparation procedures. The material was continually wetted during
all operations and samples were immediately resubmerged in tap water afterward.

Sample drying and the inducement of fissures.


The proposed test program required a statistically significant number of fissures
within the samples in order to enhance wetting and to, of course, examine the effects
of fissure deterioration within the clay shale. Since the Pierre shale does not naturally
exhibit an abundance of open fissures at the scale of the samples, it was necessary to
induce these fissures by some reproducible method.
101

Preliminary studies determined that an extensive network of fissures could best be


created and reproduced by the drying process itself. It was desirable to dry the samples
under a confining pressure while in the cell. However, because of the extremely low
permeability of unfissured Pierre shale, it proved impossible to initially dry the samples
in the cell within a reasonable period of time. In addition, the pattern and surfaces of
the fissures produced by oven- or air-drying the material were found to be very similar
to those observed for the Pierre shale in the field. It was therefore decided that the
samples would first be dried outside of the cell, and then wetted under confinement
within the cell.

It was further determined that the spacing of the fissures varied with the intensity of
the drying process. Oven-dried samples created fissures with spacings of 1-2 mm
(0.04-0.08 inches), while air-drying produced fissures every 6-15 mm, or every 0.2-0.6
inches. The oven-dried samples appeared too intensely fissured; the samples used in
this test program were therefore air- dried for at least 30 minutes to induce fissuration.
This was carried out inside of a laboratory cabinet in order to minimize the influence of
drafts and sudden temperature changes.

It is important to re-emphasize that the purpose of these tests was to assess the
potential effects of slaking and swelling with fissures of clay shale, and not necessarily
to provide design parameters for the Pierre shale. Although it is desirable to model
conditions in the field as closely as possible, this was not a requirement of the test
program. Still, the author feels that the conditions created in these tests are very
similar to those occurring in the field at a larger scale and over a longer time. The
samples in this test program might therefore be considered as miniature models of a
jointed clay shale mass, on which is imposed an accelerated version of the slaking
process.

Test Procedure
General Consideration. The effect of slaking on a given material is dependent to
some extent on several external variables, as well as those inherent material factors
such as clay mineralogy or amount of stored strain energy. These external factors
include: (a) the hydrostatic and deviatoric states of stress acting on the material, (b)
the length of drying periods, (c) the length of wetting periods, (d) the number of
102

wetting and drying cycles, and (e) the chemistry and temperature of the slaking fluid.
The two variables investigated in this test program were the time periods of drying and
wetting. All slaking was carried out under isotropic confining pressures of 10 psi (69
kPa) for one cycle of drying and wetting, using distilled, de-aired water at room
temperature.

The testing procedure involved four stages consisting of:


(a) drying of the sample to a predetermined water content, under zero confinement,
in order to induce fissuration and to initiate the drying-wetting cycle,
(b) wetting of the sample under constant confining pressure of 10 psi, to induce
slaking and softening,
(c) strain-controlled deviatoric loading of the sample under confining pressures of
10, 30, and 50 psi, in order to determine the strength and stress-strain response, and
(d) post-testing treatment, consisting of either the determination of the variation of
water content within the specimen, or the impregnation of the specimen with
Carbowax, in order to preserve it for fabric studies.

Throughout all stages, changes in the sample volume, sample weight, or both, were
measured so that the water content of the specimen could be determined at any time.
Two sets of tests were performed in the conventional triaxial cell. The data from the
first series of tests were affected by some undesirable consequences of the testing
procedure. The test procedure was then improved, resulting in some very informative
data from the second series of testing.

Initial conventional triaxial tests. During the initial testing series in the conventional
triaxial cell, eight tests were performed in order to assess the feasibility of the test
program, and to determine which variables should be investigated during the primary
test program. The three variables which were studied during these initial tests included
(a) the confining pressure acting during the slaking stage, (b) the length of the drying
and wetting period, and (c) the number of wetting and drying cycles. After the slaking
stage, all of these samples were loaded to failure under a confining pressure of 30 psi,
with a back pressure of 15 psi in the sample. The maximum rate of loading was
0.0018"/min.
103

As will be discussed in the following chapter, these initial tests provided useful
information regarding the effects of slaking on the strain response of Pierre shale.
However, the utility of these initial results is somewhat limited by two factors. First,
portions of some of the samples were not thoroughly saturated before testing, as was
later evident by the presence of well defined saturation lines near the top of a few of
the specimens. Failure in these samples was typically restricted to the saturated zones.

Furthermore, evidence indicates that negative pore pressures, generated during


dilation, were not sufficiently dissipated by the samples. As discussed previously,
volume changes were measured for both the cell and sample water. If the volume
changes in the cell water are corrected for the volume of the piston, then these changes
result solely from compression or dilation of the sample. Changes in the sample
volume similarly result from the flow of water out of and into the sample, in response
to compression and dilation, respectively. Under ideal test conditions, in which the
degree of saturation of the sample does not change, the changes in these two volumes
should be of equal magnitude but opposite in signs.

Typical changes in the volume of sample water are compared to the corrected volume
changes of cell water for several of the initial tests are plotted in Fig. 6.4. As shown in
the figure, these volume changes are rarely equal and indicate that the volume of water
in the sample changes at a slower rate than the corrected volume of cell water.
Therefore, the sample is decreasing in volume without the removal of an equal amount
of pore water. Since water is essentially incompressible, these results probably indicate
the presence of gases within the pores, and therefore the incomplete saturation of the
sample. In addition, after the peak strength has been reached, dilation of the sample is
evident from the sharp increase in volume as exhibited by the cell water. However, in
most of the initial test, the volume of water in the sample changes very little after
dilation begins, possibly indicating that the flow of water back into the sample is not
rapid enough to balance the negative pore pressures created during dilation.

In summary, the results from these initial tests indicated that the wetting was in general
not adequate to sufficiently saturate the samples. Furthermore, the negative pore
pressures generated during dilation at failure, were not dissipated rapidly enough
104

Figure 6.4. Irregular volumetric data from tests on two poorly saturated
samples.
by the sample. These two undesired effects result from two major causes: (a) the back
pressure acting on the pore water was too low, and (b) the rate of loading at failure is
too high relative to the rate of pore pressure dissipation.
105

The benefits of increasing the back pressure of the pore water are three-fold. First, at
higher pore pressures, it has been shown experimentally, as well as theoretically, that
less gas will come out of solution given an equal drop in fluid pressure (Bishop and
Henkel, 1962; Lowe and Johnson, 1969). Second, cavitation is essentially prevented at
higher back pressures since these pressures are unlikely to become tensile with
dilation. Finally, higher back pressures allow for greater pressure gradients during the
percolation process, and therefore more effectively counteract any gravitational effects
which might prevent complete saturation.

Pore pressures can be eliminated or minimized during testing either by reducing the
rate of pore pressure generation or by increasing the rate of pressure
dissipation. In practical terms, these can be accomplished by decreasing the rate of
loading or by increasing the flow of water into and out of the sample. The problems
encountered during these initial triaxial tests were avoided in the final test series by
improvements in the testing procedure.

Improved conventional triaxial tests. The final series of tests similarly employed the
conventional triaxial cell, but included several improvements in the testing procedure:
(a) the back pressure was increased from 15 psi to 60 psi (or 414 kPa),
(b) the initial rate of loading remained at 0.0018" per minute, but was significantly
reduced to 0.0008"/min well before failure,
(c) filter paper strips were placed along the length of the sample in order to greatly
facilitate flow to and from the sample,
(d) the time periods allowed for wetting and consolidation were increased, and
(e) the procedure for measuring the final water content of the sample after failure was
refined.

The changes in back pressure and the addition of filter paper strips resulted in rapid
and more complete saturation of the initially dried samples, as well as more rapid
dissipation of excess pore pressures imposed by hydrostatic and axial loading.
According to Bishop and Henkel (1962), the radial drainage generated by the presence
of the vertical filter strips can alone increase the rate of flow into and out of
106

Table 6.2. Procedure checklist for triaxial tests on intact and softened Pierre
shales samples.

the sample by a factor of 10. In addition, the rate of loading near failure, and therefore
the rate of excess pore pressure generation, was reduced to less than half of the initial
value.

In contrast to previous tests, the data from this third series of tests showed no
significant deviations between the amounts of water flowing into and out of the
sample, and the amounts flowing to and from the cell. It can therefore be assumed
107

Table 6.2 (cont)

that the samples were sufficiently saturated and that any dissolved air in the pore fluid
remained in solution during the dilation of the sample. This at the least indicates that
any excess pore water pressures (negative or positive) that might have been generated
were not high enough to alter the solubility of the pore water. For reference, the
108

procedure check-list for the final series of tests is presented in Table 6.2. All steps
involving critical measurements are boldfaced. Great care was given to the
measurement of sample volume changes before, during, and after deviatoric loading of
the sample. The procedure for determining the water contents will be discussed in
more detail in the next section.

Water content measurements. In order to determine the initial volumes of water and
solids in the samples, water content measurements were performed on six randomly
sampled disks of saturated Pierre shale. The samples were weighed before and after
oven drying. Initial sample volumes were determined both by direct measurement with
calipers and by recording the volume displaced as the sample is immersed in calibrated
cylinder. The results of these measurements gave the following average values:

unit density = 2.23 - 2.26 gm/cm3

porosity (n) = 26%

void ratio (e) = 0.35 - 0.36

water content = 13.2%

specific gravity of solids 2.65 .

As indicated by the procedures check-list in Table 6.2, four critical values were
determined at various stages of testing: (1) the initial sample volume, (2) the initial
sample weight, (3) the final sample weight after removal of the sample from the cell,
and (4) the volumetric changes recorded by the pipettes during deviatoric loading of
the sample. By assuming the initial porosity (n) of 0.26, as determined above, the
sample's specific volume can be calculated at any point during the consolidation or
shear testing procedures, as follows:

(VW)i = (VT)i * n (7.1)

(VS)i = (VT)i - (VW)I (7.2)


109

(VW)f = (VW)i + [(WT)f - (WT)i] /γw (7.3)

VW(t) = (VW)f - Vp(t), (7.4)

where V and W are volume and weight, respectively; the subscripts i and f indicate
initial and final values, respectively, while the inner subscripts T, W, and S indicate
total, water, or solids parameters. The values n and γ w are the initial porosity (=0.26)
and the unit density of water; Vp(t) is the difference between the final volume reading
of the pipette and the pipette reading at any time, t; and VW(t) is similarly the water
volume of the sample at any time. In addition, the amount of water lost during the
drying stage can be determined from the volume and weight measurements recorded
immediately before the sample is placed in the cell.
110

CHAPTER VII

TEST RESULTS AND ANALYSIS

The test program described in the previous chapter provided several types of data,
including volumetric changes during drying and rewetting, volumetric and stress-strain
response during shear, and both peak and ultimate strength values. Analyses using
these data examined: (a) anisotropic shrinkage and swelling in response to drying and
wetting, respectively, (b) rate of drying, (c) changes in Youngs Modulus and Poissons
Ratio in response to slaking and swelling, (d) reduction of strength in response to
slaking and swelling, (e) normalization of stress-invariants, P and Q, to establish the
existence of a Hvorslev surface and thus assess the role of slaking in the critical state
model, and (f) assessment of strength anisotropy resulting from fissure orientation.

In this dissertation, the individual tests will be coded according to the following
scheme:
CP/TD-TW (for example 30/4-23),

where CP indicates the confining pressure during shear loading in psi units, TD the
approximate time of drying in days, and TW the approximate time of wetting in days.
Therefore, the example above indicates that the sample was dried for 4 days, wetted for
1 day, and then tested at a confining pressure of 30 psi. In addition, a value of TD =
"H" indicates that the time of drying was one half hour, while the letter "A" or "B" at
the end of the test code designates the first or second run of duplicate tests. Finally,
"HA" at the beginning of the code indicates one of two special "high angle" tests, in
which the samples were cored at a highly oblique angle.

All data processing and analyses were performed on a Corona PC computer (IBM PC
compatible). Most plots were produced on a Hewlett-Packard HP7470A plotter.

Volumetric Changes During Drying and Wetting


Drying. As previously discussed samples of Pierre shale were air dried under
conditions of zero confinement, for periods of up to 28 days. These samples were
111

weighed at various times during the drying process, and their dimensions measured
with caliber’s at the beginning and end of the drying period.

Since the drying procedure was independent of the test apparatus, the shrinkage data
includes results from all three-test series. In addition, continuous weighing of the
samples during drying allowed fairly extensive records of the changes in water content
with time. These results, plotted in Fig. 7.1, exhibit a relatively rapid exponential
decrease in water content from the initial value of 13.2% to a value of 4% to 6% after
100 hours, or about 4 days. After this, the rate of drying is greatly reduced, such that
the water content is only 3% after 677 hours, or about 28 days.

Values for void ratio, saturation, and shrinkage strains, measured at the end of the
drying periods, are presented in Table 7.1. These results indicate that the drying
process is accompanied by a decrease in the void ratio from an initial value of .35 to
about .24 after 4 days, and .22 after 28 days. Again, the major shrinkage is shown to
occur within a 4 day period. Similarly, the saturation decreases rapidly in 4 days to 50
or 60%, but continues to show a significant decrease in 28 days to a value of 34%.

Figure 7.1. Drying curve showing the rate of decrease in water content with
increased time of open-air drying of Pierre shale samples; based on data from
several test
112

Table 7.1 Water content (w%), specific volume (v), saturation (S%), and
shrinkage strain data for Pierre shale samples after open-air drying.

The values of saturation and void ratio measured during the drying process are
determined from measurements of the sample volume, as well as its weight, and are
somewhat more scattered than the values for water content. This scatter results
primarily from the opening and closing of fissures which occur during the drying
113

process. Although fissures are generally not evident in intact cores of Pierre shale,
microfissures may exist at a scale too small to see with the unaided eye. Visual
inspection during drying indicates that these fissure tend to open, or possibly form,
within 10 to 30 minutes after the initiation of drying. Many of the fissures open about 1
mm during the drying period between 1 hour to 40 hours, presumably as a result of
differential shrinkage between the exterior and interior of the sample. After 45 hours,
the fissures begin to close, such that by 70 hours, or less than 4 days, the fissures are
once again tightly closed.

Thus, three observations suggest that although drying may be initially concentrated
along fissures, drying within the sample is for the most part uniform after four days.
These include the rapid drying to low water contentswithin four days, the absence of
significant changes in sample volume after four days, and the uniform shrinkage
indicated by the reclosing of the fissures. In addition, the lack of "barreling" in the dried
samples indicates that the drying along the vertical axis was uniform, even though the
top and bottom of the samples acted as free surfaces.

As was demonstrated in Fig. 5.3, the Pierre shale exhibits a preferred direction for
fissuration upon drying. All samples except two were cored perpendicular to this
direction, which probably represents the horizontal plane of deposition. The data
presented in Table 7.1, indicate that the shrinkage strain is anisotropic, with axial
shrinkage being about two times greater than in the radial direction. The magnitudes of
shrinkage are rather high with about 2% vertical strain in one day, about 4.5% in 4
days, and 5% in 28 days. Some negative strains, indicating expansion, were measured
in samples which had dried for less than a day, and probably represent the influence of
fissure opening during initial drying. The shrinkage ratios reported above are in
agreement with those reported by Fleming et al (1970) for the Pierre shale in South
Dakota. The implications of these shrinkage strains will be discussed later in this
section.

Rewetting. As discussed previously, the samples which had undergone drying, were
then placed in the triaxial cell and rewetted for periods ranging from one to 23 days.
Because of testing difficulties, only the data from the final series of tests are considered
reliable, and thus, all rewetting of samples occurred under 10 psi confining pressures.
114

The samples were free to swell in all directions in response to rewetting in the triaxial
cell.

The vertical displacement of the sample undergoing wetting was measured directly
using strain gages. In addition, the volumes of the samples at the beginning of shear,
were calculated using the method described in Chapter VI. Thus, the axial and radial
strains resulting from rewetting were determined from these measurements, as were the
values for void ratio, water content, and degree of saturation. The results of these
analyses are presented in Table 7.2, along with the results reported previously for the
drying phase. By necessity, the values were calculated for test conditions immediately
before the application of shear stresses, and therefore include the effects of
consolidating the sample from 10 psi to the confining pressure applied during shear
testing. Therefore, the values for samples tested at 30 and 50 psi, probably indicate
somewhat less swell than what had actually occurred.

The data in Table 7.2 indicate that the amount of swelling during rewetting is more
affected by the length of the previous drying period, rather than the length of the
wetting period. Samples which had dried for periods of only half an hour, merely
regained the volume lost during the drying period. In fact, the final water content and
void ratio values for these samples are essentially equivalent to those for samples which
had not experienced any drying period.

However, samples which had been previously dried for 4 days or more, swelled
significantly upon rewetting, and in fact swelled to water content values which were
much higher than the initial value of 0.35. This increased water content is undoubtedly
the result of the breaking of bonds associated with drying-induced slaking. The amount
of swell resulting from slaking was high, indicated by water content values of about .40
for samples which had dried for 4 days, and a value of .51 for the sample dried for 28
days.

In Fig. 7.2, the rate of drying is indicated by plots of void ratio and degree of saturation
over time. As indicated previously, the data show a rapid rate of drying prior to 4-5
days, followed by a much slower rate. As the degree of saturation returns to 100%
during rewetting, the sample swells beyond the initial void ratio of 0.35. It is
115

Table 7.2 Water content (w%),specific volume (v), saturation (S%), and
shrinkage strain data for Pierre shale samples after open-air drying and
rewetting.
116

Figure 7.2. Drying and wetting curves as a function of drying time for the
Pierre shale.

Also apparent that the minor amounts of water lost after four days of drying can have a
very significant effect of the amount of swelling during rewetting.

The axial and radial strains presented in Table 7.2 are scattered and somewhat
unreliable. However, it appears that the axial strain is again greater than the radial
strain, although the swell ratio is generally less than the value of 2 to 3 reported for the
117

shrinkage ratio. Finally, the rather large value of almost 9% for the axial swell of
sample 30/28-3 should be noted.

Summary of the drying and wetting data. Core samples of Pierre shale exhibited
significant shrinkage and swelling in response to drying and rewetting, respectively.
Axial strains of over 4% were measured after 4 days of drying, while one sample
exhibited axial strain of over 5% during 28 days of drying. Rewetting of these samples
resulted in excessive swell beyond that required to return the sample to its original
state. Axial swell of almost 9% was measured for the sample which had undergone 28
days of drying, resulting in an increase in the void ratio from an initial value of 35% to
the final value of 51%. This significant increase in water content after drying and
rewetting is certainly the result of bond breakage during drying-induced slaking.

The length of the period of drying experienced prior to rewetting was far more
significant than the subsequent time of wetting. Samples which had dried for periods
less than one hour, behaved essentially the same as those samples which had not been
previously dried. Still, the most significant changes in void ratio, water content, and
strain occurred within 4 days of drying. On the other hand, the degree of saturation
continued to decrease significantly even after 28 days of drying.

The Pierre shale exhibits a strong preferred orientation of fissuration during the drying
process. In addition, anisotropic behavior was observed during drying and rewetting.
Shrinkage in the axial direction was generally 2 to 3 times that in the radial direction,
while the axial to radial swell ratio during rewetting was typically slightly less than 2.
Visual inspection of the samples during drying did not indicate any distortion of the
core samples, implying that the Pierre shale may be transversely isotropic with respect
to the coring axis. This axis is assumed to correspond to the vertical direction in the
field.

The strains observed during drying are probably elastic, and can thus be represented by
the following equations:

εr = [σr(1-ν ) - νσ a ]/E (7.1)

εa = [ σa - 2 νσr ]/E (7.2)


118

where σr and σa are the radial and axial stresses, respectively, εr and εa are the radial
and axial strains,
and E and ν are Youngs Modulus and Poissons Ratio, respectively. Thus, assuming
that ν and E are constant for both directions

εa/εr = [ σa - 2 νσr ]/[ σr(1-ν ) - νσa ]. (7.3)

Using values of εa/εr = 2, and ν = .25, which


are consistent with data presented in this dissertation, rearranging and solving eqn. 7.3
for σa/ σr indicates that the axial stress acting on the sample would need to be 1/3
greater than the radial stress in order to obtain the difference in strain observed during
drying.

However, the axial and radial stresses are presumably equal during the drying process,
implying that the differences in axial and radial strain must result from directional
variations in ν and E. This anisotropic nature could indicate (1) the transverse
alignment of clay particles during deposition and compaction, (2) previous in-situ
swelling in the vertical direction during unloading, or both.

Stress-Strain Response During Shear


After drying and rewetting, the samples of Pierre shale were tested in triaxial
compression under conditions of vertical strain-controlled loading. In Table 7.3 is an
outline of the series III tests, from which the remaining information has been derived.
The primary set of tests (Set A) involved systematic variations of two parameters: the
time of drying and the confining pressure during shear loading. For Set A tests, the
confining pressure varied between 10, 30, and 50 psi, while the length of drying varied
between zero, one half hour, and 4 days. The period of rewetting during these tests was
1 day. In addition, a second set of tests (Set B), were all performed at 30 psi confining
pressure on samples which had undergone other variations in the drying and wetting
times. These allowed evaluation of the relative significance of the length of
119

Table 7.3. Table of testing program on Pierre shale.

drying and wetting periods on the engineering behavior of the Pierre shale. Finally, two
samples (Set C) were cored at angles which were highly oblique to the other samples,
thus allowing assessment of the effects of fissure orientation.

Various results for all series III tests are plotted in Appendix E, including plots relating
axial, shear, and volumetric stresses to relevant strains. In addition, other test results
will be presented and discussed in the following sections.
120

Axial stress-strain response. In Fig. 7.3 are presented representative plots of axial
stress, axial strain, and volumetric strain, which indicate that there is essentially no
difference in response between samples which have experienced no previous drying and
those which have undergone only half an hour of drying. This implies that the
fissuration which becomes visible after about 15 minutes of drying, may actually
already be present, but tightly closed, in unaltered samples.

The stress-strain response for unaltered Pierre shale is typical for heavily
overconsolidated clays and shales. It is characterized by a steep increase of axial stress
up to a load sufficient for abrupt failure. With continued shear, the strength rapidly
decreases from this peak value to a lower value, sometimes referred to as the "ultimate"
strength (note: the term "residual" strength is not used here, because it is questionable
whether the true residual strength is obtained in any such triaxial compression tests). In
the unaltered Pierre shale, this decrease in strength after failure is quite drastic, with the
ultimate strength ranging from 28% to as low as 20% of the peak strength.

The volumetric strain during shear is characterized by an initial compression of up to


1.5 to 2.5% at failure, followed by sample dilation immediately before and following
failure. In most cases, the sample dilates to volumes well above the initial sample
volume. In heavily overconsolidated or bonded materials, the dilation following failure
is generally concentrated along the plane or zone of failure, and is therefore indicative
of the opening of a crack rather than a general expansion of the entire sample. For
these reasons, volumetric measurements after peak failure are somewhat misleading,
and should be examined with caution.

In Fig. 7.4, overlapping stress-strain plots of "well-behaved" samples illustrate that, for
samples which have undergone similar drying and rewetting schedules, the curves are
very similar even at different confining pressures. However, with higher confining
pressures, the samples tend to fail at higher axial strains, and thus sustain greater axial
stress and volumetric strain. The relationship between the axial stress and axial strain is
best described by the Youngs modulus, E, defined here as the ratio of the change of
axial stress to the change of axial strain, while ratio of the change in radial to axial
strain is given by Poissons ratio, ν . Thus, for samples with the same drying and
121

Figure 7.3. Representative stress-strain plots of samples having undergone no


drying (50/0-1) and half-an-hour of drying (50/H-1) prior to wetting and shear
testing.
122

Figure 7.4. Representative stress-strain plots of samples having undergone


similar wetting- drying cycles but tested at different confining pressures.
123

rewetting history, the values for Youngs modulus and Poissons ratio essentially remain
constant.

However, samples which have undergone longer periods of drying and rewetting
exhibit significantly lower values of E, while the values of ν do not change as
drastically. This is further illustrated by the plots presented in Fig. 7.5, which show very
significant softening of the samples as the time of drying is increased from half an hour
to 28 days.

Upon closer inspection, the prefailure portions of these curves can be divided into three
parts, as in Fig. 7.6. The first stage of the stress-strain curve is generally short and very
shallow, and probably represents initial fissure closing during loading. The remainder of
the loading curve exhibits two well-defined linear segments. These segments appear in
the plots of axial strain versus volumetric strain, as well, although the location of the
break between these curves may be slightly shifted.

The values of E and ν have been calulated for the second and third segments of these
curves, and are presented in Table 7.4. For segment II, the value of Youngs modulus,
E, decreases with softening from an average of about 28,000 psi for unaltered Pierre
shale, to about 9,000 psi for samples which had undergone 4 days drying, and finally to
a low of only 4,500 psi with 28 days of drying. This represents about an 84% increase
in the sample compressibility. Youngs modulus for segment III exhibits a similar
decrease from 22,000 psi to about 5,000 psi with softening.

As illustrated in Fig. 7.7, similar two- or three-segment loading curves have been
observed for other stiff, or cemented, clays (Uriel and Sarrano, 1973; Conlon, 1966;
Davidson, 1977). It has been suggested that the shape of such curves results from the
superposition of frictional and bonding components of material strength. This effect is
illustrated schematically in Fig. 7.8. From both Table 7.4 and the plots discussed above,
the loading curves for unaltered samples of Pierre shale are similar to curves for other
bonded clays, in that the initial loading segment is steeper than the subsequent segment.
Thus, in view of the interpretation discussed above, shearing of the unaltered Pierre
shale is initially resisted by bonding between the clay particles. After this bonding is
disrupted, probably as a result of shearing through asperities, further resistance is
provided by the frictional component of strength.
124

Figure 7.5. Comparative stress-strain plots showing drastic softening of Pierre


shale after slaking.
125

Figure 7.6. Representative stress-strain plot showing three distinct segments


in the prefailure curve.
126

Table 7.4. Values for Youngs modulus, E and Poissons ratios, ν, for loading
segments II and III for unaltered and softened Pierre shale samples.

However, for samples of Pierre shale which have previously undergone 4 days of
drying, the slopes of these two segments are essentially the same. With further
softening, the loading curve actually becomes steeper in segment III than it was in
segment II. In Chapter IV, it was proposed that the softening of the fissure walls might
progress at a faster rate than that of the interior cores of the clay shale. If one considers
127

Figure 7.7. Two- and three-segment loading curves for other stiff, or
cemented, clays (Uriel and Sarrano, 1973).
128

Figure 7.8. Schematic explaining two-segment loading curves, resulting from


the combining of frictional and bonding resistance forces.

the theoretical implications discussed in Chapter IV, the strain hardening observed in
softened Pierre shale may still be in agreement with the interpretation presented above.
However, in such samples, the bonded material remaining between fissures is separated
by the softened fill along the fissures. Thus, the shear load is initially resisted only by
the weaker frictional component of the softened fissure material, as represented by
segment II. Then with further shear displacement, the less softened clay clumps
between fissures begin to come in contact, thereby adding a slightly stronger bonding
component to the overall strength. This would result in a slight steepening of the
loading curve, as observed in the samples of softened Pierre shale.

The values of Young’s modulus discussed above were calculated using the least-
squares method for each loading segment. However, Young’s modulus was also
calculated incrementally for each step of axial loading. In Fig. 7.9, these values for all
129

Figure 7.9 Plot showing the relationship of Young’s modulus to specific


volume for loading curve segment III in Pierre shale.

Figure 7.10. Prefailure P'-v paths for all Pierre samples tested at 30 psi
confining pressure
130

tests are plotted against the specific volume, v, indicating a well defined relationship
between Young’s modulus and the specific volume. The modulus for segment III
remains greater than that for segment II, for specific volumes less than about 1.4. At
values greater than 1.4, the Young’s moduli are essentially equal for both segments.

Volumetric strain during shear. The changes in sample volume during shear are
reflected in the changes of specific volume and volumetric strain. In Appendix E, the
specific volume measured during testing is plotted against the natural log of the stress
invariant, P, for each test. In addition, the prefailure P-v paths are plotted collectively in
Fig. 7.10 for samples tested at 30 psi confining pressure.

These curves exhibit a relatively stiff behavior during initial loading, followed by an
abrupt decrease in specific volume with continued loading. This abrupt change in
volumetric compression probably indicates a change from relatively stiff elastic
behavior to more plastic deformation. With confining pressures of 30 psi, the change
from elastic behavior occurs at P'= 90 psi (Q'= 180 psi) for unaltered Pierre shale, and
at very low values of P'= 35 psi (Q'= 15 psi) for softened samples. In addition, these
curves indicate that the volumetric strains associated with plastic behavior, increase for
samples which have undergone softening in response to slaking. This is evident from
the increased slopes for samples with higher specific volumes.

The values for ∆v/∆P and ∆v/∆ln(P) have been tabulated in Table 7.5, for the loading
curve segments II and III. These data give some indication of the relative volume
changes that occur during the shear tests. However, the volumetric changes measured
by the specific volume, v, reflect changes resulting from both shear and isotropic ("all-
around") stresses. In contrast, the values for the volumetric strain reflect the changes in
volume resulting only from changes in the all-around stress and are given by

εv = - ∆V/Vo = εa + 2 εr , (7.4)

where V and Vo are the present and initial sample volumes, respectively. Values for
∆εv/∆ln(P) which are also tabulated in Table 7.5, range from a low of .0033 psi-1 for
unaltered samples to a high of .0296 psi-1 for the highly softened sample 30/28-3.
131

Table 7.5. Volumetric changes related to P′for loading segments II and III in
Pierre shale.
132

The Shear Strength of Progressively Softening Pierre Shale


The stress-strain plots presented in Appendix E and in the previous Fig. 7.5, exhibit a
very significant decrease in shear strength associated with progressive slaking in the
Pierre shale. Results from tests at 30 psi confining pressure, indicate a 45% reduction in
the peak strength upon rewetting of a samples which have been dried only 4 days, and a
drastic 75% reduction in strength for a sample having undergone only one cycle of
drying for 28 days and rewetting for 3 days. As observed for sample 30/28-3 in
Fig. 7.5, the peak strength has effectively been reduced to a value very near the value
of the ultimate strength of unaltered samples. It is quite remarkable and significant that
such drastic changes can be simply reproduced by a single cycle of drying and
rewetting!

In Fig. 7.11, the peak and ultimate values of P', Q', and specific volume, v, are plotted
for unaltered samples. The peak values of stresses and specific volume exhibit
significant scatter for unaltered samples. In contrast, the same values for samples which
had previously undergone 4 days of drying are well behaved, as seen in Fig. 7.12, and
exhibit a well defined slope of 1.8 and an intercept of 90 psi. These values can be
related to the more familiar Mohr-Coulomb parameters, c' and φ' by the equations

sin φ' = 3 M / ( 6 + M ) , (7.5)

and

c' = [(Q'/M)-P'] tan φ' (7.6)

where M is the slope, ∆Q'/∆P'. These equations give the values of φ' = 43.8o and c' =
54 psi. These values will be discussed in more detail in a later section.

The values for all of the tests are plotted in Fig. 7.13. Of particular interest is the fact
that the peak value for test 30/28-3 is near the value of the ultimate strength as
discussed above. The scatter of data in this plot is indicative of the scatter of shear
strength values typically measured in the field for the Pierre shale. With such a scatter
of data, the difficulties of establishing strength parameters for design become quite
133

Figure 7.11. Peak and ultimate values of P', Q', and v for unaltered samples of
Pierre shale.
134

Figure 7.12. Peak and ultimate values for P', Q', and v for Pierre shale samples
dried for 4 days and wet for 1 day.
135

Figure 7.13. Final and ultimate values for P', Q', and v for all Pierre samples,
unaltered and softened.
136

apparent. It is important to be able to relate the variation of strength values observed in


Fig. 7.13, to some parameter which might be more easily measured and predicted in the
field.

The Effect of Sample Orientation On the Strength of Pierre Shale


All of the results discussed thus far have concerned tests on samples which were all
cored such that the plane of primary fissuration was oriented perpendicular to the core
axis. However, as discussed previously, two samples, designated as "high angle", or
"HA", were cored such that the main fissure direction was oriented at an angle of 57o
to 61o from the axis. In Figs. 7.14 and 7.15, the stress-strain curves for these tests are
compared with the curves from samples with a similar slaking history, but different
orientation. The curves for the unaltered Pierre shale in Fig. 7.14, indicate that the
"high angle" sample exhibits a higher stiffness, but fails at a 40% lower stress. This
lower strength results from the orientation of the main fissures, being more coincident
to the direction of maximum shear. Thus, failure probably occurs entirely along this low
strength fissure.

The curve of the volumetric strain for the "high angle" sample is identical to that for the
other sample, except that failure is initiatiated at an earlier stage. The fact that failure of
the "high angle" sample occurs at the exact location of the break of the εv: εa curve for
the other sample, may indicate that the component of strength responsible for segment
III curves in other samples, is lacking or very weak for samples fissured along the
maximum shear direction.

Thus, for the "high angle" sample, the value of Young’s modulus is significantly higher
than samples cut perpendicular to the main fissure direction. However, the values for
the Poissons ratio are identical. This data supports the implication of the drying data,
that Young’s modulus for Pierre shale varies with the orientation of the sample.

The loading curves in Fig. 7.15 for softened Pierre shale exhibit similar results. The
"high angle" sample, HA30/4-1, is again stiffer than the sample, 30/4-1A, cored
perpendicular to the major plane of fissuration, while the initial Poissons ratio is again
equal for both samples. Similarly, the strength of sample, HA30/4-1, is 50% lower
137

Figure 7.14. Comparison of loading curves for unaltered Pierre shale samples
cored perpendicular (30/H-1) and at a highly oblique angle (HA30/H-1) to
major plane of fissuration
138

Figure 7.15. Comparison of loading curves for softened Pierre shale samples
cored perpendicular (30/4-1) and at a highly oblique angle (HA30/4-1) to
major plane of fissuration
139

than that of sample, 30/4-1A, which has undergone similar slaking history, but is
oriented differently. The data indicate, however, that the reduction in shear strength,
after a four day drying and wetting cycle, is higher in "high-angle" samples. In other
words, the strength of 30/4-1A is 60% of the unaltered sample, 30/H-1, while the
strength of sample, HA30/4-1, is only 43% of the unaltered, "high-angle" sample,
HA30/H-1.

Summary of the Test Results


The drying of Pierre shale results in significant shrinkage. Initially during the drying
process, fissures are either produced or already existing microfissures are opened. With
further drying, these fissures begin to close tightly. Upon drying, the Pierre shale
exhibits a strong preferred orientation for fissuration. The fissure pattern is transversely
isotropic, with one well-defined plane of continuous fissures, crossed at right angles by
one or more sets of rougher, more irregular, and less continuous fissures.

The anisotropic behavior of Pierre shale is further exemplified by the directional


variation of shrinkage and swelling strains. For samples cored perpendicular to the
major plane of fissuration, the shrinkage strain in the vertical direction is 2 to 3 times
greater than the radial strain. Upon rewetting, the vertical swell is slightly less than
twice the radial swell. These results indicate that the values for Young’s modulus, and
possibly the Poissons ratio, vary with orientation for the Pierre shale.

The drying process is accompanied by a significant decrease in the void ratio from an
initial value of .35 to about .24 after 4 days, and .22 after 28 days. Most of the change
in void ratio and water content occur within 4 days drying time. However, the degree
of saturation continued to show a substantial decrease even up to 28 days drying. The
degree of saturation dropped from about 100% to 50% in 4 days, and to 34% after 28
days.

Samples of Pierre shale which have dried longer than about 24 hours, exhibit a strong
tendency to swell beyond that necessary to return the material to its original state, even
under confining pressures of 10 psi. This excessive swell results from the breaking of
bonds during drying-induced slaking, and is probably a major factor in the reduction of
strength observed in the field. The amount of swell during rewetting is much more
140

dependent on the length of time over which drying occurred, than on the time period
for wetting. The amount of swell resulting from slaking is remarkably high. Swelling
strains of almost 9% were observed for one sample which had been previously dried for
28 days. The final void ratio after drying and rewetting changed from the initial value of
.35 to .40 for samples which had dried for 4 days, to .51 for the sample which had dried
for 28 days.

Triaxial compression tests were performed on unaltered samples of Pierre shale, as well
as samples which had undergone various histories of drying and rewetting. The stress-
strain curves for these tests exhibited a two- to three-segment linear curve up to an
abrupt failure. The first segment exhibited low stiffness, and probably reflected the
initial closing of fissures. With further shear along segment II, the loading curve
exhibited an abrupt change in the value of Young’s modulus, which was similarly
reflected in the value of Poissons ratio. The loading curves further exhibited a drastic
decrease in Young’s modulus in response to slaking, while Poissons ratio was not
significantly affected.

After peak failure, unaltered samples of Pierre shale experienced a rapid decrease in
strength to an ultimate value as low as 20% of the peak strength. The peak strengths of
unaltered Pierre shale are rather high, with a value of about 750 psi for 30 psi confining
pressures. However, this strength was shown to be drastically reduced by a single cycle
of partial drying, followed by rewetting at 10 psi. For the sample which had dried for
28 days, the peak strength exhibited an 80% reduction to values very near the ultimate
strength.

The results of all tests showed a significant amount of scatter, even for samples of
unaltered Pierre shale. The scatter was similar to that observed even at single sites in
the field. The results from these tests re-emphasized the need for a model which relates
these data to some parameter which is easily and inexpensively measured in the field.
Without such a model, the assessment of design parameters from such data proves to
be very unreliable, and in many cases impossible.

The critical state model has proved quite successful in relating the strength of normal
and overconsolidated clays to the specific volume of the soil. The following chapter
141

will evaluate the test results within the concept of critical state soil mechanics, in order
to assess whether such a model can account for the effects of slaking in clay shales.
143

CHAPTER VIII

EVALUATION OF TESTS RESULTS WITHIN


THE CONCEPT OF CRITICAL STATE MECHANICS

As introduced in the discussions of Chapter V, the effects of slaking on the engineering


behavior of clay shales might be closely related to the water content, or the specific
volume, of the softened sample. Indeed, closer examination of Fig. 7.13, reproduced as
Fig. 8.1, suggest that the scatter in the values for strength might be correlated with the
similar scatter observed for specific volume. In Fig. 8.2, the peak values for specific
volume are plotted against P'. The data, although somewhat sparse, seem to further
suggest the existence of a relationship between peak strength and the specific volume at
failure. In this chapter, the triaxial compression data is analyzed within the context of
the critical state model. Recommendations for further studies and for practical
application of these results is also included at the end of the chapter.

Assessment of the Effects of Slaking Within the Critical State Concept


As discussed previously, the critical state model is based on well-defined relationships
between the peak and ultimate strengths of clays and their specific volume at failure.
This concept has proved extremely useful in evaluating the engineering behavior of
clays which have undergone consolidation and swelling in response to changes in the
stress state. It would be equally useful if the effects of the swelling which results from
slaking, could be similarly incorporated into the critical state concept.

For heavily overconsolidated clays and clay shales, the peak strength is constrained by
the Hvorslev surface, which has been defined in previous chapters. It is therefore
assumed that proving the existence of a common Hvorslev surface for unaltered and
softened clay shale, is a major step toward assessing the ability of the critical state
concept to account for the effects of slaking in these materials. As discussed in
Chapter V, the existence of the Hvorslev surface can be proved if the values of P and
144

Figure 8.1. Peak and ultimate values for P', Q', and v for all samples of Pierre
shale, unaltered and softened (reproduced from Fig. 7.13).
145

Figure 8.2. Relationship between the specific volume and ln P’ at peak


strength for Pierre shale samples at confining pressures of 10, 30, 50 psi.

Q at failure can be successfully normalized by the function, exp(Γ-v)/λ. Thus, if the


peak values for P' and Q' in Fig. 8.1 can be normalized such that they lie along a
common failure surface, this would suggest that the peak strength of the clay shale is
dependent only on the final water content, and therefore independent on whether this
water content resulted from slaking, simple swell, compression, or a combination of all
of these processes.

Obtaining the normalization parameters. The normalization of P' and Q' commonly
employs the parameters Γ and λ. The values of these parameters define the critical
state line in v:ln P' space, and can best be obtained from triaxial compression tests on
normally consolidated, or lightly overconsolidated samples. Since such test results were
not presently available to the author, it was necessary to obtain these parameters in a
slightly modified manner, using triaxial compression data from the heavily-
overconsolidated samples tested here.

After peak failure has occurred during shear of an overconsolidated clay, the values for
P' and Q' tend to move toward the critical state line in both P':Q' and P':v space.
146

However, the amount of shear during triaxial compression tests on highly


overconsolidated clays is generally not sufficient for the material to reach the residual
strength defined by the critical state. In addition, because failure in these materials is
often concentrated along a narrow shear zone, the ultimate values obtained for P', Q',
and especially v, may not necessarily define the true critical state line, even after
extensive shear displacement. Still, the values obtained at the end of these tests, may
allow the calculation of equivalent parameters necessary to normalize the stresses, P'
and Q'.

As shown in Fig. 8.1, the P' and Q' values measured at the end of the tests appear to be
progressing toward a common straight line in P':Q' space. The approximate slope of the
line along the lower bound equals 1.4. Using Eqn. 7.5, these results indicate an ultimate
friction angle of about 34o.

However, the final values for v and ln(P') shown in Fig. 8.3, do not all appear to be
progressing toward a common line. As discussed previously, this is primarily a
reflection of the uneven distribution of water content within the sample, and the fact
that the value of specific volume measured indicates an average value, rather than the
value along the failure plane. If the actual specific volume along the failure planes could
be measured instead of the average values, then the data would probably indicate a
common critical state line in v:ln P' space. Indeed, the scattering of final specific
volumes is minimized for samples whose values of specific volume at the beginning of
the tests were nearly equal.

For example, the data for unaltered samples in Fig. 8.4, define a critical state line with
slope of -0.020, and an intercept value, vo, value of 1.455. Thus, for the purpose of
normalization, these values will taken as equivalent values for Γ and λ, respectively.
However, because of the difficulties in measuring the actual specific volume along the
failure zone, and to avoid possible confusion, the designation, vo, is used here to
replace the more tightly defined parameter, Γ, but is similarly defined as the value of
the final specific volume at ln(P) = 0. Similarly, while the slope of the line in Fig. 8.4
may be functionally equivalent to λ, it may or may not be the slope of the "true"
147

Figure 8.3 Values of v and ln P’ at the ultimate (final) strength of all Pierre
shale samples tested at 10, 30, and 50 psi.

Figure 8.4. Values of v and ln P’ at the ultimate (final) strength of unaltered


Pierre shale samples tested at 10, 30, and 50 psi.
148

critical state line. Therefore, the slope of this line will be designated as capital lambda,
Λ, instead of the more conventional, λ.

The equation for the Hvorslev equation has been previously derived as

Q = (M-h) exp[(Γ-v)/λ] + hP , (8.1)

or by normalizing P and Q, as

Q/exp[(Γ-v)/λ] = (M-h) + hP/exp[( Γv)/λ] . (8.2)

Thus, if the peak values of P and Q are normalized using the exp[(Γ-v)/λ], the resulting
values should define a straight line along which all peak strength of a heavily
overconsolidated should lie. By replacing Γ and λwith vo and Λ, Eqn. 8.2 becomes

Q/exp[(vo-v)/Λ] = b + m P/exp[(vo-v)/Λ] . (8.3)

Thus, the peak values of P and Q, which were measured for the unaltered and softened
samples of Pierre shale, can be normalized by the exp[(vo-v)/Λ]. If the resulting values
define a common failure surface for both unaltered and softened Pierre shale, then the
critical state concept appears to account for the effects of drying-induced slaking in this
clay shale.

Normalization of data. The results of the normalization, using the values of vo =


1.455 and Λ = 0.020, are listed in Table 8.1 and plotted in Fig. 8.5. As demonstrated by
this plot, the normalized data strongly define a single surface for the peak strengths of
unaltered samples of Pierre shale, as well as for samples which have undergone highly
varying degrees of slaking. Considering the extreme scatter and drastic differences in
strength shown in the previous Fig. 8.1, the degree of fit of the normalized data is quite
remarkable.

The plot in Fig. 8.5 does not include results from the extremely softened sample 30/28-
3, since its normalized strength is more than 50 times greater than the highest value in
the plot. In order to observe the results from this test, the normalized data have been
replotted on a log-log scale in Fig. 8.6. Even with such extreme softening,
149

Figure 8.5. Normalized peak values of P' and Q' for all samples of Pierre shale,
unaltered and softened (excluding 30/28-3), showing well defined failure surface.

the normalized strengths for sample 30/28-3 lie along the previously defined line. These
results strongly support the hypothesis that the effects of slaking can be considered
within the established critical state concept.

A least-squares fit of the normalized data for unaltered samples indicates a correlation
coefficient, r2, of 0.995 and the parameters
m1 = 2.3
150

Table 8.1. P’ and Q’ values for Pierre shale samples, along with normalized
values for P’ and Q’ using vο = 1.455 and Λ=0.020.
151

Figure 8.6. Normalized peak values of P' and Q' on log-log scale for all
samples of Pierre shale, unaltered and softened (including 30/28-3), showing
extension of failure surface.

B1 = 0.2 psi,

where b and m are defined in Eqn. 8.3. As demonstrated in Fig. 8.5, the normalized
data for the moderately softened samples (i.e. xx/4-1, xx/4-3) also lie along this line.
The three more intensely softened samples, however, lie along a slightly shallower line,
having a correlation coefficient of 1.000, and defined by the parameters

m2 = 2.0

B2 = 0.9 psi.
152

It is significant that the segment of the failure surface for unaltered samples also
contains data points for samples which have undergone slaking. This suggest that the
process of slaking merely acts to altered the water content of the sample within the
confines of the critical state model, and does not alter the model itself. If the break in
the failure surface occurred between the unaltered and softened samples, such a
conclusion might have been suspect.

The parameter m can be related to the Hvorslev parameters, φe and ce, which are
defined as the effective cohesion and internal friction angle for samples at the same
specific volume. As with Eqns. 7.5 and 7.6,

sin φe = 3m / (6+m) , (8.4)

and

ce = [(Q/m) - P'] tan φe . (8.5)

From Eqn. 8.4, the value for (φe)1 is calculated as 55o and (φe)2 as 48o. The value of
the effective cohesion, ce, varies with the value of Qo and thus depends on the specific
volume of the sample. Using the test data for σ3 = 30 psi, values of ce for the Pierre
shale have been calculated. Table 8.2 tabulates these values for the Hvorslev
parameters, as well as for the more conventional Mohr-Coulomb parameters.

These results again illustrate the drastic reduction in the cohesion that can occur with
the softening of the Pierre shale. The major reduction of strength during these tests,
results from a drastic decrease in the cohesion from 123 psi to 0 psi. However, in
samples which have dried for longer than 4 days, the strength is further reduced by a
decrease in the friction angle. These reductions in shear strength will be discussed in
more detail in the following chapter.

In addition to the normalization of P' and Q' at peak failure, the values for P' and Q' can
be normalized for the entire loading path during testing. Fig. 8.7 shows the normalized
stress paths from the beginning of shear until peak failure. These paths are similar in
shape for all samples, and are characterized by an initial steep rise in
153

Table 8.2. Cohesion and internal friction angle values for unaltered Pierre
shale, as well as for samples which have been rewetted after various periods of
drying.

Q'/exp[(vo-v)/Λ], followed by a gradual decrease in P'/exp[(vo-v)/Λ] up to a distinct


peak. These peaks probably define a prefailure yield surface for the Pierre shale. After
this peak, both of the normalized stresses decrease rapidly until they intercept the
Hvorslev surface where peak failure occurs. Interestingly, many of these paths appear
to briefly follow the Hvorslev surface before experiencing total loss of shear strength
Because of the localization of dilation within the failure zone, the post-peak paths are
highly unreliable and inconsistent, and have not been plotted.

Normalized strengths for "high-angle" tests. Normalization of the peak stresses for
the "high angle" samples gave the values

P/exp[(ν0 -v)/ ] Q/exp[(ν0 -v)/Λ]

HA30/H-1 0.5 1.3


HA30/4-1 4.6 9.2 .
154

Figure 8.7. Normalized pre-failure loading paths for all samples of Pierre
shale, unaltered and softened (excluding 30/28-3).
155

Comparison of these results with the failure surface for the samples in Fig. 8.5,
indicates that the strength of the unaltered "high angle" sample lies along the same line
as the previous data, but the softened "high angle" sample lies somewhat below.
Assuming that these samples are representative, and that their positions define a failure
surface for "high angle" samples, then this failure surface has an intercept, B = 0.3 psi
and a slope, m = 2.0, which is equal to the slope, m2, for obtained for highly softened
samples in the previous tests. These limited results imply an effective Hvorslev friction
angle of 48o and an effective cohesion of 3.1 psi. The significance of these "high angle"
test will be discussed in more detail in the following section.

Discussion on Slaking and the Shape of Failure Surfaces for Anisotropic


Materials
The previous section demonstrated that the critical state concept can successfully
account for the often drastic effects of slaking in the Pierre shale. It is important to
stress, however, that the present data is insufficient to conclusively relate these results
to those that might be obtained on remolded and artificially consolidated samples of
Pierre shale material.

The effects of fissuration and anisotropy on the shape of failure surfaces. The
samples tested in this program were initially highly overconsolidated, anisotropic, and
fissured clay shales. As noted before, the rather high values obtained for the effective
friction angle are more indicative of rock than soil. The materials typically considered
within critical state models are homogeneous and isotropic. In such materials, the
failure occurs along a simple plane of maximum shear stress, whose orientation is
determined only by the surrounding stress state. The strength of the material is
therefore independent of sample orientation.

However, in fissured materials, or in any anisotropic materials containing directional


elements of weakness, two additional factors may influence the measured shear
strength. First, the plane of failure may be controlled by the position and orientation of
fissures or other weaknesses, and might therefore not coincide with the plane of
maximum shear strength. Secondly, for the same reason, the failure plane may not be
simple and smooth, but may be rough or "sawtooth", similar to the joint model of
Patton discussed in Appendix C. Both of these factors tend to add a extra component
156

of strength, depending on the orientation of the sample, the magnitude of the stress
state, and the relative strength of the "intact" material.

It is quite informative to compare the simple model of Patton with the results that the
present author has obtained for the Pierre shale. The model of Patton is graphically
illustrated in the strength envelopes of Fig. 8.8a, while the normalized failure surface
for the Pierre shale is presented for comparison in Fig. 8.8b. In the Patton model, the
strength of a simple joint (Curve B) is dependent only on the frictional resistance along
the walls, and is given by φ. However, for a sawtooth joint, the resistance to shear
(Curve A) has an additional component, i, equal to the angle of the tooth, and resulting
from the sliding of one tooth over the other (i.e. dilation). At low stresses, this added
component results in a steeper failure envelope, with an total friction angle of (φ+ i). At
higher stresses, however, the shear stress exceeds the strength of the teeth (i.e.
asperities), and shear displacement is accomplished by breaking through the teeth rather
than sliding over them. At this point, the friction angle decreases to φ, the value of the
friction angle for the smooth joint.

The normalized shear strength data for the Pierre shale (Curve C) show a similar two-
segment failure envelope, suggesting the influence of fissuration. For the samples cored
perpendicular to the major plane of fissuration, the failure plane is probably step-like, as
illustrated in Fig. 8.8b. As in the Patton model, these irregularities in the failure plane
add an extra component of shear resistance at lower stress levels, resulting in a steep
failure envelope, and therefore a higher friction angle. At higher stresses, this friction
angle is reduced as shearing occurs through the irregularities, rather than over them.

This model is further supported by the data for the "high angle" tests illustrated by
Curve D. When the orientation of the major fissures are coincident with the plane of
maximum shear stress, as is almost the case for the "high-angle" tests, then the failure
plane is essentially equivalent to the planar joint in the Patton model. As would be
expected if such were the case, the limited data from the "high angle" tests suggest a
failure envelope which is void of a steeper portion and is parallel to the upper portion
of Curve C.
157

Figure 8.8. Schematic comparing (a) the Patton model for a sawtooth joint, to
(b) normalized failure envelopes for Pierre shale.

Of greater interest to the present study, however, is the fact that the failure envelope
for the Pierre shale in Fig. 8.8b, is normalized using a function of the specific volume.
Therefore, an increase in the normalized stresses can also result from an increase in the
158

Figure 8.9. Normalized strength envelope showing low stress strengths,


overconsolidated strengths, and critical state strengths for heavily-
overconsolidated clays in Canada (Graham and Li, 1985).

specific volume (i.e. water content), even if the actual effective stresses are constant or
decreased. In effect, an increase in the specific volume indicates an increase in the
degree of softening within the material. Therefore, with regard to the Patton model, an
increase in the stress state or an increase in the water content, favors the tendency for
shear through irregularities, rather than over them.

The results from the triaxial tests on Pierre shale, therefore support the fissure
deterioration model that the author proposed in Chapter IV. They further indicate that
the influence from fissures is minimized by the either increasing the stress state or
increasing the water content of the clay shale.

Similar two-segment boundary surfaces have been reported in two recent papers for
natural heavily overconsolidated clays in Canada (Graham and Li, 1985; Graham and
Au, 1985). In Fig. 8.9, the strength values were normalized using the preconsolidation
159

pressure, σvc, defined as the maximum pressure to which the soil has been subjected to
in the past or present. Although the actual values of the normalized strengths are
different, this is equivalent to the normalization procedure used in the present paper.
The normalized data for natural samples defines a two-segment boundary surface, A-B-
C, for overconsolidated samples. As predicted above, the initial steep failure surface
changes abruptly to a more shallow surface, and at its extreme, curves over to intersect
the critical state line. At lower stresses, or lower specific volume, the material strength
is greatly influenced by the presence of fissuration, and is defined by the steeper
surface. With higher stresses, or at higher specific volumes, the influence of these
fissures is greatly reduced, and the strengths can be defined by the flatter slope.

In these two papers, the authors have also investigated the relationship between the
strengths of natural clays, and that of clays which have been artificially remolded and
consolidated in the laboratory. As is evident in Fig. 8.9, if the strengths of the
artificially remolded samples are normalized by their preconsolidation pressures, then
the resulting values lie well above the failure envelope for the natural materials.

Figure 8.10. Normal consolidation and critical state lines in P':v space for
natural and remolded clays of Fig. 8.9 (Graham and Li, 1985).
160

Figure 8.11. Failure envelope for natural and remolded samples, with data
normalized according to respective normal consolidation lines (Graham and Li,
1985).
However, as shown in Fig. 8.10, the normal consolidation lines for the natural and
artificially remolded samples are essentially parallel, but are shifted in position. This has
been attributed to aging or cementation. Interestingly, if the artificially remolded
samples are normalized in reference to their own consolidation lines, then the
normalized strength lie on the same failure surface as the natural samples (Fig. 8.11).
This should be considered when attempting to compare results from natural and
artificially-remolded materials.

Slaking and the intensity of natural remolding. The test results presented in Chapter
VII, indicate that a 75% decrease in strength can occur in Pierre shale after only a
single cycle of drying and rewetting. In the present chapter, this has been interpreted as
a reduction of 6o in the internal angle of friction, and a very significant loss of cohesion
161

from 123 psi to zero. Although the reductions in shear strength experienced in these
samples are rather drastic, the calculated friction angles for the softened clay shale are
still more indicative of rock-like materials than of soft soil. Even the ultimate strength
of these samples is defined by a rather high friction angle of about 34o.

Although these angles are similar to limited strength data for firm Pierre shale at Oahe
Dam (Fleming et al, 1970), they are still much higher than those obtained for highly
weathered Pierre shale in the field. This indicates that the degree of natural remolding
experienced in the field must be much more intense than that which has been
accomplished with the samples in this test program. Indeed, the softening, which was
produced in the laboratory by a single cycle of drying and wetting, was probably
primarily confined to zones along the vertical and horizontal fissures. Therefore, the
higher values for friction angles observed in these tests probably reflect the influence of
remaining zones of stiff, intact material, and therefore the continued influence of fissure
orientation and roughness.

In summary, more intense remolding is required before the influence of fissures is


totally destroyed in the Pierre shale. Furthermore, the intensity of remolding in the field
is apparently greater than has been observed in these test. This does not suggest that
the processes other than slaking are required to achieve this intensely remolded state.
Indeed, more intense natural remolding might be achieved in the Pierre shale, and other
clay shales as well, by (a) increasing the number of cycles of drying and wetting, (b)
increasing the amount of drying for each cycle, (c) allowing more time for the clay
structure to adjust itself to the new state of increased water content, or (d) slaking the
material in the presence of shear stresses.

Since the sample 30/28-3 has already reached a state of zero cohesion, with more
intense slaking and softening of the clay shale, we should expect to begin to see a
further reduction in the internal friction angle. This would be reflected by either a
gradual bending over of the failure envelope, or an abrupt break in its slope to values,
more indicative of soft clay soil.
162

Figure 8.12. Complex P':v history followed by Pierre shale sample 30/28-3,
prior to and during triaxial testing.
163

Slaking and the Critical State Concept: Recommendations for Further


Studies and Practical Application
The test results presented in this dissertation are very encouraging. Considering the
complex history of these clay shale samples illustrated in Fig. 8.12, it is significant and
useful that the resulting peak strengths are dependent only on the final states of stress
and specific volume, and not on the path itself.

Still, these findings are preliminary, and general conclusions based on these results are
limited. More extensive investigations are required before the critical state concept
proves applicable to all clay shales undergoing slaking under various conditions. In the
following sections, the author presents suggestions regarding the potential application
of these results to practical problems in the field, and discusses recommendations for
future studies, both in the laboratory and in the field.

Practical application of the Critical State Concept in Clay Shales. The literature
reviews and discussions in Chapters II and III, attest to the great difficulties
encountered by engineers who attempt to accurately and reliably assess the engineering
behavior of clay shales in the field. Two major complications are recognized: (a) the
extreme variability in the strength and stress-strain parameters of a clay shale, even
within very localized areas, and (b) the drastic changes in the engineering behavior
which clay shales can undergo within geologically-short time periods. The author
believes, and has supported the belief in this dissertation, that in many clay shales, such
complications result primarily from the swelling associated with slaking.

In order to account for the softening effects of slaking, the engineer must either be able
to associate the slaking history to the final engineering parameters of the clay shale, or
he must be able to measure these final parameters directly. Unfortunately, the process
of slaking in a clay shale deposit can be extremely variable and complex, depending on
such factors as the local topography and stratigraphic layering, the frequency and
orientation of jointing, and the local climatic and hydrologic environment, including the
amount of rainfall, the length and frequency of wetting and drying cycles, the
infiltration and evaporation characteristics of the overlying soil, and the flow
characteristics and chemical composition of the ground water, for example. Clearly,
164

reliable assessment of the slaking history, or slaking future, of any soil is presently
impractical.

Therefore, in order to assess the present engineering behavior of a clay shale, or to


predict future changes in this behavior, we must obtain useful strength and stress-strain
parameters by direct measurement, or indirectly using some other model. For practical
field applications, direct measurement of strength parameters through laboratory testing
of clay shale samples has proved to be costly, inadequate, and unreliable. The extreme
spatial and temporal variability of clay shale strength requires a very extensive testing
program if the strength and stress-strain behavior of the clay shale deposit is to be
adequately determined. Many of the past engineering difficulties with clay shale
deposits can probably be attributed to inadequate sampling programs.

Because of high expense and limited range, in-situ testing is generally impractical for
most engineering projects in clay shales. Even with laboratory testing, the cost of
coring, retrieving, and testing of core samples of clay shale generally prohibits a
sampling program which is extensive enough to adequately assess the strength of the
clay shale deposit. This becomes particularly evident when one considers the need for
continuous monitoring of the material strength with time.

However, test results presented in this dissertation suggest that the effects of slaking
within a clay shale are restrained by the critical state model. Therefore, once the critical
state model has been defined, the strength of a clay shale which has undergone slaking
can be determined if one knows the water content and the surrounding stress state. If,
as suggested, the strength of a clay shale can be indirectly determined from its water
content, then mapping and monitoring of the strength within a clay shale deposit might
be accomplished, adequately and inexpensively.

In contrast to the testing of strength, the sample retrieval and laboratory procedures
associated with the determination of water content are uncomplicated and inexpensive.
Because of such low expense, literally thousands of samples could feasibly be retrieved
and their water contents determined. This could allow very extensive mapping of clay
shale strength, both laterally and with depth, as well as provide an inexpensive means of
monitoring strength changes with time. In addition, the mapping and monitoring of
strength variations might be further enhanced by the use of electrical resistivity, self
165

potential, or seismic measurements, which can provide indirect means of assessing


variations in water content. These methods are discussed in detail by Scully (1973), as
well as in several geology texts.

Thus, the major advantage of using the critical state model to constrain the effects of
slaking, is that the model provides us with a parameter (i.e. water content) which can
easily and cheaply be measured at a large number of locations within a clay shale
deposit. In addition, changes in strength with time can more cheaply be monitored
using water content measurements, rather than direct measurements of material
strength.

This relationship between water content and strength in a slaking clay shale does not at
the present allow for the prediction of strength changes which might occur in the clay
shale deposit with time. However, such this relationship to water content may facilitate
future studies attempting to relate environmental conditions to the rate of slaking in a
clay shale deposit. This is a result of two factors: (1) the ease with which water
contents can be monitored with time could result in more field data relating water
content changes to environmental parameters, and (2) laboratory studies which relate
water content to various slaking processes should be much easier to perform than those
which require shear strength measurements after slaking.

As envisioned by the author, the analysis of a project site within a clay shale deposit
might proceed as follows:

(a) preliminary electrical resistivity, self-potential, and seismic refraction studies


carried out at the surface, in order to detect possible stratigraphic and hydrologic
boundaries, as well as potential zones of abnormally high water contents;

(b) an extensive program of coring to obtain samples to be used for water content
measurements; in addition, these samples should be used to recognize stratigraphic
boundaries by determining mineralogy, grain size, etc.; retrieved samples large enough
to perform shear strength tests should be set aside; however, an infinite number of
determinations of water content should be possible along each core since the integrity
of the sample is not as critical for these measurements as it is for shear strength testing;
166

furthermore, the shafts produced by this coring should be employed for further
electrical resistivity logging; such studies would help recognize potential failure planes;
(c) if possible, measurement of the natural stress state within the deposit, using
relaxation determinations on core samples or other more sophisticated techniques;
though such measurements may not be necessary, it should be kept in mind that
excessive horizontal stresses are common in slaking and swelling materials, such as clay
shales; these stress state determinations would be important for input of initial stress
conditions in any later finite element analysis;

(d) determination of the normalized failure envelope, using intact samples retrieved
from the coring program, and possibly remolded samples; if the material behavior can
be constrained by a critical state model, then a large number of intact samples is
probably not necessary since scatter of the data should be minimized by the
normalization process; if more than one potentially unstable stratigraphic layer exist,
failure envelopes should be determined for each layer; the procedure for obtaining the
failure envelope would be similar to that used in this dissertation:

(i) determine the normal consolidation line coefficients from consolidation


tests, if possible, or determine the coefficients of the critical state line from
intact samples, as in this dissertation (or from remolded samples, if this proves
successful); these various methods will be discussed in more detail in the
following section;

(ii) using these parameters, normalize the shear strength data from triaxial
tests on intact samples, on samples which have undergone induced slaking, as in
this dissertation, and possibly on artificially remolded samples;

(iii) plot normalized data and calculate coefficients for the failure
envelope;

(e) once the normalized failure surface is defined, finite element analyses can be
performed on the deposit by assuming constant volume; thus the water content which
has been determined for each point based on the actual measurements, is assumed to
remain constant during shear loading, and is used to normalize the stress state at that
point.
167

Although such analyses are extremely useful for predicting the material stability
following changes in the stress state, it should remembered that such changes in the
stress state may lead to increased slaking in clay shales ; thus, it is important to
continue to monitor water contents at critical locations during and after construction;
the changes in water content, if any, can then be used in updated finite element analyses
to recognize any problems which may be developing. With experience and further
experimentation, it may become possible to predict the increased slaking which might
occur in response to changes in stress states. With such information, iterative
calculations using finite element analyses will allow the long-term stability analysis of
clay shale slopes.

Before applying the procedures described above to actual field conditions, much more
preliminary research is necessary. The intent of this discussion has been to provide a
hopeful summary of the potential application of the critical state model to the analysis
of a clay shale deposits which have undergone, or are undergoing, softening by slaking.
The results presented in this dissertation are, at most, only suggestive of the
relationship between the slaking process and the critical state model, and then only for
the Pierre shale. More research is required before such relationships can be established
for all clay shales under various conditions. Even then, further research must be
conducted before these results are successfully applied to practical problems in the
field. In the following section, the author presents some suggestions for further
research with regard to the effects of slaking, and the application of the critical state
model.

Recommendations for further studies. This dissertation has presented significant


preliminary results which suggest that the effects of the slaking process can, at least
under certain conditions, be constrained by the critical state model. However, further
research is needed before these results can be validated and applied to practical
problems in deposits of the Pierre shale and other clay shales.

As with most naturally-occurring geological materials, clay shale deposits can be


structurally complex. Therefore, the testing and analysis of clay shale deposits entail the
same complications encountered when working with other geological materials,
including the scale effects, inhomogeneity, and anisotropy which result from the
168

presence of joints and the variation of mineralogy and fabric. While these complications
are certainly a concern in any geotechnical analyses, they do not invalidate the potential
application of the critical state model to practical problems in a progressively slaking
clay shale deposit. In fact, the application of the critical state concept to slake-
susceptible clay shale deposits, alleviates the complications of inhomogeneity and
anisotropy resulting from water content variation. Still, scale effects, inhomogeneity,
and anisotropy affect geotechnical analysis in all geological materials to some extent,
and should therefore continue to be investigated.

The problems inherent to the present investigation of clay shales can be divided into
two major groups: (1) defining the normalized failure envelope, and (2) understanding
the process and effects of slaking. Future studies involving the definition of the failure
envelope for slake-susceptible clay shales should proceed along four paths:

(a) determination of the normalization parameters: In critical state soil mechanics,


the parameters used to normalize the shear strength envelope are typically determined
from the position and slope of either the normal consolidation line or the critical state
line in v:ln P' space. The simplest method for determining these parameters use
consolidation tests or triaxial tests on normally-consolidated, remolded samples. Initial
studies should employ such procedures to determine if parameters derived from
remolded clay shale material can be used successfully to normalize data from natural
clay shale deposits. Using remolded samples to determine the normalization parameters
would not only be the simplest method due to the ease of retrieving and preparing
remolded samples, but the use of a standard method would also allow for better
comparison of results between different clay shale deposits. However, as discussed
previously, data from Graham and Li (1985), indicates that the normal consolidation
lines for artificially remolded and natural clay deposits often differ due to aging effects.
In addition, it is questionable whether the normal consolidation line remains linear at
the high preconsolidation stresses often associated with clay shale deposits. Initial
studies need to determine what affects, if any, these factors will have on the
normalization process.

If remolded samples cannot be used to determine the normalization parameters, then


these parameters must be derived from the normal consolidation line or critical state
line of natural samples. Of the two lines, the normal consolidation line should in general
169

be more easily determined. Relative to strength tests, consolidation tests are fairly
direct and require fewer intact samples. If possible, consolidation testing of natural
samples is probably the best method for obtaining the necessary parameters for
normalization. However, due to the high preconsolidation pressures associated with
many clay shales, the pressures applied during standard consolidation tests are
insufficient to return the clay shale to a normally consolidated stress state, thereby
preventing the determination of the normalization parameters. In such cases, a possible
alternative would be to extensively slake the clay shale under low confining pressures
prior to consolidation testing.

As a last resort, parameters which would allow normalization could be derived using
the ultimate strengths of intact clay shale samples. This is the method used in the
present dissertation, and which results in defining a "pseudo-critical" state line. The
critical state line from which the parameters in this dissertation were determined defines
a true material state of the Pierre shale, and the use of the term "pseudo" is not meant
to imply otherwise. However, because failure in highly overconsolidated clays occurs
along narrow failure zones, the water content within the failure zone is generally much
different from the average total water content measured experimentally.

Therefore, the parameters derived from the ultimate strengths of overconsolidated


samples are likely to vary significantly from those derived from the "true" critical state
line of normally consolidated samples. This results in two complications which are
absent in normalization methods which employ remolded samples: (i) while there is
only one "true" critical state line, there can be many "pseudo-critical" state lines,
resulting in nonstandard results which cannot be easily compared between deposits, and
(ii) a "pseudo- critical" state line must be determined for samples having the same initial
stress-void ratio state prior to testing. Therefore, the samples used for determining the
normalization parameters must have undergone the same degree of slaking and
compression. In the present dissertation, unaltered samples were used, since they had
similar values for specific volume and were retrieved from the same horizon. However,
in some instances, it may be difficult to retrieve a sufficient number of samples which
have undergone the same degree of slaking.

If the critical state method is to be of practical use in clay shales, it is essential that
methods be developed for successfully normalizing the failure envelope. Thus, it is
170

suggested that future research concentrate initially on developing and testing the
normalization methods discussed above. Such research should preceed or be carried
out concurrently with the research discussed below.

(b) extending the range of the failure envelope: Although the results presented in
this dissertation indicate a very drastic reduction of strength with a single wetting and
drying cycle, the data also indicated that the strength of even the most softened sample
was still higher than the fully softened strength of the Pierre shale. Within the limited
range of softening induced in this study, a two-segment failure envelope was exhibited,
apparently resulting from the effects of jointing.

It is important to derive the failure envelope for the entire range of softening, from
intact to fully softened conditions. How does the envelope change before intercepting
the critical state line? At what point are the effects of jointing and aging erased? The
answers to these questions could come from studies on the Pierre shale or other clay
shales, which are extensively softened, either naturally or by induced slaking. As
discussed previously, more extensive softening and natural remolding might be
produced in the laboratory or field by increasing the number of wetting and drying
cycles, by increasing the amount of drying during each cycle, by allowing more time for
the clay structure to adjust itself after rewetting, and by slaking in the presence of shear
stresses. Some success might also be achieved in the laboratory by physically remolding
the material, although the results of this method should be checked closely against
results of extensive softening of natural samples.

(c) extending the method to other clay shales: The results presented in these initial
studies of the Pierre shale suggest that the changes in the strength which result from
slaking, are constrained by the critical state model. More studies are needed to confirm
these findings for the Pierre shale, and to extend the range of the failure envelope of the
Pierre shale. As discussed previously, the Pierre shale is an excellent material for such
studies, being highly susceptible to slaking, having a wide range of strengths, being
structurally simple, and being easily sampled. It is therefore suggested that further
research into the relationship between slaking and the critical state model be initially
conducted using the Colorado Pierre shale as the model material.
171

However, similar investigations should be conducted on other clay shales, in order to


determine the extent to which the critical state model can be applied to all slaking
argillaceous materials. Other clay shales, because of greater degrees of isotropy or
simpler joint patterns, may prove to be even better research materials than the Pierre
shale. Furthermore, if similar results are obtained for the London clay and the Cucarcha
clay shales, then the historical records associated with these materials may provide
invaluable information regarding the effects of slaking with time.

(d) effects of anisotropy and jointing. Theoretical and experimental evidence


presented in this dissertation indicate that anisotropy and jointing can have not only a
significant effect on the initial strength of a clay shales, but may also greatly affect the
amount of strength lost during the process of slaking. More extensive experimental
studies should be carried out on the Pierre shale and other clay shale, in order to
determine complete failure envelopes for a number of sample orientations. These
studies would provide important information on the effects that material anisotropy
might have on the shape of the failure envelope, and therefore on the rate of strength
loss with increased water content. With a series of failure envelopes for various
orientations, the normalized strength of a clay shale sample could be determined or
interpolated regardless of the orientation of the material anisotropy relative to the
potential failure plane.

In addition to the experimental testing of samples at various orientations, more direct,


systematic investigations on the effects of softening along single and multiple joints
would be quite useful. Such investigations might entail experimental testing, as well as
computer modeling. Questions to be answered include: How much softening is needed
to remove the effect of roughness along a joint? How does softening progress into a
joint wall; under what conditions is there a distinct boundary between softened and
intact material, and under which is this transition more gradual? What effect do these
different boundaries have on altering the strength of the joint? How is the shape of the
failure zone altered during the softening of various multiple joint systems? How does
this affect the overall strength of the material? How well can the Patton model for
jointed materials be applied to a progressively softening, jointed clay shale? At what
levels of stress and softening are the effect of joints insignificant?
172

In order to apply the critical state model to a progressively slaking clay shale deposit,
we must be able to define a normalized failure envelope which accounts for changes in
the water content (or specific volume) occurring during the slaking process. The
definition of such a failure envelope requires that suitable parameters be obtained from
the normal consolidation or critical state lines. It is therefore important that future
investigations develop simple and repeatable methods for determining the parameters
needed for successful normalization of clay shale strengths. In addition, the validity of
the results must be tested over the entire range of the failure envelope and under
various conditions, including slaking under in the presence of shear stresses, as well as
slaking under different confining pressures, temperatures, and wetting and drying
cycles. Additional failure envelopes should be determined and compared for various
orientations and for several different clay shale materials.

In addition to the investigations discussed above regarding the failure envelope, further
investigations into the actual process of slaking are needed in order to predict the rate
of softening in a clay shale which is susceptible to slaking. Simple laboratory tests could
provide valuable information regarding the changes in water content resulting from
slaking under various factors, including for example, changes in the frequency and
length of wetting and drying cycles, changes in the confining pressures and shear stress
states, changes in the chemical composition of the slaking fluid, and changes in
temperature. These laboratory studies could be complimented with studies in the field
which attempt to relate lateral and temporal variations in the water content to similar
variations in rainfall, temperature, ground water levels, and ground water chemistry, as
well as to surface and stratigraphic topography.

With extensive research into the nature of the slaking process and its relationship to the
critical state model, it may become possible to not only accurately determine the lateral
variation of shear strength in a given clay shale deposit, but to predict the changes in
the strength over the lifetime of an engineering structure, as well.
173

CHAPTER IX

CONCLUSIONS

Clay shale deposits have been a source of much confusion and trouble for engineers
who must deal with these materials. Our present understanding of the engineering
behavior of clay shales is primitive relative to the current state-of-the-art for other
geological materials. The author feels that most of our past inadequacies in dealing with
these materials have resulted from our lack of appreciation of their transitional nature.
This has resulted in inadequate and confusing classification schemes, as well as
dangerous and costly slope and foundation designs.

The transitional nature of clay shales is both temporal and physical. Physically, clay
shales are transitional between rock and soil, and therefore exhibit properties of both.
This has been a source of problem for geotechnical engineers, who traditionally view
geological materials in terms of rock mechanics or soil mechanics, but rarely in terms of
both. In addition, clay shales are transitional in time, and tend to transgress from rock-
like behavior to soil-like behavior within a relatively short time period. Such rapid
changes in material properties create challenges in classification and in engineering
design, both of which are traditionally based on material properties as they exist at the
present and not on possible future properties.

The engineering problems related to clay shale deposits have been worldwide and
involve both natural and man-made structures. These have been discussed in some
detail within this paper using examples associated with four very different clay shale
groups: (1) the British clay shales, (2) the vast clay shale deposits of the north-central
plains of North America, (3) the Cucaracha clay shale in the Panama Canal zone, and
(4) the clay shales of southern Italy. These materials range in their initial characteristics
from the "simple" stiff fissured clays of England, to the "soapy" clays of Panama, to the
scaley, tectonically-sheared clays of Italy, and to the rock-like shales of North America.
Still, they all share the common tendency to rapidly lose shear resistance with time.

Engineering designs based on traditional methods for analyzing geological materials


have been dangerously inadequate. From the discussions presented in this paper, two
174

conclusions become apparent: (1) any analysis which assumes only soil-like behavior,
or only rock-like behavior, will fail to adequately determine the engineering behavior of
clay shales, and (2) any successful analysis must account not only for the material
properties of the clay shale as can be measured at the present, but must also account for
changes in these properties as the clay shale deteriorates into a more soil-like state.

The evidence presented in this dissertation suggest the importance of fissure


deterioration as a viable model for explaining much of the long-term reductions of
strength exhibited by clay shales in the field. Furthermore, much discussion in the
literature has alluded to the importance of fissure deterioration in reducing the strength
of clay shale deposits. However, surprising few, if any, theoretical or experimental,
investigations regarding fissure deterioration have been reported in the geotechnical
literature.

Similarly, the susceptibility and importance of slaking within clay shales has often been
recognized by engineers concerned with clay shale deposits. Yet, the author has found
the literature essentially void of studies regarding the slaking of materials under
conditions of confinement which occur in the field. Similarly, there is a great need for
systematic investigations into the effects of slaking on the strength and stress-strain
response of clay shales. For these reasons, the author has carried out theoretical and
laboratory investigations regarding fissure deterioration in clay shales, with particular
emphasis on the effects of slaking and swelling within fissures.

In Chapter IV, the author presented a simple model for the deterioration of clay shales,
and considered the theoretical implications of a jointed material, with progressive
softening of the fissure walls. The clay shale was depicted as progressing through four
stages of alteration, including (a) a rock-like mass in which the strength was controlled
primarily by the orientation of and strength along the intact fissures, (2) a partially-
softened rock-like mass with the strength controlled primarily by the strength and
orientation of soft, filled fissures, (3) a highly-softened mass consisting of a matrix of
soft clay surrounding stiff, intact cores, and finally (d) a fully-softened, remolded clay.
Thus, the analysis of a clay shale undergoing progressive softening along fissures is
complex, and generally requires some understanding of principles from both soil and
rock mechanics.
175

In Chapter V, the author introduced the idea that the effects of slaking in clay shales
might be incorporated into the important and useful critical state soil mechanics
concept. It was proposed that the process of slaking may simply provide an alternative
path for altering the void ratio of the clay shale within the constraints of the critical
state model. Thus, by increasing the void ratio of the clay shale, slaking may act to
rejuvenate the clay shale, thereby reversing the effects of consolidation and aging which
might have occurred in the past.

Triaxial compression tests were carried out on unaltered Pierre shale samples, and on
samples which were subjected to various degrees of slaking under confinement. The
purpose of these tests were to provide initial data on the effects of slaking on the
strength and stress-strain response of a clay shale, and to assess the potential for
incorporating the effect of slaking into the critical state concept.

The drying and rewetting of the Pierre shale is accompanied by significant excess
swelling well beyond that required to return the shale to its original state. This excess
swell results from the destruction of bonds in response to drying-induced slaking. For
one sample which was dried for 28 days prior to rewetting, the void ratio increased
from an initial value of .35 to a final value of .51.

The drying and rewetting data indicate the Pierre shale is transversely isotropic.
Shrinkage and swelling strains are 2 to 3 times greater in the direction perpendicular to
the major plane of fissuration than those coincident to this plane. These results imply
that the values for E and ν may vary with orientation.

The stress-strain response of the Pierre shale is characterized by an initial closing of


fissures, followed by a two-segment loading curve up to an abrupt failure. After peak
failure, strain softening reduces the strength of Pierre shale to less than 20% of the
peak strength. In addition, the slaking of the Pierre shale significantly reduced the
values of Youngs modulus, while Poissons ratio changed little.

The strength of unaltered Pierre shale is rather high, with a peak strength of about 750
psi for confining pressures of 30 psi. However, this strength was drastically reduced by
a single cycle of partial drying followed by rewetting under confinement. For the
176

sample which had dried for 28 days, the peak strength exhibited an 80% reduction to
values very near the ultimate strengths of unaltered samples.

The strength results of all tests exhibit a significant amount of scatter, similar to that
observed in the field. Using concepts taken from critical state soil mechanics, these data
were normalized using an exponential function of the specific volume. The normalized
strengths strongly defined a two-segment Hvorslev failure surface, with friction angles
indicative of a rock-like material. The effective cohesion decreased from a value of
69 psi to zero. Thus the softening of Pierre shale in response to slaking is accompanied
by a decrease in both the effective cohesion and effective friction angle.

The definition of a single failure envelope for both unaltered and slaked samples, is
highly significant. With such a tool, the variation of strength observed in the field can
be accounted for and mapped using simple and inexpensive measurements of the water
content. In addition, by monitoring changes in the water content with time we should
be better able to assess the reduction of strength in the clay shale, and might be able to
reasonably predict the rate of softening in the future.

The results from the present test program indicate that the slaking and natural
remolding capable of occurring in the field, may be much more intense than that which
has been achieved in the laboratory by a single cycle of drying and wetting. Further
research is needed to more completely define the failure surface for the Pierre shale,
over the range of conditions observed in the field. Such studies should investigate the
strength of samples which have been intensely remolded and softened, either artificially
or naturally, and should account for the effects of fissure orientation.

Finally, the results presented in this dissertation are very encouraging in that they
suggest a simple method which might alleviate many of the difficulties presently
encountered during geotechnical analyses of clay shales. Still, the investigations
reported in this dissertation were only initial studies into a complex engineering
problem. It is the author's hope that the results presented here will encourage continued
research into an interesting and important geotechnical problem.
178

BIBLIOGRAPHY

1. Agnesi, V., Carrara, A., Macaluso, T., Monteleone, S., Pipitone, G., and
Sorriso-Valvo, M. (1982). Osservazioni preliminari sui fenomeni di instabilità
del versanti indotti dal sisma del 1980 nell'alta valle del Sele, Geol. Appl.
Idrogeol., v. XVII, 79-93.
2. American Geological Institute (1962). Dictionary of geological Terms,
Doubleday and Co., New York.
3. Associazione Geotecnica Italiana, A.G.I. (1977, 1979). Some Italian
experience on the mechanical characterization of structurally-complex
formations, Proc. Intl. Symp. Geot. Struct. Complex Form., Capri, 1977, v. 1,
1-20; Proc. IV Intl. Conf. Intl. Soc. Rock Mech., Montreux, 1979, v. 2.
4. Atkinson, J.H. and Bransby, P.L. (1978). The Mechanics of Soils; An
Introduction to Critical State Soil Mechanics, McGraw-Hill, London.
5. Attewell, P.B., and Farmer, I.W. (1976). Principles of Engineering Geology,
Wiley & Sons, Inc.
6. Attewell, P.B. and Taylor, R.K. (1973). Clay shale and discontinuous rock
mass studies, Final Report to European Research Office, U.S. Army Contract
Report No. DA-ERO-591-72-G0005.
7. Bagnold, R.A. (1956). The flow of cohesionless grains in fluids, Phil. Trans.
R. Soc. London, Ser. A, v.249, 235-297.
8. Bagnold, R.A. (1954). Experiments on a gravity-free dispersion of large solid
spheres in a Newtonian fluid under shear, Proc. R. Soc. London, Ser. A,
v.225, 49-63.
9. Bandis, S., Lumsden, A., and Barton, N. (1981). Experimental studies of scale
effects on the shear strength behavior of rock joints, Intl. J. Rock Mech. Min.
Sci. and Geomech. Abstr., v.18, 1-21.
10. Banks, D.C. (1971). Study of clay shale slopes, Rock Mechanics, 303-328.
11. Barton, N. (1973). Review of a new shear strength criterion for rock joints,
Eng. Geol., v.7, 287-232.
12. Barton, N. (1974). A review of the shear strength of filled discontinuities in
rock, Publication no. 105, Norwegian Geotech. Inst., Oslo, 1-38.
13. Belviso, R., Cherubini, C., Cotecchia, V., Del Prete, M., and Ferderico, A.
(1977). Effects of tectonization on the geo-mechanical behavior of multi-
179

coloured clays, Proc. Intl. Symp. Geot. Struct. Complex Form., Capri, v. 1,
71-78.
14. Bishop, A.W. (1966). The strength of soils as engineering materials, 6th
Rankine Lecture, Geotechnique, 16, No.2, 91-128.
15. Bishop, A.W. and Henkel, D.J. (1962). The Measurement of Soil Properties in
the Triaxial Test, Edward Arnold, London.
16. Binger, W.V. (1948). Analytical studies of Panama Canal slides, Proc. 2nd
Intl. Conf. Soil Mech. Found. Eng., v. 2, 54-60.
17. Bjerrum, L. (1967). Progressive failure in slopes of overconsolidated clay and
clay shales, J. Soil Mech. Found. Div., ASCE, v. 93, No. SM5, 1-49.
18. Bjerrum, L. (1972). Embankments on soft ground, Proc. ASCE Spec. Conf.
on Performance of Earth and Earth-Supported Structures, v2, 1-54.
19. Bjerrum, L. (1973). Problems of soil mechanics and construction on soft clays
and structurally unstable soils (collapsible, expansive, and others), State-of-
the-Art Report, Proc. 8th Intl. Conf. Soil Mech. Found. Eng., Moscow, v.3,
111-160.
1. Blatt, H., Middleton, G., and Murray, R. (1972). Origin of Sedimentary Rocks,
Prentice-Hall, Inc., New Jersey.
20. Blight, G.E. (1967). Horizontal stresses in stiff and fissured lacustrine clays,
Proc. 4th Regional Conf. for Africa on Soil Mech. Found. Eng., Cape Town,
95-99.
21. Blight, G.E. and Williams, A.A.B. (1971). Cracks and fissures by shrinkage
and swelling, Proc. 5th Regional Conf. for Africa on Soil Mech. Found. Eng.,
Luanda, v.1, 15-21.
22. Braddock, W.A. and Machette, M. (1976). Experimental deformation of
Pierre shale, Air Force Geophysics Final Report, Hanscom Air Force Base,
Massachusetts.
23. Brekke, T.L. (1965). On the measure of the relative potential swellability of
hydro-thermal montmorillonite clay from joints and faults in Pre-Cambrian and
Paleozoic rocks in Norway, Intl. J. Rock Mech. Min. Sci., v.2, 155-165.
24. Brekke, T.L. and Howard, T.R. (1972). Stability problems caused by seams
and faults, Proc. North Amer. Rapid Excav. Tunn. Conf., Chicago, v.1, SME-
AIME, 25-41.
180

25. Brekke, T.L. and Selmer-Olsen, R. (1965). Stability problems in underground


constructions caused by montmorillonite-carrying joints and faults, Eng.
Geol., v.1, 3-19.
26. British Standards Institute (1957). Site investigation, British Standard Code of
Practice, CP2001.
27. Brooker, E.W. (1967). Strain energy and behavior of overconsolidated soils,
Can. Geotech. J., v.4, no.3, 326-333.
28. Brooker, E.W. and Ireland, H.O. (1965). Earth pressures at rest related to
stress history, Can. Geotech. J., v.2, no.1, 1-15.
29. Bruce, R.L., and Bump, V.L. (1967). Coordination of landslide research by
South Dakota State agencies, 5th Ann. Symp. Eng. Geol. Soils Eng., Idaho
State University.
30. Bucher, F., and Kyulule, A.L. (1980). Residual shear strength of tropical soils,
7th Reg. Conf. Africa Soil Mech. Found. Eng., Accra, 83-94.
31. Burnett, A.D., and Fookes, P.G. 91974). A regional engineering geology
study of the London clay in London and Hampshire basins, Q. J. Eng. Geol.,
v. 7, 257-295.
32. Cancelli, A. (1981). Evolution of slopes in over- consolidated clays, Proc.
10th Intl. Conf. Soil Mech. Found. Eng., Stockholm, v. 3, 377-380.
33. Casagrande, A. (1949). Notes on the swelling characteristics of clay shales,
Harvard University, Cambridge, Mass., pp.16.
34. Casagrande, A., and Wilson, S.D. (1951). Effect of rate of loading on the
strength of clays and shales at constant water content, Geo-technique, 251-
263.
35. Cassel, F.L. (1948). Slips in fissured clay, Proc. 2nd Intl. Conf. Soil Mech., v.
2, 46-49.
36. Chandler, R.J. (1970). Shallow slab slide in the Lias clay near Uppingham,
Rutland, Geo-technique 20, 253-260.
37. Chandler, R.J. (1972). Discussion on Session 1, Stress-Strain Behavior of
Soils, R.H.G. Perry (ed), G.T. Foulis & Co., LTD, Oxfordshire.
38. Chandler, R.J., and Skempton, A.W. (1974). The design of permanent slopes
in stiff fissured clays, Gèo-technique, v. 24, 457-466.
39. Conlon, R.J. (1966). Landslide on the Toulnustouc River, Quebec, Can.
Geotechnique, v. 15, No. 2, 161-173.
181

40. Cotecchia, V. (1982). Phenomena of ground instability produced by the


earthquake of November 23, 1980 in southern Italy, IV Intl. Cong. Intl.
Assoc. Eng. Geol. (IAEG), New Delhi.
41. Crandell, D.R. (1952). Landslides and rapid- flowage phenomena near Pierre,
South Dakota, Econ. Geol., v. 47, 548-581.
42. Crosby, W.O. (1893). The origin of intersecting and parallel joints,
Technology Quarterly, v. VI, 230-236.
43. D'Argenio, B., Pescatore, T., and Scandone, P. (1975). Structural pattern of
the Campania- Lucania Apennines, Geology of Italy, C. Squyres (ed), The
Earth Science, Soc. of the Lybyan Arabic Republic, 317-328.
44. Davidson, J.L. (1977). A quasi-preconsolidation clay model, Proc. 9th Intl.
Conf. Soil Mech. Found. Eng., Tokyo, v.1, 75-79.
45. Deen, R.C. (1981). The need for a schema for the classification of transitional
(shale) materials, Geot. Test. J., v. 4, No. 1, 3-10.
46. Deere, D.U. and Patton, F.D. (1971). Slope stability in residual soils, Proc.
4th Pan. Conf., San Juan, SMFE, 87-170.
47. Delabeche, H. (1844). Discussion on paper by C.H. Gregory (1844).
48. D'Elia, B. (1980). Processi di ammorbidimento di un'argilla marnosa con
struttura a scaglia, XIV Conv. Naz. di Geot., Firenze, 409-417.
49. Denness, B. (1969). Fissure and related studies in selected Cretaceous rocks
of S.E. England, PhD. Thesis, Imperial College, London.
50. Deo, P. (1972). Shales as embankment materials, Joint Highway Research
Project, Purdue University, W. Lafeyette, Ind..
51. Duncan, J.M., and Dunlop, P. (1968). Slopes in stiff-fissured clays and shales,
J. Soil Mech. Found. Div., ASCE, v. 95, No. SM2, 467-492.
52. Erskine, C.F. (1973). Landslides in the vicinity of the Fort Randall Reservoir,
South Dakota, USGS Prof. Paper 675, pp. 65.
53. Esu, F. (1966). Short term stability of slopes in unweathered jointed clays,
Géotechnique, v.16:4, 321-328.
54. Esu, F. (1977). Behavior of slopes in structurally complex formations, Proc.
Intl. Symp. Geot. Struct. Complex Form., Capri, v. 2, 292-304.
55. Esu, F., and Grisolia, M. (1977). Creep characteristics of an overconsolidated
jointed clay, Proc. 9th Intl. Conf. Soil Mech. Found. Eng., Tokyo, v. 1, 93-
100.
182

56. Evangelista, A., Paparo Filomarino, M., and Pellegrino, A. (1977). On the
mechanical behavior of variegated clay shales of Irpinia, Proc. Intl. Symp.
Geot. Struct. Complex Form., Capri, v. 1, 229-237.
57. Fairhurst, C. (1964). On the validity of the Brazilian test for brittle materials,
Intl. J. Rock Mech. Min. Sci., v.1, 535-546.
58. Farrington, J.J. (1983). On the characterization of rock joint roughness, M.S.
thesis, University of Colorado, Boulder, Colorado.
59. Fenelli, G.B., Paparo Filomarino, L., Picarelli, L., and Rippa, F. (1981). Sulla
resistenza di picco delle argille varicolori, Atti della Riun. Gruppo Naz. di
Coord. per gli Studi di Ing. Geot., Roma.
60. Fenelli, G.B., Paparo Filomarino, M., Picarelli, L., and Rippa, F. (1982).
Proprietà fisiche e meccaniche di argille varicolori dell'Irpinia, Rivista Ital. di
Geot., Anno XVI, No. 3, 110-124.
61. Fleming, R.W., Spencer, G.S., and Banks, D.C. (1970). Empirical study of
behavior of clay shale slopes, U.S. Army Eng. Nuclear Cratering Group Tech.
Report No. 15, v. 1 & 2.
62. Fookes, P.G. (1965). Orientation of fissures in stiff overconsolidated clay of
Siwalik system, Géotechnique, v.15, 195-206.
63. Franklin, J.A., and Chandra, R. (1972). The slake- durability test, Intl. J. Rock
Mech. Min. Sci., v. 9, 325-341.
64. Gamble, J.C. (1971). Durability-plasticity classification of shales and other
argillaceous rocks, Ph.D. Thesis, Geology, University of Illinois, Urbana.
65. Geuze, E.C.W.A. (1948). Horizontal earth pressures against a row of piles,
Proc. 2nd Intl. Conf. Soil Mech. Found. Eng., Rotterdam, v. 4, 135-140.
66. Geuze, E.C.W.A. (1960). The effect of time on the shear strength of clays,
Presentation at ASCE Conv., New Orleans.
67. Ghaboussi, J., Wilson, E.L., and Isenberg, J. (1973). Finite elements for rock
joints and interfaces, J. Soil Mech. Found. Div., ASCE, v.99, SM10, 833-848.
68. Goodman, R.E. (1966). On the distribution of stresses around circular tunnels
in non- homogeneous rocks, Proc. 1st Cong. Intl. Soc. Rock Mech., Lisbon,
v.2, 249-255.
69. Goodman, R.E. (1970). The deformability of joints, ASTM Spec. Tech. Publ.
477, 174-196.
2. 71. Goodman, R.E. (1974). The mechanical properties of joints, Proc. 3rd Cong.
Intl. Soc. Rock Mech., Denver, v. 1-A, 127-140.
183

72. Goodman, R.E. and Dubois, J. (1978). Duplication of dilatancy in analysis of


jointed rocks, J. Soil Mech. Found. Div., ASCE, v.98, SM4, 399-422.
73. Goodman, R.E., Taylor, R.L., and Brekke, T.L. (1968). A model for the
mechanisms of jointed rock, J. Soil Mech. Found. Div., ASCE, v.94, SM3,
637-659.
74. Graham, J. and Au, V.C.S. (1985). Effects of freeze-thaw and softening on a
natural clay at low stresses, Can. Geotech. J., v.22, No.1.
75. Graham, J. and Li, E.C.C. (1985). Comparison of natural and remolded plastic
clay, J. Geotech. Eng., ASCE, v.111, No.7, 865-881.
76. Gregory, C.H. (1844). On railway cuttings and embankments; with an account
of some slips in the London clay, on the line of the London and Croydon
Railway, Min. Proc. Inst. C.E., v. 3, 135-145.
3. 77. Hampton, M.A. (1979). Buoyancy in debris flows, J. Sed. Pet., v.49, no.3,
735-758.
77. Henkel, D.J. (1957). Investigations of two long- term failures in London clay
slopes at Wood Green and Northolt, Proc. 4th Intl. Conf. Soil Mech. Found.
Eng., London, v. 1, 315-320.
78. Hodgson, R.A. (1961). Regional study of jointing in Comb Ridge - Navajo
Mountain area, Arizona and Utah, Bull. Amer. Assoc. Petrol. Geol., v.45:1, 1-
38.
79. Hutchinson, J.N. (1970). A coastal mudflow on the London clay cliffs at
Beltinge, North Kent, Gèotechnique, v. 20, 412-438.
80. Hvorslev, M.J. (1937). On the strength properties of remolded cohesive soils,
Danmarks Naturvidenskabelige Samfund, Ingeniörvidenskabelige Skrifter,
Series A. No. 45, Copenhagen.
81. Ingram, R.L. (1953). Fissility of mudrocks, Bull. Geol. Soc. Amer., v. 64, No.
8, 869-878.
82. Insley, A.E., Chatterji, P.K., and Smith, L.B. (1977). Use of residual strength
for stability analyses of embankment foundations containing preexisting failure
Surfaces, Can. Geot. J., v. 14, 408-428.
83. Japelli, R., Liguori, V., Umiltà, G., and Valore, C. (1979). A survey of
geotechnical properties of a stiff highly fissured clay, The Geotechnics of
Structurally Complex Formations, Capri, v.2, 91-106.
184

84. Johnson, S.J. (1969). Report of Chairman of Specialty Session no. 10,
Engineering properties and behavior of clay shales, Proc. 7th Intl. Conf. Soil
Mech. Found. Eng., Mexico City, v. 3, p. 483.
85. Kallstenius, T. (1963). Studies on clay samples taken with standard piston
sampler, Proc. No. 21, Swedish Geotech. Inst., Stockholm.
86. Kenney, T.C., Moum, J., and Berre, T. (1967). An experimental study of the
bonds in a natural clay, Proc. Geotech. Conf. on Shear Strength Prop. of
Natural Soils and Rocks, Oslo, v.1, p.65.
87. Kindle, E.M. (1923). A note on mud crack and associated joint structure,
Amer. J. Sci., 5th series, v.23, 233-234.
88. Krahn, J., Johnson, R.F., Frelund, D.G., and Clifton, A.W. (1979). A highway
cut failure in Cretaceous sediments at Maymont, Saskatchewan, Can. Geot. J.,
v. 16, 703-715.
89. Lachenbruch, A.H. (1962). Mechanics of thermal contraction cracks and ice-
wedge polygons in permafrost, Spec. Paper No. 70, Geol. Soc. Amer., 1-69.
90. Ladanyi, B. and Archambault, G. (1970). Simulation of shear behavior of a
jointed rock mass, Proc. 11th Symp. Rock Mech., AIME, 105-125.
91. Ladanyi, B. and Archambault, G. (1977). Shear strength and deformability of
filled indented joints, The Geotechnics of Structurally Complex Formations,
Capri, v.1, 317-326.
92. LeConte, J. (1882). Origin of jointed structure in undisturbed clay and marl
deposits, Amer. J. Sci., 3rd series, v.23, 233-234.
93. Leonards, G.A. and Ramiah, B.K. (1960). Time effects in the consolidation of
clays, ASTM Spec. Tech. Pub. 254, 116-130.
94. Leussink, H. and Muller-Kirchenbauer, H. (1967). Determination of the shear
strength behavior of sliding planes caused by geological features, Proc.
Geotech. Conf. Shear Strength Properties of Natural Soils and Rock, Oslo,
v.1, 131-137.
95. Lowe, J. and Johnson, T.C. (1969). Use of back pressure to increase degree
of saturation of triaxial test specimens, Research Conf. Shear Strength of
Cohesive Soils, Boulder, Colorado, ASCE, 819-836.
96. Manfredini, G., Martinetti, S., Ribacchi, R., Santoro, V.M., Sciotti, M., and
Silvestri, T. (1981). An earthflow in the Sinni Valley (Italy), Proc. 10th Intl.
Conf. Soil Mech. Found. Eng., v. 3, 457-462.
185

97. Maugeri, M., Motta, E., and Sorriso Valvo, M. (1982). The Senerchia
landslide triggered by the 23 November 1980 earthquake, 4th Intl. Conf.
IAEG, New Delhi.
98. Matheson, D.S., and Thomson, S. (1973). Geological implications of valley
rebound, Can. J. Earth Sci., v. 10, 961-978.
99. McClure (1980). Ph.D Thesis, University of California, Berkley.
100. Middlebrooks, T.A. (1942). Fort Peck slide, Trans. ASCE, v. 107, p. 723.
101. Mitchell, J.K. (1976). Fundamentals of Soil Behavior, John Wiley & Sons,
Inc., NY.
102. Morgenstern, N.R., and Eigenbrod, K.D. (1974). Classification of argillaceous
soils and rock, J. Geot. Eng. Div., v. 100, 1137-1156.
103. Nichols, T.C., Jr. (1980). Rebound, its nature and effect on engineering
works, Q. J. Eng. Geol. London, v. 3, 133-152.
104. Ortolani, F., and Torre, M. (1981). Guida all' escursione nell'area interessata
del terremoto del 23 novembre 1980, Seduta Soc. Geol. Ital., Napoli, pp. 78.
105. Palladino, D.J., and Peck, R.B. (1972). Slope failures in an overconsolidated
clay, Seattle, Washington, Gèotechnique, v. 22, 563-595.
106. Patton, F.D. (1966). Multiple modes of shear failure in rock and related
materials, Ph.D. dissertation, University of Illinois, Urbana, Illinois.
107. Peterson, R. (1954). Studies of Bearpaw shale at a dam site in Saskatchewan,
Proc. ASCE, New York, v.80, separate no. 476.
108. Peterson, R., Jasper, J.L., Rivard, P.J., and Iverson, N.L. (1960). Limitations
of laboratory shear strength in evaluating stability of highly plastic clays,
ASCE Res. Conf. on Shear Strength of Cohesive Soils, Boulder, 765-791.
109. Picarelli, L. (1981). Sulla resistenza residua di argille varicolori dell'Italia
meridionale, Atti del XIV Conv. Ital. di Geot., Firenze.
110. Pratt, H.R., Black, A.D., and Brace, W.F. (1974). Friction and deformation of
jointed quartzite diorite, Proc. 3rd Cong. of Intl. Soc. Rock Mech., Denver, v.
I.A, 243-317.
111. Price, N.J. (1966). Fault and Joint Development in Brittle and Semi-Brittle
Rock, Pergamom Press, New York, pp.176.
112. Rippa, F., and Picarelli, L. (1977). Some considerations on index properties of
southern Italy shales, Proc. Intl. Symp. Geot. Struct. Complex Form., Capri,
v. 2, 401-406.
186

113. Roberds, W.J. and Einstein, H.H. (1978). Comprehensive model for rock
discontinuities, J. Geotech. Eng. Div., ASCE, v.104, GT5, 553-569.
114. Rocha, M. (1964). Mechanical behavior of rock foundation in concrete dams,
Trans. 8th Cong. on Large Dams, Edinburgh, v.1, 785-831.
115. Rodine, J.D. (1974). Analysis of the mobilization of debris flows, Final Report
to U.S. Army Research Office, Grant no. DA-ARO-D-31-124- 71-G158,
pp.226.
116. Rodine, J.D. and Johnson, A.M. (1976). The ability of debris, heavily
freighted with coarse clastic materials, to flow on gentle slopes,
Sedimentology, v.23, 213-234.
117. Rowe, P.W. (1962). The stress-dilatancy relationship for static equilibrium of
an assembly of particles in contact, Proc. R. Soc London, ser. A, v.269, 500-
527.
118. Sandroni, S.S. (1977). The strength of London clay in total and effective
stress terms, Ph.D. Thesis, University of London.
119. Schuster, R.L. (1965). Minors structures in the London clay, Thesis, Imperial
College, London.
120. Scully, J. (1973). Landslides in the Pierre shale in Central South Dakota,
South Dakota Dept. Highways Study No. 635(67), Final Report, pp. 707.
121. Serafim, J.L. and Lopez, J.J.B. (1961). In-situ shear tests and triaxial tests of
foundation rocks of concrete dams, Proc. 5th Cong. Soil Mech Found. Eng.,
Paris, v.1, 533-539.
122. Shamburger, J.H., Patrick, D.M., and Lutten, R.J. (1975). Survey of problem
areas and current practices, Design and Construction of Compacted Shale
Embankments, v. 1, Federal Highway Administration, FHWA-RD-75-61,
Washington, D.C..
123. Sinclair, S.R. and Brooker, E.W. (1967). The shear strength of Edmonton
shale, Proc. Geotech. Conf. Oslo, v.1, 295-299.
4. 125. Singh, R., Henkel, D.J., and Snagrey, D.A. (1973). Shear and Ko
swelling of overconsolidated clay, Proc. 8th Intl. Conf. Soil Mech. Found. Eng.,
Moscow, v. 1.2, 367-376.
126. Skempton, A.W. (1942). The failure of a retaining wall at Kensal Green, BRS
Soil Mechanics Record No. 3, Loc. 110.
187

127. Skempton, A.W. (1948). The rate of softening in stiff fissured clays, with
special reference to London clay, Proc. 2nd Intl. Conf. Soil Mech., v. 2, 50-
53.
128. Skempton, A.W. (1953). Soil mechanics in relation to geology, Proc. of teh
Yorkshire Geol. Soc., v.29:1, 33-62.
129. Skempton ,A.W. (1961). Horizontal stresses in an overconsolidated Eocene
clay, Proc. 5th Intl. Conf. Soil Mech. Found. Eng., v.1, p.351.
130. Skempton, A.W. (1964). Long-term stability of clay slopes, Fourth Rankine
Lecture, Gèotechnique, v. 14, p. 77.
131. Skempton, A.W. (1977). Slope stability of cuttings in brown London clay,
Proc. 9th Intl. Conf. Soil Mech. Found. Eng., Tokyo, v. 3, 261-270.
132. Skempton, A.W., and Delory, F.A. (1957). Stability of natural slopes in
London clay, Proc. 4th Inst. Conf. Soil Mech., London, v. 2, 378-381.
133. Skempton, A.M., and Hutchinson, J.N. (1969). Stability of natural slopes and
embankment foundations, Proc. 7th Intl. Conf. Soil Mech. Found. Eng., State-
of-the-Art Volume, Mexico City, p. 291.
134. Skempton, A.W. and La Rochelle, P. (1965). The Bradwell slip; a short-term
failure in London clay, Géotechnique, v.15, no.3, p.221.
135. Skempton, A.W. and Northerly, R.D. (1952). The sensitivity of clays,
Géotechnique, v.3:1, 100-106.
5. 136. Stimpson, B. and Walton, G. (1970). Clay mylonites in English coal
measures, Proc. 1st Cong. Intl. Assoc. Eng. Geol., Paris, v.2, 1388-1393.
136. Sture, S. and Ko, H.Y. (1978). Strain-softening of brittle geologic materials,
Intl. J. Num. Analyt. Meth. Geomech., v.2, 237-253.
137. Sture, S. (1976). Strain softening behavior of geologic materials and its effects
on structural response, Ph.D. dissertation, University of Colorado, Boulder,
Colorado.
138. Tavenas, F. and Leroueil, S. (1977). Effects of stresses and time on yielding of
clays, Proc. 9th Intl. Conf. Soil Mech. Found. Eng., Tokyo, v.1, 319-326.
139. Terzaghi, K. (1936). Stability of slopes of natural clay, Proc. Intl. Conf. Soil
Mech., Harvard, v. 1, 161-165.
140. Terzaghi, K. (1961). Discussion on "Horizontal stresses on an
overconsolidated Eocene clay, Proc. 5th Intl. Conf. Soil Mech. Found. Eng.,
v.3, 144-145.
188

141. Terzaghi, K., and Peck, R.B. (1967). Soil Mechanics in Engineering Practice,
2nd ed., John Wiley & Sons, NY.
142. Thomson, S., and Hayley, D.W. (1975). The Little Smoky Landslide, Can.
Geot. J., v. 12, 379-392.
143. Tourtelot, H.A. (1962). Preliminary investigation of the geological setting and
chemical composition of the Pierre shale, Great Plains Region, USGS Prof.
Paper 390, pp. 72.
144. Tulinov, R. and Molokov, L. (1976). Role of joint filling material in shear
strength of rocks, Symp. Intl. Soc. Rock Mech., Nancy, paper II-24.
145. Twenhofel, W.H. (1950). Principles of Sedimentation, McGraw-Hill, New
York.
146. Underwood, L.B. (1967). Classification and identification of shales, J. Soil
Mech. Found. Div., ASCE, v. 93, 97-116.
147. Uriel, S. and Serrano, A.A. (1973). Geotechnical properties of two collapsible
volcanic soils of low bulk density at the site of two dams in Canary Islands
(Spain), Proc. 8th Intl. Conf. Soil Mech. Found. Eng., Moscow, v.2.2, 257-
264.
148. Vallejo, L.E. (1979). An explanation for mud- flows, Géotechnique, v.29,
no.3, 351-354.
149. Vallejo, L.E. (1980). Mechanics of mudflow mobilization in low-angled clay
slopes, Eng. Geol., v.16, 63-70.
150. Ward, W.H., Samuels, S.G., and Butler, M.E. (1959). Further studies of the
properties of London clay, Géotechnique, v.9, 33-58.
151. Weaver, C.E. (1967). The significant of clay minerals in sediments,
Fundamental Aspects of Petroleum Geochemistry, B. Nagy and U. Colombo
(ed), New York, Elsevier Pub. Co., 37-76.
152. Weisner, W.R. (1969). residual strength of over- consolidated clay shales,
M.Sc. Thesis, Dept. Civil Eng., Univ. of Saskatchewan, Saskatoon.
153. Wentworth, C.K. (1922). Geology, W.C. Putnam, Oxford University Press,
Inc., NY.
154. White, W.A. (1961). Colloid phenomena in sedimentation of argillaceous
rocks, J. Sed. Petrology, v.31:4, 560-570.
155. Widger, R.A. and Frelund, D.G. (1979). Stability of swelling clay
embankments, Can. J. Geotech., v.16, 140-151.
189

APPENDIX A

CLASSIFICATION SCHEMES

The general characteristics of clay shales include (1) prestressed (i.e. highly
overconsolidated), (2) commonly fissured with slickensides often present, (3) strong
diagenetic bonding, (4) tendency to slake when rewetted after drying, and (5) high
swelling pressures in the presence of water. They have been referred to in the literature
as "stiff", "fissile", "intact", "compacted", or "brittle" clays, as well as "soil-like shale".
Beyond this general description of clay shales, the classification of these materials has
become complicated and confusing. Numerous classification schemes for argillaceous
materials have been proposed, and most are reviewed by Shamburger, Patrick, and
Cutten (1975) and Deen (1981), as well as in the following section.

Many of these classifications are geological and depend on such properties as quartz
content, grain size, color, and the "degree of compaction". Although these provide
important information regarding geological history of these materials, such
classifications can be highly deceptive when concerned with engineering behavior. This
is particularly evident when evaluating the behavior of clay shales. For example, Pierre
shale can be described as a grey-black, highly-compacted shale with no visible tendency
toward fissility on a fine scale, whereas the argille varicolori might be described as a red
or green, highly-tectonized, scaly clay. Yet these two materials have similarities in their
engineering behavior which would classify them together as clay shales.

Most engineering classification schemes for argillaceous materials employ such material
properties as grain size, shear strength derived from simple tests, overconsolidation
ratio, and Atterberg limits. These classification schemes have proved useful for
normally consolidated and slightly overconsolidated clays, but have generally not been
adequate for transitional materials between highly overconsolidated clays and shales.
As will be discussed below, the major deficiency of most engineering classification
schemes of argillaceous materials is the absence of an adequate time factor. For
transitional materials, such as clay shales, it is important to not only consider how these
materials behave at the present, but also how they will behave within a reasonable
engineering time frame.
190

Geological Classification
Most classification schemes of crustal materials are either geological or engineering
oriented, depending on the intended application. Although geological classifications
have proven useful for many practical problems the major objective of these schemes is
the determination of the geological history of the deposits. With this in mind, the
various geological classification schemes for argillaceous materials are reviewed below.

Initial geological classifications were developed primarily on the basis of grain size. The
scheme of Wentworth (1922) arbitrarily set 0.0625mm as the boundary between
argillaceous materials (shale or mudstone) and the remaining clastic, or fragmental,
sedimentary rocks. Ingram (1953) subdivided clayey materials based on percentages of
silt and clay components, and on their breaking characteristics (Table A.1). In this
scheme, the presence of fissility distinguishes "shales" from "stone", while the prefixes
"clay", "silt", or "mud" are derived from the relative percentages of the grain size
components. Therefore, such terms as claystone, siltstone, and clay shale began to be
entrenced in the literature.

Folk (1968) clarified Ingram's scheme by refining "mudstone" as an argillaceous


materials with subequal amounts of clay and silt (Table A.2). The classification scheme
of Gamble (1972) is essentially the same as those of Folk and Ingram, except that the
terms "clay shale" and "silt shale" have been changed to "clayey shale" and "silty shale".
Although this change may seem insignificant, the term "clayey shale" does help to
distinguish a clay-rich shale from a "clay shale" which, in engineering usage, implies
certain engineering behavior and not simply a fissile rock which is rich in clay content.

The classification of Underwood (1967) is the first attempt to divide shales into "soil-
like" shale and "rock-like" shale (Fig. A.1). Although this scheme is essentially
geological, it begins to approach the concept of a soil-like and rock-like behavior.
191

Table A.1. Geological classification of mudrocks by Ingram (1953).

Table A.2. Geological classification of mudstone (Folk, 1968).


192

Figure A.1. Classification scheme of Underwood (1967).

However, the division between these two groups is poorly defined. Similarly,
Skempton and Hutchinson (1969) attempt to crudely relate geological origin of
materials to their potential engineering behavior (Table A.3). The usefulness of this
scheme for purposes other than for providing a general understanding of possible
relationships is quite limited.
193

Table A.3. Classification scheme of Skempton and Hutchinson (1969).


194

Although the classification schemes above can provide some useful information for
engineers, they are generally inadequate for assessing potential engineering behavior of
argillaceous materials. Regardless, the above review traces the use of the term "clay
shale" in the geological literature to describe a fissile rock, rich in clay-sized
components. As will be discussed below, this use of the term "clay shale" is not
synonymous with its use in engineering literature.

Engineering Classification
The objective of an engineering classification scheme is to categorize geological
materials according to their potential engineering behavior. In this regard, an
engineering classification is often oriented toward specific applications. This tends to
cause some confusion among investigators when schemes employed for one application
are considered valid for all applications.

Classification of argillaceous materials for engineering purposes has been particularly


difficult. Many of these difficulties have resulted because of the transitional nature of
some of these materials. As will be discussed further, these transitional materials create
further confusion for many geotechnical engineers who are accustom to viewing a
material as either a rock or a soil, but not as a material that can have proper ties of
both. In addition, few engineering classification schemes account for the potential
changes in material behavior which can occur in a relatively short time in many of these
deposits.

Terzaghi (1936) divided clays based on stiffness and the presence or absence of
fissures:
(a) soft clays free from joints and fissures,
(b) stiff clays also free from joints and fissures,
(c) stiff fissured clays.
Bjerrum (1967) proposed an overlapping three-fold classification, based on bond
strength and extending up to shale materials:
(a) overconsolidated clays (i.e. overconsolidated clays with weak or no bonds),
(b) clay shales (i.e. overconsolidated plastic clays with well-developed diagenetic
bonds), and
(c) shales (i.e. overconsolidated plastic clays with strongly developed diagenetic
bonds).
195

These two schemes have significant, but poorly distinguished overlap between them
creating some confusion of terms. Further confusion has developed from the use of the
British Standard Institute classification, which uses similar terms based on consistency
or strength:

Consistency Field indication Strength


Very stiff Brittle or very tough >150 kN/m2
Stiff Cannot be molded in fingers 75 - 150
Firm Molded in fingers by firm pressure 40 - 75
Soft Easily molded in fingers 20 - 40
Very soft Extrudes between fingers <20 kN/m2

Since the introduction of these classifications, more ambiguity in terminology has


resulted from the liberal use of the terms "overconsolidated shale" (Johnson, 1969;
Fleming et al, 1970), and "stiff, fissured clay" (Chandler, 1970) to indicate a weakly-
bonded shale. This inconsistency in terminology has been most pronounced and most
confusing for the argillaceous materials which are transitional between normally
consolidated clays and intact shales. Most of these difficulties result from inability’s to
adequately account for potential changes in material behavior with time. Some
investigators have attempted to account for this effect by including a "durability", or
slaking factor in their classification schemes.

Gamble (1971) carried out extensive investigations on the durability of shales of all
ages, location, and consistency. Based on correlation’s of material properties, such as
water content, liquid limit, dry density, etc., he determined that these materials could
best be grouped according to a relationship between a two- cycle slake durability index
and their plastic index (Fig. A.2). A clay shale is generally considered as having high
plasticity and a low slake durability index. However, Gamble does not attempt to relate
his classification scheme to established terminology, and concludes that more work is
necessary to correlate laboratory results with field behavior.
196

Figure A.2. Classification scheme of Gamble (1971), based on the relationship


between slaking durability and plastic index.

Similarly realizing the importance of shale deterioration, Deo (1972) proposed a


scheme based entirely on shale durability. Three tests, all of which measure shale
durability (i.e. slaking, slake durability, and sulfate soundness), were performed on
various shales from Paleozoic deposits in Indiana. Using indices derived from these
three tests, Deo categorized Paleozoic argillaceous materials into soil-like shales, two
types of intermediate shales, and rock-like soils (Fig. A.3). The most significant aspect
of this scheme is the classification of the materials according to the their susceptibility
to deterioration rather than the initial state of the material.

Morgenstern and Eigenbrod (1974) were the first to attempt to combine earlier
classification schemes based on initial properties with those schemes based on
durability. They present two classification schemes, one based entirely on the slaking
characteristics (i.e. the rate of slaking versus the amount of slaking), and a more
significant scheme based on the undrained shear strength, strength loss after softening,
changes of water content after softening, and the time of softening
197

Figure A.3. Classification scheme of Deo (1972), based entirely on resistance


to slaking.

(Fig. A.4). The major importance of this scheme is the emphasis on changes of strength
and water content with softening.

Unfortunately, this scheme first divides argillaceous material into either soil or rock
based on three potentially conflicting properties, the undrained shear strength, the
degree of strength loss after softening, and the degree of change in water content after
softening. Only after this division are slaking characteristics used to determine if any of
the "soil-like" materials are clay shales. According to this scheme, Pierre shale could be
classified as "rock-like" according to its initial strength characteristics, or as "soil-like"
based on its response to softening. In addition, the Pierre shale, although rock-like in
initial strength, slakes completely to a soft mud with only one cycle of the slake-
durability test.

The Transitional Nature of Clay Shale


The term "clay shale" has become widely adopted in engineering practice. Yet, the
discussion above illustrates that the term has also become poorly defined and nebulous.
In its original meaning, the term "clay shale" did not necessarily imply that the material
was particularly troublesome. However, the term "clay shale" has
198

Figure A.4. Two part classification scheme of Morgenstern and Eigenbrod


(1974).
199

become increasing used within engineering practice to imply a stiff to rock-like


argillaceous material that is susceptible to significant detrimental changes in its
engineering behavior as a result of its interactions with water.

For example, Fleming et al (1970) refer to clay shales as clayey or silty,


overconsolidated, sedimentary materials, which may or may not be slightly cemented.
They continue by describing clay shale materials as tending to slake when exposed to
cyclic wetting and drying. Materials cemented to the extent that they do not slake when
exposed to cyclic wetting and drying are termed siltstone or claystone ....

Furthermore the relationship of deterioration to the clay shales of Great Britain is


illustrated by Attewell and Farmer (1976):

In a freshly excavated state, the Lias would be considered a true clay shale. Where it
outcrops ... it tends to degrade very readily into a clay/mud flow.... Such dessication
fractures, together with any jointing and laminations ... can rapidly reduce most clay
shales ... to an aggregate in a few months, the actual length of time being a function
of wet-dry (slaking) cyclic frequency together with the compositional mineralogy and
fabric of the rock.

It must be remembered that the purpose of any engineering classification scheme is to


provide terms which aid the user in distinguishing materials which have similar
engineering properties. The more recent classification schemes for argillaceous
materials have attempted to account for the important role of durability by including
slaking factors. However, these schemes have failed to recognize a unique property of
clay shales.

Other engineering materials are classified according to the engineering properties that
they presently exhibit. Yet, a "clay shale" is unique not in its present properties, but
rather in its potential for significant deterioration of these properties as a result of
interactions with water. Classification schemes to date have failed to recognize that, as
the term now implies, a "clay shale" can encompass a stiff clay, such as the London
clay, or a clayey shale, such as the Pierre shale. The terms "stiff clay" and "clayey shale"
define these materials according to their present engineering properties. However, the
200

London clay and Pierre shale can be further classified as "clay shales" based on their
potential for significant reduction of strength upon wetting, or wetting and drying.

A clay shale is therefore defined by the present author as a stiff clay, or shale, which
can undergo significant deterioration of its engineering properties as a result of
interactions with water. This definition recognizes that the term "clay shale" implies a
transitional material rather than a material with well-defined initial properties.

The classification scheme proposed here in Fig. A.5 employs previous classification
schemes, while placing clay shales within the proper perspective. According to this
scheme, the Pierre shale is an argillaceous material, a mudstone, and a clayey shale, as
well as a clay shale. Similarly, the London clay is a stiff, fissured clay because of its
present engineering properties, and a clay shale because of its potential for rapid
deterioration.

Of course, an adequate factor is still needed to determine whether a stiff clay or clayey
shale could "significantly deteriorate as a result of interaction with water". Like
previous schemes, this factor should most likely be a measure of the slaking potential of
the material. However, as discussed in the main body of this dissertation, the author

Figure A.5. Modified classification scheme of the present author, showing clay
shale as a unique class of argillaceous materials.
201

questions whether the present practices of measuring slaking potential are adequate for
determining the realistic response of these materials to slaking conditions in the field.

Summary
The term "clay shale" has become widely adopted, but increasingly nebulous in
meaning. This is partly the result of confusion between the use of "clay shale" in the
geological sense of a shale with a predominant clay fraction, and the use of the term in
the engineering sense of a clayey material that is intermediate between a firm clay and
schemes to date have proved inadequate in establishing a widely accepted distinction
between clay shales and other argillaceous materials.

The term "clay shales" has become increasingly used in the engineering practice to
imply a stiff to rock-like argillaceous material that is susceptible to relatively rapid
degradation with time. The most recent schemes have recognized the need for an
adequate temporal factor in the classification of clay shales. This factor is typically
some measure of the slaking characteristics of the material. As discussed within this
paper, there are some inadequacies in the present procedures for determining the slake
durability. Still, even considering these inadequacies, slake durability provides a
reasonable temporal factor for the classification of clay shales.

However, confusion has remained because the classification schemes to date have made
futile attempts to distinguish clay shales from stiff clays or clayey shales. Instead, it
must be recognized that a clay shale can be a "stiff clay" or "clayey shale" based on its
present properties, and still be a "clay shale" based on its tendency to rapidly degrade
with time. The author has proposed a modified engineering classification scheme for
argillaceous materials, which places clay shales within this proper perspective.

The slaking tests employed in the present classification schemes of argillaceous


materials include (1) the modified jar slaking test, (2) the slake durability test, (3) the
"one-dimensional free swell test", and (4) the sulfate soundness test. In the modified jar
slaking test (Moriwaki, 1974), an undisturbed or dried sample is placed on a wire mesh,
which is then lowered into a jar of water. The slaking potential is indicated by the loss
of weight as the material crumples and falls through the mesh. The slake durability test
of Franklin and Chandra (1972) is performed by rotating six 40 to 60 gram samples in a
wire mesh drum, which is partly immersed in water. The weight percentage of material
202

remaining in the drum after one cycle (200 revolutions in ten minutes) defines the
"slake durability index". The sulfate soundness test is similar to the above tests, except
that the previously dried samples are wetted in a sodium or magnesium sulfate solution.
The percentage of material retained on a 3/8" sieve indicates the sulfate soundness
index.

The slaking factor of Morgenstern and Eigenbrod (1974) is unique in that it is based on
the one-dimensional free swell of a laterally confined sample. In these tests, the change
of height, and therefore the change of water content, were measured as a function of
wetting and drying cycles. Increased swelling is assumed to indicate progressive slaking
within the specimen.

The modified jar slaking, slake durability, and sulfate soundness tests described above,
all provide very useful information on the relative "weatherability" of various
argillaceous materials. However, the extent to which these tests significantly model
conditions existing in the field is limited. All of these tests measure the amount of
slaking under conditions of zero confining pressure. The review of slaking processes in
Chapter IV, discusses several different mechanisms of slaking, each caused by various
magnitudes of disruptive stresses. It is conceivable that slaking forces exhibited by
some materials may be balanced by very low confining pressures. In such materials,
slaking would only affect the very top surface of deposits in the field. In addition, the
slake durability test employs a tumbling factor, which generally does not exist in the
field.

More important, however, is the fact that none of these tests measure the relative
reduction of shear strength that occurs as these materials undergo deterioration in the
field. Although these tests do measure the important tendency of a material to degrade,
they do not necessarily measure the relative reduction of strength due to degradation.

Present evidence regarding the relative importance of weathering of clay shales in the
field is inconclusive. There is debate as to whether the long term behavior of clay shales
is controlled predominantly by material deterioration or by the reduction of negative
pore pressures with time. The modes by which weathering affects the engineering
behavior of clay shales is also not well understood. The purpose of this dissertation is
to investigate the role of slaking in altering the strength and stress-strain response of
203

clay shales. Such research is necessary before a classification factor can be developed
which will adequately distinguish the unique and troublesome clay shales.
204

APPENDIX B

PROCESSES ACTING DURING THE FORMATION


OF CLAY SHALES

The unique characteristics of clay shales can be attributed to a combination of several


processes occurring during deposition and during subsequent unloading. Several of the
factors responsible for the present behavior of clay shales, such as lithology,
compaction, and bonding, were developed prior to the unloading of the original
material. This appendix therefore examines the geological development of a clay shale
deposit. It is important to develop a firm understanding of the physico-chemical
processes acting on clay shale before attempting to account for the unique behavior of
clay shale deposits.

Formation of Clay Shales


The behavior of clay shales is closely related to their lithology, which includes the
mineralogy and the amount of clay fraction present, as well as the degree of
compaction, and the extent and strength of internal bonding. An increase in the clay
fraction and in the percentage of montmorillonite results in increases in the plasticity
and swelling potential and a decrease in the residual strength of clay shales. The
mineralogy and grain size distribution of any deposit is originally controlled by the
mineralogy and climatic conditions existing in the source area from which the material
is eroded and transported, and by the energy regime of the transporting agent.

Most clay shales were deposited in a marine environment, although some are lacustrine
(i.e. lake) deposits. Most of the present clay shales can further be characterized by
relatively high amounts of montmorillonite. Several studies presented by Blatt et al
(1972), indicate that the relative amounts of montmorillonite, illite, and kaolinite in
modern marine deposits, can be directly related to the climatic environment of adjacent
shores. As temperature and rainfall increase, illite is converted to montmorillonite, with
greater rainfall and temperature, montmorillonite might be degraded further to
kaolinite. Therefore, kaolinite is abundant where tropical rivers empty into the ocean,
whereas an abundance of montmorillonite reflects more moderate climatic conditions.
205

However, the relative amount of montmorillonite in an area can be greatly increased by


nearby volcanic activity, and most deposits rich in montmorillonite can be attributed to
such a source. In particular, highly plastic bentonitic layers, such as those found in the
clay shales of the upper Missouri River basin, are the result of in-situ alteration of
volcanic ash.

In addition, chemical analyses on thousands of rocks in the U.S. indicate that the
relative abundance’s of the illite, kaolinite, and smectite (i.e. montmorillonite) vary with
the age of the rock (Weaver, 1967). Such variations can be attributed to changes in the
paleoenvironment or to diagenetic processes which act to change the material after
deposition. As shown in Fig. B.1, expandable clays have been relatively abundant from
100 million year ago to the present. This may be a major factor accounting for the fact
that most of the clay shales in the U.S. are Upper Cretaceous or younger.

An abundance of clay sized particles in a deposit can result from a similar abundance of
clay particles in the source area or from deposition in an environment far from high-
energy streams (i.e. farther offshore). In a marine environment, sand and silt-sized
particles are generally deposited nearer to shore, whereas clay sized particles can be
temporarily suspended and carried farther offshore. In addition, kaolinite has been
shown to be more abundant closer to shore, whereas montmorillonite becomes more
abundant farther offshore (Blatt et al, 1972). This can be attributed to the small size of
montmorillonite particles. Even though montmorillonite is highly reactive and tends to
readily flocculate in the marine environment, these floccules normally do not exceed the
size of kaolinite particles.
In addition to the lithological effects discussed above, the behavior of clay shales is
related to the degree of consolidation and the degree of unloading to which they have
been exposed. Consolidation can be defined as the progressive decrease in the volume
of voids in a soil in response to loading, and in saturated materials is accompanied by
expulsion of water from the voids. Overconsolidation is a condition where the present
overburden load is less than any previous load which the material has experienced. The
history of consolidation and unloading in clay shales can be very complex, involving the
cyclic deposition and erosion of overlying materials, the advance and retreat of glaciers,
cyclic wetting and drying, and loading and unloading by tectonism.
206

Figure B.1. Relative abundance of major groups of clay minerals in


Phanerozoic mudrocks (Blatt et al, 1972; Weaver, 1967).

The probable stress history and resulting variation of water content, of the Bearpaw
clay shale is graphically illustrated in Fig. B.2. Consolidation loads are generally
estimated from measurements in laboratory consolidation tests. However, Kenney et al
(1967) have shown that the measured consolidation load can be greatly dependent on
the composition of pore fluid and cementing compounds, and can therefore be
erroneous.

Still it appears that the Bearpaw, Claggett, Pierre, and Colorado clay shales in the
northern plains of North America were consolidated by loads of a hundred tons/sq ft
(about 10 MPa) or more (Fleming et al, 1970). As indicated by Table B.1, the loads
207

Figure B.2. Probable stress history of the Bearpaw sediments (Scott and
Brooker, 1968).

Table B.1 Estimated maximum preconsolidation loads on North


America clay shales units (Fleming et al, 1970).
208

from glacial ice were apparently less than those from earlier overlying sediments. The
London clay is estimated to have been covered with 500 to 700 ft of overburden
(Henkel, 1957), resulting in consolidation loads of about 150 tons/sq ft, or 15 MPa.
Consolidation loads of 13 to 20 MPa have been reported by Banks (1971) for the
Cucaracha clay shale in Panama, while those estimated for the Italian clay shales range
from 0.3 to 5.0 MPa (Cancelli, 1981; Esu and Grisolia, 1977; A.G.I., 1977; Fenelli et
al, 1982). The stress history of the Italian clay shales has in general been greatly
complicated by additional tectonic shearing forces.

The strength of clay shales and their subsequent softening depend on the extent and
strength of the bonds that form during and after consolidation. The physico- chemical
processes that are believed to be responsible for this bonding are discussed in great
depth throughout the book of Mitchell (1976), and will not be reviewed extensively in
this dissertation. The nature of bonding is very complex, involving interlayer forces
acting within clay particles, as well as those forces acting between individual clay
particles. In some clay minerals, such as illite and kaolinite, the interlayer bonding may
be very strong; in contrast, the interlayer bonds within particles of montmorillonite are
moderate, readily allowing separation between layers and thus swelling of the clay
particle.

Bonding between clay particles involves the complex interplay of attractive van Der
Waals forces and the repulsive forces acting between the tightly held viscous layers of
water that surround the particles. In the schematics of Fig. B.3 is illustrated the nature
of these forces as they are believed to exist for different conditions of chemical
environment and mineralogy. It can be seen that under some conditions, there is an
initial repulsion between particles as these particles are forced together. However, if
this "threshold" of repulsion is exceeded, the forces become attractive and therefore
provide a degree of bonding between particles. In addition to the forces discussed
above, very strong bonds can be developed between particles by the introduction of
chemical bonds, or cementation, between these particles. All of these bonds act to
influence the nature of the fabric during deposition and compaction, and subsequently
influence the susceptibility of the material to loss of strength during unloading and
weathering.
209

Figure B.3. Energies of repulsion, attraction, and net curves of interaction for
parallel flat plates (Mitchell, 1976).

The geological history of overconsolidated clays and clay shales was discussed by
Skempton (1964), and is illustrated in the schematic of Fig. B.4, presented by Fleming
et al (1970). Point (a) represents a unit of clay immediately after deposition. The clay
at this stage has zero horizontal and vertical stresses acting on it and is characterized by
a very high water content and essentially no strength. As the effective vertical stress is
increased by the continual deposition of overlying sediment, as at point (b), the water
content of that clay unit decreases, the shear strength increases, and the horizontal
stresses increase such that the Ko ratio of horizontal to vertical stress is constant but
less than one. This process continues to the stage represented by point (c), at which
time further deposition is halted. At the points (a), (b), and (c), the clay is "normally-
consolidated" since it has never been exposed to overburden loads greater than the load
existing at that stage.
210

For clay shales, the consolidation stage described above is followed by a period of
vertical stress release, resulting for example from the erosion of overlying sediments.
The clay unit thus "rebounds" in response to this unloading by regaining some of the
water lost, and losing some of the strength gained during the consolidation process.
The clay is now "overconsolidated" since the overburden stresses are less than they had
been in the past. The rebound effect exhibited by the clay unit does not typically result
in total recovery of water content or total loss of strength. The effective vertical stress
acting on the overconsolidated clay unit at point (d) in Fig. B.4, is the same as that
experienced by the normally consolidated clay at point (b). However, in the
overconsolidated stage, the clay has a lower water content and is stronger than the
normally consolidated clay at the same effective confining pressure. Upon reloading of
the overconsolidated clay, the void ratio response is approximated by that shown in
Fig. B.5. The clay would initially exhibit low compressibility until the stress reaches the
highest load experienced by the sample, or the "preconsolidation" load, after which the
void ratio curve again follows the curve defined by the normally consolidated clay.

The time period between the process of deposition and the beginning of unloading can
be considerable. Large overburden stresses acting over this long period of time can
result in significant reduction of the void ratio, a rearrangement of soil particles, and a
subsequent increase in the number and strength of interparticle bonds. This process has
been referred to as "secondary compression" or "aging", and its occurrence during
consolidation tests in the lab has been reported by Leonards and Rahmiah (1960). The
results showed in Fig. B.6, illustrate the effects of a sustained load on decreasing the
void ratio and compressibility of a normally consolidated clay. Upon initial reloading,
the clay specimen acts as though it was an overconsolidated clay which has experienced
a preconsolidation pressure higher than that which has been applied.

The effect of sustained loading over geological time was shown by Bjerrum (1972) to
be similar. As illustrated in Fig. B.7, Bjerrum has suggested that the rate of decrease of
the void ratio is roughly proportional to the logarithm of time. Therefore, if a clay
undergoes secondary compression for a period of 10,000 years, about 50% of the
compression will have occurred in the first year and about 80% after 100 years. The
clay which sustained a load of po for 10,000 years in Fig. B.7, will act upon reloading
211

Figure B.4. Schematic of the stress history of an overconsolidated clay over


geological time (Skempton, 1964; Fleming et al, 1970).
212

Figure B.5. Schematic showing void ratio response to reloading after rebound.

as though it were an overconsolidated clay which has experienced a greater


preconsolidation load of pc'.
This behavior can be attributed to increased bonding arising from recrystallization of
particles, adhesion between particles, or precipitation of cementing agents between
particles. Bjerrum (1967) has presented a scheme for the geological history of
overconsolidated clays and clay shales similar to that presented by Skempton (1964)
and discussed above. However, the scheme of Bjerrum accounts for the effects of
diagenetic bonds formed during or after the consolidation process. Referring back to
213

Figure B.6. Aging effects observed in the laboratory for a normally


consolidated clay (Leonards and Rahmiah, 1960).
214

Figure B.7. Schematic illustrating aging effects resulting from sustained loading over
geological time (Bjerrum, 1972).
215

the schematic in Fig. B.4, if the clay unit which has been consolidated to the point (c)
then undergoes extensive secondary compression with time, the properties of the clay
would then be represented by the point (c)'. Water content would be reduced, bonding
would be increased, and as a result, the strength would be increased. This bonding can
become so strong that the material can then be classified as a rock.

During unloading, the clay has a tendency to swell and to pull in more water. However,
the stronger the bonds formed during consolidation, the more the clay will be inhibited
from swelling and taking on more water. Therefore, some of the strain energy gained
during consolidation may be retained by the clay after unloading, depending on the
strength of the bonds. Clays which have weak bonds will release all of their strain
energy immediately upon unloading. In contrast, clays or claystones with strong bonds
may retain some strain energy upon unloading, but will release this energy slowly in
response to weathering processes. According to the definition advanced in Appendix
A, such materials would be classified as clay shales. Clays and claystones which have
formed what Bjerrum called permanent bonds, may never release their stored strain
energy.

During the loading of clay deposits, the horizontal stresses increase in direct proportion
to the vertical stress, so that the ratio of horizontal to vertical stresses, Ko, is constant
and less than one. Upon unloading, the clay is able to expand vertically more easily
than in the horizontal direction. Therefore, vertical stresses are released to a greater
extent than horizontal resulting in a relative increase in the horizontal stresses and an
increase in the ratio, K. Because clays with weak bonds tend to undergo extensive
expansion upon unloading, the horizontal stresses in these materials can be very high
relative to the vertical overburden. Similar results have been observed for clay shales
which have expanded in response to weathering near the surface.

Results obtained by Brooker (1967) suggest that the susceptibility to disintegration by


slaking is a function of the stored strain energy and that the amount of absorbed strain
energy is dependent on the mineralogy. Plastic clays containing significant percentages
of illite, and in particular, montmorillonite, exhibit a high tendency to store large
amounts of strain energy, and to release this energy in response to weathering
processes. This is consistent with the relationship between these minerals and the
response of clay shales observed in the field.
216

It can be summarized that the formation of a clay shale requires one or more cycles of
extensive loading followed by a period of unloading, during which strong but not
permanent, bonds are formed. The tendency of a clay deposit to develop into a clay
shale is greatly enhanced by the presence of montmorillonite, and to a lesser degree,
illite. Although a clay shale may appear to be much stronger than other
overconsolidated clays as a result of stronger bonds, these bonds are readily destroyed
by the processes of weathering occurring near the surface of the earth, resulting in a
rapid and drastic loss of strength.
217

APPENDIX C

THE MECHANICS OF SINGLE FISSURES

Clean joints. The shear strength along a perfectly straight and smooth joint is
controlled simply by the residual strength of the wall material. However, as illustrated
in Fig. C.1, joint surfaces may be undulated or stepped, as well as planar, and on a
smaller scale, these surfaces may be rough, smooth, or very polished. The effect of
non-planar, or rough, joint surfaces is to create a certain degree of interlocking
between the two joint surfaces, as they try to slip past each other. Therefore, in order
for slip to occur, these rough peaks or "asperites" must either slide over one another,
resulting in dilation normal to the joint, or the asperities must themselves break by
shear or tensile failure.

This behavior was emphasized by Patton (1966), who developed a model for the shear
strength along a joint based on simple geometric relationships and the assumption of
zero cohesion. From the geometric relationships shown in Fig. C.2a for an inclined,
planar joint, the normal and shear stresses acting on the surface are given by

Γi = Γcos2i + σ sin i cos i, (A.1)

σi = σ cos2i + Γ sin i cos i, (A.2)

and assuming zero cohesion, its shear strength by

Γi = σ itan φ. (A.3)

By combining these equations, one obtains the shear strength in terms of normal and
shear stresses acting along the primary joint direction

Γ = σ tan (φ+ i). (A.4)


218

Figure C.1. Schematic illustrating various shapes and roughnesses that are
possible in joints.
219

Figure C.2. Schematic of the joint models used by Patton (1966).

Patton demonstrated in a series of tests on models with regular surface projections, that
at low normal stresses this equation likewise gives the strength of the analogous joint
model shown in Fig. C.2b. At higher normal stresses, however, the shear strength of
the wall material is exceeded, and failure begins to occur through the inclinations, or
asperities. In accordance with Patton's model, the shear strength at higher stresses, is
therefore given by the shear strength of the wall material, which in terms of Mohr-
coulomb theory, is
220

ΓI = cI + σ tan φI. (A.5)

Ladanyi and Archambault (1970) noted that the transition from dilation to shearing
through asperities would be smooth rather than abrupt, and based on theory and
experimentation, proposed that the strength of a joint could be defined by the equation

(1-as)(v + tan φ) + as ΓI (A.6)


I=
1 - (1 - as)v tan φ

where as is the proportion of the discontinuity surface which is sheared through intact
rock material, v is the dilation rate at peak shear strength, and I is the shear strength of
intact material. Ladanyi and Archambault suggested that the shear strength of intact
material next to the joint be given by the parabolic equation of Fairhurst (1964):

ΓI = σj √1+n - 1 (1 + n σ )1/2 (A.7)


n σj

where σj is the uniaxial compressive strength of the rock material adjacent to the
discontinuity, and n is the ratio of uniaxial compressive strength to uniaxial tensile
strength of the adjacent rock material. It should be noted that, duet to weathering or
loosening, σ may be lower than the uniaxial strength of the material within intact
j
blocks.

The plot in Fig. C.3 compares the normalized shear strength of a joint as predicted by
the equation of Ladanyi and Archambault, to that predicted by the Patton equation.
According to Patton's analysis, the strength envelope would change abruptly once the
shear stresses exceeded the shear strength of the intact material. In contrast, the
envelope of Ladanyi and Archambault exhibits a more gradual transition from the
strength as predicted by Patton to the intact material strength, and demonstrates that
the strength may be overestimated by Patton's equation. However, problems do exist
with the equation of Ladanyi and Archambault, particularly with regard to
determination of the values for as and v. These parameters must generally be
determined empirically, leading to rather complex equations.
221

Figure C.3. Comparison of normalized shear strength of a joint as predicted by


equations of Ladanyi and Archambault and the model of Patton.

An alternative approach has been proposed by Barton (1973), who developed the Joint
Roughness Coefficient, JRC. Based on careful tests on artificially produced "joints",
Barton derived the following empirical equation for the shear strength of a joint:

Γ = σ tan [ φ+ JRC.log10(JCS/σ)] , (A.8)

where JCS is the Joint Wall Compressive Strength, which is equivalent to σj in the
equations of Ladanyi and Archambault. Thus the effects of joint roughness are more
directly seen in the equation of Barton. The value of the JRC is determined from plots
such as in Fig. C.4 and is dependent on the length of the joint over which shear occurs
(Pratt et al, 1974; Bandis et al, 1981). This dependency on scale, and the effects of
different values of the JRC, are illustrated in Fig. C.5. Therefore, even though a large-
scale fault may be extremely rough, its JRC may be very low due to the extreme
222

Figure C.4. Roughness profiles and corresponding ranges of JRC values


associated with each.
223

Figure C.5. Plot showing the dependency of shear strength on scale and JRC.

length over which slip occurs. Similarly, a joint which has a low JRC value in the field,
may have a high JRC value with regard to the size of samples tested in the laboratory.

All of these models emphasize the importance of the degree of roughness along the
joints. The dependency on the scale of roughness relative to the magnitude of
displacement has likewise been emphasized. However, no methods have been
developed to adequately measure and quantify the roughness of joint surfaces at all
scales, although the profilometric and analytical procedures presented by Farrington
(1983) are promising.

The few models reviewed above primarily attempt to define the failure envelope for the
peak strength of single joints, and therefore do not describe the stress strain
224

Figure C.6. Critical state model of Roberds and Einstein (1978) for the
behavior of rock joints.
behavior before or after failure. At least three models have also been proposed to
account for this behavior of joints based on elasto plastic concepts. The model of

Agbabian, as presented by Ghaboussi et al (1973), and the Goodman model (Goodman,


1966, 1974; Goodman and Dubois, 1972; Goodman et al, 1968) adequately model
some behavioral aspects of fissured materials, but have been shown by Roberds and
Einstein (1978) to be somewhat limited.

Roberds and Einstein (1978) proposed a comprehensive model for the behavior of rock
joints based on elasto-plastic principles and the concept of critical state mechanics. As
shown in Fig. C.6, the model accounts for the complex behavior of joints by allowing
for four yield surfaces: (a) the intact rock yield surface, which acts in nonjointed rocks
or when the normal stress is very high, (b) the discontinuity, or joint, yield surface,
which governs yield at low stresses once a joint has formed, (c) the discontinuity
residual yield surface, and (d) the ultimate residual yield surface, given by the CS line
and determined apparently as the strength of a planar, perfectly smooth joint. In
225

Fig. C.6, is illustrated the stress path corresponding to the loading of an intact rock
until fracture is initiated, followed by unloading, then reloading to the strength given
now by the discontinuity yield surface, and finally by the discontinuity residual yield
surface.

Within this scheme, the ultimate residual strength, or critical state condition, can be
obtained if the joint is sheared at sufficiently high stresses to again mobilize the intact
rock strength. Evidence for two "residual" yield surfaces has been presented by Krahn
et al (1979), Insley et al (1977), Weisner (1969), and Thomson and Hayley (1975),
who report that the strength mobilized in Canadian clay shales was found to be equal to
the residual strength of remolded, precut samples, rather than the residual strength of
undisturbed samples.

Roberd and Einstein have demonstrated that the previous models of Patton, Barton,
and Ladanyi and Archambault can be incorporated into the comprehensive critical state
model. Although much more investigation is required to define the yield surfaces of
jointed rock, the critical state model of Roberds and Einstein, provides a useful
framework within which further advances in analyzing the behavior of clean rock joints
might be made.

Filled joints. Joints in nature are often filled with materials with different mechanical
properties than the joint wall material. These can consist of materials which have been
carried into the joint by water, or material derived from the joint walls by weathering or
by crumbling during shear. Several studies have investigated the strength of filled
joints, although most have been concerned with a specific problems encountered during
construction projects. The studies of most interest to this dissertation include those
dealing with clay filling (Brekke, 1965; Brekke and Selmer-Olsen, 1965; Brekke and
Howard, 1972; Sinclair and Brooker, 1967; Leussink and Muller-Kirchenbauer, 1967;
and Stimpson and Walton, 1970), and those dealing with the effects of wall rock
alteration (Serafim and Lopez, 1961; Rocha, 1964; Patton, 1966; and Deere and
Patton, 1971).

More systematic studies have been performed by Goodman (1970), Tulinov and
Molokov (1971), and Goodman et al (1972). An extensive review of the studies prior
226

Figure C.7. Shear strength of a rough joint as a function of the joint-fill thickness.

to 1974 on filled joints was presented by Barton (1974). The findings of these, and
later, studies can be summarized, as per Ladanyi and Archambault (1977), as follows:

(1) For most filled joints, the failure envelope is located between that for the
filling and that for a similar clean joint.

(2) As illustrated in Fig. C.7, the strength and stiffness of filled joints decreases
gradually as the thickness of the filling increases relative to the amplitude of the
asperites; although the strength and stiffness approach that of the filling, they
still remain significantly higher than the strength of the filling alone, even at
100% filling thickness.
227

(3) The stress-strain curves of filled joints often display two segments, one
reflecting the deformability of the filling before rock to rock contact, and the
other reflecting contact between asperities, after the filling has been sheared.

(4) The dilation rate at failure decreases with increasing normal pressure and
eventually becomes negative (i.e. contractive) at high normal pressures.

(5) The strength of filled joints does not always depend on the thickness of the
filling. If the joint walls are flat and smooth, or covered by a coating with a
very low coefficient of friction, the shear plane is always located along the
contact between the wall and the filling.

(6) One of the most dangerous fill, or gouge, materials is a swelling clay,
because of its loss of strength due to swelling and the high pressures created if
swelling is prevented.

Thus, the thickness of filling required to significantly reduce the strength of a joint
depends on the roughness of the joint walls. For joints which are smooth and planar,
the strength of the joint can be rapidly reduced to the strength of the fill material. In
addition, if the strength of the fill material is much lower than the shear strength along
the clean joint, the reduction of strength resulting from the presence of a filling can be
rather drastic. Finally, a filling of swelling clay is particularly troublesome due to the
loss of strength and high pressures associated with swelling.
228

APPENDIX D

THE STRENGTH OF A CLAY MASS CONSISTING


OF INTACT CORES SURROUNDED
BY A MUD MATRIX

At some stage during the softening of clay shale, the deposit might be considered as an
assemblage of stiff, unaltered clay cores, surrounded by a soft, remolded clay, or mud,
matrix. In evaluating the strength of such a clay mass, we will consider three cases: (a)
the intact cores, or grains if you will, are totally separated by the clay matrix and there
is essentially no interaction between the cores, (b) the cores are in contact, but there is
no interlocking, and (c) the cores are closely packed, resulting in particle interlock. For
the first case, the strength of the mass is provided entirely by the soft, clay matrix.
Therefore it is important to note that it is not necessary to soften the entire mass of a
clay shale before it will exhibit the "fully softened", or remolded, strength. It is only
necessary that the concentration of intact clay cores is sufficiently low as to inhibit
significant interaction between intact cores.

The second and third cases have been analyzed by Rowe (1962) using Mohr-Coulomb
theory. The second case involving contact of the cores, but no interlocking, is
illustrated in Figs. D.1a. From the analogous block model in Fig. D.1b, the pressure P
required to slide the block is equal to the interparticle cohesive force,
C, and the interparticle frictional force, Q.tan µ , so
that

P = C + Q.tan µ. (D.1)

However, if adjacent intact cores are interlocked as in Fig. D.1c, the analogous block
model in Fig. D.1d indicates that force required for sliding is now equal to

P = {C / [cos β (1-tanβtanµ)]} + Q.tan(µ+β) . (D.2)


229

Figure D.1. Two dimensional friction model for granular materials (Rowe,
1962).

Thus, the effective friction angle now includes an interlocking, as well as an


interparticle, component,

φe = β + µ ,
(D.3)

and the effective cohesion, likewise, includes both interlocking and interparticle
components,
Ce = C / [cos β (1-tanβtanµ)] . (D.4)
These two cases are essentially the same as for loosely, and densely, packed granular
soils, respectively.
230

Rodine (1974) performed special cone penetration tests on various mixtures of sand-
sized particles within a clay-water slurry in order to assess the effects of particle
concentration on the strength of the mixtures. He determined that the concentration of
particles of a single size, could be 45 to 65% of the total volume before significant
interlocking occurred, and up to 64% for particles of two select sizes. However,
theoretical analysis of mixture of multi-sized particles suggests that interlocking would
be negligible at particle concentrations as high as 89 to 95%.

Once initiated, mudflows consisting of hard clay fragments surrounded by a mud


matrix, are characterized by very low strengths and the ability to transport large
fragments, and even boulders, over long distances on shallow slopes (Rodine and
Johnson, 1976; Hampton, 1979). The least slope angle at which these mudflows are
mobilized in the field is always smaller than the slope angles predicted by the method of
Skempton and DeLory (1957) based on residual strength conditions (Vallejo, 1979;
1980). Vallejo, as well as Hampton (1979), have shown that these very low strengths
can be accounted for using the particle "dispersion" theory, proposed by Bagnold
(1954; 1956).

The present author has observed the presence of very low-strength mudflows which
flow for long distances on slopes of only a few degrees, at the village of Bisacchia,
Italy. These highly fluid mudflows originate from rather steep, and apparently
metastable, slopes composed of "argille varicolori". The transition of this clay shale
into a mudflow is surprisingly drastic, with the mudflows originating from within deep
"collapse channels" which are lined by near vertical walls. These collapse channels are
very similar in appearance to those often associated with the failure of "quick clays".
Although there does appear to be some softening occurring along many of the fissures,
the wall material itself has not been softened to any great extent. The proportion of
material which has been significantly softened certainly does not exceed five percent.
Such mudflows may originate in the manner of the flows discussed above, and suggest
again that very little softening is required to develop a highly fluid clay mass.

In summary, if the concentration of intact cores is low, the strength of the mass will be
equal to the strength of the matrix material. The strength in this case would be
dependent on the water content of the matrix material and might be analyzed in
accordance to the critical state model. If the concentration of intact clay cores is
higher, such that there is interference between cores, then the material should be
231

modeled as a granular soil whose strength is controlled by the interparticle and


interlocking components of strength acting between the intact cores, as well as the
degree of packing of the cores. However, two factors make the analysis of highly-
altered clay shale more complicated than that for typical granular materials: (a) unlike
pore water which essentially has no resistance to shear, the interparticle "fluid" in an
altered clay shale has a shear strength of its own, as well as a component of strength
acting at the boundaries between the fluid and cores, and (b) the "grains", or cores, may
themselves be quite soft, and therefore cannot be assumed to be infinitely stiff.

The interlocking component of strength, which is certainly of great importance in


fissured clay shales, could in theory be eliminated if only 5% of the clay shale material
is softened. In reality, the percentage of softening required to eliminate interlocking is
probably somewhat higher, but may still be surprisingly low. The percentage of clay
shale material which must be softened before the "fully softened" strength is obtained
has not been determined. However, it is not necessary to soften the entire clay shale
mass, nor even a large percentage of the mass, before mobilizing "fully softened"
strength.
232

APPENDIX E

PRESENTATION OF DATA

THE FOLLOWING PAGES PRESENT DATA IN THE FORM OF


GRAPHS, SKETCHES, AND NOTES FOR EACH TEST.

ALL GRAPHS SHOULD BE SIMILAR IN SCALE FOR


COMPARISON.
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262

You might also like