Professional Documents
Culture Documents
38.fundamentals of Soil Behaviour 3nd Edition (James K.Mitchell&Kenichi Soga PDF
38.fundamentals of Soil Behaviour 3nd Edition (James K.Mitchell&Kenichi Soga PDF
Soil Behavior
l
ria
Third Edition
ate
James K. Mitchell
dM Kenichi Soga
hte
rig
py
Co
Copyright 2005 by John Wiley & Sons, Inc. All rights reserved
Published by John Wiley & Sons, Inc., Hoboken, New Jersey
Published simultaneously in Canada
l
ria
No part of this publication may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording, scanning, or otherwise,
except as permitted under Section 107 or 108 of the 1976 United
States Copyright Act, without either the prior written permission of
the Publisher, or authorization through payment of the appropriate
ate
per-copy fee to the Copyright Clearance Center, Inc., 222
Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978)
750-4470, or on the web at www.copyright.com. Requests to the
Publisher for permission should be addressed to the Permissions
Department, John Wiley & Sons, Inc., 111 River Street, Hoboken,
NJ 07030, (201) 748-6011, fax (201) 748-6008, e-mail:
permcoordinator@wiley.com. dM
Limit of Liability/Disclaimer of Warranty: While the publisher and
author have used their best efforts in preparing this book, they
make no representations or warranties with respect to the accuracy
or completeness of the contents of this book and specifically
disclaim any implied warranties of merchantability or fitness for a
particular purpose. No warranty may be created or extended by
sales representatives or written sales materials. The advice and
hte
Soga.—3rd ed.
p. cm.
ISBN-13: 978-0-471-46302-7 (cloth : alk. paper)
ISBN-10: 0-471-46302-7 (cloth : alk. paper)
1. Soil mechanics. I. Soga, Kenichi. II. Title.
TA710.M577 2005
624.15136—dc22
2004025690
l
Preface xi
ria
CHAPTER 1 INTRODUCTION 1
1.1 Soil Behavior in Civil and Environmental Engineering 1
ate
1.2 Scope and Organization 3
1.3 Getting Started 3
dM 2.1 Introduction 5
2.2 The Earth’s Crust 5
2.3 Geologic Cycle and Geological Time 6
2.4 Rock and Mineral Stability 7
2.5 Weathering 8
2.6 Origin of Clay Minerals and Clay Genesis 15
2.7 Soil Profiles and Their Development 16
2.8 Sediment Erosion, Transport, and Deposition 18
2.9 Postdepositional Changes in Sediments 25
hte
Engineering 35
3.2 Atomic Structure 38
3.3 Interatomic Bonding 38
3.4 Secondary Bonds 39
3.5 Crystals and Their Properties 40
py
l
Interrelationships 85
ria
4.3 Engineering Properties of Granular Soils 85
4.4 Dominating Influence of the Clay Phase 94
4.5 Atterberg Limits 95
4.6 Activity 97
4.7 Influences of Exchangeable Cations and pH 97
ate
4.8 Engineering Properties of Clay Minerals 98
4.9 Effects of Organic Matter 104
4.10 Concluding Comments 105
Questions and Problems 106
CHAPTER
dM 5 SOIL FABRIC AND ITS MEASUREMENT 109
5.1 Introduction 109
5.2 Definitions of Fabrics and Fabric Elements 110
5.3 Single-Grain Fabrics 112
5.4 Contact Force Characterization Using Photoelasticity 119
5.5 Multigrain Fabrics 121
5.6 Voids and Their Distribution 122
hte
l
7.8 Measurement of Pore Pressures in Soils 183
ria
7.9 Effective and Intergranular Pressure 184
7.10 Assessment of Terzaghi’s Equation 185
7.11 Water–Air Interactions in Soils 188
7.12 Effective Stress in Unsaturated Soils 190
7.13 Concluding Comments 193
ate
Questions and Problems 193
l
9.22 Ground Freezing 310
ria
9.23 Concluding Comments 319
Questions and Problems 320
ate
10.1 Introduction 325
10.2 General Volume Change Behavior of Soils 325
10.3 Preconsolidation Pressure 327
10.4 Factors Controlling Resistance to Volume Change 330
10.5 Physical Interactions in Volume Change 331
10.6 Fabric, Structure, and Volume Change 335
dM 10.7 Osmotic Pressure and Water Adsorption Influences on
Compression and Swelling 339
10.8 Influences of Mineralogical Detail in Soil Expansion 345
10.9 Consolidation 348
10.10 Secondary Compression 353
10.11 In Situ Horizontal Stress (K0) 355
10.12 Temperature–Volume Relationships 359
hte
l
12.8 Rate Effects on Stress–Strain Relationships 497
ria
12.9 Modeling of Stress–Strain–Time Behavior 503
12.10 Creep Rupture 508
12.11 Sand Aging Effects and Their Significance 511
12.12 Mechanical Processes of Aging 516
12.13 Chemical Processes of Aging 517
ate
12.14 Concluding Comments 520
Questions and Problems 520
Index 559
hte
rig
py
Co
l
According to the National Research Council (1989, 2005), sound geoengineering is key in
ria
meeting seven critical societal needs. They are waste management and environmental protec-
tion, infrastructure development and rehabilitation, construction efficiency and innovation, se-
curity, resource discovery and recovery, mitigation of natural hazards, and the exploration and
development of new frontiers. Solution of problems and satisfactory completion of projects in
ate
each of these areas cannot be accomplished without a solid understanding of the composition,
structure, and behavior of soils because virtually all of humankind’s structures and facilities
are built on, in, or with the Earth. Thus, the purpose of this book remains the same as for the
prior two editions; namely, the development of an understanding of the factors determining
and controlling the engineering properties and behavior of soils under different conditions,
with an emphasis on why they are what they are. We believe that this understanding and its
dM
prudent application can be a valuable asset in meeting these societal needs.
In the 12 years since publication of the second edition, environmental problems requiring
geotechnical inputs have remained very important; dealing with natural hazards and disasters
such as earthquakes, floods, and landslides has demanded increased attention; risk assessment
and mitigation applied to existing structures and earthworks has become a major challenge;
and the roles of soil stabilization, ground improvement, and soil as a construction material
have expanded enormously. These developments, as well as the introduction of new compu-
hte
tational, geophysical, and sensing methods, new emphasis on micromechanical analysis and
behavior, and, perhaps regrettably, the reduced emphasis on laboratory measurement of soil
properties have required looking at soil behavior in new ways. More and more it is becoming
appreciated that geochemical and microbiological phenomena and processes play an essential
role in many types of geotechnical problems. Some of these considerations have been incor-
rig
the flow of topics and logic of presentation. Time effects on soil strength and deformation
behavior have been separated into a new Chapter 12. Additional soil property correlations
have been incorporated. The addition of sets of questions and problems at the end of each
chapter provide a feature not present in the first two editions. Many of these questions and
Co
problems are open ended and without single, clearly defined answers, but they are designed
to stimulate broad thinking and the realization that judgment and incorporation of concepts
and methods from a range of disciplines is often needed to provide satisfactory solutions to
many geoengineering problems.
We are indebted to innumerable students and professional colleagues whose inquiring minds
and perceptive insights have helped us clarify issues and find new and better explanations for
observed processes and behavior. J. Carlos Santamarina and David Smith provided helpful
suggestions on the overall content and organization. Charles J. Shackelford reviewed and
provided valuable suggestions for the sections of Chapter 9 on chemical osmosis and advective
and diffusive chemical flows. Other important contributions to this third edition in the form
xi
of valuable comments, photos, resources, and proof checking were made by Hendrikus Al-
lersma, Khalid Alshibli, John Atkinson, Bob Behringer, Malcolm Bolton, Lis Bowman, Jim
Buckman, Pierre Delage, Antonio Gens, Henry Ji, Assaf Klar, Hideo Komine, Jean-Marie
Konrad, Ning Liu, Yukio Nakata, Albert Ng, Masanobu Oda, Kenneth Sutherland, Colin
Thornton, Yoichi Watabe, Siam Yimsiri, and Guoping Zhang.
KS thanks his wife, Mikiko, for her encouragement and special support.
We dedicate this book to the memory of Virginia (‘‘Bunny’’) Mitchell, whose continuing
love, support, encouragement, and patience over more than 50 years, made this and the prior
two editions possible.
JAMES K. MITCHELL
l
University Distinguished Professor, Emeritus
ria
Virginia Tech, Blacksburg, Virginia
KENICHI SOGA
Reader in Geomechanics
University of Cambridge, Cambridge, England
ate
March 2005
References
dM
National Research Council. 1989. Geotechnology—Its Impact on Economic Growth, the En-
vironment, and National Security. National Academy Press, Washington, DC.
National Research Council. 2005. Geological and Geotechnical Engineering in the New Mil-
lennium, National Academy Press, Washington, DC.
hte
rig
py
Co
Introduction
l
ria
ate
1.1 SOIL BEHAVIOR IN CIVIL AND mechanics, and structural analysis and design, wherein
ENVIRONMENTAL ENGINEERING problems are usually quite clear-cut and well defined,
they are suddenly confronted with situations where this
Civil and environmental engineering includes the con- is no longer the case. A first course in soil mechanics
ception, analysis, design, construction, operation, and may not, at least for the first half to two-thirds of the
maintenance of a diversity of structures, facilities, and
dM course, be mechanics at all. The reason for this is sim-
systems. All are built on, in, or with soil or rock. The ple: Analyses and designs are useless if the boundary
properties and behavior of these materials have major conditions and material properties are improperly de-
influences on the success, economy, and safety of the fined.
work. Geoengineers play a vital role in these projects Acquisition of the data needed for analysis and de-
and are also concerned with virtually all aspects of sign on, in, and with soils and rocks can be far more
environmental control, including water resources, wa- difficult and uncertain than when dealing with other
ter pollution control, waste disposal and containment, engineering materials and aboveground construction.
hte
and the mitigation of such natural disasters as floods, There are at least three important reasons for this.
earthquakes, landslides, and volcanoes. Soils and their
interactions with the environment are major consider- 1. No Clearly Defined Boundaries. An embank-
ations. Furthermore, detailed understanding of the be- ment resting on a soil foundation is shown in Fig.
havior of earth materials is essential for mining, for 1.1a, and a cantilever beam fixed at one end is
rig
energy resources development and recovery, and for shown in Fig. 1.1b. The free body of the canti-
scientific studies in virtually all the geosciences. lever beam, Fig. 1.1c, is readily analyzed for re-
To deal properly with the earth materials associated actions, shears, moments, and deflections using
with any problem and project requires knowledge, standard methods of structural analysis. However,
understanding, and appreciation of the importance what are the boundary conditions, and what is the
py
of geology, materials science, materials testing, and free body for the embankment foundation?
mechanics. Geotechnical engineering is concerned 2. Variable and Unknown Material Properties.
with all of these. Environmental concerns—especially The properties of most construction materials
those related to groundwater, the safe disposal and con- (e.g., steel, plastics, concrete, aluminum, and
Co
tainment of wastes, and the cleanup of contaminated wood) are ordinarily known within rather narrow
sites—has spawned yet another area of specialization; limits and usually can be specified to meet certain
namely, environmental geotechnics, wherein chemistry needs. Although this may be the case in construc-
and biological science are important. Geochemical and tion using earth and rock fills, at least part of
microbiological phenomena impact the composition, every geotechnical problem involves interactions
properties, and stability of soils and rocks to degrees with in situ soil and rock. No matter how exten-
only recently beginning to be appreciated. sive (and expensive) any boring and sampling
Students in civil engineering are often quite sur- program, only a very small percentage of the sub-
prised, and sometimes quite confused, by their first surface material is available for observation and
course in engineering with soils. After studying statics, testing. In most cases, more than one stratum is
l
compute often exceeds considerably our ability to un-
ria
derstand, measure, and characterize a problem or
process. Thus, understanding and the ability to con-
ceptualize soil and rock behavior become all the more
important.
ate
The objectives of this book are to provide a basis
for the understanding of the engineering properties and
behavior of soils and the factors controlling changes
with time and to indicate why this knowledge is im-
portant and how it is used in the solution of geotech-
dM nical and geoenvironmental problems.
Figure 1.1 The problem of boundary conditions in geo- It is easier to state what this book is not, rather than
technical problems: (a) embankment on soil foundation, (b)
what it is. It is not a book on soil or rock mechanics;
cantilever beam, and (c) free body diagram for analysis of
propped cantilever beam.
it is not a book on soil exploration or testing; it is not
a book that teaches analysis or design; and it is not a
book on geotechnical engineering practice. Excellent
books and references dealing with each of these im-
present, and conditions are nonhomogeneous and portant areas are available. It is a book on the com-
hte
involve not just one or two but an almost infinite • What is clay?
number of different materials. • Why are clays plastic?
• What are friction and cohesion?
Add to the above three factors the facts that soil and • What is effective stress? Why is it important?
rock properties may be susceptible to influences from • Why do soils creep and exhibit stress relaxation?
changes in temperature, pressure, water availability, • Why do some soils swell while others do not?
and chemical and biological environment, and one • Why does stability failure sometimes occur at
might conclude that successful application of mechan- stresses less than the measured strength?
ics to earth materials is an almost hopeless proposition. • Why and how are soil properties changed by dis-
It has been amply demonstrated, of course, that such turbance?
l
• How do soil properties change with time after dep- their resulting structures and relationships to geotech-
ria
osition or densification and why? nical properties and stability are covered in Chapter 8.
• How do temperature changes influence the me- The next three chapters deal with those soil properties
chanical properties of soils? that are of primary importance to the solution of most
• What is soil liquefaction, and why is it important? geoengineering problems: the flows of fluids, chemi-
• What causes frost heave, and how can it be pre- cals, electricity, and heat and their consequences in
ate
vented? Chapter 9; volume change behavior in Chapter 10; and
• What clay types are best suited for sealing waste deformation and strength and deformation behavior in
repositories? Chapter 11. Finally, Chapter 12 on time effects on
• What biological processes can occur in soils and strength and deformation recognizes that soils are not
why are they important in engineering problems?
dM inert, static materials, but rather how a given soil re-
sponds under different rates of loading or at some time
Developing answers to questions such as these re- in the future may be quite different than how it re-
quires application of concepts from chemistry, geol- sponds today.
ogy, biology, materials science, and physics. Principles
from these disciplines are introduced as necessary to
develop background for the phenomena under study. It 1.3 GETTING STARTED
is assumed that the reader has a basic knowledge of
Find an article about a problem, a project, or issue that
hte
Relationships between soil composition and engineer- marizes the important ideas and conclusions. A
ing properties are developed in Chapter 4. Soil com- descriptive abstract, on the other hand, simply
position by itself is insufficient for quantification of states the article contents.)
soil properties for specific situations, because the soil 2. Summarize the important geotechnical issues that
fabric, that is, the arrangements of particles, particle are found in the article and write down what you
Co
groups, and pores, may play an equally important role. believe you should know about to understand
This topic is covered in Chapter 5. them well enough to solve the problem, resolve
Water may make up more than half the volume of the issue, advise a client, and the like. In other
a soil mass, it is attracted to soil particles, and the words, what is in the article that you believe the
interactions between water and the soil surfaces influ- subject matter in this book should prepare you to
ence the behavior. In addition, owing to the colloidal deal with? Do not exceed two pages.
Soil Formation
l
ria
ate
2.1 INTRODUCTION (acid) rocks predominate beneath the continents, and
basaltic (basic) rocks predominate beneath the oceans.
The variety of geomaterials encountered in engineering Because of these lithologic differences, the continental
problems is almost limitless, ranging from hard, dense, crust average density of 2.7 is slightly less than the
large pieces of rock, through gravel, sand, silt, and clay oceanic crust average density of 2.8. The elemental
to organic deposits of soft, compressible peat. All these
dM compositions of the whole Earth and the crust are in-
materials may exist over a wide range of densities and dicated in Fig. 2.1. There are more than 100 elements,
water contents. A number of different soil types may but 90 percent of Earth consists of iron, oxygen, sili-
be present at any site, and the composition may vary con, and magnesium. Less iron is found in the crust
over intervals as small as a few millimeters. than in the core because its higher density causes it to
It is not surprising, therefore, that much of the sink. Silicon, aluminum, calcium, potassium, and so-
geoengineer’s effort is directed at the identification of dium are more abundant in the crust than in the core
soils and the evaluation of the appropriate properties because they are lighter elements. Oxygen is the only
hte
for use in a particular analysis or design. Perhaps what anion that has an abundance of more than 1 percent
is surprising is that the application of the principles of by weight; however, it is very abundant by volume.
mechanics to a material as diverse as soil meets with Silicon, aluminum, magnesium, and oxygen are the
as much success as it does. most commonly observed elements in soils.
To understand and appreciate the characteristics of Within depths up to 2 km, the rocks are 75 percent
any soil deposit require an understanding of what the secondary (sedimentary and metamorphic) and 25 per-
rig
material is and how it reached its present state. This cent igneous. From depths of 2 to 15 km, the rocks are
requires consideration of rock and soil weathering, the about 95 percent igneous and 5 percent secondary.
erosion and transportation of soil materials, deposi- Soils may extend from the ground surface to depths of
tional processes, and postdepositional changes in sed- several hundred meters. In many cases the distinction
py
iments. Some important aspects of these processes and between soil and rock is difficult, as the boundary be-
their effects are presented in this chapter and in Chap- tween soft rock and hard soil is not precisely defined.
ter 8. Each has been the subject of numerous books Earth materials that fall in this range are sometimes
and articles, and the amount of available information difficult to deal with in engineering and construction,
is enormous. Thus, it is possible only to summarize the
Co
70%
Iron 35%
60% Silicon 28%
50%
Aluminum 1.1%
l
40%
Silicon 15%
ria
30%
Oxygen 46%
20%
Oxygen 30%
10%
ate
Figure 2.2 Geologic cycle.
0%
Earth's Crust Whole Earth
Figure 2.1 Elemental composition of the whole Earth and from some other area. Sediment formation pertains to
the crust (percent by weight) (data from Press and Siever, processes by which accumulated sediments are densi-
1994). dM fied, altered in composition, and converted into rock.
Crustal movement involves both gradual rising of
unloaded areas and slow subsidence of depositional ba-
moves from the interior of Earth toward the surface sins (epirogenic movements) and abrupt movements
has a significant influence on the characteristics of the (tectonic movements) such as those associated with
resulting rock. The more rapid the cooling, the smaller faulting and earthquakes. Crustal movements may also
are the crystals that form because of the reduced time result in the formation of new rock masses through
hte
for atoms to attain minimum energy configurations. igneous or plutonic activity. The interrelationships of
Cooling may be so rapid in a volcanic eruption that no these processes are shown in Fig. 2.3.
crystalline structure develops before solidification, and More than one process acts simultaneously in na-
an amorphous material such as obsidian (volcanic ture. For example, both weathering and erosion take
glass) is formed. place at the surface during periods of uplift, or oro-
genic activity (mountain building), and deposition, sed-
rig
l
ternary period is subdivided into the Holocene (the of detrital components (most sedimentary rocks).
ria
10,000 years after the last glacial period) and the Pleis- Sometimes, rocks are composed entirely of one type
tocene. The deposits during this period are controlled of mineral (say flint or rock salt), but generally they
mainly by the change in climate, as it was too short a contain many different minerals, and often the rock is
time for any major tectonic changes to occur in the a collection or aggregation of small particles that are
positions of land masses and seas. There were as many themselves pieces of rocks. Books on petrography may
ate
as 20 glacial and interglacial periods during the Qua- list more than 1000 species of rock types. Fortunately,
ternary. At one time, ice sheets covered more than however, many of them fall into groups with similar
three times their present extent. Worldwide sea level engineering attributes, so that only about 40 rock
oscillations due to glacial and interglacial cycles affect names will suffice for most geotechnical engineering
soil formation (weathering, erosion, and sedimenta- purposes.
Minerals have a definite chemical composition and
solidation and leaching.
dM
tion) as well as postdepositional changes such as con-
an ordered arrangement of components (a crystal lat-
tice); a few minerals are disordered and without defin-
able crystal structure (amorphous). Crystal size and
structure have an important influence on the resistance
of different rocks to weathering. Factors controlling the
Eon Era Period Epoch stability of different crystal structures are considered
hte
Paleogene Eocene
57 rock-forming minerals. On the other hand, quartz does
Paleocene
65 not assume its final crystal structure until the temper-
Cretaceous
146 ature drops below 573C. Because of its high stability,
Phanerozoic Mesozoic Jurassic
208 quartz is the most abundant nonclay mineral in soils,
Triassic although it comprises only about 12 percent of igneous
py
245
Permian rocks.
290
Pennsylvanian
323
As magma cools, minerals may form and remain, or
Mississippian
363
they may react progressively to form other minerals at
Paleozoic Devonian lower temperatures. Bowen’s reaction series, shown in
Co
409
Silurian Fig. 2.5, indicates the crystallization sequence of
439
Ordovician the silicate minerals as temperature decreases from
510
Cambrian 1200C. This reaction series closely parallels various
570
Proterozoic
weathering stability series as shown later in Table 2.2.
2500 Precambrian For example, in an intermediate granitic rock, horn-
Archean blende and plagioclase feldspar would be expected to
chemically weather before orthoclase feldspar, which
Figure 2.4 Stratigraphic timescale column. Numbers repre- would chemically weather before muscovite mica, and
sent millions of years before the present. so on.
l
mineral is more stable than the one above it on the list.
surface layers of rocks. Exfoliation may occur
ria
during rock excavation and tunneling. The term
popping rock is used to describe the sudden spall-
ing of rock slabs as a result of stress release.
Mineralogy textbooks commonly list determinative
2. Thermal Expansion and Contraction The ef-
properties for about 200 minerals. The list of the most
ate
fects of thermal expansion and contraction range
common rock- or soil-forming minerals is rather short,
from creation of planes of weakness from strains
however. Common minerals found in soils are listed in
already present in a rock to complete fracture.
Table 2.1. The top six silicates originate from rocks by
Repeated frost and insolation (daytime heating)
physical weathering processes, whereas the other min-
may be important in some desert areas. Fires can
erals are formed by chemical weathering processes.
dM cause very rapid temperature increase and rock
Further description of important minerals found in
weathering.
soils is given in Chapter 3.
3. Crystal Growth, Including Frost Action The
crystallization pressures of salts and the pressure
associated with the freezing of water in saturated
2.5 WEATHERING rocks may cause significant disintegration. Many
talus deposits have been formed by frost action.
Weathering of rocks and soils is a destructive process However, the role of freeze–thaw in physical
hte
whereby debris of various sizes, compositions, and weathering has been debated (Birkeland, 1984).
shapes is formed.2 The new compositions are usually The rapid rates and high amplitude of tempera-
more stable than the old and involve a decrease in the ture change required to produce necessary pres-
internal energy of the materials. As erosion moves the sure have not been confirmed in the field. Instead,
ground surface downward, pressures and temperatures some researchers favor the process in which thin
rig
in the rocks are decreased, so they then possess an films of adsorbed water is the agent that promotes
internal energy above that for equilibrium in the new weathering. These films can be adsorbed so
environment. This, in conjunction with exposure to the tightly that they cannot freeze. However, the wa-
atmosphere, water, and various chemical and biological ter is attracted to a freezing front and pressures
agents, results in processes of alteration. exerted during the migration of these films can
py
A variety of physical, chemical, and biological proc- break the rock apart.
esses act to break down rock masses. Physical proc- 4. Colloid Plucking The shrinkage of colloidal
esses reduce particle size, increase surface area, and materials on drying can exert a tensile stress on
increase bulk volume. Chemical and biological proc- surfaces with which they are in contact.3
Co
esses can cause complete changes in both physical and 5. Organic Activity The growth of plant roots in
chemical properties. existing fractures in rocks is an important weath-
ering process. In addition, the activities of
worms, rodents, and humans may cause consid-
2
erable mixing in the zone of weathering.
A general definition of weathering (Reiche, 1945; Keller, 1957) is:
the response of materials within the lithosphere to conditions at or
near its contact with the atmosphere, the hydrosphere, and perhaps
more importantly, the biosphere. The biosphere is the entire space
3
occupied by living organisms; the hydrosphere is the aqueous enve- To appreciate this phenomenon, smear a film of highly plastic clay
lope of Earth; and the lithosphere is the solid part of Earth. paste on the back of your hand and let it dry.
l
Olivine (Mg,Fe)2SiO4 Easily weathered
ria
Epidote Ca2(Al,Fe)3(OH)Si3O12 Highly resistant to chemical weathering; used
Tourmaline NaMg3Al6B3Si6O27(OH,F)4 as ‘‘index mineral’’ in pedologic studies
Zircon ZrSiO4
Rutile TiO2
Kaolinite Si4Al4O10(OH)8
ate
Smectite, Mx(Si,Al)8(Al,Fe,Mg)4O20(OH)4, Abundant in clays as products of weathering;
vermiculite, where M ⫽ interlayer cation source of exchangeable cations in soils
chlorite
Allophane Si3Al4O12 nH2O Abundant in soils derived from volcanic ash
deposits
Imogolite Si2Al4O10 5H2O
Gibbsite
dM
Al(OH)3 Abundant in leached soils
Goethite FeO(OH) Most abundant Fe oxide
Hematite Fe2O3 Abundant in warm region
Ferrihydrate Fe10O15 9H2O Abundant in organic horizons
Birnessite (Na,Ca)Mn7O14 2.8H2O Most abundant Mn oxide
Calcite CaCO3 Most abundant carbonate
CaSO4 2H2O
hte
Physical weathering processes are generally the the ion enables it to enter the lattice of minerals
rig
forerunners of chemical weathering. Their main con- and replace existing cations. For feldspar,
tributions are to loosen rock masses, reduce particle Orthoclase feldspar:
sizes, and increase the available surface area for chem-
ical attack. K silicate ⫹ H⫹OH⫺
py
l
bonic acid and other acids are produced by the ered surfaces.
ria
roots of plants, by insects that live in the soil, b. The ions held by Al2O3 and SiO2 colloids in-
and by the bacteria that degrade plant and animal fluence the types of clay minerals that form.
remains. c. Physical properties of the system such as the
The pH of the system is important because it permeability may depend on the adsorbed ion
influences the amount of available H⫹, the solu- concentrations and types.
ate
bility of SiO2 and Al2O3, and the type of clay 4. Oxidation is the loss of electrons by cations, and
mineral that may form. The solubility of silica reduction is the gain of electrons. Both are im-
and alumina as a function of pH is shown in Fig. portant in chemical weathering. Most important
2.6. oxidation products depend on dissolved oxygen
2. Chelation involves the complexing and removal
dM in the water. The oxidation of pyrite is typical of
of metal ions. It helps to drive hydrolysis reac- many oxidation reactions during weathering
tions. For example, (Keller, 1957):
Muscovite:
2FeS2 ⫹ 2H2O ⫹ 7O2 → 2FeSO4 ⫹ 2H2SO4
K2[Si6Al2]Al4O20(OH)4 ⫹ 6C2O4H2 ⫹ 8H2O
FeSO4 ⫹ 2H2O → Fe(OH)2 ⫹ H2SO4
→ 2K⫹ ⫹ 6C2O4Al⫹ ⫹ 6Si(OH)40 ⫹ 8OH⫺
(hydrolysis)
hte
l
pheric CO2 is the source of the ions. Limestone
ria
made of calcite and dolomite is one of the rocks
that weather most quickly especially in humid
regions. The carbonation of dolomitic limestone
proceeds as follows:
ate
CaMg(CO3)2 ⫹ 2CO2 ⫹ 2H2O
→ Ca(HCO3)2 ⫹ Mg(HCO3)2
Microbiological Effects
Several types of microorganisms are found in soils;
there are cellular microorganisms (bacteria, archea, al-
hte
velopment of Earth (Hattori, 1973; McCarty, 2004). (e.g., acids) directly on the rock surface (Ehrlich,
Photosynthetic bacteria, cyanobacteria, or ‘‘blue-green 1998). Biofilms bind cations in the pore fluid and fa-
bacteria’’ evolved about 3.5 billion years ago (Proter- cilitate nucleation and crystal growth even at low ionic
ozoic era—Precambrian), and they are the oldest concentrations in the pore fluid (Konhauser and Urru-
known fossils. Cyanobacteria use energy from the sun tia, 1999). After nucleation is initiated, further mineral
to reduce the carbon in CO2 to cellular carbon and to growth or precipitation can occur abiotically, including
obtain the needed electrons for oxidizing the oxygen the precipitation of amorphous iron–aluminum sili-
in water to molecular oxygen. During the Archaean cates and poorly crystallized claylike minerals, such as
period (2.5 billion years ago), cyanobacteria converted allophone, imogolite, and smectite (Urrutia and Bev-
the atmosphere from reducing to oxidizing and eridge, 1995; Ehrlich, 1999; Barton et al., 2001).
changed the mineral nature of Earth. In the case of the Carsington Dam construction,
l
Eukaryotic algae evolved later, followed by the mul- Cripps et al. (1993) hypothesized that autotrophic bac-
ria
ticellular eukaryotes including plants. Photosynthesis teria greatly accelerated the oxidation rate of the pyrite,
is the primary producer of the organic particulate mat- so that it occurred within months during construction.
ter in shale, sand, silt, and clay, as well as in coal, The resulting sulfuric acid reacted with the drainage
petroleum, and methane deposits. Furthermore, cyano- blanket constructed of carboniferous limestone, which
then resulted in precipitation of gypsum and iron hy-
ate
bacteria and algae increase the water pH when they
consume CO2 dissolved in water, resulting in carbonate droxide, clogging of drains and generation of carbon
formation and precipitation of magnesium and calcium dioxide.
carbonates, leading to Earth’s major carbonate forma-
tions. Weathering Products
Aerobic bacteria live in the presence of dissolved
dM The products of weathering, several of which will gen-
oxygen. Anaerobic bacteria survive only in the absence erally coexist at one time, include:
of oxygen. Facultative bacteria can live with or without
oxygen. Some bacteria may resort to fermentation to 1. Unaltered minerals that are either highly resistant
sustain their metabolism under anaerobic conditions or freshly exposed
(Purves et al., 1997). For example, in the case of an- 2. Newly formed, more stable minerals having the
aerobic conditions, fermenting bacteria oxidize carbo- same structure as the original mineral
hydrates to produce simple organic acids and H2 that 3. Newly formed minerals having a form similar to
hte
are used to reduction of ferric (Fe3⫹) iron, sulfate re- the original, but a changed internal structure
duction, and the generation of methane (Chapelle, 4. Products of disrupted minerals, either at or trans-
2001). Microbial energy metabolism involves electron ported from the site. Such minerals might include
transfers, and the electron sources and acceptors can a. Colloidal gels of Al2O3 and SiO2
be both organic and inorganic compounds (Horn and b. Clay minerals
c. Zeolites
rig
dioxide. Therefore, biological activity mediates geo- weathering stages is given in Table 2.2. The similarity
chemical reactions, causing them to proceed at rates between the order of representative minerals for the
that are sometimes orders of magnitude more rapid different weathering stages and Bowen’s reaction se-
than would be predicted solely on the basis of the ther- ries given earlier (Fig. 2.5) may be noted.
Co
l
apatite) tions are the youthful ered coarse rock fragments).
ria
3 Olivine-hornblende soils all over the Climate determines the amount of water present, the
(also pyroxenes) world, but mainly temperature, and the character of the vegetative cover,
4 Biotite (also glauco- soils of the desert and these, in turn, affect the biologic complex. Some
nite, nontronite) regions where limited general influences of climate are:
ate
5 Albite (also anorthite water keeps chemical
microcline, ortho- weathering to a mini- 1. For a given amount of rainfall, chemical weath-
clase) mum. ering proceeds more rapidly in warm than in cool
climates. At normal temperatures, reaction rates
Intermediate Weathering Stages approximately double for each 10C rise in tem-
perature.
6 Quartz Soils dominated by
7
dM
Muscovite (also illite) these minerals in the
2. At a given temperature, weathering proceeds
more rapidly in a wet climate than in a dry cli-
8 2⬊1 layer silicates (in- fine silt and clay frac- mate provided there is good drainage.
cluding vermiculite, tions are mainly those 3. The depth to the water table influences weather-
expanded hydrous of temperate regions ing by determining the depth to which air is
mica) developed under grass available as a gas or in solution and by its effect
Montmorillonite or trees. Includes the on the type of biotic activity.
hte
11 Gibbsite ered soils of the warm these geomorphologic processes control landform as-
12 Hematite (also geothite, and humid equatorial semblages.
limonite) regions have clay During the early stages of weathering and soil for-
13 Anatase (also rutile, fractions dominated mation, the parent material is much more important
than it is after intense weathering for long periods of
py
From Jackson and Sherman (1953). chemical durability to persist over long periods of
weathering. Quartz is most abundant in coarse-grained
granular rocks such as granite, granodiorite, and
gneiss, where it typically occurs in grains in the mil-
glasses. Mineral fragments in lunar soils include pla- limeter size range. Consequently, granitic rocks are the
gioclase feldspar, pyroxene, ilmenite, olivine, and po- main source of sand.
tassium feldspar. Quartz is extremely rare because it is In addition to the microbiological activities dis-
not abundant in the source rocks. Carrier et al. (1991) cussed previously, biological factors of importance in-
present an excellent compilation of information about clude the influences of vegetation on erosion rate and
the composition and properties of lunar soil. the cycling of elements between plants and soils. Mi-
Mean Mean
Annual Annual
Morphoclimatic Temperature Precipitation
Zone (C) (mm) Relative Importance of Geomorphologic Processes
Glacial ⬍0 0–1000 Mechanical weathering rates (especially frost action)
high; chemical weathering rates low, mass
movement rates low except locally; fluvial action
confined to seasonal melt; glacial action at a
l
maximum; wind action significant
ria
Periglacial ⫺1 to 2 100–1000 Mechanical weathering very active with frost action at
a maximum; chemical weathering rates low to
moderate; mass movement very active; fluvial
processes seasonally active; wind action rates
locally high. Effects of the repeated formation and
ate
decay of permafrost.
Wet midlatitude 0–20 400–1800 Chemical weathering rates moderate, increasing to
high at lower latitudes; mechanical weathering
activity moderate with frost action important at
higher latitudes; mass movement activity moderate
to high; moderate rates of fluvial processes; wind
dM action confined to coasts.
Dry continental 0–10 100–400 Chemical weathering rates low to moderate;
mechanical weathering, especially frost action,
seasonally active; mass movement moderate and
episodic; fluvial processes active in wet season;
wind action locally moderate.
hte
Hot dry (arid 10–30 0–300 Mechanical weathering rates high (especially salt
tropical) weathering), chemical weathering minimum, mass
movement minimal; rates of fluvial activity
generally very low but sporadically high; wind
action at maximum.
Hot semidry 10–30 300–600 Chemical weathering rates moderate to low;
rig
dry season.
Hot wet 20–30 ⬎1500 High potential rates of chemical weathering;
(humid mechanical weathering limited; active, highly
tropical) episodic mass movement; moderate to low rates of
stream corrosion but locally high rates of dissolved
and suspended load transport.
Azonal Highly Highly Rates of all processes vary significantly with altitude;
Mountain variable variable mechanical and glacial action becomes significant at
zone high elevations.
From Fookes et al. (2000).
l
manganese, and sulfide oxidation results, and carbon- ture and detailed characterization of these minerals are
ria
ate precipitation can occur (McCarty, 2004). covered in Chapter 3.
The time needed to weather different materials var-
ies greatly. The more unconsolidated and permeable Kaolinite Minerals
the parent material, and the warmer and more humid Kaolinite formation is favored when alumina is abun-
the climate, the shorter the time needed to achieve dant and silica is scarce because of the 1⬊1 sil-
ate
some given amount of soil formation. The rates of ica⬊alumina structure, as opposed to the 2⬊1 silica to
weathering and soil development decrease with in- alumina structure of the three-layer minerals. Condi-
creasing time. tions leading to kaolinite formation usually include low
The time for soil formation from hard rock parent electrolyte content, low pH, and the removal of ions
materials may be very great; however, young soils can
dM that tend to flocculate silica (Mg, Ca, Fe) by leaching.
develop in less than 100 years from loessial, glacial, Most kaolinite is formed from feldspars and micas by
and volcanic parent material (Millar et al., 1965). Py- acid leaching of acidic (SiO2-rich) granitic rocks. Ka-
rite bearing rocks are known to break apart and un- olinite forms in areas where precipitation is relatively
dergo chemical and mineral transformations in only a high, and there is good drainage to ensure leaching of
few years. cations and iron.
Halloysite forms as a result of the leaching of feld-
spar by H2SO4, which is often produced by the oxi-
hte
2.6 ORIGIN OF CLAY MINERALS AND CLAY dation of pyrite, as shown earlier. The combination of
GENESIS conditions that results in halloysite formation is often
found in high-rain volcanic areas such as Hawaii and
There are three general mechanisms of clay formation the Cascade Mountains of the Pacific Northwest in the
by weathering (Eberl, 1984): (1) inheritance, (2) neo- United States.
formation, and (3) transformation. Inheritance means
rig
that a clay mineral originated from reactions that oc- Smectite Minerals
curred in another area during a previous stage in the
rock cycle and that the clay is stable enough to remain Smectites, because of their 2⬊1 silica⬊alumina struc-
in its present environment. Origin by neoformation ture, form where silica is abundant, as is the case
means that the clay has precipitated from solution or where both silica and alumina are flocculated. Condi-
py
formed from reactions of amorphous materials. Trans- tions favoring this are high pH, high electrolyte con-
formation genesis requires that the clay has kept some tent, and the presence of more Mg2⫹ and Ca2⫹ than
of its inherited structure while undergoing chemical Na⫹ and K⫹. Rocks that are high in alkaline earths,
reactions. These reactions are typically characterized such as the basic and intermediate igneous rocks, vol-
Co
by ion exchange with the surrounding environment canic ash, and their derivatives containing ferromag-
and/or layer transformation in which the structure of nesian minerals and calcic plagioclase, are usual parent
octahedral, tetrahedral, or fixed interlayer cations is materials. Climatic conditions where evaporation ex-
modified. ceeds precipitation and where there is poor leaching
The behavior of nonclay colloids such as silica and and drainage, such as in arid and semiarid areas, favor
alumina during crystallization is important in deter- the formation of smectite.
mining the specific clay minerals that form. Certain
general principles apply.5 Illite (Hydrous Mica) and Vermiculite
Hydrous mica minerals form under conditions similar
to those leading to the formation of smectites. In ad-
5
The considerations in Chapter 6 provide a basis for these statements. dition, the presence of potassium is essential; so ig-
neous or metamorphic rocks and their derivatives are guish. Their thickness may range from a few milli-
the usual parent rocks. Weathering of feldspar in cool meters to several meters. The horizons may differ in
climates often leads to the development of illite. Al- any or all of the following ways:
teration of muscovite to illite and biotite to vermiculite
during weathering is also a significant source of these 1. Degree of breakdown of parent material
minerals. Interstratifications of vermiculite with mica 2. Content and character of organic material
and chlorite are common. The high stability of illite is 3. Kind and amount of secondary minerals
responsible for its abundance and persistence in soils 4. pH
and sediments. 5. Particle size distribution
Chlorite Minerals
All the horizons considered together, including the
l
underlying parent material, form the soil profile.6 The
Chlorites can form by alteration of smectite through
ria
part of the profile above the parent material is termed
introduction of sufficient Mg2⫹ to cause formation of the solum. Eluviation is the movement of soil material
a brucitelike layer that replaces the interlayer water. from one place to another within the soil, either in
Biotite from igneous and metamorphic rocks may alter solution or in suspension as a result of excess precip-
to trioctahedral chlorites and mixed-layer chlorite– itation over evaporation. Eluvial horizons have lost ma-
ate
vermiculite. Chlorites also occur in low- to medium- terial; illuvial horizons have gained material.
grade metamorphic rocks and in soils derived from Master horizons are designated by the capital letters
such rocks. O, A, B, C, and R (Table 2.4). Subordinate symbols
are used as suffixes after the master horizon designa-
Discussion tions to indicate dominant features of different kinds
of horizons, as indicated in the table. The O horizons
dM
The above considerations are greatly simplified, and
there are numerous ramifications, alterations, and var- are generally present at the soil surface under native
iations in the processes. One clay type may transform vegetation, but they may also be buried by sedimen-
to another by cation exchange and weathering under tation of alluvium, loess, or ash fall. The A horizon is
new conditions. Entire structures may change, for ex- the zone of eluviation where humified organic matter
ample, from 2⬊1 to 1⬊1, so that montmorillonite forms accumulates with the mineral fraction. The amount of
when magnesium-rich rocks weather under humid, organic matters (fibers to humic/fulvic acids) varies
hte
moderately drained conditions, but then alters to kao- from 0.1 percent in desert soils to 5 percent or more
linite as leaching continues. Kaolinite does not form in in organic soils and affects many engineering proper-
the presence of significant concentrations of calcium. ties including compressibility, shrinkage, strength and
The relative proportions of potassium and magne- chemical sorption. The B horizon is the zone of illu-
sium determine how much montmorillonite and illite viation where clay, iron compounds, some resistant
form. Some montmorillonites alter to illite in a marine minerals, cations, and humus accumulate. The R ho-
rig
environment due to the high K⫹ concentration. Mixed- rizon is the consolidated rock, and the C horizon con-
layer clays often form by partial leaching of K or sists of the altered material from which A and B
Mg(OH)2 from between illite and chlorite layers and horizons are formed.
by incomplete adsorption of K or Mg(OH)2 in mont- Soil profiles developed by weathering can be cate-
gorized into three groups on the basis of their miner-
py
morillonite or vermiculite.
Further details of the clay minerals are given in alogy and chemical composition as shown in Fig. 2.8
Chapter 3. More detailed discussions of clay mineral (Press and Siever, 1994). Pedalfers, which are formed
formation are given by Keller (1957, 1964a & b), Wea- in moist climate, are soils rich in aluminum and iron
ver and Pollard (1973), Eberl (1984), and Velde oxides and silicates such as quartz and clay minerals.
Co
Master Horizons
O1 Organic undecomposed horizon
O2 Organic decomposed horizon
A1 Organic accumulation in mineral soil horizon
A2 Leached bleached horizon (eluviated)
A3 Transition horizon to B
AB Transition horizon between A and B—more like A in upper part
A and B A2 with less than 50% of horizon occupied by spots of B
l
ria
AC Transition horizon, not dominated by either A or C
B and A B with less than 50% of horizon occupied by spots of A2
B Horizon with accumulation of clay, iron, cations, humus; residual
concentration of clay; coatings; or alterations of original
material forming clay and structure
ate
B1 Transition horizon more like B than A
B2 Maximum expression of B horizon
B3 Transitional horizon to C or R
C Altered material from which A and B horizons are presumed to be
formed
R Consolidated bedrock
dM Subordinate Symbols
b Buried horizon
ca Calcium in horizon
cs Gypsum in horizon
cn Concretions in horizon
hte
f Frozen horizon
g Gleyed horizon
h Humus in horizon
ir Iron accumulation in horizon
m Cemented horizon
p Plowed horizon
rig
in calcium from the calcium carbonates and other sol- rizon because most of the organic matter is recycled
uble minerals originated from sedimentary bedrock. from the surface to the vegetation.
Soil water is drawn up near the surface by evaporation, Lithologic discontinuities may be common in land-
leaving calcium carbonate pellets and nodules. They scapes where erosion is severe, and these discontinui-
can be found in the southwest United States. Laterite, ties are often marked by stone layers from previous
which is formed in a wet, tropical climate, is rich in erosion cycles. In some places, soils have developed
aluminum and iron oxides, iron-rich clays, and alu- several sequences of A and B horizons, which are su-
minum hydroxides. Silica and calcium carbonates are perimposed over each other. Superimposed soil se-
leached away from the soil. It has a very thin A ho- quences are likely the result of climate changes acting
l
present) Calcium insoluble iron and
ria
carbonate aluminum oxides;
Some iron and B pellets and occasional quartz
aluminium oxides nodules
B precipitated; all precipitated Iron-rich clays and
soluble materials,
such as aluminum
ate
carbonates, hydroxides
leached away
Figure 2.8 Major soil types: (a) Pedalfer soil profile developed on granite, (b) Pedocal soil
profile developed on sedimentary bedrock, and (c) Laterite soil profile developed on mafic
igneous rock (from Press and Siever, 1994).
hte
on uniform geologic materials, or are the remnants of chanical action. The transporting agents, for example,
former soil profiles (paleosoils) that have been buried water, wind, and ice, are by themselves capable only
under younger soils (Olson, 1981). of limited wearing action on rocks, but the process is
rig
Streams, ocean currents, waves, wind, groundwater, to erode than to transport particles. Particles are eroded
glaciers, and gravity continually erode and transport when the drag and lift of the fluid exceed the gravi-
soils and rock debris away from the zone of weather- tational, cohesive, and frictional forces acting to hold
ing. Each of these transporting agents may cause them in place. The stream velocity required to erode
Co
marked physical changes in the sediment it carries. Al- does not decrease indefinitely with decreasing particle
though detailed treatment of erosion, transportation, size because small particles remain within the bound-
and depositional processes is outside the scope of this ary layer adjacent to the stream bed where the actual
book, a brief outline of their principles and their effects stream velocity is much less than the average velocity.
on the transported soil is helpful in understanding the Relationships between particle size and average stream
properties of the transported material. velocity required to erode and transport particles by
wind and water are shown in Fig. 2.9.
Erosion Ice has the greatest competency for sediment move-
Erosion includes all processes of denudation that in- ment of all the transportation agents. There is no limit
volve the wearing away of the land surface by me- to the size of particles that may be carried. Ice pushes
l
portional to the square of the particle diameter. For
ria
larger particles and turbulent fluid flow, the settling ve-
locity is proportional to the square root of the particle
diameter. Particles stay in suspension once they have
been set in motion as long as the turbulence of the
ate
stream is greater than the settling velocity.
The largest particles that can be transported by water
are carried by traction, which consists of rolling and
Figure 2.9 Comparison of erosion and transport curves for dragging along the boundary between the transporting
air and running water. The air is a slightly more effective agent and the ground surface. Particles intermediate in
erosional agent than streams for very small particles but is
dM size between the suspended load and the traction load
ineffective for those larger than sand (from Garrels, 1951).
may be carried by saltation, in which they move by a
series of leaps and bounds. Soluble materials are car-
ried in solution and may precipitate as a result of
material along in front and erodes the bottom and sides changed conditions. The combined effects mean that
of the valleys through which it flows. In an active gla- the concentration of sediment is not constant through
cier (Fig. 2.10), there is continuous erosion and trans- the depth of the transporting agent but is much greater
near the stream bed than near the top. Fine particles
hte
Maximum
Size
Approximate Eroded by
Type of Average Average Max Load Type of Relative
Agent Flow Velocity Velocity Areas Affected per m3 Transport Effect
Streams Turbulent A few km/h Sand All land A few tens of Bed load, 1
kilograms suspended
load,
l
solution
ria
Waves Turbulent A few km/h Sand Coastlines A few tens of Same as 2
kilograms streams
Wind Turbulent 15 km/h Sand Arid, semiarid, A kilogram Bed load, 3
beaches, suspended
plowed fields load
ate
Glaciers Laminar A few m/yr Large High latitudes Hundreds of Bed load, 2
boulders and altitudes kilograms suspended
load,
surface
load
Groundwater Laminar A few m/yr Colloids Soluble material A kilogram Solution 3
dM and colloids
Gravity cm/yr to a Boulders Steep slopes, 2000 kg Bed load 3
few m/s sensitive
clays,
saturated
cohesionless
hte
soils,
unconsolidated
rock
Adapted from Garrels (1951).
rig
polishing of grains, and wind-driven impact can cause ated with a beach environment, produce a relatively
frosting of grains. The shape and surface character of smooth, pitted surface texture. Aeolian sands exhibit a
particles influences a soil’s stress–deformation and rougher surface texture, particularly over small dis-
strength properties owing to their effects on packing, tances. Some, but not all, river sands may have a very
py
volume change during shear, and interparticle friction. smooth particle surface that reflects the influence of
Basic minerals, such as the pyroxenes, amphiboles, chemical action. Sand that has undergone change after
and some feldspars, are rapidly broken down chemi- deposition and burial is termed diagenetic sand. Its
cally during transport. Quartz, which is quite stable surface texture may reflect a long and stable period of
Co
because of its resistant internal structure, may be mod- interaction with the groundwater. In some cases, very
ified by mechanical action, but only at a slow rate. rough surface textures can develop. Ottawa sand, a ma-
Quartz sand grains may survive a number of successive terial that has been used for numerous geotechnical
sedimentation cycles with no more than a percent or research investigations, is such a material.
two of weight loss due to abrasion. Some effects of transportation on sediment proper-
The surface textures of quartz sand particles reflect ties are summarized in Table 2.6. The gradational
their origin, as shown by the examples in Fig. 2.12 for characteristics of sedimentary materials reflect their
different sands, each shown to three or four magnifi- transportation mode as indicated in Fig. 2.13. Sedi-
cations. The mechanical and chemical actions, associ- ments of different origins lie within specific zones of
l
ria
ate
Figure 2.11 Schematic diagram of sediment concentration with depth in a transporting
stream.
dM
the figure, which are defined by the logarithm of the The thickness of deposits formed during any one
ratio of 75 percent particle size to 25 percent particle cycle may vary from less than a millimeter to hundreds
size and the median (50 percent) grain size. of meters. The period may range from months to
thousands of years, and only one or many types of
sediments may be involved.
hte
the bed load, then the particles will settle. When ice Spring and summer thaws contributed clay and silt-
melts, the sediments may be deposited in place or car- laden meltwater to the lake. The coarsest particles set-
ried away by meltwater. Materials in solution can tled first to form the summer layer. Because of the
precipitate when exposed to conditions of changed much slower settling velocity of the clay particles,
temperature or chemical composition, or as a result of most did not settle out until the quiet winter period. A
py
evaporation of water. Sediments may be divided into photograph of a vertical section through a varved clay
is shown in Fig. 2.14. The alternating coarser-grained,
those formed primarily by chemical and biological
light-colored layers and finer-grained, darker layers are
means and those composed primarily of mineral and
clearly visible. The shear resistance along horizontal
rock fragments. The latter are sometimes referred to as varves is much less than that across the varves. Also,
Co
detrital or clastic deposits. the hydraulic conductivity is much greater in the hor-
The deposition of sediments into most areas is cy- izontal direction than in the vertical direction. Exten-
clical. Some causes of cyclic deposition are: sive deposits of varved clays are found in the northeast
and north central United States and eastern Canada.
1. Periodic earth movements Detailed description of the geology and engineering
2. Climatic cycles of various lengths, most notably properties of Connecticut Valley varved clay is given
the annual rhythm by DeGroot and Lutenegger (2003).
3. Cyclic shifting of tributaries on a delta Complex soil deposition processes occur along
4. Periodic volcanism coastlines, estuaries, and shallow shelves in relation to
l
ria
ate
dM
hte
Figure 2.12 Surface textures of four sands of differing origins: (a) river sand, (b) beach
sand, (c) aeolian sand and (d) diagenetic sand (courtesy of Norris, 1975).
rig
the location of the shoreline. Soil deposits include fore- Chemical and biochemical sediments may consist of
shore sand and gravels, which are sorted by wave ac- one or two kinds of materials. For example, calcium
tions, organic deposits, and clays preserved in lagoons, carbonate sediments are made of calcite, which origi-
py
offshore fine sands, and muds. River channels may be nates from the shells of organisms in the deep sea (Fig.
overdeepened, and soft sediments then accumulate to 2.16a). Some clays contain significant amounts of mi-
form buried valleys. Most coastlines and estuaries of crofossils due to the depositional environment as
the world were subject to sea level changes in the Qua- shown in Fig. 2.16b; such clays include Mexico City
Co
ternary period. In particular, the post glacial rise of sea clay (Diaz-Rodriguez et al., 1998), Ariake clay (Oht-
level, which ended about 6000 years ago, has had a subo et al., 1995), and Osaka Bay clay (Tanaka and
worldwide influence on the present-day coastal forms. Locat, 1999). The microfossils include diatoms (sili-
Figure 2.15 shows alternating layers of marine (Ma) ceous skeleton of eukarya cells in either freshwater or
and fluvial (Diluvial-D) sediments in the geotechnical marine environments), radiolaria (found in marine en-
profile down to 400 m depth below sea level at Osaka vironments and consisting mostly of silica), and for-
Bay, Japan (Tanaka and Locat, 1999). The observed manifera (calcium carbonate shell secreted by marine
variation corresponds well to the local relative sea level eukarya). The presence of microfossils can have a pro-
during its geological history up to 1 million years ago. found effect on the behavior of the soil mass, confer-
l
ria
ate
dM
hte
ring unusual geotechnical properties that deviate from ternary period from glacial out wash and deglaciated
general property expectations, including high porosity, till areas. The deposits are spread widely and blanket
high liquid limit, unusual compressibility, and uniquely preexisting landforms. The deposits are up to 30 m
high friction angle. For examples, see Tanaka and Lo- thick in the Missouri and Rhine River Valleys, more
py
cat (1999) and Locat and Tanaka (2001). than 180 m thick in Tajikistan, and up to 330 m thick
While streams and rivers produce deposits according in northern China.
to grain size, a glacier transports the finest dust and
large boulders side by side at the same rate of move-
Co
l
Silt: little effect frosted
ria
surfaces
Sorting Considerable sorting Very considerable Very little sorting No sorting
sorting
(progressive)
ate
Adapted from Lambe and Whitman (1969).
dM
hte
rig
py
Co
Figure 2.13 Influence of geologic history on sorting of particle sizes (adapted from Selmer-
Olsen, 1964).
l
ria
ate
Figure 2.14 Vertical section through varved clay from the
New Jersey meadowlands (courtesy of S. Saxena).
dM
hte
l
Flow deposits Clay to gravel
ria
Marls Silt (fossils)
Shallow ocean Estuarine Silt and clay
Littoral Silt and sand
Shelf Silt and clay
Deep ocean Pelagic Silt and clay
ate
Oozes—calcareous Silt and clay
Oozes—siliceous Silt and clay
Flow Clay to gravel
Glacier Subglacial till Clay to boulders
Supraglacial till Sand to boulders
Residual Land Tropical soils Clay to sand
dM Saprolite Clay to boulders
Decomposed granite Clay to boulders
Colluvial soils Clay to boulders
Chemical and biochemical Lake Evaporites (sakkas)
Sea Evaporites
Limestone
hte
Gas hydrates
Adapted from Locat et al. (2003).
where the material is exposed to new conditions of solidation pressure greater than present overburden ef-
rig
temperature, pressure, and chemistry. An understand- fective pressure) at shallow depths. Other important
ing of postdepositional changes is essential for under- mechanisms include partial consolidation under in-
standing of properties, interpreting soil profile data, creased overburden and the effects of weathering.
and in reconstructing geologic history. A brief outline
py
Figure 2.17 Properties of Thames estuary clay. The overconsolidation in the upper 10 ft
was caused by surface drying (Skempton and Henkel, 1953).
27
l
ria
ate
dM
hte
rig
py
l
Erosion of overlying sediments due to glacial process ily determined quantitatively. It is known to contribute
ria
leads to mechanical overconsolidation. A typical ex- to clay sensitivity, and it may be responsible for an
ample of this is London clay, a marine clay deposited apparent preconsolidation pressure. Removal of iron
during the Eocene period. The erosion took place in compounds from a very sensitive clay from Labrador,
late Tertiary and Pleistocene times and the amount of Canada, by leaching led to a 30-ton/m2 decrease in
apparent preconsolidation pressure (Kenney et al.,
ate
erosion is estimated to be about 150 m in Essex
(Skempton, 1961) to 300 m in the Wraybury district 1967). Coop and Airey (2003) show for carbonate soils
(Bishop et al., 1965). After the unloading, small re- that cementation develops soon after deposition and
loading occurred by new deposition of gravels in the enables the soil to maintain a loose structure.
late Quaternary period. Within the London clay, five Failure to recognize cementation has resulted in con-
major transgressive–regressive cycles are recognized
dM struction disputes. For example, a soil on a major proj-
during its deposition. The postdepositional processes ect was marked on the contract drawings as glacial
are site specific; that is, the degree of weathering and till. It proved to be so hard that it had to be blasted.
desiccation and the amount of erosion vary depending The contractor claimed the soil was cemented because
on location. This variation in depositional and post- during digging failure took place through pebbles as
depositional processes results in complex mechanical well as the clay matrix. The owner concluded that this
behavior (Hight et al., 2003). happened because the pebbles were weathered. Proper
evaluation of the material before the award of the con-
hte
Authigenesis, Diagenesis, Cementation, and tract could have avoided the problem.
Recrystallization Clay particles adhere to the surfaces of larger silt
and sand particles, a process called clay bounding.
Authigenesis is the formation of new minerals in place Eventually the larger grains become embedded into a
after deposition. Authigenesis can make grains more clay matrix and their influence on the geotechnical be-
angular, lower the void ratio, and decrease the per- havior becomes limited. The clay bounding provides
rig
meability. Small crystals and rock fragments may grow arching of interparticle forces, maintaining a large void
into aggregates of coarser particles. ratio even at high effective stresses.
Diagenesis refers to such phenomena as changes in
particle surface texture, the conversion of minerals
from one type to another, and the formation of inter- Time Effects
py
particle bonds as a result of increased temperature, Even freshly deposited or densified sands can develop
pressure, and time. Many diagenetic changes are con- significant increases in strength and stiffness over rel-
trolled by the pH and redox potential of the deposi- atively short time periods, that is, by a factor of 2 or
tional environment. With increasing depth of burial in more within a few months (Mitchell and Solymar,
Co
a sedimentary basin, clayey sediments may undergo 1984). Time effects and the aging of both cohesive
substantial transformation. Expansive clay minerals and cohesionless soils are analyzed and reviewed
can transform to a nonexpansive form, for example, by Schmertmann (1991). Uncertainty remains as to
montmorillonite to mixed layer to illite, as a result of whether the mechanisms for the observed increases in
the progressive removal of water layers under pressure apparent preconsolidation pressure, strength, and stiff-
(Burst, 1969). Burial depths of 1000 to 5000 m may ness are chemical, physical, or both. Research is con-
be required, and the transformation process appears tinuing on this important aspect of soil behavior so that
thermally activated as a result of the increased tem- it will be possible to predict both the amount and the
perature at these depths. Chlorite can form in mud and rate of property changes for use in the analysis of geo-
shale during deep burial (Weaver and Pollard, 1973). technical problems. The aging process is of particular
interest in connection with hydraulic fills and ground shock. Cracks up to 2 ft wide, of unknown depth, and
improvement projects, more details are given in Chap- spaced several meters apart have caused damage to
ter 12. buildings and highways.
l
activity of microorganisms are controlled partly by
salts. This process is important in the formation of
ria
pore geometry and local physicochemical conditions.
highly sensitive, quick clays, as discussed in more de- Therefore, apart from its impact on life itself, biolog-
tail in Chapter 8. ical activity has influenced the evolution of the earth
Materials can be removed from sediments by differ- surface, impacted mineral, sediment, and rock forma-
ential solution and subsequent leaching. Calcareous tion, accelerated the rate of rock weathering and al-
ate
and gypsiferous sediments are particularly susceptible tered its products, influenced the composition of
to solution, resulting in the formation of channels, sink groundwater, and participated in the formation of gas
holes, and cavities. and petroleum hydrocarbons.
Bioturbance refers to the action of organisms living
Jointing and Fissuring of Clay Soils on or in sediments. By organic cementation, they mod-
dM
Some normally consolidated clays, almost all flood-
plain clays, and many preconsolidated clays are weak-
ify grain size, density, or cohesion (Richardson et al.,
1985; Locat et al., 2003). The aggregation activity of
ened by joints and fissures. Joints in floodplain clays various worms densifies deposits by changing the grain
result from deposition followed by cyclic expansion size of the sediment. Tubes that form can provide local
and contraction from wetting and drying. Joints and drainage and decrease the bulk density. The active zone
fissures in preconsolidated clays result from unloading of bioturbance is usually to depths less than 30 cm.
or from shrinkage cracks during drying. Closely Sticky organic mucus or polymer bridging binds to-
hte
spaced joints in these types of clays may contribute to gether clay–silt particles, producing clusters.
slides some years after excavation of cuts. The unload- Chemical transformation processes are mediated by
ing enables joints to open, water to enter, and the clay organisms. Some notable processes are summarized as
to soften. follows (Mitchell and Santamarina, 2005):
Fissures have been found in normally consolidated
clays at high water contents that could not have been 1. Sulfur Cycle Elemental sulfur (S0) and sulfides
rig
caused by drying or unloading (Skempton and Northey, (S2⫺) are the stable forms of sulfur under anaer-
1952), and increased brittleness has been observed in obic conditions, whereas sulfates (SO42⫺) are the
soft clay chunks that have been stored for some time. stable forms of sulfur under aerobic conditions.
These effects may be caused by syneresis, which is the Sulfides form under anaerobic conditions from
mutual attraction of clay particles to form closely knit sulfates already present in seawater and sedi-
py
aggregates with fissures between them. Similar behav- ments or introduced by diffusion and ground-
ior is many times observed in gelatin after aging. water flow. The sulfate ion is not reduced to
Weathering and the release of potassium may also re- sulfide at Earth surface temperature and pressure
sult in fissuring. unless biologically mediated. Sulfate-reducing
Co
Vegetation, especially large trees, can cause shrink- bacteria are anaerobic and grow best at neutral
age and fissuring of clays (Barber, 1958; Holtz, 1983). pH but are known to exist over a broad range of
The root systems suck up water, causing large capillary pH and salt content. When exposed to aerobic
shrinkage pressures. When rain falls on crusted surface conditions, reduced sulfur compounds, hydrogen
layers of dried-up saline lakes, it is rapidly absorbed sulfides (H2S), and elemental sulfur are used as
by capillarity. The air may become so compressed that an energy source by sulfide-oxidizing bacteria
it causes tension cracking or blowouts in a form similar and converted to sulfates.
in appearance to root holes. These sediments may also 2. Iron Cycle Iron in the subsurface exists pre-
undergo severe cracking, apparently as a result of dominantly in the reduced or ferrous (Fe2⫹) state
l
by humankind. On October 21, 1966, 144 people, 116
ria
low solubility, the formation of H2SO4 provides a
of them children, were killed when a tip of coal waste
source of important reactions in the solid and pore wa-
slid onto the village of Aberfan in South Wales, United
ter phases. The total dissolved solids increases owing
to the dissolution of carbonates in the soil. Gypsum Kingdom. The collapse was caused by tipping of coal
can form, with an associated volume increase, at the waste over a natural underground spring, and the coal
ate
expense of carbonate minerals. The precipitated ferric slag slowly turned into a liquid slurry. The tragedy was
hydroxide is thermodynamically unstable and rapidly caused by two days of continual heavy rain loosening
transforms to yellow goethite, FeO–OH. Geothite, the coal slag. As a result of the disaster at Aberfan,
while stable under wet conditions, will slowly dehy- the Mines and Quarries Tips Act of 1969 was intro-
drate to red hematite, Fe2O3, under dry conditions. duced. This act was passed in order to prevent disused
Microorganisms have a limited effect on the for-
dM tips from becoming a danger to members of the public.
mation of coarse grains. However, bioactivity can af- Over 8000 million tons of ore have been mined in
fect diagenetic evolution, promote the precipitation of the South African deep-level underground gold mining
cementing agents, cause internal weathering, and alter industry (Blight et al., 2000). Considerations for dis-
fines migration, filter performance, and drainage in posing these wastes into tailings ponds and dams in-
silts and sands. clude the physicochemical nature of the extracted
Severely water-limited environments distress micro- minerals as well as the topography and climate of the
organisms and hinder biological activity. Nonetheless, disposal sites. Tailings dams have failed, resulting in
hte
there is great bacterial activity in the unsaturated or- destructive mudflows (Blight, 1997). One reported case
ganic surface layer of a soil where plant roots are was the failure of the Merriespruit ring-dyke gold tail-
found. Fierer et al. (2002) observed that bacterial ac- ings dam in South Africa in 1994, which killed 17
tivity decreases by 1 or 2 orders of magnitude by 2 m people in a village nearby. Overtopping of the tailings
of depth. Horn and Meike (1995) conclude that micro- dyke occurred after a significant rainfall event, and ap-
rig
bial activity requires 60 to 80 percent saturation. proximately 500,000 m3 of tailings flowed through this
Hence, there is less reduction in bacterial count with breach. The liquefied tailings flowed for a distance of
depth in saturated sediments. Hindered biological ac- about 2 km. A large volume of tailings was in a me-
tivities in unsaturated soils may reflect lack of nutrients tastable state in situ, and overtopping and erosion of
in isolated water at menisci, slow nutrient flow in per- the impoundment wall exposed this material, resulting
py
colating water paths, and increased ionic concentration in static liquefaction of the tailings and a consequent
in the pore fluid as water evaporates and dissolved salts flow failure (Fourie et al., 2001).
approach ion saturation conditions. The urban underground in major cities is congested
The physical scales over which the physicochemical, by utility lines, tunnels, and building foundations.
Co
bioorganic, and burial diagenetic processes act range Much may be more than 100 years old; for example,
from atomic dimensions to kilometers, and the time- more than 50 percent of the water supply pipes in Lon-
scales range from microseconds to years. Table 2.8 don were built using cast-iron during Victorian time.
summarizes the processes, fabric characteristics, and Aging infrastructure changes the in situ stress condi-
scales associated with different mechanisms. tion, as well as groundwater chemistry, and this can
lead to changes in the stress–strain–time behavior of
Human Effects the subsoil. Underground openings are sources or sinks
The global human population has grown from approx- of different environments; tunnels can act as a ground-
imately 600 million at the beginning of the eighteenth water drain as well as source for air into the ground.
Table 2.8 Summary of Processes and of the Fabric Signature and Temporal Scales Associated with Various
Mechanisms
Fabric
Signaturesa Physical
Processes Mechanisms (predominant) Scales Time Remarks
Physicochemical Electromechanical EF Atomic and s to ms Two particles may rotate
molecular to FF
⬃ 4 m
Thermomechanical FF Molecular to ms to min Initial contacts EF then
l
(some EF) 0.2 mm rotations to FF:
ria
common in selective
environments
Interface dynamics FF and EF m to ⬃ 0.5 s Some large compound
mm particles may be
possible at high
ate
concentrations
Bioorganic Biomechanical EF ⬃ 0.5 mm to s to min Some FF possible
⬎ 2.0 mm during bioturbation
Biophysical EE and FF m to mm s to min Some very large clay
organic complexes
possible
dM
Biochemical Nonunique m to mm h to yr New chemicals formed,
(unknown) some altered
Burial Mass gravity FF localized cm to km yr Can operate over large
diagenesis swirl physical scales
Diagenesis- Nonunique Molecular yr New minerals formed,
cementation (unknown) some altered, changes
hte
in morphology
a
EF, edge-to-face; EE, edge-to-edge; FF, face-to-face.
Adapted from Mitchell and Santamarina (2005) and Bennett et al. (1991).
rig
Detailed studies of the geotechnical impacts of such to infer clay mineral types, to detect the presence of
problems have, so far, been limited (e.g., Gourvenec organic and high clay content layers, to locate borrow
et al., 2005), and further studies of the impacts of aging materials for construction, and to estimate the depth to
on existing infrastructure are needed. unaltered parent material. Pedological data can be used
py
aids in anticipating and understanding the probable and grain arrangement. Thus, knowledge of the trans-
composition, structure, properties, and behavior of a portation and deposition history provides insight into
soil. Along with site investigation data, characteri- geotechnical engineering properties.
zation of the landforms, that is, understanding of the In short, the soil and its properties with which we
former and current geomorphological processes asso- deal today are a direct and predictable consequence of
ciated with the past and present climatic conditions, the parent material of many years ago and of all the
often helps to define ground conditions for designing things that have happened to it since. The better our
geotechnical structures and anticipating the long-term knowledge of what that parent material was and what
performance. For example, the knowledge can be used the intervening events have been, the better our ability
l
physical and chemical weathering processes on the
for use as chemical, radioactive, and mixed (chem-
ria
mechanical and flow properties of the rocks and
ical and radioactive) waste containment barriers?
soils on which they act?
Why?
3. Describe the chemical reactions of pyrite oxidation
9. Prepare diagrams showing your estimates as a func-
and explain how bacteria can mediate the chemical
tion of elevation of the following soil characteristics
ate
processes.
that you would expect to encounter between the
4. Discuss what types of clay minerals are likely to be bottom and the top of Mount Kilimanjaro in Tan-
produced under each morphoclimatic zone listed in zania. Give a brief explanation for each.
Table 2.3. a. Soil plasticity
5. Using Stokes’s law, derive the sedimentation speeds
dM b. Soil gradation and mean particle size
of spherical particles with different sizes in fresh- c. Angularity–roundness of sand and gravel parti-
water under hydrostatic condition. Would they cles
change in saltwater? Compare the results to the data d. Iron content
given in Fig. 2.9 and discuss the comparison. e. Cementation between particles
6. List and discuss human activities that may poten- f. Organic matter content
tially change the properties of soils. g. Water content
hte
rig
py
Co
Soil Mineralogy
l
ria
ate
3.1 IMPORTANCE OF SOIL MINERALOGY IN naked eye. Clay can refer both to a size and to a class
GEOTECHNICAL ENGINEERING of minerals. As a size term, it refers to all constituents
of a soil smaller than a particular size, usually 0.002
Soil is composed of solid particles, liquid, and gas and mm (2 m) in engineering classifications. As a mineral
ranges from very soft, organic deposits through less term, it refers to specific clay minerals that are distin-
compressible clays and sands to soft rock. The solid
dM guished by (1) small particle size, (2) a net negative
particles vary in size from large boulders to minute electrical charge, (3) plasticity when mixed with water,
particles that are visible only with the aid of the elec- and (4) high weathering resistance. Clay minerals are
tron microscope. Particle shapes range from nearly primarily hydrous aluminum silicates. Not all clay par-
spherical, bulky grains to thin, flat plates and long, ticles are smaller than 2 m, and not all nonclay par-
slender needles. Some organic material and noncrys- ticles are coarser than 2 m; however, the amount of
talline inorganic components are found in most natural clay mineral in a soil is often closely approximated by
fine-grained soils. A soil may contain virtually any el- the amount of material finer than 2 m. Thus, it is
hte
ement contained in Earth’s crust; however, by far the useful to use the terms clay size and clay mineral con-
most abundant are oxygen, silicon, hydrogen, and alu- tent to avoid confusion. A further important difference
minum. These elements, along with calcium, sodium, between clay and nonclay minerals is that the nonclays
potassium, magnesium, and carbon, comprise over 99 are composed primarily of bulky particles; whereas,
percent of the solid mass of soils worldwide. Atoms the particles of most of the clay minerals are platy, and
in a few cases they are needle shaped or tubular.
rig
amount of clay. Nonetheless, clay and organic matter composed of water containing various types and
in a soil usually influence properties in a manner far amounts of dissolved electrolytes. Organic compounds,
greater than their abundance. both soluble and immiscible, are found in soils at sites
Mineralogy is the primary factor controlling the size,
Co
35
l
ria
ate
dM
hte
rig
py
Co
Figure 3.2 Characteristics of particles and particle dispersoids (adapted from Stanford Re-
search Institute Journal, Third Quarter, 1961).
Co
37
that have been affected by chemical spills, leaking tronic energy can jump to a higher level by the ab-
wastes, and contaminated groundwater. The gas phase, sorption of radiant energy or drop to a lower level by
in partially saturated soils, is usually air, although or- the emission of radiant energy. No more than two elec-
ganic gases may be present in zones of high biological trons in an atom can have the same energy level, and
activity or in chemically contaminated soils. the spins of these two electrons must be in opposite
The mechanical properties of soils depend directly directions. Different bonding characteristics for differ-
on interactions of these phases with each other and ent elements exist because of the combined effects of
with applied potentials (e.g., stress, hydraulic head, electronic energy quantization and the limitation on the
electrical potential, and temperature). Because of these number of electrons at each energy level.
interactions, we cannot understand soil behavior in An atom may be represented in simplified form by
terms of the solid particles alone. Nonetheless, the a small nucleus surrounded by diffuse concentric
l
structure of these particles tells us a great deal about ‘‘clouds’’ of electrons (Fig. 3.3). The maximum num-
ria
their surface characteristics and their potential inter- ber of electrons that may be located in each diffuse
actions with adjacent phases. shell is determined by quantum theory. The number
Interatomic and intermolecular bonding forces hold and arrangement of electrons in the outermost shell are
matter together. Unbalanced forces exist at phase of prime importance for the development of different
boundaries. The nature and magnitude of these forces types of interatomic bonding and crystal structure.
ate
influence the formation of soil minerals, the structure, Interatomic bonds form when electrons in adjacent
size, and shape of soil particles, and the physicochem- atoms interact in such a way that their energy levels
ical phenomena that determine engineering properties are lowered. If the energy reduction is large, then a
and behavior. In this chapter some aspects of atomic strong, primary bond develops. The way in which the
and intermolecular forces, crystal structure, structure
dM bonding electrons are localized in space determines
stability, and characteristics of surfaces that are perti- whether or not the bonds are directional. The strength
nent to the understanding of soil behavior are sum- and directionality of interatomic bonds, together with
marized simply and briefly. This is followed by a the relative sizes of the bonded atoms, determine the
somewhat more detailed treatment of soil minerals and type of crystal structure assumed by a given compo-
their characteristics. sition.
hte
l
are common in gases. If outer shell electrons are rep- Metallic bonds are nondirectional and can exist only
ria
resented by dots, then examples for (1) hydrogen gas, among a large group of atoms. It is the large group of
(2) methane, and (3) chlorine gas are: electrons and their freedom to move that make metals
such good conductors of electricity and heat. The me-
1. H ⫹ H ⫽ H:H tallic bond is of little importance in most soils.
ate
H
2. C ⫹ 4H ⫽ H:C:H
Bonding in Soil Minerals
H
A combination of ionic and covalent bonding is typical
⫹ Cl:
3. :Cl ⫽ :Cl:Cl:
in most nonmetallic solids. Purely ionic or covalent
bonding is a limiting condition that is the exception
rather than the rule in most cases. Silicate minerals are
dM
In the solid state, covalent bonds form primarily be-
tween nonmetallic atoms such as oxygen, chlorine, the most abundant constituents of most soils. The in-
teratomic bond in silica (SiO2) is about half covalent
nitrogen, and fluorine. Since only certain electrons
participate in the bonding, covalent bonds are direc- and half ionic.
tional. As a result, atoms bonded covalently pack in
such a way that there are fixed bond angles.
Ionic Bonds Ionic bonds form between positively 3.4 SECONDARY BONDS
hte
als, alkalies (e.g., sodium, potassium), and alkaline liquids and solid particles.
earths (e.g., calcium, magnesium) form cations. Anions
(negatively charged atoms that are attracted to the an- The Hydrogen Bond
ode) are those atoms requiring only a few electrons to If a hydrogen ion forms the positive end of a dipole,
complete their outer shell. Because the outer shells of then its attraction to the negative end of an adjacent
py
ions are complete, structures cannot form by electron molecule is termed a hydrogen bond. Hydrogen bonds
sharing as in the case of the covalent bond. Since ions form only between strongly electronegative atoms such
are electrically charged, however, strong electrical at- as oxygen and fluorine because these atoms produce
tractions (and repulsions) can develop between them. the strongest dipoles. When the electron is detached
Co
The ionic bond is nondirectional. Each cation at- from a hydrogen atom, such as when it combines with
tracts all neighboring anions. In sodium chloride, oxygen to form water, only a proton remains. As the
which is one of the best examples of ionic bonding, a electrons shared between the oxygen and hydrogen at-
sodium cation attracts as many chlorine anions as will oms spend most of their time between the atoms, the
fit around it. Geometric considerations and electrical oxygens act as the negative ends of dipoles, and the
neutrality determine the actual arrangement of ioni- hydrogen protons act as the positive ends. The positive
cally bonded atoms. and negative ends of adjacent water molecules tie them
As ionic bonding causes a separation between the together forming water and ice.
centers of positive and negative charge in a molecule, The strength of the hydrogen bond is much greater
the molecule will orient in an electrical field forming than that of other secondary bonds because of the small
a dipole. The strength of this dipole is expressed in size of the hydrogen ion. Hydrogen bonds are impor-
tant in determining some of the characteristics of the Examples of some common crystals are shown in
clay minerals and in the interaction between soil par- Fig. 3.4.
ticle surfaces and water.
Characteristics of Crystals
van der Waals Bonds Certain crystal characteristics are used to distinguish
different classes or groups of minerals. Variations in
Permanent dipole bonds such as the hydrogen bond are these characteristics result in different properties.
directional. Fluctuating dipole bonds, commonly
termed van der Waals bonds, also exist because at any 1. Structure The atoms in a crystal are arranged
one time there may be more electrons on one side of in a definite orderly manner to form a three-
the atomic nucleus than on the other. This creates weak dimensional network termed a lattice. Positions
l
instantaneous dipoles whose oppositely charged ends within the lattice where atoms or atomic groups
ria
attract each other.
Although individual van der Waals bonds are weak,
typically an order of magnitude weaker than a hydro-
gen bond, they are nondirectional and additive between
ate
atoms. Consequently, they decrease less rapidly with
distance than primary valence and hydrogen bonds
when there are large groups of atoms. They are strong
enough to determine the final arrangements of groups
of atoms in some solids (e.g., many polymers), and
they may be responsible for small cohesions in fine-
dM
grained soils. Van der Waals forces are described fur-
ther in Chapter 7.
Crystal Formation
py
l
ria
ate
dM
hte
rig
py
Co
Figure 3.5 Unit cells of the 14 Bravais space lattices. The capital letters refer to the type
of cell: P, primitive cell; C, cell with a lattice point in the center of two parallel faces; F,
cell with a lattice point in the center of each face; I, cell with a lattice point in the center
of the interior; R, rhombohedral primitive cell. All points indicated are lattice points. There
is no general agreement on the unit cell to use for the hexagonal Bravais lattice; some prefer
the P cell shown with solid lines, and others prefer the C cell shown in dashed lines (modified
from Moffatt et al., 1965).
l
sizes of most soil particles.
ria
4. X-ray and Electron Diffraction The orderly
atomic arrangements in crystals cause them to
behave with respect to X-ray and electron beams
in much the same way as does a diffraction grat-
ing with respect to visible light. Different crystals
ate
yield different diffraction patterns. This makes X-
ray diffraction a powerful tool for the study and
identification of very small particles, such as clay
that cannot be seen using optical means.
5. Symmetry There are 32 distinct crystal classes
dM
based on symmetry considerations involving the
arrangement and orientation of crystal faces.
These 32 classes may be grouped into 6 crystal
systems with the classes within each system bear-
ing close relationships to each other.
Figure 3.6 The six crystal systems.
The six crystal systems are illustrated in Fig. 3.6.
Crystallographic axes parallel to the intersection edges
hte
axis perpendicular to the other three and of dif- with the distances that separate parallel planes is im-
ferent length. Examples are quartz, brucite, cal- portant for the identification and classification of dif-
cite, and beryl. ferent minerals. All lengths are expressed in terms of
Tetragonal System There are three mutually per- unit cell lengths. Any plane through a crystal may be
pendicular axes, with two horizontal of equal described by intercepts, in terms of unit cell lengths,
length, but different than that of the vertical axis. on the three or four crystallographic axes for the sys-
Zircon is an example. tem in which the crystal falls. The reciprocals of these
Orthorhombic System There are three mutually intercepts are used to index the plane. Reciprocals are
perpendicular axes, each of different length. Ex- used to avoid fractions and to account for planes par-
amples include sulfur, anhydrite, barite, diaspore, allel to an axis (an intercept of infinity equals an index
and topaz. value of 0).
l
ria
ate
dM
hte
rig
py
Co
Figure 3.7 Miller indices: (a) Unit cell of muscovite, (b) (002) plane for muscovite, (c)
(014) plane for muscovite, and (d) (623) plane for muscovite.
theses are always used to indicate crystallographic Table 3.1 Atomic Packing, Structure, and Structural
planes, whereas brackets are used to indicate direc- Stability
tions. For example, [111] designates line oq in Fig.
3.7a. Additional examples of Miller indices for planes Radius
through the muscovite crystal are shown in Figs. 3.7b, Ratioa Nb Geometry Example Stability
3.7c, and 3.7d. A plane that cuts a negative axis is
designated by placing a bar over the index that pertains 0–0.155 2 Line — —
to the negative intercept (Fig. 3.7d). The general index 0.155– 3 Triangle (CO3)2⫺ Very high
(hkl) is used to refer to any plane that cuts all three 0.225
axes. Similarly (h00) designates a plane cutting only 0.225– 4 Tetrahedron (SiO4)4⫺ Moderately
the a axis, (h0l) designates a plane parallel to the b 0.414 high
l
axis, and so on. For crystals in the hexagonal system, 0.414– 6 Octahedron [Al(OH)6]3⫺ High
ria
the Miller index contains four numbers. The (001) 0.732
planes of soil minerals are of particular interest be- 0.732– 8 Body-cen- Iron Low
cause they are indicative of specific clay mineral types. 1.0 tered cube
1.0 12 Sheet K–O bond Very low
in mica
ate
3.7 FACTORS CONTROLLING CRYSTAL a
Range of cation to anion diameter ratios over which
STRUCTURES stable coordination is expected.
b
Coordination number.
Organized crystal structures do not develop by chance.
The most stable arrangement of atoms in a crystal is
dM
that which minimizes the energy per unit volume. This
is achieved by preserving electrical neutrality, satisfy- Table 3.2 Relative Stabilities of Some Soil Mineral
ing bond directionality, minimizing strong ion repul- Structural Units
sions, and packing atoms closely together.
If the interatomic bonding is nondirectional, then the Approximate
relative atomic sizes have a controlling influence on Relative Bond
packing. The closest possible packing will maximize Strength
hte
the number of bonds per unit volume and minimize the Structural Unit (Valence/N)
bonding energy. If interatomic bonds are directional,
as is the case for covalent bonds, then both bond angles Silicon tetrahedron, (SiO4)4⫺ 4/4 ⫽ 1
and atomic size are important. Aluminum tetrahedron, [Al(OH)4]1⫺ 3/4
Anions are usually larger than cations because of Aluminum octahedron, [Al(OH)6]3⫺ 3/6 ⫽ 1/2
electron transfer from cations to anions. The number Magnesium octahedron, [Mg(OH)6]4⫺ 2/6 ⫽ 1/3
rig
ratio of cationic to anionic radii, coordination number, trically neutral. In crystals formed by ionic bonded pol-
and the geometry formed by the anions are indicated yhedra, the packing maintains electrical neutrality and
in Table 3.1. minimizes strong repulsions between ions with like
Most solids do not have bonds that are completely charge. In such cases, the valence of the central cation
Co
nondirectional, and the second nearest neighbors may equals the total charge of the coordinated anions, and
influence packing as well as the nearest neighbors. the unit is really a molecule. Units of this type are held
Even so, the predicted and observed coordination num- together by weaker, secondary bonds. An example is
bers are in quite good agreement for many materials. brucite, a mineral that has the composition Mg(OH)2.
The valence of the cation divided by the number of The Mg2⫹ ions are in octahedral coordination with six
coordinated anions is an approximate indication of the (OH)⫺ ions forming a sheet structure in such a way
relative bond strength, which, in turn, is related to the that each (OH)⫺ is shared by 3Mg2⫹. In a sheet con-
structural stability of the unit. Some of the structural taining N Mg2⫹ ions, therefore, there must be 6N/3 ⫽
units common in soil minerals and their relative bond 2N (OH)⫺ ions. Thus, electrical neutrality results, and
strengths are listed in Table 3.2. the sheet is in reality a large molecule. Successive oc-
l
then the units arrange in a variety of ways in response was formed with its present proportions of the different
ria
to the repulsions. The silicon cation is in this category. cations in the structure.
Double chains of indefinite length may form with
(Si4O11)6⫺ as part of the structure. The amphiboles fall
3.8 SILICATE CRYSTALS into this group (Fig. 3.8). Hornblendes have the same
basic structure, but some of the Si4⫹ positions are filled
ate
Small cations form structures with coordination num- by Al3⫹. The cations Na⫹ and K⫹ can be incorporated
bers of 3 and 4 (Table 3.1). These cations are often into the structure to satisfy electrical neutrality; Al3⫹,
highly charged and generate strong repulsions between Fe3⫹, Fe2⫹, and Mn2⫹ can replace part of the Mg2⫹ in
adjacent triangles or tetrahedra. As a result, such struc- sixfold coordination, and the (OH)⫺ group can be re-
tures share only corners and possibly edges, but never
dM placed by F⫺.
faces, since to do so would bring the cations too close In sheet silicates three of the four oxygens of each
together. The radius of silicon is only 0.039 nm, tetrahedron are shared to give structures containing
whereas that of oxygen is 0.132 nm. Thus silicon and (Si2O5)2⫺. The micas, chlorites, and many of the clay
oxygen combine in tetrahedral coordination, with the minerals contain silica in a sheet structure. Framework
silicon occupying the space at the center of the tetra- silicates result when all four of the oxygens are shared
hedron formed by the four oxygens. The tetrahedral with other tetrahedra. The most common example is
arrangement satisfies both the directionality of the quartz. In quartz, the silica tetrahedra are grouped to
hte
bonds (the Si–O bond is about half covalent and half form spirals. The feldspars also have three-dimensional
ionic) and the geometry imposed by the radius ratio. framework structures. Some of the silicon positions are
Silicon is very abundant in Earth’s crust, amounting to filled by aluminum, and the excess negative charge
about 25 percent by weight, but only 0.8 percent by thus created is balanced by cations of high coordina-
volume. Almost half of igneous rock by weight and tion such as potassium, calcium, sodium, and barium.
91.8 percent by volume is oxygen. Differences in the amounts of this isomorphous sub-
rig
Silica tetrahedra join only at their corners, and stitution are responsible for the different members of
sometimes not at all. Thus many crystal structures are the feldspar family.
possible, and there is a large number of silicate min-
erals. Silicate minerals are classified according to how
the silica tetrahedra (SiO4)4⫺ associate with each other,
py
3.9 SURFACES
as shown in Fig. 3.8. The tetrahedral combinations in-
crease in complexity from the beginning to the end of All liquids and solids terminate at a surface, or phase
the figure. The structural stability increases in the same boundary, on the other side of which is matter of a
direction. different composition or state. In solids, atoms are
Co
Island (independent) silicates are those in which the bonded into a three-dimensional structure, and the ter-
tetrahedra are not joined to each other. Instead, the four mination of this structure at a surface, or phase bound-
excess oxygen electrons are bonded to other positive ary, produces unsatisfied force fields. In a fine-grained
ions in the crystal structure. In the olivine group, the particulate material such as clay soil the surface area
minerals have the composition R22⫹ SiO44⫺. Garnets may be very large relative to the mass of the material,
contain cations of different valences and coordination and, as is emphasized throughout this book, the influ-
numbers R32⫹ R23⫹(SiO4)3. The negative charge of the ences of the surface forces on properties and behavior
SiO4 group in zircon is all balanced by the single Zr4⫹. may be very large.
Ring and chain silicates are formed when corners of Unsatisfied forces at solid surfaces may be balanced
tetrahedra are shared. The formulas for these structures in any of the following ways:
l
ria
ate
dM
hte
rig
py
Co
Figure 3.8 Silica tetrahedral arrangements in different silicate mineral structures. Reprinted
Gillott (1968) with permission from Elsevier Science Publishers BV.
l
ria
ate
dM
hte
rig
py
Co
1. Attraction and adsorption of molecules from the Each unsatisfied bond force is significant relative to
adjacent phase the weight of atoms and molecules. The actual mag-
2. Cohesion with the surface of another mass of the nitude of 10⫺11 N or less, however, is infinitesimal
same substance compared to the weight of a piece of gravel or a grain
3. Solid-state adjustments of the structure beneath of sand. On the other hand, consider the effect of re-
the surface. ducing particle size. A cube 10 mm on an edge has a
surface area of 6.0 ⫻ 10⫺4 m2. If it is cut in half in 1. Very abundant in the source material
the three directions, eight cubes result, each 5 mm on 2. Highly resistant to weathering, abrasion, and im-
an edge. The surface area now is 12.0 ⫻ 10⫺4 m2. If pact
the cubes are further divided to 1 m on an edge, the 3. Weathering products
surface becomes 6.0 m2 for the same 1000 mm3 of
material. Thus, as a solid is subdivided into smaller The nonclays are predominantly rock fragments or
and smaller units, the proportion of surface area to mineral grains of the common rock-forming minerals.
weight becomes larger and larger. For a given particle In igneous rocks, which are the original source mate-
shape, the ratio of surface area to volume is inversely rial for many soils, the most prevalent minerals are the
proportional to some effective particle diameter. feldspars (about 60 percent) and the pyroxenes and
For many materials when particle size is reduced to amphiboles (about 17 percent). Quartz accounts for
l
1 or 2 m or less the surface forces begin to exert a about 12 percent of these rocks, micas for 4 percent,
ria
distinct influence on the behavior. Study of the behav- and other minerals for about 8 percent.
ior of particles of this size and less requires consider- However, in most soils, quartz is by far the most
ations of colloidal and surface chemistry. Most clay abundant mineral, with small amounts of feldspar and
particles behave as colloids, both because of their mica also present. Pyroxenes and amphiboles are sel-
small size and because they have unbalanced surface dom found in significant amounts. Carbonate minerals,
ate
electrical forces as a result of isomorphous substitu- mainly calcite and dolomite, are also found in some
tions within their structure. soils and can occur as bulky particles, shells, precipi-
Montmorillonite, which is one of the members of tates, or in solution. Carbonates dominate the compo-
the smectite clay mineral group (see Section 3.17), sition of some deep-sea sediments. Sulfates, in various
may break down into particles that are only 1 unit cell forms, are found primarily in soils of semiarid and arid
regions, with gypsum (CaSO4 2H2O) being the most
dM
thick (1.0 nm) when in a dispersed state and have a
specific surface area of 800 m2 /g. If all particles con- common. Iron and aluminum oxides are abundant in
tained in about 10 g of this clay could be spread out residual soils of tropical regions.
side by side, they would cover a football field. Quartz is composed of silica tetrahedra grouped to
form spirals, with all tetrahedral oxygens bonded to
silicon. The tetrahedral structure has a high stability.
In addition, the spiral grouping of tetrahedra produces
hte
3.10 GRAVEL, SAND, AND SILT PARTICLES a structure without cleavage planes, quartz is already
an oxide, there are no weakly bonded ions in the struc-
The physical characteristics of cohesionless soils, that ture, and the mineral has high hardness. Collectively,
is, gravel, sand, and nonplastic silts, are determined these factors account for the high persistence of quartz
primarily by particle size, shape, surface texture, and in soils.
size distribution. The mineral composition determines Feldspars are silicate minerals with a three-
rig
hardness, cleavage, and resistance to physical and dimensional framework structure in which part of the
chemical breakdown. Some carbonate and sulfate min- silicon is replaced by aluminum. The excess negative
erals, such as calcite and gypsum, are sufficiently sol- charge resulting from this replacement is balanced by
uble that their decomposition may be significant within cations such as potassium, calcium, sodium, strontium,
the time frame of many projects. In many cases, how- and barium. As these cations are relatively large, their
py
ever, the nonclay particles may be treated as relatively coordination number is also large. This results in an
inert, with interactions that are predominantly physical open structure with low bond strengths between units.
in nature. Evidence of this is provided by the soils on Consequently, there are cleavage planes, the hardness
the Moon. Lunar soils have a silty, fine sand gradation; is only moderate, and feldspars are relatively easily
Co
however, their compositions are totally different than broken down. This accounts for their lack of abun-
those of terrestrial soils of the same gradation. The dance in soils compared to their abundance in igneous
engineering properties of the two materials are sur- rocks.
prisingly similar, however. Mica has a sheet structure composed of tetrahedral
The gravel, sand, and most of the silt fraction in a and octahedral units. Sheets are stacked one on the
soil are composed of bulky, nonclay particles. As most other and held together primarily by potassium ions in
soils are the products of the breakdown of preexisting 12-fold coordination that provide an electrostatic bond
rocks and soils, they are weathering products. Thus, of moderate strength. In comparison with the intralayer
the predominant mineral constituents of any soil are bonds, however, this bond is weak, which accounts for
those that are one or more of the following: the perfect basal cleavage of mica. As a result of the
l
develop preferred orientations, which can be respon- in civil engineering projects. Biochemical processes
ria
sible for anisotropic properties within a soil mass. The form marl, which ranges from relatively pure calcium
surface texture of the grains influences the stress– carbonate to mixtures with mud and organic matter.
deformation and strength properties. Iron oxide is formed in some lakes. Diatomite or dia-
tomaceous earth is essentially pure silica formed from
the skeletal remains of small (up to a few tenths of a
ate
millimeter) freshwater and saltwater organisms. Owing
3.11 SOIL MINERALS AND MATERIALS to their solubility limestone, calcite, gypsum, and other
FORMED BY BIOGENIC AND GEOCHEMICAL salts may cause special geotechnical problems.
PROCESSES Oxidation and reduction of pyrite-bearing earth ma-
terials, that is, soils and rocks containing FeS2, can be
Evaporite deposits formed by precipitation of salts
dM the source of many types of geotechnical problems,
from salt lakes and seas as a result of the evaporation including ground heave, high swell pressures, forma-
of water are sometimes found in layers that are several tion of acid drainage, damage to concrete, and corro-
meters thick. The major constituents of seawater and sion of steel (Bryant et al., 2003). The chemical and
their relative proportions are listed in Table 3.3. Also biological processes and consequences of pyritic re-
listed are some of the more important evaporite de- actions are covered in Sections 8.3, 8.11, and 8.16.
More than 12 percent of Canada is covered by a
peaty material, termed muskeg, composed almost en-
hte
CHARACTERISTICS
Important compositional, structural, and morphological
characteristics of the important nonclay minerals found
in soils are summarized in Table 3.4. Of these miner-
Co
Figure 3.9 Swelling index as a function of mica content for Clay minerals in soils belong to the mineral family
coarse-grained mixtures (data from Terzaghi, 1931). termed phyllosilicates, which also contains other layer
l
ria
ate
dM
hte
Figure 3.10 Photomicrographs of sand and silt particles from several soils: (a) Ottawa stan-
dard sand, (b) Monterey sand, (c) Sacramento River sand, (d) Eliot sand, and (e) lunar soil
mineral grains (photo courtesy Johnson Space Center). Squares in background area are 1⫻1
mm. (ƒ) Recrystallized breccia particles from lunar soil (photo courtesy of NASA Johnson
Space Center). Squares in background grid are 1⫻1 mm.
rig
silicates such as serpentine, pyrophyllite, talc, mica, units and the manner in which two successive two- or
and chlorite. Clay minerals occur in small particle three-sheet layers are held together.
py
sizes, and their unit cells ordinarily have a residual Differences among minerals within clay mineral
negative charge that is balanced by the adsorption of groups result primarily from differences in the type and
cations from solution. amount of isomorphous substitution within the crystal
The structures of the common layer silicates are structure. Possible substitutions are nearly endless in
Co
made up of combinations of two simple structural number, and the crystal structure arrangement may
units, the silicon tetrahedron (Fig. 3.12) and the alu- range from very poor to nearly perfect. Fortunately for
minum or magnesium octahedron (Fig. 3.13). Different engineering purposes, knowledge of the structural and
clay mineral groups are characterized by the stacking compositional characteristics of each group, without
arrangements of sheets1 (sometimes chains) of these detailed study of the subtleties of each specific mineral,
is adequate.
1
In conformity with the nomenclature of the Clay Minerals Society Silica Sheet
(Bailey et al., 1971), the following terms are used: a plane of atoms,
a sheet of basic structural units, and a layer of unit cells composed In most clay mineral structures, the silica tetrahedra
of two, three, or four sheets. are interconnected in a sheet structure. Three of the
l
ria
ate
dM Figure 3.11 Sand and silt size particle shapes as seen in
silhouette.
Silica Chains
hte
the tips all point in the same direction. The structure minum in octahedral coordination with oxygens or hy-
has the composition (Si4O10)4⫺ and can repeat indefi- droxyls. In some cases, other cations are present in
nitely. Electrical neutrality can be obtained by replace- place of Al3⫹ and Mg2⫹, such as Fe2⫹, Fe3⫹, Mn2⫹,
ment of four oxygens by hydroxyls or by union with Ti4⫹, Ni2⫹, Cr3⫹, and Li⫹. Figure 3.13b is a schematic
Co
a sheet of different composition that is positively diagram of such a sheet structure. The oxygen-to-
charged. The oxygen-to-oxygen distance is 2.55 ang- oxygen distance is 2.60 Å, and the space available for
stroms (Å),2 the space available for the silicon ion is the octahedrally coordinated cation is 0.61 Å. The
0.55 Å, and the thickness of the sheet in clay mineral thickness of the sheet is 5.05 Å in clays (Grim, 1968).
structures is 4.63 Å (Grim, 1968). If the cation is trivalent, then normally only two-
thirds of the possible cationic spaces are filled, and the
structure is termed dioctahedral. In the case of alu-
2
minum, the composition is Al2(OH)6. This composition
In conformity with the SI system of units, lengths should be given
in nanometers. For convenience, however, the angstrom unit is re- and structure form the mineral gibbsite. When com-
tained for atomic dimensions, where 1 Å ⫽ 0.1 nm. bined with silica sheets, as is the case in clay mineral
Percent by Weight
Ion Grams per Liter of Total Solids Important Evaporite Deposits
Sodium, Na⫹ 10.56 30.61 Anhydrite CaSO4
Magnesium, Mg2⫹ 1.27 3.69 Barite BaSO4
Calcium, Ca2⫹ 0.40 1.16 Celesite SrSO4
Potassium, K⫹ 0.38 1.10 Kieserite MgSO4 H2O
Strontium, Sr2⫹ 0.013 0.04 Gypsum CaSO4 2H2O
Chloride, Cl⫺ 18.98 55.04 Polyhalite Ca2K2Mg(SO4) 2H2O
Ma2Mg(SO4)2 4H2O
l
Sulfate, SO42⫺ 2.65 7.68 Bloedite
ria
Bicarbonate, HCO3⫺ 0.14 0.41 Hexahydrite MgSO4 6H2O
Bromide, Br⫺ 0.065 0.19 Epsomite MgSO4 7H2O
Fluoride, F⫺ 0.001 — Kainite K4Mg4(Cl/SO4) 1 1H2O
Boric Acid, H3BO3 0.026 0.08 Halite NaCl
34.485 100.00 Sylvite KCl
ate
Flourite CaF2
Bischofite MgCl2 6H2O
Carnallite KMgCl3 6H2O
Adapted from data by Degens (1965).
dM
structures, an aluminum octahedral sheet is referred to units does not necessarily form the naturally occurring
as a gibbsite sheet. minerals. The ‘‘building block’’ approach is useful,
If the octahedrally coordinated cation is divalent, however, for the development of conceptual models.
then normally all possible cation sites are occupied and
the structure is trioctahedral. In the case of magne-
hte
sium, the composition is Mg3(OH)6, giving the mineral 3.14 SYNTHESIS PATTERN AND
brucite. In clay mineral structures, a sheet of magne- CLASSIFICATION OF THE CLAY MINERALS
sium octahedra is termed a brucite sheet.
Schematic representations of the sheets are useful The manner in which atoms are assembled into tetra-
for simplified diagrams of the structures of the differ- hedral and octahedral units, followed by the formation
ent clay minerals:
rig
Silica sheet or
Occurrence
in Soils of
Crystal Particle Specific Engineering
Mineral Formula System Cleavage Shape Gravity Hardness Interest
l
Plagioclase NaAlSi3O8 Triclinic 2 planes Bulky— 2.62–2.76 6 Common
ria
feldspar CaAl2Si3O8 (variable) elongate
Muscovite Kal3Si3O10(OH)2 Monoclinic Perfect basal Thin plates 2.76–3.1 2–21⁄2 Common
mica
Biotite mica K(Mg,FE)3AlSi3O10(OH)2 Monoclinic Perfect basal Thin plates 2.8–3.2 21⁄2–3 Common
Hornblende Na,Ca,Mg,Fe,Al silicate Monoclinic Perfect Prismatic 3.2 5–6 Uncommon
ate
prismatic
Augite Ca(Mg,Fe,Al)(Al,Si)2O6 Monoclinic Good prismatic Prismatic 3.2–3.4 5–6 Uncommon
(pyroxene)
Olivine (Mg,Fe)2SiO4 Orthorhombic Conchoidal Bulky 3.27–3.37 61⁄2–7 Uncommon
fracture
Calcite CaCO3 Hexagonal Perfect Bulky 2.72 21⁄2–3 May be
dM abundant
locally
Dolomite CaMg(CO3)2 Hexagonal Perfect Bulky 2.85 31⁄2–4 May be
rhombohedral abundant
locally
Gypsum CaSO4 2H2O Monoclinic 4 planes Elongate 2.32 2 May be
hte
abundant
locally
Pyrite FeS2 Isometric Cubical Bulky cubic 5.02 6–61⁄2
Data from Hurlbut (1957).
rig
py
Co
Figure 3.12 Silicon tetrahedron and silica tetrahedra arranged in a hexagonal network.
l
ria
ate
dM
hte
rig
py
erally similar engineering properties. The minerals structure. Such material is referred to as allophane or
have unit cells consisting of two, three, or four sheets. noncrystalline clay.
The two-sheet minerals are made up of a silica sheet The bottom row of Fig. 3.15 shows that the 2⬊1
and an octahedral sheet. The unit layer of the three- minerals differ from each other mainly in the type and
sheet minerals is composed of either a dioctahedral or amount of ‘‘glue’’ that holds the successive layers to-
trioctahedral sheet sandwiched between two silica gether. For example, smectite has loosely held cations
sheets. Unit layers may be stacked closely together or between the layers, illite contains firmly fixed potas-
water layers may intervene. The four-sheet structure of sium ions, and vermiculite has somewhat organized
chlorite is composed of a 2⬊1 layer plus an interlayer layers of water and cations. The chlorite group repre-
hydroxide sheet. In some soils, inorganic, claylike ma- sents an end member that has 2⬊1 layers bonded by an
terial is found that has no clearly identifiable crystal organized hydroxide sheet. The charge per formula
l
ria
ate
Figure 3.15 Synthesis pattern for the clay minerals.
dM
unit is variable both within and among groups, and clay mineral layers. Bonding between these sheets is
reflects the fact that the range of compositions is great of the primary valence type and is very strong. How-
owing to varying amounts of isomorphous substitution. ever, the bonds holding the unit layers together may
Accordingly, the boundaries between groups are some- be of several types, and they may be sufficiently weak
what arbitrary. that the physical and chemical behavior of the clay is
influenced by the response of these bonds to changes
hte
filled, and all of the cations are aluminum. In an ideal of the particles. Many of these cations are exchange-
brucite sheet, all the octahedral spaces are filled by able cations because they may be replaced by cations
magnesium. In an ideal silica sheet, silicons occupy all of another type. The quantity of exchangeable cations
tetrahedral spaces. In clay minerals, however, some of is termed the cation exchange capacity (cec) and is
py
the tetrahedral and octahedral spaces are occupied by usually expressed as milliequivalents (meq)3 per 100 g
cations other than those in the ideal structure. Common of dry clay.
examples are aluminum in place of silicon, magnesium Five types of interlayer bonding are possible in the
instead of aluminum, and ferrous iron (Fe2⫹) for mag- layer silicates (Marshall, 1964).
nesium. This presence in an octahedral or tetrahedral
Co
position of a cation other than that normally found, 1. Neutral parallel layers are held by van der Waals
without change in crystal structure, is isomorphous forces. Bonding is weak; however, stable crystals
substitution. The actual tetrahedral and octahedral cat- of appreciable thickness such as the nonclay min-
ion distributions may develop during initial formation
or subsequent alteration of the mineral.
erals of pyrophyllite and talc may form. These layers, which are greater in these minerals be-
minerals cleave parallel to the layers. cause of a smaller interlayer distance.
2. In some minerals (e.g., kaolinite, brucite, gibb-
site), there are opposing layers of oxygens and Whatever the reason, the smectite minerals are the
hydroxyls or hydroxyls and hydroxyls. Hydrogen dominant source of swelling in the expansive soils that
bonding then develops between the layers as well are so prevalent throughout the world.
as van der Waals bonding. Hydrogen bonds re-
main stable in the presence of water.
3. Neutral silicate layers that are separated by
highly polar water molecules may be held to- 3.16 THE 1⬊1 MINERALS
gether by hydrogen bonds.
l
4. Cations needed for electrical neutrality may be in The kaolinite–serpentine minerals are composed of al-
ria
positions that control interlayer bonding. In mi- ternating silica and octahedral sheets as shown sche-
cas, some of the silicon is replaced by aluminum matically in Fig. 3.16. The tips of the silica tetrahedra
in the silica sheets. The resulting charge defi- and one of the planes of atoms in the octahedral sheet
ciency is partly balanced by potassium ions be- are common. The tips of the tetrahedra all point in the
same direction, toward the center of the unit layer. In
ate
tween the unit cell layers. The potassium ion just
fits into the holes formed by the bases of the the plane of atoms common to both sheets, two-thirds
silica tetrahedra (Fig. 3.12). As a result, it gen- of the atoms are oxygens and are shared by both sili-
erates a strong bond between the layers. In the con and the octahedral cations. The remaining atoms
chlorites, the charge deficiencies from substitu- in this plane are (OH) located so that each is directly
tions in the octahedral sheet of the 2⬊1 sandwich
dM below the hole in the hexagonal net formed by the
are balanced by excess charge on the single-sheet bases of the silica tetrahedra. If the octahedral layer is
layer interleaved between the three-sheet layers. brucite, then a mineral of the serpentine subgroup re-
This provides a strongly bonded structure that sults, whereas dioctahedral gibbsite layers give clay
while exhibiting cleavage will not separate in the minerals in the kaolinite subgroup. Trioctahedral 1⬊1
presence of water or other polar liquids. minerals are relatively rare, usually occur mixed with
5. When the surface charge density is moderate, as kaolinite or illite, and are hard to identify. A diagram-
in smectite and vermiculite, the silicate layers matic sketch of the kaolinite structure is shown in Fig.
hte
readily adsorb polar molecules, and also the ad- 3.17. The structural formula is (OH)8Si4Al4O10, and the
sorbed cations may hydrate, resulting in layer charge distribution is indicated in Fig. 3.18.
separation and expansion. The strength of the in- Mineral particles of the kaolinite subgroup consist
of the basic units stacked in the c direction. The bond-
terlayer bond is low and is a strong function of
ing between successive layers is by both van der Waals
charge distribution, ion hydration energy, surface
forces and hydrogen bonds. The bonding is sufficiently
ion configuration, and structure of the polar mol-
rig
the nonclay minerals, pyrophyllite and talc, which have network. As a result, kaolinite, which is the most abun-
comparable structures, do not. There are two possible dant member of the subgroup and a common soil min-
reasons (van Olphen, 1977): eral, is triclinic instead of monoclinic. The unit cell
dimensions are a ⫽ 5.16 Å, b ⫽ 8.94 Å, c ⫽ 7.37 Å,
Co
1. The interlayer cations in smectite hydrate, and ⫽ 91.8, ⫽ 104.5, and ⫽ 90.
the hydration energy overcomes the attractive Variations in stacking of layers above each other,
forces between the unit layers. There are no in- and possibly in the position of aluminum ions within
terlayer cations in pyrophyllite; hence, no swell- the available sites in the octahedral sheet, produce dif-
ing. ferent members of the kaolinite subgroup. The dickite
2. Water does not hydrate the cations but is ad- unit cell is made up of two unit layers, and the nacrite
sorbed on oxygen surfaces by hydrogen bonds. unit cell contains six. Both appear to be formed by
There is no swelling in pyrophyllite and talc be- hydrothermal processes. Dickite is fairly common as
cause the surface hydration energy is too small secondary clay in the pores of sandstone and in coal
to overcome the van der Waals forces between beds. Neither dickite nor nacrite is common in soils.
l
ria
Figure 3.16 Schematic diagrams of the structures of kaolinite and serpentine: (a) kaolinite
ate
and (b) serpentine.
dM Halloysite
Halloysite is a particularly interesting mineral of the
kaolinite subgroup. Two distinct endpoint forms of this
mineral exist, as shown in Fig. 3.19; one, a hydrated
form consisting of unit kaolinite layers separated from
each other by a single layer of water molecules and
having the composition (OH)8Si4Al4O10 4H2O, and
hte
Figure 3.19 Schematic diagrams of the structure of halloysite: (a) halloysite (10 Å) and (b)
l
halloysite (7 Å).
ria
values, 2.9 Å, is the approximate thickness of a single determinations at high pH. This suggests that broken
layer of water molecules. bonds are at least a partial source of exchange capacity.
ate
The recommended terms for the two forms of hal- That a positive cation exchange capacity is measured
loysite are halloysite (7 Å) and halloysite (10 Å). under low pH conditions when edges are positively
Transformation from halloysite (10 Å) to halloysite (7 charged indicates that some isomorphous substitution
Å) by dehydration can occur at relatively low temper- must exist also.
atures and is irreversible. Halloysite is often found in As interlayer separation does not occur in kaolinite,
soils formed from volcanic parent materials in wet en-
dM balancing cations must adsorb on the exterior surfaces
vironments. It can be responsible for special properties and edges of the particles.
and problems in earthwork construction, as discussed
later in this book.
Morphology and Surface Area
Isomorphous Substitution and Exchange Capacity Well-crystallized particles of kaolinite (Fig. 3.20), na-
crite, and dickite occur as well-formed six-sided plates.
Whether or not measurable isomorphous substitution
The lateral dimensions of these plates range from
exists within the structure of the kaolinite minerals is
hte
exchange reactions.
3. Broken bonds around particle edges may give un-
satisfied charges that are balanced by adsorbed
cations.
Kaolinite particles are charged positively on their
edges when in a low pH (acid) environment, but neg-
atively charged in a high pH (basic) environment. Low Figure 3.20 Electron photomicrograph of well-crystallized
exchange capacities are measured under low pH con- kaolinite from St. Austell, Cornwall, England. Picture width
ditions and high exchange capacities are obtained for is 17 m (Tovey, 1971).
l
(OH) layer to revert to 8.62 Å, resulting in a curvature The theoretical composition in the absence of
ria
with the hydroxyls on the inside and the bases of the isomorphous substitutions is (OH)4Si8Al4O20
silica tetrahedra on the outside. The outside diameters n(interlayer)H2O. The structural configuration and cor-
of the tubular particles range from about 0.05 to 0.20 responding charge distribution are shown in Fig. 3.24.
m, with a median value of 0.07 m. The wall thick- The structure shown is electrically neutral, and the
ness is about 0.02 m. The tubes range in length from atomic configuration is essentially the same as that in
ate
a fraction of a micrometer to several micrometers. Dry- the nonclay mineral pyrophyllite.
ing of halloysite (10 Å) may result in splitting or un-
rolling of the tubes. The specific surface area of Isomorphous Substitution in the Smectite Minerals
kaolinite is about 10 to 20 m2 /g of dry clay; that of
halloysite (10 Å) is 35 to 70 m2 /g.
dM Smectite minerals differ from pyrophyllite in that there
is extensive isomorphous substitution for silicon and
aluminum by other cations. Aluminum in the octahe-
dral sheet may be replaced by magnesium, iron, zinc,
3.17 SMECTITE MINERALS nickel, lithium, or other cations. Aluminum may re-
place up to 15 percent of the silicon ions in the tetra-
Structure hedral sheet. Possibly some of the silicon positions can
The minerals of the smectite group have a prototype be occupied by phosphorous (Grim, 1968).
hte
structure similar to that of pyrophyllite, consisting of Substitutions for aluminum in the octahedral sheet
an octahedral sheet sandwiched between two silica may be one-for-one or three-for-two (aluminum oc-
sheets, as shown schematically in Fig. 3.22 and dia- cupies only two-thirds of the available octahedral sites)
grammatically in three dimensions in Fig. 3.23. All the in any combination from a few to complete replace-
tips of the tetrahedra point toward the center of the ment. The resulting structure, however, is either almost
unit cell. The oxygens forming the tips of the tetra- exactly dioctahedral (montmorillonite subgroup) or
rig
hedra are common to the octahedral sheet as well. The trioctahedral (saponite subgroup). The charge defi-
anions in the octahedral sheet that fall directly above ciency resulting from these substitutions ranges from
0.5 to 1.2 per unit cell. Usually, it is close to 0.66 per
unit cell. A charge deficiency of this amount would
result from replacement of every sixth aluminum by a
py
l
ria
Figure 3.22 Schematic diagrams of the structures of the smectite minerals: (a) montmoril-
ate
lonite and (b) saponite.
dM
hte
rig
Figure 3.23 Diagrammatic sketch of the montmorillonite in Fig. 3.25. Particles range in thickness from 1-nm
Co
l
ria
Na0.66
Nontronite Al for Si Fe3⫹ for Al (OH)4(Si7.34Al0.66) Fe43⫹O20
↓
Na0.66
Trioctahedral, Smectites,
ate
or Saponites
Hectorite None Li for Mg (OH)4Si8(Mg5.34Li0.66) O20
↓
Na0.66
Saponite Al for Si Fe3⫹ for Mg (OH)4(Si7.34Al0.66) Mg6O20
↓
dM Na0.66
Sauconite Al for Si Zn for Mg (OH)4(Si8⫺yAly)(Zn6⫺xMgx) O20
↓
Na0.66
a
Two formula units are needed to give one unit cell.
After Ross and Hendricks (1945); Marshall (1964); and Warshaw and Roy (1961).
hte
Bentonite
A very highly plastic, swelling clay material known as
bentonite is very widely used for a variety of purposes,
ranging from drilling mud and slurry walls to clarifi-
py
sure due to the high content of bentonite. Stability in Fig. 3.28. The unit cell is electrically neutral and
problems in underground construction may be caused has the formula (OH)4K2(Si6Al2)Al4O20. Muscovite is
by the presence of montmorillonite in joints and faults the dioctahedral end member of the micas and contains
(Brekke and Selmer-Olsen, 1965). only Al3⫹ in the octahedral layer. Phlogopite (brown
mica) is the trioctahedral end member, with the octa-
hedral positions filled entirely by magnesium. It has
3.18 MICALIKE CLAY MINERALS the formula (OH)4K2(Si6Al2)Mg6O20. Biotite (black
mica) is trioctahedral, with the octahedral positions
Illite is the most commonly found clay mineral in soils filled mostly by magnesium and iron. It has the general
encountered in engineering practice. Its structure is formula (OH)4K2(Si6Al2)(MgFe)6O20. The relative pro-
quite similar to that of muscovite mica, and it is some- portions of magnesium and iron may vary widely.
l
times referred to as hydrous mica. Vermiculite is also Illite differs from mica in the following ways (Grim,
ria
often found as a clay phase constituent of soils. Its 1968):
structure is related to that of biotite mica.
1. Fewer of the Si4⫹ positions are filled by Al3⫹ in
Structure illite.
ate
The basic structural unit for the muscovite (white mica) 2. There is some randomness in the stacking of lay-
is shown schematically in Fig. 3.26a. It is the three- ers in illite.
layer silica–gibbsite–silica sandwich that forms pyro- 3. There is less potassium in illite. Well-organized
phyllite, with the tips of all the tetrahedra pointing illite contains 9 to 10 percent K2O (Weaver and
toward the center and common with octahedral sheet Pollard, 1973).
ions. Muscovite differs from pyrophyllite, however, in
dM 4. Illite particles are much smaller than mica parti-
that about one-fourth of the silicon positions are filled cles.
by aluminum, and the resulting charge deficiency is
balanced by potassium between the layers. The layers Some illite may contain magnesium and iron in the
are continuous in the a and b directions and stacked in octahedral sheet as well as aluminum (Marshall, 1964).
the c direction. The radius of the potassium ion, 1.33 Iron-rich illite, usually occurring as earthy green pel-
Å, is such that it fits snugly in the 1.32 Å radius hole lets, is termed glauconite.
formed by the bases of the silica tetrahedra. It is in 12- The vermiculite structure consists of regular inter-
hte
fold coordination with the 6 oxygens in each layer. stratification of biotite mica layers and double molec-
A diagrammatic three-dimensional sketch of the ular layers of water, as shown schematically in Fig.
muscovite structure is shown in Fig. 3.27. The struc- 3.26b. The actual thickness of the water layer depends
tural configuration and charge distribution are shown on the cations that balance the charge deficiencies in
rig
py
Co
Figure 3.26 Schematic diagram of the structures of muscovite, illite, and vermiculite: (a)
muscovite and illite and (b) vermiculite.
l
ria
ate
dM
Figure 3.27 Diagrammatic sketch of the structure of muscovite.
(OH)4(MgCa)x(Si8
xAlx)(MgFe)6O20 yH2O
x ⬇ 1 to 1.4 y⬇8
rig
l
Morphology and Surface Area
ria
Illite usually occurs as very small, flaky particles
mixed with other clay and nonclay materials. High-
purity deposits of illite are uncommon. The flaky par-
ticles may have a hexagonal outline if well crystallized. Figure 3.30 Schematic diagram of the structure of chlorite.
ate
The long axis dimension ranges from 0.1 m or less
to several micrometers, and the plate thickness may be
as small as 3 nm. An electron photomicrograph of illite
is shown in Fig. 3.29. Vermiculite may occur in nature
as large crystalline masses having a sheet structure of vermiculite, except that an organized octahedral
somewhat similar in appearance to mica. In soils, ver-
dM sheet replaces the double water layer between mica
miculite occurs as small particles mixed with other layers. The layers are continuous in the a and b direc-
clay minerals. tions and stacked in the c direction. The basal spacing
The specific surface area of illite is about 65 to 100 is fixed at 14 Å.
m2 /g. The primary surface of vermiculites is 40 to 80 Isomorphous Substitution The central sheet of the
m2 /g, and the secondary (interlayer) surface may be as mica layer is trioctahedral, with magnesium as the pre-
high as 870 m2 /g. dominant cation. There is often partial replacement of
Mg2⫹ by Al3⫹, Fe2⫹ and Fe3⫹. There is substitution of
hte
l
from volcanic ash and in tropical residual soils. Some
ria
soils rich in allophane and oxides may exhibit signif-
icant irreversible decreases in plasticity and increases
Figure 3.31 Electron photomicrograph of attapulgite from in strength when dried. Many are susceptible to break-
Attapulgis, Georgia. Picture width is 4.7 m (Tovey, 1971).
down and strength loss when subjected to traffic or
manipulation during earthwork construction (Mitchell
ate
and Sitar, 1982; Mitchell and Coutinho, 1991).
Mixed-Layer Clays
More than one type of clay mineral is usually found 3.20 SUMMARY OF CLAY MINERAL
in most soils. Because of the great similarity in crystal
dM CHARACTERISTICS
structure among the different minerals, interstratifica-
tion of two or more layer types often occurs within a The important structural, compositional, and morpho-
single particle. Interstratification may be regular, with logical characteristics of the important clay minerals
a definite repetition of the different layers in sequence, are summarized in Table 3.6. Data on the structural
or it may be random. According to Weaver and Pollard characteristics of the tetrahedral and octahedral sheet
(1973), randomly interstratified clay minerals are sec- structures are included.
ond only to illite in abundance.
hte
stratified clay with high charge, micalike layers with Identification of the fine-grained minerals in a soil is
fixed interlayer cations alternating in a regular manner usually done by X-ray diffraction. Simple chemical
with low-charge montmorillonite-like layers containing tests can be used to indicate the presence of organic
exchangeable cations capable of hydration. matter and other constituents. The microscope may be
used to identify the constituents of the nonclay frac-
py
are termed allophane. Such material is amorphous to while owing to our inability to make exact quantitative
X-rays because there is insufficient long-range order of links from composition to properties. Accordingly,
the octahedral and tetrahedral units to produce sharp from knowledge of grain size distribution, the relative
diffraction effects, although in some cases there may intensities of different X-ray diffraction peaks, and a
be diffraction bands. Allophane has no definite com- few other simple tests a semiquantitative analysis may
position or shape and may exhibit a wide range of be made that is usually adequate for most purposes.
physical properties. Some noncrystalline clay material A general approach is given in this section for the
is probably contained in all fine-grained soils. It is determination of soil composition, some of the tech-
common in volcanic soils because of the abundance of niques are described briefly, and criteria for identifi-
glass particles. cation of important soil constituents are stated.
Structure
Subgroup and Octahedral Layer Tetrahedral Layer
a
Type Schematic Structure Mineral Complete Formula / Unit Cell Cations Cations Isomorphous Substitution Interlayer Bond
l
Dickite (OH)8Si4Al4O10 Al4 Si4 Little O—OH
Hydrogen Strong
ria
Nacrite (OH)8Si4Al4O10 Al4 Si4 Little
O—OH
1⬊1 Halloysite (OH)8Si4Al4O10 Al4 Si4 Little Hydrogen Strong
(dehydrated) O—OH
Halloysite (OH)8Si4Al4O10 4H2O Al4 Si4 Little Hydrogen Strong
(hydrated) O—OH
Hydrogen Strong
ate
Montmorillonite Montmorillonite (OH)4Si8(Al3.34Mg.66O20nH2O Al3.34Mg.66 Si8 Mg for Al, Net charge O—O
(OH)4Si8Al4O20 NH2O ↓ * always ⫽ 0.66- / unit Very weak
(Theoretical Na.66 cell expanding lattice
Unsubsitituted)
Beidellite
dM (OH)4(Si7.34Al66)(Al4)O20nH2O Al4 Si7.34Al.66 Al for Si, Net charge O—O
↓ always ⫽ 0.66- / for Very weak
Na.66 unit cell expanding lattice
Nontronite (OH)4(Si7.34Al.66)Fe43⫹O20nH2O Fe4 Si7.34Al.66 Fe for Al, Al for Si, Net O—O
↓ charge always ⫽ 0.66- Very weak
Na.66 / for unit cell expanding lattice
2⬊1 Saponite Hectorite (OH)4Si8(Mg5.34Li.66)P20nH2O Mg5.34Li.66 Si8 Mg, Li for Al, Net O—O
↓ charge always ⫽ 0.66- Very weak
Na.66 / unit cell expanding lattice
Saponite (OH)4(Si7.34Al.66)Mg6O20nH2O Mg, Fe3⫹ Si7.34Al.66 Mg for Al, Al for Si, O—O
hte
2⬊1⬊1 Chlorite Chlorite (OH)4(SiAl)8(Mg.Fe)6O20 (2⬊1 layer) (Mg,Fe)6(2⬊1 layer) (Si,Al)8 Al for Si in 2⬊1 layer
(Several varieties (MgAl)6(OH)12 interlayer (Mg,Al)6 interlayer Al for Mg in interlayer
known)
a
Arrows indicate source of charge deficiency. Equivalent Na listed as balancing cation. Two formula units (Table 3.4) are required per unit cell.
b
Electron microscope data.
Triclinic
a ⫽ 5.14, b ⫽ 8.93, c ⫽ 7.37
⫽ 91.6, ⫽ 104.8, ⫽ 89.9
7.2 Å 6-sided flakes 0.1–4 ⫻
0.05–2 冎 single
3–15 2.60–2.68 10–20 Very common
to 3000 ⫻ 4000
l
(stacks)
0.07–300 ⫻ 2.5– 1–30
ria
Monoclinic 14.4 Å Unit cell 6-sided flakes Rare
a ⫽ 5.15, b ⫽ 8.95, c ⫽ 14.42 contains 2 1000
⫽ 9648 unit layers
Almost Orthorhombic 43 Å Unit cell Rounded flakes 1 ⫻ 0.025– Rare
a ⫽ 5.15, b ⫽ 8.96, c ⫽ 43 contains 6 0.15
⫽ 9020 unit layers
a ⫽ 5.14 in O Plane 7.2 Å Random Tubes 0.07 O.D. 5–10 2.55–2.56 Occasional
a ⫽ 5.06 in OH Plane stacking of 0.04 I.D.
ate
b ⫽ 8.93 in O Plane unit cells 1 long.
b ⫽ 8.62 in OH Plane 10.1 Å Water layer Tubes 5–40 2.0–2.2 35–70 Occasional
⬖ layers curve between unit
cells
9.6Å—Complete Dioctahedral Flakes (equi- ⬎10 Å ⫻ up to 80–150 2.35–2.7 50–120 Primary Very common
separation dimensional) 10 700–840 Secondary
several ⫻
unit cell
mont.
Trioctahedral Brand laths 50 Å Thick Rare
a ⫽ 5.34, b ⫽ 9.20 10.5–14 Å Alternating Similar to illite 100–150 40–80 Primary Fairly common
c ⫽ 28.91, ⫽ 9315 Mica and 870 Secondary
py
double H2O
layers
Co
From Grim, R. E. (1968) Clay Mineralogy, 2d edition, McGraw-Hill, New York. Brown, G. (editor) (1961) The X-ray Identification and Crystal Structure of Clay Materials, Mineralogical
Society (Clay Minerals Group), London.
Methods for Compositional Analysis properties of the mineral in the soil are the same as
Methods and techniques that may be employed for de- those of a reference mineral. However, different sam-
termination of soil composition and study of soil grains ples of any given clay mineral may exhibit significant
include: differences in composition, surface area, particle size
and shape, and cation exchange capacity. Thus, selec-
1. Particle size analysis and separation tion of ‘‘standard’’ minerals for reference is arbitrary.
2. Various pretreatments prior to mineralogical Quantitative clay mineral determinations cannot be
analysis made to an accuracy of more than about plus or minus
3. Chemical analyses for free oxides, hydroxides, a few percent without exhaustive chemical and min-
amorphous constituents, and organic matter eralogical tests.
4. Petrographic microscope study of silt and sand
l
grains General Scheme for Compositional Analysis
ria
5. Electron microscope study A general scheme for determination of the components
6. X-ray diffraction for identification of crystalline of a soil is given in Fig. 3.32. Techniques of the most
minerals value for qualitative and semiquantitative analysis are
7. Thermal analysis indicated by a double asterisk, and those of particular
ate
8. Determination of specific surface area use for explaining unusual properties are indicated by
9. Chemical analysis for layer charge, cation a single asterisk. The scheme shown is by no means
exchange capacity, exchangeable cations, pH, the only one that could be used; a feedback approach
and soluble salts is desirable wherein the results of each test are used to
10. Staining tests for identification of clays plan subsequent tests. Brief discussions of the various
dM techniques listed in Fig. 3.32 are given below. X-ray
Procedures for determination of soil composition are diffraction analysis is treated in more detail in the next
described in detail in publications of the American So- section because of its particular usefulness for the
ciety of Agronomy. Part 1—Physical and Mineralog- identification of fine-grained soil minerals.
ical Methods provides a set of procedures for Grain Size Analysis Determination of particle size
mineralogical analyses for use by soil scientists and and size distribution is usually done using sieve anal-
engineers. Part 2—Microbiological and Biochemical ysis for the coarse fraction [sizes greater than 74 m
Properties, published in 1994, is useful for determi- (i.e., 200 mesh sieve)] and by sedimentation methods
hte
nations needed for bioremediation and other geoen- for the fine fraction. Details of these methods are pre-
vironmental purposes. Part 3—Chemical Methods, sented in standard soil mechanics texts and in the stan-
published in 1996 contains methods for characterizing dards of the American Society for Testing and
soil chemical properties as well as several methods for Materials (ASTM). Determination of sizes by sedi-
characterizing soil chemical processes. Part 4— mentation is based on the application of Stokes’s law
Physical Methods, published in 2002, is an updated
rig
given.
where s ⫽ unit weight of particle, w ⫽ unit weight
of liquid, ⫽ viscosity of liquid, and D ⫽ diameter
Accuracy of Compositional Analysis of sphere. Sizes determined by Stoke’s law are not ac-
Techniques for chemical analysis are generally of a tual particle diameters but, rather, equivalent spherical
Co
high order of accuracy. However, this accuracy does diameters. Gravity sedimentation is limited to particle
not extend to the overall compositional analysis of a sizes in the range of about 0.2 mm to 0.2 m, the
soil in terms of components of interest in understand- upper bound reflecting the size limit where flow around
ing and quantifying behavior. This is because knowl- the particles is no longer laminar, and the lower bound
edge of the chemical composition of a soil is of limited representing a size where Brownian motion keeps par-
value by itself. Chemical analysis of the solid phase of ticles in suspension indefinitely.
a soil does not indicate the organization of the ele- The times for particles of 2, 5, and 20 m equivalent
ments into crystalline and noncrystalline components. spherical diameter to fall through water a distance of
For quantitative mineralogical analysis of the clay 10 cm are about 8 h, 1.25 h, and 5 min, respectively,
fraction, it is usually necessary to assume that the at 20C. At 30C the required times are about 6.5 h, 1
l
ria
ate
dM
hte
rig
Figure 3.32 Flow sheet for compositional analysis of soils (adapted from Lambe and Martin,
1954).
h, and 4 min. A centrifuge can be used for accelerating from laboratory preparation of the sample to that of
py
the settlement of small particles and is the most prac- the particles and aggregates in the natural soil is un-
tical means for extracting particles smaller than about known.
a micrometer in size. Optical and electron microscopes are sometimes
Sedimentation methods call for treatment of a soil– used to study particle sizes and size distributions and
Co
water suspension with a dispersing agent and thorough to provide information on particle shape, aggregation,
mixing prior to the start of the test. This causes break- angularity, weathering, and surface texture.
down of aggregates of soil particles, and the degree of Pore Fluid Electrolyte The total concentration of
breakdown may vary greatly with the method of prep- soluble salts may be determined from the electrical
aration. For example, the ASTM standard method of conductivity of extracted pore fluid. Chemical or pho-
test permits the use of either an air dispersion cup or tometric techniques may be used to determine the el-
a blender-type mixer. The amount of material less than emental constituents of the extract (Rhoades, 1982).
2 m equivalent spherical diameter may vary by as Removal of excess soluble salts by washing the sample
much as a factor of 2 by the two techniques. The re- with water or alcohol may be necessary before pro-
lationship between the size distribution that results ceeding with subsequent analysis. If they are not re-
moved, the soil may be difficult to disperse, it may be Exchange Complex Determination of the cation
difficult to remove organic matter, reliable cation exchange capacity (expressed in milliequivalents per
exchange capacity determinations will be impossible, hundred grams of dry soil) is made after first freeing
and mineralogical analyses will be complicated (Kunze the soil of excess soluble salts. The adsorbed cations
and Dixon, 1986). are then replaced by a known cation species, and the
pH Determination of the acidity or alkalinity of a amount of the known cation needed to saturate the
soil in terms of the pH is a relatively simple measure- exchange sites is determined analytically (Rhoades,
ment that can be made using a pH meter or special 1982). The composition of the original cation complex
indicators (American Society for Testing and Materi- can be determined by chemical analysis of the original
als, 1970; McLean, 1982). The value obtained depends extract (Thomas, 1982).
on the ratio of soil to water, so it is usual to standardize Potash The hydrous mica minerals (illites) are the
l
the measurement using a 1⬊1 ratio of soil to water by only minerals commonly found in the clay size fraction
ria
weight. For highly plastic soils a lower soil-to-water of soils that contain potassium in their crystal structure.
ratio may be required to produce a suspension suitable Thus, knowledge of the K2O content is useful for quan-
for pH measurement. The pH decreases with increas- titative determination of their abundance. A method for
ing concentration of neutral salts in solution and with potassium determination is given by Knudsen et al.
(1986). Well-organized 10-Å illite layers contain 9 to
ate
increasing amounts of dissolved CO2.
Carbonates Carbonates, in the form of calcite 10 percent K2O (Weaver and Pollard, 1973).
(CaCO3), dolomite [CaMg(CO3)2], marl, and shells are Specific Surface Area Ethylene glycol and glycerol
frequently found in soils, and they can be readily de- adsorb on clay surfaces. As different clay minerals
tected by effervescence when the soil is treated with have different values of specific surface, the amount of
dilute HCl. Many methods for determining inorganic
dM glycol or glycerol retained under controlled conditions
carbonates, calcite, and dolomite in soils are available can be used to aid in the quantitative determinations
(Nelson, 1982). These include dissolution in acid, dif- of clay minerals and for estimation of specific surface
ferential thermal analysis, X-ray diffraction, and chem- area (Martin, 1955; Diamond and Kinter, 1956; and
ical analyses. American Society for Testing and Materials, 1970).
Gypsum Gypsum (CaSO4 2H2O) can be deter- Use of ethylene glycol monoethyl ether (EGME) as
mined by a simple heating test. Visible grains will turn the polar molecule for determining surface area offers
white when heated on a metal plate as a result of de- the advantages of the attainment of adsorption equilib-
hte
hydration to form ‘‘dead-burnt gypsum’’ (Shearman, rium more rapidly and with greater precision (Carter
1979). Quantitative determinations can be made using et al., 1982). A monomolecular layer of EGME is as-
procedures described by Nelson (1982). sumed to form in vacuum on a predried clay sample.
Organic Matter Organic matter can be readily de- The weight of EGME adsorbed after equilibrium is
tected by treatment of the soil with a 15 percent hy- reached is converted to specific surface using a factor
of 0.000286 g EGME per square meter of surface.
rig
l
independent of the nature of the bombarded at-
ria
oms and appear as a band of continuously vary- Figure 3.35 Composite relationship for X-ray intensity as a
function of wavelength.
ing wavelength as shown in Fig. 3.34.
The resulting output of X-rays from these two ef-
fects acting together is shown in Fig. 3.35. X-rays are
ate
generated using a tube in which electrons stream from length. The wavelengths of monochromatic radiation
a filament to a target material across a voltage drop of (usually K, Fig. 3.33) produced from commonly used
20 to 50 kV. Curved crystal monochrometers can be target materials range from 0.71 Å for molybdenum to
used to give X-rays of a single wavelength. Alterna- 2.29 Å for chromium. Copper radiation, which is most
tively, certain materials are able to absorb X-rays of frequently used for mineral identification, has a wave-
different wavelengths, so it is possible to filter the out-
dM
put of an X-ray tube to give rays of only one wave-
length of 1.54 Å.
Diffraction of X-rays
Because wavelengths of about 1 Å are of the same
order as the spacing of atomic planes in crystalline
materials, X-rays are useful for analysis of crystal
hte
ters designate shells in which electron transfer takes place. orientation of parallel atomic planes, relative to the di-
rection of the incident beam, at which radiations are in
phase depends on the wave length of the X-rays and
the spacing between atomic planes.
Figure 3.36 shows a parallel beam of X-rays of
Co
l
ria
Figure 3.36 Geometrical conditions for X-ray diffraction according to Bragg’s law.
ate
From symmetry, BC ⫽ CD, and by trigonometry, the X-ray beam. Thus, each mineral will produce a
CD ⫽ d sin . Thus the necessary condition is given characteristic set of reflections at values of corre-
by sponding to the interatomic spacings between the
prominent planes. The intensities of the different re-
n ⫽ 2d sin (3.3) flections vary according to the density of atomic pack-
dM ing and other factors.
This is Bragg’s law. It forms the basis for identification When the oriented aggregate method is used, platy
of crystals using X-ray diffraction. Since no two min- clay particles are precipitated onto a glass slide, usu-
erals have the same spacings of interatomic planes in ally by drying from a deflocculated suspension or sep-
three dimensions, the angles at which diffractions oc- arated from a suspension on a porous ceramic plate.
cur (and the atomic spacings calculated from them) can With most particles oriented parallel to the slide, the
be used for identification. X-ray diffraction is partic- (001) reflections are intensified, whereas reflections
ularly well suited for identification of clay minerals from (hk0) planes are minimized.
hte
because the (001) spacing is characteristic for each In the Bragg equation, n may be any whole number.
clay mineral group. The basal planes generally give the The reflection corresponding to n ⫽ 1 is termed the
most intense reflections of any planes in the crystals first-order reflection. If the first-order reflection for a
because of the close packing of atoms in these planes. mineral gives d(001) ⫽ 10 Å, then for n ⫽ 2 there can
The common nonclay minerals occurring in soils are be a reflection at 5 Å, for n ⫽ 3 there can be a reflec-
rig
also detectable by X-ray diffraction. tion at 3.33 Å, and so on. It is common to refer to
these as higher-order reflections due to the (002) plane,
Detection of Diffracted X-rays the (003) plane, and so on, even though atomic planes
Because the small size of most soil particles prevents do not exist at these spacings. They are, in reality, val-
the study of single crystals, use is made of the powder ues of d/n ⫽ /(2 sin ) for integer values of n ⬎ 1.
py
tensities are scanned by a Geiger, proportional, or scin- values of 2. Each reflection must be assigned to some
tillation tube and recorded automatically to produce a component of the sample. The first step in the analysis
chart showing the intensity of diffracted beam as a is to determine all values of d/n for the particular type
function of angle 2. As an example, the diffraction of radiation (which determines ) using Eq. (3.3). The
pattern for quartz is shown in Fig. 3.37. The powder test pattern may be compared directly with patterns for
method works because the very large number of par- known materials. The American Society for Testing
ticles in a sample ensures that some will always be and Materials maintains a file of patterns for many
properly oriented to produce a reflection. materials indexed on the basis of the strongest lines in
All prominent atomic planes in a crystal will pro- the pattern. X-ray diffraction data for the clay minerals
duce a reflection if properly positioned with respect to and other common soil minerals are given in Grim
l
ria
ate
dM
hte
rig
py
Co
Figure 3.37 X-ray diffractometer chart for quartz. Peaks occur at specific 2 angles, which
can be converted to d spacings by Bragg’s law. Numbers in parentheses are the Miller indices
for the crystal planes responsible for the indicated peak.
(1968), Carroll (1970), Brindley and Brown (1980), to 11.6 Å when heated at 70C. All interlayer water
Whittig and Allardice (1986), and Moore and Reyn- can be expelled at 500C, but rehydration is rapid on
olds (1997). The most intense reflections for minerals cooling. Permanent dehydration and collapse to 9.02
commonly found in powder samples of soils are listed Å can be achieved by heating to 700C.
in Table 3.7. Basal spacings for different clay minerals Chlorite Minerals The basal spacing of chlorite
associated with different pretreatments are listed in Ta- minerals is fixed at 14 Å because of the strong ordering
ble 3.8 and shown pictorially in Fig. 3.38. of the interlayer complex. Chlorites often have a clear
sequence of four or five basal reflections. The third-
order reflection at 4.7 Å is often strong. Iron-rich chlo-
Criteria for Clay Minerals rites have a weak first-order reflection but strong
The different clay minerals are characterized by first- second-order reflections and, thus, may be confused
with kaolinite. The facts that chlorite is destroyed
l
order basal reflections at 7, 10, or 14 Å. Positive iden-
when treated with 1 N HCl at 60C while kaolinite is
ria
tification of specific mineral groups ordinarily requires
specific pretreatments. Separation of size fractions re- unaffected, and that kaolinite is destroyed but chlorite
quires thorough dispersion of the sample. As cement- may not be affected on heating to 600C, are useful
ing compounds may both inhibit dispersion and for distinguishing the two clay mineral types.
adversely affect the quality of the diffraction patterns,
ate
Criteria for Nonclay Minerals
their removal may be necessary. To ensure uniform ex-
pansion due to hydration for all crystals of a particular Strong X-ray diffraction reflections for some of the
mineral, the clay should be made homoionic. Magne- nonclay minerals are listed in Table 3.7. These include
sium and potassium are most frequently used for sat- feldspar, quartz, and carbonates. More detailed listings
of X-ray powder data for specific iron oxide minerals,
uration of the exchange sites. Detailed procedures for
dM silica minerals, feldspars, carbonates, and calcium sul-
pretreatments useful in X-ray diffraction analysis of
fate minerals are given in Brindley and Brown (1980)
clay soils are given by Whittig and Allardice (1986) as well as in standard reference files.
and Moore and Reynolds (1997).
Kaolinite Minerals The kaolinite basal spacing of Quantitative Analysis by X-ray Diffraction
about 7.2 Å is insensitive to drying or moderate heat- Quantitative determination of the amounts of different
ing. Heating to 500C destroys kaolinite minerals, but minerals in a soil on the basis of simple comparison
not the other clay minerals. Hydrated halloysite has a of diffraction peak heights or areas are uncertain
hte
basal spacing of 10 Å, which collapses irreversibly to because of differences in mass absorption coefficients
7 Å on drying at 110C. Organic chemical treatments of different minerals, particle orientations, sample
are sometimes used to distinguish dehydrated halloy- weights, surface texture of the sample, mineral crys-
site from kaolinite (MacEwan and Wilson, 1980). The tallinity, hydration, and other factors. Estimates based
electron microscope can also be used to distinguish on X-ray data alone are usually at best semiquantita-
dehydrated halloysite with its tubular morphology from tive; however, in some cases techniques that account
rig
sive character of this group of minerals provides the compositions and mixed layering. For more detailed
basis for their positive identification. When air dried, treatment of X-ray diffraction theory, identification cri-
these minerals may have basal spacings of 12 to 15 Å. teria, and techniques, particularly as related to the
After treatment with ethylene glycol or glycerol, the study of clays, see Klug and Alexander (1974), Carroll
Co
smectites expand to a d(001) value of 17 to 18 Å. When (1970), Brindley and Brown (1980), Whittig and Al-
oven dried, d(001) drops to about 10 Å as a result of the lardice (1986), and especially Moore and Reynolds
removal of interlayer water. (1997).
Vermiculite Although an expansive mineral, the
greater interlayer ordering in vermiculite results in less 3.23 OTHER METHODS FOR COMPOSITIONAL
variability in basal spacing than occurs in the smectite ANALYSIS
minerals. When Mg saturated, the hydration states of
vermiculite yield a discrete set of basal spacings, re- Thermal Analysis
sulting from a changing but ordered arrangement of Principle Differential thermal analysis (DTA) con-
Mg cations and water in the interlayer complex. When sists of simultaneously heating a test sample and a
fully saturated, the d spacing is 14.8 Å, which reduces thermally inert substance at constant rate (usually
l
6.39 Felds. 2.60 Verm., Sepiol.
ria
4.90–5.00 Illite, Mica, Halloysite 2.56 Illite (VS), Kaol.
4.70–4.79 Chlor. (S) 2.53–2.56 Chlor., Felds., Mont.
4.60 Verm. (S) 2.49 Kaol. (VS)
4.45–4.50 Illite (VS), Sepiolite 2.46 Quartz, heated Verm.
4.46 Kaol. 2.43–2.46 Chlorite
ate
4.36 Kaol. 2.39 Verm., Illite
4.26 Quartz (S) 2.38 Kaol.
4.18 Kaol. 2.34 Kaol. (VS)
4.02–4.04 Felds. (S) 2.29 Kaol. (VS)
3.85–3.90 Felds. 2.28 Quartz, Sepiol.
3.82 Sepiol. 2.23 Illite, Chl.
3.78
dM
Felds. 2.13 Quartz, Mica
3.67 Felds. 2.05–2.06 Kaol. (WK)
3.58 Carbonate, Chl. 1.99–2.00 Mica, Illite (S), Kaol. Chl.
3.57 Kaol. (VS), Chl. 1.90 Kaol.
3.54–3.56 Verm. 1.83 Carb.
3.50 Felds., Chlor. 1.82 Quartz
hte
b
Italics indicates (001) spacing.
about 10C/min) to over 1000C and continuously ygen, are also used to some extent. Thermal analysis
measuring differences in temperature between the sam- techniques are described in detail by Tan et al. (1986).
ple and the inert material. Differences in temperature The results of differential thermal analysis are pre-
between the sample and the inert substance reflect re- sented as a plot of the difference in temperature
actions in the sample brought about by the heating. between sample and inert material (
T) versus tem-
Thermogravimetric analyses, based on changes in perature (T) as indicated in Fig. 3.39. Endothermic re-
weight caused by loss of water or CO2 or gain in ox- actions are those wherein the sample takes up heat,
Table 3.8 X-ray Identification of the Principal Clay Minerals (⬍ 2 m) in an Oriented Mount of a Clay
Fraction Separated from Sedimentary Material
Glycolation Effect
Mineral Basal d Spacings (001) (1 h, 60C) Heating Effect (1 h)
Kaolinite 7.15 Å (001); 3.75 Å (002) No change Becomes amorphous 550–
600C
Kaolinite, disordered 7.15 Å (001) broad; 3.75 Å No change Becomes amorphous at lower
broad temperatures than kaolinite
l
Halloysite, 4H2O 10 Å (001) broad No change Dehydrates to 2H2O at 110C
ria
(hydrated)
Halloysite, 2H2O 7.2 Å (001) broad No change Dehydrates at 125–150C;
(dehydrated) becomes amorphous 560–
590C
Mica 10 Å (002); 5 Å (004) No change (001) becomes more intense on
ate
generally referred to as heating but structure is
(001) and (002) maintained to 700C
Illite 10 Å (002), broad, other No change (001) noticeably more intense
basal spacings present on heating as water layers
dM but small are removed; at higher
temperatures like mica
Montmorillonite group 15 Å (001) and integral (001) expands to 17 At 300C (001) becomes 9 Å
series of basal spacings Å with rational
sequence of
higher orders
Vermiculite 14 Å (001) and integral No change Dehydrates in steps
series of basal spacings
hte
spacings
Mixed-layer minerals Regular, one (001) and No change unless Various, see descriptions of
integral series of basal an expandable individual minerals
spacings component is
py
present
Random, (001) is addition Expands if Depends on minerals present in
of individual minerals montmorillonite is interlayered mineral
and depends on amount a constituent
of those present
Co
l
ria
ate
dM
hte
rig
Reproduced with permission from The American Society of Agronomy, Inc., Madison, WI.
and in exothermic reactions, heat is liberated. Analysis quired is about 1 g. Although the temperatures at
Co
of test results consists of comparing the sample curve which thermal reactions take place are a function only
with those for known materials so that each deflection of the sample, the size and shape of the reaction peaks
can be accounted for. depend also on the thermal characteristics of the ap-
Apparatus Apparatus for DTA consists of a sample paratus and the heating rate.
holder, usually ceramic, nickel, or platinum; a furnace; Reactions Producing Thermal Peaks The impor-
a temperature controller to provide a constant rate of tant thermal reactions that generate peaks on the ther-
heating; thermocouples for measurement of tempera- mogram are:
ture and the difference in temperature between the
sample and inert reference material; and a recorder for 1. Dehydration Water in a soil may be present in
the thermocouple output. The amount of sample re- three forms in addition to free pore water: (1)
l
such as the quartz inversion at 573C and the kaolinite
ria
endotherm at 650C, the amplitude can be used for
quantitative analysis. In either case, calibration of the
Figure 3.39 Thermogram of a sandy clay soil.
apparatus is necessary, and the overall accuracy is of
the order of plus or minus 5 percent.
ate
adsorbed water or water of hydration, which is Optical Microscope
driven off at 100 to 300C, (2) interlayer water
such as in halloysite and expanded smectite, and Both binocular and petrographic microscopes can be
(3) crystal lattice water in the form of (OH) ions, used to study the identity, size, shape, texture, and con-
the removal of which is termed dehydroxylation. dition of single grains and aggregates in the silt and
dM
Dehydroxylation destroys mineral structures. The sand size range; for study in the thin section of the
temperature at which the major amount of crystal fabric, that is, the spatial distribution and interrelation-
lattice water is lost is the most indicative property ships of the constituents; and for study of the orien-
for identification of minerals. Dehydration reac- tations of groups of clay particles. Because the in-focus
tions are endothermic and occur in the range of depth of field decreases sharply as magnification in-
500 to 1000C. creases, study of soil thin sections is impractical at
2. Crystallization New crystals form from amor- magnifications greater than a few hundred. Thus, in-
dividual clay particles cannot usually be distinguished
hte
from one form to another at a specific tempera- to identify the various minerals; however, relatively
ture, and the energy of transformation shows up simple diagnostic criteria that can be used for identi-
as a peak on the thermogram. For example, fication of over 80 percent of the coarse grains in most
quartz changes from the to form reversibly soils are given by Cady et al. (1986). These criteria
at 573C. The peak for the quartz phase change are based on such factors as color, refractive index,
py
is sharp, and its amplitude is nearly in direct pro- birefringence, cleavage, and particle morphology. The
portion to the amount of quartz present. The nature of surface textures, the presence of coatings,
quartz peak is frequently masked within the peak layers of decomposition, and so on are useful both for
for some other reacting material, but may be interpretation of the history of a soil and as a guide to
Co
readily identified by determining the thermogram the soundness and durability of the particles.
during cooling of the sample or by letting it cool
first and then rerunning it. The other minerals are Electron Microscope
destroyed during the initial run while the quartz With modern electron microscopes it is possible to re-
reaction is reversible. solve distances to less than 100 Å, thus making study
4. Oxidation Exothermic oxidation reactions in- of small clay particles feasible. Electron diffraction
clude the combustion of organic matter and the study of single particles may also be useful. Electron
oxidation of Fe2⫹ to Fe3⫹. Organic matter oxi- diffraction is similar to X-ray diffraction except an
dizes in the 250 to 450C temperature range. electron beam instead of an X-ray beam is used.
l
imen preparation. col retention of 100 mg/g, and a cation exchange ca-
ria
In the scanning electron microscope (SEM), second- pacity of 35 meq/100 g. Then, assuming 9 percent as
ary electrons emitted from a sample surface form what an average value of for pure illite (Table 3.9), the con-
appear to be three-dimensional images. The SEM has tent of illite is estimated at 4.0/9.0, or 44 percent. Be-
a ⫻20 to ⫻150,000 magnification range and a depth cause only the illite and smectite will contribute to the
of field some 300 times greater than that of the light glycol adsorption, the amount of smectite may be es-
ate
microscope. These characteristics, coupled with the timated:
fact that clay particles themselves and fracture surfaces
through soil masses may be viewed directly, have led 0.44 ⫻ 60 ⫹ S ⫻ 300 ⫽ 100
to extensive use of the SEM for study of clays. Ex-
amples of electron photomicrographs of clays and soils 100 ⫺ 26.4
dM ⬖S⫽ ⫽ 25%
are given earlier in this chapter and in Chapter 5. Prin- 300
ciples of electron microscopy techniques and addi-
tional examples are presented in McCrone and Delly The remaining 31 percent can be ascribed to quartz
(1973) and Sudo et al. (1981). and other nonclay components. For this clay mineral
composition, the theoretical cation exchange capacity
should be, based on the reference values in Table 3.9:
3.24 QUANTITATIVE ESTIMATION OF SOIL
hte
identification criteria and reference values for the clay eralogical analysis is the uncertainty in the reference
minerals are given in Table 3.9. values for the different clay minerals.
After the quantities of organic matter, carbonates, A semiquantitative analysis is sufficient for most ap-
free oxides, and nonclay minerals have been deter- plications. This may be done as follows. The silt and
Co
mined, the percentages of clay minerals are estimated sand fraction can be examined by microscope and the
using the appropriate glycol adsorption, cation ex- approximate proportion of nonclay minerals deter-
change capacity, K2O, and DTA data. The nonclays mined. The amount of clay size material ⫺2 m can
can be identified, and their abundance determined, us- be estimated by grain size distribution analysis. As a
ing the microscope, grain size distribution analysis, X- first approximation, it may be assumed that the amount
ray diffraction, and DTA. The amount of illite is of clay mineral equals at least the amount of clay size.
estimated from the K2O content since this is the only This assumption is justified for the following reasons.
clay mineral containing potassium. The amount of ka- Nonclay minerals, principally quartz, are found in the
olinite is most reliably determined from the 600C clay size fraction. On the other hand, for most soils,
Table 3.9 Summary of Clay Mineral Identification Criteria—Reference Data for Clay Mineral Identification
(⫺2-m fraction)
l
ria
Illite 10 60 25 8–10 End. 500–650 Broad
ate
Smectite 10–18 300 85 End. 600–750
End. 900
Exo. 950
Chlorite 14c 30 40 0 End. 610 10 or 720 20
a
For clays prepared at same relative humidity the size of the 100–300C endotherm (adsorbed water removal) increases
dM
in the order kaolinite–illite–smectite.
b
For samples started at 50% RH the amplitude of 600 peak/amplitude of adsorbed water peak ⬎⬎⬎1.
c
Heat treatment will accentuate 14 Å line and weaken 7 Å line.
the amount of clay mineral exceeds the amount of clay Interatomic bonding, crystal structure, and surface
hte
size. This most probably results from cementation of characteristics determine the size, shape, and stability
small clay particles into aggregates larger than 2 m of soil particles and the interactions of soil particles
in diameter. Approximate proportions of the different with liquids and gases. The structural stability of the
clay minerals in the clay fraction can be estimated different minerals controls their resistance to weath-
from the relative intensities of the X-ray diffraction ering and hence accounts in part for the relative abun-
reflections for each mineral. The presence of organic
rig
their mineralogy. Mineralogy also determines interac- although grain crushing becomes important in coarse-
tions with fluid phases. Together, these factors deter- grained soils such as sands and gravels when they are
mine plasticity, swelling, compression, strength, and under very high stresses.
fluid conductivity behavior. Thus, mineralogy is fun- The type of bonding between the unit layers of the
damental to the understanding of geotechnical prop- clay minerals, coupled with the adsorption properties
erties, even though mineralogical determinations are of the particle surfaces, controls soil swelling. Ad-
not made for many geotechnical investigations. In- sorption and desorption processes are important in
stead, other characteristics that reflect both composi- interactions between chemicals and soils. These inter-
tion and engineering properties, such as Atterberg actions in turn determine the flow and attenuation of
limits and grain size distribution, are determined. various substances through soil. Changes in surface
l
Since about 1980, environmental problems, espe- 7. As the geotechnical engineer on a project, you find
ria
cially those related to the safe disposal and contain- an inorganic soil containing 15 percent by weight
ment of municipal, hazardous, and nuclear waste and of particles finer than 100 m, as measured by
to the clean up of contaminated sites and the protection hydrometer analysis. What soil components do
of groundwater, have assumed a major role in geo- you expect? Why?
technical engineering practice. This has required a How could you confirm this expectation? Be spe-
ate
greatly increased focus on the compositional charac- cific in terms of tests and diagnostic criteria.
teristics of soils and their relation to the long-term 8. What is the smallest interplanar spacing that can
physical and chemical properties that control soil be- be measured by X-ray diffraction using copper K
havior under changed and extreme environmental con-
radiation?
ditions. dM 9. You suspect that a fine-grained soil sample con-
tains kaolinite, illite, and smectite minerals. De-
scribe in logical sequence the tests you would do
QUESTIONS AND PROBLEMS to verify that these clay minerals are present. In-
1. A montmorillonite has a cation exchange capacity dicate the reasons why you choose these tests and
of 130 meq/100 g and a total external and internal the criteria for distinguishing among the minerals.
surface area of 800 m2 /g. 10. An inorganic clay has a liquid limit of 350 percent.
hte
a. How many calcium ions will there be on a par- a. What is the most probable predominant clay
ticle that is 0.4 m ⫻ 0.2 m ⫻ one unit cell mineral in this soil?
in thickness? b. Explain the high liquid limit in terms of the
b. What percentage of the dry weight of the clay crystal structure of this mineral.
is composed of calcium? c. Would you recommend founding light struc-
rig
2. An orthorhombic crystal has axial ratios of 0.6, tures on shallow footings above this soil? Why?
0.3, and 1.0. The (500) plane is 2.0 Å horizontally 11. A soil sample has a cation exchange capacity of
from the origin. This crystal is irradiated with
30 meq/100 g and a specific surface area of 50
CuK X-rays (wave length of 1.54 Å). At what
m2 /g. You wish to determine the type of clay min-
py
clay mineral in the sample by weight percent- 14. The gradation curve for a sandy clay soil is shown
age? in Fig. 3.40.
c. What concerns would you have about this soil a. What are the percentages by weight of sand,
as a foundation material? silt, and clay size material?
d. How could you minimize any problems iden- b. Consider a 100-g sample of the soil and assume
tified in part (c)? that all sand particles are of a size equal to the
average particle size in the sand size range, the
13. In general the average clay particle size as repre- silt particles are of a size equal to the average
sented by some effective diameter D, for smectite particle size in the silt size range, and all clay
particles (S) is less than that of hydrous mica (il- particles are of a size equal to the average par-
lite) (HM) particles, which, in turn, is less than ticle size in the clay size range. Base your de-
l
that of kaolinite (K) particles. In addition, the av- termination of average particle size in each
ria
erage particle thicknesses are in the order range on equal weights of particles coarser and
finer than the average for each size range.
tS ⬍ tHM ⬍ tK Estimate the number of sand, silt, and clay
particle in the sample. For purposes of this
estimate, the sand and silt particles can be
ate
and values of the thickness-to-diameter ratio (t/D)
assumed to be spherical. Assume the clay
are in the order
particles to be flat disks having a diameter-to-
thickness ratio of 10. Assume the average size
(t/D)S ⬍ (t/D)HM ⬍ (t/D)K of clay particles on the gradation curve to rep-
resent the disk diameter.
What are some implications of these relationships c. Estimate the specific surface area of this soil in
dM
with respect to the relative values of plasticity, hy- square meters per gram. Determine the per-
draulic conductivity, compression–swell behavior, centages of this total that are contributed by the
and strength characteristics of three soils: one con- sand, silt, and clay fractions.
taining a large amount of smectite, one containing d. Are the estimates of the numbers of particles
a large amount of hydrous mica (illite), and one and specific surface area made in this way too
containing a large amount of kaolinite? high, too low, or correct? Why?
hte
rig
py
Co
l
Engineering Properties
ria
ate
4.1 INTRODUCTION ceeds 50 percent.1 The engineering properties of co-
hesionless soil are often determined by applied
The engineering properties of a soil depend on the confining pressure and looseness or denseness as in-
composite effects of several interacting factors. These dicated by the relation of the current void ratio to the
factors may be divided into two groups: compositional
dM lowest and highest possible values of void ratio for the
factors and environmental factors. Compositional fac- soil. The engineering properties of cohesive soil are
tors determine the potential range of values for any often characterized by stiffness and strength and by
property. They include: relating the current water content and past consolida-
tion history to the compositional characterization pro-
1. Types of minerals vided by the plasticity index. Some engineering
2. Amount of each mineral characteristics of coarse-grained and fine-grained soils
3. Types of adsorbed cations
hte
The influences of compositional factors on engineering Quantitative determination of soil behavior com-
properties can be studied using disturbed samples. pletely in terms of compositional and environmental
Environmental factors determine the actual value of factors is impractical for several reasons:
any property. They include:
1. Most natural soil compositions are complex, and
py
83
75 mm 5 mm 0.07 mm 0.002 mm
3 in. 0.2 in. 0.003 in. 0.00008 in.
l
ria
Particles visible without magnification Particles not visible without magnification
Grain size measurable with sieves Grain size not measurable with sieves
Grain size measured by sedimentation rate
Grains do not form a coherent mass Grains stick together when mixed with water due
ate
even when wet – cohesionless to pore water suction and physicochemical pore
fluid-mineral interaction – cohesive
Nonplastic – there is no range of water Plastic – deforms without cracking over a range
content where the soil can be deformed of water content between the liquid limit and the
without cracking or crumbling. plastic limit
dM Liquid (pancake batter)
Plastic (modeling clay)
Liquid Limit (LL)
Plastic Limit (PL)
Semisolid (chocolate bar)
Shrinkage Limit
Solid (chalk)
Permeability is moderate to high (10 -6 to 10-1 m/s). Permeability is low to very low (<10-7 m/s).
hte
Water flows easily through the voids. Water flows slowly through the voids.
Drainage takes weeks to tens of years.
Drainage occurs rapidly except under dynamic Both “drained” and “undrained” strengths are important.
loading; e.g., earthquakes. “Undrained” strength is low when preconsolidation
Only “drained” strength is important for conditions pressure is low.
other than earthquake loading or rapid landslides.
rig
l
ical, the liquid limit values for the bentonite–illite
ria
Study of soil composition in relation to soil properties mixtures were much less than predicted. This resulted
may be approached in two ways. In the first, natural from excess salt in the illite that, when mixed with the
soils are used, the composition and engineering prop- bentonite, prevented full interlayer expansion of the
erties are determined, and correlations are made. This montmorillonite particles in the presence of water.
method has the advantage that measured properties are
ate
those of naturally occurring soils. Disadvantages, how-
ever, are that compositional analyses are difficult and
time consuming, and that in soils containing several 4.3 ENGINEERING PROPERTIES OF
minerals or other constituents such as organic matter, GRANULAR SOILS
silica, alumina, and iron oxide the influence of any one
dM The mechanical behavior of granular materials is gov-
constituent may be difficult to isolate. erned primarily by their structure and the applied ef-
In the second approach, the engineering properties fective stresses. Structure depends on the arrangement
of synthetic soils are determined. Soils of known com- of particles, density, and anisotropy. Particle sizes,
position are prepared by blending different commer- shapes, and distributions, along with the arrangement
cially available clay minerals of relatively high purity of grains and grain contacts comprise the soil fabric.
with each other and with silts and sands. Although this The packing characteristics of granular materials are
approach is much easier, it has the disadvantages that discussed further in Chapter 5.
hte
ples difficult. Variations in composition and texture oc- the usual size distribution (or grading) curve where
cur in sediments within distances as small as a few particle diameters are shown to a logarithmic scale.
centimeters. Residual soils, in particular, are likely to The origin of a cohesionless soil can be reflected by
be very nonhomogeneous. its grading. Alluvial terrace deposits and aeolian de-
Co
A second difficulty is that the different constituents posits tend to be poorly graded or sorted. Glacial de-
of a soil may not influence properties in direct or even posits such as Boulder clays and tills are often well
predictable proportion to the quantity present because graded, containing a wide variety of particle sizes.
of physical and physicochemical interactions. As an Small particles in a well-graded soil fit into the voids
example of physical interactions, blending of equal between larger particles. Well-graded cohesionless
proportions of uniform sand and clay, each having a soils are relatively easy to compact to a high density
compacted unit weight of 17 kN/m3, would not nec- by vibration. The loss of fine fraction by internal ero-
essarily yield a mixture also having a unit weight of sion can lead to large changes in engineering proper-
17 kN/m3 after compaction. The resulting unit weight ties. Uniformly graded soils are usually used for
might be as high as 20 kN/m3 because the clay can controlled drainage applications because they are not
fill void spaces between sand particles. susceptible to loss of fines by internal erosion and their
l
ria
ate
dM
Figure 4.2 Interactions between clay minerals as indicated by liquid limit (data from Seed
et al., 1964).
hte
hydraulic conductivity can be maintained within defin- minimum void ratios change by mixing sand and silt
able and narrow limits. in different proportions. At low silt contents, silt par-
The slope of the grain size distribution curve is char- ticles fit into the voids between larger sand particles,
acterized by the coefficient of uniformity Cu: so the void ratio of sand–silt mixtures decreases with
increase in silt content. However, at a certain silt con-
rig
d60 tent, the silt fully occupies the voids, and the increase
Cu ⫽ (4.1) in silt content results in sand particles floating inside
d10
the silt matrix. Then, the void ratios increase with fur-
ther increase in silt content.
where d60 and d10 correspond to the sieve sizes that 60
py
density is defined by
(or minimum and maximum densities) reflecting the
loosest and densest states, respectively. Uniformly
graded soils tend to have a narrower range of possible emax ⫺ e
DR ⫽ ⫻ 100% (4.2)
densities compared to well-graded soils. Soils contain- emax ⫺ emin
ing angular particles tend to be less dense than soils
with rounded particles, as discussed later in this sec- in which emax, emin, and e are the maximum, minimum,
tion. However, angular and weak materials may crush and actual void ratios.
significantly more during compression, compaction, or The relative density correlates well with other prop-
deformation. Figure 4.4 shows how the maximum and erties of granular soils. As different standard test meth-
2.0
1.8 Maximum void ratio
1.4
Void ratio
1.2
1.0
0.8
0.6
0.4
l
ria
0.2
0.0
0 10 20 30 40 50 60 70 80 90 100
Silt content (%)
ate
Figure 4.4 Maximum and minimum void ratios of Monterey
sand–silt mixtures (from Polito and Martin, 2001).
properties.
such as surface smoothness, roundness of edges and
corners, and asperities, is important.
Particle Shape
With the exception of mica, most nonclay minerals
Particle shape is an inherent soil characteristic that in soils occur as bulky particles.2 Most particles are
plays a major role in mechanical behavior of soils.
Characterization of particle shape is scale dependent,
as shown in Fig. 4.5. At larger scales, that is, that of 2
Quartz particles become flatter with decreasing size and may have
the particle itself, the particle morphology might be a platy morphology when subdivided to a fineness approaching clay
described as spherical, rounded, blocky, bulky, platy, size (Krinsley and Smalley, 1973).
not equidimensional, however, and are at least slightly tion of aspect ratio and roundness. A convenient way
elongate or tabular. A frequency histogram of particle to characterize particle shapes in more detail is by a
length-to-width ratio (L/W) for Monterey No. 0 sand Fourier mathematical technique. For instance, the (R,
is shown in Fig. 4.6. This well-sorted beach sand is ) Fourier method is in the following form:
composed mainly of quartz with some feldspar. The
冘 (a cos n ⫹ b sin n)
mean of all the particle measurements is an L/W ratio N
of 1.39. This distribution is typical of that for many R() ⫽ a0 ⫹ n n (4.3)
n⫽1
sands and silty sands.
Particle morphology in soil mechanics has histori-
cally been described using standard charts against where R() is the radius at angle , N is the total num-
which individual grains may be compared. A typical ber of harmonics, n is the harmonic number, and a and
l
chart and some examples are shown in Fig. 4.7 (Krum- b are coefficients giving the magnitude and phase for
ria
bein, 1941; Krumbein and Sloss, 1963; Powers, 1953). each harmonic. The lower harmonic numbers give the
Sphericity is defined as the ratio of the diameter of a overall shape; for instance, the sphericity is expressed
sphere of equal volume to the particle to the diameter by the first and second harmonics. The coefficient val-
of the circumscribing sphere. Roundness is defined as ues for higher-order descriptors generally decay with
the ratio of the average radius of curvature of the cor- increasing descriptor or harmonic number, which ex-
ate
ners and edges of the particle to the radius of the max- presses smaller features (i.e., texture) (Meloy, 1977).
imum sphere that can be inscribed (Wadell, 1932). Other mathematical methods to curve-fit particle
Sphericity and roundness are measures of two very dif- shapes are listed in Table 4.1. Further discussion on
ferent morphological properties. Sphericity is most de- particle shape characterization is given by Barrett
pendent on elongation, whereas roundness is largely
dM (1980), Hawkins (1993), Santamarina et al. (2001), and
dependent on the sharpness of angular protrusions Bowman et al. (2001).
from the particle. Different definitions of sphericity In an assembly of uniform size spherical particles,
and roundness are available, as shown in Table 4.1. the loosest stable arrangement is the simple cubic
Due to the variety of definitions available, the quanti- packing giving a void ratio of 0.91. The densest pack-
fication of particle shape requires accurate specifica- ing is the tetrahedral arrangement giving a void ratio
tion of their definition. of 0.34. Particle shape affects minimum and maximum
In recent years, techniques for computer analysis of void ratios as shown in Fig. 4.8 (Youd, 1973). The
hte
shape data by digital imaging have improved greatly, values increase as particles become more angular or
and standard software applications include determina- the roundness (defined as roundness 1 in Table 4.1)
rig
py
Co
Figure 4.6 Grain shape distribution of Monterey No. 0 sand. Results are based on study of
277 particles, d50 ⫽ 0.43 mm, Cu ⫽ 1.4 (Mahmood, 1973).
0.9
0.7
Sphericity
0.5
l
ria
0.3
ate
0.1 0.3 0.5 0.7 0.9
Roundness
(a)
dM
High Sphericity
Low Sphericity
hte
Figure 4.7 Particle shape characterization: (a) Chart for visual estimation of roundness and
sphericity (from Krumbein and Sloss, 1963). (b) Examples of particle shape characterization
rig
decreases. When R ⫽ 1, the particle is a sphere. As the following linear fit to the relationship is proposed
py
particles become more angular, R decreases to zero. (Santamarina and Cho, 2004);
Void ratios are also a function of particle size distri-
bution; the values decrease as the range of particle repose ⫽ 42 ⫺ 17R (4.4)
sizes increases (increase in the coefficient of unifor-
mity Cu).
Co
Method Definition
Morphology—Sphere
Sphericity 1 Diameter of a sphere of equal volume
Diameter of circumscribing sphere
l
Volume of circumscribing sphere
ria
Sphericity 3
Projection sphericity Area of particle outline
Area of a circle with diameter equal to the longest length of outline
ate
Inscribed circle sphericity Diameter of the largest inscribed circle
Diameter of the smallest inscribed circle
dM Morphology—Ellipse
Eccentricity p /Rap, where the ellipse is characterized by Rp ⫹ p cos 2 in polar coordinates
Texture—Roundness
Mean radius
Morphology—Texture
Co
Fourier method Eq. (4.3), first and second harmonics, characterize sphericity, whereas higher harmonics
(around 10th) characterizes roundness. Surface texture is characterized by much
higher harmonics.
Fourier descriptor More flexible than the Fourier method by using the complex plane (Bowman et al.,
method 2001). Lower harmonics give shape characteristics such as elongation, triangularity,
squareness, and asymmetry. Higher harmonics (larger than 8th) give textural features.
Fractal analysis Use as a measure of texture (Vallejo, 1995; Santamarina, et al. 2001).
From Hawkins (1993), Santamarina et al. (2001), and Bowman et al. (2001).
1.2
the particle assembly. Further details on the relation
An
between particle stiffness and particle assemblage stiff-
gu ness are given in Chapter 11.
1.0 lar
R=
0.2
0
Particle Strength
0.8 0.20
Sub
l
an The crushability of soil particles has large effects on
Sub gular
ria
rou
nde 0.25 the mechanical behavior of granular materials. At high
Rou d
0.6 nde 0.30 stresses, the compressibility of sand becomes large as
d
0.35 a result of particle crushing, and the shape of an e–log
0.49 p compression curve becomes similar to that of nor-
Minimum Void Ratio, emin
0.8
0.70 mally consolidated clay (Miura et al., 1984; Coop,
ate
Ang
ular
R=0
1990; Yasufuku et al., 1991). Under constant states of
.20
0.6 stress, the amount of particle breakage increases with
Ang
ular
R=0 time, contributing to creep of the soil (Lade et al.,
.17
0.20
1996). The amount of crushing in a soil mass depends
0.4 Suba both on the stiffness and strength of the individual
ngula
r 0.25 0.35 grains and how applied stresses are transmitted through
0.2
Subrounded
Rounded
dM 0.30 0.49
0.70 the assemblage of soil particles.
1 2 3 4 6 10 15 Particle strength or hardness is characterized by
Coefficient of Uniformity, Cu crushing at contacts or particle tensile splitting. There
is a statistical variation in grain strength for particles
Figure 4.8 Maximum and minimum void ratios of sands as of a specified material and of a given size (Moroto and
a function of roundness and the coefficient of uniformity Ishii, 1990; McDowell, 2001). Random variation in
(from Youd, 1973).
hte
50
Table 4.2 Elastic Properties of Geomaterials at
Angle of repose φrepose
Room Temperature
40
Young’s Shear Modulus Poisson’s
Co
l
1.18 37.9 35.6
ria
1.4 46.7 42.4
1.7 39.6 38.5
Silica sand 0.5 147.4 132.5 McDowell (2001)
1 66.7 59.0
ate
2 41.7 37.3
Silica sand 0.28 110.9 147.3 Nakata et al. (2001)
0.66 72.9 73.1
1.55 31.0 29.7
dM Feldspar
Aio feldspar sand 0.85 20.9 24.6 Nakata et al. (1999)
1.0 24.3 22.8
1.18 18.1 18.2
1.4 23.1 21.4
1.7 18.9 18.3
Calcareous Sand
hte
50 — 1.1
Carboniferous limestone 5 — 14.9
particle 8 — 12.2 Lee (1992)
12 — 10.3
20 — 8.3
py
30 — 7.0
40 — 6.2
50 — 5.7
Quiou sand 1 109.3 96.19 McDowell and Amon (2000)
Co
2 41.4 36.20
4 4.2 3.87
8 0.73 0.63
16 0.61 0.54
Others
Masado decomposed granite
soil 1.55 24.2 22.1 Nakata et al. (2001)
Glass beads 0.93 365.8 339.6 Nakata et al. (2001)
Angular glass 0.93 62.1 60.0 Nakata et al. (2001)
a
Stress below which 37% of the particles do not fracture.
b
Force/d 2 at which particle of size d is crushed.
l
ment of different size particles. It can be argued that 20
ria
larger size particles are more likely to break because
the normal contact forces in a soil element increase
with particle size and the probability of a defect in a 0
0.01 0.1 1.0
given particle increases with its size as shown in Fig
Grain size (mm)
4.10 (Hardin, 1985). However, if a larger particle has
ate
contacts with neighboring particles (i.e., larger coor- Figure 4.11 Evolution of particle size distribution curve
dination number), the load on it is distributed, and the upon crushing (from Hagerty et al., 1993).
probability of facture is less than for a condition with
fewer contacts. Experimental evidences suggest that
fines increase as particles break by increase in applied
dM
pressure. For example, the evolution of particle size tact with many smaller particles. The very smallest
distribution curves for Ottawa sand in one-dimensional particles have a lower coordination number because
compression is shown in Fig. 4.11 (Hagerty et al., there are fewer smaller particles available for contact.
1993). Hence, the coordination number dominates over Hence, the largest particles in the aggregate become
size-dependent particle strength. Larger particles have protected by the surrounding newly formed smaller
higher coordination numbers because they are in con- particles, and smaller particles are more likely to break
hte
Carboniferous Limestone
rig
10
Particle Strength (MPa)
5.0
py
Co
0.5
0.2
1 5 10 50 100
Average Particle Size (mm)
Figure 4.10 Relationship between tensile strength and particle size (from Lee, 1992).
w WS C WS
VW ⫹ VC ⫽ ⫹ (4.6)
4.4 DOMINATING INFLUENCE OF THE CLAY 100 w 100 GSC w
PHASE
If clay and water completely fill the voids in the gran-
In general, the more clay in a soil, the higher the plas- ular phase, then
ticity, the greater the potential shrinkage and swell, the
冉 冊
l
lower the hydraulic conductivity, the higher the com- w Ws C W C Ws
ria
⫹ ⫽ 1⫺ e (4.7)
pressibility, the higher the cohesion, and the lower the 100 w 100 GSC w 100 GSG w G
internal angle of friction. Whereas surface forces and
their range of influence are small relative to the weight which simplifies to
and size of silt sand particles, the behavior of small
冉 冊
ate
and flaky clay mineral particles is strongly influenced w C C eG
by surface forces, as discussed in Chapter 6. Water is ⫹ ⫽ 1⫺ (4.8)
100 100GSC 100 GSG
strongly attracted to clay particle surfaces, also dis-
cussed in Chapter 6, and results in plasticity, whereas The void ratio of a granular material composed of
nonclay particles have much smaller specific surface bulky particles is of the order of 0.9 in its loosest pos-
and less affinity for water and do not develop signifi-
dM sible state. The specific gravity of the nonclay fraction
cant plasticity, even when in finely ground form. in most soils is about 2.67, and that of the clay fraction
If it is assumed as a first approximation that all of is about 2.75. Inserting these values in Eq. (4.8) gives
the water in a soil is associated with the clay phase,
the amount of clay required to fill the voids of the C ⫽ 48.4 ⫺ 1.42w (4.9)
granular phase and prevent direct contact between
granular particles can be estimated for any water con- This relationship indicates that for water contents typ-
tent. The weight and volume relationships for the dif-
hte
gravity of the granular particles. The volume of voids icantly influence properties. For example, just 1 or 2
in the granular phase is eG VGS, where eG is the void percent of highly plastic clay present in gravel used as
ratio of the granular phase and VGS is the volume of a fill or aggregate may be sufficient to clog handling
granular solids, given by and batching equipment.
py
Co
l
effective size, uniformity) only in a far more complex responds approximately to the water content at which
ria
manner.’’ a soil has an undrained shear strength of about 2.5 kPa.
Casagrande (1932b) developed a standard device for Subsequent studies have indicated that the liquid limit
determination of the liquid limit and noted that the for all fine-grained soils corresponds to shearing resis-
nonclay minerals quartz and feldspar did not develop tance of about 1.7 to 2.0 kPa and a pore water suction
plastic mixtures with water, even when ground to sizes of about 6 kPa (Russell and Mickle, 1970; Wroth and
ate
smaller than 2 m. Further studies led to the formation Wood, 1978; Whyte, 1982).
of a soil classification system based on the Atterberg Liquid limit values are determined using both the
limits for identification of cohesive soils (Casagrande, Casagrande liquid limit device and the fall cone device.
1948). This system was adopted, with minor modifi- Different standards adopt different devices and, there-
cations, as a part of the Unified Classification System. fore, correlations based on liquid limit should be used
A plot of plasticity index as a function of liquid limit with some caution. The variation of undrained shear
dM
that is divided into different zones, as shown in Fig. strength with water content can be obtained from a
4.13, is termed the plasticity chart. This chart forms series of fall cone tests and solutions are available us-
an essential part of the Unified Soil Classification Sys- ing the theory of plasticity for various geometries used
tem. in fall cones (Houlsby, 1982; Koumoto and Houlsby,
hte
rig
py
Co
2001). Furthermore, with the aid of critical state soil greater the total amount of water required to reduce
mechanics (see Chapter 11), some other engineering the strength to that at the liquid limit. The specific
properties, such as compressibility, can be deduced surface areas of the different clay minerals (Table 3.6)
(Wood, 1990). are consistent with the liquid limit values of different
Values of hydraulic conductivity at the liquid limit clay minerals in Table 4.5. Additional support for this
for several clays are given in Table 4.4, from Nagaraj concept is given by the following relationship found
et al. (1991). The striking aspect of these data is that, for 19 British clays:
although the water contents and void ratios at the liq-
uid limit for the different clays vary over a very wide LL ⫽ 19 ⫹ 0.56As (20%) (4.10)
range, the hydraulic conductivity is very nearly the
same for all of them. This means that the effective pore where LL is the liquid limit and As is the specific sur-
l
sizes controlling fluid flow must be about the same for face in square meters per gram (Farrar and Coleman,
ria
all the clays at their liquid limit. Such a microfabric is 1967).
consistent with the cluster model for hydraulic con- The effects of electrolyte concentration, cation va-
ductivity discussed in Chapter 9. In this model, the lence and size, and dielectric constant of the pore fluid
individual clay particles associate into aggregates or on the liquid limit of kaolinite and montmorillonite are
flocs, as shown schematically in Fig. 9.11. The size of
ate
illustrated and discussed by Sridharan (2002). The ef-
voids between the clusters or aggregates controls the fects are generally consistent with the above interpre-
flow rate according to either model. tation and can be explained also through double-layer
The approximately equal strengths, pore water suc- (see Chapter 6) influences on swelling, flocculation
tions, and hydraulic conductivities for all clays at their and deflocculation of clay particles, and shear strength.
liquid limit can be explained by the concepts that (1)
dM
the aggregates or clusters are the basic units that in-
teract to develop the strength, that is, the aggregates Plastic Limit
act somewhat like single particles, (2) the average ad- The plastic limit has been interpreted as the water con-
sorbed water layer thickness is about the same on all tent below which the physical properties of the water
particle surfaces, and (3) the average size of interclus- no longer correspond to those of free water (Terzaghi,
ter pores is the same for all clays. Concept 2 provides 1925a) and as the lowest water content at which the
the key to why different clays have different values of cohesion between particles or groups of particles is
hte
liquid limit. All clays have essentially the same surface sufficiently low to allow movement, but sufficiently
structures, that is, a layer of oxygen atoms in tetrahe- high to allow particles to maintain the molded posi-
dral coordination with silicon, or a layer of hydroxyls tions (Yong and Warkentin, 1966). Whatever the struc-
in octahedral coordination with aluminum or magne- tural status of the water and the nature of the
sium. The forces of interaction between these surfaces interparticle forces, the plastic limit is the lower
and adsorbed water should be about the same for the boundary of the range of water contents within which
rig
different clay minerals. Thus, the amount of water ad- the soil exhibits plastic behavior; that is, above the
sorbed per unit area of surface that corresponds to a plastic limit the soil can be deformed without volume
pore water suction of 6 kPa should be about the same. change or cracking and will retain its deformed shape;
This means that the greater the specific surface, the below the plastic limit it cannot. Plastic limit values
py
l
Dehydrated halloysite (3) 35–55 30–45
ria
Attapulgite (4) 160–230 100–120
Chlorite (5) 44–47 36–40
Allophane (undried) 200–250 130–140
a
(1) Various ionic forms. Highest values are for monovalent; lowest
ate
are for di- and trivalent. (2) All samples 10% clay, 90% sand and silt.
(3) Various ionic forms. Highest values are for di- and trivalent; lowest
are for monovalent. (4) Various ionic forms. (5) Some chlorites are
nonplastic.
Data Sources: Cornell University (1950), Samuels (1950), Lambe
and Martin (1955), Warkentin (1961), and Grim (1962).
dM
for different clay minerals are listed in Table 4.5. The plasticity index
Activity ⫽ (4.12)
undrained shear strength at the plastic limit is reported % ⬍ 2 m
to be in the ranges of 100 to 300 kPa with an average
value of 170 kPa (Sharma and Bora, 2003). For many clays, a plot of plasticity index versus clay
hte
plasticity index fluence of the clay fraction on properties and the more
susceptible their values to changes in such factors as
wherein the plasticity index is given by PI ⫽ LL ⫺ type of exchangeable cations and pore fluid composi-
PL. The liquidity index is useful for expressing and tion. For example, the activity of Belle Fourche mont-
py
comparing the consistencies of different clays. It nor- morillonite varies from 1.24 with magnesium as the
malizes the water content relative to the range of water exchangeable cation to 7.09 for sodium saturation of
content over which a soil is plastic. It correlates well the exchange sites. On the other hand, the activity of
with compressibility, strength, and sensitivity proper- Anna kaolinite only varies from 0.30 to 0.41 for six
ties of fine-grained soils as illustrated in later chapters different cation forms (White, 1955).
Co
of this book.
l
ria
ate
dM
Figure 4.14 Relationship between plasticity index and clay fraction (from Skempton, 1953).
hte
Table 4.6 Activities of Various Clay Minerals sions ordinarily flocculate in the presence of divalent
and trivalent cations.
Mineral Activity pH influences interparticle repulsions because of its
effects on clay particle surface charge. Positive edge
Smectites 1–7 charges can exist in low pH environments. These ef-
Illite 0.5–1 fects are of greatest importance in kaolinite, lesser
rig
ther in Chapter 6.
concentration is low. On the other hand, divalent and 4.8 ENGINEERING PROPERTIES OF CLAY
Co
l
age on drying (the lower the shrinkage limit).
ria
Plasticity values for different clay minerals are listed
in Table 4.5 in terms of ranges in the liquid, plastic,
and shrinkage limit values. Most of the values were Particle Size and Shape
determined using samples composed of particles finer Different clay minerals occur in different size ranges
than 2 m. Several general conclusions can be made (Table 3.6) because mineralogical composition is a ma-
ate
concerning the Atterberg limits of the clay minerals. jor factor in determining particle size. There is some
concentration of different clay minerals in different
1. The liquid and plastic limit values for any one bands within the clay size range (less than 2 m), as
clay mineral species may vary over a wide range. indicated in Table 4.7. The shapes of the most common
2. For any clay mineral, the range in liquid limit clay minerals are platy, except for halloysite, which
values is greater than the range in plastic limit occurs as tubes (Fig. 3.21). Particles of kaolinite are
values.
dM relatively large, thick, and stiff (Fig. 3.13). Smectites
3. The variation in values of liquid limit among dif- are composed of small, very thin, and filmy particles
ferent clay mineral groups is much greater than (Fig. 3.25). Illites are intermediate between kaolinite
the variation in plastic limits. and smectite (Fig. 3.29) and are often terraced and thin
hte
Particle
Size Predominating Common Rare
rig
l
erals is less than about 1 ⫻ 10⫺7 m/s and may range
ria
to values less than 1 ⫻ 10⫺12 m/s for some of the
monovalent ionic forms of smectite minerals at low Figure 4.15 Ranges in effective stress failure envelopes for
porosity. The usual measured range for natural clay pure clay minerals and quartz (from Olson, 1974). Reprinted
soils is about 1 ⫻ 10⫺8 to 1 ⫻ 10⫺10 m/s. For clay with permission of ASCE.
ate
minerals compared at the same water content, the hy-
draulic conductivities are in the order smectite (mont-
morillonite) ⬍ attapulgite ⬍ illite ⬍ kaolinite.
Shear Strength
There are many ways to measure and express the shear
dM
strength of a soil, as described in most geotechnical
engineering textbooks. In most cases, a Mohr failure
envelope, where shear strength (usually peak, critical
state, or residual) is plotted as a function of the direct
effective stress on the failure plane, or a modified Mohr
diagram, in which maximum shear stress is plotted ver-
hte
⫽ c ⫹ n tan (4.13)
strength, for pure clay minerals and quartz are shown a clay is composed of two distinct parts: a cohesion
in Fig. 4.15. The increase in shear strength with in- that depends only on void ratio (water content), and a
crease in effective stress, that is, the friction angle, is frictional contribution, dependent only on normal ef-
greatest for the nonclay mineral quartz, followed in fective stress. Evaluation of these two parts was done
descending order by kaolinite, illite, and montmoril- by measurement of the strength of two samples both
lonite. The ranges in the position of a failure envelope at the same void ratio or water content, but at different
for a given mineral result from differences in such fac- levels of effective stress. This condition is obtained by
tors as fabric, adsorbed cation, pH, and overconsoli- using one normally consolidated and one overconsol-
dation ratio. A similar pattern of failure envelopes for idated sample. The strength parameters determined in
some natural soils is shown in Fig. 4.16. The finer this way, often termed the Hvorslev parameters or true
l
cohesion intercept is either zero or very small, except contents less than 50 percent. This is because with ex-
ria
for cemented soils. Thus, a significant true cohesion, pansive clay minerals (montmorillonite) or flocculated
if defined as strength in the absence of normal stress fabrics (30 g salt/liter) the ratio of volume of wet clay
on the failure plane, does not exist in the absence of to volume of quartz is greater than the ratio of dry
chemical bonding. These considerations are discussed volumes. It is further illustration of the dominating in-
in more detail in Chapter 11. fluence of the clay phase discussed earlier.
ate
dM
hte
rig
py
Co
Figure 4.17 Residual friction angles for clay–quartz mixtures and natural soils (from Ken-
ney, 1967).
l
Compression index values for a number of different Approximate ranges of the coefficient of consoli-
ria
natural clays are shown in Fig. 4.18 as a function of dation for natural clays are given in Fig. 4.19. The
plasticity index (Kulhawy and Mayne, 1990). The val- above values for pure clays and clay mineral mixtures
ues for pure clays plot generally within the defined are within the same general ranges. One conclusion
ranges in Fig. 4.18. The compression index for un- that can be drawn from the comparability of compres-
sion index and coefficient of consolidation values for
ate
loading and reloading is about 20 percent of the value
for virgin compression. natural clays with those for pure clays is that the clay
As both compressibility and hydraulic conductivity phase dominates the compression and consolidation
are strong functions of soil composition, the coefficient behavior, with the nonclay material playing a passive
of consolidation cv is also related to composition be- role as relatively inert filler.
cause cv is directly proportional to hydraulic conduc-
dM
tivity and inversely proportional to the coefficient of Swelling and Shrinkage
compressibility.4 Values of cv determined in one study The actual amount of volume change of a soil in re-
sponse to a change in applied stress depends on the
4
The coefficient of compressibility av is the negative of the rate of environmental factors listed in Section 4.1 as well as
change of void ratio with effective stress. on the cation type, electrolyte type and concentration,
hte
rig
py
Co
Figure 4.18 Compression and unload–reload indices as a function of plasticity index (from
Kulhawy and Mayne, 1990). Reprinted with permission from EPR1.
p.s.i.)
re - 1
p.s.i.)
ressu
re - 1
ge P
ressu
char
ge P
(Sur
.)
.s.i
l
char
1p
ria
e-
(Sur
sur
res
Figure 4.19 Coefficient of consolidation as a function of
eP
liquid limit (from NAVFAC, 1982).
rg
ha
c .
i.)
ur p.s
(S 4
ate
6.9
e-
sur
and pore fluid dielectric constant. However, the poten- Pres
e
harg
tial total amount of swell or shrinkage is determined (Surc
by the type and amount of clay. From a consideration
of the clay mineral structures and interlayer bonding
dM
(Chapter 3), it would be expected that smectite and
vermiculite should undergo greater volume changes on
wetting and drying than do kaolinite and hydrous mica. Figure 4.20 Four correlations between swelling potential
Experience indicates clearly that this is indeed the and plasticity index (from Chen, 1975).
case. In general, the swelling and shrinking properties
of the clay minerals follow the same pattern as their
plasticity properties, that is, the more plastic the min-
hte
eral, the more potential swell and shrinkage. Illustra- sure during swelling is clearly shown. The tests by
tions of the influences of adsorbed cation type and pore Seed et al. (1962b) were done using artificial mixtures
fluid composition are given in Chapter 10 and by Srid- of sand and clay minerals compacted at optimum water
haran (2002). content using Standard AASHTO compactive effort al-
Because of the many problems encountered in the lowed to swell under a surcharge pressure of 1 psi (7
performance of structures founded on high volume kPa). The measurements by Holtz and Gibbs (1956)
rig
change soils, numerous attempts have been made to were made using both undisturbed and remolded sam-
develop reliable methods for their identification. The ples allowed to swell from an air-dry state to saturation
most successful of these are based on the determina- under a surcharge of 1 psi (7 kPa).
tion of some factor that is related directly to the clay The results of the tests on artificial sand–clay min-
mineral composition, such as shrinkage limit, plasticity eral mixtures obtained by Seed et al. (1962b) correlate
py
index, activity, and percentage finer than 1 m. well with compositional factors that reflect both the
Simple, unique correlations between swell or swell type and amount of clay, that is, the activity A, defined
pressure and these parameters that reflect only the type as
PI/
C, and the percent clay size C (% ⬍ 2 m),
and amount of clay are not possible because of the according to
Co
l
normal strength test. The variation in creep rate for
ria
these specimens as a function of plasticity index is
shown in Fig. 4.23. The correlation is reasonably
unique because the plasticity index reflects both the
type and amount of clay.
ate
4.9 EFFECTS OF ORGANIC MATTER
Organic matter in soil may be responsible for high
Figure 4.21 Classification chart for swelling potential (mod- plasticity, high shrinkage, high compressibility, low
ified from Seed et al., 1962b).
dM
S ⫽ 2.16 ⫻ 10⫺3 (PI)2.44 (4.15)
Time-Dependent Behavior
Different soil types undergo varying amounts of time-
dependent deformations and stress variations with
time, as exhibited by secondary compression, creep,
Co
l
Increasing the organic carbon content by only 1 or
ria
2 percent may increase the limits by as much as an
increase of 10 to 20 percent in the amount of material
finer than 2 m or in the amount of montmorillonite
(Odell et al., 1960). The influences of organic matter
content on the classification properties of a soft clay
ate
from Brazil are shown in Fig. 4.24.
The maximum compacted densities and compressive
strength as a function of organic content of both nat-
ural samples and mechanical mixtures of inorganic
dM soils and peat are shown in Figs. 4.25 and 4.26, re-
spectively. Both the compacted density and strength
decrease significantly with increased organic content
and the relationships for natural samples and the
mixtures are about the same. Increased organic content
also causes an increase in the optimum water content
for compaction.
The large increase in compressibility as a result of
hte
hydraulic conductivity, and low strength. Soil organic in time after the end of primary consolidation.
matter is complex both chemically and physically, and The effect of organic matter on the strength and
many reactions and interactions between the soil and stiffness of soils depends largely on whether the or-
the organic matter are possible (Oades, 1989). It may ganic matter is decomposed or consists of fibers that
occur in any of five groups: carbohydrates; proteins;
py
down to 0.1 m in size. The specific properties of the creasing the strength.
colloidal particles vary greatly depending upon parent
material, climate, and stage of decomposition.
The humic fraction is gel-like in properties and neg- 4.10 CONCLUDING COMMENTS
atively charged (Marshall, 1964). Organic particles can
strongly adsorb on mineral surfaces, and this adsorp- Knowledge of soil composition is a useful indicator of
tion modifies both the properties of the minerals and the probable ranges of geotechnical properties and
the organic material itself. Soils containing significant their variability and sensitivity to changes in environ-
amounts of decomposed organic matter are usually mental conditions. Although quantitative values of
characterized by a dark gray to black color and an odor properties for analysis and design cannot be derived
of decomposition. At high moisture contents, decom- from compositional data alone, information on com-
l
ria
ate
dM
Figure 4.24 Influence of organic content on classification properties of Juturnaiba organic
clay, Brazil (from Coutinho and Lacerda, 1987).
position can be helpful for explaining unusual behav- and more direct evaluation of their significance is
hte
ior, identification of expansive soils, selection of needed. Examples of some soil types in which these
sampling and sample handling procedures, choice of factors may be especially important are decomposed
soil stabilization methods, and prediction of probable granite, tropical residual soils, volcanic ash soils, col-
future behavior. lapsing soils, loess, and carbonate sand, as discussed
For example, if it is known that a soil to be used in in more detail by Mitchell and Coutinho (1991).
earthwork construction contains either hydrated halloy-
rig
as a result of leaching and solutioning processes. larger particles and why angular particles are more
In many cases, the effects of composition on behav- susceptible to breakage than round particles.
ior are reflected by information on particle size, shape,
3. Using Figs. 4.8 and 4.11, show how the maximum
and size distribution of the coarse fraction, and the
and minimum void ratio changes with applied load
Atterberg limits of the fine fraction. On large projects
as particles progressively break and the coefficient
and whenever unusual behavior is encountered, how-
of uniformity Cu increases. Plot the data in e–log
ever, compositional data are valuable aids for interpre-
v space and discuss the result.
tation of observations. Furthermore, the influences of
compositional and structural factors are not always ad- 4. Using Eq. (4.8), derive a relationship between C
equately reflected by the usual classification properties, (the percentage of clay) versus w (water content)
l
ria
ate
dM Figure 4.27 Effect of organic content on the compressibility
properties of Juturnaiba organic clay, Brazil (from Coutinho
and Lacerda, 1987).
instead of clay?
5. Using the reported undrained shear strengths at liq-
uid limit and plastic limit, derive a relationship be-
tween the compression index Cc and plasticity index
PI. Assume that the ratio of undrained shear
strength to vertical effective stress, su / v, is 0.3.
rig
e. Kaolinite at its liquid limit (specific surface ⫽ soil—discuss the strengths and weaknesses of the
15 m2 /g of dry clay), liquid limit ⫽ 70 percent Unified Soil Classification System (USCS) in pro-
f. Kaolinite at its plastic limit (specific surface ⫽ viding a clear and unambiguous picture of the prob-
15 m2 /g of dry clay), plastic limit ⫽ 30 percent able behavior of the following soil types. In
developing your answer, be specific concerning
7. By examining the data presented in Figs. 4.24 and what is measured and the terms of reference used
4.29, discuss why organic clays exhibit larger com- in the USCS and what is most important in deter-
pressibility compared to inorganic clays (see Fig. mining any property being discussed. (Note: Some
4.18). of the information in Chapter 8 may be useful in
8. Assume that you are able to determine accurate, developing your answer to this question.)
reliable quantitative values for all details of the min- a. Clean sand
l
eralogical, chemical, and biological constituents of b. Decomposed granite
ria
a given soil. All particle sizes, shapes, and distri- c. Calcareous sand
butions are also known. Speculate on your ability
d. Organic silt
to predict the volume change, strength, and per-
meability properties of this soil over a range of wa- e. Expansive clay
f. Glacial till
ate
ter contents. Give reasons for why you would have
low or high confidence in your predictions. g. Loess
9. In light of what is known about the dependence of h. Dispersive clay
engineering properties on soil composition—both i. Volcanic ash
of the particles and of the other phases present in a
dM j. Estuarine mud
hte
rig
py
Co
l
Measurement
ria
ate
5.1 INTRODUCTION niques that incorporate small-scale features such as
inhomogeneity and microfractures into continuum
Although soils are composed of discrete soil particles models became available (Mura, 1987; Nemat-Nasser
and particle groups, a soil mass is almost always and Hori, 1999). Increased computational speeds al-
treated as a continuum for engineering analysis and
dM lowed simulation of an assembly of individual soil par-
design. Nonetheless, the specific values of properties ticles by modeling particle contact behavior, and this
such as strength, permeability, and compressibility de- led to the development of numerical methods such as
pend on the size and shape of the particles, their ar-
the discrete/distinct element method and contact dy-
rangements, and the forces between them. Thus, to
namics (Cundall and Strack, 1979; Moreau, 1994;
understand a property requires knowledge of these fac-
Cundall, 2001). In the early developments, simulations
tors. Furthermore, new theories of particulate mechan-
were limited to an assembly of two-dimensional cir-
ics and computational methods based on these theories
hte
their relationships to mechanical properties. In the late be investigated using distinct element methods.
1960s, knowledge expanded rapidly, sparked by im- Other innovations in the past two decades have led
proved techniques of sample preparation and the de- to improved material measurement techniques and
velopment of the scanning electron microscope. In the their interpretation using computers. These include the
Co
early 1970s attention was directed also at particle ar- environmental scanning electron microscopy (ESEM),
rangements in cohesionless soils. From this work came nanoindentation and probing, complex digital image
a realization that characterization of the properties of analysis, magnetic resonance imaging (MRI), X-Ray
sands and gravels cannot be done in terms of density tomography, and laser-aided tomography. Some of
or relative density alone, as had previously been them have been used to characterize the microscopic
thought. Particle arrangements and stress history must properties of soils (Oda and Iwashita, 1999).
be considered in these materials as well. The more established methods for studying and,
In the 1970s and 1980s, micromechanics theories where possible, quantifying the arrangements of par-
were developed that aimed to relate microstructure to ticles, particle groups, and voids in different soils are
macroscopic behavior. Various homogenization tech- described and illustrated in this chapter. Some ele-
109
l
to the combined effects of fabric, composition, and in-
ria
terparticle forces. Methods for determination of soil
fabric are described and examples of different fabric
types are given in the following sections. The impor-
tance of soil fabric as a factor determining soil prop-
ate
erties and behavior is discussed and illustrated in
Chapter 8. In practice, special problems, unusual soils,
and the need to ensure that measured properties prop-
erly reflect the in situ conditions may require appli-
cation of these testing and interpretation methods.
It is necessary to consider the size, the form, and
dM
the function of different fabric units and to keep in
mind the scale at which the fabric is of interest. For
example, a carefully compacted clay liner for an im-
poundment may have uniformly and closely packed
particle groups within it, thus giving a material with
very low hydraulic conductivity. If, however, the liner
becomes broken into sections measuring a meter or so
hte
l
Particle Associations in Soils
ria
Particle associations in sediments, residual soils, and
compacted clays assume a variety of forms; however,
most of them are related to combinations of the con-
figurations shown in Fig. 5.1 and reflect the difference
ate
in water content between a suspension and a denser
soil mass. Fine-grained soils are almost always com-
posed of multiparticle aggregates. Overall, three main
groupings of fabric elements may be identified (Collins
and McGown, 1974):
dM
1. Elementary Particle Arrangements Single
forms of particle interaction at the level of indi-
vidual clay, silt, or sand particles
2. Particle Assemblages Units of particle organi-
zation having definable physical boundaries and
a specific mechanical function, and which consist
of one or more forms of the elementary particle Figure 5.2 Schematic representation of elementary particle
hte
number of additional terms have also been used to de- cluster and total void ratios. The term ped (Brewer,
scribe the same or similar features. 1964) has a similar meaning to cluster.
Cardhouse is an edge-to-face arrangement forming
an open fabric similar to the edge-to-face flocculated Fabric Scale
Co
but dispersed arrangement of Fig. 5.1c (Goldschmidt, The fabric of a soil may be viewed relative to three
1926). A domain (Aylmore and Quirk, 1960, 1962) or levels of scale. From smallest to largest they are:
packet or book (Sloane and Kell, 1966) is an aggregate
of parallel clay plates. An array of such fabrics is 1. Microfabric The microfabric consists of the
termed a turbostratic fabric and is similar to the inter- regular aggregations of particles and the very
weaving bunches of Fig. 5.3h. An edge-to-face asso- small pores between them. Typical fabric units
ciation of packets or books is termed a bookhouse and are up to a few tens of micrometers across.
is similar to the arrangement of Fig. 5.1e. A cluster is 2. Minifabric The minifabric contains the aggre-
a grouping of particles or aggregates into larger fabric gations of the microfabric and the interassem-
l
ria
ate
dM
hte
rig
Figure 5.3 Schematic representations of particle assemblages (Collins and McGown, 1974).
(a) connectors, (b) connectors, (c) connectors, (d ) irregular aggregations by connector as-
semblages, (e) irregular aggregations in a honeycomb, (ƒ ) regular aggregation interacting
with particle matrix, (g) interweaving bunches of clay, (h) interweaving bunches of clay with
silt inclusions, (i) clay particle matrix, and ( j ) granular particle matrix.
py
blage pores between them. Minifabric units may and minifabrics. Time-dependent deformations such as
be a few hundred micrometers in size. creep and secondary compression are controlled most
Co
3. Macrofabric The macrofabric may contain strongly by the mini- and microfabric.
cracks, fissures, root holes, laminations, and the
like that correspond to the transassemblage pores
shown in Fig. 5.6. 5.3 SINGLE-GRAIN FABRICS
Soil mechanical and flow properties depend on de- Sand and gravel particles are sufficiently large and
tails of these three levels of fabric to varying degrees. bulky that they ordinarily behave as independent units.
For example, the hydraulic conductivity of a fine- Attempts to describe the stress–deformation behavior
grained soil is almost totally dominated by the macro- of granular soils using particulate mechanics theories
l
ria
ate
dM
Figure 5.4 Schematic representation of pore space types (Collins and McGown, 1974).
[e.g., Newland and Allely (1957), Rowe (1962, 1973), In some cases, sand samples can be dried prior to
Horne (1965), Matsuoka (1974), Murayama (1983), impregnation since sand fabrics are not generally af-
Nemat-Nasser and Mehrabadi (1984), and Wan and fected by capillary stresses. A procedure for doing this
hte
Guo (2001)] have met with some success. The devel- to enable study of the fabrics produced in Monterey
opment of discrete element methods for numerical No. 0 sand by different methods of compaction is
modeling of granular soils has greatly extended the given by Mitchell et al. (1976).
potential for these methods as discussed in Section 5.1.
These theories are based on elastic distortion of par-
rig
ticles and the sliding and rolling of particles, usually Packing of Equal-Sized Spheres
assumed of spherical or disk shape. In real granular Regular packing of spheres of the same size provides
soils, the irregular particle shapes and distribution of insight into the maximum and minimum possible den-
sizes mean that packing is usually far from regular. sities, porosities, and void ratios that are possible in
Nonetheless, the theories and computations can pro- single-grain fabrics. Five different possible packing ar-
py
vide valuable insights into behavior, and knowledge of rangements are shown in Fig. 5.7, and properties of the
the characteristics of ideal systems can be useful for arrangements shown are listed in Table 5.1. The range
interpreting data on real soils (see Chapter 11). of possible porosities is from 25.95 to 47.64 percent,
and the corresponding range of void ratios is from 0.35
Co
l
ria
ate
dM
hte
Figure 5.5 Scanning electron photomicrograph features of undisturbed soil fabrics (Collins
and McGown, 1974). (a) Partly discernible particle systems in Lydda silty clay, Israel (fresh-
water alluvial deposit); (b) grain–grain contacts in Ford silty loess, England (aeolian deposit);
rig
(c) connector assemblages in Breidmerkur silty till, Iceland (glacial ablation deposit); (d )
particle matrix assemblage in Immingham silty clay, England (estuarine deposit); (e) regular
aggregation assemblage in Holon silty clay, Israel (consisting of elementary particle arrange-
ments interacting with each other and silt) (freshwater alluvial deposit); ( ƒ ) interweaving
py
bunch assemblage in Hurlford organic silty clay, Scotland (freshwater lacustrine deposit);
and (g) irregular aggregation assemblage in Sundland silty clay, Norway (marine deposit).
Glass balls allowed to fall freely form an anisotropic to accumulate near the base. Makse and co-workers’
Co
assembly, with the balls tending to arrange themselves (1997) experiments produced the interesting additional
in chains (Kallstenius and Bergau, 1961). The number result that if the large grains in a binary mixture have
of balls per unit area in contact with a vertical plane a greater angle of repose than the small grains, then
can be different from the number in contact with a the mixture stratifies into alternating layers of small
horizontal plane. The same behavior is observed for and large grains. If the small grains have a larger angle
sand pluviated through air and water. of repose than the large grains, then segregation with-
Spontaneous segregation and stratification has been out stratification results. This type of behavior is rel-
observed when granular mixtures of particles of two evant to such geoengineering problems as the stability
different predominant sizes are dumped into a pile of dumped mine waste piles, geological formations
(Makse et al., 1997; Fineberg, 1997). When a mixture susceptible to static liquefaction, and the processing
of sizes is poured into a pile, the larger particles tend and transport of granular materials.
l
ria
ate
dM
hte
rig
py
Particle Packings in Granular Soils spheres shown by the values in Table 5.1, that is, po-
Particle sizes in soil vary, and as a result, smaller par- rosity in the range of 26 to 48 percent and void ratio
ticles can occupy pore spaces between larger particles. in the range of 0.35 to 0.91. This is illustrated by the
This results in a tendency toward higher densities and data in Table 5.2. The lower values of porosity and
lower void ratios than for uniform spheres. On the density and higher unit weight for silty sand and gravel
other hand, irregular particle shapes produce a ten- can be attributed to silt filling the large voids between
dency toward lower densities and higher porosities and the gravel particles.
void ratios. The net result is that the range of porosities Many studies have shown that a given cohesionless
and void ratios in real soils with single-grain fabrics soil can have different fabrics at the same void ratio
may not be much different from that for uniform or relative density. Characterization of this fabric can
l
ria
ate
dM
hte
Figure 5.6 Overall microfabric in Tucson silty clay, United States (freshwater alluvial de-
posit) (Collins and McGown, 1974).
rig
be done in terms of grain shape factors, grain orien- self with respect to the sample and to the field deposit
tations, and interparticle contact orientations (Lafeber, is also an essential part of the fabric description.
1966; Oda, 1972a; Mahmood and Mitchell, 1974; Orientations of long axes for a large number of
py
Mitchell et al., 1976). More recently, application of grains can be expressed by a histogram or rose dia-
image analysis techniques (Section 5.8) has led to bet- gram. A frequency histogram for a sand having a mean
ter understanding and quantification of fabric features. axial ratio equal to 1.65 and placed by tapping the side
The orientation of grains in a sand deposit can be of a vertical, cylindrical mold is shown in Fig. 5.9. The
Co
described in terms of the inclination of the particle orientation of each grain was assigned to one of the
axes to a set of reference axes. For example, the ori- 15 intervals between 0 and 180. The V-section refers
entation of the particle shown in Fig. 5.8 can be de- to a thin section from a vertical plane (oriented parallel
scribed by the angles and . In most studies, to the cylinder axis). The H-section refers to orienta-
however, a thin section is studied to give the orienta- tions in the horizontal plane.
tions of apparent long axes. The long axes of particles Orientations of long axes in the vertical plane for
are referred to a single horizontal reference axis by an two samples of well-graded crushed basalt [mean
angle .1 The spatial orientation of the thin section it- (length)/(width) ⫽ 1.64] are shown by the rose dia-
grams in Figs. 5.10 and 5.11. In this study, the orien-
1
This method underestimates the value of L / W for elongate particles tations of at least 400 grains were measured for each
having their long axis out of the plane of the thin section. sample, and the orientation of each was assigned to
l
ria
ate
dM
hte
Figure 5.7 Ideal packings of uniform spheres: (a) simple cubic, (b) cubic tetrahedral, (c)
tetragonal sphenoidal, (d ) pyramidal, and (e) tetrahedral.
rig
one of the eighteen 10 intervals between 10 and 180. paction, however, resulted in a more nearly random
A completely random distribution would yield the fabric (Fig. 5.11).
dashed circles shown in the figures. There is a strong Interparticle contact orientations and their distribu-
preferred orientation in the horizontal direction in the tion influence deformation and strength properties and
sample prepared by pouring (Fig. 5.10). Dynamic com- anisotropy. These orientations can be described in
Table 5.2 Maximum and Minimum Void Ratios, Porosities, and Unit Weights for Several Granular Soils
l
Fine to coarse sand 0.95 0.20 49 17 13.4 21.7
ria
Micaceous sand 1.2 0.40 55 29 11.9 18.9
Silty sand and gravel 0.85 0.14 46 12 14.0 22.9
Modified from Lambe and Whitman (1969).
ate
dM
hte
rig
Figure 5.9 Frequency histograms of long particle axis orientations in two planes for a
uniform fine sand. Reprinted from Oda (1972a), with permission of The Japanese Society
of SMFE.
l
sity is 1600 kg / m3 and the relative density is 62 percent. els (e.g., Tobita, 1989; Muhunthan et al., 1996; Yimsiri
ria
and Soga, 2000; Wan and Guo, 2001; Li and Dafalias,
2002).
The mean value of the particle coordination number
and its standard deviation are additional important fab-
ric features in granular soils. The coordination number
ate
is the number of adjacent particles in contact with any
given particle, and it is dependent on particle size,
shape, size distribution, and void ratio. Relationships
between the different orientation and packing param-
dM eters and mechanical properties of cohesionless soils
are given in Chapter 8.
particle contact points rarely occur in the analyzed with the direction of the principal stresses. Utilizing
plane, measurement of contact normals can be prone this technique, the analysis of a photoelastically sen-
to detection errors. sitive particle assembly under different boundary load-
The orientation of Ni is defined by angles and ing conditions gives information about the internal
as shown in Fig. 5.12. A procedure for determination force structure through particle contacts. Averaging the
py
of the angular distributions of normals E(, ) is contact forces over a number of particles in a region
given by Oda (1972a). For a fabric with axial sym- of interest gives the average effective stress. The down-
metry around the vertical axis, the function E(, ) side of this technique is that actual soil particles cannot
is independent of , so the distribution of E() as a be used. However, the force information obtained from
Co
function of can be used to characterize the distri- a transparent particulate assembly is useful for under-
bution of interparticle contact normals. Contact normal standing how actual soil particle systems are likely to
distributions for four sands deposited in water and behave.
compacted by tapping on the sides of their containers Light propagates in a vacuum or in air at a speed C
are shown in Fig. 5.13. The horizontal dashed lines of 3 ⫻ 108 m/s. In other transparent materials, the
represent the distributions for an isotropic fabric. In speed V is lower and the ratio C/V is called the re-
each case there is a greater proportion of contact plane fractive index. In photoelastic materials, the change in
normals in the near vertical direction; that is, there is refractive index in the i direction (ni) is proportional to
a preferred orientation of contact planes near the hor- the change in normal stress i in the same direction;
izontal.
ni ⫽ Kso
i, where Kso is the stress-optical material
l
ria
ate
Figure 5.12 Characterization of interparticle contact orientation.
dM
constant. Hence, the velocity becomes direction depen-
dent when the material is stressed in an anisotropic
manner.
Using a polarizer, the incoming light is polarized
hte
l
ticles become opaque due to refraction and reflection
ria
where I0 is a constant and is the wavelength of the of light at the particle surfaces, which are often opti-
light. The light intensity becomes zero when ⫽ N cally damaged. This adds difficulty in examining the
(N ⫽ fringe order ⫽ 1, 2, ...), and hence the magnitude contact force distributions. However, if the pores are
of principal stress difference at a given point can be filled with a fluid that has the same refractive index as
evaluated from Eq. (5.3). Photoelastic images of a cir- the photoelastic material, the assembly becomes more
ate
cular disk squeezed between two contacts are shown transparent. Figure 5.16 shows the force distribution in
in Fig. 5.14 (Howell et al., 1999). crushed glass particles when a cone penetrometer is
The forces applied to particles are not equal. Instead, pushed into the material (Allersma, 1999). Again, de-
the spatial distribution of forces varies significantly due velopment of a strong force network is evident.
to random positions of the particles. Figure 5.15 shows
dM
images in an assemblage of pentagonal-shaped disks
under (a) geostatic stresses by gravity and (b) both 5.5 MULTIGRAIN FABRICS
gravity loading and point loading at the center of the
model (Geng et al., 2001). A chainlike force distribu- In Section 5.2, it was emphasized that single-grain fab-
tion, indicated by large light intensity paths, exists rics are rare in soils containing clay-size particles. This
even under geostatic stress conditions. Strong force is often true also for silts (particle sizes in the range
chains can develop in an assembly of pentagonal- of 2 to 74 m). For example, experiments have shown
hte
shaped polymer particles as shearing progresses (Geng that silt-size quartz particles sedimented in water can
2
rig
1.5
py
1
Co
0.5
0
-1 -0.5 0 0.5 1
(a) (b)
Figure 5.14 Photoelastic image of a circular disk squeezed between two contacts: (a) the-
oretically expected image and (b) actual image (from Howell et al., 1999).
l
ria
ate
Figure 5.16 Cone penetration test in photoelastic particles
(from Allersma, 1999).
dM
hte
upper limit of about 1.0 for single-grain assemblages sorb on nonclay particle surfaces, and clay surfaces are
of bulky particles. However, silt-size particles form often chemically reactive. In addition, clay particle
multigrain arrangements during slow sedimentation, groups in many soils may be remnants of a preexisting
because they are sufficiently small that their arrange- rock from which the soil was derived.
Co
l
out disturbance of the original fabric is difficult, and
ria
5.7 SAMPLE ACQUISITION AND in most cases there is no way to determine how much
PREPARATION FOR FABRIC ANALYSIS disturbance there may have been.
Pore Fluid Removal Air drying without significant
Obtaining representative soil samples with minimal disruption of the natural fabric may be possible for
disturbance is essential if reliable measurements of en- soils that do not undergo much shrinkage. For soft
ate
gineering properties are to be made. The same consid- samples at high water content, oven drying may cause
erations apply in the selection and preparation of less fabric change than air drying, evidently because
samples for the study of fabric. Accordingly, the sam- the longer time required for air drying allows for
pling and preparation phases of fabric study are criti- greater particle rearrangement (Tovey and Wong,
cal, and special methods are many times needed.
dM 1973). On the other hand, the stresses induced during
Proven methods for reliable determination of fabric oven drying may result in some particle breakage.
can also be used for evaluation of the effects of dif- Water removal by drying at the critical point has also
ferent sampling procedures used in engineering prac- been used. If the temperature and pressure of the sam-
tice, although there does not appear to be much record ple are raised above the critical values, which for water
of this having been done. are 374C and 22.5 MPa, respectively, the liquid and
Both direct and indirect methods are used to study vapor phases are indistinguishable. The pore water can
the fabric and fabric features of soils, as listed in Table then be distilled off without the presence of air–water
hte
5.3. An illustrative schematic diagram prepared by interfaces that can lead to shrinkage. The high tem-
R. N. Yong that summarizes methods for analysis of perature and pressure may change the clay particles,
soil composition and fabric using various parts of the however. To avoid this, replacement by carbon dioxide
electromagnetic spectrum is shown in Fig. 5.18. In in- has been used. The critical temperature and pressure
terpreting the results from any of these methods, judg- of carbon dioxide are 31.1C and 7.19 MPa, respec-
ment is required to be sure that the conclusions pertain tively. The procedure requires prior impregnation of
rig
to properties and behavior of interest. For example, the sample with acetone, which may cause swelling in
discontinuities, fractures, and anisotropy on a macro- partly saturated and expansive soils (Tovey and Wong,
scale can override the influences of microfabric details. 1973). Both critical point and freeze-drying cause less
Of the methods listed in Table 5.3, optical and elec- sample disturbance and shrinkage than do air or oven
tron microscopy, X-ray diffraction, and pore size dis- drying, but they are more difficult and time consuming.
py
tribution offer the advantage of providing direct Freeze-drying can be used for removal of water.
(usually) unambiguous information about specific fab- Sublimation of the ice in a soil that has been rapidly
ric features, provided the samples are representative frozen avoids the problem of air–water interfaces and
and the sample preparation procedures have not de- shrinkage that accompany water removal by drying.
Co
stroyed the original fabric. On the other hand, these Sample size must be small, usually thinner than about
techniques are limited to small samples, and they are 3 mm, if disruption due to nonuniform freezing is to
destructive of the samples studied. The other tech- be avoided. Quick freezing is best done in a liquid that
niques are nondestructive, at least in principle, and can has been cooled to its melting point in liquid nitrogen,
be used for the study of soil fabric in situ and for the such as isopentane at ⫺160C or Freon 22 at ⫺145C.
study of changes in fabric that accompany compres- This avoids gaseous bubbling caused by direct immer-
sion, shear, and fluid flow. However, with most of these sion in liquid nitrogen at ⫺196C (Delage et al., 1982).
methods interpretation is seldom straightforward or un- The freezing temperature should be less than ⫺130C
ambiguous. The use of several methods of fabric anal- to avoid formation of crystalline ice. Sublimation of
l
ria
ate
dM
hte
rig
py
Co
dM
Replica Transmission
hte
Electron Micrograph
or Diffraction Pattern
rig
py
Scanning Electron
Micrograph
Figure 5.18 Methods for examining mineralogy, fabric, and structure of soils using parts of
the electromagnetic spectrum (prepared by R. N. Yong, McGill University Soil Mechanics
Co
Laboratory).
125
l
scale or larger, large pores, shear
ria
zones
Useful upper limit of magnification
about 300⫻
Electron microscope Direct observation of particles Resolution to about 100 Å; large
or fracture surfaces through depth of field with SEM; direct
ate
soil sample (SEM) observation of particles; particle
observation of surface groups and pore space; details of
replicas (TEM) microfabric; environmental SEM
can be used to observe
dM specimens containing water
and gas
X-ray diffraction Groups of parallel clay plates Orientation in zones several square
produce stronger diffraction millimeters in area and several
than randomly oriented micrometers thick; best in single
plates mineral clays
Pore size distribution (1) Forced intrusion of a (1) Pores in range from ⫺0.01 to
nonwetting fluid (usually ⬎10 m
hte
mercury)
(2) Capillary condensation (2) 0.1 m maximum
Wave propagation Particle arrangement, density, Anisotropy; measures fabric
and stress influences wave averaged over a volume equal to
velocity sample size
Dielectric Variation of dielectric Assessment of anisotropy,
rig
l
tigations, the effects of freeze-drying on the fabric are 5.19.
ria
small. Additional considerations in sample preparation The disturbed zone at the surface of Carbowax-
by freeze-drying are given by Tovey and Wong (1973) treated samples extends to a maximum depth of about
and Gillott (1976). 1 m in kaolinite (Barden and Sides, 1971). As thin
Pore Fluid Replacement If thin sections are re- sections used for polarizing microscope study are of
quired, as for optical microscopy or when drying the order of 30 m thick, this disturbed zone is of little
ate
shrinkage must be minimized, but the presence of a consequence. It is also insignificant for X-ray diffrac-
material in pore spaces is not objectionable, replace- tion studies.
ment of the pore water may be necessary. Various res- Fracture surfaces in dried specimens are sometimes
ins and plastics have been used for this purpose. taken as representative of the undisturbed fabric. Ad-
High-molecular-weight ethylene glycol such as Car-
dM ditional preparation, such as gentle blowing of the sur-
bowax 6000 is miscible with water in all proportions face or peeling is needed following fracture because
and has been used for many studies. Carbowax 6000 (1) there may be loose particles on the surface, and (2)
melts at 55C but is solid at lower temperatures. a fracture surface may be more representative of a
Impregnated samples are prepared by immersing an plane of weakness than of the material as a whole. An
undisturbed cube sample, 10 to 20 mm on a side, in alternative approach to avoid these problems is to frac-
melted Carbowax at 60 to 65C. The top surface of the ture a frozen wet specimen as described by Delage et
specimen should be left exposed to vapor for the first al. (1982).
hte
day of immersion to allow escape of trapped gases and The method of sample preparation should be se-
prevent specimen rupture. The wax should be changed lected after consideration of scale of fabric features of
after 2 or 3 days to ensure water-free wax in the sample interest, method of observation to be used, and the soil
pores. Replacement of all water by the Carbowax is type and state as regards water content, strength, dis-
usually complete in a few days. After removal from turbance, and so forth. With these factors in mind, the
the liquid wax and cooling, the sample is ready for probable effects of the preparation methods on the fab-
rig
soluble liquids cannot be used at any stage of the 5.8 METHODS FOR FABRIC STUDY
grinding or section mounting process. Measurements
by X-ray diffraction have shown that Carbowax re- Once suitable samples and surfaces have been pre-
placement of water has essentially no effect on the fab- pared, direct study of different fabric features is pos-
ric of wet kaolinite (Martin, 1966). sible using one or more of several methods, as
Co
Gelatins or water-soluble resins may be used in lieu indicated in Fig. 5.18. Details of these methods are
of Carbowax, or the sample may be impregnated with discussed in this section as well as the advantages and
methanol or acetone before replacement with resins or limitations of each.
plastics. Further details on resin impregnation are
given by Smart and Tovey (1982) and Jang et al. Polarizing Microscope
(1999).
Individual particles of silt and sand can be seen using
petrographic and binocular microscopes, and the sizes,
Preparation of Surfaces for Study orientations, and distributions of particles and pore
Surfaces chosen for study should reflect the original spaces can be described systematically. Thin sections
fabric of the soil and not the preparation method. or polished surfaces can be used for two-dimensional
l
ria
ate
Figure 5.19 Effect of surface preparation on fabric seen by the scanning electron microscope
(a) before peeling and (b) after one peeling, ⫻5000 (from Tovey and Wong, 1973).
dimensional representation, Fig. 5.20 shows the pore resolution is to a few micrometers using magnifications
pattern in a section of a stony desert tableland soil from up to about 300 times. If, however, clay plates are
near Woomera, Australia, which suggests some degree aligned parallel to each other in a group, then they
rig
py
Co
Figure 5.20 Pore pattern of a section from a stony tableland soil from Woomera, Australia.
Pores in white, clay matrix in gray, and silt sand grains in black (from Lafeber, 1965).
Reprinted with permission of AJSR.
L1 s1
R s1 R
L1
s2
Figure 5.21 Distribution of elongated pores (white figure) and of elongated skeleton grains
(black figure) in different directions for the pattern in Fig. 5.20. The broken circle represents
an even distribution of lengths over all directions. s1 and s2 are the major maxima of the
elongated pores, L1 is the major maximum of elongated grains, and R is the reference di-
l
rection (from Lafeber, 1965). Reprinted with permission of AJSR.
ria
behave optically as one large particle with definite op- singular orientations of particles (Lafeber, 1968), the
tical properties. semiquantitative scale proposed by Morgenstern and
ate
The optical axes and the crystallographic axes of the Tchalenko (1967c) given in Table 5.4 is useful.
clay minerals are almost coincident. For plate-shaped A vertical section taken through varved clay is
particles, the refractive indices in the a and b directions shown in Fig. 5.22. The upper half shows the winter-
are approximately equal, but different from that in the deposited clay varve and the lower half the summer-
c-axis direction. The difference in refractive indices
dM deposited silt varve. Strong preferred orientation of the
along different optical axes of a crystal determines the
optical property termed the birefringence.
If a group of parallel particles is viewed in plane Table 5.4 Orientation Scale for Clay Aggregates
polarized light looking down the c axis, a uniform field Viewed in Plane Polarized Light
is seen as the group is rotated around the c axis. If the
same particle group is viewed with the c axis normal Birefringence Ratio Particle Parallelism
to the direction of the light, no light is transmitted
hte
when the basal planes are parallel to the direction of 1.0 Random
polarization, and a maximum is transmitted when they 1.0–0.9 Slight
are at 45 to it. Thus there are four positions of ex- 0.9–0.5 Medium
tinction and illumination when the sample is viewed 0.5–0.1 Strong
using light passed through crossed nicols and the mi- 0 Perfect
croscope stage is rotated through 360. For rod-shaped
rig
l
too large to be appreciated using an electron micro-
ria
scope, but important to understanding soil behavior.
Some of these features include distributions of silt and
sand grains, silt and sand particle coatings, homoge-
neity of fabric and texture, discontinuities of various
types, and shear planes (e.g., Mitchell, 1956; Morgen-
ate
stern and Tchalenko, 1967b, 1967c; McKyes and
Yong, 1971; Oda and Kazama, 1998). A thin section
from a shear zone through a soft silty clay at the site
of a foundation failure under an embankment at Fid-
dler’s Ferry on the floodplain of the Mersey River,
dM
England, is shown in Fig. 5.23a. Details of the shear
zone deduced from the photomicrograph are shown in Figure 5.23b Details of Fiddler’s Ferry shear zone (Mor-
Fig. 5.23b. genstern and Tchalenko, 1967c). F is a fragment of ambient
material; the hatched areas indicate the shear matrix where
the birefringence ratio ⫽ 0.45; and the direction of hatch-
Electron Microscope ing is the average particle orientation over the stippled areas
The electron microscope can reveal clay particles and where ⫽ 1.00.
hte
l
technique and the freeze–fracture technique appear the tions of uninterrupted sediment accumulation and have
ria
best of the available methods for obtaining represen- quite high porosities (60 to 90 percent). Sediments of
tative sample surfaces. this type are very compressible and weak.
That careful techniques are successful in preserving Progressive collapse of microfabric of a sensitive
delicate fabrics is evidenced by Fig. 5.24, which shows Champlain clay with increasing vertical loading is
the microstructures of six artificial clay sediments (Os- shown in Fig. 5.26 (Delage and Lefebvre, 1984). The
ate
ipov and Sokolov, 1978). These samples were obtained preconsolidation pressure of the clay was 54 kPa. The
by gradual sedimentation of clay particles ⬍1 m in SEM photos were taken along the vertical plane and
the distribution of macropores at each loading stage
was derived from the photos as shown in the figure.
dM Aggregate structure is apparent at the intact stage be-
low the preconsolidation pressure. At a loading of 124
hte
rig
py
Co
l
ria
10 μm intact 10 μm 124 kPa 10 μm 421kPa 10 μm 1452 kPa
ate
Voids due to
Pores Solid particles pulling out of
particles
Figure 5.26 SEM photographs of a sensitive Champlain clay under consolidation at (a)
intact state, (b) 124 kPa, (c) 421 kPa, and (d ) 1452 kPa. The preconsolidation pressure of
dM
the clay is 54 kPa (from Delage and Lefebvre, 1984).
kPa, the collapse of macropores in the horizontal di- which the SEM electromagnetic lens must exist. This
rection is observed. Aggregates are also aligning in the pressure differentiation is achieved by a special device
horizontal direction. As the loading increases (421 and called a pressure-limiting aperture. Examination of
hte
1452 kPa), aggregates become less apparent by the samples can be done under a range of gaseous envi-
complete collapse of macropores and the particles are ronments (H2O, CO2, N2, etc.), relative humidities (0
aligning in the horizontal direction. Although the field to 100 percent), pressures (up to 6.7 kPa), and tem-
of view at high magnification is limited, mosaics of peratures (⫺180 to 1500C). ESEM images are taken
photomicrographs may be prepared to show larger fab- using an electrical current detector that collects and
ric features. Such a composite is shown in Fig. 5.6.
rig
niques of electron microscopy used to examine the trons. The cascading amplification of the signal from
structures of soils can be found in Smart and Tovey the original secondary electrons enables the secondary
(1981, 1982). electron detector to create an image. The positive ions
are attracted to the negatively charged sample surface
Co
Environmental SEM and suppress the charging artefacts. This charge sup-
Conventional SEM samples have to be dry, vacuum pression allows imaging of nonconductive samples.
compatible, and electrically conductive. To examine A significant feature of ESEM is its ability to ob-
liquid and hydrated samples, the pressure has to be at serve liquids inside the samples. The rate of sublima-
least 612 Pa, the minimum vapor pressure required to tion and condensation of water can be controlled by
maintain liquid water at 0C. An environmental scan- manipulating the pressure and temperature. Figure 5.27
ning electron microscope (ESEM) allows wet, natural, is an ESEM image of a sample containing illite clays
and nonconductive samples to be examined by having (left side) and quartz grains (right side). Water droplets
the specimen chamber at higher pressure separated were placed on the sample by condensation of distilled
from the high-vacuum electron optical regions in water present as a gaseous phase in the testing cham-
Image Analysis
Image analyzers can be used with both optical and
electron microscopes for quantification of fabric fea-
tures. Digital imaging cameras can resolve reflected or
transmitted light from the sample into pixels. The
amount of light per pixel is then converted into an
analog signal. After the entire image is acquired, the
l
analog signal for each pixel is converted to digital form
ria
for analysis, manipulation, and storage. Image analysis
offers greatly increased potential for quantitative de-
Figure 5.27 ESEM image of illite clay (left side) and quartz scription of different fabric elements. Details of the
grains (right side). Water droplets placed on the samples method are beyond the scope of this book. Examples
of image analysis of soil specimens are given by Frost
ate
show that the quartz surface is hydrophilic and the illite sur-
face is hydrophobic (from Buckman et al., 2000). and Wright (1993), Tovey and Hounslow (1995), and
Frost and McNeil (1998).
ber. The photo shows the wettability of fluids on soil X-ray Diffraction
dM
minerals. Spherical water droplets are observed on the
clay surface, indicating that this illite is hydrophobic.
As discussed in Section 3.22, crystallographic planes
in minerals refract X-rays at an intensity that depends
Quartz sand, on the other hand, is hydrophilic as low on (1) the amount of mineral in the volume of soil
domed droplets of water are formed on the surface. irradiated and (2) the proportion of the mineral grains
As pressure and temperature can be varied in the that are properly oriented. For clay minerals, parallel
specimen chamber, the ESEM allows studies of dy- orientation of plates enhances the basal reflections but
namic changes in samples such as wetting, drying, ab- decreases the intensity of reflection from lattice planes
hte
sorption, melting, corrosion, and crystallization. Figure oriented in other directions. The intensity of (001) re-
5.28 shows ESEM images of the swelling of bentonite flections provides a measure of clay particle orienta-
in a sand–bentonite mixture (Komine and Ogata, tion.
2004). Initially, the bentonite particles are attached to The relative heights of basal peaks for different sam-
the sand grains and macropores can be observed. As ples of the same material give a measure of particle
rig
py
Co
Figure 5.28 ESEM images showing swelling process of bentonite clay in a sand–bentonite
mixture (from Komine and Ogata, 2004).
orientation differences. A fabric Index (FI) based on features as well as on texture and disturbance (Kenney
areas of diffraction peaks is defined as (Gillott, 1970): and Chan, 1972). A number of laboratories routinely
X-ray sample tubes prior to selection of samples for
FI ⫽ V/(P ⫹ V) (5.4) removal and testing for determination of deformation
and strength properties. The procedure is simple, rapid,
where V is the area of the basal peak in a section cut and inexpensive (apart from the initial cost of the
perpendicular to the orientation plane, and P is the area equipment).
of the same peak from a section cut parallel to the X-radiography is also useful for the study of defor-
plane of parallel orientation of particles. The value of mation patterns in soils. Lead shot is placed in regular
FI ranges from zero for perfect preferred orientation to patterns in samples or in blocks of soil used for model
0.5 for perfectly random orientation. A similar proce- tests. The positions of the shot are determined at var-
l
dure that retains the concept of peak areas, but does ious stages throughout a test by comparison of succes-
ria
not require their exact measurement, is given by Yo- sive radiographs. The results can be used to locate
shinaka and Kazama (1973). shear zones and compute strains and their variation
The peak ratio (PR), defined as the ratio of the (002) throughout the material.
reflection to that of the (020) reflection, can also be X-ray computed tomography (CT) allows construc-
tion of a three-dimensional density profile inside
ate
used as a measure of orientation. The PR has the ad-
vantages of being independent of the particle concen- a material by assembling X-ray radiographic two-
tration within the total soil and of minimizing the dimensional images taken at different angles. The res-
effects of mechanical and instrumentation variables olution of a CT scanner is determined by the
(Martin, 1966). The PR of kaolinite with completely dimensions of a source and a detector as well as their
random particle orientations is about 2.0. For maxi-
dM positions in relation to the test specimen. The tech-
mum parallel orientation the PR is about 200. The rea- nique has been used to examine the locations of shear
sons for choosing the (002) and (020) reflections are zones within a specimen as local dilation inside the
that (1) they are strong and (2) the corresponding 2 shear band gives low electron density (Desrues et al.,
values are not too far apart, thus ensuring that about 1996; Otani et al., 2000; Alshibi et al., 2003; Otani
the same sample volume will be irradiated for deter- and Obara, 2004). Figure 5.29 shows the locations of
mination of both peaks. shear zones in cylindrical sand specimens that were
X-ray diffraction methods had the advantage of sheared to different axial strains in triaxial compres-
hte
quantification of data in a way that was not possible sion. The specimens showed strain-softening behavior
with optical and electron microscope methods. How- and exhibited uniform bulging with no apparent single
ever, the development of image analysis techniques for or multiple shear bands. The CT images were taken at
use with the latter has largely overcome this problem. strains greater than the peak axial strain of approxi-
X-ray methods have some disadvantages, including (1) mately 2 percent. No apparent shear zones are ob-
served at an axial strain of 4.6 percent, indicating that
rig
Thus, X-ray diffraction is best suited for fabric anal- are apparent (Desrues et al., 1996; Alshibi et al., 2003):
ysis of single mineral clays in which particle orienta-
1. Cone-Shaped Shear Zone The images of the
tions over regions the size of the X-ray beam (a few
horizontal plane show black circles appearing at
millimeters) are of interest or in conjunction with other
the center, and they become smaller in diameter
Co
l
ria
ate
dM
hte
Figure 5.29 CT scans of a dense sand specimen under triaxial compression: (a) Horizontal
slice at the midheight, (b) vertical slice, and (c) 3D image (from Alshibi et al., 2003).
rig
sliced images (Fig. 5.29b). Close examination of determinations and from image analysis of thin sec-
these lines reveal that there are several pairs of tions and SEM pictures.
conjugate shear bands at two different inclined
angles as shown in Fig. 5.29c. Volumetric Pore Size Distribution Determinations
py
Further details of shear bands are given in Chapter Volumetric pore size distributions can be determined
11. Other noninvasive techniques reported to observe using forced intrusion of a nonwetting fluid, a capillary
particle packing arrangements include nuclear mag- condensation method based on interpretation of ad-
netic resonance imaging (Ehrichs et al., 1995; Ng and sorption and desorption isotherms, and by removal of
Co
Wang, 2001) and laser-aided tomography (Matsushima water by suction or air pressure.
et al., 2002). The maximum pore size that can be measured using
the capillary condensation method is about 0.1 m.
With the possible exception of intraaggregate pores
5.9 PORE SIZE DISTRIBUTION ANALYSIS most soil pores are larger, so this method is of limited
usefulness. The mercury intrusion method, however, is
The shape and distribution of voids are one of the three useful for measurement of pore sizes from about 0.01
most important measures of fabric, along with contact m to several tens of micrometers. The basis of the
distributions and particle orientations. Pore information method is that a nonwetting fluid (fluid-to-solid contact
can be obtained by volumetric pore size distribution angle ⬎90) will not enter pores without application of
pressure. For pores of cylindrical shape, the capillary 4. The apparatus may not have the capacity to pen-
pressure equation applies, and etrate the smallest pores in a sample.
l
and sample preparation method result in changes in
ria
using successively higher pressures. The total volume pore size distributions.
of mercury intruded at any pressure gives the total vol- Pore size distributions may be estimated for sands,
ume of pores with an equivalent diameter larger than which are too coarse for mercury intrusion, by deter-
that corresponding to that pressure. The surface tension mination of the pore water volume that is drained ei-
of mercury is 4.84 ⫻ 10⫺4 N/mm at 25C. The contact ther by application of suction to the sample or by
ate
angle is about 140; measurements by Diamond application of air pressure to the pore water. Equation
(1970) gave 139 for montmorillonite and 147 for (5.5) applies. The surface tension of water, 7.5 ⫻ 10⫺5
other clay mineral types. N/mm at ordinary temperature, and a contact angle
Limitations of the mercury intrusion method are: of 0 should be used.
1. Pores must be dry initially. Freeze-dried samples
dM
are often used to minimize the effect of volume Image Analysis
change upon drying. The spatial distribution of local voids inside a soil
2. Isolated pores are not measured. specimen can be obtained by analyzing the images ob-
3. Pores accessible only through smaller pores will tained from thin sections. Generally, two image anal-
not be measured until the smaller pores are pen- ysis methods are available: (1) method of polygons and
etrated. (2) mean free path. In the first method the centroids of
hte
rig
py
Co
Figure 5.30 Pore size distributions in crushed basalt as affected by compaction method.
l
path between particles by use of a scanning line that Vp ⫽ 兹M/ (5.7)
ria
passes through both particles and voids as shown in
Fig. 5.31b. The spacing and orientation of the line are where is the mass density.
varied, and a representative void is then produced by The constrained modulus M is related to the more
summing over the void lines found on a number of familiar Young’s modulus according to
scanned lines in each direction (Kuo et al., 1998). Us-
ate
ing this method, Masad and Muhunthan (2000) found 1⫺
that larger local voids exist in the horizontal direction M⫽ E (5.8)
(1 ⫹ )(1 ⫺ 2)
than the vertical for a pluviated specimen.
in which is Poisson’s ratio. Young’s modulus and
dM the shear modulus are related to each other by
5.10 INDIRECT METHODS FOR FABRIC
CHARACTERIZATION E ⫽ 2(1 ⫹ )G (5.9)
All physical properties of a soil depend in part on the The moduli depend on the applied effective stresses,
fabric; therefore, the measurement of a property pro- stress history, void ratio, and plasticity index. For co-
vides indirect measure of the fabric. Some of the mea- hesionless soils the modulus varies approximately as
surements that are particularly useful are listed in Table the square root of the effective confining pressure. For
hte
5.3 and are discussed briefly in this section. cohesive soils the modulus varies as the effective con-
fining pressure to a power between 0.5 and 1.0. The
Elastic Wave Propagation small strain shear modulus of soil depends on contact
The propagation velocities of compression and shear stiffness and fabric state. Therefore, the change in
waves through a soil depend on the density, confining shear wave velocity with confining pressure provides
rig
py
Co
Figure 5.31 Image analysis methods to determine void fabric: (a) polygon method (after
Bhatia and Soliman, 1990) and (b) mean free path method (Kuo et al., 1998).
Vp & Vs (m/s)
b
ferent modulus values. This difference will be reflected
by differences in shear and compression wave veloci- 1000
vb = 0.35
ties. These velocities can be measured, and this pro- vb = 0.4
l
vides a means for assessing fabric. The shear wave
ria
velocity is the more useful of the two because shear
waves are only transmitted through the solid grain 500
vb = 0.25
structure of the soil mass, that is shear waves cannot vb = 0.5
be transmitted through water. Anisotropic soil structure
ate
and stress states can be detected on the basis of dif-
Vs = 212 (ms)
ferent shear wave velocities in different directions. Fur-
0
ther details of the relationships between small strain
moduli and compositional and environmental factors 0 0.2 0.4 0.6 0.8 1
are given in Chapter 11. B-value
If the material is dry, the bulk modulus of the skel-
dM Figure 5.32 Variation in P- and S-wave velocities with B
eton can be derived using both shear wave and com-
value in loose Toyoura sand under an isotropic compression
pression wave velocity measurements. If the material stress of 98 kPa (after Tsukamoto et al., 2002).
includes water, the P-wave velocity depends on the
elastic properties of soil solids and water, saturation,
and porosity. For fully saturated conditions, solutions
are available for two-phase media (Biot, 1956a, 1956b; whereas that at B ⫽ 0.05 (Sw ⫽ 90 percent) is only
Stoll, 1989; Mavko et al., 1998; Santamarina et al., 500 m/s. The S-wave velocity, on the other hand, is
hte
2001). The solutions show that there are two P-waves independent of the water saturation. Kokusho (2000)
and one S-wave. The fast P-wave and S-wave are the derives the following relationship that relates the fast
standard waves and the velocities have weak depend- P-wave velocity to B value:
ency on frequency. The slow P-wave (or Biot wave),
which is associated with the diffusional process of wa-
冪43 ⫹ 3(1 2(1 ⫺)
rig
gates with a velocity that is 10 to 15 percent faster There is a dramatic decrease in P-wave velocity with
than the velocity through water. This is because the even a very small decrease in B value from fully sat-
stiffness of the soil skeleton contributes to increasing urated conditions.
P-wave velocity. In very loose saturated soil, the P-
Co
wave velocity is essentially controlled by the bulk Dielectric Dispersion and Electrical Conductivity
modulus of water and has a value of about 1500 m/s. The flow of electricity through a soil is a composite of
When air is introduced, P-wave velocity decreases. (1) flow through the soil particles alone, which is
Even with a small amount of air, the reduction is dra- small, because the solid phase is a poor conductor, (2)
matic due to a large decrease in bulk modulus of the flow through the pore fluid alone, and (3) flow through
fluid–air mixture. The effect of B-value (or water sat- both solid and pore fluid. The total electrical flow also
uration ratio Sw) on P- and S-wave velocities of Toy- depends on the porosity, tortuosity of flow paths, and
oura sand specimen (Dr ⫽ 30 percent) is shown in Fig. conditions at the interfaces between the solid and liq-
5.32 (Tsukamoto et al., 2002). The fast P-wave veloc- uid phases. These factors are, in turn, dependent on the
ity at B ⫽ 0.95 (Sw ⫽ 100 percent) is 1700 m/s, particle arrangements and the density. Thus, a simple
l
eration of how the measurement method may influence 5.33.
ria
what is being measured. At the same time, however, The electrical response characteristics in the low-
measurement of the frequency dependence of electrical frequency range depend on particle size and size dis-
properties can be useful for evaluation of fabric and as tribution, water content, direction of current flow
an index for engineering properties. relative to the direction of preferred particle orienta-
The capacitance C and the resistance R can be mea- tion, type and concentration of electrolyte in the pore
ate
sured relatively easily. If electrical flow is in one di- water, particle surface characteristics, and sample dis-
mension only, then the electrical conductivity is turbance. Relationships between dielectric properties
given by and compositional and state parameters such as poros-
ity, particle shape, fabric anisotropy, and specific sur-
⫽ L/(RA)
dM (5.11) face area are given by Arulanandan (1991). The theory
is based on Maxwell’s (1881) relationship between po-
where L is the sample length and A is the cross- rosity and the dielectric properties of a mixture of so-
sectional area. lution and spherical particles, and its extension to
The capacitance can be converted to the relative di- ellipsoidal particles that are all oriented in one direc-
electric constant D (see Chapter 6) using tion by Fricke (1953). Extensive discussion of electro-
magnetic properties of soils is given in Santamarina et
al. (2001).
D ⫽ CL/(A0) (5.12)
hte
long axis orientation, porosity, and degree of satura- ture caused by mechanically and environmentally
tion. If a soil has an anisotropic fabric, then the for- induced changes in state of the soil.
mation factor is different in different directions.
Mechanical Properties
Thermal Conductivity
The mechanical properties of soil, including stress–
Heat transfer through soils is through soil grains, wa- deformation behavior, strength, compressibility, and
ter, and pore air. As the thermal conductivity of soil permeability, depend on fabric in ways that are rea-
minerals is about 2.9 W/(m C), and the values for sonably well understood, as considered in Chapter 8.
water and air are 0.6 and 0.026 W/(m C), respec- Therefore, information about fabric can be deduced
tively, heat transfer is mainly through the soil particles. from measurements of these properties and known in-
Accordingly, the lower the void ratio, the greater the
l
terrelationships between properties and fabric.
number and area of interparticle contacts and the
ria
higher the degree of saturation, the higher is the ther-
mal conductivity. The thermal conductivity of a typical
soil is likely to be in the range of 0.5 to 3.0 W/(m 5.11 CONCLUDING COMMENT
C). This property is considered in more detail in Sec-
Fabric analyses are useful in research to show how
ate
tion 9.6.
Thermal conductivity can be determined using a rel- mechanical properties are dependent on particle asso-
atively simple transient heat flow method in which a ciations and arrangements. Fabric information can be
line heat source, called a thermal needle, is inserted used to deduce details of the depositional and postdep-
into the soil. The needle contains both a heating wire ositional history of a deposit. The effects of different
and a temperature sensor. When heat is introduced into
dM sampling methods can be assessed through the study
the needle at a constant rate, the temperatures T2 and of fabric changes. Insights can be gained into the me-
T1 at times t2 and t1 are related to the thermal conduc- chanics of strength mobilization, the nature of peak
tivity k according to and residual strengths, and the stress–strain behavior
of soils from fabric studies.
4 ln(t2) ⫺ ln(t1) The indirect methods for fabric study are often use-
k⫽ (5.13) ful for determination of properties, homogeneity, and
Q T2 ⫺ T1 anisotropy in situ. They may be of value also for as-
hte
example, the ratios of thermal conductivity in the states, each having its own unique set of geotechnical
horizontal direction kh to that in the vertical direction properties.
kv for three clays with preferred particle orientations in
the horizontal direction were in the range of 1.05 to
1.70, depending on the clay type, consolidation pres-
py
sure, and sample disturbance (Penner, 1963b). For the QUESTIONS AND PROBLEMS
probe in the vertical position in a cross anisotropic
fabric, the value of k determined from Eq. (5.13) is kh. 1. Two samples of the same remolded clay have been
For the probe in the horizontal direction, a value of ki consolidated from the liquid limit to the same water
Co
is measured that is related to kv and kh according to content. One was consolidated under an isotropic
(Carlslaw and Jaeger, 1957) set of stresses and the other under anisotropic
stresses. What differences in fabric would you an-
k2i ticipate? Why?
kv ⫽ (5.14)
kh 2. Two slurries of the same clay, one with flocculated
clay particles and the other with deflocculated par-
Thermal probe measurements can also be used to ticles, have been consolidated under an effective
detect differences in density at different locations in stress of 100 kPa. Which will have the higher (a)
the same material (Bellotti et al., 1991) and for eval- void ratio, (b) sensitivity, (c) strength? Explain your
uation of changes in density, water content, and struc- answer.
l
Exhibit 5.1 Soil fabrics.
ria
3. A series of shrinkage tests was done on a fine- specific soil types and states for which each of these
grained soil mass, and it was found that the shrink- methods might be useful for gaining insights and
ate
age was a maximum in the Z direction and was a understanding of the macro- and microfabrics and
minimum in all directions lying in a plane perpen- their influences on volume change, strength, and
dicular to the Z direction. permeability properties.
a. Was the soil mass likely to have been isotropi- 7. To obtain an essentially undisturbed sample of co-
cally consolidated or anisotropically consoli- hesionless soil from the field that preserves the in
dated?
dM
b. If anisotropically consolidated, what was the ma-
situ fabric is usually impossible without resorting
to expensive and time-consuming procedures such
jor principal stress direction? as ground freezing or injection followed by setting
c. Would you expect the soil to be isotropic with of a grout or resin. Suppose that you do not have
respect to hydraulic conductivity? Why? If ani- the time or budget that will allow this, but wish to
sotropic, in which direction would the hydraulic reconstitute disturbed specimens of the soil in the
conductivity be greatest? Why? laboratory by forming them in such a way that they
hte
4. Could X-ray diffraction alone be used to distinguish will have fabrics that reasonably duplicate the un-
among the fabrics shown in Exhibit 5.1? Explain disturbed condition in the field. Suggest practical
your answer. Pertinent geometrical parameters of laboratory procedures that might be used, starting
typical X-ray diffractometers are: distance from X- with dry and disturbed soil of the type indicated, to
ray source to sample ⫽ 17 cm, divergence of X-ray reproduce specimens that could then be used for
rig
beam ⫽ 1, angle of incidence of X-ray beam to fabric studies and measurements of mechanical
the sample surface in the range of 10 to 35. properties:
a. Beach sand
5. You are analyzing a new type of laboratory strength
test that imposes unusual boundary conditions on b. Alluvial deposit
py
the sample being tested. What methods of fabric c. Wind-blown dune sand
study would you use to examine the location, di- d. Uniform sand placed as a hydraulic fill
rection, thickness, and fabric of shear zones within e. Uniform sand placed as a hydraulic fill and then
specimens? What would these methods tell you? densified using vibratory probes
Co
6. Several methods for study and characterization of f. Sand fill placed as a pavement base and densified
soil fabric are listed in Table 5.3. Indicate some by a vibratory roller
l
Total Stress
ria
ate
7.1 INTRODUCTION phases (usually water and air) can carry normal stress
but not shear stress. Furthermore, whenever the total
The compressibility, deformation, and strength prop- head in the fluid phases within the soil mass differs
erties of a soil mass depend on the effort required to from that outside the soil mass, there will be fluid flow
distort or displace particles or groups of particles rel-
dM into or out of the soil mass until total head equality is
ative to each other. In most engineering materials, reached.
resistance to deformation is provided by internal In this chapter, the relationships between stresses in
chemical and physicochemical forces of interaction a soil mass are examined with particular reference to
that bond the atoms, molecules, and particles together. stress carried by the assemblage of soil particles and
Although such forces also play a role in the behavior stress carried by the pore fluid. Interparticle forces of
of soils, the compression and strength properties de- various types are examined, the nature of effective
pend primarily on the effects of gravity through self
hte
other engineering materials, which, for practical pur- The principle of effective stress is the keystone of
poses, do not change density when loaded or unloaded. modern soil mechanics. Development of this principle
Because of the stress dependencies of the state, a was begun by Terzaghi about 1920 and extended for
given soil can exhibit a wide range of properties. For- several years (Skempton, 1960a). Historical accounts
tunately, however, the stresses, the state, and the prop-
py
173
This fraction of the total principal stresses will be called ticle forces in a soil mass. Interparticle forces at the
the effective principal stresses . . . . A change in the neutral microscale can be separated into the following three
stress u produces practically no volume change and has categories (Santamarina, 2003):
practically no influence on the stress conditions for failure
. . . . Porous materials (such as sand, clay, and concrete) 1. Skeletal Forces Due to External Loading These
react to a change of u as if they were incompressible and forces are transmitted through particles from the
as if their internal friction were equal to zero. All the meas- forces applied externally [e.g., foundation load-
urable effects of a change of stress, such as compression,
distortion and a change of shearing resistance are exclu-
ing) (Fig. 7.1a)].
sively due to changes in the effective stresses 1, 2 and 2. Particle Level Forces These include particle
3. Hence every investigation of the stability of a saturated weight force, buoyancy force when a particle is
body of soil requires the knowledge of both the total and submerged under fluid, and hydrodynamic forces
l
the neutral stresses. or seepage forces due to pore fluid moving
ria
through the interconnected pore network (Fig.
In simplest terms, the principle of effective stress 7.1b).
asserts that (1) the effective stress controls stress– 3. Contact Level Forces These include electrical
strain, volume change, and strength, independent of the forces, capillary forces when the soil becomes
magnitude of the pore pressure, and (2) the effective unsaturated, and cementation-reactive forces (Fig.
ate
stress is given by ⫽ ⫺ u for a saturated soil.1 7.1c).
There is ample experimental evidence to show that
these statements are essentially correct for soils. The When external forces are applied, both normal and
principle is essential to describe the consolidation of a tangential forces develop at particle contacts. All par-
liquid-saturated deformable porous solid, as was done ticles do not share the forces or stresses applied at the
for the one-dimensional case by Terzaghi and further boundaries in equal manner. Each particle has different
dM
developed for the three-dimensional case by others
such as Biot (1941). It is also an essential concept for
skeletal forces depending on the position relative to the
neighboring particles in contact. The transfer of forces
the understanding of soil liquefaction behavior during through particle contacts from external stresses was
earthquakes. shown in Fig. 5.15 using a photoelastic model. Strong
The total stress can be directly measured or com- particle force chains form in the direction of major
puted using the external forces and the body force due principal stress. The evolution and distribution of in-
to weight of the soil–water mixture. A pore water pres-
hte
not mean that u ⫽ u0, as discussed in Section 7.7. The strength is given in Chapter 11.
effective stress is a deduced quantity, which in practice Particle weights act as body forces in dry soil and
is taken as ⫽ ⫺ u0. contribute to skeletal forces, observed in the photo-
elastic model shown in Fig. 5.15. When the pores are
filled with fluids, the weight of the fluids adds to the
py
7.3 FORCE DISTRIBUTIONS IN A body force of the soil–fluids mixture. However, hydro-
PARTICULATE SYSTEM static pressure results from the fluid weight, and the
uplift force due to buoyancy reduces the effective
The term intergranular stress has become synonymous weight of a fluid-filled soil. This leads to smaller skel-
Co
with effective stress. Whether or not the intergranular etal forces for submerged soil compared to dry soil.
stress i is indeed equal to ⫺ u cannot be ascertained Seepage forces that result from additional fluid pres-
without more detailed examination of all the interpar- sures applied externally produce hydrodynamic forces
on particles and alter the skeletal forces.
1
The terms and are the principal total and effective stresses.
For general stress conditions, there are six stress components (11, 7.4 INTERPARTICLE FORCES
22, 33, 12, 23, and 31), where the first three are the normal stresses
and the latter three are the shear stresses. In this case, the effective
stresses are defined as 11 ⫽ 11 ⫺ u, 22 ⫽ 22 ⫺ u, 33 ⫽ 33 ⫺ Long-range particle interactions associated with elec-
u, 12 ⫽ 12, 23
⫽ 23, and 31
⫽ 31. trical double layers and van der Waals forces are dis-
l
Forces
ria
Seepage Electrical Forces
ate
(a) (b) (c)
Figure 7.1 Interparticle forces at the particle level: (a) skeletal forces by external loading,
(b) particle level forces, and (c) contact level forces (after Santamarina, 2003).
dM
cussed in Chapter 6. These interactions control the
flocculation–deflocculation behavior of clay particles
layers of water when clay plates are pressed together
may be 0.05 to 0.1 J/m2. The corresponding pressure
in suspension, and they are important in swelling soils required to squeeze out one molecular layer of water
that contain expanding lattice clay minerals. In denser may be as much as 400 MPa (4000 atm) (van Olphen,
soil masses, other forces of interaction become impor- 1977).
tant as well since they may influence the intergranular Thus, pressure alone is not likely to be sufficient to
stresses and control the strength at interparticle con- squeeze out all the water between parallel particle sur-
hte
tacts, which in turn controls resistance to compression faces in naturally occurring clays. Heat and/or high
and strength. In a soil mass at equilibrium, there must vacuum are needed to remove all the water from a fine-
be a balance among all interparticle forces, the pres- grained soil. This does not mean, however, that all the
sure in the water, and the applied boundary stresses. water may not be squeezed from between interparticle
contacts. In the case of interacting particle corners,
edges, and faces of interacting asperities, the contact
rig
gap between parallel particle surfaces separated by dis- There is also an increase in resistance to tangential
tance d at potentials V1 and V2 is conductive, there is force at particle contacts. However, when the bond
an attractive force per unit area, or tensile strength, breaks, the shear capacity at a contact reduces to that
given by (Ingles, 1962) of the uncemented contacts.
An analysis of the strength of cemented bonds
4.4 ⫻ 10⫺6 (V1 ⫺ V2)2 should consider three cases: (i) failure in the cement,
F⫽ N/m2 (7.1) (ii) failure in the particle and (iii) failure at the ce-
d2
ment–particle interface. The following equation can be
derived (Ingles, 1962) for the tensile strength T per
where F is the tensile strength, d is in micrometers, unit area of soil cross section:
and V1 and V2 are in millivolts. This force is indepen-
冉 冊冘
l
dent of particle size and becomes significant (greater 1 n
T ⫽ Pk
ria
than 7 kN/m2 or 1 psi) for separation distances less (7.2)
1⫹e n
than 2.5 nm. Ai
Electromagnetic Attractions Electromagnetic at- 1
ate
6.12. Anandarajah and Chen (1997) proposed a method mean coordination number of a grain, e is the void
to quantify the van der Waals force between particles ratio, n is the number of grains in an ideal breakage
specifically for fine-grained soils with various geomet- plane at right angles to the direction of T, and Ai is
ric parameters such as particle length, thickness, ori- the total surface area of the ith grain.
entation, and spacing. For a random and isotropic assembly of spheres of
diameter d, Eq. (7.2) becomes
dM
Primary Valence Bonding Chemical interactions
between particles and between the particles and adja-
cent liquid phase can only develop at short range. Co- Pk
T ⫽ (7.3)
valent and ionic bonds occur at spacings less than 0.3 d (1 ⫹ e)
2
ing. The activation energy for soil deformation is high, shown that
in the range characteristic for rupture of chemical
bonds, and strength behavior appears in reasonable
(R ⫺ cos )
conformity with the adhesion theory of friction (Chap- cosh ⫽ R sin (7.5)
ter 11). Thus, interatomic bonding between particles
py
l
clear.
ria
Capillary Stresses Because water is attracted to
soil particles and because water can develop surface
tension, suction develops inside the pore fluid when a
saturated soil mass begins to dry. This suction acts like
a vacuum and will directly contribute to the effective
ate
stress or skeletal forces. The negative pore pressure is
usually considered responsible for apparent and tem-
porary cohesion in soils, whereas the other attractive
forces produce true cohesion.
dM When the soil continues to dry, air starts to invade
into the pores. The air entry pressure is related to the
pore size and can be estimate using the following equa-
tion, assuming a capillary tube as shown in Fig. 7.3a:
Figure 7.2 Contact zone failures for cemented spheres. 2aw cos
P̂c ⫽ (7.9)
rp
hte
obtain a value for P in Eq. (7.2) enabling computation (⫽ ua ⫺ uw, where ua and uw are the air and water
of the tensile strength T of a cemented soil. pressures, respectively) is larger than P̂c, then air in-
The behavior of cemented soils can depend on the vades the pore.2 Since soil has pores with various sizes,
timing of cementation development. Artificially ce- the water in the largest pores is displaced first followed
mented soils are often loaded after cementation has
py
Capillary Tube
Representing a Pore
2 rp
θ
ua
^ 2σaw cosθ
uw Pc = ρw gdc = rp
dc
l
ria
ate
(a) (b)
Figure 7.3 Capillary tube concept for air entry estimation: (a) capillary tube and (b) bundle
of capillary tubes to represent soil pores with different sizes.
dM
Fc ⫽ r 2 Pˆ c ⫽
2 r 2 aw cos
rp
(7.10)
teractions of all these forces in a soil is beyond the
present state of knowledge. Nonetheless, their exis-
tence bears directly on the magnitude of intergranular
where rp is the size of the pore into which the air has pressure and the relationship between intergranular
entered. Since the fluid acts like a membrane with neg- pressure and effective stress as defined by ⫽
ative pressure, this force contributes directly to the ⫺ u.
A simplified equation for the intergranular stress in
hte
curved air–water interface produces a pore water ten- 7.5a) is of interest. The proportion of the total wavy
sion, which, in turn, generates interparticle compres- surface area that is comprised of intergrain contact area
sive forces. The force only acts at particle contacts in is very small (Fig. 7.5c).
contrast to the funicular regime, as shown in Fig. 7.4b. The two particles in Fig. 7.5 that contact at point A
The interparticle force generally depends on the sep- are shown in Fig. 7.6, along with the forces that act in
py
aration between the two particles, the radius of the liq- a vertical direction. Complete saturation is assumed.
uid bridge, interfacial tension, and contact angle (Lian Vertical equilibrium across wavy surface x–x is con-
et al., 1993). Once the water phase becomes discontin- sidered.3 The effective area of interparticle contact is
uous, evaporation and condensation are the primary ac; its average value along the wavy surface equals the
Co
mechanisms of water transfer. Hence, the humidity of total mineral contact area along the surface divided by
the gas phase and the temperature affect the water va- the number of interparticle contacts. Define area a as
por pressure at the surface of water menisci, which in
turn influences the air pressure ua.
3
Note that only vertical forces at the contact are considered in this
simplified analysis. It is evident, however, that applied boundary nor-
mal and shear stresses each induce both normal and shear forces at
7.5 INTERGRANULAR PRESSURE interparticle contacts. These forces contribute both to the develop-
ment of soil strength and resistance to compression and to the slip-
ping and sliding of particles relative to each other. These interparticle
Several different interparticle forces were described in movements are central to compression, shear deformations, and creep
the previous section. Quantitative expression of the in- as discussed in Chapters 10, 11, and 12.
Interparticle
Forces
Soil Particles
Soil Particles
Air
l
ria
Liquid
Bridges
Pores of Radius
rp Filled with Air
ate
Suction forces act only at particle
contacts and the magnitude of the
Negative pore pressure acts all forces depends on the size of liquid
around the particles bridges.
dM (a) (b)
Figure 7.4 Microscopic water–soil interaction in unsaturated soils: (a) funicular regime and
(b) pendular regime.
hte
rig
py
2. u(a ⫺ ac), the force carried by the hydrostatic where ⫽ aw /a. Although it is clear that for a dry soil
pressure u. Because a ⬎⬎ ac and ac is very small, ⫽ 0, and for a saturated soil ⫽ 1.0, the usefulness
the force may be taken as ua. Long-range, of Eq. (7.15) has been limited in practice because of
double-layer repulsions are included in ua. uncertainties about for intermediate degrees of sat-
3. A(a ⫺ ac) ⬇ Aa, the force caused by the long- uration. Further discussion of the effective stress con-
range attractive stress A, that is, van der Waals cept for unsaturated soils is given in Section 7.12.
and electrostatic attractions. Limiting the discussion to saturated soils, two ques-
4. Aac, the force developed by the short-range at- tions arise:
tractive stress A, resulting from primary valence
(chemical) bonding and cementation. 1. How does the intergranular pressure i relate to
5. Cac, the intergranular contact reaction that is gen- the effective stress as defined for most analyses,
l
erated by hydration and Born repulsion. that is, ⫽ ⫺ u?
ria
2. How does the intergranular pressure i relate to
Vertical equilibrium of forces requires that the measured quantity, m ⫽ ⫺ u0, that is taken
as the effective stress, recalling (Section 7.2) that
a ⫹ Aa ⫹ Aac ⫽ ua ⫹ Cac (7.11) pore pressure can only be measured at points out-
side the true interparticle zone?
ate
Division of all terms by a converts the forces to
Answers to these questions require a more detailed
stresses per unit area of cross section,
consideration of the meaning of fluid pressures in soils.
ac
⫽ (C ⫺ A) ⫹u⫺A (7.12)
a
dM 7.6 WATER PRESSURES AND POTENTIALS
The term (C ⫺ A)ac /a represents the net force across Pressures in the pore fluid of a soil can be expressed
the contact divided by the total cross-sectional area in several ways, and the total pressure may involve
(soil plus water) that is served by the contact. In other several contributions. In hydraulic engineering, prob-
words, it is the intergrain force divided by the gross lems are analyzed using Bernoulli’s equation for the
area or the intergranular pressure in common soil me- total heads and head losses associated with flow be-
hte
chanics usage. Designation of this term by i gives tween two points, that is,
i ⫽ ⫹ A ⫺ u (7.13) p1 v2 p v2
Z1 ⫹ ⫹ 1 ⫽ Z2 ⫹ 2 ⫹ 2 ⫹
h1–2 (7.16)
w 2g w 2g
Equations analogous to Eqs. (7.11), (7.12), and (7.13)
rig
w 2g
aw
i ⫽ ⫹ A ⫺ ua ⫺ (u ⫺ ua) (7.14) Flow results only from differences in total head;
a w
conversely, if the total heads at two points are the
same, there can be no flow, even if Z1 ⫽ Z2 and p1 ⫽
In the absence of significant long-range attractions, p2. If there is no flow, there is no head loss and
h1–2
this equation is similar to that proposed by Bishop ⫽ 0.
(1960) for partially saturated soils The flow velocity through soils is low, and as a re-
sult v 2 /2g → 0, and in most cases it may be neglected.
i ⫽ ⫺ ua ⫹ (ua ⫺ uw) (7.15) Therefore, the relationship
l
water in soils, for example, potential, pressure, the pressure head in normal hydraulic usage. It
ria
and head. results from that part of the boundary stresses
2. The possible existence of tensions in the pore wa- that is transmitted to the water phase, from pres-
ter. sures generated by capillarity menisci, and from
3. Compositional differences in the water from water adsorption forces exerted by particle sur-
faces. A piezometer measures the matrix poten-
ate
point-to-point and adsorptive force fields from
particle surfaces. tial if it contains fluid of the same composition
4. Differences in interparticle forces and the energy as the soil water.
state of the pore fluid from point to point owing 3. Osmotic (or solute) potential s (head hs, pres-
to thermal, electrical, and chemical gradients. sure ps) is the work per unit quantity of water to
Such gradients can cause fluid flows, deforma-
dM transport reversibly and isothermally an infinites-
tions, and volume changes, as considered in more imal quantity of water from a pool of pure water
detail in Chapter 9. at a specified elevation and atmospheric pressure
to a pool containing a solution identical in com-
Some formalism in definition and terminology is position to the soil water, but in all other respects
necessary to avoid confusion. The status of water in a identical to the reference pool. This component
soil can be expressed in terms of the free energy rel- is, in effect, the osmotic pressure of the soil wa-
ative to free, pure water (Aitchison, et al., 1965). The ter, and it depends on the composition and ability
hte
free energy can be (and is) expressed in different ways, of the soil particles to restrain the movement of
including adsorbed cations. The osmotic potential is nega-
tive, that is, water tends to flow in the direction
1. Potential (dimensions—L2T⫺2: J/kg) of increasing concentration.
2. Head (dimensions—L: m, cm, ft)
3. Pressure (dimensions—ML⫺1 T⫺2: kN/m2, dyn/ The total potential, head, and pressure then become
rig
l
ria
ate
dM
Figure 7.7 Schematic representation of a saturated soil for analysis of pressure conditions.
pore pressure, Z ⫽ 0, hm ⫽ hm0, and hs0 ⫽ 0 if pure At point 2, which is between the same two clay par-
water is used in the piezometer. Thus, ticles as point 1 but closer to a particle surface, there
hte
It follows that u0
hm2 ⫹ hs2 ⫽ hm1 ⫹ hs1 ⫽ hm0 ⫽
w
rig
P0 ⫽ hm0 w ⫽ u0 (7.22)
A similar analysis could be applied to any point in the
system. If point 3 were midway between two clay par-
the measured pore pressure. ticles spaced the same distance apart as the particles
Point 1 is at the same elevation as point 0, except it on either side of point 1, then hs3 ⫽ hs1, but Z3 ⫽ 0.
py
is inside the soil mass and midway between two clay Thus,
particles. At this point, Z1 ⫽ 0, but hs ⫽ 0 because the
electrolyte concentration is not zero. Thus,
u0
⫽ Z3 ⫹ hm3 ⫹ hs3 ⫽ Z3 ⫹ hm3 ⫹ hs1 (7.24)
w
Co
H1 ⫽ 0 ⫹ hm1 ⫹ hs1
A partially saturated system can also be analyzed,
If no water is flowing, H1 ⫽ H0, and but the influences of curved air–water interfaces must
be taken into account in the development of the hm
hm1 ⫹ hs1 ⫽ hm0 terms.
The conclusions that result from the above analysis
Also, because p1 ⫽ p0 ⫽ u0 of component potentials are:
1. As the osmotic and gravitational components
u0 ⫽ hm1 w ⫹ hs1 w (7.23) vary from point to point in a soil at equilibrium,
l
3. A pore pressure measurement using a piezometer 4. Consolidation Tests The consolidation pressure
ria
containing pure water gives a pressure u0 ⫽ wh, on a sample at equilibrium is the soil water suc-
where h is the pressure head at the piezometer. tion. If the consolidation pressure were instanta-
When referred back to points between soil par- neously removed, then a negative water pressure
ticles, u0 is seen to include contributions from or suction of the same magnitude would be
osmotic pressures as well as matrix pressures. needed to prevent water movement into the soil.
ate
Since osmotic pressures are the cause of long- 5. Vapor Pressure Methods The relationship be-
range repulsions due to double-layer interactions, tween relative humidity and water content is used
measured pore water pressures may include con- to establish the relationship between suction and
tributions from long-range interparticle repulsive water content.
forces. dM 6. Osmotic Pressure Methods Soil samples are
equilibrated with solutions of known osmotic
pressure to give a relationship between water
7.8 MEASUREMENT OF PORE PRESSURES IN content and water suction.
SOILS 7. Dielectric Sensors Such as Capacitance Probes
and Time Domain Reflectometry Soil moisture
Several techniques for the measurement of pore water can be indirectly determined by measuring the
pressures are available. Some are best suited for lab- dielectric properties of unsaturated soil samples.
hte
oratory use, whereas others are intended for use in the With the knowledge of soil water characteristics
field. Some yield the pore pressure or suction by direct relationship (Section 7.11), the negative pore
measurement, while others require deduction of the pressure corresponding to the measured soil
value using thermodynamic relationships. moisture can be determined. The capacitance
probe measures change in frequency response of
1. Piezometers of Various Types Water in the pi- the soil’s capacitance, which is related to dielec-
rig
ezometer communicates with the soil through a tric constants of soil particle, water, and air. The
porous stone or filter. Pressures are determined capacitance is largely influenced by water con-
from the water level in a standpipe, by a manom- tent, as the dielectric constant of water is large
eter, by a pressure gauge, or by an electronic compared to the dielectric constants of soil
pressure transducer. A piezometer used to mea-
py
cially prepared gypsum block or porous ceramic erties. Volumetric water content can therefore be
block are measured. The water held by the block indirectly measured from the travel time mea-
determines the resistance or permittivity, and the surement.
moisture tension in the surrounding soil deter-
mines the amount of moisture in the block Piezometer methods are used when positive pore
(Whalley et al., 2001). The same principle can be pressures are to be measured, as is usually the case in
applied by placing a dry filter paper on a soil dams, slopes, and foundations on soft clays. The other
specimen and allowing the soil moisture to ab- methods are suitable for measurement of negative pore
sorb into the paper. When the suction in the filter pressures or suction. Pore pressures are often negative
paper is equal to the suction in the soil, the two in expansive and partly saturated soils. More detailed
descriptions and comparisons of these and other meth- defined effective stress ⫽ ⫺ u0 differ by the net
ods are given by Croney et al. (1952), Aitchison et al. interparticle stress due to physicochemical contribu-
(1965), Richards and Peter (1987), and Ridley et al. tions,
(2003).
i ⫺ ⫽ A ⫺ R (7.30)
7.9 EFFECTIVE AND INTERGRANULAR
PRESSURE When A and R are both small, as would be true in
granular soils, silts, and clays of low plasticity, or in
In Section 7.5, it was shown that the intergranular pres- cases where A ⬇ R, the intergranular and effective
sure is given by stress are approximately equal. Only in cases where
l
either A or R is large, or both are large but of signifi-
i ⫽ ⫹ A ⫺ u
ria
(7.25) cantly different magnitude, would the intergranular and
effective stress be significantly different. Such a con-
where u is the hydrostatic pressure between particles dition appears not to be common, although it might be
(or hm w in the terminology of Section 7.7). General- of importance in a well-dispersed sodium montmoril-
ized forms of Eq. (7.24) are lonite, where compression behavior can be accounted
ate
for reasonably well in terms of double-layer repulsions
u0 ⫽ Z w ⫹ hm w ⫹ hs w (7.26) (Chapter 10).4
The derivation of Eq. (7.30) assumed vertical equi-
and librium, with contributing forces parallel to each other,
that is, the intergranular stress i is the sum of the
u ⫽ hm w ⫽ u0 ⫺ Z w ⫺ hs w (7.27) skeletal forces (defined as ⫽ ⫺ u0) and the elec-
dM trochemical stress (A ⫺ R), as illustrated in Fig. 7.8a.
Thus, Eq. (7.25) becomes, for the case of no elevation This implies that the deformation induced by the elec-
difference between a piezometer and the point in ques- trochemical stress (A ⫺ R) is equal to the deformation
tion (i.e., Z ⫽ 0), induced by the skeletal forces at contacts [i.e., a ‘‘par-
allel’’ model as described by Hueckel (1992)]. The
i ⫽ ⫹ A ⫺ u0 ⫹ hs w (7.28) change in pore fluid chemistry at constant confinement
hte
pore water pressure between particles u. Equation arrangement. Increase in intergranular stress i or ef-
(7.29) indicates that if intergranular pressure i is to fective stress changes the interparticle spacing,
be expressed in terms of a measured pore pressure u0, which may contribute to changes in strength properties
then the long-range repulsion R must also be taken into upon shearing.
account. The actual hydrostatic pressure between par-
ticles u ⫽ u0 ⫹ R includes the effects of long-range
repulsions as required by the condition of constant to-
4
tal potential for equilibrium. A detailed analysis of effective stress in clays is presented by Chat-
topadhyay (1972), which leads to similar conclusions, including Eq.
In the general case, therefore, the true intergranular (7.29). i was termed the true effective stress and it governed the
pressure i ⫽ ⫹ A ⫺ u0 ⫺ R and the conventionally volume change behavior of Na–montmorillonite.
Skeletal Force
Skeletal Force
Electrochemical Force Electrochemical Force
Skeletal Force
Skeletal Force Electrochemical Force
Electrochemical Force
σi σi
l
ria
Skeletal Force
σ = σ _ u0
Skeletal Force Electrochemical
Particle Deformation
σ = σ _ u0 Force A _ R
by Skeletal Force
Electrochemical
Force A _ R
ate
Deformation at
the Contact Total Deformation
σi
σi at the Contact
σi = σ _ u0 + A _ R σi = σ _ u0 = A _ R
dM (a) (b)
Since the particles are arranged in parallel as well of saturated soils. Skempton proposed three possible
as nonparallel manner, the chemomechanical coupling relationships for effective stress in saturated soils:
behavior of actual soils can be far from the predictions
made by the above two models. In fact, Santamarina 1. The true intergranular pressure for the case when
(2003) argues that the impact of skeletal forces by ex- A⫺R⫽0
ternal forces, particle-level forces, and contact-level
rig
with measurable interparticle repulsive and attractive 2. The solid phase is treated as a real solid that has
forces]. compressibility Cs and shear strength given by
i ⫽ k ⫹ tan (7.33)
Co
冉 冊
stress that governs consolidation and strength behavior
of soils. However, its usefulness has been established ac tan
⫽ ⫺ 1 ⫺ u (7.34)
from the experience of many years of successful ap- tan
plication in practice. Skempton (1960b) showed that
the Terzaghi equation does not give the true effective where is the effective stress angle of shearing
stress but gives an excellent approximation for the case resistance. For volume change,
⫽ ⫺ 1 ⫺ 冉 Cs
C 冊
u (7.35)
Table 7.1 Compressibility Values for Soil, Rock,
and Concrete
l
acted upon by both a total stress and a pore water Dense sand 18 0.028 0.0015
ria
pressure according to Loose sand 92 0.028 0.0003
London clay (over cons.) 75 0.020 0.00025
V Gosport clay (normally cons.) 600 0.020 0.00003
⫽ ⫺C
(7.37)
V
After Skempton (1960b).
ate
and also satisfied the Coulomb equation for drained
a
Compressibilities at p ⫽ 98 kN/m2; water Cw ⫽ 0.49
shear strength d : ⫻ 10⫺6 per kN/m2.
d ⫽ c ⫹ tan (7.38)
dM
when both a total stress and a pore pressure are acting.
It may be noted that this approach assumes that the
A more rigorous evaluation of the contribution of
soil particle compressibility to effective stress was
Coulomb strength equation is valid a priori. made by Lade and de Boer (1997) using a two-phase
The results of Skempton’s analysis showed that Eq. mixture theory. The volume change of the soil skeleton
(7.32) was not a valid representation of effective stress. can be separated into that due to pore pressure incre-
Equations (7.34) and (7.35) give the correct results for ment
u and that due to the change in confining pres-
soils, concrete, and rocks. Equation (7.36) accounts sure
( ⫺ u) (or
⫺
u). The effective stress
hte
well for the behavior of soils but not for concrete and increment
is defined as the stress that produces the
rock. The reason for this latter observation is that in same volume change,
soils Cs /C and ac tan /tan approach zero, and,
thus, Eqs. (7.34) and (7.35) reduce to Eq. (7.36). In CV0
⬅
Vsks ⫹
Vsku ⫽ CV0(
⫺
u)
rock and concrete, however, Cs /C and ac tan /tan
are too large to be neglected. The value of tan /tan ⫹ CuV0
u (7.40)
rig
5
A more general expression has been proposed as ij ⫽ ij ⫹ iju, Lade and de Boer (1997) used this equation to de-
where ij is the tensor that accounts for the constitutive characteristics rive an effective stress equation for granular materials
of the solid such as complex kinematics associated with anisotropic
elastic materials (Carroll and Katsube, 1983; Coussy, 1995; Did- under drained conditions. Consider a condition in
wania, 2002). which the total confining pressure is constant [
( ⫺
l
rock with small interconnected pores and and de Boer (1997)
ria
low porosity (Lade and de Boer, 1997)
Cs Equation (7.43) Suklje (1969); Lade and de Boer (1997)
1 ⫺ (1 ⫺ n)
C
After Lade and de Boer (1997).
ate
u) ⫽ 0], but the pore pressure changes by
u.6 The
volume change of soil skeleton caused by change in
pore pressure (
Vsku) is attributed solely from the vol-
dM
umetric compression of the solid grains (
Vgu). Hence,
⫽
⫺ 冋
1 ⫺ (1 ⫺ n)
Cs
C 册
u or
rig
冋
⫽ 1 ⫺ (1 ⫺ n)
Cs
C 册 (7.43)
quartz sand and gypsum sand (Lade and de Boer, Figure 7.9 Variation of with stress for quartz sand and
1997). For a stress level less than 20 MPa, is essen- gypsum sand (Lade and de Boer, 1997).
tially one. Thus, Terzaghi’s effective stress equation,
while not rigorously correct, is again shown to be an
Co
excellent approximation in almost all cases for satu- Can the effective stress concept also be applied for
rated soils (i.e., solid grains and pore fluid are consid- undrained conditions where drainage is prevented?
ered to be incompressible compared to soil skeleton That is, when an isotropic total stress load of
iso is
compressibility). applied, is
u equal to
iso? Using a two-phase mix-
ture theory, the total stress increment (
iso) is sepa-
rated into partial stress increments for the solid phase
6
(
s) and the fluid phase (
ƒ) (Oka, 1996). Consid-
An example of this condition is a soil under a seabed, in which the
sea depth varies. This condition is often called the ‘‘unjacked con- ering that the macroscopic volumetric strains by two
dition.’’ phases are equal but of opposite sign for undrained
l
Air
Cl is the compressibility of pore fluid. θ
ria
If the excess pore pressure generated by undrained
Solid surface
isotropic loading
is
u, the partial stress increment
for the fluid phase becomes (Oka, 1996) (b)
ate
ƒ ⫽ n
u (7.45) Water
Air
Combining Eqs. (7.45) and (7.46),
Solid
C ⫺ Cs
u ⫽
iso (7.46)
dM
C ⫺ Cs ⫹ n(Cl ⫺ Cs)
(c)
The multiplier in the right-hand side of the above Figure 7.10 Wettability of two fluids (water and air) on a
equation is in fact Bishop’s pore water pressure coef- solid surface: (a) contact angle less than 90, (b) contact an-
ficient B (Bishop and Eldin, 1950).7 For typical soils gle more than 90, and (c) unsaturated sand with water as the
(Cs ⬇ 1.9 ⫺ 2.7 ⫻ 10⫺8 m2 / kN, Cl ⬇ 4.9 ⫻ 10⫺9 wetting fluid and air as the nonwetting fluid.
m2 /kN, C ⬇ 10⫺5 ⫺ 10⫺4 m2 /kN), so the values of B
hte
7.11 WATER–AIR INTERACTIONS IN SOILS The contact angle is a property related to interac-
tions of solid and two fluids (water and air, in this
Wettability refers to the affinity of one fluid for a solid case).
surface in the presence of a second or third fluid or
gas. A measure of wettability is the contact angle, as ⫺ ws
py
which was introduced in Eq. (7.9). Figure 7.10 illus- cos ⫽ (7.47)
trates a drop of the reference liquid (water for Fig. aw
7.10a and air for Fig. 7.10b) resting on a solid surface
in the presence of another fluid (air for Fig. 7.10a and where as is the interfacial tension between air and
solid, ws is the interfacial tension between water
Co
l
rating two fluids. This pressure difference exists be- 102 4
3
ria
cause of the interfacial tension present in the fluid–
fluid interface. Matric suction is a property that causes 101 2
porous media to draw in the wetting fluid and repel 1
the nonwetting fluid and is defined as the difference 100
between the nonwetting fluid pressure and the wetting
ate
fluid pressure. For a two-phase system consisting of 10-1
water and air, the matric suction is 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Volumetric Water Content θ w
⫽ un ⫺ uw (7.48)
Figure 7.11 Soil–water characteristic curves for some Dutch
soils (from Koorevaar et al., 1983; copied from Fredlund and
dM
where un is the pressure of the nonwetting fluid (air)
and uw is the pressure of the wetting fluid (water).
Rahardjo, 1993).
the interfacial tension between the two liquids. The as shown in Fig. 7.12a. One cause of hysteresis is the
pressure drop across the interface is directly propor- existence of ‘‘ink bottle neck’’ pores at the microscopic
tional to the interfacial tension and inversely propor- scale as shown in Fig. 7.12b. Larger water-filled pores
tional to the radius of curvature. It follows that higher can remain owing to the inability of water to escape
air pressure is required for air to enter water-saturated through smaller openings below in the case of drainage
fine-grained than coarse-grained materials. or above in the case of evaporation. Another cause is
rig
Soil contains a range of different pore sizes, which irreversible change in soil fabric and shrinkage during
will drain at different capillary pressure values. This drying.
leads to a soil–water characteristic relationship in The curves in Fig. 7.11 have two characteristic
which the matric suction is plotted against the volu- points—the air entry pressure a and residual volu-
py
metric water content (or sometimes water saturation metric water content r as defined in Fig. 7.12a. The
ratio) such as shown in Fig. 7.11.9 The curves are often entry pressure is the matric suction at which the air
determined during air invasion into a previously water- begins to enter the pores and the pores become inter-
saturated soil. As the volumetric water content de- connected (Corey, 1994). At this point, the air per-
meability becomes greater than zero. Corey (1994)
Co
9 10
The soil–water characteristic curve is referred to by a variety of For the Dense NAPL–water two-phase system (often Dense NAPL
names depending on different disciplines. They include moisture re- is the nonwetting fluid and water is the wetting fluid), the displace-
tention, soil–water retention, specific retention, and moisture char- ment pressure may be important to examine the potential of DNAPL
acteristic. invading into a noncontaminated water-filled porous media.
Scanning
Curve
Suction
Scanning
Hysteresis
Curve
ψa
ψd
Main Drying Curve
l
Main Wetting
ria
Curve
θr Water Content
Wetting
θ r Residual Water Content
ate
ψ a Air Entry Value
ψ d Displacement pressure
(b)
(a)
Figure 7.12 Hysteresis of a soil–water characteristic curve: (a) effect of hysteresis and (b)
dM
ink bottle effect: a possible physical explanation for the hysteresis.
greater than the displacement pressure because pore algebraic formulas have been proposed to fit the mea-
throats smaller than the maximum must be penetrated sured soil-water characteristic relations. The most pop-
to establish air connectivity. The air entry pressure is ular ones are (a) the Brooks–Corey (1966) equation:
hte
when d (7.50)
rig
冋冉 冊 册
that the residual saturation defined here is a mathe- ⫺1 / m 1⫺m
matical fitting parameter without a specific quantitative ⫺ r
⫽ 0 ⫺1 (7.51)
value. m ⫺ r
The shape of the soil–water characteristic curve de-
Co
pends on many factors, including the grain size distri- where 0 and m are curve-fitting parameters.
bution, soil fabric, the contact angle, and the interfacial Various modifications have been proposed to these
tension [see Eq. (7.11)]. If the material is uniform with equations to include behaviors such as hysteresis, non-
a narrow range of pore sizes, the curve has three dis- wetting fluid trapping, and three-phase conditions.
tinct parts: a straight part up to the air entry pressure,
a relatively horizontal middle part, and an end part that
is almost vertical (soil 1 in Fig. 7.11). On the other 7.12 EFFECTIVE STRESS IN UNSATURATED
hand, if the material is well graded, the curve is SOILS
smoother (soils 3, 4, and 5 in Fig. 7.11). The capillary
pressure increases gradually as the water saturation de- Although it seems clear that the volume change and
creases and the middle part is not horizontal. Many strength behavior of partly saturated soils are con-
l
the effective stress since uw is negative. Thus, the
ria
Bishop equation is appealing intuitively because neg-
ative pore pressures are known to increase strength and
decrease compressibility. Using Eq. (7.52), the shear 0.80
strength of unsaturated soil can be expressed as
ate
0.76
⫽ {( ⫺ ua) ⫹ (ua ⫺ uw)}tan (7.53)
Void Ratio e
Initially Soaked Test
where is the effective friction angle of the soil. 0.72
However, difficulties in the evaluation of the parameter
, its dependence on saturation ( ⫽ 1 for saturated
dM Air Dry (8 specimens)
soils and ⫽ 0 for dry soils), and that the relationship
0.68 Soaked at Constant Void Ratio
between and saturation is soil dependent, as shown
Soaked at Constant Applied
in Fig. 7.13a, all introduce problems in the application Pressure
of Eq. (7.53). Since water saturation is related to matric
0.64
suction as described in Section 7.11, it is possible that 10 100 1000
depends on matric suction as shown in Fig. 7.13b.
Applied Pressure (kPa )
Nonetheless, because of the complexity in determining
hte
, the attempt to couple total stress and suction to- Figure 7.14 Oedometer compression curves of unsaturated
gether into a single equivalent effective stress is un- silty soils (after Jennings and Burland, 1962 in Leroueil and
certain (Toll, 1990). Hight, 2002).
rig
1. Compacted
Boulder Clay – 0.55
(ua – uw)
2. Compacted Shale χ=
(ua – uw)
Coefficient χ
3. Breadhead silt
py
Coefficient χ
4. Silt
5. Silty clay
6. Sterrebeek silt
7. White clay
Co
Figure 7.13 Variation of parameter with the degree of water saturation Sr for different
soils: (a) versus water saturation (after Gens, 1996) and (b) versus suction (after Khalili
and Khabbaz, 1998).
The void ratio decreased upon soaking and the final in which a and b are material parameters that may also
state was very close to the compression curve of the depend on degree of saturation and stress. For exam-
saturated silt. Additional tests in which constant vol- ple, Fredlund et al. (1978) propose the following equa-
ume during soaking was maintained by adjusting the tion:
applied load were also done. Again, after equilibrium,
the state of soaked samples was close to the compres- ⫽ ( ⫺ ua)tan ⫹ (ua ⫺ uw)tan b (7.55)
sion curve of the saturated silt. Soaking reduces the
suction and, hence, Bishop’s effective stress decreases. where b is the angle defining the rate of increase in
This decrease in effective stress should be associated shear strength with respect to soil suction. An example
with an increase in void ratio. However, the experi- of this parameter as a function of water content, fric-
mental observations gave the opposite trend (i.e., a de- tion angle, and matric suction is given by Fredlund et
l
crease in void ratio is associated with irreversible al. (1995).
ria
compression). The presence of meniscus water lenses Similarly, the change in void ratio
e of an unsat-
in the soil before wetting was stabilizing the soil struc- urated soil can be given by (Fredlund, 1985)
ture, which is not taken into account in Bishop’s equa-
tion (7.52).
⫽ at
( ⫺ ua) ⫹ am
(ua ⫺ uw) (7.56)
An alternative approach is to describe the shear
ate
strength/deformation and volume change behavior of where at is the coefficient of compressibility with re-
unsaturated soil in terms of the two independent stress spect to changes in ⫺ ua and am is the coefficient of
variables ⫺ ua and ua ⫺ uw (Coleman, 1962; Bishop compressibility with respect to changes in capillary
and Blight, 1963; Fredlund and Morgenstern, 1977; pressure. A similar equation, but with different coef-
Fredlund, 1985; Toll, 1990, Fredlund and Rahardjo,
dM ficients, can be written for change in water content.
1993; Tarantino et al., 2000). Figure 7.15 shows the For a partly saturated soil, change in water content and
results of isotropic compression tests of compacted ka- change in void ratio are not directly proportional.
olin. Different compression curves are obtained for The two stress variables, or their modifications that
constant suction conditions, and relative effects of ⫺ include porosity and water saturation, have been used
ua and ua ⫺ uw on volume change behavior can be in the development of elasto-plastic-based constitutive
observed. Furthermore, the preconsolidation pressure models for unsaturated soils (e.g., Alonso et al., 1990;
(or yield stress) increases with suction. Wheeler and Sivakumar, 1995; Houlsby, 1997; Gallip-
hte
On this basis, the dependence of shear strength on oli et al., 2003). The choice of stress variables is still
stress is given by equations of the form in debate; further details on this issue can be found in
Gens (1996), Wheeler and Karube (1996), Wheeler et
⫽ a( ⫺ ua) ⫹ b(ua ⫺ uw) (7.54)
al. (2003), and Jardine et al. (2004).
Bishop’s parameter in Eq. (7.52) is a scalar quan-
tity, but microscopic interpretation of water distribution
rig
1.25
in pores can lead to an argument that is directional
Preconsolidation dependent (Li, 2003; Molenkamp and Nazemi,
pressure
1.20
2003).11 During the desaturation process, the number
of soil particles under a funicular condition decreases,
py
1.15
and they change to a pendular condition with further
drying. For particles in the funicular region, the suction
Void Ratio e
1.10
pressure acts all around the soil particles like the water
ua _ uw (kPa) pressure as illustrated in Fig. 7.4a. Hence, the effect is
Co
1.05 300 kPa isotropic even at the microscopic level. However, once
the microscopic water distribution of a particle changes
Curves are Averages of 200 kPa to the pendular condition, the capillary forces only act
1.00 Several Tests 100 kPa on a particle at locations where water bridge forms and
0 kPa the contribution to the interparticle forces becomes
0.95
25 50 100 200 400
σ _ ua (kPa)
11
A microstructural analysis by Li (2003) suggests the following ef-
Figure 7.15 Isotropic compression tests of compacted kaolin fective stress expression:
(after Wheeler and Sivakumar, 1995 in Leroueil and Hight,
2002). ij ⫽ ij ⫺ uaij ⫹ ij (ua ⫺ uw)
l
electrochemical forces, creating large matric suction. water content relation of the sand is given by soil
ria
Although uw ⫽ u0 is used in practice, the actual pore 1 in Fig. 7.11.
pressure u acting at interparticle contacts may be dif- a. Calculate the saturated unit weight and dry unit
ferent from u0, as discussed in Section 7.9. The con- weight.
tribution of the long-range interparticle forces to b. Evaluate the unit weights at different saturation
mechanical behavior of unsaturated clayey soils re-
ate
ratios Sw.
mains to be fully evaluated.
c. Plot the hydrostatic pore pressures with depth
down to a depth of 10 m and evaluate the satu-
ration ratios above the water table.
7.13 CONCLUDING COMMENTS
d. Along with the hydrostatic pore pressure plot,
sketch the vertical total stress with depth using
dM
The concepts in this chapter provide insight into the
meanings of intergranular pressure, effective stress, the unit weights calculated in parts (a) and (b).
and pore water pressure and the factors controlling e. Estimate the vertical effective stress with depth.
their values. Because soils behave as particulate ma- Use Bishop’s equation (7.52) with ⫽ Sw. Com-
terials and not as continua, knowledge of these stresses ment on the result.
and of the factors influencing them is a necessary pre-
requisite to the understanding and quantification of 2. Repeat the calculations done in Question 1 with soil
5 in Fig. 7.11. The specific gravity of the soil is
hte
structure. These factors are, in turn, influenced by the diameter of 0.2 mm and the void ratio is 0.7. As-
composition and ambient conditions. The relationships sume k/(1 ⫹ e) ⫽ 3.1. Consider the following two
between soil structure and effective stress are devel- cases: (a) ⫽ 0.0075 mm and ⫽ 5 and (b)
oped further in Chapter 8. Chemical, electrical, and ⫽ 0.025 and ⫽ 30. Comment on the results.
thermal influences on effective pressures and fluid
py
pressures in soils have not been considered in the de- 4. Compute the following contact forces at different
velopments in this chapter. They may be significant, particle diameters d ranging from 0.1 to 10 mm.
however, as regards soil structure stability fluid flow, Comment on the results in relation to the effective
volume change, and strength properties. They are an- and intergranular pressure described in Section 7.9.
Co
alyzed in more detail in subsequent chapters. a. Weight of the sphere, W ⫽ –61 Gs wd 3, where Gs
An understanding of the components of pore water is the specific gravity (say 2.65) and w is the
pressure is important to the proper measurement of unit weight of water.
pore pressure and interpretation of the results. Inclu- b. Contact force by external load, N ⫽ d 2, where
sion of the effect of pore water suction and air or gas is the external confining pressures applied.
pressure on the mechanical behavior of unsaturated The equation is approximate for a simple cubic
soils requires modification of the effective stress equa- packing of equal size spheres (Santamarina,
tion used for saturated soils. Complications arise from 2003). Consider two cases, (i) ⫽ 1 kPa (⬇
the difficulty in the choice of stress variables and in depth of 0.1 m) and (ii) ⫽ 100 kPa (⬇ depth
treatment of contact-level forces (i.e., capillary forces of 10 m).
c. Long-range van der Waals attraction force, A ⫽ 7. Give a microscopic interpretation for why an un-
Ahd/(24t 2), where Ah is the Hamaker constant saturated soil can collapse and decrease its volume
(Section 6.12) and t is the separation between upon wetting as shown in Fig. 7.14 even though the
particles (Israelachvili, 1992, from Santamarina, Bishop’s effective stress decreases.
2003). Use Ah ⫽ 10⫺20 N-m and t ⫽ 30 Å.
8. Clay particles in unsaturated soils often aggregate
5. Discuss why it is difficult to measure suction using creating matrix pores and intraaggregate pores. Air
a piezometer-type tensiometer for long-term moni- exists in the matrix pores, but the intraaggregate
toring of pore pressures. Describe the advantages of pores are often saturated by strong water attraction
other indirect measurement techniques such as po- to clay surfaces. The total potential of unsaturated
rous ceramic and dielectric sensors. soil can be extended from Eq. (7.19) to ⫽ g ⫹
m ⫹ s ⫹ p, where p is the gas pressure poten-
l
6. For the following cases, compare the effective
tial.12 Discuss the values of each component of the
ria
stresses calculated by the conventional Terzaghi’s
equation and by the modified equation (7.39) with above equation in the matrix pores and the intraag-
values presented in Fig. 7.8. Discuss the possible gregate pores.
errors associated with effective stress estimation by
Terzaghi’s equation.
ate
12
a. Pile foundation at a depth of 20 m. This was proposed by a Review Panel in the Symposium on Mois-
ture Equilibrium and Moisture Changes in Soils Beneath Covered
b. A depth of 5 km from the sea level where the Areas in 1965.
subsea soil surface is 1 km deep.
dM
hte
rig
py
Co
l
Structure, Geotechnical
ria
Properties, and Stability
ate
8.1 INTRODUCTION structures with characteristic properties, how these
properties and the associated behavior are interelated,
Long before methods were available to confirm spe-
dM and why these processes and properties are relevant
cific soil fabrics and structures, hypotheses were ad- to geotechnical applications are the subjects of this
vanced about them, their formation, and their stability chapter.
in an attempt to account for such phenomena as
strength loss of clays on remolding, differences in
properties of soils deposited in different environments,
collapsing soils and soil liquefaction, creep and sec- 8.2 STRUCTURE DEVELOPMENT
ondary compression, pore pressure generation during
Early Concepts
hte
both fabric and its stability into account, the term bance of many natural clays at constant water content.
structure is used. Particles, particle groups, and their This sensitivity of the undisturbed structure, which is
associations, together with interparticle forces and ap- quantified as the ratio of the undisturbed to fully re-
plied stresses, determine the overall soil structure. molded strength at the same water content, can be great
enough to give the strength loss due to remolding
py
original stress state. Thoroughly remolded and re- at points of contact between particles. Disturbance of
worked soil is said to be destructured. Virtually every the clay caused contacts to rupture, more water to fill
natural, undisturbed soil has structure. As emphasized in around the old contact points, and the strength to
by Leroueil and Vaughan (1990), the structure can be drop. Different adsorbed ions were also recognized as
as significant in determining engineering behavior as possibly responsible for differences in strength and
can such important factors as porosity and stress his- sensitivity (Terzaghi, 1941). Goldschmidt (1926) hy-
tory. pothesized that particles in sensitive clay are arranged
How residual and transported soil deposits are in a ‘‘cardhouse’’ that collapses on remolding.
formed, how the formative processes and subsequent A load-carrying skeleton consisting of highly com-
changes over time act to produce unique types of soil pressed ‘‘bond clay’’ trapped between silt and fine sand
195
l
Structure-determining factors and processes are sum-
ria
marized in Fig. 8.2. Initial conditions dominate the
structure of young deposits at high porosity or freshly
compacted soils, whereas older soils at lower porosity
are likely to be influenced more by the postdeposi-
ate
tional changes.
Single-grain fabrics are uncommon in soils contain-
ing clay. Complex fabric units of micrometer-to-
millimeter size or greater consisting of skeleton grains,
clay aggregates, and pores are characteristic of most
dM fine-grained soil structures.
The principle of chemical irreversibility of clay fab-
ric (Bennett and Hurlbut, 1986) applies generally to
fine-grained soil deposits. This principle recognizes
that the chemical environment is critical during the in-
itial stages of sediment fabric formation in water. How-
Figure 8.1 Strength loss of a clay that is extremely sensitive
ever, after the initial flocculation of particles and
deposition, the chemistry is much less important in in-
hte
assumed to form by simultaneous deposition of floc- The texture of residual soils formed by the in-place
culated clay particles and silt and sand grains in the weathering of crystalline rocks may be quite similar to
saltwater environment. The clay deposited in the inter- that of the parent rock. Clay particles may form coat-
stices between the elements of the skeleton, termed ings over silt and sand grains as a result of repeated
matrix clay, is assumed to be only partly consolidated wetting and drying. Open, porous fabrics form in some
py
and remain at high water content. Remolding mixes zones, while dense, low-porosity fabrics form in oth-
the matrix and bond clays, thereby destroying the pri- ers, and heterogeneity is common.
mary load-carrying structure and causing a reduction Intense weathering and leaching, coupled with an
in strength. abundance of aluminum and iron oxides, produces fab-
Co
Winterkorn and Tschebotarioff (1947) suggested rics and textures ranging from open granular to dense
that sensitivity resulted from a cementation similar to and clayey in tropical and subtropical soils. Concre-
that in loess and sandstone. This cementation was at- tions and nodules are common in some of these ma-
tributed to slow recrystallization or formation of ce- terials. For example, a red kaolinitic clay from Kenya
menting materials from inorganic substances of low is composed of ‘‘crumbs’’ made up of ‘‘subcrumbs’’
solubility. that can in turn break up into ‘‘sub-subcrumbs’’ that
In the years since the formulation of these ideas of contain a random arrangement of individual particles
soil structure, it has been possible to determine fabrics (Barden, 1973). Pores are in two classes: irregular
and compositions in more detail and, along with a bet- pores of about 1 m and very small pores of about
ter understanding of the stress–deformation–strength 5 nm.
l
ria
ate
Figure 8.2 Structure-determining factors and processes.
Alluvial Soils
dM work of filmy particles may form. Ca–montmorillonite
Alluvial soils can be deposited in marine, brackish wa- particles are usually made up of several unit layers.
ter, or freshwater basins. Single clay particles are rare, Some heavily consolidated montmorillonites exhibit
and flocculated fabrics of particle groups can form in surprisingly little preferred orientation.
water over the full range of salinities. Edge-to-face There is little or no preferred particle orientation in
soft marine and brackish water illitic clays, except
hte
uted, except in varved or stratified clays, and the larger within which the clay plates are intensely oriented.
grains are not usually in contact with each other. Open The open packing of sensitive postglacial clay may
initial fabrics are characteristic of water-laid sediments, be due in part to the presence of very small quartz
with the degree of openness dependent on clay min- particles of platy morphology (Krinsley and Smalley,
eralogy, particle size, and water chemistry, including 1973; Smalley et al., 1973). Below a critical size of
py
both the total salt content and the monovalent/divalent about a cleavage mechanism appears to exist, so platy
cation ratio. The intensity of flocculation may be less particles of quartz and possibly other nonclay minerals
in brackish and freshwater deposits, so subsequent form as a result of grinding.
consolidation can cause greater preferred orientation of Organic matter in the form of microscopic animal
Co
platy particles and particle groups than in saltwater and plant fragments, microorganisms, and organic
clays. Very slow accumulation rates allow for more compounds can have a profound effect on the structure
stability in open fabrics than is possible when the sed- and properties of postglacial clays (Söderblom, 1966;
iment accumulates rapidly. Pusch, 1973a, 1973b). The number of bacteria in the
Aggregates in illitic clay contain particle arrange- oceans is from 1 ⫻ 109 to 3 ⫻ 1011 per m3 at depths
ments ranging from random to booklike. Booklike of 10 to 50 m beneath the surface (Reinheimer, 1971).
aggregates are most common in kaolinite. The concen- It is probable that microorganisms were prevalent in
tration and type of adsorbed cations usually controls the ocean at the time postglacial clays were formed as
the basic fabric units in smectite. Na–montmorillonite well. As organic material and clay surfaces interacted,
can separate into unit layers, and an interwoven net- organic matter was attached to the sedimentary aggre-
gates. In the new environment most of the organisms mineral types are characteristic of these materials.
died or became dormant because of the absence of nu- Well-developed domains of clay are common, and
trients, subsequently contributing humic acids, fulvic there may be soft clay zones that bridge over some
acids, and humus. Either aggregating or dispersing ten- pores caused by the arching action of the large parti-
dencies result, depending on the environment. Electron cles. High past stresses on some boulder clays have
micrographs of ultrathin sections (Pusch, 1973a) show developed macrofabric features that include shear
organic matter both as fluffy bodies and as distinct ob- zones and shear planes.
jects associated with aggregates.
Fabric anisotropy as a result of one-dimensional Remolded and Compacted Soil Fabrics
compression after deposition will ordinarily result in The fabric immediately after remolding or compacting
some anisotropy of mechanical properties. Currents, a soil depends on several factors, including strength of
l
waves, and slopes may also cause preferred orienta- preexisting fabric units, compaction method, and com-
ria
tions of particles. An example for Portsea Beach sand paction or remolding effort. The general effects of dis-
is shown in Fig. 8.3. The long axes of elongate parti- turbance and remolding at constant water content are
cles show preferred orientations parallel to the coast- to break down flocculated aggregations, destroy shear
line and dipping landward at an angle of about 10. planes, eliminate large pores, and produce a more ho-
ate
mogeneous fabric (on a macroscopic scale). Whether
Aeolian Soils or not there will be a preferred direction of particle
Wind-deposited soils such as loess are characterized orientation depends on the methods used. When well-
by particles in the silt and fine sand ranges, although defined shear planes are formed, there usually is an
small amounts of clay are often present. These depos- alignment of platy particles or particle groups along
its, which are usually partly saturated, are often subject
dM the shear plane.
to collapse if saturated. The loose metastable fabric is Under anisotropic consolidation conditions, plates
maintained by clay and light carbonate cementation at align with their long axes in the plane acted on by the
grain contacts. The overall macrofabric can be de- major principle stress. An isotropic (hydrostatic) con-
scribed as bulky granular. solidation stress produces an isotropic fabric, provided
Directional, preferred orientation in Vicksburg (Mis- the fabric was isotropic at the start of consolidation.
sissippi) loess was observed and described by Matal- Soil compaction can be done using different meth-
ucci et al. (1969). The long axes of grains concentrated ods, including impact, kneading, vibratory, and static.
hte
in an azimuth direction of 285 to 289, with an incli- The method used and the initial state of the soil can
nation of 3 to 8. A prevailing wind direction of 290 have profound effects on the fabrics of both sands and
at the time of deposition was deduced from the thin- clays and on the properties of the compacted soil. In
ning pattern of the loess in the area, thus accounting clays, the water content is important; it controls the
for the observed three-dimensional anisotropy. ease with which particles and particle groups can be
rig
types. The presence of small, platy quartz particles de- (static compaction) does not penetrate the soil, as is
rived by glacial grinding was noted earlier. Many silty usual for compaction dry of optimum water content,
and sandy ablation tills have a multimodal grain size then there may be a general alignment of particles or
distribution, with coarser particles distributed through particle groups in horizontal planes. If the soil is suf-
Co
a fine-particle matrix (McGown, 1973). The fabric of ficiently wet of optimum that the compaction rammer
the matrix is variable. Many fabric forms are similar penetrates the soil surface as a result of a bearing ca-
to those observed in collapsing soils (Barden et al., pacity failure under the rammer face, there is an align-
1973). ment of particles along the failure surfaces. A series
Boulder clays differ from soft, sedimentary clays in of such zones is developed as a result of successive
that they contain a wider range of grain sizes, with rammer blows, and a folded or convoluted fabric may
some particles extending into the gravel to boulder result, as shown, for example, by Fig. 8.4.
ranges, and they are much denser. Many boulder clays
have been subjected to high vertical and tangential Effects of Postformational Changes
stresses as a result of readvancing ice sheets. Poor sort- As listed in Fig. 8.2, a large number of postformational
ing and the presence of a large number of different factors can modify the initial structure of a soil.
l
ria
ate
dM
hte
rig
py
Co
Figure 8.3 Fabric and particle orientation in Portsea Beach sand (Lafeber and Willoughby,
1971). (a) Vertical cross section (perpendicular to the coastline) where B is the dip direction
of bedding plane, H is the horizontal plane, and I is the imbrication plane. (b) Distribution
of long axis orientations.
l
particle contacts. However, in some soils that possess
ria
bonding and cementation in their initial states, consol-
idation stresses greater than some critical value can
break down the structure, thus causing weakening and
collapse.
Temperature Transformations of structure associ-
ate
ated with leaching, precipitation, cementation, weath-
Figure 8.4 Microfabric of Takahata kaolin compacted wet ering, and pressure increase develop more rapidly at
of optimum using impact compaction ⫻1000. Reprinted from high temperatures than at low temperatures.
Yoshinaka and Kazama (1973) with permission of The Jap- Shearing Shearing collapses some structures,
anese Society of SMFE. dM whereas in others, such as heavily overconsolidated
clay, it may change the structure significantly only in
the immediate vicinity (a few millimeters) of the shear
Time Chemical diffusion and chemical reactions plane.
are time dependent. Following deposition, remolding, Unloading Stress relief as a result of unloading can
or compaction, the interparticle forces, and therefore allow elastic rebound of particles and particle groups
the mechanical properties, can also change simply as and the onset of swelling. Some very stiff materials
a result of pore pressure redistribution in the new en- may split and/or spall after unloading.
hte
1. Move particles.
2. Cause compression due to seepage forces. 8.3 RESIDUAL SOILS
rig
Weathering In the zone of weathering, some ma- with these materials has been in areas of temperate
terials are broken down and others are formed. climate. However, the need for knowledge and under-
Changes in pore water chemistry influence the inter- standing of the engineering behavior of tropical resid-
particle forces and flocculation–deflocculation tenden- ual soils is great, owing to the extensive construction
cies. Weathering can disrupt the initial soil fabric. worldwide in areas covered by these soils.
Cyclical wetting and drying and freezing and thaw- Residual soils differ from sedimentary soils in that
ing disrupt weak particle assemblages and intergroup they have formed in place in response to the local
associations. Wetting generally means weakening and parent material, climate, topography, and drainage
may lead to collapse of some structures, particularly conditions. They may retain elements of the parent
those with open fabrics where particles are only material structure; they are usually nonuniform and
l
opment of a single engineering classification system
ria
that has universal applicability is unlikely.
Tropical Soils
In regions of high temperature and abundant rainfall,
ate
rock weathering is intensive and is characterized by
the rapid breakdown of feldspars and ferromagnesian
minerals, the removal of silica and bases (Na2O, K2O,
MgO), and the concentration of iron and aluminum
oxides. This process is termed laterization (Gidigasu,
dM
1972; Grant, 1974; and others) and involves leaching
of SiO2 and deposition of Fe2O3 and Al2O3. A laterite
is a soil whose ratio of SiO2 to Al2O3 is less than 1.33,
whereas a lateritic soil has a ratio between 1.33 and
2.00 (Bawa, 1957).
With abundant rainfall, high temperature, good Figure 8.5 Schematic diagram of a typical tropical residual
drainage, and crystalline parent materials, feldspars soil profile (from Little, 1969).
hte
Allophane as a clay mineral type is described in Continued mechanical working or the removal of
Chapter 3. The term allophane soil is also used to refer sesquioxides from such soils can result in sig-
to andisols. They occur commonly in the Caribbean, nificant changes in properties. The effects of
the Andes, and the Pacific areas of the United States, remolding and sesquioxide removal on the
Indonesia, Japan, and New Zealand. A comprehensive classification properties of a lateritic soil are
presentation of the structure and properties of allo- shown in Table 8.1.
phane soils is given by Maeda et al. (1977) and Wesley 2. Air drying may cause clay size particles to form
(1977). aggregates of silt and sand size and a loss of
A typical deep weathering profile in the tropics is plasticity, as shown by the data in Table 8.2. The
shown schematically in Fig. 8.5. Boundaries between significant decrease in plasticity that resulted
Sesquioxide
Property Unremolded Remolded Free
Liquid limit (%) 57.8 69.0 51.3
Plastic limit (%) 39.5 40.1 32.1
Plasticity index (%) 18.3 28.0 19.2
Specific gravity 2.80 2.80 2.67
Proctor density 13.3 13.0 13.8
l
(kN/m3)
ria
Optimum moisture 35.0 34.5 29.5
content (%)
From Townsend et al. (1971).
ate
Table 8.2 Effect of Air Drying on Index Properties of a Hydrated Laterite Clay from the Hawaiian Islands
Wet
dM (at Natural Moist Dry
Moisture (Partial Air (Complete
Index Properties Content) Drying) Air Drying) Remarks
Sand content (%) 30 42 86 Dispersion prior to hydrometer test
with sodium silicate
Silt content (%) 34 17 11
(0.05–0.005 mm)
hte
from drying a number of different tropical soils 6. Soils in tropical areas exist at water contents
is shown in Fig. 8.6. higher than those that are desirable for most
py
3. Drying may cause hardening, and this hardening earthwork construction. As a result, difficulties in
may be irreversible in some cases. British Stan- soil handling and compaction are common.
dard BS1377 (1990) recognizes the irreversible The yielding and strength of residual soils reflect their
changes that occur during drying and recom- bonded structure. The preconsolidation pressure may
Co
mends that tropical residual soils be tested in have no connection with the stress history or overbur-
their natural state wherever possible. den pressure on the soil. Typical preconsolidation pres-
4. The compacted dry density, plasticity index, and sure values of residual soils are given in Table 8.3.
compressibility of tropical residual soils are After yielding, residual soils exhibit large compressi-
likely to be less than the values for temperate bility as a result of structure degradation and particle
soils of comparable liquid limit. On the other breakage. A relationship between compression index
hand, the strength and permeability may be and in situ void ratio for several soils is given in Fig.
higher. 8.7. Extensive discussion on the mechanical behavior
5. Tropical residual soils commonly are heteroge- of residual soils in relation to their bonded structure is
neous in structure and texture. given by Vaughan (1988).
l
ria
Figure 8.6 Effect of drying on the Atterberg limits of some tropical soils (from Morin and
Todor, 1975).
ate
Table 8.3 Yield Stresses of Various Residual Soils
a)
ine
Obtained from Odometer or K0 Triaxial Tests
Gu
w
Ne
dM
Soil Type and Location Yield Stress (kPa) 1.5
a
ap
Soft Clay (Canada) Sensitivity 8
(P
Sensitivity 4
ite
Compression Index, Cc
ys
llo
Seprolitic (Sea USA)
New Guinea 100–350
Ha
Tuccarul, various
d
Volcanic clay 110–270
an
1.0 (Brazil)
ne
Gneiss, basalt, and sandstone, From Basalt
ha
(Brazil)
lop
Brazil 60–450
Al
Granite, basalt, and sandstone,
hte
Zealand 200–500
1.0 2.0 3.0 4.0
After Fookes (1997). In Situ Void Ratio
Granitic rock weathers in general accordance with als that are coarse and uniformly graded, and for
Bowen’s reaction series. Biotite decomposes first, highly angular particles and particles with high intra-
followed by plagioclase feldspar. When part of the granular void content. Consequences of this may in-
plagioclase has decomposed and breakdown of the clude substantial reductions in the peak frictional
orthoclase begins, the rock breaks into fragments of strength with increasing confining pressure. For ex-
decomposed granite called gruss. When most of the ample, Yapa et al. (1995) found a reduction in friction
orthoclase has weathered to kaolinite, the gruss crum- angle of 25 percent in densely compacted specimens
bles to silty sand, which typically contains mica flakes. and about 15 percent in loose specimens over the con-
Apart from some mechanical breakdown, the quartz fining pressure range from 100 to 1500 kPa. Friction
fragments remain unchanged. angles assigned to decomposed granites used in 12 em-
Decomposed granite profiles generally contain four bankment dam fills constructed in California in the
l
zones as shown in Fig. 8.8. The deepest zone consists 1960s were conservatively selected and ranged from
ria
of angular granitic blocks. The amount of residual de- 29 to 38. Compaction to greater than 90 percent mod-
bris is small, although the rock may be relatively ified Proctor maximum relative compaction at opti-
highly altered. The next zone above contains abundant mum water content is recommended to minimize
angular to subangular core stones in a matrix of gruss settlement due to postconstruction hydrocompression
and residual debris. The upper middle zone is the most when the fill is wetted.
ate
variable part of the weathering profile and typically During the 1995 Kobe, Japan, earthquake, many re-
contains about equal amounts of rounded core stones, claimed land sites in Kobe liquefied extensively. The
gruss, and residual debris. The topmost zone usually soil used for reclamation was decomposed granite
consists of an unstructured mass of clayey sand with called Masado, which is a well-graded material with
a highly variable grain size distribution.
dM particles ranging from gravels to fines. The liquefac-
Construction can be difficult in areas underlain by tion of this soil was surprising because of its higher
decomposed granite. The bedrock profile is highly ir- uniformity coefficient and greater dry density than
regular, and competent bedrock may be located at vari- sandy soils. The weak and crushable character of Ma-
able depths below the ground surface. The core stones sado particles is considered to be one of the causes.
can present significant obstacles to excavation. Seem- The undrained cyclic shear strength of the decomposed
ingly sound pieces of rock and gravel break down granite was found to be much smaller than that of a
when excavated or used in earthwork construction. The gravelly soil that had a similar particle size distribution
hte
presence of mica may cause cohesionless soils com- but with strong particles (Kokusho et al., 2004).
posed of decomposed granite to be highly compressi-
ble. Colluvial Soils
Decomposed granite can be used successfully as an Colluvium is soil that has formed in place but subse-
embankment fill material provided it is remembered quently has been transported down slope by gravity.
that particles may undergo substantial breakage under
rig
Pyritic Soils
Pyrite (FeS2) bearing rocks and soils are responsible
for foundation heave, concrete degradation, steel cor-
rosion, environmental damage, acid drainage, acceler-
ated weathering of rock, and loss of strength and
stability of geomaterials. Sulfur occurs in rock and soil
in the forms of sulfide (S⫺ or S2⫺), sulfate (SO42⫺), and
Figure 8.8 Zones of a mature profile of decomposed granite. organic sulfur. The amount of sulfide sulfur (also
l
illary zones and localize along discontinuities due to properties. These soils are of particular importance in
ria
reduced confining stress in these regions. Volume in- highway, airfield, and land development projects. Sur-
crease from the growth of sulfate minerals along bed- face soils are classified so that they can be aggregated
ding planes is a dominant factor in the vertical heave into categories that are useful for understanding gen-
that occurs in shale and other layered materials. The esis, properties, and behavior, especially in relation to
production of sulfates by pyrite oxidation also in- agriculture. All soils in the United States (more than
ate
creases the potential for further deleterious reactions, 11,000 in 1980) and numerous soils in other countries
such as the formation of gypsum (CaSO4 2H2O) and have been classified according to soil taxonomy (Soil
other expansive sulfate minerals (e.g., ettringite). Survey Staff, 1975).
Pyrite oxidation processes proceed in the following Soil taxonomy is a multicategory system of soil clas-
way: dM sification that includes 10 orders, about 47 suborders,
200 great groups, 1000 subgroups, 2000 families, and
10,000 series. Unlike most classification systems, each
FeS2 ⫹ –72 O2 ⫹ H2O → Fe2⫹ ⫹ 2SO42⫺ ⫹ 2H⫹
category of soil taxonomy carries elements of the
Fe2⫹ ⫹ –14 O2 ⫹ H⫹ → Fe3⫹ ⫹ –21 H2O higher category so that when a soil is classified at the
family level, the family name indicates the order, sub-
Fe3⫹ ⫹ 3H2O → Fe(OH)3 ⫹ 3H⫹ order, great group, and subgroup to which the soil
belongs. The soil family name also may contain infor-
FeS2 ⫹ 14Fe3⫹ ⫹ 8H2O → 15Fe2⫹
hte
H2SO4 ⫹ CaCO3 ⫹ H2O → CaSO4 2H2O ⫹ CO2 imperfectly weathered, consolidated rocks. Entisols
include some recent, young soils formed in poorly
and is accompanied by very large volume increases, as drained areas. In general, geotechnical engineers en-
the products of pyrite oxidation reactions are signifi- counter these soils more than any other because large
Co
cantly less dense than the initial sulfide (pyrite). Pyrite, construction activities tend to concentrate in areas
of specific gravity (Gs ⫽ 4.8–5.1), reacts with calcite where these soils accumulate, such as in river valleys
(Gs ⫽ 2.7) to create gypsum (Gs ⫽ 2.3) (Hawkins and and in areas bounded by water. The majority of large
Pinches, 1997). urban areas are located in such regions. To understand
Mitigation options that are useful for preventing or the characteristics of these soils requires consideration
reducing sulfide-induced problems include controlling of transportation, deposition, and postdepositional sed-
the pyrite oxidation process, use of restraining forces imentary processes. These topics are considered in
to prevent ground movement, design measures that al- Section 2.8.
low for movement, and removal or neutralization of Vertisols (inverted soils) are deep and clayey and are
acid. A recent review of geotechnical problems, heave known also as black cotton, black earth, and blackland
soils. They are associated with a climate that has very deep. A relatively thick B horizon may be brightly col-
dry and very wet seasons. The texture of all horizons ored (red and yellow) as a result of oxidation and hy-
is clayey, and the dominant clay mineral is smectite. dration of iron. The B horizon has more than twice the
The soils are expansive. clay content of the A horizon. The cation exchange
Inceptisols, or new soils, include tundra and selected capacity is low in all horizons, and the clay fraction is
soils of marshes, swamps, and flat areas. Tundra is a composed mainly of kaolinite, illite, and quartz. Many
dark gray, peaty accumulation over gray mottled min- lateritic soils of subtropical regions are ultisols.
eral horizons. The soil is poorly drained and boggy. Oxisol is an iron oxide and aluminum oxide-rich,
The clay mineral content is low. Permafrost (perma- highly weathered clayey material that changes irre-
nently frozen soil) is frequently present in the substra- versibly to concretions, hardpans, or crusts when de-
tum. Humic-gley inceptisols are mineral soils formed hydrated. Clay minerals are rapidly broken down and
l
in poorly drained areas that possess a sticky, compact, removed. What little clay remains is usually kaolinitic.
ria
gray, or olive-gray B or C horizon. The A horizon may Deposits of these soils may be up to 30 m or more in
contain 5 to 10 percent organic matter. depth and may range in texture from friable soils to
Aridisols (arid soils) are characterized by surface ac- hard rock. Some oxisols are strong and resistant to
cumulations of salts from upward movement of water, breakdown; however, others may lose their granular
characteristics when worked, becoming soft, clayey,
ate
and usually consist of several centimeters of soil over
a calcareous parent material. The soils may be alkaline, and impervious. Most laterites of the tropics are oxi-
with high concentrations of soluble salts of calcium, sols.
magnesium, and sodium near the surface. Illite and Histosols, or organic soils, are bog soils whose char-
smectite are common in these soils. acteristics depend largely on the nature of the vegeta-
Mollisols generally form in cool areas having annual
dM tion from which they form.
rainfall of 400 to 650 mm. They typically have a dark An 11th order, andisols was also proposed to ac-
A1 horizon, and the horizon boundaries are indistinct. commodate the soil developed from volcanic ash.
Smectites predominate in the clay fraction over illite.
There may be local accumulations of sepiolite, paly-
gorskite, and attapulgite, and calcium salts may be 8.5 TERRESTRIAL DEPOSITS
present.
Aeolian Deposits
Spodosols are found south of the tundras in areas
hte
where rainfall exceeds 600 mm/yr, and summers are Of the various sediment transporting agents, wind is
short and cool. Spodosols are characterized by mod- the only one that can move material uphill for any
erate humus accumulation, a thin A1 horizon, and a distance. Wind is most easily able to move sand. It is
strongly eluviated A2 horizon. The B horizon is dark not a universal agent of erosion, as its effects are re-
brown to reddish brown and often cemented by organic stricted to areas of a particular climate such as deserts
rig
compounds and iron oxides. The texture of all horizons or to specific places such as beaches and plowed fields.
except O is often sandy. The soils are acid, have a low The load suspended by the wind, which is composed
cation exchange capacity, and illite dominates the clay primarily of silt-size particles, is carried high above the
fraction. ground and may be transported for great distances. The
Alfisols are found south of the spodosol region and bed load, moved by saltation and traction, moves
py
east of the prairies in northeastern United States and slowly and as a unit.
southeastern Canada and in the humid, temperate areas Deposition from wind occurs with reduction in wind
of western Europe and eastern Asia, where rainfall av- velocity. Consequently, accumulations are found in the
erages 750 to 1300 mm annually. These soils are char- lee of desert areas. Coarser particles of sand, carried
Co
acterized by a thin A1 horizon (50 to 150 mm) and a by saltation and traction, pile in dunes with their long
well-developed gray to yellowish A2 horizon. The B axis parallel to the wind. Loess deposits, composed of
horizon is gray to reddish brown, darker, and of finer silt-size particles, are of particular interest because of
texture than either the A or C horizons. They are acid their unique structure and properties and are described
soils, and kaolinite is the dominant clay mineral. more fully in Section 8.16.
Ultisols are found in areas of high temperature and
high rain (1000 to 1500 mm/yr). Leaching is great, Glacial Deposits
and mineral decay is rapid. Surface accumulation of Several types of deposit form from glacial melting, as
organic matter is small, and the leached A horizon is listed in Table 8.4. Moraines are dropped directly from
l
d. Subglacial moraine—material at sole of glacier
ria
e. Ground moraine—deposited subglacial moraine
f. Terminal or end moraine—ridge of deposits built up at end of glacier
g. Recessional moraine—terminal moraine of receding glacier
2. Drumlins
Mounds of boulder clay formed under deep ice
ate
II. Glacio-Fluvial Deposited Material
A. Sediments
1. Coarse gravel to clay, progressively sorted dams and deltas
2. Crudely bedded gravel and sand in kames and eskers
B. Structures dM
1. Alluvial fans for glaciers terminating on land
2. Outwash plains merged with fans
3. Deltas for glaciers terminating in standing water
4. Kettle holes caused by melting of stranded ice blocks
5. Kames—mounds of crudely bedded sand and gravel caused by stream from melting ice
6. Esker—winding ridge of sand and gravel from meltwater stream in ice tunnel or from receding ice
III. Glacial Lake Deposited Material
hte
A. Sediments
1. Sands to clay
2. Poor sorting and stratification of channel deposits
3. Excellent stratification of lake floor deposits
B. Structures
1. Overflow channels where lake water escaped
rig
the melting ice. There are several types of moraine, sorted gravel and sand deposits. Many lateral moraines
depending on where the material is dumped relative to and dead ice deposits are mixed glacial and glacio-
the ice mass, as indicated in the table. Moraines usu- fluvial deposits.
Co
ally contain a wide range of unsorted particle sizes, Glacial lake deposits are quiet water sediments that
and the material is known as till. When large amounts are usually composed of fine-grained materials. Varved
of boulders and clay are present, the deposit is referred clay is an example (see Fig. 2.13). The formation and
to as boulder clay. Some glacial moraines are densely characteristics of varved clay are discussed in Section
compacted owing to compression under advancing ice 2.8.
masses. The characteristics of a specific glacial deposit de-
Glacio-fluvial deposits are transported from the pend on the type and erodability of the parent material,
melting point by flowing meltwater; kames and eskers the type and distance of transportation, gradients, and
(Fig. 8.9) are examples. Kames and eskers are poorly pressures. For example, bottom moraines are usually
l
stages of glaciation led to scouring of a valley beneath
ria
the present floodplain surface. Rising sea at the end of
the glacial period resulted in deposition of sands and
gravels in the bottom of the valley followed by finer
material above. In the 25,000 years since the last gla-
ciation, the Mississippi River has changed from an
ate
overloaded, shallow, braided stream to a deep, single-
Figure 8.9 Glacio-fluvial sediments (Selmer-Olsen, 1964). channel, and meandering river.
The variety of deposits found within the Mississippi
River Valley is great, and their interrelationships are
dM complex; however, each can be accounted for in a log-
finer grained and more consolidated than lateral or end ical way in terms of the factors governing its deposi-
moraines. Finely ground (silt and clay size) rock flour tion and history, as described by Kolb and Shockley
is produced by the grinding action of the ice and may (1957).
be a major constituent of postglacial marine and lake The coarser materials were laid down initially in the
clays found in Canada and Scandinavia. bottom of the valley. Occasional lenses of clay, sandy
More extensive and detailed information on glaciers silt, and silty sand are found in these substratum de-
and the characteristics of glacial deposits can be found posits. The depths to these materials vary from about
hte
in Leggett and Hatheway (1988) and West (1995), 3 m in the north to 30 m in the southern part of the
among many other texts and references. river, and the thickness varies from 15 to 125 m in the
same direction.
Braided stream deposits are usually remote from
Alluvial Deposits
present large streams. Most are relatively dense, sandy
Alluvial deposits form from pluvial (high rain area) silts and clayey sands. Natural levees rise to 5 m or
rig
and fluvial (river) deposition and are generally char- more above the floodplain and decrease in grain size
acterized by laterally discontinuous, lenticular beds away from the crest and in a downstream direction.
that are oriented downstream and have different parti- Point bar deposits composed of silts and silty sands
cle size characteristics. Gravels are often in contact form on the inside of river bends during high-water
with sand and silt.
py
energy, and all particles larger than a certain size are of the river, left behind as oxbow lakes, fill with weak
dumped in a jumble of large and small particles. The and compressible clay and silty clay layers with thick-
flow then slips to one side following the steepest slope. nesses up to 30 m or more. Abandoned river courses
The channel may subsequently fill, and the flow shifts many miles long fill with materials similar to those of
again. When this process occurs at the base of a slope, the oxbow lakes.
the result is an alluvial fan, a temporary feature that is Medium- to high-plasticity clays, often organic,
a symmetrical pile of material spread out radially from termed backswamp deposits, form in shallow areas
the point of slope change. during flood stage. Because of desiccation between pe-
In advanced stages of stream development, the riods of deposition, they have water contents lower
stream occupies only a small part of a broad, flat val- than the abandoned channel deposits.
l
erally continuous and thick clay beds. The Corcoran mud in the seaward delta face, which is later covered
ria
clay, which covers an area of about 15,000 km2 in by alluvial, lacustrine, and beach deposits as the delta
California’s San Joaquin Valley, forms an extensive grows.
confining bed and aquiclude in the valley and is a sig- The complex formations of the Mississippi River
nificant feature influencing groundwater development. delta reflect the composite effects of the advancing
delta and the encroaching sea. Pleistocene sediments
ate
Paludal, or swamp, deposits usually consist of plas-
tic silts, muds, and clays with high water content and consisting of dense clays, sands, and gravels underlie
organic matter. Difficult problems may be associated the delta. Sand and shell beaches, often 5 m high or
with these deposits because of their low strength and more, are among the most suitable deltaic formations
high compressibility and from the formation of marsh for foundation support. Conversely, difficult geotech-
gas. dM nical problems are associated with fine-grained and or-
ganic delta sediments because of their low strength and
high compressibility.
8.6 MIXED CONTINENTAL AND MARINE
DEPOSITS 8.7 MARINE DEPOSITS
Littoral Deposits An averaged and idealized profile through the marine
hte
Littoral deposits form in the tidal zone and consist of environment is shown in Fig. 8.10. The continental
tidal lagoon, tidal flat, and beach sediments. Lagoon shelf extends from low tide to an average water depth
sediments include fine-grained sands and silts in the of about 130 m (nearly 450 ft). The steeper continental
channels and organic-rich silt and clay in the quiet ar- slope (average of 4 leads down to the more gently
eas. Organic matter and carbonates may be abundant. sloping continental rise. The average water depth in
Tidal flat deposits consist of fine-grained dark muds, the deep ocean is more than 3500 m (11,500 ft).
rig
with lenses or stringers of sand and gravel, and are There are three main types of marine sediments:
free of intermediate-size sediments. Beach deposits lithogenous (derived from terrestrial, volcanic, or cos-
consist of clean fine- to coarse-grained sand with oc- mic sources), biogenous (remains of marine organ-
casional stringers of gravel. isms), and hydrogenous (precipitates from the seawater
or interstitial water). An engineering classification sys-
py
l
ria
ate
dM
hte
Figure 8.10 Idealized profile of the continental margin, with vertical exaggeration (after
Heselton, 1969).
rig
bulence decreases. Generally there is a decrease in sediments are (Semple, 1988) that they are composed
particle size and increased influences of biological and of weak, angular particles, particle sizes and size dis-
chemical factors in the seaward direction, although the tributions are variable, there is uneven cementation
sediment distributions may be irregular due to tidal over short distances, and they have high void ratio rel-
Co
currents and seasonal climatic variations. Neritic de- ative to silicate sediments. As a result, these materials
posits reflect sediment source areas and climatic con- may be the source of special geotechnical problems.
ditions, with sandstone, shale, and limestone typically For example, the side friction developed on driven
present in shelf areas. With the exception of the bi- piles in calcareous sands is often very much lower than
ogenous sediments, the physical properties of conti- anticipated based on the behavior of piles in quartz
nental shelf deposits are essentially the same as those sand (Noorany, 1985; Murff, 1987; Jewell et al., 1988).
of comparable terrestrial soils.
Calcareous Sands Calcareous bioclastic sands are Bathyal Deposits
formed from the skeletal remains of corals, shells of The bathyal environment includes the continental slope
mollusks, and algae. They are widely distributed in the and the continental rise. Bathyal sediments are typi-
Figure 8.11 Chart for classification of marine sediments (from Noorany, 1989). Reprinted
with permission of ASCE.
211
Abyssal Deposits
Deep-ocean (abyssal) deposits consist primarily of
brown clays and calcareous and silicious oozes, with
thicknesses of 300 to 600 m. Terrigenous deposits are
derived from land, whereas pelagic sediments settle
from the water alone and contain the shells and skeletal
l
ria
remains of tiny marine organisms and plants. Accu-
mulation rates range from less than a millimeter per
thousand years in the deep sea to a few tens of centi-
meters per year in near-shore areas close to the mouths
of large rivers (Griffin et al., 1968). Oozes contain
ate
more than 50 percent biotic material.
Calcareous ooze, composed of empty shells or tests,
covers about 35 percent of the seafloor for water
depths up to about 5 km. It is usually nonplastic, cream
to white in color, and composed of easily crushed
sand- to silt-size particles. Brown clay is found beneath
dM
Figure 8.12 Electron photomicrographs of calcareous sand most of the deeper ocean areas. Its origin is believed
from Guam. Magnification is 45⫻ (courtesy of I. Noorany). to be atmospheric dust and fine material circulated by
ocean currents. About 60 percent of this material is
finer than 60 m, and the clay fraction contains chlo-
rite, smectite, illite, and kaolinite, with illite often the
cally fine sand, silt, and mud of high water content and most abundant. Brown clays have high water contents,
hte
low shear strength. The tectonic setting of the depo- moderate-to-high plasticity, and low strength. Siliceous
sitional area and the characteristics of the continental ooze, composed of plant remains, is found mainly in
source materials largely control the distribution, ge- the Antarctic, northeast of Japan, and in some areas of
ometry, and properties of these sediments. the equatorial Pacific.
Erosion, transport, and deposition of these sediments Except near their surface, deep-sea deposits are nor-
may be caused by the frictional effects of contour- mally consolidated and highly compressible. There is
rig
following undercurrents that result in thick sequences an apparent overconsolidation of the near-surface ma-
of sediment ‘‘drift’’ consisting of alternating thin layers terial at many locations. This evidently reflects bond-
of very fine sands, silts, and muds (Leeder, 1982). Ap- ing developed as a result of the extremely slow rate of
preciable quantities of sediments can be transported deposition and physicochemical effects (Noorany and
from the continental slope and rise to the deep-ocean Gizienski, 1970). Much of the available data on the
py
abyssal plains by slumps, debris flows, and turbidity mechanical properties of deep-seafloor soils pertains to
flows. material from the upper 6 m.
Detailed exploration of the ocean margins indicates
that debris flows are probably a much more important
Co
depositional process on the seafloor than has been pre- 8.8 CHEMICAL AND BIOLOGICAL DEPOSITS
viously suspected. For example, debris flow deposits
of enormous extent have been identified that were gen- Evaporite deposits formed by precipitation of salts
erated by large sediment slides on the northwestern from salt lakes and seas as a result of the evaporation
African continental margin. The flow traveled on a of water are sometimes found in layers that are up to
slope as flat as 0.1 for a distance of several hundred several meters thick. The major constituents of sea-
kilometers. The deposits cover an area of about 30,000 water, their relative proportions, and some of the more
km2 and originated from a massive slump of about 600 important evaporite deposits are listed in Table 8.5. In
km3 on the upper continental rise where a prominent some areas alternating layers of evaporite and clay or
Percent by Weight
Ion Grams per Liter of Total Solids Important Evaporite Deposits
Sodium, Na⫹ 10.56 30.61 Anhydrite CaSO4
Magnesium, Mg2⫹ 1.27 3.69 Barite BaSO4
Calcium, Ca2⫹ 0.40 1.16 Celesite SrSO4
Potassium, K⫹ 0.38 1.10 Kieserite MgSO4 H2O
Strontium, Sr2⫹ 0.013 0.04 Gypsum CaSO4 2H2O
Chloride, Cl⫺ 18.98 55.04 Polyhalite Ca2K2Mg(SO4 ) 2H2O
Bloedite Ma2Mg(SO4)2 4H2O
l
Sulfate, So42⫺ 2.65 7.68
Hexahydrite MgSO4 6H2O
ria
Bicarbonate, HCO3⫺ 0.14 0.41
Epsomite MgSO4 7H2O
Bromide, Br⫺ 0.065 0.19 Kainite K4Mg4(Cl/SO4 ) 11H2O
Fluoride, F⫺ 0.001 — Halite NaCl
Boric Acid, H3BO3 0.026 0.08 Sylvite KCl
34.485 100.00 Flourite CaF2
ate
Bischofite MgCl2 6H2O
Carnallite KMgCl3 6H2O
Adapted from data by Degens (1965).
dM
other fine-grained clastic sediments are formed during about 50 kPa, owing to the roughness and interlocking
cyclic wet and dry periods. of the diatoms, but becomes more compressible under
Many limestones have been formed by precipitation higher stresses owing to crushing of the diatoms (Day,
or from the remains of various organisms. Because of 1995).
the much greater solubility of limestones than of most
other rock types, they may be the source of special
hte
problems caused by solution channels and cavities un- 8.9 FABRIC, STRUCTURE, AND PROPERTY
der foundations. RELATIONSHIPS: GENERAL CONSIDERATIONS
More than 12 percent of Canada is covered by a
peaty material, termed muskeg, composed almost en- The variety of possible soil fabrics and the many pos-
tirely of decaying vegetation. Peat and muskeg may sible interparticle force systems associated with each
have water contents of 1000 percent or more, they are mean that the potential number of soil structures is
rig
very compressible, and they have low strength. The almost limitless. The mechanical properties of a soil
special properties of these materials and methods for reflect the influences of the structure to a degree that
analysis of geotechnical problems associated with depends on the soil type, the structure type, and the
them are given by MacFarlane (1969), Dhowian and particular property of interest. The effects of structure
Edil (1980), and Edil and Mochtar (1984). can be of equal importance to those of initial void ratio
py
Chemical sediments and rocks in freshwater lakes, and stress. In this sense, structure refers to the differ-
ponds, swamps, and bays are occasionally encountered ences between the actual void ratio and effective stress
in civil engineering projects. Biochemical processes and the corresponding values for the same soil in the
form marls ranging from relatively pure calcium car- destructured state. The difference between void ratio
Co
bonate to mixtures with mud and organic matter. Iron under a given effective stress for a soil with some
oxide is formed in some lakes. Diatomite or diatoma- structure, which is the case for consolidation of virtu-
ceous earth is essentially pure silica formed from the ally all sediments from a high void ratio, and the void
skeletal remains of small (up to a few tenths of a mil- ratio of a completely destructured soil is illustrated in
limeter) freshwater and saltwater organisms. Com- Fig. 8.13.
pacted fills of diatomaceous earth can have very low It is possible that a soil can be at state to the right
dry unit weights (1.0 to 1.2 Mg/m3) and high moisture of the virgin compression curve in Fig. 8.13 as a result
contents (40 percent or more). The material may be- of bonding by chemical cementation or aging effects.
have as a dense granular material at stresses below Thus the full range of possible states in void ratio–
l
ria
ate
dM
Figure 8.13 The influence of metastable fabric on void ratio under and effective consoli-
dation pressure.
hte
rig
py
Co
l
Figure 8.15a shows one-dimensional compression
ria
curves for various reconstituted clays with a wide
range of plasticities. The void index, was proposed by
Burland (1990) for correlating the compression behav-
ior of different clays and for assessing the influence of
structure on properties. The void index Iv is defined as
ate
e ⫺ e*
Iv ⫽ 100
(8.1)
C*c
dM
in which e is the void ratio, e*100 is the ‘‘intrinsic’’ void
ratio under an effective vertical stress of 100 kPa in
the one-dimensional odometer test, and C* c is the in-
trinsic compression index. The intrinsic properties are
determined for a reconstituted samples of clay that
have been prepared at a water content of about 1.25
times the liquid limit. The intrinsic compression curves
hte
termed the sedimentation compression line (SCL), are Figure 8.15 (a) One-dimensional compression curves for
shown in Fig. 8.16b along with the intrinsic compres- several clays. (b) Normalized compression curves defining
sion curve, termed the intrinsic compression line the intrinsic compression line (ICL) (from Burland, 1990).
(ICL).
Co
l
ria
ate
dM
Figure 8.16 (a) Compression curves for several clays (from Skempton, 1970). (b) Normal-
ized compression curves for clays in (a) showing the intrinsic compression line (ICL) and
sedimentation compression line (SCL).
hte
1. Under a given effective consolidation pressure, particles and particle groups may be essentially
a soil with a flocculated fabric is less dense than stress free as a result of arching by surrounding
the same soil with a deflocculated structure. fabric elements, as discussed further in Chapter
rig
causes a greater change in fabric of a flocculated at point a on the virgin compression curve can
soil structure than in a deflocculated soil struc- deform to point b as a result of disturbance and
ture. reconsolidation or by secondary compression
4. The average pore diameter and range of pore under stress a stained for a long time. A sam-
Co
sizes is smaller in deflocculated and/or destruc- ple initially at c can reach point b as a result of
tured soils than in flocculated and/or undis- unloading from c. The stress–deformation
turbed soils. properties of the two samples will differ. The
5. Shear displacements usually orient platy parti- overconsolidation ratio (OCR), defined as the
cles and particle groups with their long axes ratio of the maximum past consolidation effec-
parallel to the direction of shear. tive stress to the present overburden effective
6. Anisotropic consolidation stresses tend to align stress is a good measure of stress history. The
platy particles and particle groups with their OCR of sample 2 in Fig. 8.19 is c / a.
long axes in the major principal plane. 9. Volume change tendencies determine pore pres-
7. Stresses are usually not distributed equally sure development during undrained deforma-
among all particles and particle groups. Some tion.
l
ria
ate
dM
hte
10. Changes in structure of a saturated soil at con- the rate of deformation is slow or positive pore pres-
stant volume are accompanied by changes in ef- sures will be generated if deformation is rapid. On the
fective stress. These effective stress changes are other hand, if the soil is initially at a state in the di-
py
water to drain from or enter the soil. whereas heavily overconsolidated clays and dense
sands are dilative.
Figure 8.20 illustrates points 9, 10, and 11. For any
saturated, destructured soil there is a unique relation-
ship between combinations of void ratio and effective 8.10 SOIL FABRIC AND PROPERTY
consolidation pressure termed the critical state or ANISOTROPY
steady state line, as discussed in more detail in Chapter
11. If the soil is on this line, there is no tendency for Anisotropic consolidation, shear, directional transpor-
change in volume during shear deformation. However, tation components, method of remolded or compacted
if the state of the soil is in the region above and to the soil preparation, and compaction of soil in layers each
right of the critical state line it will either contract if may produce anisotropic fabrics. Fabric anisotropy on
l
ria
ate
dM
Figure 8.17 Compression curves for freshwater glacial lake clay at pressures below and
above yield (from Burland, 1990).
hte
rig
py
Co
Figure 8.18 Void index in relation to stress states for dif- Figure 8.19 Illustration of different paths to reach the same
ferent clay types (from Chandler et al., 2004). present void ratio–effective stress state.
l
ria
ate
Figure 8.20 Initial state in relation to the critical-state or steady-state line and its influence
on pore pressure or volume changes during deformation.
dM
a macroscale usually leads to mechanical property an- the finding that as the density increased the intensity
isotropy, and the property differences in different di- of preferred orientation decreased. The sample stiff-
rections may be significant. Examples of anisotropic ness, as measured by the ratio of stress to shear dis-
fabrics in sands are given in Figs. 5.9 and 5.10. placement at 50 percent of peak strength, was about
Some examples are presented in this section to il- twice as high for shear across the direction of preferred
lustrate the general nature and magnitudes of aniso- orientation than parallel to it.
hte
tropy in properties that may be associated with a Figure 8.22 shows the variation in friction angle as
homogeneous anisotropic fabric. These considerations a function of the loading direction in plane strain and
are separate from property anisotropy caused by strat- triaxial compression in relation to the initial bedding
ification of different soil layers, although the latter may plane measured on dense Toyoura sand specimens
be very important in the field, especially with respect (Park and Tatsuoka, 1994). The term is the angle of
rig
to fluid flow. Additional analysis and discussion of the the bedding plane relative to the maximum principal
effects of fabric and stress anisotropy on soil stress– stress direction, and the measured friction angles are
deformation and strength are given in Chapter 11. normalized by the friction angle in plane strain com-
pression with ⫽ 90. The friction angle is the lowest
Sands and Silts when the loading direction is approximately at ⫽ 30.
py
The strength of crushed basalt, both along and across This is partly because the failure shear plane coincides
the direction of preferred orientation of grains, is with the bedding plane. The friction angles in triaxial
shown in Fig. 8.21. Preferred orientation of the some- compression are generally less than those in plain
what elongated particles (mean particle length to width strain compression due to the intermediate stress effect
Co
ratio ⫽ 1.64) was obtained by pouring the soil into a (see Chapter 11). Less bedding effect is also observed
shear box. Intense preferred orientation was obtained in triaxial compression because multiple shear planes
at moderate relative densities, as shown by Fig. 5.11. at different directions are often produced in triaxial
At the lower relative densities the strength was about compression samples, whereas fewer, but more dis-
40 percent greater across the plane of particle orien- tinct, shear plane are observed in plane strain com-
tation than along it. As shown by Fig. 8.21, this dif- pression.
ference decreased with increasing density, and for The orientations of contact planes between particles
relative densities above 90 percent, the strengths in the have significant influence on the stress–strain and vol-
two directions were the same. This is consistent with ume change behavior of granular soils when they are
l
Triaxial compression tests were done on samples of
ria
these sands with different maximum principal stress
directions relative to the original horizontal plane.
The results of these tests for Toyoura sand (b in Fig.
5.13) are shown in Fig. 8.23. Toyoura sand is com-
posed of elongated, flat particles having an axial ratio
ate
of 1.65, but similar results were obtained also for the
Tochigi sand (Fig. 5.13d). The results of these and
other tests reported by Oda (1972a) included tests at
different relative densities. They illustrate important
dM aspects of anisotropic granular soil fabric on mechan-
ical properties, for example:
l
ria
ate
dM
hte
rig
py
Figure 8.23 Effect of initial fabric anisotropy on stress–strain and volume change behavior
of Toyoura sand. Angle is between major principal stress direction and the original hori-
zontal plane (from Oda, 1972a). Reprinted with permission of The Japanese Society of
Co
SMFE.
21 for moist samples as the direction of the major testing undisturbed samples1 cut as shown in Fig. 8.24.
principal stress was changed from normal to the pre-
ferred orientation of particles to 45 to it.
1
Anisotropic fabric in undisturbed Portsea Beach To handle undisturbed sand samples, Lafeber and Willoughby
(1971) used a two-stage replacement of the original seawater by pol-
sand is shown in Fig. 8.3. The effect of this anisotropy yethylene glycol (Carbowax 4000). Triaxial tests were done after first
on the behavior in triaxial compression was studied by heating the samples to melt the Carbowax.
l
ria
ate
Figure 8.24 Orientations of triaxial cylinders of Portsea Beach sand in relation to in situ
conditions (Lafeber and Willoughby, 1971).
dM
Values of mean secant modulus for samples at different 1. Anisotropic fabric, as indicated by particle ori-
orientations are given in Table 8.6. There are signifi- entations and interparticle contact orientations, is
cant differences among samples tested in different di- likely in natural deposits, compacted fills, and
rections, and there is no horizontal plane of isotropy laboratory samples.
for deformation modulus. 2. Anisotropic fabrics produce anisotropic mechan-
Collectively, the results of studies of the effects of ical properties.
hte
fabric anisotropy on properties of granular soils show 3. Strengths and deformation moduli are higher for
the following: shear directions across planes of preferred ori-
entation than along them.
4. The magnitude of strength and modulus aniso-
Table 8.6 Effect of Sample Orientation on Secant tropy depends on density and the extent to which
particles are platy and elongated. Differences in
rig
Direction Azimuth (kN/m2) (kN/m2) greater than differences in peak strength. Moduli
in different directions may differ by a factor of 2
Vertical 5.41 ⫻ 104 0.27 ⫻ 104 or 3.
Horizontal Parallel to 6. The effect of fabric anisotropy on mechanical
4.01 ⫻ 104 0.24 ⫻ 104
Co
冋冉 冊 冉 冊
2
for most soil deposits, and this topic is discussed in
cu cu more detail in Section 9.3.
⫻ ⫺ (1 ⫺ K0) cos2 45 ⫹ ⫺
p p 2
冉 冊册
1 ⫺ K0
2 1/2
l
⫹ (8.2) 8.11 SAND FABRIC AND LIQUEFACTION
ria
2
If saturated sand is at a void ratio above the critical-
where cu is undrained shear strength, p is vertical con- state or steady state line (Fig. 8.20) and sheared rap-
solidation pressure, K0 is the coefficient of lateral Earth idly, it will try to densify. As water cannot escape from
pressure at rest, and is the inclination of the failure the pores instantaneously, the collapsing structure will
ate
plane to the horizontal. The pore pressure parameter transfer normal stress to the pore water. The accom-
Aƒ is defined as panying decrease in effective stress reduces the shear
strength to a low value, and the soil mass liquefies.
Cyclic loading due to earthquakes is perhaps the most
uƒ
Aƒ ⫽ (8.3) common cause of dynamic liquefaction. The resistance
(
1 ⫺
3)ƒ
dM to liquefaction depends on characteristics of the sand,
including gradation, particle size, and particle shape;
where
uƒ is the change in pore water pressure at fail- relative density; confining pressure; and initial stress
ure, and (
1 ⫺
3)ƒ is the deviator stress at failure. state. A comprehensive review of the state of knowl-
The degree of mobilization of c and at peak edge of the causes and effects of soil liquefaction
stress difference and the strain at failure in an un- during earthquakes was published by the National
drained test vary with orientation of principal stresses. Research Council (NRC, 1985) and by Kramer (1996).
hte
Data on the variation of undrained compressive Liquefaction depends on a sand’s resistance to de-
strength with orientation of the failure plane are sum- formation and the degree to which rapidly applied
marized in Fig. 8.25. Strengths in the vertical and hor- shear stresses cause a tendency for the structure to re-
izontal directions may differ by as much as 40 percent duce in volume or collapse. Since samples of the same
as a result of fabric anisotropy. The differences in un- sand at the same density but having different fabrics
drained strength in the different directions result from have different stress–strain and volume change prop-
rig
differences in pore pressures developed during shear erties, see Section 8.8, it follows that different fabrics
(Duncan and Seed, 1966; Bishop, 1966; Nakase and should influence liquefaction resistance as well. Figure
Kamei, 1983; Kurukulasuriya et al., 1999). The effec- 8.27 shows for three sands that preparation of samples
tive stress strength parameters are independent of sam- by two different methods produced distinctly different
ple orientation. The drained strength is independent of resistances to liquefaction, as measured by the number
py
shear stress orientation relative to fabric orientation, as of load cycles to cause liquefaction at a particular value
demonstrated by tests on kaolin (Duncan and Seed, of cyclic stress ratio. The cyclic stress ratio for these
1966; Morgenstern and Tchalenko, 1967b). Stress tests was defined as the ratio of half the cyclic deviator
paths for two samples from a clay with anisotropic stress to the initial effective confining pressure.
Co
fabric but isotropic initial stresses are shown schemat- The differences in liquefaction behavior result from
ically in Fig. 8.26. differences in the sand fabric owing to different sample
The facts that both the effective stress strength pa- preparation methods (Mitchell et al., 1976). Results
rameters and the drained strength are independent of similar to those in Fig. 8.27 are shown in Fig. 8.28 for
fabric anisotropy, but that pore pressures developed in samples of Monterey No. 0 sand at a relative density
undrained shear are strongly influenced by anisotropy, of 50 percent prepared by three different methods.
suggest that the effect of fabric anisotropy on strength Similar behavior was measured for samples of the
is the same for both sands and clays. Changes in stress same sand at a relative density of 80 percent (Mulilus
orientation relative to fabric orientation influence vol- et al., 1977). Monterey No. 0 sand is a uniform me-
l
ria
ate
dM
hte
Figure 8.25 Variation of compressive strength with orientation of failure plane (from Dun-
can and Seed, 1966). Reprinted with permission of ASCE.
rig
py
Co
Figure 8.26 Stress paths in triaxial compression for differently oriented samples for clay
with anisotropic fabric.
l
ria
ate
dM
Figure 8.27 Influence of sand sample preparation method on liquefaction resistance (from
Mulilis et al., 1977). Reprinted with permission of ASCE.
hte
rig
py
Co
Figure 8.28 Liquefaction resistance of Monterey No. 0 sand prepared to a relative density
of 50 percent by three methods (Mulilus et al., 1977).
dium sand with rounded to subrounded grains consist- of undisturbed sand in the field since the field fabric
ing predominantly of quartz with some feldspar and is not usually known, and undisturbed samples are vir-
mica. tually impossible to obtain. It also explains partially
That the fabrics were different for the different prep- why such heavy reliance is placed on the results of in
aration methods was determined by analysis of particle situ tests such as the standard penetration test and the
long axis and interparticle contact on normal orienta- cone penetration test for assessment of the in situ liq-
tions measured on thin sections cut through the uefaction resistance of sand deposits.
samples. Pluviation resulted in distinct preferred Of several laboratory methods that can be used to
orientation of particle long axes in the horizontal di- prepare sand samples, pluviation usually produces the
rection. Moist vibration produced the most random ori- most compressible and weakest fabrics at any relative
entation of particle long axes, with moist tamping density. Thus this method can be used to obtain a lower
l
giving intermediate values. The results of static triaxial bound or most conservative estimate of the properties
ria
compression tests (Fig. 8.29) showed stress–strain and that the same sand at the same relative density can
volume change behavior consistent with the observed have in the undisturbed state in the field. Most sands
fabrics and liquefaction resistance. That is, the weakest in situ are stronger because of prestressing effects, ag-
and least dilative material was that prepared by dry ing, and cementation. The difference between the plu-
pluviation, and the strongest and most dilative material viated sample lower bound values and the actual in situ
ate
was prepared by moist vibration. values can be large. A corollary of this is that undis-
From results such as these, it is clear that relative turbed sand deposits can suffer a stress loss on distur-
density by itself is insufficient for characterization of bance; that is, they are sensitive in the same way as
the sand properties. This means that sand samples re- many clay deposits owing to loose metastable struc-
constituted in the laboratory ordinarily cannot be used
dM tures.
for determination of properties that are representative
St
Insensitive ⬃1.0
Slightly sensitive clays 1–2
Medium sensitive clays 2–4
Very sensitive clays 4–8
Slightly quick clays 8–16
Medium quick clays 16–32
Very quick clays 32–64
⬎64
l
Extra quick clays
ria
From Rosenqvist (1953).
Figure 8.30 Photomicrograph of undisturbed Leda clay, air
dried. Picture width is 8 m (Tovey, 1971).
creases the liquid limit of low-activity clays and con-
ate
sequently the remolded strength, while the void ratio
remains essentially constant or decreases only a small
amount. quick clays as a consequence of delayed or secondary
compression. This compression can be accelerated as
Composition of Sensitive Clays a result of leaching of salts during formation of the
deposit (Torrance, 1974).
dM
Quick clays may not differ from clays of low sensitiv-
ity in terms of mineral composition, grain size distri-
Causes of Sensitivity
bution, or fabric. Most quick clays are postglacial
deposits, with the mineralogy of the clay fraction dom- At least six different phenomena may contribute to the
inated by illite and chlorite and that of the nonclay development of sensitivity:
fraction by quartz and feldspar. Amphibole and calcite
are also common. The activity of quick clays is usually 1. Metastable fabric
hte
less than 0.5. The pore fluid composition and the 2. Cementation
changes in composition that have developed between 3. Weathering
the time of deposition and the present are of paramount 4. Thixotropic hardening
importance. Changes in the type and amount of elec- 5. Leaching, ion exchange, and change in the
trolyte, organic compounds, and small quantities of monovalent/divalent cation ratio
6. Formation or addition of dispersing agents
rig
the undisturbed fabric of sensitive clays is composed in a cardhouse arrangement of elongate and platy par-
of flocculated assemblages of particles or aggregates. ticles. A consequence of this is well illustrated by the
Electron photomicrographs show open and flocculated sedimentation compression line relative to the intrinsic
particle arrangements in medium sensitive to quick compression line in Fig. 8.16b. During consolidation
Co
clays. The contribution of fabric to high sensitivity is this fabric can carry effective stress at a void ratio
through open networks of particles and aggregates that higher than would be possible if the particles and par-
are linked by unstable connections. The fabric of un- ticle groups were arranged in an efficient, parallel ar-
disturbed Leda clay is shown in Fig. 8.30. A very wide ray. When saturated soil is mechanically remolded
range of particle sizes may be seen. from a state such as represented by point 1 in Fig. 8.13,
The microfabric of quick clay and that of adjacent the fabric is disrupted, effective stresses are reduced
zones of much less sensitive clay may be the same. because of the tendency for the volume to decrease,
Thus, while an open flocculated fabric is necessary, it and the strength is less.
is not a sufficient condition for quick clay develop- If the original consolidation stress is reapplied, then
ment. Some preferred orientation might develop in there will be additional consolidation, and the void ra-
tio will decrease to a point as represented by 2 in Fig. effective stress to almost zero from the initial value of
8.13. Mechanical remolding and reapplication of 200 kPa. This illustrates the interdependence of effec-
stresses will cause consolidation to point 3, and con- tive stress and structure, as well as the effects of struc-
tinued repetition of the process will lead ultimately to ture metastability.
a minimum void ratio for the fully destructured soil at A point of practical importance is that the continu-
n. Thus, if the soil is at any state within the shaded ing generation of metastable fabrics following distur-
zone of Fig. 8.13, it will have some degree of metas- bance explains why some sand deposits have been
tability of structure and could be further consolidated observed to reliquefy at the same locations in succes-
if disturbed and recompressed. sive earthquakes.
Sensitivity values resulting from metastable particle Cementation Many soils contain carbonates, iron
arrangements were measured in undrained triaxial tests oxide, alumina, and organic matter that may precipitate
l
on saturated kaolinite samples consolidated from high at interparticle contacts and act as cementing agents.
ria
initial water content (Houston, 1967). They decreased On disturbance, the cemented bonds are destroyed
from 12 at high water content and low consolidation leading to a loss of strength. Four naturally cemented
pressure to 2 at low water content and high consoli- Canadian clays tested by Sangrey (1970) had sensitiv-
dation pressure. Consolidated, undrained triaxial com- ities of 45 to 780.
pression tests on saturated sand–kaolinite mixtures
ate
Late glacial plastic clay from near Lilla Edit in the
consolidated initially under an effective stress of 200 Gota Valley of Sweden has a sensitivity of 30 to 70.
kPa gave the results shown in Fig. 8.31. The loss in The apparent preconsolidation pressure as determined
strength due to disturbance was accompanied by a by odometer tests is much greater than the maximum
large increase in pore water pressure and decrease in past overburden pressure (Bjerrum and Wu, 1960).
dM When consolidation pressure greater than this apparent
maximum past pressure is applied, there is a marked
reduction in cohesion. This was interpreted to result
from a rupture of cemented interparticle bonds that
were created by carbonation of microfossils and or-
ganic matter and precipitation of pore water salts at
particle contacts. Removal of carbonates, gypsum, and
iron oxide by leaching with EDTA (a disodium salt of
hte
l
Freshwater leaching following a drop in sea level or
ria
rise in land level results in removal of the seawater
environment. Percolating freshwater in silt and sand
lenses is sufficient to remove salt from the clay by
Figure 8.32 Properties of a purely thixotropic material. diffusion without the requirement that the water flow
through all the pores of intact clay (Torrance, 1974).
ate
Although leaching causes little change in fabric, the
interparticle forces are changed, resulting in a decrease
tropic hardening may account for low to medium sen- in undisturbed strength of up to 50 percent, and such
sitivity and for a part of the sensitivity of quick clays a large reduction in remolded strength that quick clay
(Skempton and Northey, 1952).
dM forms. The large increase in interparticle repulsion is
The mechanism of thixotropic hardening is ex- responsible for the deflocculation and dispersion of the
plained as follows (Mitchell, 1960). Sedimentation, clay on mechanical remolding. It results in part from
remolding, and compaction produce soil structures the decrease in electrolyte concentration causing in-
compatible with these processes. Once the externally crease in double-layer thickness. Changes in strength
applied energy of remolding or compaction is re- and the increase in sensitivity accompanying the leach-
moved, however, the structure is no longer in equilib- ing of salt from a Norwegian marine clay are shown
rium with the surroundings. If the interparticle force in Fig. 8.34. The relationship between sensitivity and
hte
balance is such that attraction is somewhat in excess salt content for several Norwegian marine clays is
of repulsion, there will be a tendency toward floccu- shown in Fig. 8.35. Confirmation of the leaching hy-
lation of particles and particle groups and for reorgan- pothesis was obtained by means of leaching tests on
ization of the water–cation structure to a lower energy artificially sedimented clays (Bjerrum and Rosenqvist,
state. Both effects, which have been demonstrated ex- 1956). Åsrum clay sedimented in saltwater (35 g/liter)
perimentally, take time because of the viscous resis- and then leached of salt exhibited an increase in sen-
rig
tance to particle and ion movement. sitivity from 5 to 110. A sample sedimented in fresh-
The effect of time after disturbance on the pressure water had a sensitivity of 5 to 6.
in the pore water is particularly significant. Several Although leaching of salt is necessary, it may not be
studies show that there is a continual decrease in pore sufficient for the development of quick clay. The salt
water pressure, or increase in pore water tension, with
py
slow regain during periods of rest. The concurrent range is that the essential condition for development
time-dependent increase in effective stress accounts for of quick clay is an increase in interparticle repulsions.
the observed increase in undrained strength. Considerations in Chapter 6 show that the type of cat-
The importance of thixotropic hardening in contrib- ions and the relative amounts of monovalent and
uting to the sensitivity of clay in the field is impossible divalent cations have a controlling influence on
to determine. Laboratory studies start with a specific equilibrium particle arrangements.
present composition and density. The initial state of a
clay deposit in nature is usually far different than at
the present time, and the history of an undisturbed clay 2
Sherard (1975, personal communication) indicated that thixotropic
bears little resemblance to that of a remolded sample strength ratios of up to 100 have been measured in Champlain clay.
l
ria
ate
dM
hte
rig
py
Figure 8.33 Effect of shear on pore water tensions for various clays (after Day, 1955).
Co
The electrokinetic or zeta potential in Champlain also shown in Fig. 8.36. The percent monovalent cat-
clay, as determined using electroosmosis (see Chapter ions in the pore water is given by
9), correlates well with sensitivity, as shown in Fig.
8.36 (Penner, 1965). The electrokinetic potential is a Na⫹ ⫹ K⫹
measure of the double-layer potential, with higher val- ⫻ 100
ues associated with thicker double layers and higher Na ⫹ K⫹ ⫹ Ca2⫹ ⫹ Mg2⫹
⫹
l
tem, possibly by organic matter (Söderblom, 1969;
ria
Lessard and Mitchell, 1985). The mechanism by which
these changes occur as deduced by Lessard (1981) is
summarized as follows.
Organic matter from marine organisms deposits si-
multaneously with the illite, feldspar, and quartz that
ate
constitute the bulk of a postglacial marine clay. Iron
oxide minerals are also present in small quantities. As
the depth of burial increases with continued deposition,
so does the distance to oxygen supply from the sea-
dM water above.
Bacterial oxidation of the organic matter depletes
the oxygen content of the pore water, and an anaerobic
environment develops that reduces ferric oxides to sol-
uble ferrous iron. Simultaneously, sulfates in the pore
water are reduced to hydrogen sulfide by the organic
matter with the aid of sulfate-reducing bacteria. The
formation of iron sulfide materials then follows:
hte
l
ria
ate
dM
Figure 8.35 Relationship between sensitivity and salt concentration for some Norwegian
hte
Aging of Quick Clay Samples From studies on the quick clay from LaBaie, Que-
Important changes in the properties of quick clays have bec, it was possible to explain the transformations that
been observed to develop with time after sampling, cause changes in properties after sampling, such as
rig
including increases in remolded strength and liquid those shown in Fig. 8.37 (Lessard and Mitchell, 1985).
limit and decrease in the liquidity index, all without Geotechnical properties of the LaBaie clay determined
change in water content. For example, the changes that within one month after sampling are shown in Fig.
occurred in remolded quick clay from Outardes-2 in 8.38. This clay is composed primarily of rock flour
containing plagioclase, K-feldspar, quartz, amphibole,
py
the consistencies of different clays, as it normalizes the liquidity index, pH, and concentrations of several ion
water content relative to the plasticity index.3 types as a function of storage time are shown in Fig.
8.39. These results show that aging leads to increases
in both pore water salinity and the concentrations of
3
Similarly, the void index, Iv [Eq. (8.1)] is often used for correlating divalent cations in the pore water and decreases in pH.
the compression behavior of different clays and for assessing the
influence of structure on properties (Burland, 1990; Cotecchia and
Collectively, the compositional changes are responsible
Chandler, 2000; Jardine et al., 2004). for increase in remolded strength (Fig. 8.37) and de-
crease in liquidity index (Fig. 8.39) because each
e ⫺ e 100
* depresses the double layer, thereby decreasing the
Iv ⫽
C *c interparticle repulsive forces. The remolded strength
l
strength and the liquid limit and decreases in the sen-
ria
sitivity and liquidity index. More complete descrip-
tions of the reactions, including phase diagrams and
reaction kinetics are given by Lessard (1981) and Les-
sard and Mitchell (1985). An important role of bacteria
in mediating the oxidation and reduction reactions
ate
associated with quick clay formation and aging is
suggested. The importance of geochemical and
microbiological processes in geotechnical engineering
has been given little attention in the past. Future stud-
dM ies of the phenomena and processes are likely to pro-
vide important new insights and understanding.
and electrokinetic potential (data from Penner, 1965). testing can provide a rapid and easy means for assess-
ing whether aging processes have occurred. To mini-
mize the aging effects the exposure of samples to air
correlates well with both the concentration of divalent should be minimized, thick wax caps should be used
cations and the total cation concentration, as shown in with rust-free sample tubes, and samples should be
py
Fig. 8.40. The method of storage (see Fig. 8.39) does stored at low temperatures to slow down reaction rates.
not affect the correlations shown in Fig. 8.40; rather,
it influences the time required for the chemical con- Summary of Sensitivity-Causing Mechanisms
centration changes to occur. The six causes for the development of sensitivity dis-
Co
These changes in chemistry and properties are cussed above are summarized in Table 8.8. An estimate
caused by the following sequence of events. When of the upper limit of sensitivity for each mechanism is
quick clay is sampled or exposed, some contact with also given. Virtually all natural soils, including many
the air and oxygen is inevitable. This air causes some sands, are sensitive in that they lose some strength on
of the remaining organic matter to oxidize and form disturbance and remolding. Exceptions are heavily ov-
carbonic acid, which, in turn, dissolves calcium car- erconsolidated stiff fissured clays that can gain strength
bonate, thus increasing the concentrations of calcium because of the elimination of fissures and planes of
and bicarbonate in the pore water. Even extremely low weakness. Quick clays are formed from soft glacial
partial pressures of O2 are sufficient to initiate oxida- marine clays only after removal of excess salt by leach-
tion phases of the sulfur cycle. The oxidation of pyrite ing and further increase in double-layer repulsions as
l
ria
ate
dM
Figure 8.37 Changes in the remolded strength and consistency of a Canadian quick clay as
a function of time (Lessard, 1978).
hte
rig
py
Co
Figure 8.38 Geotechnical characteristics of the quick clay from LaBaie as a function of
depth (after Lessard, 1981).
l
dictable in terms of sensitivity, liquidity index, and ef-
ria
fective stress using the concepts given in the preceding
sections.
ate
Glacial and postglacial clays of high and low sensitiv-
ity exhibit significant differences, as shown by the pro-
files in Fig. 8.41 for a normal clay from Drammen and
a quick clay from Manglerud, both in Norway. One of
dM the most important of these differences is that at Man-
glerud the water content is well above the liquid limit;
that is, the liquidity index is greater than 1.0. This is
characteristic of quick clays.
Plasticity and Activity When normal clays are con-
verted to highly sensitive or quick clays by the chem-
ical changes described in Section 8.12, the liquid limit,
plasticity index, and activity decrease. These changes
hte
l
monovalent/divalent cation ratio
ria
Formation or addition of dispersing Extra quick (⬎64) Inorganic clays containing organic compounds
agents in solution or on particle surfaces
a
Adjectival descriptions according to Rosenqvist (1953).
b
Pertains to samples starting from present composition and water content. Role of thixotropy in causing sensitivity in
ate
situ is indeterminate.
burden vertical effective stress, vo, and CIUC means Sensitivity As each point on the curves for fully
isotropically consolidated undrained compression tests
dM destructured soil in Figs. 8.13 and 8.14 represents com-
were used for determination of strength. pletely remolded material, the sensitivity at any point
Stress–Strain Relationships In general, strain at on this curve must be unity. Thus this curve is a line
failure decreases with increasing sensitivity. Some of constant sensitivity or sensitivity contour. Saturated
quick clays are quite brittle during unconfined loading clay at a given water content and pore fluid composi-
and fracture at very low strains, sometimes by axial tion cannot be made weaker than its thoroughly re-
splitting. Further working of the fractured specimen molded strength. Therefore, a water content–effective
may cause it to turn into a fluid mass. stress relationship to the left of that for the fully de-
hte
Compressibility The compressibility of highly sen- structured soil is not possible. The undisturbed strength
sitive clays is relatively low until the consolidation increases with increasing effective stress at constant
stress exceeds the preconsolidation pressure. It then in- water content (Fig. 8.45), and the sensitivity at all
creases sharply as shown by Fig. 8.44 for Champlain points to the right of the fully destructured soil curve
clay. As the void ratio reduces under higher consoli- is greater than 1. Thus, the maximum gradient of sen-
dation pressures, the compressibility eventually as-
rig
Consolidation Because the initial structure depends content at constant initial effective stress. At constant
on many factors and the volume changes under pres- water content, the lower the effective stress, the easier
sure are a function of structure, a soil does not have a it is for the soil to dilate since less energy is required
unique consolidation curve. All states and compression for expansion against low pressures than high. There-
Co
curves must be above the curve for the fully destruc- fore, the maximum gradient of Aƒ is as shown in Fig.
tured material. 8.46.
Strength of Normally Consolidated Soil The higher Strain at Failure Restrained dilation increases ef-
the effective stress at a given water content, the greater fective stress, thus increasing shearing resistance. Con-
the undrained strength because of increased frictional sequently, the deformation required to cause failure
resistance between particles. For constant effective increases with increasing dilation. On the other hand,
stress, strength increases with decreasing water content strain at failure should decrease with increase in Aƒ
because of increased dilatancy. Thus the general be- because Aƒ varies inversely with dilation tendencies.
havior shown in Fig. 8.45 is observed. Consequently, the maximum positive gradient of strain
l
ria
ate
dM
hte
rig
Figure 8.41 Soil profiles for marine clays of low and high sensitivity (from Bjerrum, 1954).
py
at failure should be opposite to the maximum gradient sensitivity, Aƒ , and strain at failure. Contours based on
for Aƒ . these values are shown in Fig. 8.47. The variations in
Co
Example of Relationships The results of triaxial the measured values are in general accord with the
compression tests on kaolinite (Houston, 1967) illus- predictions stated previously.
trate the above relationships. By consolidating different
samples from several different initial water contents
and remolding and reconsolidating them in various Sensitivity–Effective Stress–Liquidity Index
ways, samples covering a range of initial effective Relationship
stress and water content values, each reflecting a dif- General relationships between sensitivity, effective
ferent structure, were obtained. The results of un- stress, and water content can be established based on
drained triaxial tests yield values of strength, normalization of the remolded strength versus water
l
ria
Figure 8.42 Sensitivity as a function of liquidity index for
Norwegian marine clays. Relationship was averaged from Figure 8.44 Consolidation curves for Champlain (Leda)
ate
many more data points than those shown (data from Bjerrum, clay. Reproduced with permission from the National Re-
1954). search Council of Canada, from the Canadian Geotechnical
Journal, Vol. 3, pp. 61–73, 1966.
dM
hte
ASCE.
l
Some fine-grained soils are structurally unstable, easily
ria
dispersed, and, therefore, easily eroded. Soils in which
the clay particles will detach spontaneously from each
other and from the soil structure and go into suspen-
sion in quiet water are termed dispersive clays. The
consequences of the exposure of dispersive clays to
ate
water may be several, as shown by Figs. 8.50 and 8.51.
The surface erosion pattern on an excavated slope,
which is characteristic of ‘‘badlands’’ topography, is
Figure 8.46 Gradient of pore pressure parameter Aƒ with shown in Fig. 8.50. Erosion tunnels in a flood control
water content and effective stress variations. dike are shown in Fig. 8.51. Failures of this type have
dM occurred in well-constructed, low homogeneous dams.
In each case shown, the soil contained readily dis-
persed clay particles that went easily into suspension
1
Su ⫽ (kPa) (8.4) in flowing water. Failures of this type have occurred in
(LI ⫺ 0.21)2 embankments, dams, and slopes composed of clays
with low-to-medium plasticity (CL and CL–CH) that
The following equation that can be deduced from contain montmorillonite. Dispersive piping in dams
Sharma and Bora (2003) also fits the relationship de-
hte
In Eq. (8.5) is the undrained strength and w, wL, and placed dry of optimum and not well compacted. Set-
wp are the water content, liquid limit, and plastic limit tlement below the phreatic surface and arching above
values. it can result in crack formation. Water moving through
By averaging data for several clays, the relationship the crack picks up dispersive clay particles, with the
between liquidity index, effective stress, and sensitivity rate of removal increasing as the seepage velocity and
py
shown in Fig. 8.49 is obtained. Figure 8.49 is also valid size of opening increase. This is a fundamentally dif-
for moderately overconsolidated clays, provided the ferent mechanism than erosive piping, which develops
preconsolidation pressure is used instead of the present and works backward from the discharge face. Tunnel-
effective stress. This is because the water content and ing has been initiated in soils with a hydraulic con-
Co
undrained strength depend more on the preconsolida- ductivity as low as 1 ⫻ 10⫺7 m/s.
tion pressure than on the present effective stress. Some Visual classification, Atterberg limits, and particle
deviations from the values in Fig. 8.49 are to be ex- size analyses do not provide a basis for differentiation
pected because of the extensive averaging used in its between dispersive clays and ordinary erosion-resistant
preparation. These deviations may be greatest for extra clays. However, relatively simple chemical tests, a dis-
quick clays because of the very low remolded strength, persion test, a ‘‘crumb’’ test, and the pinhole test (Sher-
the difficulty in determining it accurately, and its con- ard et al., 1976) can be used for identification of
trolling influence on the calculated value of sensitivity. dispersive clays.
Nonetheless, the relationships in Figs. 8.48 and 8.49 In the pinhole test, distilled water is allowed to flow
can be used to estimate sensitivity and strength when through a 1.0-mm-diameter hole drilled through a
undisturbed samples or in situ strength data are not compacted specimen. If the soil is dispersive, the water
l
ria
ate
dM
hte
rig
py
becomes muddy and the hole rapidly erodes. For non- indicating probable dispersive clay behavior in soils of
dispersive clays the water remains clear and there is relatively low total salt concentration in the pore water.
no erosion. The pinhole test and test procedure are As determination of the ESP requires measurement of
described in ASTM Standard D4647-93 (1998) both the cation exchange capacity and the amount of
(ASTM 2000). sodium in the exchange complex, it is not a simple or
As noted in Section 6.15, the exchangeable sodium rapid method for identification of dispersive clay. A
percentage (ESP) is a strong indicator of potential dis- simpler chemical measure of potential dispersivity,
persive behavior, with an ESP greater than 2 indicating supported by the results of tests on many samples, was
possible dispersion, and an ESP greater than 10 to 15 proposed by Sherard et al. (1972, 1976) that is based
l
ria
Figure 8.50 Erosion pattern in excavated slope of sensitive
clay (courtesy of J. L. Sherard).
ate
on the percent sodium in the saturation extract from a
Figure 8.48 Relation between shear strength of remolded
clay and liquidity index (from Leroueil et al., 1983). Repro- soil–water paste. This correlation is shown in Fig.
duced with permission from the National Research Council 8.52.
of Canada. Many subsequent tests have shown, however, that
dM the zones in Fig. 8.52 are not always reliable indicators
of dispersibility. For example, Craft and Acciardi
(1984) found that only 62.3 percent of 223 samples
were classified correctly. This is not surprising because
hte
rig
py
Co
Figure 8.49 General relationships between sensitivity, liquidity index, and effective stress.
l
erosion of the dispersive clay to stop.
ria
In the dispersion test the percentage of particles finer
than 5 m is determined by hydrometer analyses of
samples with and without dispersing agent in the sus-
pension water (Sherard et al., 1972). The higher the
ratio of percentage material finer than 5 m by weight
ate
Figure 8.51 Erosion damage on the crest of 5-m-high flood measured in the test without dispersing agent to that
control dike caused by rain runoff concentrating in drying measured in the test with a dispersing agent, the
cracks, Rio Zulia, Venezuela (courtesy of J. L. Sherard).
greater the probability of dispersion in the field. This
ratio, when expressed as a percentage, is termed the
dM percent dispersion. Values greater than 20 to 25 percent
whether or not a soil will exhibit dispersive behavior indicate that dispersion may be a problem. Values
depends not only on its chemical and mineralogical greater than 50 percent are nearly always indicative of
composition but also on its state, as reflected by water soils susceptible to severe erosion damage initiated by
content, density, and structure, on the chemistry of the clay dispersion.
water to which it is exposed, and on the specific con- In the crumb test a small clod of the soil is placed
ditions of exposure, including temperature, confining in a beaker and submerged in water. If the soil clod is
hte
rig
py
Co
Figure 8.52 Relationship between dispersibility (susceptibility to colloidal erosion) and dis-
solved pore water salts based on pinhole tests and field observations. SAR ⫽ sodium ab-
sorption ratio, Eq. (6.33). Concentrations in meq / liter (from Sherard et al., 1976). Reprinted
with permission of ASCE.
l
convert a soil to a nondispersive form. Filters that are stresses, then the material splits apart. Seedsman
ria
designed to retain small particle sizes should be used (1986) found that the slaking mechanism was related
on the discharge side of dams and dam cores. For an to the bulk density, and the higher the density the more
existing dam, in which tunneling erosion is expected resistant the material to slaking by any mode.
to develop, lime can be added at the upstream face to
be carried inward by the percolating water. Additional
ate
strategies were suggested and evaluated by Sherard and
Decker (1977). 8.16 COLLAPSING SOILS AND SWELLING
SOILS
Large areas of Earth’s surface, particularly in the Mid-
8.15 SLAKING dM west and Southwest United States, parts of Asia, South
Most fine-grained soils slake after exposure to air and America, and southern Africa, are covered by soils that
subsequent unconfined immersion in water; an initially are susceptible to large decreases in bulk volume when
intact piece of soil will disintegrate into a pile of pieces they become saturated. Such materials are termed col-
or sediment of small particles. This disintegration may lapsing soils. Collapse may be triggered by water alone
begin immediately upon immersion or develop slowly or by saturation and loading acting together. Soils with
with time. Slaking usually is more rapid and vigorous collapsible structures may be residual, water deposited,
hte
in materials that have been dried prior to immersion or aeolian. In most cases, the deposits have a loose
compared to the same material immersed at its initial structure of bulky shaped grains, often in the silt-to-
water content. Whether a material slakes or not has fine-sand range. Collapsible grain structures are left
been proposed as a basis for distinguishing between behind in residual soils as a result of leaching of sol-
soil and rock (Morgenstern and Eigenbrod, 1974). The uble and colloidal material. Water- and wind-deposited
slaking of hard clays and clay shale is a concern in the collapsing soils are usually found in arid and semiarid
rig
stability of open excavations and the shale durability regions and are a consequence of the loose fabrics and
when it is used as an aggregate or rockfill for construc- weak structures that form.
tion. Debris flows (mudflows and torrential stream depos-
From controlled tests on relatively pure samples of its) are deposited suddenly and locally, and may form
a loose, metastable structure. Torrential stream sedi-
py
from the surface of the intact clay by dispersion where such deposits form, water moves upward to
into the adjacent water. evaporate, leaving behind its content of dissolved salts.
2. Swelling Slaking Water is adsorbed by the clay If subsequently wetted, the loose structure can collapse
and the material swells and softens. and cause large settlements.
3. Surface Slaking Aggregates of clay particles When a large canal was constructed through the San
spall off the surface and accumulate as sediment Joaquin Valley during the 1960s to carry water from
in the adjacent water. northern California to southern California, it was nec-
4. Body Slaking The material splits and disinte- essary to cross many collapsible debris flow deposits.
grates into pieces, and the failure appears to de- In order to minimize future settlements of the canal
velop from the inside out. and appurtenant structures as a result of canal leakage,
extensive preponding was carried out before construc- The low density and light cementation of the loess
tion of the canal. structure make it susceptible to collapse. When main-
Soils susceptible to large collapse as a result of wet- tained dry, it is reasonably strong and incompressible.
ting can be identified using a density criterion. If the The porous structure may persist even beneath 60 m
density is sufficiently low that the void space is larger of overburden. When saturated, however, loess depos-
than needed to hold the liquid limit water content, then its may lose their stability. Compression due to satu-
collapse problems are likely (Gibbs and Bara, 1967). ration alone may be small, but with a surcharge, it may
If the void space is less than that needed for the liquid be very large, as shown in Fig. 8.54. Watering lawns
limit water content, then collapse is not likely unless around new houses founded on loess has been known
the soil is loaded. to cause large settlements. If saturated loess deposits
Loess deposits are widespread throughout the mid-
l
western United States and parts of Asia. This material,
ria
which is wind-blown silt, is light brown in color, crum-
bly, and essentially devoid of stratification. The par-
ticles are predominantly silt size and composed of
feldspar and quartz. A small amount of clay, usually
less than 15 percent, may be present. Smectite is the
ate
usual clay mineral type. Calcite may be present in
amounts up to 30 percent and can act as weak cement
that precipitates along the sides of vertical root holes
and at interparticle contacts. Densities of undisturbed
loess may be as low as 1.2 g/cm3, and the natural
dM
water content of metastable deposits of loess is low,
on the order of 10 percent. Most loesses plot near the
A-line on the plasticity chart.
Because of vertical root holes formed by gradual
burial of grassy plains, the absence of stratification,
and light cementation, loess cleaves on vertical planes,
and vertical faces cut in loess are quite stable, as shown
hte
in Fig. 8.53. In fact, if inclined slopes are cut, they will Figure 8.54 Compression properties of Missouri River basin
gradually erode back to a series of steplike vertical loess (from Clevenger, 1958). Reprinted with permission of
faces. ASCE.
rig
py
Co
l
(1977, 1983).
ria
Among the continuum of soils and rocks that are en- Knowledge of the geologic history of a deposit, the
countered in engineering and construction, very heav- mineralogical and chemical composition, and the new
ily overconsolidated fine-grained soils and mudstones, loading and exposure conditions provides initial in-
shales, and siltstones are sometimes among the most sights about whether shales, siltstones, and sandstones
difficult to deal with. It is not always clear whether can be expected to degrade. Accelerated weathering
ate
such materials should be treated as soil or rock. If be- and durability tests are used to classify shale durability.
havior is rocklike, the material can be used in earth- Tests used for this purpose have been described and
work construction like a rock and placed in thick lifts reviewed by Huber (1997). They include water adsorp-
without much compaction. If the shale is susceptible tion, wet–dry, freeze–thaw, jar slaking, crushing, point
to break down, however, it must be treated as a soil
dM load strength, ultrasonic disaggregation, and slake
and placed in thin, well-compacted lifts. If considered durability tests in which the breakdown of shale
a rock and subsequent deterioration under the actions submerged in a rotating wire basket (Franklin and
of stress, water, and chemical change causes break- Chandra, 1972) is determined. The results of these
down, loss of strength and increase in compressibility, tests form the basis of several shale durability classi-
then there can be failures. Conversely, if the durability fication systems that have as their goal to distinguish
and mechanical properties are too conservatively as- shales that cause problems from those that do not. One
signed, then unnecessary overdesigns and excessive of the first such systems was developed by Underwood
hte
costs may result. (1967) for engineering evaluations. Table 8.9, adapted
Shale is a prime example of a material that illus- from Underwood’s study, is a listing of physical and
trates the soft rock-hard soil problem. According to composition properties associated with the indicated
Terzaghi et al. (1996): types of unfavorable behavior. It may be seen that the
range of most properties within which unfavorable be-
Shale is a clastic sedimentary rock mainly composed of
havior is likely to develop is rather broad, which means
rig
used as an embankment fill or stable pavement sub- processes and subsequent changes over time act to
grade. On the other hand, many shales that appear in- produce unique types of soil structures with character-
tact and rocklike when exposed or excavated can have istic properties, and how these properties and the as-
properties that deteriorate with time. The problems, sociated behavior are interrelated. Several illustrations
then, are to determine whether degradation is likely, of the relevance of these processes and properties to
and if so, how much and how fast. geotechnical applications are among the subjects of
Degradation, apart from that caused by mechanical this chapter.
processes such as unloading, compression, crushing, The structure of a soil depends on its fabric and
and shearing, is usually by slaking (see Section 8.15) interparticle force system. It reflects all facets of the
initiated by exposure to air, moisture, and changed soil’s composition, history, present state, and environ-
chemical environment. Cementation shales are likely mental influences. Soil particles come in a great variety
Table 8.9 Properties and Conditions Likely to Cause Problems with Shale
l
ria
Physical and Compositional Properties Probable in Situ Behavior
Lab tests and in situ Unfavorable behavior High pore Low Tendency to Slope Rapid Rapid Tunnel
observations probable for values pressure bearing rebound stability slaking erosion support
in indicated range capacity problems problems
ate
Compressive ⬍300–1800 ⫻ ⫻
strength, kPa
Modulus of ⬍140–1400 ⫻ ⫻
elasticity, MPa
Cohesive strength, ⬍30–700 ⫻ ⫻ ⫻
dM
kPa
Angle of internal ⬍10–20 ⫻ ⫻ ⫻
friction, deg.
Dry unit weight, ⬍11.0–17.3 ⫻ ⫻(?)
kN/m3
Potential swell, % ⬎3–15 ⫻ ⫻ ⫻ ⫻
Natural moisture ⬎20–35 ⫻ ⫻
hte
content, %
Hydraulic ⬍10⫺5 ⫻ ⫻ ⫻
conductivity, m/s
Predominant clay Smectite or illite ⫻ ⫻
minerals
Activity ⬎0.75–2.0 ⫻
rig
Wetting and drying Reduces to grain ⫻ ⫻
cycles sizes
Spacing of rock Closely spaced ⫻ ⫻ ⫻(?) ⫻
defects
Orientation of rock Adverse ⫻ ⫻ ⫻
py
defects
State of stress ⬎Existing ⫻ ⫻ ⫻
overburden
Co
l
c. If this kaolinite were to be used for a structural
ria
fill, which method of compaction would you
specify? Why?
QUESTIONS AND PROBLEMS d. If this kaolinite were to be used for the core of
an earth dam, which method of compaction
ate
1. Indicate whether each of the following statements would you specify? Why?
is True or False. Justify your answer with a brief e. If each material was saturated without further
statement. change in dry unit weight and then sheared un-
a. Rearrangement of particles during shear pro- drained, sketch the curves of pore water pressure
vides an important contribution to the residual versus strain that you would expect to obtain for
strength of a highly plastic clay.
dM
b. The relationship between critical void ratio and
each.
3. Given heavily overconsolidated clay under a present
effective confining pressure is the same for un- vertical effective overburden stress of 200 kPa, the
disturbed and reconstituted samples of the same maximum past effective stress on the horizontal
sand. plane was 600 kPa. Consider this to be state I. It
c. The sensitivity of a clay can be explained by the has been determined that this clay has a peak fric-
change in effective stress caused by remolding. tion angle of 30 and a residual friction angle
hte
d. Collapse of structure in a saturated soil is usually r of 17. There is no cementation of the clay struc-
accompanied by an increase in effective stress. ture. Rapid shear is defined as deformation at a rate
e. Relative density is a suitable single parameter for fast enough so there can be no change in pore pres-
characterizing sand properties. sure in the shear zone. Slow shear is deformation
f. Strength loss when a sensitive clay is disturbed at a rate slow enough that there can be no change
rig
l
bilized in situ, that is, made less susceptible to large
ria
strength loss. Na⫹ ⫽ 4.0 meq/liter
5. It is shown in Fig. 8.48 that there is a unique re- Ca2⫹ ⫽ 0.5 meq/liter
lationship between remolded shear strength and li-
quidity index that seems to be independent of the Mg2⫹ ⫽ 0.3 meq/liter
ate
particular clay. Explain why this should be so.
6. Which is easier to determine, the exchangeable so- Comment on the possible consequences of pro-
dium percentage or the sodium adsorption ratio? longed percolation by this water. Justify your
Why? conclusions numerically.
7. Salt-affected soils are classified by agronomists as
dM 8. State special geotechnical characteristics of the
follows: following soil types and relate them to their
(1) formational processes, (2) composition, (3) en-
vironmental setting, and (4) structure. By all means
consult references in addition to the relevant sec-
Exchangeable Usual tions in this book to enhance the quality of your
Sodium Usual Structural answers.
Soil Group Percentage pH State a. Loess
hte
g. Carbonate sand
a. Complete the above table by filling in the last h. Glacial moraine
column on the right. i. Saprolite
b. Which, if any, of these soil types may be a prob-
py
l
soil on Earth.
Use principles relating to geologic and soil-forming
ria
h. The optimum size of particles that might form
processes, soil composition, surfaces, fabric, structure, metastable honeycomb structures on the Moon
and any other relevant concepts to make reasoned es- as opposed to a finding by Terzaghi that silt-size
timates of or comments on the following: particles in the range of 6- to 20-m diameter
a. Soil particle composition are most susceptible to this effect on Earth.
ate
dM
hte
rig
py
Co
Conduction Phenomena
l
ria
ate
9.1 INTRODUCTION flow are important under a variety of circumstances. A
coupled flow is a flow of one type, such as hydraulic,
Virtually all geotechnical problems involve soil or rock driven by a potential gradient of another type, such as
deformations and stability and/or the flow through
dM electrical.
earth materials of fluids, chemicals, and energy in var- This chapter includes a review of the physics of di-
ious forms. Flows play a vital role in the deformation, rect and coupled flow processes through soils and their
volume change, and stability behavior itself, and they quantification in practical form, an evaluation of rele-
may control the rates at which the processes occur. vant parameters, their magnitudes, and factors influ-
Descriptions of these flows, predictions of flow quan- encing them, and some examples of applications.
tities, their rates and changes with time, and associated
changes in the properties and composition of both the
hte
form a major part of engineering analysis and design. Fig. 9.1) relates linearly to its corresponding driving
As a result, much is known about the hydraulic con- force Xi according to
ductivity and permeability of earth materials. Chemi-
cal, thermal, and electrical flows in soils are also Ji ⫽ Lii Xi (9.1)
important. Chemical transport through the ground is a
py
major concern in groundwater pollution, waste dis- in which Lii is the conductivity coefficient for flow.
posal and storage, remediation of contaminated sites, When written specifically for a particular flow type and
corrosion, leaching phenomena, osmotic effects in clay using familiar phenomenological coefficients, Eq. (9.1)
layers, and soil stabilization. Heat flows are important becomes, for cross section area A
Co
251
vh ⫽ kh ih (9.6)
qh ⫽ kh ih A (9.7)
l
terial. Steady-state and transient flow analyses in soils
ria
are based on Darcy’s law. In many instances, more
attention is directed at the analysis than at the value of
kh. This is unfortunate because no other property of
importance in geotechnical problems is likely to ex-
ate
hibit such a great range of values, up to 10 orders of
magnitude, from coarse to very fine grained soils, or
show as much variability in a given deposit as does
the hydraulic conductivity. Some soils exhibit 2 or 3
orders of magnitude variation in hydraulic conductivity
dM as a result of changes in fabric, void ratio, and water
content. These points are illustrated by Fig. 9.2 in
which hydraulic conductivity values for a number of
soils are shown.
Figure 9.1 Four types of direct flow through a soil porous Different units for hydraulic conductivity are often
mass. A is the total cross-section area normal to flow; n is used by different groups or agencies; for example, cen-
porosity. timeters per second by geotechnical engineers, feet per
year by groundwater hydrologists, and Darcys by pe-
hte
respectively. Typical ranges of values for these prop- Fluid flow through soils finer than coarse gravel is lam-
erties are given later. The driving forces for flow are inar. Equations have been derived that relate hydraulic
conductivity to properties of the soil and permeating
given by the respective hydraulic, thermal, electrical,
fluid. A usual starting point for derivation of such
and chemical gradients, ih, it, ie, and ic, respectively.
equations is Poiseuille’s law for flow through a round
py
The terms in Eqs. (9.2) through (9.5) are identified capillary, which gives the average flow velocity, vave,
in Fig. 9.1 and in Table 9.1, which also shows analogs according to
between the various flow types. As long as the flow
rates and gradients are linearly related, the mathemat- p R2
vave ⫽
Co
ical treatment of each flow type is the same, and the i (9.8)
8 h
equations for flow of one type may be used to solve
problems of another type provided the property values where is viscosity, R is tube radius, and p is unit
and boundary conditions are properly represented. Two
well-known practical illustrations of this are the cor-
respondence between the Terzaghi theory for clay con- 1
This ‘‘law’’ was established empirically by Darcy based on the re-
solidation and one-dimensional transient heat flow, and sults of flow tests through sands. Its general validity for the descrip-
tion of hydraulic flow through most soil types has been verified by
the use of electrical analogies for the study of seepage many subsequent studies. Historical accounts of the development of
problems. Darcy’s law are given by Brown et al. (2003).
l
(m/s)
ria
Flow qh (m3 /s) qt (J/s) Current I (amp) jD (mol/s)
Flux qh /A (m3 /s/m2) qt /A (J/s/m2) I/A (amp/m2) JD ⫽ jD /A (mol s⫺1 m⫺2)
h T V c
Gradient ih ⫽ ⫺ (m/m) it ⫽ ⫺ (C/m) ie ⫽ ⫺ (v/m) ic ⫽ ⫺ (mol m⫺4)
x x x x
Conduction Darcy’s law Fourier’s law Ohm’s law Fick’s law
ate
h T V V c
qh ⫽ ⫺kh A qt ⫽ ⫺kt A I ⫽ ⫺e A⫽ JD ⫽ ⫺D A
x x x R x
Capacitance Coefficient of Volumetric heat Capacitance C (farads ⫽ Retardation factor, Rd
volume change C(J/ C/m3) coul/volt) (dimensionless)
dW a dQ
M⫽ dM⫽ w v ⫽ C⫽
dh 1⫹e dT
kh
cv
Continuity
W
t 冉冊 q
⫹ h ⫽0
A
u
t 冉冊
q
⫹ t ⫽0
A
Q
t
⫹ 冉冊
I
A
⫽0
(m)
t
⫹ JD ⫽ 0
Steady state 2qh ⫽ 0 2qt ⫽ 0 2I ⫽ 0 2JD ⫽ 0
h k 2h T k 2T V 2V c D* 2c
hte
Diffusion ⫽ h 2 ⫽ t 2 ⫽ ⫽
t M x t C x t C x2 t RD x2
冉 k
M
⫽ cv 冊 冉 冊
k
C
⫽a
rig
dius RH
where a is the cross-sectional area of the tube. For
other shapes of cross section, an equation of the same
flow channel cross-section area form will apply, differing only in the value of a shape
RH ⫽
wetted perimeter coefficient Cs, so
Co
is useful. p RH2
For a circular tube flowing full, q ⫽ Cs ia (9.11)
h
R2 R
RH ⫽ ⫽ (9.9) For a bundle of parallel tubes of constant but irreg-
2R 2 ular cross section contributing to a total cross-sectional
area A (solids plus voids), the area of flow passages Aƒ
so Poiseuille’s equation becomes filled with water is
l
ria
ate
Figure 9.2 Hydraulic conductivity values for several soils. Soil identification code: 1, com-
pacted caliche; 2, compacted caliche; 3, silty sand; 4, sandy clay; 5, beach sand; 6, compacted
Boston blue clay; 7, Vicksburg buckshot clay; 8, sandy clay; 9, silt—Boston; 10, Ottawa
sand; 11, sand—Gaspee Point; 12, sand—Franklin Falls; 13, sand–Scituate; 14, sand–Plum
dM
Island; 15, sand–Fort Peck; 16, silt—Boston; 17, silt—Boston; 18, loess; 19, lean clay; 20,
sand—Union Falls; 21, silt—North Carolina; 22, sand from dike; 23, sodium Boston blue
clay; 24, calcium kaolinite; 25, sodium montmorillonite; 26–30, sand (dam filter) (From
Lambe and Whitman (1969). Copyright 1969 by John Wiley & Sons. Reprinted with
permission from John Wiley & Sons.
hte
rig
py
Co
l
where P is the wetted perimeter, L is the length of flow sizes, an effective diameter Deff can be computed from
ria
channel in the direction of flow, Vs is the volume of the particle size distribution (Carrier, 2003) according
solids and S0 is the wetted surface area per unit volume to
of particles. The wetted surface area depends on the
100%
particle sizes and the soil fabric and may be considered Deff ⫽ (9.19)
兺(ƒi /Dave,i)
ate
as an effective surface area per unit volume of solids.
It is less than the total specific surface area of the soil
where fi is the fraction of particles between two sizes
since flow will not occur adjacent to all particle sur-
(Dli and Dsi) and Dave,i is the average particle size be-
faces.
tween two sizes (⫽D0.5 0.5
li Dsi ); S0 can also be estimated
For void ratio e and volume of solids Vs, the volume
from the specific surface area. Methods for nonplastic
of water Vw is
soils and clayey soils are given in Chapter 3 and also
dM
Vw ⫽ eVs S (9.14)
are summarized by Chapuis and Aubertin (2003). Var-
ious modifications for S0 are available to take irregular
particle shapes (Loudon, 1952; Carrier, 2003) into ac-
Equation (9.11) becomes count.
冉冊 冉冊 冉 冊
The Kozeny–Carman equation accounts well for the
p e dependency of permeability on void ratio in uniformly
q ⫽ Cs RH2 Snih A ⫽ Cs p RH2 S i A
1⫹e h graded sands and some silts; however, serious discrep-
hte
冉 冊冉 冊 冉 冊
ters or aggregates that have large intercluster pores and
rig
冉冊 冉 冊
factor, and noting that total surface area per unit vol-
1 e3 ume is inversely proportional to particle size, then
K ⫽ kh ⫽ (9.18)
p k0 T S 0 1 ⫹ e
2 2
Like the Kozeny–Carman equation, Eq. (9.20) de- and clays (Khalifa et al., 2002). These nonsteady and
scribes the behavior of cohesionless soils reasonably turbulent flow conditions are not treated herein.
well, but it is inadequate for clays. For a uniform sand As early as 1898, instances were cited in which hy-
with bulky particles and a given permeant, Eqs. (9.17) draulic flow velocity in fine-grained materials in which
and (9.20) indicate that kh should vary directly with laminar flow can be expected increased more than pro-
e3 /(1 ⫹ e) and D2s , and experimental observations sup- portionally with increases in gradient (King, 1898).
port this. The absence of water flow at finite hydraulic gradients
Despite the inability of the theoretical equations to in ceramic filters of 0.1-m average pore diameter
predict the hydraulic conductivity accurately in many was reported by Derjaguin and Krylov (1944). Oakes
cases, they do reflect the overwhelming importance of (1960) found no detectable flow through a 30-cm-long
pore size. Flow velocity depends on the square of pore suspension of 6 percent Wyoming bentonite subjected
l
radius, and hence the flow rate depends on radius to to a 50-cm head of water. Experiments by Miller and
ria
the fourth power. The specific surface in the Kozeny– Low (1963) led to the conclusion that there was a
Carman equation and the representative grain size term threshold gradient for flow through sodium montmo-
in Eq. (9.20) are both measures of pore size. All other rillonite. Flow rates through clay-bearing sandstones
factors equal, the hydraulic conductivity depends far were found to increase more than directly with gradi-
ate
more on the fine particles than on the large. A small ent up to gradients of 170 by von Englehardt and Tunn
percentage of fines can clog the pores of an otherwise (1955). Deviations from Darcy’s law in pure and nat-
coarse material and result in a manyfold lower hydrau- ural clays up to gradients of 900 were measured by
lic conductivity. On the other hand, the presence of Lutz and Kemper (1959). Apparent deviations from
fissures, cracks, root holes, and the like can result in Darcy’s law for flow in undisturbed soft clay are shown
enormous increases in the rate of water flow through
dM in Fig. 9.4.
an otherwise compact soil layer. The reported deviations from linearity between flow
Equation (9.20) predicts that the hydraulic conduc- rate and hydraulic gradient are most significant in the
tivity should vary with the cube of the degree of sat- lower range of gradients. Hydraulic gradients in the
uration, and some, but not all, experimental data field are seldom much greater than one. Thus, de-
support this, even in the case of fine-grained soils. viations from Darcy’s law, if real, could have very
Consideration of flow through unsaturated soils is important implications for the applicability of
given in Section 9.4. steady-state and transient flow analyses, including con-
hte
and pore sizes and flow rates are sufficiently great, then Three hypotheses have been proposed to account for
flow is turbulent, and Darcy’s law no longer applies. nonlinearity between flow velocity and gradient: (1)
Turbulent flow conditions are likely in flows through non-Newtonian water flow properties, (2) particle mi-
gravel and rockfill (Ahmed and Sunada, 1969; Ar- grations that cause blocking and unblocking of flow
bhabhirama and Dinoy, 1973; George and Hansen, passages, and (3) local consolidation and swelling that
py
1992; Hansen et al., 1995; Li et al., 1998).2 Some mod- is inevitable when hydraulic gradients are applied
ification of Darcy’s law is needed also to account for across a compressible soil. The apparent existence of
nonsteady and wave-induced flows through sands, silts, a threshold gradient below which flow was not de-
tected was attributed to a quasi-crystalline water struc-
Co
l
ria
ate
dM
Figure 9.4 Dependence of flow velocity on hydraulic gradient. Undisturbed soft clay from
Skå Edeby, Sweden (from Hansbo, 1973).
hte
obeyed exactly in several of these studies. Thus it is electrolyte concentration was changed from 0.6 to 0.1
unlikely that unusual water properties are responsible N NaCl are shown in Fig. 9.5. The cumulative through-
for non-Darcy flow behavior. put is the ratio of the total flow volume at any time to
On the other hand, particle migrations leading to the sample pore volume. The hydraulic conductivities
void plugging and unplugging, electrokinetic effects, for these materials ranged from more than 1 ⫻ 10⫺7 to
rig
and chemical concentration gradients can cause appar- less than 1 ⫻ 10⫺9 m/s.
ent deviations from Darcy’s law. Analysis of interpar- Practical Implications Evidence indicates that
ticle bond strengths in relation to the magnitude of Darcy’s law is valid, provided that all system variables
seepage forces shows that particles that are not partic- are held constant. However, unless fabric changes, par-
ipating in the load-carrying skeleton of a soil mass can ticle migrations, and internal void ratio redistributions
py
be moved under moderate values of hydraulic gradient. caused by effective stress and chemical changes can
Soils with open, flocculated fabrics and granular soils be shown to be negligible, hydraulic conductivity mea-
with a relatively low content of fines appear particu- surements in the laboratory should be made under con-
larly susceptible to the movement of fine particles dur- ditions of temperature, pressure, hydraulic gradient,
Co
l
ria
Figure 9.5 Reduction in hydraulic conductivity as a result of internal swelling (from Hard-
castle and Mitchell, 1974).
ate
gradient. If hydraulic gradients are low enough to du- u0
i⫽ p (9.24)
plicate those in most field situations, then the labora-
dM wH
tory testing time usually becomes unacceptably long.
In such cases, tests over a range of gradients are de- The real gradient for any layer thickness or loading
sirable in order to assess the stability of the soil struc- intensity can be obtained by using actual values of u0
ture against changes due to seepage forces. and H and the appropriate value of p from Fig. 9.6.
Similarly, the gradients that are developed in labo- For small values of u0 / w H, as is the case in the
ratory consolidation tests on thin samples are many field, for example, for u0 ⫽ 50 kPa, H ⫽ 5m, then
times greater than exist in thick layers of the same clay u0 / w H ⫽ 1, and the field gradients are low throughout
hte
in the field. The variation of hydraulic gradient i with most of the layer thickness during the entire consoli-
time factor T during one-dimensional consolidation ac- dation process. On the other hand, for a laboratory
cording to the Terzaghi theory is shown in Fig. 9.6. sample of 10 mm thickness and the same stress in-
The solution of the Terzaghi equation gives excess crease, u0 / w H is 500, and the hydraulic gradients are
pore pressure u as a function of position (z/H) and very large. In this case a gradient-dependent hydraulic
time factor conductivity could be the cause of significant differ-
rig
冉冊 冉 冊
Anisotropy
冘
⬁
Co
u 2u0 Mz ⫺M2T
i⫽ ⫽ cos e (9.22) Anisotropic hydraulic conductivity results from both
z w wH m⫽0 H preferred orientation of elongated or platy particles and
stratification of soil deposits. Ratios of horizontal-to-
If a parameter p is defined by vertical hydraulic conductivity from less than 1 to
more than 7 were measured for undisturbed samples
冘 cos冉MzH 冊e
⬁ of several different clays (Mitchell, 1956). These ratios
⫺M2T
p⫽2 (9.23) correlated reasonably well with preferred orientation of
m⫽0
the clay particles, as observed in thin section. Ratios
of 1.3 to 1.7 were measured for kaolinite consolidated
Eq. (9.22) becomes one dimensionally from 4 to 256 atm, and 0.9 to 4.0
l
ria
ate
dM
Figure 9.6 Hydraulic gradients during consolidation according to the Terzaghi theory.
were measured for illite and Boston blue clay consol- drains are used. On the other hand, lateral drainage
hte
idated over a pressure range up to more than 200 atm beneath a loaded area may not be greatly influenced
(Olsen, 1962). A ratio of approximately 2 was mea- by a high ratio of horizontal to vertical conductivity if
sured for kaolinite over a range of void ratios corre- the width of loaded area is large compared to the thick-
sponding to consolidation pressures up to 4 atm ness of the drainage layer.
(Morgenstern and Tchalenko, 1967b). Thus, an average
hydraulic conductivity ratio of about 2 as a result of
rig
microfabric anisotropy may be typical for many clays. Fabric and Hydraulic Conductivity
Large anisotropy in hydraulic conductivity as a re- The theoretical relationships developed earlier in this
sult of stratification of natural soil deposits or in earth- section indicate that the flow velocity should depend
work compacted in layers is common. Varved clays on the square of the pore radius, and the flow rate is
have substantially greater hydraulic conductivity in the proportional to the fourth power of the radius. Thus,
py
horizontal direction than in the vertical direction owing fabrics with a high proportion of large pores are much
to the presence of thin silt layers between the thin clay more pervious than those with small pores. For ex-
layers. The ratio of horizontal values to vertical values ample, remolding several undisturbed soft clays re-
determined in the laboratory, rk, is 10 5 for Con- duced the hydraulic conductivity by as much as a
Co
necticut Valley varved clay (Ladd and Wissa, 1970). factor of 4, with an average of about 2 (Mitchell,
Similar values were measured for the varved clay in 1956). This reduction results from the breakdown of a
the New Jersey meadows. Values less than 5 were mea- flocculated open fabric and the destruction of large
sured for New Liskeard, Ontario, varved clay (Chan pores.
and Kenney, 1973). An illustration of the profound influence of com-
The practical importance of a high hydraulic con- paction water content on the hydraulic conductivity of
ductivity in the horizontal direction depends on the dis- fine-grained soil is shown in Fig. 9.7. All samples were
tance to a drainage boundary and the type of flow. For compacted to the same density. For samples compacted
example, the rate of groundwater flow will clearly be using the same compactive effort, curves such as those
affected, as will the rate of consolidation when vertical in Fig. 9.8 are typical. For compaction dry of optimum,
l
ria
ate
dM
hte
rig
tion water content for samples of silty clay prepared to con- Three levels of fabric are important when consid-
stant density by kneading compaction. ering the hydraulic conductivity of finer-grained soils.
The microfabric consists of the regular aggregations of
particles and the very small pores, perhaps with sizes
Co
clay particles and aggregates are flocculated, the resis- up to about 1 m, between them through which very
tance to rearrangement during compaction is high, and little fluid will flow. The minifabric contains these ag-
a fabric with comparatively large pores is formed. For gregations and the interassemblage pores between
higher water contents, the particle groups are weaker, them. The interassemblage pores may be up to several
and fabrics with smaller average pore sizes are formed. tens of micrometers in diameter. Flows through these
Considerably lower values of hydraulic conductivity pores will be much greater than through the intraag-
are obtained wet of optimum in the case of kneading gregate pores. On a larger scale, there may be a ma-
compaction than by static compaction (Fig. 9.8) be- crofabric that contains cracks, fissures, laminations, or
cause the high shear strains induced by the kneading root holes through which the flow rate is so great as
compaction method break down flocculated fabric to totally obscure that through the other pore space
units. types.
l
The wide range of values of hydraulic conductivity The sizes of clusters depend on the mineralogical
ria
of compacted fine-grained soils that results from the and pore fluid compositions and the formational proc-
large differences in fabric associated with compaction ess. Conditions that favor aggregation of individual
to different water contents and densities is illustrated clay plates produce larger clusters than deflocculating,
by Fig. 9.9. The grouping of contours means that se- dispersing environments. There is general consistency
lection of a representative value for use in a seepage with the interparticle double-layer interactions de-
ate
analysis is difficult. In addition, if it is required that scribed in Chapter 6. When a fine-grained soil is
the hydraulic conductivity of earthwork not exceed a sedimented in or mixed with waters of different
certain value, such as may be the case for a clay liner electrolyte concentration or type or with fluids of dif-
for a waste pond, then specifications must be carefully ferent dielectric constants, quite different fabrics result.
drawn. In so doing, it must be recognized also that
dM This explains why the / term in Eqs. (9.18) and
other properties, such as strength, also vary with com- (9.20) is inadequate to account for pore fluid differ-
paction water content and density and that the com- ences, unless comparisons are made using samples
paction conditions that are optimal for one property having identical fabrics. This will only be the case
may not be suitable for the other. A procedure for the when a pore fluid of one type replaces one of another
development of suitable specifications for compacted type without disturbance to the soil.
clay liners is given by Daniel and Benson (1990). The cluster model developed by Olsen (1962) ac-
The primary reason equations such as (9.18) and counts for discrepancies between the predicted and
hte
(9.20) fail to account quantitatively for the variation of measured variations in flow rates through different
the hydraulic conductivity of fine-grained soils with soils. The following equation can be derived for the
change in void ratio is unequal pore sizes (Olsen, ratio of estimated flow rate for a cluster model, qCM to
1962). A typical soil has a fabric composed of small the flow rate predicted by the Kozeny–Carman equa-
tion (9.18) qKC:
rig
py
Co
Figure 9.9 Contours of constant hydraulic conductivity for Figure 9.10 Cluster model for permeability prediction (after Ol-
silty clay compacted using kneading compaction. sen, 1962).
冉 冊
the variations of ec with eT that accompany compres-
sion and rebound. Olsen (1962) considered the relative z
vi ⫽ ⫺k(S) ⫹ (9.26)
compressibility of individual clusters and cluster as- xi xi
semblages. The compressibility of individual clusters
is small at high total void ratios, so compression is where k(S) is saturation-dependent hydraulic conduc-
accompanied by reduction in the intercluster pore tivity, is the matric suction equivalent head (L), and
sizes, but with little change in intracluster void ratio. z/ xi is the unit gravitational vector measured upward
l
This assumption is supported by the microstructure in direction z (1.0 if xi is the direction of gravity z).
ria
studies of Champlain clay by Delage and Lefebvre When percolating water infiltrates vertically into dry
(1984) Thus, the actual hydraulic conductivity de- soil, the hydraulic gradient near the sharp wetting front
creases more rapidly with decreasing void ratio during can be very large because of a large value of the /
compression than predicted by the Kozeny–Carman x term. However, the wetting front becomes less sharp
ate
equation until the intercluster pore space is comparable as the infiltration proceeds and the gravity term then
to that in a system of closely packed spheres, when dominates. The hydraulic gradient then is close to one
the clusters themselves begin to compress. Further de- and the magnitude of flux is equal to the hydraulic
creases in porosity involve decreases in both ec and eT. conductivity k(S).
As the intercluster void ratio now decreases less rap- Using Eq. (9.26), the equation of mass conservation
idly, the hydraulic conductivity decreases at a slower
dM becomes
rate with decreasing porosity than predicted by the
Kozeny–Carman equation. During rebound increase in
porosity develops mainly by swelling of the clusters,
whereas the flow rate continues to be controlled pri-
(nS)
t
⫽
xi 冋 冉
k(S)
xi
⫹
z
xi 冊册 ⫹
R
w
(9.27)
冋 冉 冊册
ing in terms of surface properties and solution com-
S z
position derived using DLVO theory (see Chapter 6), n ⫽ k() ⫹
combined with assumed geometrical representations of t xi xi xi
rig
negative. However, the hydraulic conductivity is not ing equation can be solved for (often numerically by
constant and depends on the amount and connectivity the finite difference or finite element method).
of water in the pores. For instance, Eq. (9.20) predicts The hydraulic conductivity of unsaturated soils can
that hydraulic conductivity should vary as the cube of be a function of saturation, water content, matric suc-
the degree of saturation.3 This relationship has been tion, or others. Measured hydraulic conductivities of
well-graded sand and clayey sand as a function of (a)
matric suction and (b) saturation ratio are shown in
3
The hydraulic conductivity can also be a function of volumetric Fig. 9.11. Both figures are related to each other, as the
moisture content or matric suction . These variables are related
to each other by the soil–water characteristic curve as described in matric suction is a function of saturation ratio by the
Chapter 7. soil moisture characteristic curve as described in Sec-
l
Matric Suction (kPa) Saturation (%)
ria
(a) (b)
1.E+00
1.E+00
Sand 1.E-02
Relative Permeability kr
1.E-02
Relative Permeability kr
Clayey Sand
1.E-04
ate
1.E-04
1.E-06
1.E-06
1.E-08 1.E-08
1.E-10 1.E-10
80 100
Matric Suction (kPa) Saturation (%)
(c) (d)
Figure 9.11 Hydraulic conductivity of partially saturated sand and clayey sand as a function
of matric suction and degree of saturation (from Stephens, 1996).
hte
tion 7.12. Various methods to measure the hydraulic meability, no interconnected path for the permeating
conductivity of unsaturated soils are available (Klute, fluid) to 1 (⫽ permeating fluid at full saturation). The
1986; Fredlund and Rahardjo, 1993). However, the equation can be used for a nonwetting fluid (e.g., air)
rig
measurement in unsaturated soils is more difficult to by substituting the values of and of the nonwetting
perform than in saturated soils because the hydraulic fluid.
conductivity needs to be determined under controlled The data in Fig. 9.11a and 9.11b can be replotted
water saturation or matric suction conditions. as the relative permeability against matric suction in
A general expression for the hydraulic conductivity Fig. 9.11c and against saturation ratio in Fig. 9.11d.
py
k of unsaturated soils can be written as The two different curves in Fig. 9.11d clearly show
that kr ⫽ S3 derived from Eq. (9.20) is not universally
g applicable. At very low water contents, the water in
k ⫽ krK ⫽ kr ks (9.29)
the pores becomes disconnected as described in Chap-
Co
for hydraulic conductivity as a function of pore sizes in which k(i) is the calculated hydraulic conductivity
for an isotropic material: for a specified water content i; is i the last water con-
tent class on the wet end, for example, i ⫽ 1 denotes
n2 r 21 ⫹ 3r 22 ⫹ 5r 23 ⫹ ⫹ (2m ⫺ 1)r 2m the pore class corresponding to the saturated water
K⫽ content S, and i ⫽ l denotes the pore class correspond-
m2 8
ing to the lowest water content L for which hydraulic
(9.30) conductivity is calculated; ks /ksc is the matching factor,
defined as the measured saturated hydraulic conductiv-
in which K is the specific hydraulic conductivity (per- ity divided by the calculated saturated hydraulic con-
meability) (L2), n is the porosity, m is the total number ductivity; and l is the total number of pore classes (a
of pore classes, and ri is the mean radius of the pores pore class is a pore size range corresponding to a water
l
in pore class i. Pore sizes can be measured from data content increment) between ⫽ L and S. Thus
ria
on the amount of water withdrawn as the suction on
the soil is progressively increased. Using the capillary m S
equation, the radius of the largest water-filled pore un- ⫽ (9.35)
l S ⫺ L
der a suction of (L) is given by
ate
A constant value of l is used at all water contents, and
2
r⫽ (9.31) the value of l establishes the number of pore classes
wg for which ⫺2 terms are included in the calculation at
j
saturation. Other pore size distribution models for un-
in which is the surface tension of water, w is the saturated soils are available, and an excellent review
density of water, and g is the acceleration of gravity.
dM of these models is given by Mualem (1986).
As it is usually more convenient to use moisture suc- Equation (9.34) can be written in an integration form
tion than pore radius, Eq. (9.29) can be rewritten as as (after Fredlund et al., 1994)
K⫽
2 n2 ⫺2
[ ⫹ 32⫺2 ⫹ 5⫺2
22wg2 m2 1 3 k() ⫽
ks 2 Sp
ksc 2wg
冕
L
⫺x
2(x)
dx (9.36)
⫹ ⫹ (2m ⫺ 1)⫺2
m ] (9.32)
hte
k⫽ [ ⫹ 3⫺2
2 ⫹ 53
⫺2
sizes (Fredlund et al., 1994).
2wg m2 1
From Eq. (9.36), the relative permeability kr is a
⫹ ⫹ (2m ⫺ 1)⫺2
m ] (9.33) function of water content as follows:
冒冕 S
py
equate the calculated and measured hydraulic conduc- the residual water content r. If the moisture content
tivities. Matching at full saturation is preferable to –suction relationship (or the soil–water character-
matching at a partial saturation point because it is sim- istic curve) is known, the relative permeability kr can
pler and gives better results. Rewriting Eq. (9.33) and be computed from Eq. (9.37) by performing a numer-
introducing a matching factor gives ical integration. The hydraulic conductivity k is then
estimated from Eq. (9.29) with the knowledge of sat-
冘 [(2j ⫹ 1 ⫺ 2i)
ks 2 2S l urated hydraulic conductivity ks.
k(i) ⫽ ⫺2
] The use of the soil–water characteristic curve to es-
ksc 2wg m2 j⫽1
j
timate the hydraulic conductivity of unsaturated soils
(i ⫽ 1, 2, . . . , l) (9.34) is attractive because it is easier to determine this curve
r
s
r
d
() 冊
2
(9.38)
through for a given set of fine and coarse soil hydraulic
properties and interface inclination (Ross, 1990; Steen-
huis et al., 1991; Selkar, 1997; Webb, 1997).
Capillary barriers have received increased attention
where q describes the degree of connectivity between as a means for isolating buried waste from ground-
l
the water-conducting pores. Mualem (1976) states that water flow and as part of landfill cover systems in dry
q ⫽ 0.5 is appropriate based on permeability measure-
ria
climates (Morris and Stormont, 1997; Selkar, 1997;
ments on 45 soils. van Genuchten et al. (1991) substi- Khire et al., 2000). The barrier can be used to divert
tuted the soil–water characteristic equation (7.52) into the flow laterally along an interface and/or to store
Eq. (9.38) and obtained the following closed-form so- infiltrating water temporarily in the fine layer so that
lution4: it can be removed ultimately by evaporation and tran-
ate
冉 冊再 冋 冉 冊 册冎
p 1/m m 2
spiration. Capillary barriers are constructed as simple
⫺ r ⫺ r two-layer systems of contrasting particle size or mul-
kr() ⫽ 1⫺ 1⫺
S ⫺ r S ⫺ r tiple layers of fine- and coarse-grained soils. If the
(9.39) thickness of the overlying fine layer is too small, cap-
dM illary diversion is reduced because of the confining
flow path in the fine layer. The minimum effective
Both Eq. (9.39) as well as Eq. (9.37) using the soil– thickness is several times the air-entry head of the fine
water characteristic curve by Fredlund and Xing soil (Warrick et al., 1997; Smersrud and Selker, 2001).
(1994) give good predictions of measured data as Khire et al. (2000) stress the importance of site-specific
shown in Fig. 9.12. metrological and hydrological conditions in determin-
The two hydraulic conductivity–matric suction ing the storage capacity of the fine layer. The soil for
curves shown in Fig. 9.11a cross each other at a matric the underlying coarse layer should have a very large
hte
suction value of approximately 50 kPa (or 5 m above particle size contrast with the fine soil, but fines
the water table under hydrostatic condition). Below migrations into the coarse sand should be avoided.
this value, the hydraulic conductivity of sand is larger Smesrud and Sekler (2001) suggest the d50 particle size
than that of the clayey sand. However, as the matric ratio of 5 to be ideal. The thickness of the coarse sand
suction increases, the water in the sand drains rapidly layer does not need to be great, as the purpose of the
toward its residual value, giving a very low hydraulic layer is simply to impede the downward water migra-
rig
matric suction at the interface is larger than 50 kPa, Heat flow through soil and rock is almost entirely by
and the water infiltrating downward through the finer conduction, with radiation unimportant, except for sur-
clayey sand cannot enter into the coarser sand layer face soils, and convection important only if there is a
because the underlying sand layer is less permeable high flow rate of water or air, as might possibly occur
Co
than the overlying clayey sand. The water will instead through a coarse sand or rockfill. The thermal conduc-
move laterally along the bedding interface. This phe- tivity controls heat flow rates. Conductive heat flow is
nomenon is called a capillary barrier (e.g., Zaslavsky primarily through the solid phase of a soil mass. Values
and Sinai, 1981; Yeh et al., 1985; Miyazaki, 1988). of thermal conductivity for several materials are listed
The barrier will be maintained as long as the lateral in Table 9.2. As the values for soil minerals are much
discharge along the interface (preferably inclined) is higher than those for air and water, it is evident that
the heat flow must be predominantly through the sol-
ids. Also included in Table 9.2 are values for the heat
4
m ⫽ 1 ⫺ 1 / n is assumed (van Genuchten et al., 1991). See Eq. capacity, volumetric heat, heat of fusion, and heat of
(7.52). vaporization of water. The heat capacity can be used
10
Predicted coefficient
0.01 of permeability
(wetting)
l
Measured coefficient
ria
of permeability
(drying)
0.001
Measured coefficient
of permeability
(wetting)
0.0001
ate
20 30 40 50 60
Volumetric Water Content
(a)
Figure 9.12 Comparisons of predicted and measured relationships between hydraulic con-
ductivity and volumetric water content for two soils. (a) By Eq. (9.37) with the measured
data for Guelph loam (from Fredlund et al., 1994) and (b) by Eq. (9.39) with the measured
dM
data for crushed Bandelier Tuff (van Genuchten et al., 1991).
to compute the volumetric heat using the simple rela- to carry heat away from buried electrical transmission
tionships for frozen and unfrozen soil given in the ta- cables, or as an insulating material, for example, for
ble. Volumetric heat is needed for the analysis of many underground storage of liquefied gases. The water con-
hte
types of transient heat flow problems. The heat of fu- tent below which the thermal resistivity begins to rise
sion is used for analysis of ground freezing and thaw- with further drying is termed the critical water content,
ing, and the heat of vaporization applies to situations and below this point the system is said to have lost
where there are liquid to vapor phase transitions. thermal stability (Brandon et al., 1989).
The denser a soil, the higher is its composite thermal The following factors influence the thermal resistiv-
conductivity, owing to the much higher thermal con- ity of partly saturated soils (Brandon and Mitchell,
rig
which may be used for estimates of the thermal con- Dry Density The higher the dry density of a soil,
ductivity for many cases. If a more soil-specific value the higher is the thermal conductivity.
is needed, they may be measured in the laboratory us- Gradation Well-graded soils conduct heat better
ing the thermal needle method (ASTM, 2000). More than poorly graded soils because smaller grains
Co
detailed treatment of methods for the measurement of can fit into the interstitial spaces between the
the thermal conductivity of soils are given by Mitchell larger grains, thus increasing the density and the
and Kao (1978) and Farouki (1981, 1982). mineral-to-mineral contact.
The relationship between thermal resistivity (inverse Compaction Water Content Some sands that com-
of conductivity) and water content for a partly satu- pacted wet and then dried to a lower water content
rated soil undergoing drying is shown in Fig. 9.14. If have significantly higher thermal conductivity
drying causes the water content to fall below a certain than when compacted initially at the lower water
value, the thermal resistivity increases significantly. content.
This may be important in situations where soil is used Time Sands containing high percentages of silica,
as either a thermally conductive material, for example, carbonates, or other materials that can develop ce-
Thermal
Conductivity Material Btu/h/ft2 / F/ft W/m/K
Air 0.014 0.024
Water 0.30 0.60
Ice 1.30 2.25
Snow
(100 kg m⫺3) 0.03 0.06
(500 kg m⫺3) 0.34 0.59
l
Shale 0.90 1.56
ria
Granite 1.60 2.76
Concrete 1.0 1.8
Copper 225 389
Soil 0.15–1.5 (⬇1.0) 0.25–2.5 (⬇1.7)
Polystyrene 0.015–0.035 0.03–0.06
ate
Heat Capacity Material Btu/lb/ F kJ/kg/K
Water 1.0 4.186
Ice 0.5 2.093
dM Snow
(100 kg m⫺3) 0.05 0.21
(500 kg m⫺3) 0.25 1.05
Minerals 0.17 0.710
Rocks 0.20–0.55 0.80–2.20
Volumetric Heat Material Btu/ft3 / F kJ/m3 /K
Unfrozen Soil d (0.17 ⫹ w/100) d (72.4 ⫹ 427w/100)
hte
Soil
Frozen soil d (0.17 ⫹ 0.5w/100) d (72.4 ⫹ 213w/100)
Snow
(100 kg m⫺3) 3.13 210
(500 kg m⫺3) 15.66 1050
rig
l
cal equations and theoretical expressions based on
ria
simplified models may provide satisfactory results,
depending on the particular soil and conditions. They
differ in assumptions about the possible flow paths for
electric current through a soil–water matrix, the path
lengths and their relative importance, and whether
ate
charged particle surfaces contribute to the total current
flow.
T
F⫽ ⫽ W (9.41)
W T
F ⫽ n⫺m (9.42)
py
and water content for a compacted sand. and ‘‘apparent’’ formation factor at saturation less than
100 percent is
l
an equation for the formation factor as a function of
ria
the porosity n and the tortuosity T ⫽ Le /L is tem is equivalent to two electrical resistors in parallel
(Waxman and Smits, 1968). The result is that the total
electrical conductivity T is
T2
F⫽ (9.44)
n
T ⫽ X(W ⫹ s) (9.49)
ate
For S ⬍ 1, and assuming that the area available for
electrical flow is nSA, then F ⫽ T 2 /nS. In principle, if in which s is a surface conductivity term, and X is a
F is measured for a given soil and n is known, a value constant analogous to the reciprocal of the formation
of tortuosity can be calculated to use in the Kozeny– factor that represents the internal geometry.
Carman equation for hydraulic conductivity.
dM This approach yields better fits of T versus W data
Cluster Model As discussed earlier in connection for clay-bearing soils. However, it assumes a constant
with hydraulic conductivity, the cluster model (Olsen, value for the contribution of the surface ions that is
1961, 1962) shown in Fig. 9.10 assumes unequal pore independent of the electrolyte concentration in the pore
sizes. Three possible paths for electrical current flow water, and it fails to include a contribution for the sur-
can be considered: (1) through the intercluster pores, face conductance and pore water conductance in a se-
(2) through the intracluster pores, and (3) alternately ries path.
through inter- and intracluster pores. On this basis the Three-Element Network Model A third path is in-
hte
following equations for formation factor as a function cluded in this formulation that considers flow along
of the cluster model parameters can be derived (Olsen, particle surfaces and through pore water in series in
1961): addition to the paths included in the two-parallel-
resistor model. The flow paths and equivalent electrical
冉 冊冉 冊
circuit are shown in Fig. 9.15. Analysis of the electrical
1 ⫹ eT 1
F ⫽ T2 (9.45) network for determination of T gives
rig
eT ⫺ ec 1⫹X
aWs
X⫽Y⫹Z (9.46) T ⫽ ⫹ bs ⫹ cW (9.50)
(1 ⫺ e)W ⫹ es
py
l
ria
ate
Figure 9.15 Three-element network model for electrical conductivity: (a) current flow paths
and (b) equivalent electrical circuit.
dM
for conductivity values in the range of about 0.3 to 0.6 frequency, and 0 is the permittivity of vacuum [8.85
S/m. ⫻ 10⫺12 C2 /(Nm2)]. The frequency-dependent effective
conductivities of deionized water and kaolinite–water
Alternating Current Conductivity and Dielectric mixtures at two different water contents (0.2 and 33
Constant percent) are shown in Fig. 9.18a. The complicated in-
teractions of different polarization mechanisms are re-
The electrical response of a soil in an AC field is fre-
hte
σeff (S/m)
33%
10–4 No Data
Available
0.2%
10–6
100 102 104 106 108 1010
Frequency (Hz)
(a)
l
ria
Electrode Effect
106 33% Deionized Water
ate
100
accumulate at the electrode–specimen interface (Klein 100 102 104 106 108 1010
and Santamarina, 1997). Similarly to the observations Frequency (Hz)
made for the effective conductivities, the real permit-
dM (b)
tivity values of the mixtures show complex trends of
Figure 9.18 (a) Conductivity and (b) relative permittivity as
frequency dependency.
a function of frequency for deionized water and kaolinite at
For analysis of AC conductivity and dielectric con- water contents of 0.2 and 33 percent (from Santamarina et
stant as a function of frequency in an AC field, Smith al., 2001).
and Arulanandan (1981) modified the three-element
model shown in Fig. 9.15 by adding a capacitor in
parallel with each resistor. The resulting equations can
hte
Figure 9.17 Frequency ranges associated with different polarization mechanisms (from San-
tamarina et al., 2001).
ductivity and apparent dielectric constant by computer Table 9.3 Self-Diffusion Coefficients for Ions at
optimization of geometrical and compositional param- Infinite Dilution in Water
eters. The resulting parameter values are useful for
characterizing mineralogy, porosity, and fabric. More Anion D0 ⫻ 1010(m2 /s) Cation D0 ⫻ 1010(m2 /s)
detailed discussions on electrical models, data inter- (1) (2) (3) (4)
pretation, and correlations with soil properties are
given by Santamarina et al. (2001). OH⫺ 52.8 H⫹ 93.1
F⫺ 14.7 Li⫹ 10.3
Cl⫺ 20.3 Na⫹ 13.3
Br⫺ 20.8 K⫹ 19.6
9.7 DIFFUSION I⫺ 20.4 Rb⫹ 20.7
HCO3⫺ Cs⫹
l
11.8 20.5
ria
Chemical transport through sands is dominated by ad- NO3⫺ 19.0 Be2⫹ 5.98
vection, wherein dissolved and suspended species are SO42⫺ 10.6 Mg2⫹ 7.05
carried with flowing water. However, in fine-grained CO32⫺ 9.22 Ca2⫹ 7.92
soils, wherein the hydraulic flow rates are very small, — — Sr2⫹ 7.90
for example, kh less than about 1 ⫻ 10⫺9 m/s, chemical — — Ba2⫹ 8.46
ate
diffusion plays a role and may become dominant when — — Pb2⫹ 9.25
kh becomes less than about 1 ⫻ 10⫺10 m/s. Fick’s law, — — Cu2⫹ 7.13
Eq. (9.5), is the controlling relationship, and D(L2T⫺1), — — Fe2⫹a 7.19
the diffusion coefficient, is the controlling parameter. — — Cd2⫹a 7.17
Diffusive chemical transport is important in clay — — Zn2⫹ 7.02
barriers for waste containment, in some geologic — — Ni2⫹a 6.79
dM
processes, and in some forms of chemical soil — — Fe3⫹a 6.07
stabilization. Comprehensive treatments of the diffu- — — Cr3⫹a 5.94
sion process, values of diffusion coefficients and — — Al3⫹a 5.95
methods for their determination, and applications, a
Values from Li and Gregory (1974). Reprinted with
especially in relation to chemical transport and waste
permission from Geochimica et Cosmochimica Acta, Vol.
containment barrier systems, are given by Quigley et
38, No. 5, pp. 703–714. Copyright 1974, Pergamon
hte
slowing down the faster and speeding up the slower. The diffusion coefficient could increase with time of
This may be seen in Table 9.4, which contains values flow through a soil as a result of such processes as
of some limiting free solution diffusion coefficients for (Quigley, 1989):
some simple electrolytes. 1. K⫹ fixation by vermiculite, which would decrease
Co
Diffusion through soil is slower and more complex the cation exchange capacity and increase the
than diffusion through a free solution, especially when free water pore space
adsorptive clay particles are present. There are several 2. Electrical imbalances that act to pull cations or
reasons for this (Quigley, 1989): anions
3. The attainment of adsorption equilibrium, thus
1. Reduced cross-sectional area for flow because of eliminating retardation of some species
the presence of solids
2. Tortuous flow paths around particles In an attempt to take some of these factors, espe-
3. The influences of electrical force fields caused by cially geometric tortuosity of interconnected pores,
the double-layer distributions of charges into account, an effective diffusion coefficient D* is
l
NaBr 16.25
An error function solution for Eq. (9.53) (Ogata,
ria
NaI 16.14
1970; Freeze and Cherry, 1979), for the case of one-
KCl 19.93
dimensional diffusion from a layer at a constant source
KBr 20.16
concentration C0 into a layer having a sufficiently low
KI 19.99
initial concentration that it can be taken as zero at t ⫽
CsCl 20.44
ate
0, is
CaCl2 13.35
BaCl2 13.85
C x x
⫽ erfc ⫽ 1 ⫺ erf (9.54)
Reported by Shackelford and Daniel, 1991a after Rob- C0 2兹D*t 2兹D*t
inson and Stokes, 1959. Reprinted from the Journal of
Geotechnical Engineering, Vol. 117, No. 3, pp. 467–484.
dM where C is the concentration at any time at distance ⫻
Copyright 1991. With permission of ASCE. from the source.
Curves of relative concentration as a function of
depth for different times after the start of chloride dif-
used. Several definitions have been proposed (Shack- fusion are shown in Fig. 9.19a (Quigley, 1989). An
elford and Daniel, 1991a) in which the different factors effective diffusion coefficient for chloride of 6.47 ⫻
are taken into account in different ways. Although 10⫺10 m2 /s was assumed. Also shown (Fig. 9.19b) is
the migration velocity of the C/C0 front within the soil
hte
in which a is an ‘‘apparent tortuosity factor’’ that takes When there are adsorption–desorption reactions,
several of the other factors into account, and use values chemical reactions such as precipitation–solution, ra-
of D* measured under representative conditions. The dioactive decay, and/or biological processes occurring
effective coefficient for diffusion of different chemicals during diffusion, the analysis becomes more complex
through saturated soil is usually in the range of about than given by the foregoing equations. For adsorption–
py
2 ⫻ 10⫺10 to 2 ⫻ 10⫺9 m2 /s, although the values can desorption reactions and the assumption that there is
be one or more orders of magnitude lower in highly linearity between the amount adsorbed and the equi-
compacted clays and clays, such as bentonite, that can librium concentration, Eq. (9.53) is often written as
behave as semipermeable membranes (Malusis and
Co
l
chemical flows are given in Table 9.5. These ranges
ria
are for fine-grained soils, that is, silts, silty clays,
clayey silts, and clays. They are for full saturation;
values for partly saturated soils can be much lower.
Also listed in Table 9.5 are values for electroosmotic
ate
conductivity, osmotic efficiency, and ionic mobility.
These properties are needed for analysis of coupling
of hydraulic, electrical, and chemical flows, and they
are discussed further later.
dM 9.9 SIMULTANEOUS FLOWS OF WATER,
CURRENT, AND SALTS THROUGH
SOIL-COUPLED FLOWS
Usually there are simultaneous flows of different types
through soils and rocks, even when only one type of
hte
different times and (b) velocity of migration of the front hav- there are nonuniform distributions of cations and ani-
ing a concentration C / C0 of 0.5. ons within soil pores resulting from the attraction of
cations to and repulsion of anions from the negatively
charged particle surfaces. The net negativity of clay
particles is caused primarily by isomorphous substitu-
py
volumetric water content, that is, the volume of water tions within the crystal structure, as discussed in Chap-
divided by the total volume (porosity in the case of a ter 3, and the ionic distributions in the pore fluid are
saturated soil), and Kd is the distribution coefficient. described in Chapter 6. Because of the small pore sizes
The distribution coefficient defines the amount of a in fine-grained soils and the strong local electrical
Co
given constituent that is adsorbed or desorbed by a soil fields, clay layers exhibit membrane properties. This
for a unit increase or decrease in the equilibrium con- means that the passage of certain ions and molecules
centration in solution. Other reactions influencing the through the clay may be restricted in part or in full at
amount in free solution relative to that fixed in the soil both microscopic and macroscopic levels.
(e.g., by precipitation) may be included in Kd, depend- Owing to these internal nonhomogeneities in ion
ing on the method for measurement and the conditions distributions, restrictions on ion movements caused by
being modeled. Distribution coefficients are usually de- electrostatic attractions and repulsions, and the de-
termined from adsorption isotherms, and they may be pendence of these interactions on temperature, a vari-
constants for a given soil–chemical system or vary ety of microscopic and macroscopic effects may be
with concentration, pH, and temperature. More de- observed when a wet soil mass is subjected to flow
l
Electro osmotic ke
ria
conductivity
Diffusion D m2 s⫺1 2 ⫻ 10⫺10 2 ⫻ 10⫺9
coefficient
Osmotic " — 0 1.0
efficiencyb
ate
Ionic mobility u m2 s⫺1 V⫺1 3 ⫻ 10⫺9 1 ⫻ 10⫺8
a
The above values of flow coefficients are for saturated soil. They may be
much less in partly saturated soil.
b
0 to 1.0 is the theoretical range for the osmotic efficiency coefficient. Values
greater than about 0.7 are unlikely in most fine-grained materials of geotechnical
dM
interest.
gradients of different types. A gradient of one type Xj flow is developed later. Electroosmosis has been used
can cause a flow of another type Ji, according to for many years as a means for control of water flow
and for consolidation of soils. Chemicalosmosis, the
hte
terms commonly used to describe them.6 creation of frozen soil barriers in the presence of flow-
Of the 12 coupled flows shown in Table 9.6, several ing groundwater. Electrically driven heat flow, the Pel-
are known to be significant in soil–water systems, at tier effect, and chemically driven heat flow, the Dufour
least under some conditions. Thermoosmosis, which is effect, are not known to be of significance in soils;
py
water movement under a temperature gradient, is im- however, they appear not to have been studied in any
portant in partly saturated soils, but of lesser impor- detail in relation to geotechnical problems.
tance in fully saturated soils. Significant effects from Streaming current, the term applied to both hydrau-
thermally driven moisture flow are found in semiarid lically driven electrical current and ion flows, has im-
and arid areas, in frost susceptible soils, and in expan- portance to both chemical flow through the ground
Co
sive soils. An analysis of thermally driven moisture (advection) and the development of electrical poten-
tials, which may, in turn, influence both fluid and ion
flows as a result of additional coupling effects. The
complete roles of thermoelectricity and diffusion and
6
Mechanical coupling also occurs in addition to the hydraulic, ther- membrane potentials are not yet known; however, elec-
mal, electrical, and chemical processes listed in Table 9.6. A common
manifestation of this in geotechnical applications is the development trical potentials generated by temperature and chemical
of excess pore pressure and the accompanying fluid flow that result gradients are important in corrosion and in some
from a change in applied stress. This type of coupling is usually most groundwater flow and stability problems.
easily handled by usual soil mechanics methods. A few other types
of mechanical coupling may also exist in soils and rocks (U.S. Na- Whether thermal diffusion of electrolytes, the Soret
tional Committee for Rock Mechanics, 1987). effect, is important in soils has not been evaluated;
Gradient X
Chemical
Flow J Hydraulic Head Temperature Electrical Concentration
Fluid Hydraulic Thermoosmosis Electroosmosis Chemical
conduction osmosis
Darcy’s law
Heat Isothermal heat Thermal Peltier effect Dufour effect
transfer or conduction
l
thermal filtration Fourier’s law
ria
Current Streaming current Thermoelectricity Electric Diffusion and
Seebeck or conduction membrane
Thompson effect Ohm’s law potentials or
sedimentation
ate
current
Ion Streaming current Thermal diffusion Electrophoresis Diffusion Fick’s
ultrafiltration of electrolyte or law
(also known as Soret effect
hyperfiltration)
dM
however, since chemical activity is highly temperature
dependent, it may be a significant process in some
systems. Finally, electrophoresis, the movement of
qhe
qh
⫽ 冉冊
ke
E
kh
H
(
C ⫽ 0) (9.59a)
charged particles in an electrical field, has been used The ratio (kc /kh) in Fig. 9.20 indicates the hydraulic
for concentration of mine waste and high water content
hte
description of coupled flows is given later. A practical the layer. During consolidation, the hydraulic conduc-
form of Eq. (9.57) for fluid flow under combined hy- tivity decreases dramatically. However, the ratios kc /kh
draulic, chemical, and electrical gradients is and ke /kh increase significantly, indicating that the rel-
ative importance of osmotic and electroosmotic flows
H log(CB /CA)
E to the total flow increases. Although the data shown in
py
qh ⫽ ⫺kh A ⫹ kc A ⫺ ke A (9.58)
L L L Fig. 9.20 are shown as a function of the consolidation
pressure, the changes in the values of kc /kh and ke /kh
in which kh, kc, and ke are the hydraulic, osmotic, and are really a result of the decrease in void ratio that
electroosmotic conductivities,
H is the hydraulic head accompanies the increase in pressure, as may be seen
Co
冉冊
lonite-based bentonites. In systems containing confined
qhc k log(CB /CA) clay layers acted on by chemical and/or electrical gra-
⫽⫺ c (
E ⫽ 0) (9.59)
qh kh
H dients, Darcy’s law by itself may be an insufficient
basis for prediction of hydraulic flow rates, particularly
and, in the absence of a chemical gradient, the ratio of if the clay is highly plastic and at a very low void ratio.
electroosmotic flows to hydraulic flows is Such conditions can be found in deeply buried clay
l
ria
ate
dM
hte
rig
Figure 9.20 Hydraulic, osmotic, and electroosmotic conductivities of kaolinite (data from
Olsen 1969, 1972): (a) consolidation curve, (b) conductivity values, and (c) conductivities
as a function of void ratio.
py
and clay shale and in densely compacted clays. For Each approach has advantages and limitations. It is as-
more compressible clays, the ratios kc /kh and ke /kh may sumed in the following that the soil properties remain
be sufficiently high to be useful for consolidation by unchanged during the flow processes, an assumption
Co
electrical and chemical means, as discussed later in this that may not be justified in some cases. The effects of
chapter. flows of different types on the state and properties of
a soil are discussed later in this chapter. However,
when properties are known to vary in a predictable
9.10 QUANTIFICATION OF COUPLED FLOWS manner, their variations may be taken into account in
numerical analysis methods.
Quantification of coupled flow processes may be done
by direct, empirical determination of the relevant pa-
rameters for a particular case or by relationships de- Direct Observational Approach
rived from a theoretical thermodynamic analysis of the In the general case, there may be fluid, chemical, elec-
complete set of direct and coupled flow equations. trical, and heat flows. The chemical flows can be sub-
divided according to the particular chemical species ductivity coefficient kh is readily determined.7 The co-
present. Each flow type may have contributions caused efficient of electroosmotic hydraulic conductivity is
by gradients of another type, and their importance usually determined by measuring the hydraulic flow
depends on the values of Lij and Xj in Eq. (9.57). A rate developed in a known DC potential field under
complete and accurate description of all flows may be conditions of ih ⫽ 0. The electrical conductivity e is
a formidable task. obtained from the same experiment through measure-
However, in many cases, flows of only one or two ment of the electrical current.
types may be of interest, some of the gradients may The main advantage of this empirical, but direct, ap-
not exist, and/or some of the coupling coefficients may proach is simplicity. It is particularly useful when only
be either known or assumed to be unimportant. The a few of the possible couplings are likely to be im-
matrix of flows and forces then reduces significantly, portant and when some uncertainty in the measured
l
and the determination of coefficients is greatly simpli- coefficients is acceptable.
ria
fied. For example, if simple electroosmosis under iso-
thermal conditions is considered, then Eq. (9.57) yields General Theory for Coupled Flows
When several flows are of interest, each resulting from
qw ⫽ LHH(⫺H) ⫹ LHE(⫺E) ⫹ LHC(⫺C) (9.60) several gradients, a more formal methodology is nec-
ate
I ⫽ LEH(⫺H) ⫹ LEE(⫺E) ⫹ LEC(⫺C) (9.61) essary so that all relevant factors are accounted for
properly. If there are n different driving forces, then
JC ⫽ LCH(⫺H) ⫹ LCE(⫺E) ⫹ LCC(⫺C) (9.62) there will be n direct flow coefficients Lii and n(n ⫺
1) coupling coefficients Lij(i ⫽ j). The determination
of these coefficients is best done within a framework
where qw ⫽ water flow rate that provides a consistent and correct description of
I
Jc
⫽
⫽
dM
electrical current
chemical flow rate
each of the flows. Irreversible thermodynamics, also
termed nonequilibrium thermodynamics, offers a basis
H ⫽ hydraulic head for such a description. Furthermore, if the terms are
E ⫽ electrical potential properly formulated, then Onsager’s reciprocal rela-
C ⫽ chemical concentration tions apply, that is,
Lij ⫽ coupling coefficients; the first subscript
indicates the flow type and the second de-
hte
If there are no chemical concentration differences and the number of coefficients to be determined is sig-
across the system, then the last terms on the right-hand nificantly reduced. In addition, the derived forms for
side of Eqs. (9.60), (9.61), and (9.62) do not exist. In the coupling coefficients, when cast in terms of meas-
rig
this case, Eqs. (9.60) and (9.61) become, when written urable and understood properties, provide a basis for
in more familiar terms, rapid assessment of their importance.
The theory of irreversible thermodynamics as ap-
plied to transport processes in soils is only outlined
qw ⫽ khih ⫹ keie (9.63) here. More comprehensive treatments are given by
py
I ⫽ hih ⫹ eie (9.64) DeGroot and Mazur (1962), Fitts (1962), Katchalsky
and Curran (1967), Greenberg, et al. (1973), Yeung
and Mitchell (1992), and Malusis and Shackelford
where kh ⫽ hydraulic conductivity (2002a).
ke ⫽ electroosmotic hydraulic conductivity
Co
l
3. Validity of the Onsager reciprocal relations, a
ria
condition that is satisfied if the Ji and Xj are for- 1. Finding the dissipation function # for the flows
mulated properly (Onsager, 1931a, 1931b). Ex- 2. Defining the conjugated flows Ji and driving
perimental verification of the Onsager reciprocity forces Xi from Eq. (9.69)
for many systems and processes has been ob- 3. Formulating the phenomenological equations in
tained and is summarized by Miller (1960). the form of Eq. (9.66)
ate
4. Applying the Onsager reciprocal relations
Both the driving forces and flows vanish in systems
5. Relating the phenomenological coefficients to
that are in equilibrium, so the deviations of thermo-
measurable quantities
dynamic variables from their equilibrium values pro-
vide a suitable basis for their formulation. The When the Onsager reciprocity is used, the number
deviations of the state parameters Ai from equilibrium of independent coefficients Lij reduces from n2 to
are given by
dM [(n ⫹ 1)n]/2.
i ⫽ Ai ⫺ A 0i (9.67) Application
The quantitative analysis and prediction of flows
where A 0i is the value of the state parameter at equilib- through soils, for a given set of boundary conditions,
rium and Ai is its value in the disturbed state. depends on the values of the various phenomenological
hte
Criteria for deriving the forces and flows are then coefficients in the above flow equations. Unfortunately,
developed on the basis of the second law of thermo- these are not always known with certainty, and they
dynamics, which states that at equilibrium, the entropy may vary over wide ranges, even within an apparently
S is a maximum, and i ⫽ 0. The change in entropy homogeneous soil mass. The direct flow coefficients,
S that results from a change in state parameter gives that is, the hydraulic, electrical, and thermal conduc-
the tendency for a variable to change. Thus S/ i is tivities, and the diffusion coefficient, exhibit the
rig
a measure of the force causing i to change, and is greatest ranges of values. Thus, it is important to ex-
called Xi. amine these properties first before detailed analysis of
The flows Ji, termed fluxes in irreversible thermo- coupled flow contributions. For many problems, it may
dynamics, are given by i / t, the time derivative of be sufficient to consider only the direct flows, provided
py
i. On this basis, the resulting entropy production the factors influencing their values are fully appreci-
per unit time becomes ated.
冘JX
n
dS
⫽ ⫽ (9.68) 9.11 SIMULTANEOUS FLOWS OF WATER,
Co
i i
dt i⫽1
CURRENT, AND CHEMICALS
The entropy production can be related explicitly to Use of irreversible thermodynamics for the description
various irreversible processes in terms of proper forces of coupled flows as developed above is straightforward
and fluxes (Gray, 1966; Yeung and Mitchell, 1992). If in principle; however, it becomes progressively more
the choices satisfy Eq. (9.68), then the Onsager reci- difficult in application as the numbers of driving forces
procity relations apply. and different flow types increase. This is because of
It has been found more useful to use # ⫽ T, the (1) the need for proper specification of the different
dissipation function, in which T is temperature, than coupling coefficients and (2) the need for independent
l
1972). The theory for coupled salt and water flows was
ria
developed by Greenberg (1971) and applied to flows
in a groundwater basin (Greenberg et al., 1973) and to
chemicoosmotic consolidation of clay (Mitchell et al.,
1973).
ate
Equations for the simultaneous flows of water, elec- Figure 9.21 Schematic diagram of system for analysis of
tricity, cations, and anions under hydraulic, electrical, simultaneous flows of water, electricity, and ions through a
and chemical gradients were formulated by Yeung soil.
(1990) using the formalism of irreversible thermody-
namics as outlined previously. The detailed develop-
ment is given by Yeung and Mitchell (1993). The
dM Jv ⫽ L11(⫺P) ⫹ L12(⫺E) ⫹ L13(⫺cc)
results are given here. The chemical flow is separated
into its anionic and cationic components in order to ⫹ L14(⫺ca) (9.71)
permit determination of their separate movements as a
function of time. This separation may be important in I ⫽ L21(⫺P) ⫹ L22(⫺E) ⫹ L23(⫺cc)
some problems, such as chemical transport through the
⫹ L24(⫺ca) (9.72)
ground, where the fate of a particular ionic species, a
hte
dependent parts of the chemical potential gradients of pling coefficients reduces because
the cation (cc) and of the anion (ca). The fluxes
are the volume flow rate of the solution per unit area L12 ⫽ L21
Jv, the electric current I, and the diffusion flow rates
of the cation Jdc and the anion Jda per unit area relative L13 ⫽ L31
Co
l
conductivity, osmotic efficiency, and effective diffusion T ⫽ absolute temperature (K)
ria
coefficients done in the usual manner in geotechnical
and chemical laboratories, the detailed application of Subsequently, Manassero and Dominijanni (2003)
irreversible thermodynamic theory led Yeung (1990) pointed out that the practical equations for diffusion
and Yeung and Mitchell (1993) to the following defi- L33 and L44 do not take the osmotic efficiency " (Sec-
nitions for the Lij. It was assumed in the derivations
ate
tion 9.13) into account, so Eqs. (9.82) and (9.84) more
that the solution is dilute and there are no interactions properly should be
between cations and anions.8
k L L
L11 ⫽ h ⫹ 12 21
wn L22
(9.75) L33 ⫽ cc 冋(1 ⫺ ")D*
RT
c c k"2
⫹ c
wn 册 (9.85)
冋 册
L12 ⫽ L21 ⫽
dM ke
(9.76) L44 ⫽ ca
(1 ⫺ ")D*
a c k"2
⫹ a (9.86)
n RT wn
⫺"cckh L L
L13 ⫽ L31 ⫽ ⫹ 12 23 (9.77) This modification becomes important in clays wherein
wn L22
osmotic efficiency, that is, the ability of the clay to
⫺"cakh L L restrict the flow of ions, is high.
hte
L14 ⫽ L41 ⫽ ⫹ 12 24 (9.78) As the flows of ions relative to the soil are of more
wn L22
interest than relative to the water, Eq. (9.70) and Eqs.
e (9.73) and (9.74) can be combined to give
L22 ⫽ (9.79)
n
L23 ⫽ L32 ⫽ ccu* (9.80) Jc ⫽ (L31 ⫹ ccL11) w(⫺h) ⫹ (L32 ⫹ ccL12)(⫺E)
rig
ca
L34 ⫽ L43 ⫽ 0 (9.83)
D*
a ca Ja ⫽ (L41 ⫹ caL11) w(⫺h) ⫹ (L42 ⫹ caL12)(⫺E)
L44 ⫽ (9.84)
RT
Co
RT
⫹ (L43 ⫹ caL13) (⫺cc)
where kh ⫽ hydraulic conductivity as usually mea- cc
sured (no electrical short circuiting)
RT
ke ⫽ coefficient of electroosmotic hydraulic ⫹ (L44 ⫹ caL14) (⫺ca) (9.88)
conductivity ca
l
ents (Mitchell and Yeung, 1990). of actual motion of fluid particles from the overall or
ria
The analysis of multicomponent systems is more average movement described by Darcy’s law. More de-
complex. The use of averaged chemical properties and tails can be found in groundwater and contamination
the assumption of composite single species of anions textbooks such as Freeze and Cherry (1979) and Dom-
and cations may yield reasonable approximate solu- inico and Schwartz (1997).
Numerical models are available for groundwater
ate
tions in some cases. Malusis and Shackelford (2002a)
present a more general theory for coupled chemical flow and contaminant transport into which the above
and hydraulic flow, based on an extension of the Yeung flow equations can be introduced (e.g., Anderson and
and Mitchell (1993) formulation, which accounts for Woessner, 1992; Zheng and Bennett, 2002). The most
multicomponent pore fluids and ion exchange proc- widely used groundwater flow numerical code is
esses occurring during transport.9
dM MODFLOW developed by the United States Geolog-
The flow equations can be incorporated into numer- ical Survey (USGS); various updated versions are
ical models for the solution of transient flow problems. available (e.g., Harbaugh et al., 2000). To solve
Conservation of mass of species i requires that single-species contaminant transport problems in
groundwater, MT3DMS (Zheng and Wang, 1999) can
ci be used. The code utilizes the flow solutions from
⫽ ⫺Ji ⫺ Gi (9.90) MODFLOW. More complex multispecies reactions can
t
be simulated by RT3D (Clement, 1997). POLLUTE
hte
冋 册
the USGS, the U.S. Environmental Protection Agency
rig
Kd Kd ci (U.S. EPA), and the U.S. Salinity Laboratory.
Gi ⫽ 1 ⫹ ici ⫹ (9.90a)
n n t
l
ria
ate
dM
Figure 9.22 Electrokinetic phenomena: (a) electroosmosis, (b) streaming potential, (c) elec-
trophoresis, and (d) migration or sedimentation potential.
hte
carry their water of hydration and exert a viscous drag of the electrodes and repelled from the other. Nega-
on the water around them. Since there are more mobile tively charged clay particles move toward the anode as
cations than anions in a soil containing negatively shown in Fig. 9.22c. This is called electrophoresis.
charged clay particles, there is a net water flow toward Electrophoresis involves discrete particle transport
the cathode. This flow is termed electroosmosis, and through water; electroosmosis involves water transport
rig
its magnitude depends on ke, the coefficient of elec- through a continuous soil particle network.
troosmotic hydraulic conductivity and the voltage gra-
dient, as considered in more detail later. Migration or Sedimentation Potential
The movement of charged particles such as clay rela-
py
Streaming Potential
tive to a solution, as during gravitational settling, for
When water flows through a soil under a hydraulic example, generates a potential difference, as shown in
gradient (Fig. 9.22b), double-layer charges are dis- Fig. 9.22d. This is caused by the viscous drag of the
placed in the direction of flow. This generates an elec- water that retards the movement of the diffuse layer
Co
trical potential difference that is proportional to the cations relative to the particles.
hydraulic flow rate, called the streaming potential, be- Of the four electrokinetic phenomena, electroos-
tween the opposite ends of the soil mass. Streaming mosis has been given the most attention in geotechni-
potentials up to several tens of millivolts have been cal engineering because of its practical value for
measured in clays. transporting water in fine-grained soils. It has been
used for dewatering, soft ground consolidation, grout
Electrophoresis injection, and the containment and extraction of chem-
If a DC field is placed across a colloidal suspension, icals in the ground. These applications are considered
charged particles are attracted electrostatically to one in a later section.
l
In Table 9.5 the diffusion coefficients and ionic mo-
ria
bilities for cations and anions are considered together
since they lie within similar ranges for most species. Coupling Influences on Hydraulic Flow
Values of ionic mobility for specific ions in dilute so- In the absence of applied electrical and chemical gra-
lution are given in standard chemical references, for dients, flow under a hydraulic gradient is given by the
example, Dean (1973), and values of diffusion coeffi- first bracketed term on the right-hand side of Eq.
ate
cients are given in Tables 9.3 and 9.4. Ionic mobility (9.92). It contains the quantity k2e w /ne, which com-
is related to the diffusion coefficient according to pensates for the electroosmotic counterflow generated
by the streaming potential, which causes the measured
Di兩zi兩F value of kh to be slightly less than the true value of
ui ⫽ (9.91)
dM RT L11.
As it is not usual practice to short-circuit between
in which zi is the ionic valence and F is Faraday’s the ends of samples during hydraulic conductivity test-
constant. Similarly to the diffusion coefficients, the ing, the second bracketed term on the right-hand side
ionic mobilities are considerably less in a soil than in of Eq. (9.92) is not zero. This term represents an elec-
a free solution, especially in a fine-grained soil. troosmotic counterflow that results from the streaming
The importance of coupled flows to fluid, electrical potential and acts in the direction opposite to the hy-
current, and chemical transport through soil under dif- draulically driven flow. Analysis based on the values
hte
ferent conditions can be examined by study of the con- of properties in Table 9.5, as well as the results of
tributions of the different terms in Eqs. (9.71), (9.72), measurements, for example, Michaels and Lin (1954)
(9.87), and (9.88). For this purpose, the equations have and Olsen (1962) show that this counterflow is negli-
been rewritten in one-dimensional form and in terms gible in most cases, but it may become significant rel-
of the hydraulic, electrical, and chemical concentration ative to the true hydraulic conductivity for soils of very
gradients: ih ⫽ ⫺dh/dx, ie ⫽ ⫺dV/dx, and ic ⫽ ⫺dc/ low hydraulic conductivity, for example, kh ⬍ 1 ⫻
rig
dx, respectively. In addition, the chemical flows have 10⫺10 m/s. For example, for a value of ke of 5 ⫻ 10⫺9
been represented by a single equation. This assumes m2 /s-V, an electrical conductivity of 0.01 mho/m, and
that all dissolved species are moving together. Terms a porosity of 35 percent, the counterflow term is 0.7
involving the ionic mobility u do not exist in such a ⫻ 10⫺10 m/s.
py
formulation because the cations and anions move to- In the presence of an applied DC field the second
gether, with the effects of electrical fields assumed to bracketed term on the right-hand side of Eq. (9.92) can
accelerate the slower moving ions and to retard the be very large relative to hydraulic flow in soils finer
faster moving ions. Thus there is no net transfer of than silts, as ke, which typically ranges within only
electric charge due to ionic movement. The Lij coeffi- narrow limits, is large relative to kh; that is, kh is less
Co
cients have been replaced by the physical and chemical than 1 ⫻ 10⫺8 m/s in these soils. The relative effect-
quantities that determine them, as given by Eqs. (9.74) iveness of hydraulic and electrical driving forces for
through (9.85). The resulting equations are the follow- water movement can be assessed by comparing gra-
ing. For fluid flow: dients needed to give equal flow rates. They will be
equal if
Jv ⫽ 冋 kh
n
k2
册 冋册 k
⫹ e w ih ⫹ e ie ⫹ RT ⫺
en n
"kh
冋i
wn c 册 keie ⫽ khih (9.95)
(9.92)
The hydraulic gradient required to balance the elec-
For electrical current flow: troosmotic flow then becomes
l
fective in fine-grained soils, as discussed further in
ria
Section 9.15.
Chemically driven hydraulic flow is given by the last
Figure 9.23 Theoretical values of osmotic pressure as a
term on the right-hand side of Eq. (9.92). It depends
function of concentration difference across a clay layer for
primarily on the osmotic efficiency ". Osmotic effi- different values of osmotic efficiency coefficient, ". (T ⫽
ate
ciency has an important influence on the movement of 20C).
chemicals through a soil, the development of osmotic
pressure, and the effectiveness of clay barriers for
chemical waste containment.
Osmotic Efficiency The osmotic efficiency of clay, Values of osmotic efficiency coefficient, ", or
a slurry wall, a geosynthetic clay liner (GCL), or other
dM membrane efficiency (" expressed as a percentage),
seepage and containment barrier is a measure of the have been measured for clays and geosynthetic clay
material’s effectiveness in causing hydraulic flow un- liners; for example, Kemper and Rollins (1966), Letey
der an osmotic pressure gradient and of its ability to et al. (1969), Olsen (1969), Kemper and Quirk (1972),
act as a semipermeable membrane in preventing the Bresler (1973), Elrick et al. (1976), Barbour and Fred-
passage of ions, while allowing the passage of water. lund (1989), and Malusis and Shackelford (2002b,
The osmotic pressure concept can be better appreciated 2002c). Values of membrane efficiency from 0 to 100
by rewriting the last term in Eq. (9.92): percent have been determined, depending on the clay
hte
ference for different values of osmotic efficiency are surface areas corresponding to different clay types and
shown in Fig. 9.23. noting that volumetric water content equals surface
The van’t Hoff equation for osmotic pressure is area times layer thickness. The relationship between
specific surface area and liquid limit (LL) obtained by
冘 (n
Co
⫽ kT ⫺ niB) ⫽ RT(ciA ⫺ ciB) (9.98) Farrar and Coleman (1967) for 19 British clays
iA
where k is the Boltzmann constant (gas constant per LL ⫽ 19 ⫹ 0.56As (20%) (9.99)
molecule), R is the gas constant per molecule, T is the
absolute temperature, ni is concentration in particles in which the specific surface area As is in square meters
per unit volume, and ci is the molar concentration. The per gram, was then used to obtain the relationships
van’t Hoff equation applies for ideal and relatively di- shown in Fig. 9.25. The computed efficiencies shown
lute solution concentrations (Malusis and Shackelford, in Fig. 9.25 should be considered upper bounds be-
2002c). According to Fritz (1986) the error is low cause the assumption of uniform water distribution
(⬍5%) for 1⬊1 electrolytes (e.g., NaCl, KCl) and con- over the full surface area underestimates the effective
centrations 1.0 M. particle spacing in most cases. In most clays, espe-
l
2001).
ria
Coupling Influences on Electrical Flow
Substitution of values for the parameters in Eq. (9.93)
indicates, as would be expected, that electrical current
ate
flow is dominated completely by the electrical gradient
ie. In the presence of an applied voltage difference, the
other terms are of little importance, even if the move-
ments of anions and cations are considered separately
and the contributions due to ionic mobility are taken
dM into account. On the other hand, when a soil layer be-
haves as an open electrical circuit, small electrical po-
tentials, measured in millivolts, may exist if there are
hydraulic and/or chemical flows. This may be seen by
setting I ⫽ 0 in Eq. (9.93) and solving for ie, which
must have value if ih has value. These small potentials
Figure 9.24 Osmotic efficiency coefficient as a function of and flows are important in such processes as corrosion
b兹c where c is concentration of monovalent anion in nor- and electroosmotic counterflow.
hte
clay plates associate in clusters giving an effective spe- one for chemical transport under an electrical gradient,
cific surface that is less than that determined by most and one for transport of chemical under a chemical
methods of measurement. This means that the curves gradient. The first term in the first bracket of the right-
in Fig. 9.25 should in reality be displaced to the left. hand side of Eq. (9.94) describes advective transport.
High osmotic efficiencies are developed at low water As would be expected, the smaller the osmotic effi-
py
contents, that is, in very dense, low-porosity clays, and ciency, the more chemical flow through the soil is pos-
in dilute electrolyte systems. Malusis and Shackelford sible. The second term in the same bracket simply
(2002a, 2002b, 2002c) found that the osmotic effi- reflects the advective flow reduction that would result
ciency decreases with increasing solute concentration from electroosmotic counterflow caused by develop-
Co
and attribute this to compression of the diffuse double ment of a streaming potential. As noted earlier, this
layers adjacent to the clay particles. flow will be small, and its contribution to the total flow
Water flow by osmosis can be significant relative to will be small, except in clays of very low hydraulic
hydraulically driven water flow in heavily overconsol- and electrical conductivities. Advective transport is the
idated clay and clay shale, where the void ratio is low dominant means for chemical flow for soils having a
and the hydraulic conductivity is also very low. Such hydraulic conductivity greater than about 1 ⫻ 10⫺9
flow may be important in geological processes (Olsen m/s.
1969, 1972). Densely compacted clay barriers for The importance of an electrical driving force for
waste containment, usually composed of bentonite, chemical flow depends on the electrical potential gra-
possess osmotic membrane properties. As the chemical dient. For a unit gradient, that is, 1 V/m, chemical flow
l
ria
ate
dM
hte
quantities are comparable to those by advective flow der conditions where the value of " is low such that
under a unit hydraulic gradient in a clay having a hy- the second term in the bracket is negligible, chemical
draulic conductivity of about 1 ⫻ 10⫺9 m/s. Electri- transport by diffusion is significant relative to advec-
cally driven chemical flow is relatively less important tive chemical transport in soils with hydraulic conduc-
py
in higher permeability soils and more important in tivity values less than about 1 ⫻ 10⫺9 to 1 ⫻ 10⫺10
soils with lower kh. In cases where the electrically m/s for chemicals with diffusion coefficients in the
driven chemical transport is of interest, as in electro- range given by Table 9.7, that is, 2 ⫻ 10⫺10 to 2 ⫻
kinetic waste containment barrier applications, anion, 10⫺9 m2 /s.
Co
cation, and nonionic chemical flows must be consid- This is illustrated by Fig. 9.26 from Shackelford
ered separately using expanded relationships such as (1988), which shows the relative importance of advec-
given by Eqs. (9.87) and (9.88). tive and diffusive chemical flows on the transit time
The last bracketed quantity of Eq. (9.94) represents through a 0.91-m-thick compacted clay liner having a
diffusive flow under chemical gradients. The quantity porosity of 0.5 acted on by a hydraulic gradient of
D*ic gives the normal diffusive flow rate. The second 1.33. A diffusion coefficient of 6 ⫻ 10⫺10 m2 /s was
term represents a restriction on this flow that depends assumed. The transit time is defined as the time re-
on the clay’s osmotic efficiency, "; that is, if the clay quired for the solute concentration on the discharge
acts as an effective semipermeable membrane, diffu- side to reach 50 percent of that on the upstream side.
sive flow of chemicals is restricted. However, even un- For hydraulic conductivity values less than about 2 ⫻
l
lateral confining pressure prevents cracks from open-
ria
Figure 9.26 Transit times for chemical flow through a 0.91-
ing; thus there is risk of underestimating the hydraulic
m-thick compacted clay liner having a porosity of 50 percent conductivity of some soils. The consolidometer per-
and acted on by a hydraulic gradient of 1.33 (from Shack- meameter system allows for testing clays under a range
elford, 1988). of overburden stress states that are representative of
ate
those in the field and for quantitative assessment of the
effects of chemical interactions on volume stability and
hydraulic conductivity. More details of these perme-
10⫺9 m/s the transit time in the absence of diffusion ameters are given by Daniel (1994).
would be very long. For diffusion alone the transit time
The effects of chemicals on the hydraulic conduc-
would be about 47 years. dM tivity of high water content clays such as used in slurry
Most compacted clay barriers and geosynthetic clay
walls are likely to be much greater than on lower water
liners are likely to have hydraulic conductivity values
content, high-density clays as used in compacted clay
in the range of 1 ⫻ 10⫺11 to 1 ⫻ 10⫺9 m/s, with the
liners. This is because of the greater particle mobility
latter value being the upper limit allowed by the U.S.
and easier opportunity for fabric changes in a higher
EPA for most waste containment applications. In this
water content system. A high compactive effort or an
range, diffusion reduces the transit time significantly
in comparison to what it would be due to advection effective confining stress greater than about 70 kPa can
make properly compacted clay invulnerable to attack
hte
wall barriers constructed to contain them must be con- alumina octahedral layers of clay minerals. Bases can
sidered in the design of waste containment barriers. dissolve silica tetrahedral layers, and to a lesser extent,
Numerous studies have been done to evaluate chemical alumina octahedral layers of clay minerals. Removal
effects on clay hydraulic conductivity because of fears of dissolved material can cause increases in hydraulic
that prolonged exposure may compromise the integrity conductivity, whereas precipitation can clog pores and
of the liners and barriers and because tests have shown reduce hydraulic conductivity.
that under some conditions clay can shrink and crack The most important factors controlling the effects of
when permeated by certain classes of chemicals. Sum- organic chemicals on hydraulic conductivity are (1)
maries of the results of chemical compatibility studies water solubility, (2) dielectric constant, (3) polarity,
are given by Mitchell and Madsen (1987) and Quigley and (4) whether or not the soil is exposed to the pure
and Fernandez (1989), and factors controlling the long- organic or a dilute solution. Exposure of clay barriers
l
ria
ate
dM
hte
Figure 9.27 Three types of permeameter for compatibility testing: (a) rigid wall, (b) flexible
wall, and (c) consolidometer permeameter (from Day, 1984).
rig
to water-insoluble pure or concentrated organics is 3. Many water-insoluble organic liquids (i.e., non-
likely only in the case of spills, leaking tanks, and with aquoues-phase liquids, NAPLs) can cause shrink-
py
dense non-aqueous-phase liquids (DNAPLs) or ‘‘sink- age and cracking of clays, with concurrent
ers’’ that accumulate above low spots in liners. Some increases in hydraulic conductivity.
general conclusions about the influences of organics 4. Hydraulic conductivity increases caused by per-
on the hydraulic conductivity are: meation by organics are partly reversible when
water is reintroduced as the permeant.
Co
itation and pore clogging downstream. However, ative to that measured with water. The decreases in
after long time periods these precipitates may be hydraulic conductivity for dioxane concentrations up
redissolved and removed, thus leading to an in- to 70 percent can be accounted for in terms of fluid
crease in hydraulic conductivity. density and viscosity, as may be seen in Fig. 9.28b
8. Pure bases can cause a large increase in the hy- where the intrinsic values of permeability are shown.
draulic conductivity, whereas concentrations at or As noted earlier in this chapter, the intrinsic permea-
below the solubility limit in water have no effect. bility is defined by K ⫽ k / .
9. Organic acids do not cause large-scale dissolu- Although many chemicals do not have significant
tion of clay particles. effect on the hydraulic conductivity of clay barriers,
this does not mean that they will not be transported
The combined effects of confining pressure and con- through clay. Unless adsorbed by the clay or by or-
l
centration, as well as permeant density and viscosity, ganic matter, the chemicals will be transported by ad-
ria
are illustrated by Fig. 9.28 (Fernandez and Quigley, vection and diffusion. Furthermore, the actual transit
1988). The data are for water-compacted, brown Sarnia time through a barrier by advection, that is, the time
clay permeated by solutions of dioxane in domestic for chemicals moving with the seepage water, may be
landfill leachate. Increased hydrocarbon concentration far less than estimated using the conventional seepage
caused a decrease in hydraulic conductivity up to con- velocity. The seepage velocity is usually defined as the
ate
centrations of about 70 percent, after which the hy- Darcy velocity khih, divided by the total porosity n. In
draulic conductivity increased by about three orders of systems with unequal pore sizes the flow is almost to-
magnitude for pure dioxane (Fig. 9.28a), for samples tally through mini- and macropores, which comprise
that were unconfined by a vertical stress (v ⫽ 0). On the effective porosity ne, which may be much less than
the other hand, the data points for samples maintained
dM the total porosity. Thus effective compaction of clay
under a vertical confining stress of 160 kPa indicated barriers must break down clods and aggregates to de-
no effect of the dioxane on hydraulic conductivity rel- crease the effective pore size and increase the propor-
hte
rig
py
Co
Figure 9.28 (a) Hydraulic conductivity and (b) intrinsic permeability of compacted Sarnia
clay permeated with leachate–dioxane mixtures. Initial tests run using water (●) followed by
leachate–chemical solution (䉱). (from Fernandez and Quigley, 1988). Reproduced with per-
mission from the National Research Council of Canada.
l
known electrical gradient. Alternatively, a null indicat- and the center of the plane of mobile charge, then the
ria
ing system may be used or it may be deduced from a velocity gradient between the wall and the center of
streaming potential measurement. From experience it positive charge is v / ; thus, the drag force per unit
is known that ke is generally in the range of 1 ⫻ 10⫺9 area is dv /dx ⫽ v / , where is the viscosity. The
to 1 ⫻ 10⫺8 m2 /s V (m/s per V/m) and that it is of force per unit area from the electrical field is
E/
the same order of magnitude for most soil types, as
L, where is the surface charge density and
E/
L
ate
may be seen by the values for different soils and a is the electrical potential gradient. At equilibrium
freshwater permeant given in Table 9.7.
Several theories have been proposed to explain elec-
troosmosis and to provide a basis for quantitative pre- v
E
⫽ (9.100)
diction of flow rates.
L
dM
Helmholtz and Smoluchowski Theory or
This theory, based on a model introduced by Helm-
holtz (1879) and refined by Smoluchowski (1914), is
one of the earliest and most widely used. A liquid- 10
A derivation using a Poisson–Boltzmann distribution of counter-
filled capillary is treated as an electrical condenser with ions adjacent to the wall gives the same result.
hte
l
ria
Figure 9.29 Helmholtz–Smoluchowski model for electrokinetic phenomena.
L D E
ate
⫽ v (9.101) qA ⫽ Nqa ⫽ Na (9.104b)
E
L
From electrostatics, the potential across a condenser If the porosity is n, then the cross-sectional area of
is given by voids is nA, which must equal Na. Thus,
dM
⫽
(9.102) qA ⫽
D
E
n
L
A (9.105)
D
where D is the relative permittivity, or dielectric con- By analogy with Darcy’s law we can write Eq.
stant of the pore fluid. Substitution for in Eq. (9.105) as
(9.102) gives
hte
qA ⫽ keie A (9.106)
v⫽ 冉 冊
D
E
L
(9.103) in which ie is the electrical potential gradient
E/
L
and ke the coefficient of electroosmotic hydraulic con-
ductivity is
The potential is termed the zeta potential. It is not
rig
If an electrical potential gradient of 20 V/m is used, where A0 is the concentration of wall charges in ionic
substitution in Eq. (9.108) shows that ih is 0.01 for the equivalents per unit volume of pore fluid, and F0 is the
fine sand and 1000 for the clay. This means that a Faraday constant. Replacement of FH by FE in Eq.
hydraulic gradient of only 0.01 can move water as ef- (9.111) gives
fectively as an electrical gradient of 20 V/m in fine
sand. However, for the clay, a hydraulic gradient of r 4
E F A
qa ⫽ A F ⫽ 0 0 r 2iea (9.113)
1000 would be needed to offset the electroosmotic 8 0 0
L 8
flow.
l
so for a total cross section of N capillaries and area A
ria
However, it does not follow that electroosmosis will
always be an efficient means to move water in clays
because the above analysis does not take into account A0 F0r 2
qA ⫽ nie A (9.114)
the power requirement to develop the potential gradient 8
of 20 V/m or energy losses in the system. These points
ate
are considered further later. This equation shows that ke should vary as r 2, whereas
the Helmholtz–Smoluchowski theory leads to ke in-
Schmid Theory dependent of pore size, as previously noted. Of the two
theories, the Helmholtz gives the better results for
The Helmholtz–Smoluchowski theory is essentially a
soils, perhaps because most clays have a cluster or ag-
large-pore theory because it assumes a negligible ex-
gregate structure with electroosmotic flow controlled
dM
tension of the counterion layer into the pore. Also, it
does not account for an excess of ions over those
more by the larger pores than by the intracluster pores.
needed to balance the surface charge. A model that
Spiegler Friction Model
overcomes the first of these problems was proposed by
Schmid (1950, 1951). It can be considered a small- A completely different concept for electrokinetic proc-
pore theory. esses takes into account the interactions of the mobile
The counterions are assumed to be distributed uni- components (water and ions) on each other and of the
hte
formly throughout the fluid phase in the soil. The elec- frictional interactions of these components with pore
trical force acts uniformly over the entire pore cross walls (Spiegler, 1958). This theory provides insight
section and gives the same velocity profile as shown into conditions leading to high electroosmotic effi-
by Fig. 9.29. The hydraulic flow rate through a single ciency.
capillary of radius r is given by Poiseuille’s law: The assumptions include:
rig
The hydraulic seepage force per unit length causing The following equation for electroosmotic transport
flow is of water across a fine-grained porous material contain-
ing adsorbed and free ions can be derived:
FH ⫽ r 2 wih (9.110)
Co
C3
⫽ (W ⫺ H) ⫽ (9.115)
so C1 ⫹ C3(X34 /X13)
(moles/faraday), H is the water transport by ion hy- opposite sign. The greater the difference between the
dration (moles/faraday), C3 is the concentration of free concentrations of cations and anions, the greater the
water in the material (mol/m3), C1 is the concentration net drag on the water in the direction toward the cath-
of mobile counterions m2, X34 is the friction coefficient ode. The efficiency and economics of the process de-
between water and the solid wall, and X13 is the friction pend on the volume of water transported per unit
coefficient between cation and water. electrical charge passed. If the volume is high, then
Concentrations C1 and C3 are hypothetical and prob- more water is transported for a given expenditure of
ably less than values measured by chemical analysis electrical energy than if it is low. This volume may
because some ions may be immobile. Evaluation of X13 vary over several orders of magnitude depending on
and X34 requires independent measurements of diffu- such factors as soil type, water content, and electrolyte
sion coefficients, conductance, transference numbers, concentration.
l
and water transport. Thus Eq. (9.115) is limited as a In a low exchange capacity soil at high water content
ria
predictive equation. Its real value is in providing a rel- in a low electrolyte concentration solution, there is
atively simple physical representation of a complex much more water per cation than in a high exchange
process. capacity, low water content soil having the same pore
From Eq. (9.115), water electrolyte concentration. This, combined
with cation-to-anion ratio considerations, leads to
ate
1 the predicted water transport–water content–soil
⫽ (W ⫺ H) ⫽ (9.116) type–electrolyte concentration relationships shown
(C1 /C3 ⫹ X34 /X13)
schematically in Fig. 9.30, where increasing electrolyte
concentration in the pore water results in a much
At high water contents and for large pores, X34 /X13 →
0 because X34 becomes negligible. Then
dM
X34→0 ⫽ C3 /C1 (9.117)
Ion Hydration
Water of hydration is carried along with ions in a direct
current electric field. The ion hydration transport H is
given by
py
H ⫽ t⫹N⫹ ⫺ t
N
(9.118)
where t⫹ and t
are the transport numbers, that is, num-
Co
l
rium the potentials of the internal and external A0 ⫽ (9.121)
ria
solutions are equal and that electroneutrality is re- w
quired in both phases. It may be shown (Gray, 1966;
Gray and Mitchell, 1967) that the ratio R of cations to where w is the density of water and w is the water
anions in the internal phase for the case of a symmet- content. The higher R, the greater is the electroosmotic
rical electrolyte (z⫹ ⫽ z⫺) is given by
ate
water transport, all other things equal.
From Eqs. (9.119) to (9.121) it may be deduced that
C⫹ 1 ⫹ (1 ⫹ y2)1 / 2 exclusion of anions is favored by a high exchange
R⫽ ⫽ (9.119)
C⫺ ⫺1 ⫹ (1 ⫹ y2)1 / 2 capacity (active clay), a low water content, and low
salinity in the external solution. However, the concen-
where dM tration of anions in the double layer builds up more
hte
rig
py
Co
Figure 9.31 Electroosmotic water transport versus concentration of external electrolyte so-
lution for homoionic kaolinite and illite at various water content (from Gray, 1966).
l
shown are for open systems, that is, solution was ad-
ria
mitted at the anode at the same time it was extracted so
from the cathode. Electrochemical effects (Section
9.18) and water content changes were minimized in
qh LHE
these tests. Thus, the values can be interpreted as upper ⫽ (9.127)
bounds on the flow rates to be expected in practice. I LEE
ate
Values of water content, electrolyte concentration in
the pore water, and type of clay are required for elec- By Onsager’s reciprocity theorem LEH ⫽ LHE so
troosmosis efficiency estimation. Water content is read-
ily measured, the electrolyte concentration is easily
determined using a conductivity cell, and the clay type
dM
can be determined from plasticity and grain size in-
冉冊qh
I
P⫽0
⫽⫺ 冉 冊
E
P I⫽0
(9.128)
l
ke directly proportional to the voltage gradient.
ria
ki ⫽ (9.133)
5. All the applied voltage is effective in moving wa-
ter.13
As ke varies within relatively narrow limits, Eq. (9.133) 6. The electrical field is constant with time.
shows that the electroosmotic efficiency, measured by 7. The coupling of hydraulic and electrical flows
ate
ki, is a sensitive function of the electrical conductivity can be formulated by Eqs. (9.63) and (9.64).
of the soil. For soils 11, 13, and 14 in Table 9.7, is 8. There are no electrochemical reactions.
in the range of 0.02 to 0.03 S. For soil 12, in which
electroosmosis was not effective, is 0.25 S. In es- Governing Equations
sence, a high value of electrical conductivity means For one-dimensional flow between plate electrodes
that the current required to develop the voltage is too (Fig. 9.33a), Eq. (9.63) becomes
dM
high for economical movement of water. In addition,
if high current is used, the generation of gas, heat, and k u V
electrochemical effects become excessive. qh ⫽ ⫺ h ⫺ ke (9.134)
w x x
for the flow rate per unit area. For radial flow for the
9.17 CONSOLIDATION BY ELECTROOSMOSIS conditions shown in Fig. 9.33b and a layer of unit
hte
thickness
If, in a compressible soil, electroosmosis draws water
to a cathode where it is drained away and no water is
k u V
allowed to enter at the anode, then consolidation of the qh ⫽ ⫺ h 2r ⫺ ke 2r (9.135)
soil between the electrodes occurs in an amount equal w r r
to the volume of water removed. Water movement
rig
away from the anode causes consolidation in the vi- Introduction of Eq. (9.134) in place of Darcy’s law
cinity of the anode. The effective stress must increase in the derivation of the diffusion equation governing
concurrently. Because the total stress in the vicinity of consolidation in one dimension leads to
the anode remains essentially unchanged, the pore wa-
ter pressure must decrease. Water drains at the cathode kh 2u 2V u
py
⫹ k ⫽ mv (9.136)
where there is no consolidation. Therefore, the total, w x2 e
x2
t
effective, and pore water pressures at the cathode re-
main unchanged. As a result, hydraulic gradient de- and
velops that tends to cause water flow from cathode to
Co
or
k
du ⫽ ⫺ e w dV (9.140)
kh
l
k
u ⫽ ⫺ e w V ⫹ C (9.141)
ria
kh
ate
k
u ⫽ ⫺ e w V (9.142)
kh
gives
increase that develops depends on ke /kh. As ke only
冉 冊
varies within narrow limits for different soils, the total
2u ke 2V 1 u k V 1 u consolidation that can be achieved depends largely on
⫹ ⫹ ⫹ e w ⫽
r 2
kh w
r 2
r r kh r cv t kh. Thus, the potential for consolidation by electroos-
py
Amount of Consolidation de de
av ⫽ ⫺ ⫽ (9.143)
When the hydraulic gradient that develops in response d du
to the differing amounts of consolidation between the
anode and cathode generates a counterflow (kh / w)/ or
(u/ x) that exactly balances the electroosmotic flow
ke(V/ x) in the opposite direction, consolidation is de ⫽ av du ⫽ ⫺av d (9.144)
complete. As there then is no flow, qh in Eqs. (9.14)
and (9.135) is zero. Thus Eq. (9.134) is in which d is the increase in effective stress.
l
ria
ate
dM
hte
rig
Figure 9.34 Assumed variation of voltage with distance during electroosmosis: (a) one-
dimensional flow and (b) radial flow.
py
Thus, the more compressible the soil, the greater pendent on the voltage at the point. Accordingly, prop-
will be the amount of consolidation for a given stress erties at the end of treatment vary along a line between
increase, just as in the case of consolidation under ap- the anode and cathode, as shown, for example, by the
Co
plied loads. It follows, also, that electroosmosis will be posttreatment variations in shear strength and water
of little value in an overconsolidated clay unless the content shown in Fig. 9.36. Values of these properties
effective stress increases are large enough to bring the before treatment are also shown for comparison. More
material back into the virgin compression range. uniform property distributions between electrodes can
The consolidation loading of any small element of be obtained if the polarity of electrodes is reversed
the soil is isotropic, as it is done by increasing the after partial completion of consolidation (Wan and
effective stress through reduction in the pore water Mitchell, 1976).
pressure. The entire soil mass being treated is not con- The results shown in Fig. 9.36 were obtained at a
solidated isotropically or uniformly, however, because site in Norway where electroosmosis was used for the
the amount of consolidation varies with position, de- consolidation of quick clay (Bjerrum et al., 1967). The
l
ria
ate
dM
Figure 9.35 Consolidation by electroosmosis and by direct loading, one-dimensional case:
(a) electroosmosis and (b) direct loading.
hte
variations in strength and water content after treatment voltage, and TV is the time factor, defined in terms of
are consistent with the patterns to be expected based the distance between electrodes L and real time t as
on the predicted variation of pore pressure decrease
and vertical strain stress increase with voltage and po- cvt
sition shown in Fig. 9.35. TV ⫽ (9.146)
L2
rig
ke 2k V
冋冉 冊 册
u⫽ V(x) ⫹ e w 2 m
kh
冘
⬁
kh w 4 (⫺1)n 1
2
冋冉 冊 册
exp ⫺ n ⫹
1 2 2
2
TV (9.145) Solutions for Eqs. (9.145) and (9.148) are shown in
Figs. 9.37 and 9.38. They are applied in the same way
as the theoretical solution for classical consolidation
where V(x) is the voltage at x, Vm is the maximum theory.
l
ria
ate
dM
hte
rig
py
Co
Figure 9.36 Effect of electroosmosis treatment on properties of quick clay at Ås, Norway
(from Bjerrum et al., 1967): (a) Undrained shear strength, (b) remolded shear strength, (c)
water content, and (d) Atterberg limits.
l
ria
Figure 9.39 Average degree of consolidation as a function
ate
of dimensionless time for radial consolidation by electroos-
mosis (from Esrig, 1968). Reprinted with permission of
ASCE.
Figure 9.37 Dimensionless pore pressure as a function of
dimensionless time and distance for one-dimensional consol-
idation by electroosmosis.
dM to pile driving, pile pulling, reduction of negative skin
friction, and recovery of buried objects. Special solu-
tions for in situ determination of soil consolidation
properties by electroosmosis measurements have also
been developed (Banerjee and Mitchell, 1980).
One of the most important points to be noted from
these solutions is that the rate of consolidation depends
completely on the coefficient of consolidation, which
hte
sionless time for one-dimensional consolidation by electroos- high enough to enable consolidation in a reasonable
mosis. time. The soil types that best satisfy these conditions
are silts, clayey silts, and silty clays. Most successful
field applications of electroosmosis for consolidation
have been in these types of materials. As noted earlier,
Co
A numerical solution to Eq. (9.138) gives the results the electrical conductivity of the soil is also important;
shown in Fig. 9.39 (Esrig, 1968, 1971). For the case if it is too high, as in the case of high-salinity pore
of two pipe electrodes, a more realistic field condition water, adverse electrochemical effects and unfavorable
than the radial geometry of Fig. 9.33b, Fig. 9.39 cannot economics may preclude use of electroosmosis for
be expected to apply exactly. Along a straight line be- consolidation.
tween two pipe electrodes, however, the flow pattern
is approximately the same as for the radial case for a
considerable distance from each electrode. 9.18 ELECTROCHEMICAL EFFECTS
A solution for the rate of pore pressure buildup at
the cathode for the case of no drainage (closed cath- The measured strength increases in the quick clay at
ode) is shown in Fig. 9.40. This condition is relevant Ås, Norway (Fig. 9.36), were some 80 percent greater
l
ria
ate
Figure 9.40 Dimensionless pore pressure at the face of a cylindrical electrode as a function
of dimensionless time for the case of a closed cathode (a swelling condition) (from Esrig
and Henkel, 1968).
dM
than can be accounted for solely by reduction in water Chlorine may also form in a saline environment. Some
content. Also, the liquid and plastic limits were of the exchangeable cations on the clay may be re-
changed as a result of treatment. Consolidation alone placed by H⫹. Because hydrogen clays are generally
should have no effect on the Atterberg limits because unstable, and high acidity and oxidation cause rapid
changes in mineralogy, particle characteristics, and/or deterioration of the anodes, the clay will soon alter to
pore solution characteristics are needed to do this. the aluminum or iron form depending on the anode
hte
In addition to movement of water when a DC volt- material. As a result, the soil is usually strengthened
age field is applied between metal electrodes inserted in the vicinity of the anode. If gas generation at the
into a wet soil, the following effects may develop: ion anode causes cavitation and heat causes desiccation,
diffusion, ion exchange, development of osmotic and cracking may occur. This will limit the negative pore
pH gradients, desiccation by heat generation at the pressure that can develop to a value less than 1 atm,
electrodes, mineral decomposition, precipitation of and also the electrical resistance will increase, leading
rig
pected. Some of them, such as electrochemical hard- Cations in solution are drawn to the cathode where
ening of the soil that results in permanent changes in they combine with (OH)⫺ that is left behind to form
plasticity and strength, may be beneficial; others, such
hydroxides. The pH may rise to values as high as 12
as heating and gas generation, may impair the effi-
at the cathode. Some alumina and silica may go into
Co
l
technical problems as well.
ria
The magnitude of self-potential between different
soil layers depends on the contents of oxidizing and
9.19 ELECTROKINETIC REMEDIATION reducing substances in the layers (F. Hilbert, in Veder,
1981). These potentials can cause a natural electroos-
The transport of dissolved and suspended constituents mosis in which water flows in the direction from the
ate
into and out of the ground by electroosmosis and elec- higher to the lower potential, that is, toward the cath-
trophoresis, as well as electrochemical, reactions have ode. The process is shown schematically in Fig. 9.41.
become of increasing interest because of their potential An oxidizing soil layer is positive relative to a reducing
applications in waste containment and removal of con- layer, thus inducing an electroosmotic water flow to-
taminants from fine-grained soils. The electrolysis re- ward the interface. If water accumulates at the inter-
face, there can be swelling and loss of strength, leading
dM
actions at the electrodes described in the preceding
section, wherein acid is produced at the anode and base ultimately to formation of a slip surface.
at the cathode, are of particular relevance. After a few
days of treatment the pH in the vicinity of the anode Generation of Self-Potentials in Soil Layers
may drop to less than 2, and that at the cathode in-
crease to more than 10 (Acar and Alshewabkeh, 1993). Soils in an oxidizing environment are usually yellow
Toxic heavy metals are preferentially adsorbed by or tan to reddish brown and are characterized by oxides
and hydrates of trivalent iron and a low pH relative to
hte
⫽ 0.771 ⫹
RT
F 冉 冊
c3⫹
ln Fe
c2⫹
Fe
(9.151)
then ke /hh ⫽ 50 m/V. If the self-potential difference is
50 mV, then from Eq. (9.142) a pore pressure value of
l
gives the electrical potential as a function of the in situ be reduced.
ria
pH, that is,
Landslide Stabilization Using Short-Circuit
Conductors
⫽ 0.186 ⫺ 0.059 pH (9.152)
If slope instability is caused by a slip surface between
ate
Reasonable agreement has been obtained between reducing and oxidizing soil layers, then a simple means
measured and calculated values of for different soil for stabilization can be used (Veder, 1981). Short-
layers. The end result is that potential differences of circuiting conductors, such as steel rods, are driven
up to 50 mV or so are developed between different into the soil so that they extend across the slip surface
layers. Potentials measured in a trench excavated in a and about 1 to 2 m into the soil below. The mechanism
dM
slide zone are shown in Fig. 9.42. that is then established is shown in Fig. 9.43.
Electric current generated by reduction reactions in
the oxidizing soil layer and oxidizing reactions in the
Excess Pore Pressure Generation by Self-Potentials reducing layer flows through the conductors. Because
The pore pressure that may develop at an interface be- of the presence of oxidizing agents such as ferric iron,
tween two different soil layers is given by Eq. (9.142) oxygen, and manganese compounds, in the upper ox-
in which V is the difference in self-potentials between idizing layer that take up electrons, electrons pass from
the layers. For a given value of V, the magnitude of the metal conductor to the soil. That is, the introduc-
hte
pore pressure depends directly on ke /kh. For example, tion of electrons initiates reducing reactions. In the re-
if ke ⫽ 5 ⫻ 10⫺9 m2 /s V and kh ⫽ 1 ⫻ 10⫺10 m/s, ducing layer, on the other hand, there is already a
rig
py
Co
Figure 9.42 Electrical potentials measured in a trench cut into a slide (from Veder, 1981).
Reprinted with permission of Springer-Verlag.
l
positive relative to the cold side. Thermoosmotic pres-
ria
sures of only a few tenths of a centimeter water head
per degree Celsius were measured in saturated soil. Net
flows in different directions have been measured in dif-
ferent investigations, evidently because of different
temperature dependencies of chemical activity coeffi-
ate
cients. These small thermoelectric and thermoosmotic
effects in saturated soils may be of little practical sig-
Figure 9.43 Mechanism for slide stabilization using short- nificance in geotechnical problems.
circuiting conductors (adapted from Veder, 1981).
On the other hand, thermally driven moisture flows
dM in partly saturated soils can be large, and that these
flows can be very important in subgrade stability,
surplus of electrons. If these pass into the conductor, swelling soils, and heat transfer and storage problems
then the environment becomes favorable for oxidation of various types. Theoretical representations of mois-
reactions. Thus, positive charges are generated in the ture flow through partly saturated soils based solely on
reducing soil layer as the conductor carries electrons the application of irreversible thermodynamics, such as
away. The oxidizing soil layer then takes up these elec- developed by Taylor and Cary (1964), have not been
trons. completely successful. They underestimate the flows
hte
Completion of the electrical circuit requires current substantially, perhaps because of the inability to ade-
flow through the soil pore water in the manner shown quately represent all the processes and interactions.
in Fig. 9.43, where adsorbed cations, shown as Na⫹, A widely used theory for coupled heat and moisture
plus the associated water, flow away from the soil layer flow through soils was developed by Philip and De
interface. This electroosmotic transport of water re- Vries (1957). It accounts for both liquid- and vapor-
duces the water content in the slip zone. Thus, short- phase flows. Vapor-phase flow depends on the thermal
rig
circuit conductors have three main effects (Veder, and isothermal vapor diffusivities and is driven by tem-
1981): perature and moisture content gradients. The liquid-
phase flow depends on the thermal and isothermal
1. Natural electroosmosis is prevented because the liquid diffusivities and is driven by the temperature
py
short-circuiting conductors eliminate the poten- gradient, the moisture content gradient, and gravity.
tial difference between the two soil layers. The two governing equations are:
2. Electrochemical reactions produce electroos-
motic flow in the opposite direction, thus helping For vapor-phase flow:
to drain the shear zone.
Co
l
冉冊 冉 冊
ria
D0 d 0
DTV ⫽ v[a ⫹ ƒ(a) ]h (9.155) The hydraulic conductivity and suction relationships
w dT
are hysteretic; that is, they depend on whether the soil
The isothermal vapor diffusivity is given by is wetting or drying. Examples of the variations of the
different properties needed for the analysis are shown
冉 冊 冉 冊冉 冊
ate
D0 hg d in Fig. 9.44 as a function of degree of saturation and
DV ⫽ va 0 (9.156) volumetric water content. The data are for a crushed
w RT d
limestone that is used for a trench backfill around bur-
where D0 ⫽ molecular diffusivity of water vapor in air ied electrical transmission cables. This material is used
(L2 /T) because of its low thermal resistivity, which makes it
v ⫽ mass flow factor ⫽ P/(P ⫺ p) suitable for effective dissipation of heat from the bur-
dM
P ⫽ total gas pressure in pore space ied cable, provided the saturation does not fall below
about 40 percent.
p ⫽ partial pressure of water vapor in pore
space The vapor flow is made up of a flow away from the
⫽ tortuosity factor high-temperature side that is driven by a vapor density
a ⫽ volumetric air content (L3 /L3) gradient and a return flow caused by variation in the
h ⫽ relative humidity of air in pores pore vapor humidity as reflected by variations in soil
suction. At moderate soil suction values, for example,
hte
⫽ suction head of water in the soil (negative portant at very high suction levels. The liquid flow con-
head) (L) sists of a capillarity-driven flow toward the heat source
ƒ(a) ⫽ a/ak for 0 ⬍ a ⬍ ak and an outward liquid flow due to variations in water
⫽ 1 for a ak surface tension with temperature. McMillan’s analysis
ak ⫽ a at which liquid conductivity is lost or showed that for both sand and clay the isothermal liq-
py
at which the hydraulic conductivity falls uid diffusivity term was 4 to 5 orders of magnitude
below some arbitrary fraction of the sat- greater than the thermal liquid diffusivity term. Thus
urated value capillarity-driven flow predominates for any significant
gradient in the volumetric moisture content. The very
Co
The thermal liquid diffusivity is given by small thermal liquid diffusivity is consistent with the
冉 冊冉 冊
observations noted earlier for saturated soils in which
d measured water flows under thermal gradients are
DTL ⫽ k (9.157)
dT small.
The total water flow q in an unsaturated soil under
The isothermal diffusivity is given by the action of a temperature gradient and its resulting
冉冊
water content gradient equals the sum of the vapor-
d phase and liquid-phase movements. Thus, from Eqs.
DL ⫽ k (9.158)
d (9.153) to (9.158),
l
ria
ate
dM
hte
Figure 9.44 Examples of properties used for analysis of thermally driven moisture flow in
a partially saturated, compacted, crushed limestone: (a) particle size distribution, (b) suction
head as a function of volumetric water content, (c) hydraulic conductivity as a function of
degree of saturation and volumetric water content, (d) isothermal liquid diffusivity as a
function of degree of saturation and volumetric water content, (e) isothermal vapor diffusivity
as a function of degree of saturation and volumetric water content, and (f) Thermal water
rig
diffusivity as a function of degree of saturation and volumetric water content. Thermal re-
sistivity as a function of water content for this soil is shown in Fig. 9.14.
l
ria
ate
dM
Figure 9.44 (Continued )
hte
where kt ⫽ thermal conductivity Table 9.2. Three topics are considered in this section:
C ⫽ volumetric heat capacity (1) the depth of frost penetration, which illustrates the
application of transient heat flow analysis, (2) frost ac-
The ratio of thermal conductivity to the volumetric tion in soils, a phenomenon of great practical impor-
rig
heat capacity is the thermal diffusivity A. tance that can be understood through consideration of
Both transient and steady-state temperature distri- interactions of the physical and physicochemical prop-
butions computed using the Philip and De Vries theory erties of the soil, and (3) some effects of freezing on
incorporated into numerical models have agreed well the behavior and properties of the soil after thawing.
with measured values in a number of cases. The actual
py
methods, the finite element method, and the integrated Accurate estimation of the depth of ground freezing
finite difference method. Cameron (1986) reformulated during the winter, the depth of thawing in permafrost
the equations in terms of suction head rather than areas during the summer, and the refrigeration and time
moisture content and incorporated them into the finite requirements for artificial ground freezing for tempo-
element model of Walker et al. (1981) for solution of rary ground stabilization are all problems involving
two-dimensional problems. transient heat flow analysis. They differ from the con-
duction analyses in the preceding sections in that the
9.22 GROUND FREEZING phase change of water to ice must be taken into ac-
count. Prediction of the maximum depth of frost pen-
Heat conduction in soils and rocks is discussed in Sec- etration illustrates this type of problem. Theoretical
tion 9.5, and values for thermal properties are given in solutions of this problem are based on a mathematical
l
The Fourier equation for heat flow is
ria
where Ls is the latent heat of fusion of water and qƒ ⫺
T qu is the net rate of heat flow away from the interface.
qt ⫽ ⫺kt (9.165) Equation (9.169) can be written
z
ate
In the absence of freezing or thawing, thermal conti- dZ T T
Ls ⫽ kƒ ƒ ⫺ ku u (9.170)
nuity and conservation of thermal energy require that dt z z
the rate of change of thermal energy of an element plus
the rate of heat transfer into the element equal zero, where the subscripts u and f pertain to unfrozen and
that is, for the one-dimensional case frozen soil, respectively. Simultaneous solution of Eqs.
(9.168) and (9.170) gives the depth of frost penetra-
u
⫹
q
dM
⫽0 (9.166)
tion.
Stefan Formula The simplest solution is to assume
t z
that the latent heat is the only heat to be removed dur-
ing freezing and neglect the heat that must be removed
Using Eqs. (9.164) and (9.165), Eq. (9.166) may be
to cool the soil water to the freezing point, that is, the
written
thermal energy stored as volumetric heat is neglected.
This condition is shown by Fig. 9.46. For this case Eq.
hte
Z⫽ 冉2kƒ 冕 T dt
Ls
s
冊
1/2
(9.172)
(9.170) has been made for the conditions shown in Fig.
9.48, assuming that the soil has a uniform initial tem-
perature that is T0 degrees above freezing and that the
surface temperature drops suddenly to Ts below freez-
ing (Aldrich, 1956). The solution is
冉 冊
The integral of Ts dt is a measure of freezing intensity. 1/2
2kTst
It can be expressed by the freezing index F, which has Z⫽ (9.173)
units of degrees ⫻ time. Index F is usually given in Ls
degree-days. It is shown in relation to the annual tem-
perature cycle in Fig. 9.47. Freezing index values are where k is taken as an average thermal conductivity
derived from meteorological data. Methods for deter- for frozen and unfrozen soil. The dimensionless cor-
l
mination of freezing index values are given by Linell rection coefficient depends on the two parameters
ria
et al. (1963), Straub and Wegmann (1965), McCormick shown in Fig. 9.49. The thermal ratio is given by
(1971), and others. Maps showing mean freezing index
values are available for some areas. It is important T0
⫽ (9.174)
when using such data sources to be sure that there are Ts
ate
not local deviations from the average values that are
given. Different types of ground cover, local topogra- and the fusion parameter is
phy and vegetation, and solar radiation all influence
the net heat flux at the ground surface. C
⫽ T (9.175)
The Stefan equation can also be used to estimate the Ls s
summer thaw depth in permafrost; that is, the thickness
dM
of the active layer. In this case the ground thawing An averaged value for the volumetric heats of frozen
index, also in degree-days and derived from meteoro- and unfrozen soil can be used for C in Eq. (9.175).
logical data, is used in Eq. (9.172) in place of the In application, the quantity Tst in Eq. (9.173) is re-
freezing index (Konrad, 2001). placed by the freezing index, and Ts in (9.175) is given
Modified Berggren Formula The Stefan formula by F/t, where t is the duration of the freezing period.
overpredicts the depth of freezing because it neglects The coefficient corrects the Stefan formula for ne-
the removal of the volumetric heats of frozen and un- glect of volumetric heat. For soils with high water con-
hte
frozen soil. Simultaneous solution of Eqs. (9.168) and tent C is small relative to Ls; therefore, is small and
rig
py
Co
l
ria
ate
dM
hte
Figure 9.48 Thermal conditions assumed in the derivation of the modified Berggren for-
rig
mula.
the Stefan formula is reasonable. For arctic climates, depths of frost penetration may be misleading if esti-
where T0 is not much above the freezing point, is mates for a proposed pavement or other structure are
py
small, is greater than 0.9, and the Stefan formula is needed because of differences in ground surface char-
satisfactory. However, in more temperate climates and acteristics and because the pavement or foundation
in relatively dry or well-drained soils, the correction base will be at different water content and density than
becomes important. the surrounding soil.
Co
A comparison between theoretical freezing depths The solutions do not account for flow of water into
and a design curve proposed by the Corps of Engineers or out of the soil or the formation of ice lenses during
is shown in Fig. 9.50 for several soil types. The the- the freezing period. This may be particularly important
oretical curves were developed by Brown (1964) using when dealing with frost heave susceptible soils or
the modified Berggren equation and the thermal prop- when developing frozen soil barriers for the cutoff of
erties given in Fig. 9.13. groundwater flow. Methods for prediction of frost
Consideration should be given to the effect of dif- depth in soils susceptible to ice lens formation and the
ferent types of surface cover on the ground surface rate of heave are given by Konrad (2001). The initia-
temperature because air temperature and ground tem- tion of freezing of flowing groundwater requires that
perature are not likely to be the same, and the effects the rate of volumetric and latent heat removal be high
of thermal radiation may be important. Observed enough so that ice can form during the residence time
l
ria
ate
dM
Figure 9.49 Correction coefficients for use in the modified Berggren formula (from Aldrich,
1956).
hte
of an element of water moving between the boundaries a complex manner. If the cooling rate is too high, then
of the specified zone of solidification. the soil freezes before water can migrate to an ice lens,
so the heave becomes only that due to the expansion
of water on freezing.
rig
Frost Heaving
Three conditions are necessary for ice lens forma-
Freezing of some soils is accompanied by the forma-
tion and frost heave:
tion of ice layers or ‘‘lenses’’ that can range from a
millimeter to several centimeters in thickness. These 1. Frost-susceptible soil
lenses are essentially pure ice and are free from large
py
2. Freezing temperature
numbers of contained soil particles. The ground sur- 3. Availability of water
face may ‘‘heave’’ by as much as several tens of cen-
timeters, and the overall volume increase can be many Frost heaving can occur only where there is a water
times the 9 percent expansion that occurs when water table, perched water table, or pocket of water reason-
Co
freezes. Heave pressures of many atmospheres are ably close to the freezing front.
common. The freezing of frost-susceptible soils be- Frost-Susceptible Soils Almost any soil may be
neath pavements and foundations can cause major dis- made to heave if the freezing rate and water supply
tress or failure as a result of uneven uplift during are controlled. In nature, however, the usual rates of
freezing and loss of support on thawing, owing to the freezing are such that only certain soil types are frost
presence of large water-filled voids. Ordinarily, ice susceptible. Clean sands, gravels, and highly plastic
lenses are oriented normal to the direction of cold-front intact clays generally do not heave. Although the only
movement and become thicker and more widely sep- completely reliable way to evaluate frost susceptibility
arated with depth. is by some type of performance test during freezing,
The rate of heaving may be as high as several mil- soils that contain more than 3 percent of their particles
limeters per day. It depends on the rate of freezing in finer than 0.02 mm are potentially frost susceptible.
l
ria
ate
Figure 9.50 Predicted frost penetration depths compared with the Corps of Engineers’ de-
sign curve (Brown, 1964). Curve a—sandy soil: dry density 140 lb / ft3, saturated, moisture
content 7 percent. Curve b—silt, clay: dry density 80 lb / ft3, unsaturated, moisture content
2 percent. Curve c—sandy soil; dry density 140 lb / ft3, unsaturated, moisture content 2
percent. Curve d—silt, clay: dry density 120 lb / ft3, moisture content 10 to 20 percent (sat-
dM
urated). Curve e—silt, clay: dry density 80 lb / ft3 saturated, moisture content 30 percent.
Curve f—Pure ice over still water.
Frost-susceptible soils have been classified by the A method for the evaluation of frost susceptibility
Corps of Engineers in the following order of increasing that takes project requirements and acceptable risks
hte
frost susceptibility: and freezing conditions into account as well as the soil
type is described by Konrad and Morgenstern (1983).
Mechanism of Frost Heave The formation of ice
lenses is a complex process that involves interrelation-
Group
ships between the phase change of water to ice, trans-
(increasing
port of water to the lens, and general unsteady heat
rig
l
molecules from the film into the ice, while additional
ria
water molecules enter the film to maintain its thick-
ness. It is energetically easier to bring water to the ice
from adjacent pores than to freeze the adsorbed water
on the particle or to propagate the ice through a pore
constriction.
ate
The driving force for water transport to the ice is an
equivalent hydrostatic pressure gradient that is gener-
ated by freezing point depression, by removal of the
water from the soil at the ice front, which creates a
higher effective stress in the vicinity of the ice than
Figure 9.51 Temperature versus depth relationships in a
dM
away from it, by interfacial tension at the ice–water
interface, and by osmotic pressure generated by the freezing soil.
high concentration of ions in the water adjacent to the
ice front. Ice formation continues until the water ten-
sion in the pores supplying water becomes great The quantity L is the latent heat. Supercooling of 1C
enough to cause cavitation, or decreased upward water is sufficient to lift 12.5 kg a distance of 10 mm. Al-
flow from below leads to new ice lens formation be- ternatively, the energy for heave may originate from
hte
neath the existing lens. the thin water films at the ice surface (Kaplar, 1970).
The processes of freezing and ice lens formation As long as water can flow to a growing ice lens fast
proceed in the following way with time according to enough, the volumetric heat and latent heat can pro-
Martin’s theory. If homogeneous soil, at uniform water duce a temporary steady-state condition so that (dT/
content and temperature T0 above freezing, is subjected dz)A ⫽ (dT/dz)B. For example, silt can supply water at
to a surface temperature Ts below freezing, then the a rate sufficient for heave at 1 mm/h. After some time
rig
variation of temperature with depth at some time is as the ability of the soil to supply water will drop because
shown in Fig. 9.51. The rate of heat flow at any point the water supply in the region ahead of the ice front
is ⫺kt(dT/dz). If dT/dz at point A is greater than at becomes depleted, and the hydraulic conductivity of
point B, the temperature of the element will drop. the soil drops, owing to increased tension in the pore
When water goes to ice, it gives up its latent heat, water. This is illustrated in Fig. 9.52, where hydraulic
py
which flows both up and down and may slow or stop conductivity data as a function of negative pore water
changes in the value of dT/dz for some time period, pressure are shown for a silty sand, a silt, and a clay,
thus halting the rate of advance of the freezing front all compacted using modified AASHTO effort, at a
into the soil. water content about 3 percent wet of optimum.
Co
Ground heave results from the formation of a lens A small negative pore water pressure is sufficient to
at A, with water supplied according to the mechanisms cause water to drain from the pores of the silty sand,
indicated above. The energy needed to lift the over- and this causes a sharp reduction in hydraulic conduc-
lying material, which may include not only the soil and tivity. Because the clay can withstand large negative
ice lenses above, but also pavements and structures, is pore pressures without loss of saturation, the hydraulic
available because ice forms under conditions of super- conductivity is little affected by increasing reductions
cooling at a temperature T X ⬍ TFP, where TFP is the in the pore pressure (increasing suction). The small
freezing temperature. The available energy is decrease that is observed results from the consolidation
needed to carry the increased effective stress required
L(TFP ⫺ T X) to balance the reduction in the pore pressure. For the
F ⫽ (9.176)
TFP silt, water drainage starts when the suction reaches
l
steady-state condition for lens growth can be main-
ria
tained for a longer time, thus enabling formation of a
thicker lens.
More quantitative analyses of the freezing and frost
heaving processes in terms of segregation potential,
rates, pressures, and heave amounts are available. The
ate
Proceedings of the International Symposia on Ground
Freezing, for example, Jones and Holden (1988),
Nixon (1991), and Konrad (2001) provide excellent
sources of information on these issues.
dM Thaw Consolidation and Weakening
When water in soil freezes, it expands by about 9 per-
cent of its original volume. Thus a fully saturated soil
increases in volume by 9 percent of its porosity, even
Figure 9.52 Hydraulic conductivity as a function of negative
in the absence of ice segregation and frost heave. The
pore water pressure (from Martin and Wissa, 1972).
expansion associated with freezing disrupts the origi-
hte
the latent heat raises the freezing temperature to the it had prior to freezing. The lower the density of the
normal freezing point. Hence, there is no supercooling soil, the greater is the amount of thaw consolidation.
and no heave. Negative pore pressure development at The total settlement of foundations and pavements as-
the ice front causes the hydraulic conductivity to drop, sociated with thawing is the sum of that due to (1) the
so water cannot be supplied to form ice lenses. Thus phase change, (2) melting of segregated ice, and (3)
py
sands freeze homogeneously with depth. In clay, the compression of the weakened soil structure.
hydraulic conductivity is so low that water cannot be Testing of representative samples under appropriate
supplied fast enough to maintain the temporary steady- boundary conditions is the most reliable means for
state condition needed for ice lens growth. Heave in evaluating thaw consolidation. Samples of frozen soil
Co
clay only develops if the freezing rate is slowed to well are allowed to thaw under specified levels of applied
below that in nature. Silts and silty soils have a com- stress and under defined drainage conditions, and the
bination of pore size, hydraulic conductivity, and freez- decrease in void ratio or thickness is determined. An
ing point depression that allow for large heave at example of the effects of freezing and thawing on the
normal freezing rates in the field. compression and strength of initially undisturbed Bos-
The freezing temperature penetrates ahead of a com- ton blue clay is shown in Fig. 9.53 from Swan and
pleted ice lens, and a new lens will start to form only Greene (1998). These tests were done as part of a
after the temperature drops to the nucleation temper- ground freezing project for ground strengthening to en-
ature. The nucleation temperature for a new lens may able jacking of tunnel sections beneath operating rail
be less than that for the one before because of reduced lines during construction of the recently completed
saturation and consolidation from the previous flows, Central Artery/Tunnel Project in Boston. Detailed
0
2
4 C1-UF
e0 = 1.064
8 C4-FT
e0 = 1.171
10
12
14
l
16
ria
18
20
22
10 100 1000 10000
ate
Effective Stress, σc (kPa)
(a)
120
dM 100
Deviator Stress, σ1 – σ3 (kPa)
UUC1-UF
80 (σ1–σ3)max = 109.6 kPa
ε1 = 2.3%
su/σ3cell = 0.36
60 e0 = 1.02; w = 37.5%
hte
40
UUC4-FT
(σ1– σ3)max = 42.4 kPa
20 ε1 = 12.8%
su/σ3cell = 0.14
e0 = 1.13; w = 43.2%
rig
0
0 5 10 15 20 25
Axial Strain. %
(b)
Figure 9.53 (a) Comparison between the compression behavior of unfrozen (C1-UF) and
py
frozen then thawed (C4-FT) samples of Boston blue clay. (b) Deviator stress vs. axial strain
in unconsolidated–undrained triaxial compression of unfrozen (UUC1-UF) and frozen and
thawed (UUC4-FT) Boston blue clay (from Swan and Greene, 1998).
Co
analysis of the thaw consolidation process and its an- cations are usually temporary, and they have the ad-
alytical representation is given by Nixon and Ladanyi vantage that the ground is not permanently altered,
(1978) and Andersland and Anderson (1978). except for such property changes as may be caused by
the freeze–thaw processes. Returning the ground to its
pristine state may be important for environmental rea-
Ground Strengthening and Flow Barriers by
sons where alternative methods for stabilization could
Artificial Ground Freezing
permanently change the state and composition of the
Artificial ground freezing has applications for formation subsoil.
of seepage cutoff barriers in situ, excavation support, Freezing is usually accomplished by installation of
and other ground strengthening purposes. These appli- freeze pipes and circulation of a refrigerant. For emer-
l
Natural Strain, ε -%
times the water content in percent (Shuster, 1972). 80
ria
However, if the rate of groundwater flow exceeds about Pa
M
1.5 m/day, it may be difficult to freeze the ground 55 Temperature
0. Effect
without a very high refrigeration capacity to ensure =
,σ
that the necessary temperature decrease and latent heat °C
0
removal can be accomplished within the time any el- =
ate
T
ement of water is within the zone to be frozen. 40
The long-term strength and stress–strain character-
istics of frozen ground depend on the ice content, tem- Pa
,σ= 0.55 M
perature, and duration of loading. The short-term T= –2.2 °C
Stress effect
strength under rapid loading, which can be up to 20
dM
MPa at low temperature, may be 5 to 10 times greater T = –2.2 °C, σ = 0.138 MPa
than that under sustained stresses. That is, frozen soils 0
0 10 tf 20 30
are susceptible to creep strength losses (Chapter 12). Time, t (hr)
The deformation behavior of frozen soil is viscoplastic,
and the stress and temperature have significant influ- Figure 9.54 Creep curves for a frozen organic silty clay
ence on the deformation at any time. The creep curves (from Sanger and Sayles, 1979).
in Fig. 9.54 illustrate these effects. The onset of the
hte
frozen soil mass. Design and implementation consid- damentals is important, not only because of the
erations for use of ground freezing in construction are insights provided but also because many of the same
given by Donohoe et al. (1998). considerations apply to the several other types of flows
that are known to be important—chemical, electrical,
and thermal. Knowledge of one is helpful in the un-
py
9.23 CONCLUDING COMMENTS derstanding and quantification of the other because the
mathematical descriptions of the flows follow similar
Conductivity properties are one of the four key dimen- force-flux relationships.
sions of soil behavior that must be understood and At the same time it is necessary to take into account
Co
quantified for success in geoengineering. The other that the flows of fluids of different composition and the
three dimensions are volume change, deformation and application of hydraulic, chemical, electrical, and ther-
strength, and the influences of time. They form the mal driving forces to soils can cause changes in com-
subjects of the following three chapters of this book. positions and properties, with differing consequences,
Water flows through soils and rocks under fully sat- depending on the situation. Furthermore, as examined
urated conditions have been the most studied, and hy- in considerable detail in this chapter, flow coupling can
draulic conductivity properties, their determination and be important, especially advective and diffusive chem-
application for seepage studies of various types, con- ical transport, electroosmotic water and chemical flow,
struction dewatering, and the like are central to geo- and thermally driven moisture flow. Considerable im-
technical engineering. One objective of this chapter has petus for research on these processes has been gener-
been to elucidate the fundamental factors that control ated by geoenvironmental needs, including enhanced
and more economical waste containment and site re- 6. How can the effects of incompatibility between
mediation strategies. chemicals in a waste repository and a compacted
Ground freezing, in addition to its importance in en- clay liner best be minimized?
gineering and construction in cold regions, is seeing
7. Two parallel channels, one with flowing water and
new applications for temporary ground stabilization
the other with contaminated water, are 100 ft apart.
needed for underground construction in sensitive urban
areas. The surface elevation of the contaminated channel
is 99 ft, and the surface elevation of the clean wa-
ter channel is at 97 ft. The soil between the two
QUESTIONS AND PROBLEMS channels is sand with a hydraulic conductivity of
1 ⫻ 10⫺4 m/s, a dry unit weight of 100 pcf, and
1. A uniform sand with rounded particles has a void a specific gravity of solids of 2.65. Estimate the
l
ratio of 0.63 and a hydraulic conductivity, k, of 2.7 time it will take for seepage from the contaminated
ria
⫻ 10⫺4 m/s. Estimate the value of k for the same channel to begin flowing into the initially clean
sand at a void ratio of 0.75. channel. Make the following assumptions and sim-
2. The soil profile at a site that must be dewatered plifications:
consists of three homogeneous horizontal layers of a. Seepage is one dimensional.
ate
equal thickness. The value of k for the upper and b. The only subsurface reaction is adsorption onto
lower layers is 1 ⫻ 10⫺6 m/s and that of the mid- the soil particles.
dle layer is 1 ⫻ 10⫺4 m/s. What is the ratio of the c. The soil–water partitioning coefficient is 0.4
average hydraulic conductivity in the horizontal cm3 /g.
direction to that in the vertical direction?
dM d. Hydrodynamic dispersion can be ignored.
3. Consider a zone of undisturbed San Francisco Bay
8. For the compacted clay waste containment liner
mud free of sand and silt lenses. Comment on the
shown below and assuming steady-state condi-
probable effect of disturbance on the hydraulic
tions:
conductivity, if any. Would this material be ex-
pected to be anisotropic with respect to hydraulic a. What is the contaminant transport for pure
conductivity? Why? molecular diffusion?
b. What is the contaminant transport rate for pure
hte
l
ate collection layer is atmospheric and that the only
meability equal to 0.3 ⫻ 10⫺8 m/s per V/m
ria
fluxes across the liner are water and electricity. The
and a hydraulic conductivity of 6 ⫻ 10⫺9 m/s. characteristics of the compacted clay liner are:
Starting from the general relationship
Hydraulic conductivity
ate
Ji ⫽ Lij Xj kh ⫽ 1 ⫻ 10⫺7 m/s
Electroosmotic coefficients
derive an expression for the pore water tension ke ⫽ 2 ⫻ 10⫺9 m2 /s V
that may be developed under ideal conditions
for consolidation of the clay by electroosmosis
dM ki ⫽ 0.2 ⫻ 10⫺6 m3 /s amp
and compute the value that should develop at a a. Wire mesh is proposed for use as electrodes.
point where the voltage is 25 V. Be sure to Where would you place the anode and cathode
indicate correct units with your answers. meshes?
b. In the absence of electrochemical effects or b. If the waste pond is to be filled to an average
cavitation, would you consider your answer to depth of 6 m, what voltage drop should be
part (a) to represent an upper or lower bound maintained between the electrodes?
estimate of the pore water tension? Why?
hte
such that the results of a hydraulic conductivity the complete set of equations that would be re-
test in which streaming potential is measured can quired to describe all the flows across the liner
be used to predict the volume flow rate during during electroosmosis. Define all terms.
electroosmosis in terms of the electrical current. e. Will maintenance of a no hydraulic flow con-
py
Starting with the general equations for coupled dition ensure that no leachate will escape
electrical and hydraulic flow, derive Saxen’s law. through the clay liner? Why?
What will be the drainage rate from a soil, in
13. a. Estimate the minimum footing depths for struc-
m3 /h amp, if the streaming potential is 25 mV/
tures in a Midwestern city where the freezing
atm? What will be the cost of electrical power per
Co
mation? Assume a frozen ground temperature through the liner as a function of the hydraulic
of 32F. conductivity. Show in the same diagram the
c. If a pavement is to be placed over the soil, what proportions of the total that are attributable to
thickness of granular base course should be diffusion and advection.
used to prevent freezing of the subgrade? The Assume that the leachate collection layer is
base course will be compacted to a dry density fully drained, but for purposes of analysis the
of 125 lb/ft3 at a water content of 15 percent. fluid level can be considered at the bottom of
If the pavement structure is to contain an 8- the clay. Determine the leakage rate through
inch-thick Portland cement concrete surface the liner per unit area as a function of the hy-
layer, will your result tend to overestimate or draulic conductivity and show it on a diagram.
underestimate the base thickness required? 15. The diagram below shows the cross section of a
l
Why? tunnel and underlying borehole in which waste
ria
14. A compacted fine-grained soil is to be used as a canisters for spent nuclear fuel are located. Such
liner for a chemical waste storage area. Free liquid an arrangement is proposed for deep (e.g., several
leachate and possibly some heavier than water hundred meters) burial of nuclear waste in crys-
free phase nonsoluble, nonpolar organic liquids talline rock. The surrounding rock can be assumed
ate
(DNAPLs) may accumulate in some areas as a re- fully saturated, and the groundwater table will be
sult of rupturing and corrosion of the drums in within a few tens of meters of the ground surface.
which they were stored. Two sources of soil for Thermal studies have shown that the temperature
use in the liner are available. They have the fol- of the waste canister will rise to as high as 150C
lowing properties: at its surface. A canister life of about 100 years is
anticipated using either stainless steel or copper
dM for the material. The surrounding environment
must be safe against leakage of radionuclides from
Property Soil A Soil B
the repository for a minimum of 100,000 years.
Unified class (CH) (CL)
Liquid limit (%) 90 45
Plastic limit (%) 30 25
Clay size (%) 50 30
hte
capacity (meg/100 g)
l
ria
ate
dM
hte
rig
py
Co
l
ria
ate
10.1 INTRODUCTION the void ratios encompass virtually the full range from
fresh sediments to shale. Mechanical and chemical
Volume changes in soils are important because they changes accompany and influence the densification
determine settlements due to compression, heave due process. In general, the void ratio–effective pressure
to expansion, and contribute to deformations caused by relationship is related to grain size and plasticity in the
shear stresses. Changes in volume cause changes in
dM manner shown by Fig. 10.2b.
strength and deformation properties that, in turn, influ- Particle size and shape, which together determine
ence stability. Volume changes are induced by changes specific surface area, are the most important factors
in applied stresses, chemical and moisture environ- influencing both the void ratio at any pressure and the
ments, and temperature. The effects of stress changes effects that physicochemical and mechanical factors
are generally the most important and have been the have on consolidation and swelling (Meade, 1964).
most studied. Particle size and shape are direct manifestations of
In this chapter, factors contributing to volume composition, with increasing colloidal activity and ex-
hte
that control the behavior are the same. come nearly equal. Swelling index values for three
clay minerals, muscovite, and sand are listed in Table
10.1. For undisturbed natural soils the swelling index
10.2 GENERAL VOLUME CHANGE BEHAVIOR
values are usually less than 0.1 for nonexpansive ma-
OF SOILS terials to more than 0.2 for expansive soils.
The compressibility of dense sands and gravels is
Soil void ratio is normally in the range of about 0.5 to far less than that of normally consolidated clays; none-
4.0, as shown in Fig. 10.2. Although the range of pres- theless, volume changes under high pressures may be
sures of interest in most cases (up to a few hundred substantial in granular materials as shown in Fig. 10.3.
kilopascals) is relatively small on a geological scale, At low stress levels, the compressibility of sand de-
325
l
ria
ate
Figure 10.1 Idealized void ratio–effective stress relationships for a compressible soil.
dM
hte
rig
py
Co
Figure 10.2 Compression curves for several soils (redrawn from Lambe and Whitman,
1969).
l
Ethyl alcohol 1.10 0.06
ria
Carbon tetrachloride 1.10 0.05
Dry air 1.36 0.04
Illite Water, sodium, 1 1.77 0.37
Water, sodium, 1 ⫻ 10⫺3 2.50 0.65
Water, calcium, 1 1.51 0.28
ate
Water, calcium, 1 ⫻ 10⫺3 1.59 0.31
Ethyl alcohol 1.48 0.19
Carbon tetrachloride 1.14 0.04
Dry air 1.46 0.04
Smectite Water, sodium, 1 ⫻ 10⫺1 5.40 1.53
Water, sodium, 5 ⫻ 10⫺4 11.15 3.60
dM Water, calcium, 1 1.84 0.26
Water, calcium, 1 ⫻ 10⫺3 2.18 0.34
Ethyl alcohol 1.49 0.10
Carbon tetrachloride 1.21 0.03
Muscovite Water 2.19 0.42
Carbon tetrachloride 1.98 0.35
hte
Figure 10.3 Compressibility of three sands under high pres- Three different relationships between the present over-
sure (from Pestana and Whittle, 1995). burden effective stress v0 and the maximum past over-
l
ria
ate
dM
hte
Figure 10.4 Field compressibility of earth and rockfill materials (from Wilson, 1973). Re-
printed with permission from John Wiley & Sons.
burden effective stress vm are possible for the soil at solidation, and overconsolidated clays become
a site: normally consolidated clays when loaded beyond
rig
pleted, and disturbance that causes a structure is larger than the maximum past pressure even
breakdown and decrease in effective stress. though the soil has not experienced a pressure
2. vm ⫽ v0 —Normally Consolidated The soil is greater than the present overburden stress.
in effective stress equilibrium with the present
overburden effective stress. Surprisingly few, if Accurate knowledge of the maximum past consoli-
any, deposits have been encountered that are ex- dation pressure is needed for reliable predictions of
actly normally consolidated. Most are at least settlement and to aid in the interpretation of geologic
very slightly overconsolidated as a result of history. If the recompression to virgin compression
processes of the type summarized in Table 10.2. curve does not show a well-defined break, such as at
Underconsolidated soil behaves as normally point B in Fig. 10.1, the preconsolidation pressure is
consolidated soil until the end of primary con- difficult to determine. Gentle curvature of the com-
Stress
History In situ Stress
Category Description Profile Condition Remarks/References
A. Mechanical one 1. Changes in total Uniform with K0, but value at Most obvious and
dimensional vertical stress constant given OCR varies easiest to identify
(overburden, p ⫺ v0 for reload versus
glaciers, etc.) unload
2. Changes in pore (except with
l
pressure seepage)
ria
(water table,
seepage
conditions,
etc.)
B. Desiccation 1. Drying due to Often highly Can deviate from Drying crusts found at
ate
evaporation, erratic K0, e.g., isotropic surface of most
vegetation, capillary stresses deposits; can be at
etc. depth within deltaic
deposits
2. Drying due to
freezing
C. Drained creep
dM 1. Long-term Uniform with K0, but not Leonards and Altschaeffl
(aging) secondary constant necessarily (1964); Bjerrum
compression p / v0 normally (1967)
consolidated
value
D. Physicochemical 1. Natural Poorly understood and
hte
pression curve over the preconsolidation pressure Fig. 10.6 from Leroueil et al. (1990). The higher values
range is characteristic of sands, weathered clays, heav- of apparent preconsolidation pressure associated with
ily overconsolidated clays, and disturbed clays. the faster rates of loading reflect the influences of the
The rate of loading and time have significant effects viscous resistance of the soil structure. The rate-
on the equilibrium void ratio–effective stress relation- dependent value of preconsolidation pressure, p can
ship, especially for sensitive structured clays as shown be approximated by (e.g., Leroueil et al., 1985)
in Fig. 10.5. It is not surprising, therefore, that rate of
loading and time influence also the measured precon- log( p) ⫽ A ⫹ B log(˙a) (10.1)
solidation pressure. The preconsolidation pressure de-
creases as the duration of load application increases where ˙ a is the vertical strain rate in one-dimensional
and as the rate of deformation decreases, as shown by consolidation, and A and B are fitting parameters. Typ-
50
Conventional
Consolidation Test at
40 Conventional Consolidation
25 °C (At End of Primary
Test at 25 °C (After 24
Consolidation State)
Hours of Loading)
l
30
10 -9 10 -8 10 -7 10 -6 10-5 10-4
ria
Volumetric Strain Rate (s-1)
ate
times after the completion of primary consolidation.
mation rate on compression curves (Leroueil et al., 1990). forces. Physicochemical interactions are most impor-
(a) Ottawa clay (data from Crawford, 1964). (b) Bäckebol tant in the formational stages of fine-grained soil de-
clay (data from Sällfors, 1975). posits when they are at low pressures and high void
ratios.
Chemical and Organic Environment Chemical
py
ical examples of the fitting for the results of different precipitates cement particles together. Organic matter
types of compression tests on Berthierville clay are influences surface forces and water adsorption prop-
shown in Fig. 10.7 (Leroueil and Marques, 1996). The erties, which, in turn, increase the plasticity and com-
effect of temperature on preconsolidation pressure can pressibility. Expansion of pyrite minerals in some
Co
also be seen, and this is further discussed in Section shales and other earth materials as a result of oxidation
10.12. The data in Figs. 10.6 and 10.7 also illustrate caused by exposure to air and water has been the
the difficulties and uncertainties in determining the true source of significant structural damage (Bryant et al.,
in situ conditions from the results of laboratory tests. 2003). Temperature changes may cause changes in hy-
dration states of some salts leading to volume changes.
Mineralogical Detail Small differences in certain
10.4 FACTORS CONTROLLING RESISTANCE characteristics of expansive clay minerals can have ma-
TO VOLUME CHANGE jor effects on the swelling of a soil.
Fabric and Structure Compacted expansive soils
Both compositional and environmental factors influ- with flocculent structures may be more expansive than
ence volume change, so meaningful quantitative pre- those with dispersed structures. Figure 10.8 is an ex-
l
has relatively little effect on the compression behavior
ria
after the initial fabric has formed and the structure has
stabilized under a moderate effective stress. This is in
accordance with the principle of chemical irreversibil-
ity of clay fabric, discussed in Section 8.2. The leach-
ing of normally consolidated marine clay at high water
ate
Figure 10.8 Effect of structure and electrolyte concentration content, however, may be sufficient to cause a small
of absorbed solution on swelling of compacted clay (adapted
reduction in volume owing to changes in interparticle
from Seed et al. 1962a).
forces (Kazi and Moum, 1973; Torrance, 1974).
Stress Path The amount of compression or swell-
ing associated with a given change in stress usually
ample. At pressures less than the preconsolidation
dM
pressure, the soil with a flocculent structure was less
compressible than the same soil with a dispersed struc-
depends on the path followed. Loading or unloading
from one stress to another in stages can give consid-
erably different volume change behavior than if the
ture. The reverse is generally true for pressures greater
stress change is done in one step. An example for
than the preconsolidation pressure.
swelling of a compacted sandy clay is shown in Fig.
Stress History An overconsolidated soil is less
10.10. Each sample was placed under water after com-
compressible but more expansive than the same ma-
paction and allowed to swell under different surcharge
terial initially at the same void ratio but normally con-
hte
l
ria
Figure 10.10 Effect of unloading stress path on swelling of a compacted sandy clay (Seed
et al. 1962a).
ate
dM
hte
rig
py
Figure 10.11 Comparison of compression and swelling curves for several clays and sand–
mica mixtures (from Terzaghi, 1931).
Co
chanical behavior of mica–sand mixtures is given in terpreted, probably erroneously, as balanced entirely
Chapter 11. by interparticle repulsion.
Cross-linking adds rigidity to soil fabric, especially The importance of grain crushing increases with in-
clays containing platy particles. Particles and particle creasing particle size and confining stress magnitude.
groups act as struts whose resistance depends both on Particle breakage is a progressive process that starts at
their bending resistance and on the strengths of the relatively low stress levels because of the wide disper-
junctions at their ends. According to van Olphen sion of the magnitudes of interparticle contact forces.
(1977), cross-linking is important even in ‘‘pure clay’’ The number of contacts per particle depends on gra-
systems, where the confining pressure is sometimes in- dation and density, and the average contact force in-
l
containing idle particles, particulate mechanics analy- rounded particles.
ria
ses of behavior that depend on such quantities as av- 3. Uniform soils compress and crush more than
erage number of particles per unit area or per unit well-graded soils with the same maximum grain
volume, average number of contacts per particle, and size.
the like lose their relevance unless the analyses allow 4. Under a given stress, compression and crushing
for their existence. continue indefinitely at a decreasing rate.
ate
The resistance to grain crushing or breakage de- 5. Volume change during compression depends pri-
pends on the strength of the particles, which, in turn, marily on the major principal stress and is inde-
depends on mineralogy and the soundness of the pendent of the principal stress ratio.
grains. Failure may be by compression, shear, or in a 6. The higher the principal stress ratio (Kc ⫽ 1c /
split tensile mode. Quartz grains are more resistant
dM 3c) during consolidation, the greater the amount
than feldspar, but there is greater variability in crushing of grain crushing.
and splitting resistance with changes in particle size
for quartz than for feldspar. Particle crushing results in increase in fines content
The amount of grain crushing to be expected for with increasing confining pressure. An example of the
rockfills and gravels is summarized in Table 10.4. In change in particle size distribution curve with increas-
this table, Bq is the proportion of the solid phase by ing confining pressure is shown in Fig. 10.13 (Fuku-
weight that will undergo breakage, and qi is the con- moto, 1992). Particle crushing can be quantified by
hte
centration of solids [Vs /V ⫽ 1/(1 ⫹ e)]. Hardin’s (1985) relative breakage parameter Br, which
Average
Grain Grain Contact
Contacts/ Contacts/ Force for
Particle Particle ⫽ 1 atm
py
Crushing Particle
Grain Size Strength Breakage
Samples Distribution of Grains Bqqia
El infiernillo Well-graded High 0.02–0.10
silicified con- rockfills and for 5
glomerate gravels 1f 80
kg/cm2
Pinzandaran sand
l
and gravel
ria
San Francisco
basalt (grada-
tions 1 and 2)
El infiernillo Somewhat uni- High 0.10–0.20
diorite form rockfills for 5
ate
1f 80
kg/cm2
El granero slate Well-graded Low
(gradation A) rockfills
Mica granitic–
gneiss
dM
(gradation X)
Mica granitic– Uniform rockfill Low Increases
gneiss produced by with 1f ⬊
(gradation Y) blasting meta- maximum
morphic rocks value ⫽
(Cu ⬍ 5) 0.30
hte
a
Bq is grain breakage parameter; qi is initial concentration of solids; 1f
is major principal stress at failure.
From Marsal, 1973. Reprinted with permission of John Wiley & Sons.
rig
py
Co
Figure 10.12 Comparison of crushing of soils with different initial grain sizes for isotropic
compression under 8 MPa (from Lee and Farhoomand, 1967). Reproduced with permission
from the National Research Council of Canada.
l
ria
0
0.01 0.1 1
Grain Size (mm)
(a)
ate
100
Landstejin Sand : Initial Grading 4-7 mm
Percent Finer by Weight
0
hte
0.01 0.1 1 10
Grain Size (mm)
(b)
Figure 10.13 Change in particle size distribution curve with increasing confining pressure:
rig
(a) Ottawa sand and (b) Landstejn sand (from Fukumoto, 1992).
is defined in Fig. 10.14. The increase in Br with iso- solidation of these clays is related to sequential break-
py
tropic compression pressure is shown in Fig. 10.15 for age of clay aggregates into smaller aggregates as con-
Dog’s Bay carbonate sand (Coop and Lee, 1993). The solidation pressure increases (Bolton, 2000).
figure also shows the increase in Br at critical-state
failure (discussed further in Chapter 11). A unique
Co
particle breakage characteristic at failure is obtained 10.6 FABRIC, STRUCTURE, AND VOLUME
irrespective of shearing conditions (i.e., undrained CHANGE
triaxial, drained triaxial, or constant mean pressure
shearing). Collapse, shrinkage, and compression are due to par-
Aggregates of clay mineral particles are often ob- ticle rearrangements from shear and sliding at inter-
served in clays, and intact aggregate clusters of clay particle contacts, disruption of particle aggregates, and
particles can be considered as the smallest units con- grain crushing. Thus, both the arrangement of particles
trolling the macroscopic mechanical behavior. These and particle groups and the forces holding them in
aggregate clusters behave in some ways similarly to place are important. Swelling depends strongly on
granular particles (e.g., Barden, 1973, and Collins and physicochemical interactions between particles, but
McGown, 1974). It can be conceptualized that the con- fabric also plays a role.
l
vertical shrinkage on drying is greater than lateral
ria
shrinkage. For example, the vertical shrinkage of
Seven Sisters clay was three times greater than the hor-
izontal shrinkage (Warkentin and Bozozuk, 1961).
Collapse
ate
Figure 10.14 Definition of relative breakage parameter Br
by Hardin (1985). Collapse, as a result of wetting under constant total
stress, is an apparent contradiction to the principal of
effective stress discussed in Chapter 7. The addition of
water increases the pore water pressure and reduces
dM the effective stress; hence, expansion might be ex-
pected. The apparent anomaly of volume decrease
under decreased effective stress is because of the
application of continuum concepts to a phenomenon
that is controlled by particulate behavior at contact lev-
els for unsaturated soils. Collapse requires:
1. An open, low-density, partly unstable, partly
hte
saturated fabric
2. A high enough total stress that the structure is
metastable
3. A strong enough clay binder or other cementing
agent to stabilize the structure when dry
rig
Compression
Drying shrinkage of fine-grained soils is caused by
particle movements resulting from pore water tensions Sands In Chapter 8 it was shown that the volume
developed by capillary menisci. If two samples of clay changes during the shear of samples of sand at the
are at the same initial water content but have different same void ratio but with different initial fabrics can be
fabrics, the one that is the more deflocculated and dis- different. Different volume change tendencies for dif-
persed shrinks the most. This is because the average ferent fabrics developed resulting from different meth-
pore sizes are smaller in the deflocculated sample, thus ods of sample preparation have also manifested
allowing greater capillary stresses, and because of eas- themselves by differences in liquefaction behavior un-
ier relative movements of particles and particle groups. der undrained cyclic loading (see Fig. 8.22).
Natural
Water Dry Void Dry Void
Content Ratio Ratio
Clay (%) Sensitivity Undisturbed Remolded
Boston blue 35.6 6.8 0.69 0.50
Boston blue 37.5 5.8 0.75 0.53
Fore River, Maine 41.5 4.5 0.65 0.46
l
Goose Bay, Labrador 29.0 2.0 0.60 0.55
ria
Chicago 39.7 3.4 0.65 0.55
Beauharnois, Quebec 61.3 5.5 0.76 0.70
St. Lawrence 53.6 5.4 0.79 0.66
ate
The compression behavior of a natural intact ce- strengths of cementation bonding and particles on the
mented calcarenite sand is shown in Fig. 10.16 (Cuc- compression behavior of structured soils.
covillo and Coop, 1997). Similarly to structured clays,
dM Clays Compression curves obtained by odometer
the initial compressibility before yielding is stiff due tests on undisturbed and remolded Leda (Champlain)
to cementation. If the cementation is stronger than the clay, illite, and kaolinite are shown in Fig. 10.17. Li-
particle crushing strength, the compression line will lie quidity index is used as an ordinate, and the sensitivity
to the right of the normal compression line of the un- curves from Fig. 8.49 are superimposed.
cemented reconstituted sand. If the cementation is Curve A is for undisturbed Leda clay at an initial
weaker than the particle crushing strength, the com- water content corresponding to a liquidity index of
pression curve will merge gradually toward that of the 1.82. Because the sensitivity contours were developed
hte
uncemented sand before yielding (Cuccovillo and for normally consolidated clays, they cannot be used
Coop, 1999). This highlights the importance of relative to estimate sensitivity for stresses less than the precon-
solidation pressure. After the preconsolidation stress
has been exceeded the curve cuts sharply across the
sensitivity contours, indicating a large decrease in sen-
p (kPa)
sitivity as the structure is broken down by compres-
rig
2.40
terparticle and interaggregate shear stresses caused by
the applied compressive stress begin to exceed the
bond strengths, the degree of structural metastability
1.20 decreases, and the sensitivity decreases.
4 5 6 7 8 9 10 11 12
Curve D is for kaolinite remolded at a liquidity in-
In p(kPa)
dex of 0.98. It differs considerably from curve B. This
Figure 10.16 Isotropic compression curves of intact and re- is consistent with the results of other studies that show
constituted calcarenite sand specimens (from Cuccovillo and that the compression behavior, and therefore also the
Coop, 1997). structure, are different for a given clay remolded at
l
ria
ate
dM
hte
developed in the kaolinite of curve D than that of curve that is initiated by reduction of effective stress by un-
B. These observations show that both the concentration loading and/or addition of water. For example, an ex-
of clay in suspension and the rate of sediment accu-
pansive soil that is compacted dry of optimum water
mulation are important in determining the initial struc-
content can swell more than if compacted to the same
ture of clay deposits. At high pressures, both curves
py
dicates a low sensitivity at all consolidation pressures. A swell sensitivity has been observed in some clays
Results of strength tests showed that the actual sensi- wherein the swelling index for the remolded clay is
tivity ranged from 1.0 to 2.6. higher than that of the same clay undisturbed. The in-
Curve C is for Leda clay remolded at a liquidity creased swelling of the disturbed material can result
index of 1.82. The sensitivity increases from 1 to about both from the rupture of interparticle bonds that inhibit
8 with reconsolidation, indicating development of me- swelling in the undisturbed state and from differences
tastability after remolding and recompression. The sen- in fabric. Old, unweathered, overconsolidated clays
sitivity decreases at high pressures as convergence with may be particularly swell sensitive. Swell sensitivities
curve A is approached. All of the above findings are as high as 20 were measured in one case (Schmert-
consistent with the principles stated in Section 8.13. mann, 1969).
l
the pressure that must be applied to prevent movement
ria
of water either in or out of clay is determined as a
function of particle spacings expressed in terms of void
ratio or water content.
The concept of osmotic pressure is illustrated by
Fig. 10.18. The two sides of the cell in Fig. 10.18a are
ate
separated by a semipermeable membrane through
which solvent (water) may pass but solute (salt) can-
not. Because the salt concentration in solution is
greater on the left side of the membrane than on the
right side, the free energy and chemical potential of
the water on the left are less than on the right.1 Because
dM
solute cannot pass to the right to equalize concentra-
tions due to the presence of the membrane, solvent
passes into the chamber on the left.
The effect of this is twofold as shown by Fig.
10.18b. First, the solute concentration on the left is
reduced and that on the right side is increased, which
hte
membrane until the free energy of the water is the equilibrium, (b) final condition: equilibrium, and (c) osmotic
same on each side. pressure equilibrium.
It would be possible in a system such as that shown
by Fig. 10.18a to completely prevent flow through the
membrane by applying a sufficient pressure to the so-
py
lution in the left chamber, as shown by Fig. 10.18c. absolute temperature, ni is the concentration (particles
The pressure needed to exactly stop flow is termed the per unit volume), and ci is the molar concentration.
osmotic pressure , and it may be calculated, for dilute Thus, the osmotic pressure difference between two so-
solutions, by the van’t Hoff equation, which was intro- lutions separated by a semipermeable membrane is di-
Co
冘(n 冘(c
In a soil, there is no true semipermeable membrane
⫽ kT iA ⫺ niB) ⫽ RT iA ⫺ ciB) (10.2) separating regions of high- and low-salt concentration.
The effect of a restrictive membrane is created, how-
where k is the Boltzmann constant (gas constant per ever, by the influence of the negatively charged clay
molecule), R is the gas constant per mole, T is the surfaces on the adsorbed cations. Because of the at-
traction of adsorbed cations to particle surfaces, the
cations are not free to diffuse, and concentration dif-
1
Formal treatment of the concepts stated here and derivation of Eq. ferences responsible for osmotic pressures are devel-
(10.1) are given in standard texts on chemical thermodynamics. oped whenever double layers on adjacent particles
overlap. The situation is shown schematically in Fig. where ca is the midplane anion concentration, and c⫹ 0
10.19. The difference in osmotic pressure midway be- and c⫺0 are the equilibrium solution concentrations of
tween particles and in the equilibrium solution sur- cations and anions. At equilibrium in dilute solutions
rounding the clay is the interparticle repulsive pressure
or swelling pressure Ps. It can be expressed in terms
cc ca ⫽ c0⫹ c0⫺ ⫽ c02 (10.6)
of midplane potentials according to the following
equation (see Section 6.11):
because c⫹ ⫺
0 ⫽ c0 . Thus Eq. (10.5) becomes
Ps ⫽ p ⫽ 2n0kT(cosh u ⫺ 1) (10.3)
l
and u is the midplane potential function. c0 cc
ria
In terms of midplane cation and equilibrium solution
concentrations cc and c0 (Bolt, 1956), Eq. (10.2) be- Midplane concentrations can be determined using
comes the relationships in Chapter 6. Equation (10.7) assumes
Ps ⫽ ⫽ RT 冘(c ⫺ ci 0) (10.4)
parallel flat plates and may be written in terms of void
ate
ic ratio for saturated clay. The water content w, in terms
of weight of water per unit weight of soil solids, di-
For single cation and anion species of the same valence vided by the specific surface of soil solids As gives the
average thickness of water layer, which is half the par-
Ps ⫽ RT(cc ⫹ ca ⫺ c⫹ ⫺
0 ⫺ c0 ) (10.5) ticle spacing or d. Thus,
dM
hte
rig
py
Co
where Gs is the specific gravity of solids. Substituting where 兩 兩 is the swelling pressure or matric suction (see
Eq. (10.9) into Eq. (10.8) gives Section 7.12) measured in centimeters of water.
Since the sum of the applied constraint 兩 兩 in con-
l
e centration units and the external solution concentration
ria
d⫽ (10.10) must equal the midplane concentration, the pressure or
Gs w As
suction is given by
Bolt (1955, 1956) showed that the double-layer equa-
tions (see Chapter 6) can be combined with Eq. (10.10) 兩 兩 ⫽
冘c ⫺ 冘c
m 0
ate
⫺5
(10.14)
to give 4 ⫻ 10
⫻ 冕/2 dM d
(10.11)
冉 冊
e 2
冉 冘c 冊
(1 ⫺ (c0 /cc)2 sin2 )1/2 v()1/2 ⫽ ⫺
冪冘c
⫽0 1/2
Gs As
m
–41 %2 ⫹ m
in which v is the cation valence and distance x0 equals
approximately 0.1/ v nm for illite, 0.2/ v nm for kao- (10.15)
linite, and 0.4/ v nm for montmorillonite. The param-
eter is given by
where ⬇ 1.0 ⫻ 1015 cm/mmol at 20C and % is the
hte
冉 冊
solidation or swelling. For any value of log[Ps /(RTc0)]
e
冉冘 冊
the swelling pressure may be calculated. The void ratio v()1/2 ⫽
py
1/2
can be computed from the corresponding value of Gs As
cm
v(c0)1/2(x0 ⫹ e/Gs w As). For a given soil, Ps depends
冉 再 冋冘 冒冉 冘c 冊册 冎冊
completely on cc and c0 and those factors that cause cc
1/2 1/2
to be large relative to c0; for example, low c0, low
⫺cos⫺1 1/a 1⫺ –14 %2⫹
Co
cm m
valence of cation, and high dielectric constant, cause
high interparticle repulsions, high swelling pressures, ⫻
and large physicochemical resistance to compression. cm
It is apparent from the values in Table 10.6 that the (10.17)
dominating influence on swelling pressure at any given
void ratio is the specific surface area, which is deter- The value of a in Eq. (10.17) is given by
mined mainly by mineralogy and particle size.
The preceding relationships were developed for soils 2cm ⫺ (c⫹ ⫹⫹ ⫹⫹ ⫹ ⫹⫹ 2 1/2
m ⫹ cm ) ⫹ [4cmcm ⫹ (cm ⫹ cm ) ]
containing a single electrolyte, and they assume ideal a⫽
2cm
behavior in accord with the DLVO theory as developed
in Chapter 6. Approximate equations for mixed-cation (10.18)
v(c0)1/2 v(c0)1/2
(x0 ⫹ e/ Gs w As) log Ps /(RTc0) (x0 ⫹ e/ Gs w As) log Ps /(RTc0)
0.050 3.596 0.997 0.909
0.067 3.346 1.188 0.717
0.100 2.993 1.419 0.505
0.200 2.389 1.762 0.212
⫺0.046
l
0.300 2.032 2.076
ria
0.400 1.776 2.362 ⫺0.301
0.500 1.573 2.716 ⫺0.573
0.600 1.405 3.09 ⫺0.899
0.700 1.258 3.57 ⫺1.301
0.801 1.130 4.35 ⫺1.955
ate
0.902 1.012
a
v is the cation valence; is 8F /1000 DRT ⬇ 1015 cm/mmol for water
at normal T; c0 is the concentration in bulk solution (mmol/cm3); x0 ⫽ 4/ vT
⬇ 1/ v Å for illite, 2/ v Å for kaolinite, and 4/ v Å for montmorillonite; e is
the void ratio; Gs w is the density of solids, As is the specific surface area of
dM
clay; Ps is the swelling pressure; R is the gas constant; T is the absolute
temperature; F is the Faraday constant; and D is the dielectric constant.
Adapted from Bolt (1956).
l
ria
Figure 10.21 Compression curves of Na-montmorillonite
and Ca-montmorillonite, fraction ⬍2 m, in equilibrium with
10⫺3 M NaCl and CaCl2, respectively. The dashed lines rep-
resent the theoretical curves for As ⫽ 800 m2 / g (Bolt, 1956).
ate
Osmotic pressure theory was used successfully for
prediction of swelling pressure developed in opalinum
shale, a Jurassic clay rock (Madsen and Müller-
dM
Vonmoos, 1985, 1989). Swelling pressure was pre-
dicted using Eq. (10.2) and compared with the
measured values, with the results shown in Fig. 10.22.
Particle spacings were calculated from specific surface
area and water content.
Agreement between theory and experiment has not
Figure 10.22 Predicted and measured swelling pressures for
been good for clays containing particles larger than a Opalinum shale (Madsen and Müller-Vonmoos, 1989).
hte
Compression and swelling curves for three size frac- short- and long-range forces such as van der Waals
tions of sodium illite are shown in Fig. 10.23. The forces are neglected.
discrepancies between theory and experiment are fairly
large for the ⬍0.2-m fraction; nonetheless, the ex-
perimental curves are in the predicted relative positions Mixed-Cation Systems
py
(Fig. 10.23a). However, for samples containing coarser Most soils contain mixtures of sodium, potassium, cal-
particles (Figs. 10.23b and 10.23c), the curves are in cium, and magnesium in their adsorbed cation com-
reverse order to theoretical prediction. This is because plex. Therefore, modifications of the double-layer and
the compression was controlled by initial particle ori- osmotic pressure equations for homoionic clays are re-
Co
entations and physical interactions between the larger quired. The extent to which the resulting equations
particles rather than by osmotic repulsive pressures. may be suitable depends on the structural status of the
The concentration of CaCl2 or MgCl2 has essentially clay as well as on the particle size. Equations for
no influence on the swelling of a 2-m fraction of mixed-cation systems are derived on the assumption
illite, and the consolidation is influenced only by how that ions of all species are distributed uniformly over
the changes in concentration change the initial struc- the clay surfaces in proportion to the amounts present.
ture (Olson and Mitronovas, 1962). However, sodium and calcium ions may separate into
Factors in addition to clay particle size may also distinct regions. This is termed demixing (Glaeser and
contribute to failure of the theory in natural soils. The Mering, 1954; McNeal et al., 1966; McNeal, 1970;
DLVO theory that serves as the basis for determination Fink et al., 1971).
l
ria
ate
dM
Figure 10.23 Influence of NaCl concentration and particle size on compression and swelling
behavior of Fithian illite.
hte
Observed behavior was good for several cases ex- ical particle interference and fabric factors related to
amined using a demixed ion model (5 out of 6) for particle size. The behavior is consistent with the prin-
values of exchangeable sodium percentage (ESP) less ciple of chemical irreversibility of clay fabric (Bennett
than about 50 (McNeal, 1970). Based on X-ray deter- and Hurlbut, 1986), which is discussed in Section 8.2.
rig
minations of interplate spacings in montmorillonite Nonetheless, when the physical and chemical influ-
(Fink et al., 1971) it appears that for ences of cation type on fabric and effective specific
surface are taken into account, the behavior can be
1. ESP ⬎ 50 percent, there is random mixing of better understood, as illustrated, for example, by Di
Na⫹ and Ca2⫹ and unlimited swelling between all Maio (1996). For those cases in which fabric changes
py
plates on addition of water. and interparticle interactions are small, such as swell-
2. 10 percent ⬍ ESP ⬍ 50 percent, there is demix- ing from a precompressed state, or for clays with very
ing on interlayer exchange sites, with progres- high specific surface area (very small particles) such
sively more sets of plates collapsing to a 20-Å as bentonite, the theory gives a reasonable description
Co
l
ever, the specific surface areas of pure illite and non-
ria
t is the average thickness of water layers, and ki ⫽ / expanded smectite are only about 100 m2 /g, which
( w As), where As is the specific surface. corresponds to a water content of 50 percent. For a
Equation (10.19) shows, as would be expected, that pure kaolinite having a specific surface of 15 m2 /g, the
the lower the water content and, therefore, the smaller water content would be only 7.5 percent for a 5-nm-
the water layer thickness, the higher is the swelling thick adsorbed layer.
ate
pressure. Whereas this approach can explain the swell- It is evident, therefore, that the specific surface dom-
ing of pure clays accurately, the osmotic pressure the- inates the amount of water required to satisfy forces
ory cannot (Low, 1987, 1992). of hydration. Except for very heavily overconsolidated
On the other hand, the influences of surface charge clays and those soils that contain large amounts of ex-
density, cation valence, electrolyte concentration, and
dM pandable smectite, there is sufficient water present
dielectric constant, which have profound influences on even at low water contents to satisfy surface hydration
swelling and swelling pressure, as shown in the pre- forces, and swelling is small. On the other hand, when
vious section, are not directly accounted for by the the clay content is high and particle dissociation into
hydration theory unless appropriate adjustments can be unit layers is extensive, the effective specific surface
made for the influences of these factors on B, , and area is large and swelling can be significant. The ten-
ki. An explanation that is consistent with both the in- dency for smectite dissociation into unit layers can be
fluences of the double-layer/osmotic pressure theory evaluated through consideration of double-layer inter-
hte
and the water adsorption theory is as follows. actions, with those conditions that favor the develop-
Charge density and cation type influence the relative ment of high repulsive forces, as discussed in Chapter
proportions of fully expandable and partially expand- 6, leading to greater dissociation.
able layers in swelling clay. For example, calcium
montmorillonite does not swell to interplate distances
greater than about 0.9 nm where the particles stabilize
rig
tric constant, interlayer swelling is suppressed, and the minerals because only these minerals have sufficient
effective specific surface is greatly reduced relative to specific surface area so that there are unsatisfied water
that for the case where interlayer swelling occurs. The adsorption forces at low water contents. Details of
amount of water required to satisfy surface hydration structure and the presence of interlayer materials may
Co
Crystal Lattice Configuration Effects properties of expansive clays, for example, Rich
Greatest swelling is observed for charge deficiencies (1968). Some aspects of interlayering between the ba-
in silicate layer structures of about one per unit cell as sic sheets in the expansive clay minerals are:
indicated in Table 10.7. Evidently, for layer silicates 1. Optimum conditions for interlayer formation are:
with sufficient isomorphous substitution to give charge a. Supply of A13⫹ ions
deficiencies greater than 1.0 to 1.5 per unit cell, the b. Moderately acid pH (⬇5)
balancing cations are so strongly held and organized c. Low oxygen content
in the interlayer regions that interlayer swelling is pre- d. Frequent wetting and drying
vented. 2. Hydroxyaluminum is the principal interlayer ma-
Within the range of charge deficiencies where swell terial in acid soils, but Fe–OH layers may be
is observed, there is no consistent relationship between
l
present.
charge, as measured by the cation exchange capacity,
ria
3. Mg(OH)2 is probably the principal interlayer
and the amount of swell (Foster, 1953, 1955). This component in alkaline soils.
finding is more consistent with the surface hydration 4. Randomly distributed islands of interlayer mate-
model for clay swelling than with the osmotic pressure rial bind adjacent layers together. The degree of
theory. interlayering in soils is usually small (10 to 20
ate
An inverse correlation exists between free swell and percent), but this is enough to fix the basal spac-
the b dimension of the montmorillonite crystal lattice ing of montmorillonite and vermiculite at 14 Å.
(Davidtz and Low, 1970). Differences in b dimension, 5. The cation exchange capacity is reduced by in-
which may be caused by differences in isomorphous terlayer formation.
substitution, evidently cause changes in water hydra- 6. Swelling is reduced.
tion forces. Furthermore, as the water content in-
dM
creases, so also does the b dimension, as shown in Fig.
6.5. Swelling ceases when the b dimension reaches Salt Heave
0.9 nm. Some saline soils with high contents of salts can un-
dergo changes in volume associated with hydration–
Hydroxy Interlayering dehydration phenomena. One example is the swelling
of some soils containing large amounts of sodium sul-
The occurrence, formation, and properties of fate (Na2SO4) found in and around the Las Vegas area
hte
hydroxyl–cation interlayers (Fe–OH, Al–OH, Mg–H) of Nevada. When the temperature falls from above
have been studied regarding their effects on physical about 32C to below about 10C, the salt hydrates to
Na2SO4 10H2O with accompanying increase in vol-
ume. This salt heave has been responsible for damage
Table 10.7 Influence of Lattice Charge on to light structures and is described in more detail by
Blaser and Scherer (1969) and Blaser and Arulanandan
rig
Expansion
(1973).
Negative Charge
Mineral per Unit Cell Tendency to Expand Impact of Pyrite
Sulfur occurs in rock and soil as sulfide (S⫺ or S2⫺),
py
Margarite 4 None
Muscovite Only with drastic sulfate (SO42⫺), and organic sulfur. The sulfide min-
chemical treat- erals, of which pyrite is one of the most common and
ment, if at all easily oxidized (Burkart et al., 1999), are of greatest
concern. The amount of sulfide sulfur is a good indi-
Co
Biotite 2
Paragonite cator of the potential for oxidation reactions and
Hydrous mica ⬎1.2
weathering that can result in expansion. Sulfide-
and illite induced heave has occurred in materials containing as
Vermiculite 1.4–0.9 Expanding little as 0.1 percent sulfide sulfur (Belgeri and Siegel
Montmorillonite 1998). Products of pyrite oxidation include sulfate
Beidellite 1.0–0.6 Readily expanding minerals, insoluble iron oxides such as goethite
Nontronite (FeOOH) and hematite (Fe2O3), and sulfuric acid
Hectorite (H2SO4). Sulfuric acid can dissolve other sulfides,
Pyrophyllite 0 None heavy metals, carbonates, and the like that are present
in the oxidation zone, thus allowing the effects of ox-
From Brindley and MacEwen (1953). idation to increase as the process builds upon itself.
l
deleterious reactions, such as the formation of gypsum several series of laboratory culture experiments. Test
ria
and expansive sulfate minerals (e.g., ettringite). Gyp- results presented by Yamanaka et al. (2002), which
sum (CaSO4 2H2O) is considered to be the primary include electron photomicrographs of the bacteria,
cause of heave resulting from sulfate expansion. Vol- showed consistent variations of hydrogen sulfide con-
ume increases associated with several sulfidic chemical centration, pH, Fe3⫹ concentration, Fe2⫹ ⫹ Fe3⫹ con-
weathering reactions are given in Table 10.8. For com- centration, and SO42⫺ concentration over time periods
ate
parative purposes, these percentages are based on the up to 50 days for both the natural mudstone and the
assumption that the altered rock was initially com- mudstone after heat treatment to 121C. The heat treat-
posed of 100 percent of the original mineral. ment prevented or greatly slowed the bacterial activity,
Sulfide oxidation reactions are usually catalyzed by whereas very significant changes in concentrations and
microbial activity. Gypsum forms when sulfate ions
dM pH were measured for tests done at 28C. For example,
react with calcium in the presence of water, resulting the concentration of H2S increased from 0.3 to 2.2 mM
in very large volume increases. The products of pyrite in 20 days, the pH decreased from about 6.5 to 1.3
oxidation reactions are significantly less dense than the in 47 days, the concentration of Fe3⫹ increased from
initial sulfide product (pyrite); for example, the specific about 6 to 125 in 5 days, and the concentration of
gravity of pyrite is 4.8 to 5.1, whereas that of gypsum SO42⫺ increased from less than 1 to about 15 mM in
is only 2.3, and that of calcium is 2.6. Acidity pro- 25 days.
duced by pyrite oxidation can also result in significant Based on their results and observations, Yamanaka
hte
quantities of acid mine and rock drainage. et al. (2002) developed the following explanation for
the processes leading to the foundation heave. The
Bacterially Generated Heave—Case History ground temperature, which had been about 18C at
About 1000 wooden houses founded on mudstone sed- depth, increased to about 25C in the summer after
iments in Iwaki City, Fukushima Prefecture, Japan, excavation. Initial anaerobic, high water content con-
ditions and the stimulation of sulfate-reducing bacteria
rig
lowing admixture stabilization of these soils using fective stress is linear, and properties of the soil do not
Portland cement and lime has developed at several sites change during the consolidation process. Deformations
(Mitchell, 1986). Although test programs showed sup- in only one dimension, usually vertical, are considered
pression of swelling and substantial strength increase since determinations of settlements caused by loadings
at short times (days) as a result of the incorporation of from structures or fills are common applications of the
the stabilizer, subsequent heave of magnitude sufficient theory. In such a case, the relationship between void
to destroy pavements developed after of exposure to ratio and vertical stress is as shown in Fig. 10.24a for
water at some later time. The mechanism associated a normally consolidated clay layer, and that in Fig.
with this process appears to be as follows. 10.24b applies for an overconsolidated clay layer.2
When cement or lime is mixed with soil and water, As shown in any basic text on soil mechanics, the
there is a pH increase to about 12.4, some calcium amount of vertical settlement
H that a homogeneous
l
goes into solution and exchanges with sodium on the clay layer of thickness H will undergo if subjected to
ria
expansive clay. This ion exchange, along with light a vertical stress increase at the surface is given by
cementation by carbonate and gypsum, if present, sup-
presses the swelling tendency of the clay. The mixed
e
and compacted soil is nonexpansive and has higher
H ⫽ H (10.20)
1 ⫹ e0
strength than the untreated material. If sodium sulfate
ate
is present, then available lime is depleted according to in which e0 is the initial void ratio and
e is the de-
crease in void ratio due to the stress increase from
Ca(OH)2 ⫹ Na2SO4 → CaSO4 ⫹ 2NaOH v0
to v1
. For convenience, the change in void ratio
is often written in terms of compression index or co-
Silica (SiO2) and alumina (Al2O3) dissolve from the
dM efficient of compressibility and change in effective
clay in the high pH environment and/or they may be stress as defined in Fig. 10.1.
present in amorphous form initially. These compounds The rate at which consolidation under the stress in-
can then combine with calcium, carbonate, and sulfate creases from v0 to v1
is determined using Terzaghi’s
to form ettringite, Ca6[Si(OH)6]2(SO4)3 26H2O, and/ solution to the one-dimensional diffusion equation ap-
or thaumasite, Ca6[Si(OH)6]2(SO4)2(CO3)2 24H2O, plied to the transient state water flow from the consol-
which are very expansive materials (Mehta and Hu, idating clay layer. It is assumed in this theory that the
1978). In addition, in the case of lime-treated soil, if rate of volume decrease is controlled totally by hydro-
hte
the available lime is depleted, the pH will drop and the dynamic lag, that is, the time required for water to flow
further dissolution of SiO2 from the clay will stop. As out of the consolidating soil under the gradients gen-
silica is needed for formation of the cement (CSH) that erated by the applied pressures. The governing equa-
is the desired end product of the pozzolanic lime sta- tion is
bilization reaction, long-term strength gain is pre-
u 2u
rig
kh(1 ⫹ e)
10.9 CONSOLIDATION cv ⫽ (10.22)
av w
Co
l
ria
Figure 10.24 Idealized compression curves for clay layers: (a) normally consolidated and
ate
(b) overconsolidated.
Solutions for Eq. (10.22) for different boundary con- greater than 0.4 indicates high compressibility. Corre-
ditions are given in standard soil mechanics texts in
dM lations between compression index and compositional
terms of a dimensionless depth z/H (where H is the and state parameters have been proposed by a number
maximum distance to a drainage boundary) and a di- of investigators. Several such relationships for cohesive
mensionless time factor T ⫽ cvt/H 2 for different soils were summarized by Djoenaidi (1985) and quoted
boundary conditions. The solution for u ⫽ ƒ(z/H, T) by Kulhawy and Mayne (1990), and these relationships
for a layer of thickness 2H that is initially at equilib- are shown in Fig. 10.26. A simple correlation between
rium and subjected to a rapidly applied uniform sur- the compression ratio, defined as Cc /(1 ⫹ e0), where
face loading is shown in Fig. 10.25a. The average e0 is the initial void ratio, and the natural water content
hte
degree of consolidation U over the full depth of the is shown in Fig. 10.27.
clay layer as a function of T for this case is shown in The large increase in compressibility that occurs
Fig. 10.25b. when sensitive clay is loaded beyond its maximum
prior effective consolidation pressure is shown in Fig.
Ranges of Compressibility and Consolidation 8.44. Values of compression index for the steepest part
Parameters of the compression curve as a function of in situ void
rig
The curves in Fig. 10.2, as well as the fact that the ratio and sensitivity are shown in Fig. 10.28. The pro-
void ratio of a soil cannot decrease without limit under found influence of structure metastability as repre-
increasing pressure, mean that the assumption of a lin- sented by high sensitivity is clearly evident.
ear relationship between void ratio and log of effective Usual ranges of coefficient of consolidation for fine-
py
consolidation pressure that defines the compression in- grained soils are given in Fig. 4.19. Owing to the direct
dex Cc is simply a useful engineering approximation dependence of the coefficient of consolidation cv on
that applies over a range of stresses and void ratios of hydraulic conductivity and its inverse proportionality
practical interest.3 Values for compression index less to coefficient of compressibility, reliable determination
of a representative value in any case is difficult. Both
Co
l
ria
ate
dM
hte
rig
are based on the above simple theory, are poor. Among 4. Changes in pore pressure dissipation rates during
the types of deviations between the observed and pre- and following construction
dicted settlement and pore pressure responses are the 5. Apparent lack of strength gain with consolidation
Co
following (Crooks et al., 1984; Becker et al., 1984; following load application
Tse, 1985; Mitchell, 1986; Duncan, 1993): There are two types of reasons for deviations from
the simple theory. In the first category are those that
1. Differences in predicted and observed initial pore relate to soil behavior and the fact that in general the
pressure development upon load applications simple relationships between effective stress shown in
2. Continued pore pressure buildup after completion Figs. 10.1 and 10.24 are neither unique nor time in-
of loading dependent. In the second category are those that relate
3. Differences between field consolidation rates and to the constitutive models and their application and the
those predicted based on the results of laboratory fact that the simplifying assumptions that may be re-
tests quired are not representative of the real conditions.
l
ria
Figure 10.26 Representative values of compression index Cc for cohesive soils (Djoenaidi,
1985).
ate
Soil Behavior Factors Characteristics of the real
behavior of fine-grained soils that are important in de-
termining the amount and rate of consolidation in-
dM clude:
1. Fabric and Structure Resistance to compression
is determined by both effective stress and struc-
ture. Structural influences that must be consid-
ered relate to the initial state, the effects of
sample disturbance, structural breakdown asso-
ciated with consolidation under pressures greater
hte
Figure 10.27 Compression ratio as a function of natural wa- than the maximum past consolidation pressure,
ter content (from Lambe and Whitman, 1969). Reprinted and the effects of anisotropic loading.
with permission from John Wiley & Sons. 2. Time and Rate of Loading The relationship be-
tween void ratio and effective consolidation pres-
sure is not unique for a fine-grained soil but is
influenced by rate of loading and time under a
rig
e ⫽ e( , t) (10.23)
py
de
dt
⫽ 冉 冊
e
t
d
dt 冉冊
⫹
e
t
(10.24)
Co
a stress increase may be controlled by either how information about them and about how to account for
rapidly the water can escape under a hydraulic them can be found in Gibson et al. (1981), Tse (1985),
gradient or by how fast the structure of the soil Mesri and Castro (1987), Leroueil et al. (1990), Scott
can deform or creep under a given magnitude of (1989), Duncan (1993), and elsewhere. Generalization
effective stress. Component (1) compression is of Terzaghi’s one-dimensional consolidation theory to
commonly referred to as primary consolidation. three dimensions was made by Biot (1941). At present,
Component (2) compression is commonly re- there are finite element and finite difference codes that
ferred to as secondary compression. In addition, solve Biot’s consolidation equation incorporating non-
aging phenomena during time under sustained linear stress–stress relationships as well as anisotropic
stress generate additional resistance to further hydraulic conductivity. The hydraulic conductivity can
compression. also be a function of void ratio or effective stress. Fur-
l
3. Temperature Owing to differential thermal ex- ther details can be found in Lewis and Schrefler (1997)
ria
pansions of soil solids and the pore fluid and and Coussy (2004). Soil behavior factors are consid-
changes in interparticle bond strength and resis- ered further in the remainder of this section.
tance to sliding that can result from changes in
temperature, temperature-induced changes in ef-
Effects of Sample Disturbance
fective stress and volume are possible. These ef-
ate
fects are considered further in Section 10.12. The effects of sample disturbance on the compression
curve of sensitive or structured clay are shown in Fig.
Modeling Factors The commonly used constitutive 8.44 and include:
models for soil compression and consolidation may not
give suitable representations of actual behavior for the 1. A lower void ratio under any effective stress.
following reasons:
dM
1. The relationship between void ratio and effective
2. Higher values of recompression index and lower
values of the compression index for a disturbed
consolidation pressure is not linear, as is assumed clay than for the undisturbed soil.
for the Terzaghi consolidation theory. In fact, the 3. Less clearly defined stress history; determination
use of compression index and swelling index to of the maximum past consolidation pressure may
characterize soil compression and swelling rec- be difficult and uncertain.
ognize the nonlinear nature of the void ratio–
hte
l
is susceptible to a viscous or creep deformation under
of clean sands is shown in Fig. 10.30, where it may
ria
the action of sustained stress as the fabric elements
adjust slowly to more stable arrangements. The rate of be seen that Ce /Cc falls in the range of 0.015 to 0.03.
secondary compression is controlled by the rate at A general relationship between void ratio, effective
which the structure can deform, as opposed to the rate consolidation pressure, and time is shown in Fig.
of primary consolidation, which is controlled by 10.31, with slopes Ce and Cc indicated. When the
ate
Darcy’s law, which determines how rapidly water can curves corresponding to different times after the end
escape from the pores under a hydraulic gradient.4 of primary consolidation are projected onto the void
The mechanism of secondary compression involves ratio–log effective stress plane, Fig. 10.5 is obtained
sliding at interparticle contacts, expulsion of water for the assumption of linearity between void ratio and
from microfabric elements, and rearrangement of ad- log . Algebraic manipulation of the secondary com-
sorbed water molecules and cations into different po-
dM pression equation and the primary compression equa-
sitions. The observed behavior is consistent with that tion shows that the preconsolidation pressure is rate
of a thermally activated rate process, which involves dependent (Soga and Mitchell, 1996), consistent with
mechanisms that are discussed in more detail in Sec- the data presented in Fig. 10.7.
tion 12.4. Both laboratory tests and field measurements, as
The relationship between void ratio and log of time well as theoretical arguments, have been made to es-
during secondary compression is linear for most soils tablish whether or not (1) the relationship between the
end-of-primary consolidation void ratio and effective
hte
l
Silty clay C 0.032
ria
Near-shore clays and silts 0.055–0.075
Mexico City clay 0.03–0.035
Hudson River silt 0.03–0.06
Organic clays Norfolk organic silt 0.05
and silts Calcareous organic silt 0.035–0.06
ate
Postglacial organic clay 0.05–0.07
Organic clays and silts 0.04–0.06
New Haven organic clay silt 0.04–0.075
Peats Amorphous and fibrous peat 0.035–0.083
Canadian muskeg 0.09–0.10
Peat 0.075–0.085
dM Peat 0.05–0.08
Fibrous peat 0.06–0.085
From Mesri and Godlewski (1977).
hte
rig
py
Co
l
times, the total settlement is taken as the consolidation 1987). The main limitation of in situ measurements is
ria
settlement increased by an amount of secondary com- that they invariably cause disturbance and allow lateral
pression derived from Eq. (10.25). This is undoubtedly deformations of the ground that change the stress being
an oversimplification of real behavior, as from the per- measured. The general ranges of in situ lateral stress
spective of the soil, there should be no difference be- for different soil types are summarized, and factors in-
tween the two types of compression. It compresses just fluencing lateral stress are reviewed in this section.
ate
sufficiently to withstand the applied stresses at any
time, and the rate at which it occurs in any element Development of Horizontal Stress
depends on whether or not the rate of water flow from
The relationship between the horizontal effective stress
the element at that time is controlled by a preexisting
and the vertical effective stress depends on the lateral
hydrostatic excess pressure gradient (primary consoli-
dM deformation that accompanies changes in vertical
dation) or by the time-dependent generation of small
stress. If the vertical stress and strain increase without
pore pressures owing to structural readjustment (sec-
any deformation in the horizontal directions (i.e., one-
ondary compression).
dimensional compression, as would be the case for an
On this basis, it would seem most likely that within
accumulating sediment), the soil is said to be in an at-
a clay layer both primary consolidation and secondary
rest state, and the horizontal stress associated with this
compression may be occurring concurrently in differ-
condition is termed the at-rest pressure.
ent elements. The major difficulty has been in the for-
The ratio between the horizontal and vertical effec-
hte
10.11 IN SITU HORIZONTAL STRESS (K0) measured in triaxial compression tests. Although cor-
relations have been published that suggest unique re-
Terzaghi’s consolidation theory considers compression lationships between K0 and liquid limit or plasticity
only in one dimension. The soil model relates the ver- index, a comprehensive set of data for 135 clay soils
Co
tical strain to the change in vertical stress, and this indicates little correlation, as shown in Fig. 10.32. This
defines the volume change under zero horizontal dis- is not surprising since the Atterberg limits depend only
placement conditions. There is no need to consider the on composition, and K0 is a state parameter that is
change in horizontal stress to calculate the deforma- dependent on composition, structure, and stress history.
tion, even though the actual horizontal stress changes When the vertical stress on a normally consolidated
during loading and unloading. However, once soil de- soil is reduced, the horizontal stress does not decrease
formation departs from the one-dimensional condition, in the same proportion as the vertical stress. Thus, the
it is necessary to consider the state and changes of the value of at-rest earth pressure coefficient for an over-
stresses in the other directions and the associated vol- consolidated soil (K0)oc is greater than that for the nor-
ume change behavior. mally consolidated soil (K0)nc, and it varies with the
l
ria
ate
Figure 10.32 Lack of correlation between coefficient of
earth pressure at rest and plasticity index for normally con-
solidated soils (from Kulhawy and Mayne, 1990). Reprinted
with permission from EPRI.
dM
amount of overconsolidation, as shown schematically
in Fig. 10.33, and in Fig. 10.34 for 48 clays. The data
in Fig. 10.34 can be approximated by the equation Figure 10.34 Dependence of (K0)oc on overconsolidation ra-
tio (from Kulhawy and Mayne, 1990). Reprinted with per-
mission from EPRI.
K0 ⫽ (1 ⫺ sin )(OCR)sin (10.27)
relations for estimation of K0. toward the K0 line coincides with the region of largest
The complicated stress paths associated with one- compression index (steepest slope on the volumetric
dimensional compression of four clays are illustrated strain versus effective mean stress diagrams), implying
in Fig 10.35. In the upper plot for each clay the de- structural degradation.
viator stress is shown as a function of the mean ef-
fective stress during one-dimensional compression.
rig
(σa – r)/2(kPa)
(σa – r)/2(kPa)
ure d
50 500 fai
l ture g)
ak s truc therin
Pe d e a
K o( by we
0 0
0 50 100 150 200 0 500 1000 1500 2000
(σa + r)/2(kPa) (σa + r)/2(kPa)
0 0
e0 = 1.97 e0 = 0.69
4
εv(%)
εv(%)
10
l
ria
8
20
12
(a) Sensitive Canadian clay (b) Unweathered Keuper marl
ate
200 50
(σa – r)/2(kPa)
(σa – r)/2(kPa)
ilure
k fa
Pea
100 25
d)
ucture
d estr
K o(
dM 0 0
0 100 100 100 100 0 25 50 75 100
(σa + r)/2(kPa) (σa + r)/2(kPa)
0 0
e0 = 1.04 e0 = 0.69
5 5
εv(%)
εv (%)
10 10
hte
15 15
(c) Artificially bonded soil (d) Chalk
Figure 10.35 Variation in lateral stress with mean stress during one-dimensional consoli-
dation of four clays (from Leroueil and Vaughan, 1990).
rig
ing a friction angle and a cohesion c, the limiting Under one-dimensional conditions, compression is
minimum and maximum values of the earth pressure usually plotted on the e–log v plane, as shown in Fig.
py
冉 冊 冉 冊
plotted on the e–ln p plane (or v –ln p plane), where
2c p is the mean effective pressure and v is the specific
Ka ⫽ tan2 45 ⫺ ⫺ tan 45 ⫺ (10.28)
v
Co
冉
Kp ⫽ tan2 45 ⫹
2
⫹
2c
v 冊
tan 45 ⫹
2 冉 冊 (10.29)
isotropically, the slope of the normal compression line
is defined as ⫽ ⫺de/d ln p(⫽ ⫺dv /d ln p) (Scho-
field and Wroth, 1968).6
Figure 10.37 shows the change in void ratio with
These limiting values are for isotropic soil and a hor- mean effective stress (p) for reconstituted kaolin clay
izontal ground surface. Standard soil mechanics texts specimens consolidated isotropically at constant stress
should be consulted for further details on limiting earth
pressure coefficients under sloping ground and the in- 6
The swelling (or recompression) line is often called the ! line on
fluences of changes in applied loads on in situ lateral e–ln p plane and the slope is defined as the recompression index
stress. !(⫽ ⫺de / d ln p)
Stress Paths
1.9 q
1
2
1.7
3 p⬘
Void Ratio
1.5
l
ria
70 100 200 300 500 1000
Mean Pressure p⬘ (kPa)
ate
umetric compression behavior of reconstituted kaolin clay.
Figure 10.36 Variation of lateral earth pressure coefficient
with deformation of a soil element.
increase or decrease in mean effective pressure p and and its variability, thus providing new insights into ge-
(ii) the other due to dilation or contraction by shearing ologic and soil formational history, as well as quanti-
of the soil by the increase in q. Further discussion of tative values for use in the analysis and prediction of
deformation behavior under combined volumetric and behavior.
deviatoric stress loading conditions is given in Chapter
py
11.
Time Dependence of Lateral Earth Pressure at Rest
Anisotropy It is usually assumed in conventional geotechnical
Unless the horizontal earth pressure coefficient is equal analyses that the coefficient of lateral earth pressure at-
Co
to 1.0, which is not the usual case, the stress condition rest K0 is a time-invariant constant. Whether or not this
is indeed the case is not known with certainty, and
there is no clear consensus on how K0 should be ex-
7
In one dimensional consolidation condition, p ⫽ (1 ⫹ 2K0) v. The pected to vary with time (Schmertmann, 1983). How-
relationship between Cc (⫽ ⫺de / d log v) and (⫽ ⫺de / d ln p) ever, if a soil is assumed to remain under a constant
is Cc ⫽ ln 10: Cc ⫽ ⫺de / d log v ⫽ ⫺ln 10[de / d(ln v)] ⫽ ⫺ln effective stress state following consolidation and there
10de / {d ln p ⫺ d[ln(1 ⫹ 2K0)]} ⫽ ⫺ln 10de / d ln p) (K0 is constant
in normally consolidated state, hence d[ln(1 ⫹ 2K0)] ⫽ 0). On the are no changes in the compositional or environmental
other hand, it is not possible to relate Cs obtained from the one- conditions, then slow changes in lateral pressure
dimensional consolidation test to ! obtained from the isotropic un- should occur in any material that is susceptible to creep
loading test. This is because the K0 value changes as the specimen
is unloaded and therefore the d[ln(1 ⫹ 2K0)] term in the above equa- and stress relaxation. Creep and stress relaxation are
tion does not become zero. analyzed in Section 12.7.
l
sure at rest in normally consolidated soil should
ria
increase toward 1.0 and that in heavily overconsoli-
dated soil should decrease toward 1.0.
Values of K0 as a function of time, as determined in
triaxial cells by Lacerda (1976), for undisturbed sam-
ples of soft San Francisco Bay mud, are shown in Fig.
ate
10.38. Also shown is a theoretical relationship between
K0 and time that was developed using the general Figure 10.39 Volume of pore water drained from saturated
stress–strain–time equations developed in Section illite under an isotropic effective stress of 200 kPa as a func-
12.9. Thus, both theory and experiment support the tion of temperature change.
above reasoning that K0 should increase with time
dM
when K0 is less than 1.0.
Fig. 10.40. Temperature effects such as these must be
considered relative to their influences on deformation
10.12 TEMPERATURE–VOLUME and stability both in the laboratory and the field.
RELATIONSHIPS
Theoretical Analysis
Temperature changes generate volume and/or effective
hte
stress changes in saturated soils. For example, the per- Drained Conditions Increase in temperature causes
centage of the original pore water volume that is thermal expansion of mineral solids and pore water. In
drained from a saturated specimen of illite subjected addition, there can be changes in soil structure. For a
to a temperature increase from 18.9 to 60C followed temperature change
T, the volume change of the pore
by cooling to 18.9C while maintaining an isotropic water is
effective stress of 200 kPa is shown in Fig. 10.39. The
rig
variation in effective stress 3 under the same temper- (
Vw)
T ⫽ wVw
T (10.30)
ature changes but with drainage prevented is shown in
where w is the thermal expansion coefficient of soil
water, and Vw is the pore water volume. The change in
volume of mineral solids is
py
Figure 10.38 K0 as a function of time for San Francisco Bay (
VDR)
T ⫽ (
Vw)
T ⫹ (
Vs)
T ⫺ (
Vm)
T (10.32)
mud. The theoretical curve was developed by Kavazanjian
and Mitchell (1984) using the general stress–strain–time Eq. in which (
Vm)
T is the change in total volume due to
(12.43) adapted for zero lateral strain.
T, with volume increases considered positive.
l
ria
ate
Figure 10.40 Effect of temperature changes on the effective stress in saturated illite under
constant confining pressure.
dM
In a soil mass with all grains in contact, and assum-
ing the same coefficient of thermal expansion for all
ity of mineral solids under hydrostatic pressure, and
the compressibility of mineral solids under concen-
soil minerals, the soil grains and the soil mass undergo trated loadings, respectively, then
the same volumetric strain s
T. In addition, the
change in temperature induces a change in interparticle (
Vw)
P ⫽ mwVw
u (10.36)
forces, cohesion, and/or frictional resistance that ne-
cessitates some particle reorientations to permit the soil (
Vs)
P ⫽ msVs
u ⫹ msVs
(10.37)
hte
of the volume changes of the soil mass due to both wVw
T ⫹ s Vs
T ⫺ (
Vm)
T ⫽ Mv Vm
temperature and pressure changes; that is
⫺ mwVw
u ⫺ Vs(ms
u ⫹ ms
) (10.39)
(
Vw)
T ⫹ (
Vs)
T ⫹ (
Vw)
P ⫹ (
Vs)
P
For constant total stress during a temperature change
⫽ (
Vm)
T ⫹ (
Vm)
P (10.35)
l
ture changes in the order indicated were carried out on
ria
a sample of saturated, remolded illite after initial con-
The left side of Eq. (10.42) is equal to (
VDR)
T , solidation to an effective stress of 200 kPa. Water
and the right side is an equivalent volume change drains from the sample during increase in temperature
caused entirely by a change in pore pressure. Because and is absorbed during temperature decrease. The
shape of the curves is similar to normal consolidation
ate
Vm ⫽ Vw ⫹ Vs (10.43) curves for volume changes caused by changes in ap-
plied stresses.
Eq. (10.42) may be written, after substitution for When the temperature is increased, two effects oc-
(
Vm)
T by Eq. (10.33), cur. If the increase is rapid, a significant positive pore
pressure develops due to greater volumetric expansion
w Vw
T ⫺ sVw
T ⫺ (
VST)
T ⫽ ⫺mv Vm
u
⫺mwVw
u
dM (10.44)
of the pore water than of the mineral solids. The lower
the hydraulic conductivity of the soil, the longer the
time required for this pore pressure to dissipate. Dis-
sipation of this pressure accounts for the parts of the
Rearrangement of Eq. (10.44) gives the pore pressure
curves in Fig. 10.41 that resemble primary consolida-
change accompanying a temperature change:
tion.
The second effect results because increase in tem-
n
T(s ⫺ w) ⫹ (
VST )
T /Vm
hte
ter, the volumetric strain due to physicochemical ef- effective stress. On subsequent temperature increases,
fects, and the compressibility of the soil structure. For the secondary effect is negligible because the structure
most soils (but not rocks) mv » mw , so has already been strengthened in prior cycles.
The final height changes and volumes of water
n(s ⫺ w)
T ⫹ ST
T drained associated with each temperature change
u ⫽ (10.47)
mv shown in Fig. 10.41 are plotted as a function of tem-
perature in Fig. 10.42, and clay structure volume
Consistency in algebraic signs is required for the changes are shown in Fig. 10.43. The forms of these
application of the above equations. Both s and w are plots are similar to conventional compression curves
positive and indicate volume increase with increasing involving virgin compression, unloading, and reload-
l
ria
ate
dM
hte
rig
py
Figure 10.41 Volume of water drained from a saturated clay as a function of time as a
Co
ing. An irrecoverable volume reduction after each ST , defined previously by Eq. (10.46). For the cases
temperature cycle is noted. Again, the effect of shown, ST has a value of about ⫺0.5 ⫻ 10⫺4 C⫺1.
temperature increase is analogous to a pressure in- The effect of temperature on clay compression de-
crease. The slope of the curves in Fig. 10.43 is the pends on the pressure range (Campanella and Mitchell,
coefficient of thermal expansion for the soil structure 1968; Plum and Esrig, 1969). Weaker structure at low
l
ria
ate
dM
hte
Figure 10.42 Effect of temperature variations on the height and volume change of saturated
illite.
rig
stresses caused by increased temperature causes con- (Leroueil and Marques, 1996). The data show that
solidation to a lower void ratio in order to carry the there is approximately 1 percent decrease in precon-
stress. The weakening effect of higher temperature is solidation pressure per one 1C temperature increase
py
compensated by the strengthening effect of lower void between 5 and 40C and somewhat less at higher tem-
ratio. As shown in Fig. 10.44, the compression index peratures (Leroueil and Hight, 2002).
Cc is found to be approximately independent of tem- Stress history or overconsolidation ratio has a major
perature. On the other hand, the isothermal swelling influence on the volume change caused by increase in
Co
index ! (⫽ ⫺de/d ln p) of reconstituted samples of temperature (Hueckel and Baldi, 1990). For normally
an illitic clay measured under isotropic confining stress consolidated to moderately overconsolidated clay, ir-
conditions is found to be temperature dependent as recoverable volume reduction was observed by struc-
shown in Fig. 10.45. ture degradation and the shear strength increased.
The preconsolidation pressure of a natural soft clay Volume expansion was observed in heavily overcon-
depends on temperature as illustrated in Fig. 10.7. Fig- solidated clay, and the expansion rate increased with
ure 10.46 shows the normalized preconsolidation pres- OCR.
sure (⫽ preconsolidation pressure at temperature T/ The effect of heating followed by cooling at two
preconsolidation pressure at 20C) with temperature stages in a consolidation test is shown in Fig. 10.47.
0.08
0.06
κT
0.04
䉭
䉭
0.02 䉭
䉭
0.00
l
10 20 40 60 80 100 200
ria
Figure 10.45 Effect of temperature on swelling index of is-
otropically consolidated illitic clay specimens. The clay con-
tained small amounts of Kaolin, chlorite and quartz and had
a liquid limit of 30 percent (after Graham et al., 2001).
ate
dM
hte
l
The preceding analyses indicate that the overall vol-
ria
ume changes that result from changes in temperature
may not be large. However, the structural weakening
and pore pressure changes that occur may be signifi-
cant in terms of their influences on shear deformation
and strength.
ate
Figure 10.47 Effect of heating and cooling on void ratio
versus pressure relationship of illite (Plum and Esrig, 1969).
defined as the change in pore pressure per unit change mechanics. The mechanical aspects of compression
in temperature per unit effective stress, or alternatively, and swelling are far better understood and quantified
the change in unit effective stress per unit change in than are those generated by physicochemical, geo-
temperature, that is, chemical, and microbiological factors, although inter-
est and research on the latter is intensifying.
py
F
(
u/
T)
T
u
Soil Type (kN/m2) (C) (kN/m2) (C⫺1)
Illite (grundite) 200 21.1–43.4 ⫹58 0.013
San Francisco 150 21.1–43.4 ⫹50 0.015
Bay mud
l
Weald Claya 710 25.0–29.0 ⫹51 0.018
ria
Kaolinite 200 21.1–43.4 ⫹78 0.017
Vicksburg 100 20.0–36.0 ⫹28 0.017
buckshot 650 20.0–36.0 ⫹190 0.018
clayb
Saturated 250 5.3–15.0 ⫹190 0.079
ate
sandstone 580 5.3–15.0 ⫹520 0.092
(porous stone)
a
From Henkel and Sowa (1963).
b
From Ladd (1961) Fig. VIII-6.
dM
U.S.) to pavements, structures, and utilities each year, Depth Range Unit Weight
is attributable to both double layer repulsions and wa- (m) Soil Type (kN/m3)
ter adsorption in soils that contain significant amounts
0–5 Surcharge fill 19.0
of high plasticity clay minerals. Other causes of soil
5–10 Rubble fill 17.0
and rock expansion have been identified as well, such
10–18 Clean sand 18.0
as pyrite related mineral transformations and sulfate
hte
characteristic controlling the consolidation and ground surface before placement of the sur-
swelling behavior of a soil? Why? charge fill. Assume that each layer is normally
consolidated.
2. If two samples of the same sand have the same b. Show profiles of vertical total, effective, and
relative density and are confined under the same water pressure as a function of depth immedi-
py
effective stress, can they have different volume ately after placement of the surcharge fill.
change properties? Why? Indicate if the clay layer is normally consoli-
3. In what soil types and under what conditions do dated, overconsolidated, or underconsolidated
physical particle interactions dominate in deter- at this time.
Co
mining the compression and swelling behavior? In c. Show profiles of vertical total, effective, and
what soil types and under what conditions do water pressure as a function of depth at a long
physicochemical factors dominate? time after the placement of the surcharge fill.
Are the sand and clay layers normally consol-
4. Provide an explanation for the differences in idated, underconsolidated, or overconsolidated?
amount of swelling associated with expansion fol-
d. Show profiles of vertical total, effective, and
lowing the different stress paths shown in Fig.
water pressure as a function of depth immedi-
10.10.
ately after removal of the surcharge fill. Are the
5. Consider the following soil profile beneath a level sand and clay layers normally consolidated, un-
ground surface: derconsolidated, or overconsolidated?
l
than the values you calculated in the preceding
ria
ground surface. In the other, the consolidation is
problem, state why?
to be effected by lowering the water table to the
bottom of the clay layer and evaporation of water 12. A normally consolidated, saturated marine clay is
from the ground surface, which will cause shrink- sampled without structural disturbance from be-
age of the clay. The ground water table is initially neath the seafloor and sealed to prevent water
ate
at the top of the clay stratum. Show profiles of movement in or out. The temperature of the clay
effective stress and water pressure versus depth for in situ is 5C. The effective stress at the time of
each stratum corresponding to the condition where sampling is 200 kPa and the void ratio of the clay
the vertical effective stress is the same in each at is 0.90. The sealed sample is taken immediately to
middepth. Will the clay structure be the same in the shipboard laboratory where the original in situ
each stratum at this depth at this time? Why?
dM confining stress is immediately reapplied.
7. Describe and contrast the compression, consoli- a. What will be the subsequent effective stress in
dation, and swelling potential properties of the fol- the laboratory at a temperature of 20C? The
lowing soil types. Assume their initial states (wa- clay has a compression index of 0.5 and a
ter content, overburden pressure, environmental swelling index of 0.05. Other properties are as
chemistry) to be representative of the indicated follows:
soil type as ordinarily encountered in nature. • Compressibility of water ⫽ ⫺4.83 ⫻ 10⫺5
hte
8. Prepare a schematic diagram of liquidity index c. If the same confining stress is maintained but
versus log effective consolidation pressure. Show drainage of the sample is then allowed, how
the positions of normally consolidated and heavily much water, expressed as a percentage of the
overconsolidated samples of a given clay on this original sample volume, will move in or out of
Co
14. What is a collapsing soil? What conditions can (consolidating) a highly plastic clay slurry that is
initiate collapse? What factors determine the mag- initially at a liquidity index considerably greater
nitude and rate of collapse? Is the process com- than 1.0. Explain how each of the methods that
patible with the principle of effective stress? Why? you have identified works.
15. Volume and temperature stability over long peri- 17. Comment on the mechanisms of primary consoli-
ods of time (thousands of years) is a very im- dation and secondary compression in terms of the
rate-controlling factors, influences of and effects
portant consideration in the utilization of earth
on soil structure, whether they occur sequentially
materials as containment barriers for various types or concurrently, and the suitability of our usual
of chemical and radioactive waste. What mineral procedures for quantifying them for geoengineer-
types, gradations, and placement conditions would ing analysis.
l
you specify for this application? Why?
ria
18. Suggest possible methods for preventing or reduc-
16. Suggest possible methods other than direct loading ing swelling on the exposure of expansive soil to
using surcharge fills for reducing the water content water and explain the mechanisms involved.
ate
dM
hte
rig
py
Co
l
Behavior
ria
ate
dM
11.1 INTRODUCTION In reality, the shearing resistance of a soil depends
on many factors, and a complete equation might be of
All aspects of soil stability—bearing capacity, slope the form
stability, the supporting capacity of deep foundations,
and penetration resistance, to name a few—depend on Shearing resistance ⫽ F(e, c, , , C, H, T, , ˙ , S)
hte
of soil behavior. The Mohr–Coulomb equation is by eters in these equations may not be independent, and
far the most widely used for strength. It states that the functional forms of all of them are not known.
Consequently, the shear resistance values (including c
and ) are determined using specified test type (i.e.,
direct shear, triaxial compression, simple shear), drain-
py
where ff is shear stress at failure on the failure plane, drained, undrained, peak strength, and residual
c is a cohesion intercept, ff is the normal stress on the strength. The shear resistance values applicable in
failure plane, and is a friction angle. Equation (11.1) practice depend on factors such as whether or not the
applies for ff defined as a total stress, and c and are problem is one of loading or unloading, whether or not
referred to as total stress parameters. Equation (11.2) short-term or long-term stability is of interest, and
applies for ff defined as an effective stress, and c and stress orientations.
are effective stress parameters. As the shear resis- Emphasis in this chapter is on the fundamental fac-
tance of soil originates mainly from actions at inter- tors controlling the strength and stress–deformation
particle contacts, the second equation is the more behavior of soils. Following a review of the general
fundamental. characteristics of strength and deformation, some re-
369
l
regions is summarized. Time-dependent deformations change tendencies of the soil. For such materials
ria
and aging effects are discussed separately in Chapter the stress–strain curve from a shearing test is
12. The details of strength determination by means of typically of the form shown in Fig. 11.1a. The
laboratory and in situ tests and the detailed constitutive maximum or peak strength of a soil (point b)
modeling of soil deformation and strength for use in may be greater than the critical state strength,
numerical analyses are outside the scope of this book.
ate
in which the soil deforms under sustained load-
ing at constant volume (point c). For some soils,
the particles align along a localized failure plane
11.2 GENERAL CHARACTERISTICS OF after large shear strain or shear displacement,
STRENGTH AND DEFORMATION and the strength decreases even further to the
residual strength (point d). The corresponding
Strength
dM
1. In the absence of chemical cementation between
three failure envelopes can be defined as shown
in Fig. 11.1b, with peak, critical, and residual
grains, the strength (stress state at failure or the friction angles (or states) as indicated.
ultimate stress state) of sand and clay is ap- 3. Peak failure envelopes are usually curved in the
proximated by a linear relationship with stress: manner shown in Fig. 4.16 and schematically in
Fig. 11.1b. This behavior is caused by dilatancy
ff ⫽ ff tan (11.4) suppression and grain crushing at higher
hte
Shear
Secant Peak
Stress τ Strength Envelope
or Stress
Ratio τ/σ Peak b Shear Tangent Peak Critical state
Stress τ Strength Envelope Strength Envelope
φpeak
py
d
a d
Strain a a Normal effective stress σ
Dense or Loose or Normally
Overconsolidated Consolidated
(a) (b)
Figure 11.1 Peak, critical, and residual strength and associated friction angle: (a) a typical
stress–strain curve and (b) stress states.
l
ria
ate
dM
hte
rig
py
Co
Figure 11.2 Variation of residual strength with stress level (after Bishop et al., 1971): (a)
Brown London clay and (b) Weald clay.
of internal rolling and sliding friction between resistance depends only on composition and ef-
grains and in part of interlocking of particles fective stress. The basic concept of the critical
(Taylor, 1948). The interlocking necessitates state is that under sustained uniform shearing at
either volume expansion (dilatancy) or grain failure, there exists a unique combination of
fracture and/or crushing if there is to be void ratio e, mean pressure p, and deviator
deformation. For loose sand, the peak friction stress q.1 The critical states of reconstituted
angle (point b in Fig. 11.1b) normally coincides Weald clay and Toyoura sand are shown in Fig.
with the critical-state friction angle (point c), 11.4. The critical state line on the p –q plane is
and there is no peak in the stress–strain curve. linear,2 whereas that on an e-ln p (or e-log p)
5. The peak strength of saturated clay is influenced plane tends to be linear for clays and nonlinear
most by overconsolidation ratio, drainage con- for sands.
l
ditions, effective confining pressures, original 7. At failure, dense sands and heavily overconsol-
ria
structure, disturbance (which causes a change in idated clays have a greater volume after drained
effective stress and a loss of cementation), and shear or a higher effective stress after undrained
creep or deformation rate effects. Overconsoli- shear than at the start of deformation. This is
dated clays usually have higher peak strength at due to its dilative tendency upon shearing. At
a given effective stress than normally consoli- failure, loose sands and normally consolidated
ate
dated clays, as shown in Fig. 11.3. The differ- to moderately overconsolidated clays (OCR up
ences in strength result from both the different to about 4) have a smaller volume after drained
stress histories and the different water contents shear or a lower effective stress after undrained
at peak. For comparisons at the same water con- shear than they had initially. This is due to its
tent but different effective stress, as for points
dM contractive tendency upon shearing.
A and A, the Hvorslev strength parameters ce 8. Under further deformation, platy clay particles
and e are obtained (Hvorslev, 1937, 1960). begin to align along the failure plane and the
Further details are given in Section 11.9. shear resistance may further decrease from the
6. During critical state deformation a soil is com- critical state condition. The angle of shear re-
pletely destructured. As illustrated in Fig. 11.1b, sistance at this condition is called the residual
the critical state friction angle values are inde- friction angle, as illustrated in Fig. 11.1b. The
pendent of stress history and original structure; postpeak shearing displacement required to
hte
for a given set of testing conditions the shearing cause a reduction in friction angle from the crit-
ical state value to the residual value varies with
the soil type, normal stress on the shear plane,
and test conditions. For example, for shale my-
Water Content w
Rebound
Overconsolidated
1
In three-dimensional stress space ⫽ ( x, y, z, xy, yz, zx) or
the equivalent principal stresses ( 1, 2, 3), the mean effective
stress p, and the deviator stress q is defined as
σff σe
τ
Co
500 4
Critical State Line Critical State Line
400 3
300
2
200
Overconsolidated
1
100 Normally Consolidated
l
ria
0 0
0 100 200 300 400 500 600 0 1 2 3 4
Mean Pressure p(kPa) Mean Pressure p(MPa)
(a-1) p versus q (b-1) p versus q
ate
0.7 Critical State Line Critical State Line
Initial State
0.95
Isotropic Normal
0.6 Compression Line
0.90
dM Void ratio e
Void ratio e
0.5 0.85
0.80
0.4
Overconsolidated
0.75
hte
(a) (b)
Figure 11.4 Critical states of clay and sand: (a) Critical state of Weald clay obtained by
drained triaxial compression tests of normally consolidated () and overconsolidated (●)
specimens: (a-1) q–p plane and (a-2) e–ln p plane (after Roscoe et al., 1958). (b) Critical
py
state of Toyoura sand obtained by undrained triaxial compression tests of loose and dense
specimens consolidated initially at different effective stresses, (b-1) q–p plane and (b-2) e–
log p plane (after Verdugo and Ishihara, 1996).
Co
shear plane of several tens of millimeters may of two samples of the same soil at the same void
be required, as shown by Fig. 11.5. However, ratio but with different fabrics are accountable
significant softening can be caused by strain in terms of different effective stresses as dis-
localization and development of shear bands, cussed in Chapter 8.
especially for dense samples under low confine- 10. Undrained strength in triaxial compression may
ment. differ significantly from the strength in triaxial
9. Strength anisotropy may result from both stress extension. However, the influence of type of test
and fabric anisotropy. In the absence of chemi- (triaxial compression versus extension) on the
cal cementation, the differences in the strength effective stress parameters c and is relatively
l
ria
ate
dM
Figure 11.5 Development of residual strength with increasing shear displacement (after
Bishop et al., 1971).
Stress–Strain Behavior
1. Stress–strain behavior ranges from very brittle
for some quick clays, cemented soils, heavily
overconsolidated clays, and dense sands to duc- Figure 11.6 Effect of temperature on undrained strength of
tile for insensitive and remolded clays and loose kaolinite in unconfined compression (after Sherif and Bur-
sands, as illustrated by Fig. 11.7. An increase in rous, 1969).
Stiffness G or E
Foundations
Tunnels
Linear Elastic
Nonlinear Elastic
Preyield Plastic
Full Plastic
l
ria
10-4 10-3 10-2 10-1 100 101
Figure 11.7 Types of stress–strain behavior.
Strain %
Dynamic Methods
Local Gauges
confining pressure causes an increase in the de-
ate
Conventional Soil Testing
formation modulus as well as an increase in
strength, as shown by Fig. 11.8. (b) Typical Strain Ranges for Laboratory Tests
2. Stress–strain relationships are usually nonlin- Figure 11.9 Stiffness degradation curve: stiffness plotted
ear; soil stiffness (often expressed in terms of against logarithm of strains. Also shown are (a) the strain
tangent or secant modulus) generally decreases
dM levels observed during construction of typical geotechnical
with increasing shear strain or stress level up to structures (after Mair, 1993) and (b) the strain levels that can
peak failure stress. Figure 11.9 shows a typical be measured by various techniques (after Atkinson, 2000).
stiffness degradation curve, in terms of shear
modulus G and Young’s modulus E, along with
typical strain levels developed in geotechnical
construction (Mair, 1993) and as associated with four zones: (1) linear elastic zone, (2) nonlinear
different laboratory testing techniques used to elastic zone, (3) pre-yield plastic zone, and (4)
full plastic zone.
hte
pendencies:
Confining
TC PSC Pressure
Toyoura Sand 78.4 kPa Confining Pressures
Secant Shear Modulus G (MPa)
40 40 σc = 30 kPa
l
20
ria
20
10-4 10-3 10-2 10-1 100 10-5 10-4 10-3 10-2 10-1 100
Shear Strain (%) Shear Strain (%)
ate
(a) (b)
Figure 11.10 Stiffness degradation curve at different confining pressures: (a) Toyoura and
Ticino sands (TC: triaxial compression tests, PSC: plain strain compression tests) (after
Tatsuoka et al., 1997) and (b) reconstituted Kaolin clay (after Soga et al., 1996).
dM
Figure 11.11 shows examples of the fitting of plastic soils at low confining pressure conditions
the above equations to experimental data. to greater than 5 ⫻ 10⫺2 percent at high confin-
4. The stiffness begins to decrease from the linear ing pressure or in soils with high plasticity (San-
elastic value as the applied strains or stresses tamarina et al., 2001).
increase, and the deformation moves into the 5. Irrecoverable strains develop in the pre-yield
nonlinear elastic zone. However, a complete cy- plastic zone. The initiation of plastic strains can
hte
cle of loading, unloading, and reloading within be determined by examining the onset of per-
this zone shows full recovery of strains. The manent volumetric strain in drained conditions
strain at the onset of the nonlinear elastic zone or residual excess pore pressures in undrained
ranges from less than 5 ⫻ 10⫺4 percent for non- conditions after unloading. Available experi-
rig
196 kPa
nG = 0.13
Evmax/FE(e) (MPa)
py
103 450
400 nE = 0.49
102 nG = 0.65
nG = 0.63 350
Co
101 300
100 250
100 101 102 103 104 100 150 200 250 300
Confining pressure, p⬘ (kPa) Vertical Effective Stress,σv⬘ (kPa)
(a) (b)
Figure 11.11 Small strain stiffness versus confining pressure: (a) Shear modulus Gmax of
cemented silty sand measured by resonant column tests (from Stokoe et al. 1995) and (b)
vertical Young’s modulus of sands measured by triaxial tests (after Tatsuoka and Kohata,
1995).
l
clay are shown in Fig. 11.12. The yield envelope 8. The stress–strain relation of clays depends
ria
expands, shrinks, and rotates as plastic strains largely on overconsolidation ratio, effective
develop. It is usually considered that expansion confining pressures, and drainage conditions.
is related to plastic volumetric strains; the sur- Figure 11.14 shows triaxial compression behav-
face expands when the soil compresses and ior of clay specimens that are first normally con-
shrinks when the soil dilates. The two inner en- solidated and then isotropically unloaded to
ate
velopes shown in Fig. 11.12b define the bound- different overconsolidation ratios before shear-
aries between linear elastic, nonlinear elastic, ing. The specimens are consolidated at the same
and pre-yield zones. When the stress state confining pressure p0, but have different void
moves in the pre-yield zone, the inner envelopes ratios due to the different stress history (Fig.
move with the stress state. This multienvelope
dM 11.14a). Drained tests on normally consolidated
concept allows modeling of complex deforma- clays and lightly overconsolidated clays show
tions observed for different stress paths (Mroz, ductile behavior with volume contraction (Fig.
1967; Prévost, 1977; Dafalias and Herrman, 11.14b). Heavily overconsolidated clays exhibit
1982; Atkinson et al., 1990; Jardine, 1992). a stiff response initially until the stress state
7. Plastic irrecoverable shear deformations of reaches the yield envelope giving the peak
saturated soils are accompanied by volume strength and volume dilation. The state of the
hte
Yield State
Initial Condition Pre-yield State
q = σ⬘a-σ⬘r Yield State q = σ⬘a-σ⬘r Initial State Surrounded by
MPa MPa Linear Elastic Boundary
Failure Line Stress Path
rig
0.8
0.6
Yield Envelope
Yield Envelope
0.6
0.4
0.4 Preyield Boundary
py
0.2
0.2 Linear Elastic
Boundary MPa
MPa
0.0 0.0
Co
-0.4
Failure Line
(a) (b)
Figure 11.12 Yield envelopes: (a) Aoi sand (Yasufuku et al., 1991) and (b) Bothkennar clay
(from Smith et al., 1992).
l
ria
Dense Soil
Dense Soil
-Δu Cavitation
+ΔV/V0
0 0
ate
Loose Soil Loose soil
-ΔV/V0 +Δu
dM (a) (b)
Figure 11.13 Volume and pore pressure changes during shear: (a) drained conditions and
(b) undrained conditions.
hte
3 Heavily
Overconsolidated 3 Heavily Overconsolidated
Initial State Deviator Deviator
Failure at Critical State Stress 2 Lightly Stress U3
(D: Drained, U: Undrained) Overconsolidated
Void 2 Lightly Overconsolidated
Ratio U2
D Critical State
rig
2 Lightly Overconsolidated
U2
D 3 Heavily Overconsolidated
+ΔV/V0 -Δu
U3 3 Heavily Overconsolidated
Co
3 Heavily
Overconsolidated
Critical
2 Lightly Overconsolidated
State Line 2 Lightly Overconsolidated
-ΔV/V0 +Δu
p0 log p 1 Normally Consolidated
1 Normally Consolidated
(a) (b) (c)
l
undrained effective stress paths of anisotropi-
ria
cally and isotropically consolidated specimens
(Ladd and Varallyay, 1965). The difference in
undrained shear strength is primarily due to dif-
ferent excess pore pressure development asso-
ciated with the change in soil fabric. At large
ate
strains, the stress paths correspond to the same
friction angle.
10. A temperature increase causes a decrease in un-
drained modulus; that is, a softening of the soil.
As an example, initial strain as a function of
dM
stress is shown in Fig. 11.16 for Osaka clay Figure 11.16 Effect of temperature on the stiffness of Osaka
clay in undrained triaxial compression (Murayama, 1969).
l
ria
ate
Figure 11.17 Stress–strain and volumetric strain relationships for sand at a void ratio of
0.64 but with different initial fabrics (after Oda, 1972a). (a) Sample saturated with water
and (b) sample saturated with water–resin solution.
dM
tamping. There is similarity between these curves and As the stress state approaches failure, a direct shear-
those for Monterey No. 0 sand shown in Fig. 8.23. A induced fabric forms that is generally composed of
statistical analysis of the changes in particle orientation regions of homogeneous fabric separated by discon-
with increase in axial strain showed: tinuities. No discontinuities develop before peak
strength is reached, although there is some particle ro-
hte
1. For samples prepared by tapping, the initial fab- tation in the direction of motion. Near-perfect preferred
ric tended toward some preferred orientation of orientation develops during yield after peak strength is
long axes parallel to the horizontal plane, and the reached, but large deformations may be required to
intensity of orientation increased slightly during reach this state.
deformation.
2. For samples prepared by tamping, there was very
rig
weak preferred orientation in the vertical direc- Compaction Versus Overconsolidation of Sand
tion initially, but this disappeared with deforma- Specimens at the same void ratio and stress state be-
tion. fore shearing, but having different fabrics, can exhibit
different stress–strain behavior. For example, consider
Shear deformations break down particle and aggre- a case in which one specimen is overconsolidated,
py
gate assemblages. Shear planes or zones did not appear whereas the other is compacted. The two specimens
until after peak stress had been reached; however, the are prepared in such a way that the initial void ratio is
distribution of normals to the interparticle contact the same for a given initial isotropic confining pres-
planes E() (a measure of fabric anisotropy) did sure. Coop (1990) performed undrained triaxial com-
Co
change with strain, as may be seen in Fig. 11.18. This pression tests of carbonate sand specimens that were
figure shows different initial distributions for samples either overconsolidated or compacted, as illustrated in
prepared by the two methods and a concentration of Fig. 11.19a. The undrained stress paths and stress–
contact plane normals within 50 of the vertical as de- strain curves for the two specimens are shown in Figs.
formation progresses. Thus, the fabric tended toward 11.19b and 11.19c, respectively. The overconsolidated
greater anisotropy in each case in terms of contact sample was initially stiffer than the compacted speci-
plane orientations. There was little additional change men. The difference can be attributed to (i) different
in E() after the peak stress had been reached, which soil fabrics developed by different stress paths prior to
implies that particle rearrangement was proceeding shearing and (ii) different degrees of particle crushing
without significant change in the overall fabric. prior to shearing (i.e., some breakage has occurred dur-
l
ria
ate
dM
Figure 11.18 Distribution of interparticle contact normals as a function of axial strain for
sand samples prepared in two ways (after Oda, 1972a): (a) specimens prepared by tapping
hte
ing the preconsolidation stage for the overconsolidated much lower peak strength for the sample prepared by
specimen). Therefore, overconsolidation and compac- kneading compaction.
rig
tion produced materials with different mechanical The recoverable deformation of compacted kaolinite
properties. However, at large deformations, both spec- with flocculent structure ranges between 60 and 90
imens exhibited similar strengths because the initial percent, whereas the recovery of samples with dis-
fabrics were destroyed. persed structures is only of the order of 15 to 30 per-
py
but more unstable than deflocculated fabrics. Similar associated with kneading compaction.
behavior may be observed in compacted fine-grained Different macrofabric features can affect the defor-
soils, and the results of a series of tests on structure- mation behavior as illustrated in Fig. 11.22 for the un-
sensitive kaolinite are illustrative of the differences drained triaxial compression testing of Bothkennar
(Mitchell and McConnell, 1965). Compaction condi- clay, Scotland (Paul et al., 1992; Clayton et al., 1992).
tions and stress–strain curves for samples of kaolinite Samples with mottled facies, in which the bedding fea-
compacted using kneading and static methods are tures had been disrupted and mixed by burrowing
shown in Fig. 11.20. The high shear strain associated mollusks and worms (bioturbation), gave the stiffest
with kneading compaction wet of optimum breaks response, whereas samples with distinct laminated fea-
down flocculated structures, and this accounts for the tures showed the softest response.
1.0
Overconsolidated
0.8
0.6
q (MPa)
Normal Compression Line
2 0.4
Compacted
0.2
Overconsolidated
Sample
Void 0
l
1.5 0 0.2 0.4 0.6
Ratio p (MPa)
ria
(b)
q (MPa)
Compacted Sample
1.0
1
Compacted
ate
0.75
0.5
(MPa) Overconsolidated
0.1 1
Mean Pressure p 0.25
dM (a)
0
0 4 8 12 16 20
Axial strain ε a (%)
(c)
If slip planes develop at failure, platy and elongated clay can also be accounted for in terms of differences
particles align with their long axes in the direction of in effective stress, provided part of the undisturbed
slip. By then, the basal planes of the platy clay parti- strength does not result from cementation. Remolding
rig
cles are enclosed between two highly oriented bands breaks down the structure and causes a transfer of ef-
of particles on opposite sides of the shear plane. The fective stress to the pore water.
dominant mechanism of deformation in the displace- An example of this is shown in Fig. 11.23, which
ment shear zone is basal plane slip, and the overall shows the results of incremental loading triaxial com-
py
thickness of the shear zone is on the order of 50 m. pression tests on two samples of undisturbed and re-
Fabrics associated with shear planes and zones have molded San Francisco Bay mud. In these tests, the
been studied using thin sections and the polarizing mi- undisturbed sample was first brought to equilibrium
croscope and by using the electron microscope (Mor- under an isotropic consolidation pressure of 80 kPa.
genstern and Tchalenko, 1967a, b and c; Tchalenko, After undrained loading to failure, the triaxial cell was
Co
1968; McKyes and Yong, 1971). The residual strength disassembled, and the sample was remolded in place.
associated with these fabrics is treated in more detail The apparatus was reassembled, and pore pressure was
in Section 11.11. measured. Thus, the effective stress at the start of com-
pression of the remolded clay at the same water con-
tent as the original undisturbed clay was known.
Structure, Effective Stresses, and Strength
Stress–strain and pore pressure–strain curves for two
The effective stress strength parameters such as c and samples are shown in Figs. 11.23a and 11.23b, and
are isotropic properties, with anisotropy in un- stress paths for test 1 are shown in Fig. 11.23c.
drained strength explainable in terms of excess pore Differences in strength that result from fabric dif-
pressures developed during shear. The undrained ferences caused by thixotropic hardening or by differ-
strength loss associated with remolding undisturbed ent compaction methods can be explained in the same
l
ria
ate
Figure 11.21 Ratio of recoverable to total strain for samples
dM of kaolinite with different structure.
0.6 0.6
hte
(σa – σr)/2 σao
Facies Facies
0.2 0.2
Mottled Mottled
rig
Bedded Bedded
Laminated Laminated
0.0 0.0
Figure 11.20 Stress–strain behavior of kaolinite compacted 0 2 4 0.4 0.6 0.8 1.0
Axial Strain (%) (σa + σr)/2σao
by two methods.
py
T
11.4 FRICTION BETWEEN SOLID SURFACES ⫽ ⫽ tan (11.8)
N
The friction angle used in equations such as (11.1),
(11.2), (11.4), and (11.5) contains resistance contri- where N is the normal load on the shear surface, T is
butions from several sources, including sliding of the shear force, and , the intergrain sliding friction
grains in contact, resistance to volume change (dila- angle, is a compositional property that is determined
tancy), grain rearrangement, and grain crushing. The by the type of soil minerals.
l
ria
ate
dM
hte
rig
Figure 11.23 (a) and (b) Effect of remolding on undrained strength and pore water pressure
in San Francisco Bay mud. (c) Stress paths for triaxial compression tests on undisturbed and
py
Basic ‘‘Laws’’ of Friction 11.24, the value of T is the same for a given value
Co
Two laws of friction are recognized, beginning with of N regardless of the size of the sliding block.
Leonardo da Vinci in about 1500. They were restated
by Amontons in 1699 and are frequently referred to as Although these principles of frictional resistance
Amontons’ laws. They are: have long been known, suitable explanations came
much later. It was at one time thought that interlocking
1. The frictional force is directly proportional to the between irregular surfaces could account for the be-
normal force, as illustrated by Eq. (11.8) and Fig. havior. On this basis, would be given by the tangent
11.24. of the average inclination of surface irregularities on
2. The frictional resistance between two bodies is the sliding plane. This cannot be the case, however,
independent of the size of the bodies. In Fig. because such an explanation would require that de-
l
ria
ate
Figure 11.23
dM (Continued )
N
Ac ⫽ (11.9)
y
py
T ⫽ Acm (11.10)
Figure 11.24 Coefficient of friction for surfaces in contact.
The coefficient of friction is given by T/N,
l
ranging from 10 nm to over 100 nm in height. The
ria
slopes of the nanoscale asperities are rather flat, with
individual angles ranging from about 120 to 175 as 1
tm ⫽ (11.12)
shown in Fig. 11.25. The average slope of asperities SZ
on metal surfaces is an included angle of 150; on
rough quartz it may be over 175 (Bromwell, 1966).
ate
When two surfaces are brought together, contact is es- where is the area occupied per molecule, S is the
tablished at the asperities, and the actual contact area fraction of molecules striking the surface that stick to
is only a small fraction of the total surface area. it, and Z is the number of molecules per second strik-
Quartz surfaces polished to mirror smoothness may ing a square centimeter of surface. For a value of S
consist of peaks and valleys with an average height of equal to 1, which is reasonable for a high-energy sur-
about 500 nm. The asperities on rougher quartz sur- face, the relationship between tm and gas pressure is
dM
faces may be about 10 times higher (Lambe and Whit- shown in Fig. 11.26. The conclusion to be drawn from
man, 1969). Even these surfaces are probably smoother this figure is that adsorbed layers are present on the
than most soil particles composed of bulky minerals. surface of soil particles in the terrestrial environment,
The actual surface texture of sand particles depends on
geologic history as well as mineralogy, as shown in
Fig. 2.12.
hte
particle.
Thus, large areas of solid contact between grains are
not probable in soils. Solid-to-solid contact is through
asperities, and the corresponding interparticle contact
stresses are high. The molecular structure and com-
py
l
tional to the normal load on the asperity as shown by friction angle () is less for small and platy particles
ria
Eq. (11.9). Because surfaces are not clean, but are cov- than for large and bulky particles. In the event that two
ered by adsorbed films, actual solid contact may de- platy particles are in face-to-face contact and the sur-
velop only over a fraction of the contact area as face waviness is insufficient to cause direct solid-to-
shown in Fig. 11.27. If the contaminant film strength solid contact, shear will be through the adsorbed films,
and the effective value of will be zero, again giving
ate
is c, the strength of the contact will be
a lower value of .
T ⫽ Ac[m ⫹ (1 ⫺ )c] (11.13) In reality, the behavior of plastic junctions is more
complex. Under combined compression and shear
Equation (11.13) cannot be applied in practice be- stresses, deformation follows the von Mises–Henky
cause and c are unknown. However, it does provide
dM criterion, which, for two dimensions, is
a possible explanation for why measured values of fric-
tion angle for bulky minerals such as quartz and feld- 2 ⫹ 3 2 ⫽ y2 (11.14)
spar are greater than values for the clay minerals and
other platy minerals such as mica, even though the For asperities loaded initially to ⫽ y, the appli-
surface structure is similar for all the silicate minerals. cation of a shear stress requires that become less
The small particle size of clays means that the load than y. The only way that this can happen is for the
per particle, for a given effective stress, will be small contact area to increase. Continued increase in leads
hte
relative to that in silts and sands composed of the bulky to continued increase in contact area. This phenome-
minerals. The surfaces of platy silt and sand size par- non is called junction growth and is responsible for
ticles are smoother than those of bulky mineral parti- cold welding in some materials (Bowden and Tabor,
1964). If the shear strength of the junction equals that
of the bulk solid, then gross seizure occurs. For the
rig
d ⫽ (NR)1 / 3 (11.15)
Co
l
N 4 tional resistance varies as ()⫺1 / 3 as predicted by Eq.
ria
(11.22) up to a normal stress of the order of 200 kPa
According to these relationships, the friction coef- (30 psi), that is, the friction angle decreases with in-
ficient for two elastic asperities in contact should de- creasing normal stress (Chattopadhyay, 1972).
crease with increasing load. Nonetheless, the adhesion There are at least two possible explanations of the
theory would still apply to the strength of the junction, normal stress independence of the frictional resistance
ate
with the frictional force proportional to the area of real of quartz, feldspar, and calcite:
contact.
If it is assumed that the number of contacting as- 1. As the load per particle increases, the number of
perities in a soil mass is independent of particle size asperities in contact increases proportionally, and
and effective stress, then the influences of particle size the deformation of each asperity remains essen-
and effective stress on the frictional resistance of a soil
dM
with asperities deforming elastically may be analyzed.
tially constant. In this case, the assumption of one
asperity per contact for the development of Eq.
For uniform spheres arranged in a regular packing, the (11.22) is not valid. Some theoretical considera-
gross area covered by one sphere along a potential tions of multiple asperities in contact are availa-
plane of sliding is 4R 2. The normal load per contacting ble (Johnson, 1985). They show that the area of
asperity, assuming one asperity per contact, is contact is approximately proportional to the ap-
plied load and hence the coefficient of friction is
N ⫽ 4R2
hte
(Ac)T ⫽ 冉 冊
1 2
R (4)2 / 3 ⫽
(4)2 / 3 Quartz is a hard, brittle material that can exhibit both
py
l
ria
Figure 11.28 Variation in friction angle with normal stress for different minerals (after
Kenney, 1967).
ate
creased from 31 for coarse silt to 22 for coarse sand. do tests on two very small particles that are sliding
This is an apparent contradiction to the independence relative to each other, and test results for particle as-
of particle size on frictional resistance predicted by Eq. semblages are influenced by particle rearrangements,
(11.22). On the other hand, the assumption of one as- volume changes, surface preparation factors, and the
dM
perity per contact may not have been valid for all par-
ticle sizes, and additionally, particle surface textures on
like. Some values are available, however, and they are
presented and discussed in this section.
a microscale could have been size dependent. Further-
more, there could have been different amounts of par-
ticle rearrangement and rolling in the tests on the Nonclay Minerals
different size fractions. Values of the true friction angle for several minerals
hte
are listed in Table 11.1, along with the type of test and
Sliding Friction conditions used for their determination. A pronounced
The frictional resistance, once sliding has been initi- antilubricating effect of water is evident for polished
ated, may be equal to or less than the resistance that surfaces of the bulky minerals quartz, feldspar, and cal-
had to be overcome to initiate movement; that is, the cite. This apparently results from a disruptive effect of
coefficient of sliding friction can be less than the co- water on adsorbed films that may have acted as a lu-
rig
efficient of static friction. A higher value of static bricant for dry surfaces. Evidence for this is shown in
friction than sliding friction is explainable by time- Fig. 11.29, where it may be seen that the presence of
dependent bond formations at asperity junctions. water had no effect on the frictional resistance of
Stick–slip motion, wherein varies more or less er- quartz surfaces that had been chemically cleaned prior
to the measurement of the friction coefficient. The
py
vidual contacts is masked by the behavior of the mass feldspar) or carbonate surface (calcite) and the for-
as a whole. However, it may be an important mecha- mation of silica and carbonate cement at interparticle
nism of energy dissipation for cyclic loading at very contacts. Many sand deposits exhibit ‘‘aging’’ effects
small strains when particles are not moving relative to wherein their strength and stiffness increase noticeably
each other. within periods of weeks to months after deposition,
disturbance, or densification, as described, for exam-
ple, by Mitchell and Solymar (1984), Mitchell (1986),
11.5 FRICTIONAL BEHAVIOR OF MINERALS Mesri et al. (1990), and Schmertmann (1991). In-
creases in penetration resistance of up to 100 percent
Evaluation of the true coefficient of friction and fric- have been measured in some cases. The relative im-
tion angle is difficult because it is very difficult to portance of chemical factors, such as precipitation at
l
Quartz Block on block Dry 7.4 Polished surfaces Horn and Deere (1962)
ria
Water saturated 24.2
Quartz Particles on Water saturated 22–31 decreasing with in- Rowe (1962)
polished block creasing particle size
Quartz Block on block Variable 0–45 Depends on roughness Bromwell (1966)
and cleanliness
ate
Quartz Particle–particle Saturated 26 Single-point contact Procter and Barton
(1974)
Particle–plane Saturated 22.2
Particle–plane Dry 17.4
Feldspar Block on block Dry 6.8 Polished surfaces Horn and Deere (1962)
Water saturated 37.6
Feldspar
dM
Free particles on flat Water saturated 37 25–500 sieve Lee (1966)
surface
Feldspar Particle–plane Saturated 28.9 Single-point contact Procter and Barton
(1974)
Calcite Block on block Dry 8.0 Polished surfaces Horn and Deere (1962)
Water saturated 34.2
hte
Muscovite Along cleavage Dry 23.3 Oven dry Horn and Deere (1962)
faces
Dry 16.7 Air equilibrated
Saturated 13.0
Phlogopite Along cleavage Dry 17.2 Oven dry Horn and Deere (1962)
faces
rig
Saturated 7.4
Chlorite Along cleavage Dry 27.9 Oven dry Horn and Deere (1962)
faces
Dry 19.3 Air equilibrated
Co
Saturated 12.4
interparticle contacts, changes in surface characteris- As surface roughness increases, the apparent anti-
tics, and mechanical factors, such as time-dependent lubricating effect of water decreases. This is shown
stress redistribution and particle reorientations, in caus- in Fig. 11.29 for quartz surfaces that had not been
ing the observed behavior is not known. Further details cleaned. Chemically cleaned quartz surfaces, which
of aging effects are given in Chapter 12. give the same value of friction when both dry and wet,
l
ria
ate
dM
Figure 11.29 Friction of quartz (data from Bromwell, 1966 and Dickey, 1966).
particles are most probably rough relative to the scale gles have been measured. Residual values as low as 4
in Fig. 11.29, and they will not be chemically clean. for sodium montmorillonite are indicated by the data
Thus, values of ⫽ 0.5 and ⫽ 26 are reasonable in Fig. 11.28.
for quartz, both wet and dry. The effective stress failure envelopes for calcium
On the other hand, water apparently acts as a lubri-
py
layer is not easily disrupted. Observations have shown over the range investigated, which was 0.001 N to 0.1
that the surfaces of the sheet minerals are scratched N. The water content at any effective stress was inde-
when tested in air (Horn and Deere, 1962). When the pendent of electrolyte concentration for calcium mont-
surfaces of the layer silicates are wetted, the mobility morillonite, but varied in the manner shown in Fig.
of the surface films is increased because of their in- 11.31 for sodium montmorillonite.
creased thickness and because of greater surface ion This consolidation behavior is consistent with that
hydration and dissociation. Thus, the values of described in Chapter 10. Interlayer expansion in cal-
listed in Table 11.1 for the sheet minerals under satu- cium montmorillonite is restricted to a c-axis spacing
rated conditions (7 –13) are probably appropriate for of 1.9 nm, leading to formation of domains or layer
sheet mineral particles in soils. aggregates of several unit layers. The interlayer spac-
Figure 11.30 Effective stress failure diagrams for calcium and sodium montmorillonite (af-
l
ter Mesri and Olson, 1970).
ria
the same time there is little increase in shear strength
because the shearing strength of water and solutions is
essentially independent of hydrostatic pressure. The
ate
small friction angle that is observed for sodium mont-
morillonite at low effective stresses can be ascribed
mainly to the few interparticle contacts that resist par-
ticle rearrangement. Resistance from this source evi-
dM dently approaches a constant value at the higher
effective stresses, as evidenced by the nearly horizontal
failure envelope at values of average effective stress
greater than about 50 psi (350 kPa), as shown in Fig.
11.30. The viscous resistance of the pore fluid may
Figure 11.31 Shear and consolidation behavior of sodium contribute a small proportion of the strength at all ef-
montmorillonite (after Mesri and Olson, 1970). fective stresses.
An hypothesis of friction between fine-grained par-
hte
films held apart by high repulsive forces that carry the nanometers, there are multiple wells of minimum
effective stress. For this case, if it is assumed that there energy between nearby surfaces and a force is
is essentially no intergranular contact, then Eq. (7.29) required to overcome the energy barrier between
becomes the wells when the particles move relative to each
Co
l
1994), offer physical insights into particle interactions
ria
and load transfers that are difficult to deduce from
physical experiments. Typical inputs for the simula-
tions are particle packing conditions and interparticle
contact characteristics such as the interparticle friction
angle . Complete details of these numerical methods
ate
are beyond the scope of this book; additional infor-
mation can be found in Oda and Iwashita (1999).
However, some of the main findings are useful for
developing an improved understanding of how stresses
dM are carried through discrete particle systems such as
soils and how these distributions influence the defor-
mation and strength properties.
more physical interference and more intergrain contact lines. This is called the strong force network and is the
than in sodium montmorillonite since the water content key microscopic feature of load transfer through the
range for the strength data shown in Fig. 11.30 was granular system. The scale of statistical homogeneity
only about 50 to 97 percent, whereas it was about 125 in a two-dimensional particle assembly is found to be
to 450 percent for the sodium montmorillonite. At a a few tens of particle diameters (Radjai et al., 1996).
py
consolidation pressure of about 500 kPa, the slope of Forces averaged over this distance could therefore be
the failure envelope for calcium montmorillonite was expected to give a stress that is representative of the
about 10, which is in the middle of the range for non- macroscopic stress state. The particles not forming a
clay sheet minerals (Table 11.1). part of the strong force network are floating like a fluid
Co
l
ticles in the weak clusters, and the particles are perhaps
ria
behaving like a viscous fluid.
ate
shear, particles in the strong force network do not slide,
but columns of particles buckle (Cundall and Strack,
1979). Particles in the strong force network collapse
upon buckling, and new force chains are formed.
Hence, the spatial distributions of the strong force net-
work are neither static nor persistent features.
dM At a given time of biaxial compression loading, par-
ticle sliding is occurring at almost 10 percent of the
contacts (Kuhn, 1999) and approximately 96 percent
of the sliding particles are in the weak clusters (Radjai
Figure 11.33 Normal force distributions of a two- et al., 1996). Over 90 percent of the energy dissipation
dimensional disk particle assembly: (a) isotropic stress con- occurs at just a small percentage of the contacts (Kuhn,
hte
dition and (b) biaxial stress condition with maximum load in 1999). This small number of sliding particles is asso-
the vertical direction (after Thornton and Barnes, 1986). ciated with the ability of particles to roll rather than to
slide. Particle rotations reduce contact sliding and dis-
sipation rate in the granular system. If all particles
could roll upon one another, a granular assembly
as six times the mean normal contact force, but would deform without energy dissipation.7 However,
rig
approximately 60 percent of contacts carry normal this is not possible owing to restrictions on particle
contact forces below the mean (i.e., weak cluster rotations. It is impossible for all particles to move by
particles). When normal contact forces are larger than rotation, and sliding at some contacts is inevitable due
their mean, the distribution law of forces can be ap- to the random position of particles (Radjai and Roux,
proximated by an exponentially decreasing function; 1995).8 Some frictional energy dissipation can there-
py
Radjai et al. (1996) show that PN ( ⫽ N/ ⬍N⬎) ⫽ fore be considered a consequence of disorder of par-
ke1.4(1⫺ ) fits the computed data well for both two-and ticle positions.
three-dimensional simulations. The exponent was As deformation progresses, the number of particles
found to change very slightly with the coefficient of in the strong force network decreases, with fewer par-
Co
interparticle friction and to be independent of particle ticles sharing the increased loads (Kuhn, 1999). Figure
size distributions.
Simulations show that applied deviator load is trans-
ferred exclusively by the normal contact forces in the 7
This assumes that the particles are rigid and rolling with a single-
strong force networks, and the contribution by the point contact. In reality, particles deform and exhibit rolling resis-
weak clusters is negligible. This is illustrated in Fig. tance. Iwashita and Oda (1998) state that the incorporation of rolling
11.35, which shows that the normal contact forces con- resistance is necessary in discrete particle simulations to generate
realistic localized shear bands.
tribute greater than the tangential contact forces to the 8
For instance, consider a chain loop of an odd number of particles.
development of the deviator stress during axisymme- Particle rotation will involve at least one sliding contact.
l
ria
Figure 11.34 Probability distributions of interparticle contact forces: (a) normal forces and
ate
(b) tangential forces. The distributions were obtained for contact dynamic simulations of
500, 1024, 1200 and 4025 particles. The effect of number of particles in the simulation on
probability distribution appears to be small (after Radjai et al., 1996).
dM
hte
rig
py
Figure 11.35 Contributions of normal and tangential contact Figure 11.36 Contributions of strong and weak contact
forces to the evolution of the deviator stress during axisym- forces to the evolution of the deviator stress during axisym-
metric compression of a dense granular assembly (after metric compression of a dense granular assembly (after
Thornton, 2000).
Co
Thornton, 2000).
11.37 shows the spatial distribution of residual defor- boundary of the group shows large residual deforma-
mation, in which the computed deformation of each tion, whereas the center shows very small residual de-
particle is subtracted from the average overall defor- formation. The rotating group of interlocked particles,
mation (Williams and Rege, 1997). A group of inter- which can be considered as a weak cluster, becomes
locked particles that instantaneously moves as a rigid more apparent as applied strains increase toward fail-
body in a circular manner can be observed. The outer ure. The bands of large residual deformation [termed
-8 -6 -4 -2 2 4 6 8 10
-0.5 Axial Strain (%)
(a)
l
ria
Fabric Anisotropy A
Contact Plane Normals
in Initial State: 0.4
A More in Vertical Direction
0.3
ate
B Same in All Directions
0.2
C More in Horizontal Direction
0.1
-8 -6 -4 -2 2 4 6 8 10
-0.1
Axial Strain (%)
-0.2
dM -0.3
-0.4
-0.5
nesses are 1.5D50 to 2.5D50 in the early stages of shear- prepared at different states of initial contact anisotropy
ing and increase to between 1.5D50 and 4D50 as under an isotropic stress condition (Yimsiri, 2001). The
deformation proceeds. This microband slip zone may initial void ratios are similar (e0 ⬇ 0.75 to 0.76) and
eventually become a localized shear band. both drained triaxial compression and extension tests
were simulated. Although all specimens are initially
py
develop anisotropy in contact orientations. An initial ple C: more in the horizontal direction). As shown in
isotropic packing of particles develops an anisotropic Fig. 11.38a, both samples A and C showed stiffer re-
contact network during compression loading. This is sponse when the compression loading was applied in
because new contacts form in the direction of com- the preferred direction of contact forces, but softer re-
pression loading and contacts that orient along the di- sponse when the loading was perpendicular to the pre-
rection perpendicular to loading direction are lost. ferred direction of contact forces. The response of
The initial state of contact anisotropy (or fabric) sample B, which had an isotropic fabric, was in be-
plays an important role in the subsequent deformation tween the two. Dilation was most intensive when the
as illustrated in Fig. 11.18. Figure 11.38 shows results contact forces were oriented preferentially in the di-
l
specimens fail. The final steady-state value is indepen- 1). Again the directional evolution of the fabric in the
ria
dent of the initial fabric, indicating that the inherent weak clusters is opposite to the direction of loading.
anisotropy is destroyed by the shearing process. The Therefore, the stability of the strong force chains
final fabric anisotropy after triaxial extension is larger aligned in the vertical loading direction is obtained by
than that after triaxial compression because the addi- the lateral forces in the surrounding weak clusters.
tional confinement by a larger intermediate stress in
ate
the extension tests created a higher degree of fabric
anisotropy. Changes in Number of Contacts and Microscopic
Close examination of the contact force distribution Voids
for the strong force network and weak clusters gives At the beginning of biaxial loading of a dense granular
interesting microscopic features. Figure 11.39 shows dM assembly, more contacts are created from the increase
the values of A determined for the subgroups of contact
in the hydrostatic stress, and the local voids become
smaller. As the axial stress increases, however, the lo-
cal voids tend to elongate in the direction of loading
as shown in Fig. 11.41. Consequently particle contacts
0.1 are lost. As loading progresses, vertically elongated lo-
cal voids become more apparent, leading to dilation in
Fabric Anisotropy Parameter A
hte
0.05
0.0
1 2 3 4 5 6
N/<N>
rig
-0.05
-0.1
py
9
The density of contact plane normals E() with direction is fitted
with the following expression (Radjai, 1999):
c
E() ⫽ {1 ⫹ A cos 2( ⫺ c)}
where c is the total number of contacts, c is the direction for which Figure 11.40 Evolution of the fabric anisotropy parameters
the maximum E is reached, and the magnitude of A indicates the
amplitude of anisotropy. When the directional distribution of contact of strong forces and weak clusters when the specimen is un-
forces is independent of , the system has an isotropic fabric and der biaxial compression loading conditions (after Thornton
A ⫽ 0. and Antony, 1998).
l
ria
ate
dM
hte
Figure 11.41 Simulated spatial distribution of local microvoids under biaxial loading (after
Iwashita and Oda, 2000): (a) 11 ⫽ 1.1% (before failure), (b) 11 ⫽ 2.2% (at failure), (c)
rig
terms of overall sample volume (Iwashita and Oda, strain is required to reach the critical state void ratio.
py
50
40
Macroscopic Friction Angle (degrees)
30
l
ria
20
Drained (Thornton, 2000)
Drained Triaxial Compression (Yimsiri, 2001)
ate
Undrained Triaxial Compression (Yimsiri, 2001)
10 Drained Triaxial Extension (Yimsiri, 2001)
Undrained Triaxial Extension (Yimsiri, 2001)
Experiment (Skinner, 1969)
0
dM 0 10 20 30 40 50 60 70 80 90
Figure 11.42 Relationships between interparticle friction angle and macroscopic friction
angle from discrete element simulations. The macroscopic friction angle was determined
from simulations of drained and undrained triaxial compression (TC) and extension (TE)
tests. The experimental data by Skinner (1969) is also presented (after Thornton, 2000, and
Yimsiri, 2001).
hte
more than 20, the contribution of increasing interpar- have like a fluid. Increased friction at the contacts in-
ticle friction to the macroscopic friction angle becomes creases the stability of the system and reduces the
relatively small; the macroscopic friction angle ranges number of contacts required to achieve a stable con-
rig
between 30 and 40, when the interparticle friction dition. As long as the strong force network can be
angle increases from 30 to 90.10 formed, however, the magnitude of the interparticle
The nonproportional relationship between macro- friction becomes of secondary importance.
scopic friction angle of the particle assembly and in- The above findings from discrete particle simula-
terparticle friction angle results because deviatoric load tions are partially supported by the experimental data
py
is carried by the strong force networks of normal given by Skinner (1969), which are also shown in Fig.
forces and not by tangential forces, whose magnitude 11.42. He performed shear box tests on spherical par-
is governed by interparticle friction angle. Increase in ticles with different coefficients of interparticle friction
interparticle friction results in a decrease in the per- angle. The tested materials included glass ballotini,
Co
centage of sliding contacts (Thornton, 2000). The in- steel ball bearings, and lead shot. Use of glass ballotini
terparticle friction therefore acts as a kinematic was particularly attractive since the coefficient of in-
constraint of the strong force network and not as the terparticle friction increases by a factor of between 3.5
direct source of macroscopic resistance to shear. If the and 30 merely by flooding the dry sample. Skinner’s
interparticle friction were zero, strong force chains data shown in Fig. 11.42 indicate that the macroscopic
could not develop, and the particle assembly will be- friction angle is nearly independent of interparticle
friction angle.
l
Critical M (triaxial
Rege, 1997). An assembly of rounded particles exhib- extension)
ria
its greater softening behavior with fabric anisotropy State Line
change with strain, whereas an assembly of elongated
particles requires more shearing to modify its initial (a)
fabric anisotropy to the critical state condition
(Nouguier-Lehon et al., 2003). Specific Compression Lines of
ate
Volume v Constant Stress Ratio q/p
Γ
11.7 CRITICAL STATE: A USEFUL REFERENCE λ
CONDITION
λ cs
After large shear-induced volume change, a soil under
dM
a given effective confining stress will arrive ultimately
Isotropic
Compression
Line
at a unique final water content or void ratio that is Critical
independent of its initial state. At this stage, the inter- State Line
locking achieved by densification or overconsolidation
1 ln p
is gone in the case of dense soils, the metastable struc-
ture of loose soils has collapsed, and the soil is fully (b)
hte
destructured. A well-defined strength value is reached Figure 11.43 Critical state concept: (a) p–q plane and (b)
at this state, and this is often referred to as the critical v–ln p plane.
state strength. Under undrained conditions, the critical
state is reached when the pore pressure and the effec-
tive stress remain constant during continued deforma-
tion. The critical state can be considered a fundamental state stress ratio. The critical state on the void ratio (or
rig
state, and it can be used as a reference state to explain specific volume)–mean pressure plane is defined by
the effect of overconsolidation ratios, relative density, two material parameters: cs, the critical state com-
and different stress paths on strength properties of soils pression index and %, the specific volume intercept at
(Schofield and Wroth, 1968). unit pressure (p ⫽ 1). The compression lines under
constant stress ratios are often parallel to each other,
py
l
and, if interpreted properly, they can provide useful pcs and qcs, but not ecs; therefore, it is not at the critical
ria
and simple phenomenological representations of com- state.
plex behavior. Converting the void ratio in Eq. (11.26) to water
The drained and undrained critical state strengths are content, a normalized critical state line can be written
illustrated in Figs. 11.44a and 11.44b for normally using the liquidity index (see Fig. 11.45).
ate
consolidated clay and heavily overconsolidated clay,
respectively. The initial mean pressure–void ratio state wcs ⫺ wPL /p)
ln(pPL
of the normally consolidated clay is above the critical LIcs ⫽ ⫽ (11.29)
wLL ⫺ wPL ln(pPL /pLL)
state line, whereas that of the heavily overconsolidated
clay is below the critical state line. When the initial
state of the soil is normally consolidated at A (Fig.
dM where wcs is the water content at critical state when
11.44a), the critical state is B for undrained loading the effective mean pressure is p. pLL and pPL
are the
Critical
Deviator Critical Deviator Drained Peak Strength State Line
Stress q State Line M Stress q M
Drained Strength Undrained Strength
hte
C
Drained Strength
E
Undrained G F
Strength
3
B 3
1 D
1
rig
Specific Specific
Volume v Volume v
py
Γ Γ
B A F
Isotropic G E Isotropic
Co
Compression Compression
Line D Line
D
C
λ λ
λ cs λ cs
Critical State Line Critical State Line
1 ln p 1 ln p
(a) (b)
Figure 11.44 Drained and undrained stress–strain response using the critical state concept:
(a) normally consolidated clay and (b) overconsolidated clay.
Water Liquidity
Critical Isotropic Critical
Content Index
State Line Compression Line State Line
Liquid Limit
wLL LI = 1
l
ria
ln(pLL) ln(p) ln(pPL) ln(pLL) ln(p) ln(pPL) Mean
Mean
pressure Pressure
(a) (b)
ate
Figure 11.45 Normalization of the critical state line: (a) water content versus mean pressure
and (b) liquidity index versus mean pressure.
ecs ⫽ e0 ⫺ s 冉冊p
pa
12
A review by Sharma and Bora (2003) gives average values of where e0 is the void ratio at p ⫽ 0, pa is atmospheric pressure, and
suLL ⫽ 1.7 kPa and suPL ⫽ 170 kPa. s and are material constants.
3 MTC
Triaxial Compression
MTC 1
l
Triaxial Extension
ria
2
-1.5 MTE
3
ate
(a)
(b)
Figure 11.46 Plastic state of clay in relation to normalized liquidity index: (a) stress ratio
when plastic state initiates for a given LIeq and (b) definition of stress ratios used in (a) (after
Schofield, 1980).
dM
emax – ecs 1
DR,cs =
emax – emin = In (σc/p)
e max 0 emax 0
hte
0.2 0.2
Relative Density Dr
Relative Density Dr
0.4 0.4
0.6 0.6
rig
0.8 0.8
e min 1 emin 1
1.2 1.2
py
Figure 11.47 Critical state line derived from Eq. (11.32): (a) e–log p plane and (b) e–p
Co
plane.
behavior of clays. Coop and Lee (1993) found that would not be sufficient to cause further breakage. Coop
there is a unique relationship between the amount of et al. (2004) performed ring shear tests (see Section
particle breakage that occurred on shearing to a critical 11.11) on a carbonate sand to find a shear strain re-
state and the value of the mean normal effective stress. quired to reach the true critical state (i.e., constant
This implies that sand at the critical state would reach particle grading). They found that particle breakage
a stable grading at which the particle contact stresses continues to very large strains, far beyond those
reached in triaxial tests. Figure 11.48a shows the vol- Early Studies
umetric strains measured for a selection of their tests, The important role of volume change during shear, es-
which were carried out at vertical stress levels in the pecially dilatancy, was recognized by Taylor (1948).
range of 650 to 860 kPa. A constant volumetric strain From direct shear box testing of dense sand specimens,
is reached at a shear strain of around 2000 percent. For he calculated the work at peak shear stress state and
specimens at lower stress levels, more shear strains showed that the energy input is dissipated by the fric-
(20,000 percent or more) were required. Similar find- tion using the following equation:
ings were made for quartz sand (Luzzani and Coop,
2002). Figure 11.48b shows the degree of particle peak dx ⫺ n dy ⫽ n dx (11.33)
breakage with shear strains in the logarithmic scale.
The amount of breakage is quantified by Hardin’s
where peak is the applied shear stress at peak, n is the
l
(1985) relative breakage parameter Br defined in Fig.
confining normal (effective) stress on the shear plane,
ria
10.14. At very large strains, the value of Br finally
dx is the incremental horizontal displacement at peak,
stabilizes. The strain required for stabilization depends
dy is the incremental vertical displacement (expansion
on applied stress level. Interestingly, less shear strain
positive) at peak stress, and is the friction coefficient.
was needed for the mobilized friction angle to reach
The energy dissipated by friction (the component in
the steady-state value (Fig. 11.48c) than for attainment
ate
the right-hand side) is equal to the sum of the work
of the constant volume condition, (Fig. 11.48a). The
done by shearing (first component in the left-hand
critical state friction angle was unaffected by the par-
side) and that needed to increase the volume (the sec-
ticle breakage.
ond component in the left-hand side). The latter com-
In summary, the critical state concept is very useful
ponent is referred to as dilatancy.
to characterize the strength and deformation properties
dM Rearranging Eq. (11.33),
of soils when it is used as a reference state. That is, a
冉冊
soil has a tendency to contract upon shearing when its
peak dy
state is above the critical state line, whereas it has a ⫽ tan m ⫽ ⫹ (11.34)
tendency to dilate when it is below the critical state dx
line. Various normalized state parameters have been
proposed to characterize the difference between the ac- Thus, the peak shear stress ratio or the peak mobilized
tual state and the critical state line, as illustrated in Fig. friction angle m consists of both interlocking (dy/dx)
and sliding friction between grains (). This equation
hte
11.8 STRENGTH PARAMETERS FOR SANDS angle m must take into account particle rearrange-
ments as well as the sliding resistance at contacts and
Many factors and phenomena act together to deter- dilation. Particle crushing, which increases in impor-
mined the shearing resistance of sands, for example, tance as confining pressure increases and void ratio
mineralogy, grain size, grain shape, grain size distri- decreases, should also be added to these components.
py
bution, (relative) density, stress state, type of tests and The general interrelationships among the strength con-
stress path, drainage, and the like [see Eq. (11.3)]. In tributing factors and porosity can be represented as
this section, the ways in which these factors have be- shown in Fig. 11.50. In this figure, f is the friction
come understood and have been quantified over the last angle corrected for the work of dilation. It is influenced
Co
several decades are summarized. Several correlations by particle packing arrangements and the number of
are given to provide an overview and reference for sliding contacts. The denser the packing, the more im-
typical values and ranges of strength parameters for portant is dilation. As the void ratio increases, the mo-
sands and the influences of various factors on these bilized friction angle decreases. The critical state is
parameters.14 defined as the condition when there is no volume
change by shearing [i.e., (dy/dx) ⫽ 0 in Eq. (11.34)].
The corresponding mobilized friction angle m is crit .
14
A number of additional useful correlations are given by Kulhawy The ‘‘true friction’’ in the figure is associated with the
and Mayne (1990). resistance to interparticle sliding.
l
20 RS13
ria
RS15
40
Luzzani & Coop, 805 kPa
ate
650-930 kPa
1.0
248-386 kPa
RS7
0.6
dM RS8
(b)
0.4
0.2 ?
800 kPa
0.0 unsheared
hte
50
Mobilized Friction Angle (degrees)
rig
40
30 RS3
RS7
py
20 RS8
(c)
RS9
10
RS10
RS15
Co
0
1 10 100 1000 10,000 100,000
Shear Strain (%)
Figure 11.48 Ring shear test results of carbonate sand: (a) volumetric strain versus shear
strain, (b) the degree of particle breakage with shear strains, and (c) mobilized friction angle
versus shear strains (after Coop et al., 2004).
l
( pD, e D) 3. State pressure index (Wang et al., 2002)
ria
Ip = p/pc
pcL pcD Loose sand Ip = pL/pcL (>1)
Dense sand Ip = pD/pcD (<1)
Log (Mean pressure p)
ate
Figure 11.49 Various parameters that relate the current state to the critical state.
Void Ratio
Packing
To Zero
⫽ crit ⫹ 0.8 (11.35)
26 30 34 38 42
Porosity n (%)
where dilation angle is the ratio of volumetric strain
py
Figure 11.50 Contributions to shear strength of granular increment dv to the axial strain da at the stress state
soils (modified from Rowe, 1962). of interest. This is similar to Taylor’s equation (Eq.
(11.34)). However, in Eq. (11.34) changes with
shear, whereas crit is a constant material property.
Co
35
Triaxial Compression
30
l
20
0.1 0.2 0.5 1 2 5 10 20 50 100
ria
Effective Mean Pressure at Peak Failure (kPa)
Figure 11.51 Dependence of the friction angle of cohesion-
less soils on relative density and gradation (after Schmert- Figure 11.53 Effect of confining pressure on peak friction
mann, 1978). angle (after Yamamuro and Lade, 1996).
ate
To take effective confining pressure into account,
Bolton (1986) proposed the normalized dilatancy index
IR, defined as
冉冊
dM IR ⫽ Dr(Q ⫺ ln p) ⫺ R ⫽ Dr ln
c
p
⫺R (11.36)
the dilation angle in Eq. (11.35) vary with effective the following correlations for the peak friction angle
confining pressure; that is, Figs. 11.51 and 11.52 apply and critical state friction angle (in degrees) from the
for a particular value of confining pressure. In general, plots shown in Fig. 11.54.
the contribution of dilation increases with increasing
Co
Data for p ≈ 300 kPa ther details are given in Section 11.12.
l
of sands is important when flow slides or earthquakes
ria
are of concern. These events are very rapid, and rapid
4 deformation of loose to medium dense cohesionless
soils can generate excess pore water pressures resulting
in loss of strength or liquefaction. The stress-strain re-
0 lationship in undrained triaxial tests of Toyoura sand
ate
0.0 0.2 0.4 0.6 0.8 1.0
at different densities are shown in Fig. 11.57a, and the
Relative Density Dr corresponding effective stress paths are shown in Fig.
11.57b (Yoshimine et al., 1998). A sudden flow failure
Figure 11.54 Difference between peak friction angle and
critical state friction angle for triaxial compression and plain
can occur in loose sand deposits by the drop in strength
strain compression data on sands (after Bolton, 1986).
dM with increase in shear strain. Typical undrained re-
sponses of sand specimens at different densities are
illustrated in Fig. 11.58a.
Loose sand exhibits peak strength and then softens.
angle crit, as determined by Bolton (1986), is shown The peak state on the p –q plane is termed the collapse
in Fig. 11.55. The values shown are appropriate for surface (Sladen et al., 1985),15 and the slope increases
quartz sands (Q ⫽ 10). with increase in initial density and decrease in confin-
Other forms to characterize the peak friction angle ing pressure, as illustrated in Fig. 11.58b. In triaxial
hte
in relation to the initial state of a sand are available. compression, the slope for many sandy soils ranges
For example, Been and Jefferies (1986) relate the peak from 0.62 to 0.90 with an upper bound of 1.0 (Olson
friction angle to the state parameter defined in Fig. and Stark, 2003). Once the soil softens, large shear
11.49, as shown in Fig. 11.56. deformation is generated by moderate shear stresses.
The softened soil eventually leads to the steady state,
in which there is no further contraction tendency. The
rig
15
Similar concepts are proposed by others. For example, the critical
stress ratio (Vaid and Chern, 1985), the instability line (Lade, 1992),
Figure 11.55 Dilatancy component as a function of mean and the yield strength ratio (Olson and Stark, 2003).
effective stress at critical state and relative density (modified 16
The basic concept of the steady state is essentially the same as the
from Bolton, 1986). critical state defined for clay by Schofield and Wroth (1969).
l
ria
ate
Figure 11.56 Peak friction angle versus state parameter (after Been and Jefferies, 1986).
dM
Even dense sands exhibit positive excess pore pres- For loose sand, the steady state is the minimum un-
sures at the beginning of deformation up to small drained shear strength associated with a rapid collaps-
strain. However, after a certain stress ratio is reached, ing of soil structure. As discussed in Section 11.7, it
the undrained stress path reverses its direction indicat- has been suggested that the stress state of the steady
ing contractive to dilative behavior as shown in Fig. state is a function of void ratio, so a unique critical
11.58c, and the stress reversal is called the state of state line exists on the e–log p plane as shown in Fig.
hte
phase transformation (Ishihara et al., 1975). The 11.4b (Castro, 1975; Poulos et al., 1985, and others).
stress–strain response thereafter is strain hardening and The shape depends on grain angularity and fines con-
does not exhibit any peak. The soil eventually reaches tent (Zlatovic and Ishihara, 1995). At a given initial
the ultimate steady state or the critical state as long as void ratio, the steady state strength can be determined
the pore water does not cavitate. from the critical state line. For a relatively small con-
Medium dense specimens initially soften after the fining pressure, a small change in void ratio can give
rig
stress state passes the collapse line as illustrated in Fig. dramatic difference in undrained shear strength be-
11.58c. The stress state then reaches a point of mini- cause the critical state line on e–log p plane is very
mum strength, which is called the quasi-steady state flat at this stress level.
(Alarcon-Guzman et al., 1988) or flow with limited liq- For medium dense sand, the quasi-steady state can
uefaction (Ishihara, 1993). At this stage, the soil is in be considered as the minimum undrained shear
py
the state of phase transformation, and the mobilized strength. As the soil continues to deform, the shearing
strength then increases gradually with further shear resistance increases. Although the stress ratios at
strain due to increase in effective stress by negative quasi-steady state, and critical state are similar on the
pore water pressure development. As shearing contin- p –q plane, the quasi-steady state on e–log p plane
Co
ues, the soil shows a strain-hardening behavior, climb- lies below the critical state line as shown in Fig. 11.59.
ing along the critical state line, and the stress state For a given initial void ratio, therefore, the stress state
finally reaches the critical or ultimate steady state at of quasi-steady state is smaller than that of the critical
very large strains. Reported data indicate that the slope state.
of the critical state on the p –q plane is approximately The location of the quasi-steady state line on e–log
the same as that of the phase transformation line (Been p plane is influenced by shear mode and sample prep-
et al., 1991; Ishihara, 1993; Zhang and Garga, 1997; aration method (i.e., soil fabric) (Konrad, 1990; Ishi-
Vaid and Sivathayalan, 2000); at least, these lines are hara, 1993; Yoshimine and Ishihara, 1998). Figure
difficult to distinguish from each other. 11.60 shows the undrained shear behavior of Toyoura
l
ria
ate
dM
hte
rig
Figure 11.57 Undrained stress–strain response of Toyoura sand specimens prepared at dif-
ferent densities by dry pluviation (after Yoshimine et al., 1998).
py
sand in triaxial compression, triaxial extension, and marily due to the anisotropic soil fabric. Further details
Co
simple shear (Yoshimine et al., 1999). The specimens are given in Section 11.12.
were prepared to similar void ratios, and an initial con- The slope of the collapsing surface and the mini-
fining pressure of 100 kPa was applied. The minimum mum undrained strength are related to both initial
undrained shear strength and the quasi-steady state density and confining pressure. Typical values of the
vary significantly depending on the mode of shearing, collapse surface stress ratio obtained from triaxial
which in turn leads to different quasi-steady state lines compression tests are plotted against state parameter
on the e–p plane as shown in Fig. 11.61. Hence, large in Fig. 11.62 (Olson and Stark, 2003). Although the
variation of minimum undrained shear strengths is of- data are scattered, a general trend for a given sand is
ten observed depending on shear mode, which is pri- that the slope decreases with decreasing state param-
Steady State
Dilative Collapse Surface for
Liquefaction Very Loose Sand
Loose
Medium Dense
Quasi-steady
State Critical State Very Loose
Mean Pressure p
Steady State (b)
l
Loose
ria
Liquefaction (σ3 = 0)
Excess Pore Pressure Δu
ate
Collapse Surface for
Medium Dense Medium Dense Sand
Dense
Medium Dense
dM Dense
Mean Pressure p
Phase Transformation
(a) (c)
Figure 11.58 Typical undrained responses of sand specimens with different densities: (a)
stress–strain–pore pressure response, (b) stress paths for loose and very loose specimens,
and (c) stress paths for medium dense and dense specimens.
hte
0.90
Void Ratio e
l
ria
ate
dM Figure 11.61 Quasi-steady state line in triaxial compression,
triaxial extension, and simple shear on e–p plane (after
Yoshimine et al., 1999).
ce ⫽ hce
hte
(11.40)
ff
⫽ hc ⫹ ff tan e (11.41)
e e
tion 11.11. The friction angle of kaolin clays tends to
remain near the above values even at large strains. Reported values of hc and e range from 0.034 to
py
ff
proposed the following relationship to model the tan m ⬅ ⫽ hc e ⫹ tan e (11.42)
strength characteristics of overconsolidated clays: ff ff
冉 冊
cohesion intercept. These are called the Hvorslev pa-
q 6 sin e p
rameters. If e is constant, ce becomes a linear function ⫽ c cot ⫹ for triaxial compression
pe 3 ⫺ sin e pe pe
of e, which is the equivalent consolidation pressure
as determined from the void ratio at failure eff as shown
in Fig. 11.3.
q
⫽
6 sin e
冉c cot ⫹
pe 3 ⫹ sin e pe
p
pe冊 for triaxial extension
l
ria
ate
Figure 11.62 Relationship between the slope of the instability line and state parameter
(after Olson and Stark, 2003).
dM
hte
rig
py
Co
Figure 11.63 Relationship between the undrained shear strength ratio and state parameter
(after Olson and Stark, 2003).
l
ria
ate
Figure 11.64 Relationship between sin and plasticity index for normally consolidated
dM
soils (adapted from Kenny, 1959). Data for pure clays from Olsen, 1974.
where m is the mobilized friction angle at failure. For Undrained Shear Strength
normally consolidated clay, (e / ff) ⫽ 1 and tan m ⫽ Undrained shear strength su coupled with total stress
tan crit
⫽ hc ⫹ tan e. Substituting this into Eq. analysis [c ⫽ su and ⫽ 0 in Eq. (11.1)] is often used
(11.42) gives
hte
冉 冊
under undrained conditions. The undrained shear
e strength of saturated normally consolidated clay deter-
tan m ⫽ hc ⫺ 1 ⫹ tan crit (11.43)
ff mined using isotropically consolidated specimens as a
function of liquidity index is shown in Fig. 8.43.
Hence the peak friction angle m for overconsolidated The undrained strength for a given initial void ratio
rig
clays depends on the overconsolidation (e / ff) (the eini can be obtained using the critical state Eqs. (11.25)
first term in the right-hand side) and the critical state and (11.26):
friction angle crit. The form of this equation is similar
冉 冊
to Eqs. (11.34) and (11.35) derived for sands.
The Hvorslev parameters, e and ce, have been M % ⫺ 1 ⫺ eini
su ⫽
py
exp (11.44)
termed true friction angle and true cohesion, and are 2
considered by some to reflect the mechanism of shear
strength in terms of interparticle forces and friction. The above equation applies to both normally consoli-
Such an interpretation is questionable, however, be- dated and overconsolidated clays.
Co
cause, as shown in Chapter 8, two samples at the same For a given soil, the initial void ratio eini can be
water content but different effective stresses must have related to the current stress state and the overconsoli-
different structures. Thus, during deformation, there dation ratio. The relationship between undrained
will be differences in volume change under drained strength normalized by the effective overburden stress
conditions or differences in pore water pressure for de- after isotropic consolidation i has been deduced from
formation under undrained conditions. Furthermore, critical state soil mechanics by Wroth and Houlsby
Eq. (11.40) shows that ce is an effective stress- (1985) as
dependent property. Present evidence is that true co-
hesion is negligible in the absence of chemical bonding
between particles caused by cementation. su / i ⫽ 0.129 ⫹ 0.00435PI (11.45)
l
In Fig. 11.65, the undrained strength is normalized by (OCR)m (11.47)
v0
v0
ria
NC
ate
vertical preconsolidation stress vp to the current ver-
tical stress v. Typical values of m range between 0.7
and 0.9. This method is particularly well suited for use
with clays of low-to-medium sensitivity, that is, clays
that do not suffer large structural breakdown when
dM consolidated beyond their preconsolidation pressure. It
is also important to note that the strength ratio depends
largely on the mode of shearing as shown in Fig. 11.67
(Ladd, 1991) and the values given in Eqs. (11.45) and
(11.46) are at the lower range (or the conservative side)
in the figure. Further details of the effect of shearing
mode on undrained strength are given in Section 11.12.
hte
Figure 11.66 Normalized undrained strength ratio as a func- Figure 11.67 Effect of shear modes on undrained shear
tion of overconsolidation ratio (after Ladd et al., 1977). strength ratios of different plasticity soils (after Ladd, 1991).
l
ria
ate
(a) (b)
Figure 11.68 Shear bands observed in plane strain compression tests: axial strain of (a) 9.6
percent and (b) 19 percent (after Alshibli and Sture, 2000).
dM
‘‘shear band’’ formation in a direction diagonal to the
principal stress directions. Examples of shear bands
observed in the laboratory are shown in Fig. 11.68 (Al-
shibli and Sture, 2000). The deformation is localized,
and the two soil bodies on opposite sides of the shear
band act as rigid bodies. Strain localization tends to
hte
at the peak stress and crit is the critical state friction does not mean that more particles are involved in the
angle. Experimental data indicate that the shear band shear band. It is more likely that the local void ratio
direction is close to /4 ⫹ crit /2 for small diameter in the shear band is growing. Examination of resin-
particles (D50 ⫽ 0.2 mm), but the inclination decreases impregnated specimens with shear bands shows that
toward /4 ⫹ /2 for larger diameter particles (Oda this local void ratio is larger than the maximum void
and Iwashita, 1999). ratio for a stable load-carrying structure (Oda and Ka-
The thickness of the shear band depends on particle zama, 1998, Frost and Jang, 2000) and the very loose
size as shown in Fig. 11.69. It increases with increas- structure in the shear band is shown in Fig. 11.70 using
ing displacement, but then reaches a constant value X-ray CT. Discrete particle simulations by Iwashita
between 7 to 10 particle diameters when the displace- and Oda (1998) show that the very large void ratios
l
ized, and the residual state is developed within a shear
ria
band, as described in the previous section. The residual
Figure 11.70 Large local voids observed along the shear
zone using microfocus X-ray computed tomography (after strength-determining factors for a given test type and
Oda et al., 2004). strain rate are reduced to the effective stress, compo-
sition, and friction angle. The friction angle corre-
ate
sponding to this strength is the residual friction angle
r . Its value depends on the mineralogy, gradation,
inside the shear band (see Fig. 11. 41d) are associated bulky particle characteristics, and rate of shear. The
with particle rotation inside and near the shear band. relationships between stress and shear displacement for
The particles inside the shear band tend to rotate, soils with low and high clay fractions sheared under
whereas the particles outside the shear band remained
dM constant effective stress on the failure plane are shown
in their original positions. A high gradient of particle in Fig. 11.73. A residual condition can also be devel-
rotation is developed at the boundary of shear bands, oped under undrained conditions. In this case, the ef-
and the rotational resistance at this high rotational gra- fective stress on the shear plane at the residual state
dient zone was able to transfer loads even at void ratios will differ from that initially or at peak stress.
larger than the maximum void ratio. As a result of the rearrangement contributions to the
Possible situations where strain localization is likely residual strength of soils with low clay contents, as in
to occur are listed in Table 11.2. Fig. 11.73b, the loss of strength, under drained con-
hte
Distinct failure planes are often observed in both the ditions, between peak and residual is small. This is
laboratory and the field. Figure 11.72 is an X-ray pho- illustrated by Fig. 11.74, which shows the critical and
tograph of shear bands observed in a model of a re- residual friction angles for sand–bentonite mixtures
taining wall failure (Milligan, 1974 and interpretation tested in ring shear. Three zones are identified in Fig.
by Leśniewska and Mroz, 2000). Shear bands are ob-
11.74 and termed rolling shear, transitional shear, and
served in the field as slip planes. The practical impli-
rig
sliding shear.
cation of strain localization is that there can be
The shear displacements required to reach the resid-
significant differences between the continuum-based
ual strength can be large, as indicated by the values in
assumptions and analyses used in common geotech-
Table 11.3. Because the shear displacements required
nical design (such as friction angle and critical state)
to reach the residual state are large, the stability of
py
the scope of this book; however, other references are value that is influenced by the amount of progressive
available (Chambon et al., 1994; Vardoulakis and Su- failure along the shear surface.
lem, 1995). Owing to the large displacements required to de-
velop a full residual condition, special testing methods
have been developed. For example, a ring shear device
11.11 RESIDUAL STATE AND RESIDUAL was developed by Bishop et al. (1971), which can be
STRENGTH used to shear specimens through large displacements
that are always in the same direction. The values of
The drop in drained strength after peak is reached in residual friction angle shown in Fig. 11.28 were ob-
intact overconsolidated clay can be attributed to (1) an tained by shearing samples in direct shear back and
1. Dilative Material Subjected to Drained Shear Localization occurs after the deviator stress reaches its peak as the
soil dilates. This type of localization occurs in dense sand under low confinement and in heavily overconsolidated
clay. Some examples are given in Desrues et al. (1996) and Saada et al. (1999).
2. Contractive Material Subjected to Undrained Shear Shear of loose sand at high confinement causes softening
after the effective stress state passes the collapse line (see Section 11.8) due to generation of excess pore pressures.
In some cases, shear bands are hidden by bulging (Santamarina and Cho, 2003). Some examples are given in Finno
et al. (1998) and Mokni and Desrues (1999).
3. Dilative Material Subjected to Undrained Shear Cavitation Shear of dense sand at low confinement generates
l
large reductions in pore pressure. If the pore pressure becomes less than the vapor pressure (⫺100 kPa) of water,
ria
cavitation occurs. The effective stress drops at locations where cavitation occurs, and, therefore, the soil softens. Some
examples are given in Schrefler et al. (1996), Roger et al. (1998), and Mokni and Desrues (1999).
4. Alignment of Platy Particles If soil particles are platy, they align at a certain angle and this reduces the shearing
resistance. The residual friction angle discussed in Section 11.11 is a good example of this type of strain localization
ate
behavior.
5. Lightly Cemented Soils When cemented sands are sheared under low confinement, interparticle cementation breaks
at low strain levels, and the shear resistance drops. Examples are given in Santamarina and Cho (2003) for artificially
cemented soils and Cuccovillo and Coop (1999) for naturally structured sands.
dM
6. Unsaturated Soils For soils at low degree of saturation, menisci form at the particle contacts, and this increases
interparticle attractions by surface tension. When the soil is sheared, some of the menisci break, and this additional
contribution to strength is lost, at least temporarily, until new menisci are formed. The loss of menisci causes local
decrease in interparticle attraction, and, therefore, the soil may undergo softening.
7. Particle Breakage When particle breakage occurs, there is a change in particle size distribution, particle shape, and
textures. The collapse of soil structure by particle breakage leads to contractive behavior upon shearing.
hte
8. Heterogeneous Soil If a soil has a layer of loose material sandwiched between denser materials, strain localizes in
the loose layer. Microlayering is observed in many natural soils due to their depositional conditions. For moist tamped
compacted soil, thin looser zones exist between tamping layers, and these can initiate localized failure.
9. Other Cases The degree of localization can be influenced significantly by experimental conditions. Influencing
factors include nonuniform specimen shape, friction at end platens, high length-to-diameter specimens, and tilted
rig
platens. The occurrence of localization also depends on applied loading rate. Figure 11.71 shows failed samples of
kaolin clay after unconfined compression tests at two different loading rates (Atkinson, 2000). Sample A, which was
loaded slowly, exhibited strain localization due to local fluid migration in the dilating shear zone. Sample B, which
was loaded more rapidly, did not show distinct shear bands, as the pore fluid did not have time to migrate within the
specimen.
py
forth through displacements of 2 to 2.5 mm each side direct shear box to enable continued deformation al-
of center until minimum values were obtained. As Ta- ways in the same direction.
ble 11.3 indicates that many tens of millimeters in the
same direction may be required, some of the values in
Fig. 11.28 may be high, especially for the layer sili- Nonclay Minerals
cates, where the protrusion of particle edges across the The residual strength of nonclay minerals is not much
shear surface could give increased resistance. A pre- different than the critical state strength, as noted above.
ferred method of testing, if a ring shear device is not Quartz, feldspar, and calcite all have the same value of
available, is to separate and reset the two halves of a r ⫽ 35, as shown in Fig. 11.28, even though the
l
ria
Figure 11.71 Samples of kaolin clay after unconfined com-
pression tests: (a) slower loading and (b) faster loading (cour-
tesy of J. H. Atkinson, 2000). Figure 11.73 Stress–shear displacement curves under con-
ate
stant effective normal stress on the shear plane (Skempton,
1985): (a) high clay fraction (⬎40 percent and (b) low clay
fraction (⬍20 percent).
dM
hte
rig
py
Co
(a) (b)
Figure 11.72 Shear bands observed in centrifuge modeling of retaining wall failure: (a) X-
ray photograph (after Milligan, 1974) and (b) interpretation of the photography by Lés-
niewska and Mroz (2000).
l
available for certain soil types (Mesri and Cepeda-
ria
Diaz, 1986; Colotta et al., 1989; Stark and Edit, 1994).
Highly plastic soils of volcanic origin are an excep-
tion to the general relationship shown in Fig. 11.75.
Figure 11.74 Influence of clay fraction on the peak and re-
These soils, which may have clay size fractions well
sidual friction angles for sand–bentonite mixtures as deter-
ate
mined by ring shear tests (after Lupini et al., 1981). above 50 percent, exhibit residual friction angles that
are several degrees higher than those shown in the fig-
ure. Both particle morphology and structural factors
have been suggested as possible causes (Sitar, 1991;
Table 11.3 Displacements Corresponding to Various Wesley, 1992). As volcanic clays often contain large
Stages of Shear in Clays with a Clay Size Fraction
dM amounts of allophane, which consists of bulky shaped
Greater Than 30 Percenta particles rather than platy particles, rolling shear may
continue to make a major contribution. Alternatively,
Displacement (mm) or in addition, physicochemical attractive forces be-
tween particles may be sufficiently strong to prevent
Normally
the development of parallel orientations of platy par-
Stage Overconsolidated Consolidated
ticles and basal plane shear. For soils with liquid limits
Peak 0.5–3 3–6 more than 50, Wesley (2003) shows that the position
hte
Volume change rate ⬇ 0 4–10 on the plasticity chart in relation to the A-line [i.e.,
At r ⫹ 1 30–200
PI ⫽ PI ⫺ 0.73(LL ⫺ 20)] is a good indicator to give
Residual r 100–500 good correlations for residual friction angle for both
clays and volcanic ashes as shown in Fig. 11.76.
a
Intact clays with ⬍ 600 kPa.
Data from Skempton (1985).
rig
l
a massive mineral than a platy mineral in shear
ria
(Chattopadhyay, 1972).
For many clays, the residual friction angle decreases
with increasing confining pressure, that is, the failure
envelopes are curved. The values in Fig. 11.28 for sev-
ate
Figure 11.76 Residual friction angle of volcanic clays and eral clay minerals, as well as the data for brown Lon-
clays in relation to their location on the plasticity chart rel- don clay and Weald clay in Fig. 11.2, show significant
ative to the A-line (from Wesley, 2003). stress dependency of the residual friction angle, that
is, the failure envelope is curved. One possible reason
for this curvature is that under low normal stresses less
dM work is required to shear the clay in the absence of
Clay Minerals perfect orientation of clay particles in the shear zone
Basal plane slip is the dominant deformation mecha- than would be required to develop parallelism during
nism at large strain in the clay minerals and other layer shear.
silicates. Compression textures with basal planes ap- An alternative explanation of the stress dependency
proximately perpendicular to the normal load direction of r for clays can be based on the elastic junction
are formed in the shear zone, and most of the defor- theory developed in Section 11.4. If deformations at
hte
mation takes place in this zone as well as in zones of particle contacts are elastic, then the area of real con-
high particle orientation that enclose it. The behavior tact between sliding surfaces increases less than pro-
Table 11.4 Bonding Along Cleavage Planes, Cleavage Mode, and Residual Strength
rig
r
Mineral Mode of Cleavage Bonding Along Cleavage Planes (deg) Particle Shape
Quartz No definite cleavage 35 Bulky
py
portionally with increase in normal effective stress; (1990) show from their database of past studies that
and according to Eq. (11.22), tan r should vary as the friction angle deduced from triaxial extension tests
(n)⫺1 / 3. Data for several clays are shown in Fig. 11.77, is 10 to 20 percent larger than that from triaxial com-
which show agreement with this theory in the low- pression tests. Similar differences have been observed
pressure range (up to 200 kPa), but at higher stresses, between plane strain friction angle and triaxial com-
r is independent of stress, indicating that the solid pression friction angle for both sands and clays. The
contact area varies in direct proportion to effective nor- peak friction angles of sands determined from plane
mal stress. strain tests are 0.5 to 4 larger than those from triaxial
Both hypotheses appear tenable, and evidence is not compression tests (Cornforth, 1964). The Mohr–
available to favor one over the other. Nonetheless, for Coulomb friction angle is based only on the major and
practical purposes it is clear that determinations of val- minor principal stresses [i.e., Eq. (11.5)]. Under tri-
l
ues of residual strength to be used for analysis of spe- axial compression conditions, the intermediate prin-
ria
cific problems should be made under stress conditions cipal stress, 2, equals the minor principal stress,
approximating those in the field. whereas under plane strain loading it is greater. The
Other important considerations include the structural higher confinement under plane strain loading can ac-
features and lithological details in the field. Examples count for the higher measured friction angle.
of the former are presheared surfaces generated by The stress states at failure in the field differ from
ate
old landslides and tectonic and glacial deformation, triaxial compression/extension and plane strain con-
whereas those for the latter are horizontal bedding ditions. The effect of intermediate principal stress can
planes, laminations, and weak seams (Mesri and Shah- be expressed by a b value, defined by
ien, 2003). They all can contribute in dropping to the
residual condition after relatively small displacements.
dM 2 ⫺ 3
b⫽ (11.48)
1 ⫺ 3
11.12 INTERMEDIATE STRESS EFFECTS AND
ANISOTROPY
where 1, 2, and 3 are maximum, intermediate, and
The friction angles shown in Fig. 11.51 and Fig. 11.52 minor principal stresses, respectively; b ⫽ 0 for triaxial
are for triaxial compression. Kulhawy and Mayne compression conditions (1 ⬎ 2 ⫽ 3), whereas b ⫽
hte
rig
py
Co
Figure 11.77 Residual friction angle as a function of normal effective stress on the shear
plane raised to the minus one-third power (replotted from data in Chattopadhyay, 1972).
l
compression) to b ⫽ 0.3 to 0.4 (plane strain), and it
ria
higher than measured in triaxial compression. The
stays constant or slightly decreases as b reaches 1 (tri- measured friction angle in triaxial extension is about
axial extension). 20 percent greater than in triaxial compression, as
The variation of measured friction angle with shown in Fig. 11.80. These results are consistent with
changes in intermediate principal stress can be attrib- the data presented in Fig. 11.78 for sands.
ate
uted to the effects of different mean stress and stress The undrained shear strength su in triaxial compres-
anisotropy on the dilatancy and particle rearrangement sion is approximately twice as large as that of triaxial
contributions to the total strength. For given maximum extension for normally consolidated clay as shown in
and minimum principal stresses, the triaxial extension Fig. 11.67. This large variation is primarily due to the
conditions have the largest mean effective stress, difference in the excess pore pressures generated and
whereas the triaxial compression conditions have the is very much related to the initial bedding structure, as
dM
smallest mean effective stress. The higher confinement discussed later.
for triaxial extension and plane strain conditions con-
tributes to the increasing friction angle for these con- Failure Envelopes
ditions.
Various models fit the experimental data showing the
Fabric anisotropy also contributes to differences be-
intermediate stress effect. Among them are
tween triaxial and plane strain strengths. Discrete par-
hte
60
Friction Angle
30
Triaxial compression 20°
20
l
Matsuoka and Nakai (1985)
ria
0
0 0.2 0.4 0.6 0.8 1
b-Value
Triaxial Plane Strain Triaxial
ate
Compression Extension
invariants.18 The models are plotted in Fig. 11.81. Mat- Positive loading in triaxial compression tests is usu-
suoka and Nakai’s model gives the same friction angle ally in the direction perpendicular to the bedding plane,
for compression and extension, whereas Lade and whereas that in triaxial extension tests tends to be ra-
Duncan’s model gives the ratio of the triaxial extension dially inward in the bedding plane direction. Figure
friction angle (TE) to the triaxial compression friction 11.82 shows the undrained shearing responses of Toy-
py
angle (TC) to be 1.08 at TC ⫽ 20 to 1.15 at TC ⫽ oura sand specimens in triaxial compression and ex-
40. Given the large scatter in the published experi- tension tests (Yoshimine et al., 1998). The specimens
mental data (see Fig. 11.78), it is not possible to con- sheared in extension showed 100 percent excess pore
clude that one model is better than the other. pressure development leading to static liquefaction,
Co
l
potential or undrained shear strength.
ria
11.13 RESISTANCE TO CYCLIC LOADING AND
LIQUEFACTION
ate
Repeated or cyclic loading of soils can be caused by
a number of natural phenomena or human activities,
including earthquakes, wind, waves, vehicular traffic,
and reciprocating machinery. Cyclic loads cause
dM stresses and deformations in much the same manner as
do slowly applied loads; however, their relatively short
duration and repetitive and dynamic nature are respon-
sible for several unique aspects of soil behavior. At-
tention is given in this section primarily to saturated
cohesionless soils and clays because these materials
are particularly susceptible to strength degradation
and/or failure during earthquakes. Soils of the type
hte
axial compression will be case A (b ⫽ 0 and ⫽ 0) ures 11.85(a-2) and 11.85(b-2) show development of
in Fig. 11.83, whereas the triaxial extension will be volumetric strain with increasing number of loading
case B (b ⫽ 1 and ⫽ 90) in Fig. 11.83. Ideally, cycles. For the loose sand specimen (Fig. 11.85a), the
therefore, the investigation of b-value effects on volumetric strain increases more or less monotonically.
Co
strength should be performed with the same value Although there is a decreasing trend of void ratio with
or vice versa (e.g., cases A, C, and D or cases C, E, increasing number of cycles, it is also interesting to
and F in Fig. 11.83). note that, for a given cyclic shear application on the
Using a hollow cylinder torsional shear apparatus, dense sand specimen, the volumetric strain does not
Yoshimine et al. (1998) examined the effects of b value increase monotonically but fluctuates with cyclic load-
and the loading direction separately on undrained ing, as illustrated in Fig. 11.85(b-2). Shear-induced di-
behavior of Toyoura sand, and some test results are lation is observed as the applied shear displacement
shown in Fig. 11.84. For sands with relative density reaches its maximum in the loading stage. During un-
between 30 and 41 percent, the effect of inherent an- loading, there is shear-induced volume contraction.
isotropy appears to be more significant than the effect The combination of dilation during loading and con-
of b. When the loading direction was in parallel to the traction during unloading leads to overall contraction.
σv σv
σh1 σh1
σσh2 σh1
σh2
σh1v> σh2 > σv σh1v= σh2 > σv
α = 90° v
F B
Parallel to the
Bedding Plane
α
σ11
Loading Direction
l
of Major Principal α
ria
Stress in Relation σσ3
to the Bedding σσ22
E
Plane σ1v> σ2 > σ3
ate
σv σv σv
v
σh1 σh1
σh1
Perpendicular to σh2 σh2 σh2
the Bedding Plane
dM σvv> σh1 = σh2 σvv> σh1 > σh2 σvv= σh1 > σh2
α = 0°
C D
0 A 1
Triaxial b - Value Triaxial
Compression Extension
Figure 11.83 Effects of inherent anisotropy and the intermediate stress.
hte
The influences of shear strain magnitude and num- ment of shear strains with cyclic loading is called cy-
ber of load cycles are shown in Fig. 11.86. The den- clic mobility. Liquefaction is said to have occurred
sification results from the disruption of the initial soil when the pore water pressure has increased to the mag-
fabric caused by the repeated shear strains followed by nitude of the initial effective confining pressure, at
rig
repositioning of the soil grains into more efficient which point the strains become very large. Similar to
packing. The higher the initial void ratio and the the undrained response in monotonic loading, the un-
greater the number of cycles, the greater the effect. drained response of sand under cyclic loading depends
on density, confining pressure, and soil fabric. The ef-
Undrained Behavior fect of density on cyclic behavior of Toyoura sand un-
py
When saturated soil is subjected to repeated cycles of der triaxial loading conditions is shown in Fig. 11.88.
loading, and provided the shear stresses are of suffi- All sands exhibit increase in pore pressure with in-
cient magnitude, the structure begins to break down, crease in number of loading cycles, but the shear strain
and part of the confining stress is transferred to the development for a given number of cycle is smaller for
Co
pore water, with a concurrent reduction of effective denser specimens. In the loose sand (Fig. 11.88a),
stress and strength. This, in turn, leads to increase in when the stress state reaches the collapse surface, the
shear strain under constant stress cyclic loading or a soil softens leading to sudden liquefaction. The me-
decrease in the cyclic stress required to cause a given dium dense sand (Fig. 11.88b) exhibits quasi-steady
value of cyclic strain. The deformation and failure be- state as the stress state reaches the phase transforma-
havior of sands in undrained cyclic loading depends on tion line. Some cycles with large stress–strain loops
initial void ratio, initial effective stress state, and the are observed and the specimen finally reaches lique-
cyclic shear stress amplitude. faction. The dense sand (Fig. 11.88c) never liquefies.
The results of an undrained cyclic simple shear test Once the stress state reaches the phase transformation
on Monterey sand are shown in Fig. 11.87. Develop- line, the stress–strain curve moves back and forth
l
ria
ate
dM
hte
Figure 11.84 Effect of and b values on undrained response of dry plluviated Toyoura
sand: (a) effect of when b ⫽ 0.5 and (b) effect of b when ⫽ 45 (after Yoshimine et
al., 1998).
rig
along and below the steady-state line and shear strain between cyclic shear stress and number of load cycles
develops gradually. to initial liquefaction depends on the relative density,
Beneath gently sloping to flat ground, liquefaction and is of the form shown in Fig. 11.89. In this figure
py
may lead to ground oscillation or lateral spread as a the cyclic shear stress applied by a simple shear ap-
consequence of either flow deformation or cyclic mo- paratus is normalized by dividing by the initial effec-
bility (Youd et al., 2001). The liquefaction suscepti- tive confining pressure 0, and the ratio is often called
bility of different types of natural and artificial the cyclic resistance ratio (CRR). Methods for deter-
Co
sedimentary soil deposits is summarized in Table 11.5. mination of the liquefaction susceptibility of a specific
As the excess pore pressure developed during lique- site are given by Kramer (1996) and by Youd et al.
faction dissipates, ground settlement is observed. Sand (2001).
boils can develop through overlying less permeable In reality, generation of pore pressure is a result of
soils in order to dissipate the excess pore pressures the breakdown of soil structure and a tendency for the
from liquefied soil below. soil to densify, and this is caused by shear deforma-
The magnitude of the cyclic shear strains that de- tions, so liquefaction is more fundamentally controlled
velop following initial liquefaction decreases with in- by shear strain than by shear stress. Furthermore, there
creasing initial relative density and increases with is a level of shear strain, or threshold shear strain be-
increasing cyclic shear stress. The general relationship low which no pore pressure is generated. This is illus-
_2 2 _2 2
Shear Strain γ (%) Shear Strain γ (%)
_1.2 _1.2
l
ria
(a-1) Stress Ratio – Shear Strain Relation (b-1) Stress Ratio – Shear Strain Relation
Volumetric Strain εv
Volumetric Strain εv
0.6
3
ate
_2 0 2
(a) (b)
Figure 11.85 Cyclic behavior of Toyoura sand in drained conditions: (a) loose sand and (b)
dense sand (after Pradhan and Tatsuoka, 1989).
hte
l
methods, and the mode of shearing (Seed, 1979; Seed
ria
and Harder, 1990). Additional information and data
can be found in Youd et al. (2001), Vaid et al. (2001),
Boulanger (2003), and Hosono and Yoshimine (2004).
ate
Residual Strength after Liquefaction
The residual strength of sands, silty sands, and silts
following liquefaction is a subject of continuing study
owing to its importance in the analysis of postearth-
dM quake stability and deformation of embankments,
dams, and structures. Detailed discussion of this topic
is outside the scope of this book; however, two ap-
proaches have been used to estimate the residual
strength, one based on steady state strength determined
by laboratory tests as described in Section 11.8 and the
other on the Standard Penetration Test (SPT) N value
(Seed, 1987; Seed and Harder, 1990). A correlation
hte
umetric strain that would occur if drainage were per- example, if there is a less permeable layer above a sand
mitted. Martin et al. (1975) give procedures to evaluate layer, excess pore pressures can develop under the
these two parameters from the results of static rebound impermeable layer leading to softening of the soil.
tests in a consolidation ring and cyclic load tests on Hence, local heterogeneity plays an important role in
dry sand, respectively. Finn (1981) reported good liquefaction-induced soil deformation and failure,
agreement between predicted and measured values us- which requires careful site investigation to identify any
ing the proposed method. low permeable layers.
The liquefaction resistance depends not only on cy- The strength degradation of clays due to cyclic load-
clic stress amplitude and density but also on the initial ing follows similar patterns to that of sand, but it is
effective stress state. For example, Fig. 11.89 is deter- much less for clays than for cohesionless and slightly
50 100
Dr = 30%
40 Initial Cyclic Loops 80 Reaching Collapse Surface
Initial p = 200 kPa
Before Failure after Several Cycles
Cyclic Stress Δq = 40 kPa 30 60
Deviator Stress q (MPa)
l
-30 -60
Surface after Several Initial State
ria
Liquefaction
Cycles -40 -80
-50 -100
Axial Strain (%) Mean Pressure p(MPa)
(a)
ate
100 100
Dr = 50% Phase Transformation Phase Transformation
80 80 Initial State
Initial p = 200 kPa
60 60
Cyclic Stress Δq = 60 kPa
Deviator Stress q (MPa)
(b)
50 50
rig
Increasing
20 Cycles 20
10 10
py
0 0
-15 -10 -5 -10 0 5 10 15 -10 0 20 40 60 80 100 120
Phase Transformation
-50 -50
Axial Strain (%) Mean Pressure p(MPa)
(c)
Figure 11.88 Cyclic behavior of Toyoura sand in undrained conditions: (a) loose sand, (b)
medium dense sand, and (c) dense sand (after Yamamoto, 1998).
l
Marine terraces and Widespread — Low Very low Very low
ria
plains
Delta and fan-delta Widespread High Moderate Low Very low
Lacustrine and playa Variable High Moderate Low Very low
Colluvium Variable High Moderate Low Very low
ate
Talus Widespread Low Low Very low Very low
Dunes Widespread High Moderate Low Very low
Loess Variable High High High Unknown
Glacial till Variable Low Low Very low Very low
Tuff Rare Low Low Very low Very low
Tephra dM Widespread High High ? ?
Residual soils Rare Low Low Very low Very low
Sabka Locally variable High Moderate Low Very low
b. Coastal Zone
Delta Widespread Very high High Low Very low
Esturine Locally variable High Moderate Low Very low
hte
Beach high wave energy Widespread Moderate Low Very low Very low
Low wave energy Widespread High Moderate Low Very low
Lagoonal Locally variable High Moderate Low Very low
Fore shore Locally variable High Moderate Low Very low
c. Artificial
rig
cohesive soils that are susceptible to liquefaction as triaxial tests (Fig. 11.96b), the initial static shear is
Co
shown in Fig. 11.95 (Hyodo et al., 1994). An assump- defined as a /su ⫽ (ac ⫺ rc )/2su, where ac and rc
tion of a strength loss of about 20 percent is sometimes are the axial and radial consolidation stresses, respec-
used in practice. Figure 11.96 shows the undrained cy- tively. The values of cy /su ⫽
(a ⫺ r)/2su show
clic shear stress ratio cy /su that brings normally con- peaks at a /su ⫽ 0.2 to 0.3, indicating that the small
solidated clays to failure after 10 loading cycles initial anisotropy gave increased cyclic resistance.
(Andersen, 2004). The data include eight clays with However, the undrained cyclic shear stress ratio at fail-
different plasticity indices. In the direct shear tests ure decreases when the initial static shear stress is
(Fig. 11.96a), the undrained cyclic shear stress ratio at higher or in triaxial extension conditions. Evidently in
failure decreases with increase in initial static shear normal clays the magnitude of cyclic shear strain is
stress a /su and increases with plasticity index. In the less than that required to cause complete remolding.
l
ria
Figure 11.89 Cyclic stress ratio and number of load cycles to cause initial liquefaction of
ate
a sand at different initial relative densities (from De Alba et al., 1976). Reprinted with
permission of ASCE.
dM
hte
rig
Cyclic stresses could cause sufficient deformations in Figure 11.91 Mechanism of pore pressure generation during
quick clay to initiate a liquefaction-type flow failure. cyclic loading (Seed and Idriss, 1982).
Some examples are given in Andersen et al. (1988).
11.14 STRENGTH OF MIXED SOILS preparation methods. Figure 11.97 shows different sce-
narios of intergranular matrix of two different size par-
The presence of fines in sands can significantly influ- ticles (Thevanayangam and Martin, 2002). Initially the
ence the strength behavior. Differing effects can be ob- maximum and minimum void ratios of a sand–silt mix-
tained depending on particle size, shapes, and sample ture decrease with increase in silt content, but then the
l
shear tests (from Seed et al., 1976). Reprinted with permis- Figure 11.95 Comparison of the cyclic resistance ratios of
ria
sion of ASCE. Itukaichi clay and Toyoura sand (Hyodo et al., 1994).
ate
Cyclic Shear Strength / Static Undrained Shear Strength
Direct Simple Shear Tests - Strength at 10 Cycles
dM OCR = 1
1.2
Offshore Africa, PI=80 -100%
Troll I, PI=37%
0.8
(τcy / suDSS)
Drammen, PI=27%
Figure 11.93 Effect of confining pressure on cyclic resis-
tance ratio (Hyodo et al., 2002). 0.4
North Sea GC, PI=16-27%
0.2
Storebælt, PI=7-12%
0.0
rig
0.4
Drammen, PI=27%
0.2
Storebælt, PI=7-10%
0.0
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
Initial Shear Stress / Static Undrained Shear Strength
(τa / s u)
Figure 11.94 Postliquefaction residual strength as a function Figure 11.96 Normalized shear stresses that give undrained
of Standard Penetration Test N values (Seed and Harder, failure after 10 cycles in (a) direct shear tests and (b) triaxial
1990). tests (Andersen, 2004).
l
ria
ate
dM
Figure 11.97 Granular mix classification (Thevanayangam and Martin, 2002).
hte
void ratios increase when the silt becomes the host soil fines content have the same granular void ratio and the
as shown in Fig. 4.4. In case (i), the fine particles fit same mechanical properties, the fines are just occu-
in the void space formed by the coarse particles. The pying the void space and are not influencing shear re-
mechanical behavior is little affected by the presence sistance.
of fines because the external forces are transferred Most reported cases show that, for a given granular
through the contacts between coarse particles. In cases void ratio, the undrained strength and cyclic shear re-
rig
(ii) and (iii), the fine particles start to fully occupy sistance are either independent of or increase with silt
some void space and separate the coarse particles and content (Shen et al., 1977; Vaid, 1994; Polito and
prevent them from touching each other. These fine par- Martin, 2001; Carraro et al., 2003). The undrained
ticles may reinforce the skeleton of coarse particles or response of sand mixed with equidimensional silt
they may make the skeleton unstable. As the propor- particles is shown in Fig. 11.98 (Kuerbis et al., 1988).
py
tion of fine particles increases, the coarse particles float Specimens of the mixture were created by slurry dep-
inside the matrix of fine particles as illustrated as case osition, and the density was controlled in such a way
(iv). The fine grains then dominate the mechanical be- that all specimens had relatively similar granular void
havior of the mixed soils, and the coarse grains may ratios eG, even though the actual void ratio decreased
Co
or may not contribute to shear resistance as a reinforc- with increasing silt content. Both undrained triaxial
ing element. Once the mixing scenario reaches case compression and extension tests were performed fol-
(iv), the void ratio increases with increasing fines con- lowing isotropic consolidation. Increased silt content
tent due to increasing specific surface of the mixture. gave stiffer response in triaxial compression. Appar-
The threshold value to become case (iv) depends on ently, the silts filled the void space and stabilized the
the specific mixture but is usually in the range of 25 soil as shown in Fig. 11.99a. However, the effect was
to 45 percent fines in most cases (Polito and Martin, small in triaxial extension.
2001). Liquefaction resistance increases with relative den-
For cases (i) to (iii), the granular void ratio eG de- sity as shown in Fig. 11.89. However, increasing silt
fined in Chapter 4 is a useful index for consideration content gives scattered relationships between relative
of the effect of fines. If two mixed soils with different density and the CRR at 20 loading cycles, as shown in
100 0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Relative Density
l
0
100 200 300 400
(a)
ria
(σa + r)/2 (kPa) 0.8
0% fines
ate
tests on silty sand with different mixing ratios (Kuerbis et 0.4
al., 1988).
0.2
dM 0.0
0.8 0.7 0.6 0.5 0.4 0.3
Void Ratio
(b)
0.8
0% fines
Cyclic Resistance Ratio
5% fines
0.6 10% fines
15% fines
hte
0.4
(a) (b)
0.2
0.0
rig
Figure 11.99 Schematic diagrams of how fine particles are increasing silt content. If the CRR values are plotted
Co
achieved if some fines are placed between the coarser bridged between the host sand particles (Fig. 11.99d),
particles as shown in Fig. 11.99b. In this case, the and increased the overall void ratio as shown in Fig.
structure is metastable, and the strength of the mixed 11.102. On the other hand, inclusion of smaller silt and
soil is reduced due to fewer sand grain contacts. clay particles decreased the overall void ratio, as also
When smaller particles such as clays are added in- shown in Fig. 4.4. The open fabric of a sand–mica
stead of silt-size particles, the clay fines act as a lu- mixture can give complicated soil deformation and
bricator at sand particle contacts as shown in Fig. strength properties depending on mica particle orien-
11.99c and make the soil unstable. Undrained re- tation and shear mode (Hight et al., 1998).
sponses of Ham river sand mixed with different kaolin Further increase in fines content leads to sand par-
contents are shown in Fig. 11.101 (Georgiannou et al., ticles floating in clay or silt as shown by case (iv) in
1991). Samples were prepared by pluviating the sand Fig. 11.97. The mixed soil then behaves more like pure
l
into distilled water with suspended kaolin particles so clay or silt. The deformation behavior then becomes
ria
that similar granular void ratios were achieved. Both more clay/silt dominated, and the coarser particles
undrained triaxial compression and triaxial extension may or may not contribute to the strength properties.
tests were performed after consolidating the samples For example, Fig. 11.103 shows that the liquefaction
under K0 stress conditions. In triaxial compression, the resistance of mixtures with fines content greater than
increase in clay content did not affect the peak stress, 35 percent was independent of silt content and granular
ate
but the strain-softening behavior was more pro- void ratio (Polito and Martin, 2001).
nounced. After the specimen passed the phase trans-
formation line, the stress increased toward the critical
state. In triaxial extension, addition of clay led to total 11.15 COHESION
liquefaction. The friction angle did not change for clay
dM
fractions up to 20 percent. This delayed the develop- True cohesion is shear strength in excess of that gen-
ment of anisotropic fabric needed to resist the increas- erated by frictional resistance to sliding between par-
ing load. ticles, the rearrangement of particles, and particle
The shape of fine particles also influences the sta- crushing. That is, true cohesion must result from ad-
bility of the mixed soil. Hight et al. (1998) report the herence between particles in the absence of any exter-
behavior of micaceous sands in connection with flow nally applied or self-weight forces. The existence of
slides that occurred during construction of the Jamuna tensile or shear strength in the absence of effective
hte
Bridge in Bangladesh. The large and platy mica flakes compressive stress in the soil skeleton or on the failure
plane might be considered true cohesion. However, the
particulate nature of soil and the fact that most inter-
particle contacts are not oriented in the plane of shear
(MPa) mean that the application of directional shear stress
0.3 Triaxial Compression Tests will induce normal forces at most interparticle con-
rig
Clay Content = 0%, eG = 0.77 tacts. These forces will, in turn, generate a resistance
Clay Content = 4.6%, eG = 0.80
Clay Content = 7.6%, eG = 0.80
0.2
py
Initial Stress
(σ’a+ σ’ r)/2
State
0.1
(σ’a+ σ’r)/2
Co
0.0 0.5
0.1 0.2 0.3 0.4
(MPa)
Triaxial Extension Tests
Clay Content = 0%, e G = 0.77
-0.1 Clay Content = 3.5%, e G = 0.80
Clay Content = 7.5%, e G = 0.80
䊉
0.15 ×
夽 夽
0.10 䊉 䉬䉬 䊏 × 䉱
䊉 × × 夽
l
ria
0.05
0.00
0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20
ate
Granular Void Ratio eG
Figure 11.103 Variation of cyclic resistance ratio with granular void ratio with silt content
above the threshold value (Polito and Martin, 2001).
dM
to sliding at the contact provided the value of is solution–redeposition processes, or they may be
greater than zero. taken from solution. An analysis of the strength
Confirmation of the existence of a true cohesion and of cemented bonds was given by Ingles (1962)
determination of its value from strength tests is diffi- and is summarized in Section 7.4 and Eqs. (7.2)
cult because projection of the failure envelope back to to (7.8). Cohesive strengths of as much as several
⫽ 0 is uncertain, owing to the curvature of most hundred kilopascals (several tens of pounds per
hte
failure envelopes, unless tests are done at very low square inch) may result from cementation.
effective stresses. Tensile tests cannot be made on most Stress–strain curves and peak failure envelopes
soils. Harison et al. (1994) performed various types of for cemented sands are shown in Fig. 11.104.
tensile tests on compacted clay specimens but found These curves show that even relatively small
that the tensile strengths decreased with increase in amounts of cement can have very large effects on
specimen size due to increase in the number of internal the deformation properties. Small values of co-
rig
flaws. There is no convenient way to run a triaxial hesion have a large effect on the stability of a
compression test while maintaining the effective stress soil and its ability to stand unsupported on steep
equal to zero on the potential failure plane. Strength slopes. However, at large strains when the ce-
can be measured by direct shear with no applied nor- mentation breaks down, the strengths become
mal stress . Some examples are given in Bishop and similar irrespective of the degree of cementation
py
Garga (1969), Graham and Au (1985), and Morris et as shown in Fig. 11.104a.
al. (1992); however, for the reason given in the previ- 2. Electrostatic and Electromagnetic Attractions
ous paragraph, the measured strength cannot be attrib- Electrostatic and electromagnetic attractions be-
uted specifically to true cohesion. tween small particles are discussed in Sections
Co
l
ria
ate
Figure 11.104 Stress–strain curves and failure envelopes for cemented and uncemented sand
at a relative density of 74 percent: (a) stress–strain curves and (b) failure envelopes based
on peak strength (from Clough et al., 1981). Reprinted with permission of ASCE.
dM
stress reduction, but a part is retained as shown stresses between particles generated by internal attrac-
in Fig. 11.3. Whether or not the higher strength tive forces. The mechanism of shear resistance result-
in the overconsolidated clay is because of the ing from these attractions should be the same as if the
lower void ratio or due to the formation of inter- contact normal stresses were derived from effective
hte
particle bonds is not known. However, a ‘‘cold compression stresses carried by the soil. It is conven-
welding’’ or adhesion may be responsible for ient, therefore, to think of cohesion (except for cemen-
some of it. This could result from the formation tation) as due to interparticle friction derived from
of primary valence bonds at interparticle con- interparticle attractions, whereas the friction term in
tacts. the Mohr–Coulomb equation is developed by interpar-
ticle friction caused by applied stresses. Essentially the
rig
tween particles in a partly saturated soil. Equation (7.9) both applied stresses and interparticle forces contrib-
can be used to estimate the magnitude of tensile uting to the effective stress that developed the frictional
strength that can be developed by capillary stresses in resistance. Present evidence indicates that cohesion
a soil. This is not a true cohesion; instead, it is a fric- due to interparticle attractive forces is quite small in
almost all cases, whereas that due to chemical cemen-
Co
Summary
11.16 FRACTURING OF SOILS
Several contributions to cohesion are summarized in
Fig. 11.105 in terms of the potential tensile strengths Soil fracturing is often observed in geotechnical prac-
that can be generated by each mechanism as a function tice. Tensile cracks develop when there is external ten-
of particle size. For all the mechanisms except chem- sion stress such as at the crest of a landslide or vertical
ical cementation, cohesion is a consequence of normal cuttings. In some cases, water can fill the cracks, lead-
l
ria
ate
Figure 11.105 Potential contributions of several bonding mechanisms to soil strength
dM
(Ingles, 1962).
ing to further instability. Soil piping can occur in a dam principal effective stress is equal to the negative value
from water flow through cracks causing internal ero- of the tensile strength (t ).19 This criterion can be writ-
sion. Hydraulic fracturing results from increase in the ten as
pressure at the crack tips. Hydraulic fractures can be
hte
ment and contaminated ground remediation. Desicca- sheared in undrained conditions and the soil cracks if
tion also causes the development of tensile cracks as Eq. (11.52) is satisfied. The tensile total stress 3 then
the suction in the soil increases by evaporation and becomes
causes shrinkage of the soil by increase in effective
stresses.
py
rapid increase in external pressure or in sands and where u0 is the initial pore pressure,
u is the excess
clays by fluid permeation. Various mechanisms for pore pressure generated during the shearing process
fracture initiation are described below. leading to fracture, A is Skempton’s pore pressure pa-
rameter (Section 8.9), and
1 and
3 are the changes
Fracture under Undrained Conditions
If particle contacts cannot carry tension, it is often as-
sumed that the tensile cracking occurs when the minor 19
Note that the tensile strength of a soil is defined in terms of effec-
principal effective stress 3 becomes zero. If the soil tive stress. Unfortunately, many tensile strength values are written in
is cemented, cracking is generated when the minor total stress since pore pressure is not measured.
in major and minor principal total stresses, respec- increases, but the circumferential stress initially de-
tively.20 creases as long as the soil behaves linear elastically
Rearrangement of Eq. (11.53) gives and does not fail in shear (see Fig. 11.106a). Cracks
develop in the radial direction when the effective cir-
1 cumferential stress becomes zero for uncemented soils
3 ⫽ ⫺ (3i ⫹ t) ⫹
1 (11.54) and equal to the negative value of the tensile strength
A
for cemented soils. Assuming that the clay behaves
linear elastically,21 the change in the radial total stress
where 3i is the initial minor principal effective stress
r (⫽
1) at the cavity is equal to the negative of
prior to applying the tensile force. the change in the circumferential stress
(⫽
3);
Injection of fluid into a cylindrical cavity surrounded
r ⫽ ⫺
. Substituting this condition in Eq. (11.54)
l
by a clay formation can lead to fracture by increase in under plane strain conditions (A ⬇ –12 )22 gives
ria
cavity pressure. Examples of this mechanism are frac-
ture grouting and soil fracturing around driven piezom- Pf ⫺ 3i ⫽
r ⫽ 3i
⫹ t or
eters (Lefebvre et al., 1981, 1991). According to cavity
expansion theory, the radial total stress at the cavity Pf ⫽ 23i ⫺ u0 ⫹ t (11.55)
ate
20
A more general case can be written as 3 ⫽ u0 ⫹ (
p ⫹ a
q) ⫺
t, where
p and
q are the changes in mean pressure and deviator 21
For simplicity, the undrained behavior of clays is assumed to be
stress, and a is the modified pore pressure parameter defined by linear elastic-perfectly plastic.
Wood (1990). dM 22
No change of intermediate principal stress (
2 ⫽ 0) is assumed.
Tension crack σr = Pf
σθ
σ0 σr
Pf
hte
σθ σ’0 σθ
Cavity
−σ σθ
Displacement
-σθ = σt t
Solid Line: Total Stresses
Dotted Line: Effective Stresses
rig
Pf
(a)
Plastic σr = Pf
Deformation
py
σθ σ0 2su
σθ 2s σr
σ0 u
Pf
Co
(b)
Figure 11.106 Fracture mechanisms of injection fluids into a cavity: (a) tensile fracture in
undrained conditions and (b) shear failure in undrained conditions.
l
of shear-induced pore pressure and a nonlinear stress– for various soils are plotted in Fig. 11.107. The m val-
ria
strain relationship (Andersen et al., 1994). ues of individual data sets are in general bounded by
As injection pressure increases, the clay at the sur- Eqs. (11.55) and (11.56).
face of the cavity may reach undrained shear failure
before the circumferential effective stress becomes
Fracture under Drained Conditions
zero in uncemented soils or reaches the tensile strength
ate
in cemented soils. In such cases, the changes in the Forced seepage flow into a cavity in permeable soil
stress state at the cavity boundary with increasing cav- leads to soil fracture if the effective stress reduces to
ity strain are shown in Fig. 11.106b). Upon shear the negative sign of the tensile strength of the soil.
failure, the difference between the radial and circum- Practical applications of this situation are in situ per-
ferential stresses (both total and effective) remains
dM meability testing and bore hole stability.
equal to 2su, and, therefore, the minimum principal ef- To interpret the fracture conditions around a driven
fective stress never reaches zero. In such circumstance, piezometer, Bjerrum et al. (1972) developed the fol-
it is difficult to see how plastic yielding initiates a frac- lowing conditions for the initiation of fracture in soils
ture. However, there is much field and experimental using the equilibrium equation with the assumptions of
evidence suggesting that fracture has indeed occurred steady state pore fluid flow from a cylindrical cavity
even though plastic deformation was observed at the and elastic soil material. Horizontal cracks may de-
cavity due to the low undrained shear strength of the velop if the injection pressure exceeds the initial total
hte
soil (Mori and Tamura, 1987; Panah and Yanagisawa, vertical stress:
1989; Au et al., 2003). A possible explanation is that
the increase in plastic shear failure zone created shear Pinj ⫽ u0 ⫹ v0 (11.58)
bands or an unstable state around the cavity. This leads
to a localized microscale crack and the injected fluid where Pinj is the injection pressure, u0 is the initial pore
can penetrate into the crack to produce local tensile pressure, and v0 is the initial vertical effective stress.
rig
stresses at the crack tips, as illustrated in Fig. 11.106b. Vertical cracks in the radial direction from the pie-
A simple cylindrical cavity expansion analysis shows zometer develop when the circumferential effective
that the cavity pressure required for the cavity bound- stress becomes smaller than the tensile strength of the
ary to reach the plastic state is material. Bjerrum et al. (1972) consider two cases: (i)
py
fied:
su is the undrained shear strength. The fracture pressure
冉 冊
Pf increases with initial confining pressure in direct
proportion (i.e., slope of 1). If the plastic zone around 1
Pinj ⫽ u0 ⫹ ⫺ 1 [t ⫹ (1 ⫺ )h0
] (11.59)
the expanding cavity increases before fracture initiates
or su increases with initial confining pressure, the frac-
ture pressure Pf would increase from the value given where is Poisson’s ratio, t is the tensile strength,
in Eq. (11.56) and, therefore, the linear proportion be- h0 is the initial horizontal effective stresses; is a
tween Pf and 3i is expected to be larger than 1. disturbance factor that considers the change in circum-
Empirical equations to estimate soil fracture under ferential effective stress due to piezometer installation.
undrained conditions are available (Jaworski et al., Typical values of are given in Table 11.6.
l
ria
ate
dM
hte
rig
Figure 11.107 Increase in fracture pressure with initial confining pressure of different soils.
py
Co
Range of
Compressibility Ratio
Soil Type E/ h0(1 ⫹ v)
l
Further increase in injection pressure leads to devel- shown in Fig. 11.109. When fluid is injected into the
ria
opment of vertical cracks in the radial direction, which
occurs when the following condition is satisfied:
ate
Desiccation Cracks
Reduction in moisture by surface evaporation from
clays leads to increase in interparticle contact forces
by suction. Soil then shrinks and desiccation cracks
may develop. The generation of cracks changes the
dM
hydraulic properties from Darcy’s-type homogeneous
flow to fracture-dominated flow. This can cause some
environmental problems, such as unexpected poor per-
formance of contaminant barrier systems. Figure
11.108 shows the crack patterns observed after desic-
cation of sensitive clays (Konrad and Ayad, 1997). The
cracks can be pentagonal and heptagonal in shape, and
hte
(b)
soil to create hydraulic fracture, a rule of thumb is that well before failure. A good example is the one-
vertical fractures are formed when K0 is less than 1 [as dimensional compression behavior discussed in Chap-
given Eq. (11.59)] and horizontal fractures develop ter 10. After the stress state becomes larger than the
when K0 is more than 1 [as given in Eq. (11.60) with preconsolidation pressure, the soil has yielded and
⫽ 0]. However, this assumes injection into a linear plastic strains develop. This leads to the concept of
elastic infinite soil medium. When multiple grout in- yield envelope (sometimes referred as yield surface or
jections are performed at close distance, horizontal limit state curve), which differentiates the state of the
fractures can be observed even though K0 is less than soil between elastic and plastic. Examples of the yield
1 (Soga et al., 2004). Natural bedding also affects frac- envelope of sands and clays were shown in Fig. 11.12.
ture orientation. In shallow formations, fractures are When the stress state reaches the yield envelope, the
often horizontally oriented or gradually dipped (Mur- total strain is governed by the development of plastic
l
doch and Slack, 2002). strain increments.
ria
Simple criteria presented as Eqs. (11.56) to (11.61) Unfortunately, for soils, there is no distinct transition
are applied for global stress conditions, where micro- from elastic to plastic behavior. Plastic strains do de-
scale cracks often develop by local tensile stresses at velop inside the yield envelope and the stiffness de-
the crack tips. Fracture mechanics have been used with grades even at very small strain levels. Figure 11.111
some success to characterize the cracking resistance of shows a schematic nonlinear stress–strain relationship
ate
the soils and to examine possible crack propagation
for a soil subjected to monotonic and cyclic deviator
(Morris et al., 1992; Harison et al., 1994; Murdoch and
loads. Some experimental data are shown in Figs.
Slack, 2002). The actual mechanisms of fracture de-
11.85 and 11.88. Under cyclic loading, the relation-
velopment in a fluid–soil system are more complicated
ships are hysteretic, which indicates energy absorption,
than in the above analyses, as illustrated in Fig. 11.110.
They may involve plastic deformation at the crack tip,
dM or damping, during each complete cycle of stress re-
soil rate effects, penetration of injection fluid into the versal. The shear modulus G and damping ratio are
cracks, and permeation of injection fluid from cracks used to characterize the curves in Fig. 11.111, and they
into the soil medium. If the clay is overconsolidated are defined by
and saturated, the negative pore pressure generated by
shearing in front of the crack could possibly lead to
cavitation and dry cracks may develop in front of pen-
etrating injection fluid.
hte
Secant Stiffness
Permeation
Fluid Penetration
Into Crack
Figure 11.110 Possible fracture propagation mechanisms in Figure 11.111 Monotonic and cyclic load stress–strain re-
soils. lationships at different strain amplitudes.
l
Understanding this pre-yield deformation behavior is the p –q plane as shown in Figs. 11.112b, 112c and
ria
very important, as most strains observed in geotech- 112d. The initial stress state is considered to be at point
nical construction practice are indeed small (less than O, and the boundaries of the zones are determined by
0.1 percent) (Burland, 1989). Site response under stress probe testing in different stress path directions.
earthquake loading is influenced by stiffness degrada- The boundaries often associated with strain levels (ax-
ate
tion and damping characteristics that are associated ial or shear strains), and the corresponding secant stiff-
with relatively small strain levels (Seed and Idriss, ness values are illustrated in Fig. 11.112a.
1982). This was illustrated in Fig. 11.9, which shows
typical strains observed in various types of geotech- 1. Zone 1 (True Elastic Region) Soil particles do
nical construction and shows that the necessary defor- not slide relative to each other under a small
dM
I II III IV
Stiffness G or E
q Critical-State Line
II III
Initial State
I
Strain p
rig
0 ’
p
(a) (c) State B
Critical State Line q Critical State Line
q
py
III III
Y2 Envelope
Co
Y2 Envelope Y1 Envelope
I II O
Y1 Envelope Initial State
I
p
Figure 11.112 Four zones of deformation characterization: (a) stiffness degradation and
plastic strain development, (b), (c), and (d) are the stress conditions and the location of the
four zones associated with three successive states (modified from Jardine, 1992).
stress increment, and the stiffness is at its maxi- elastic even though microscopically soil particles
mum. The soil stiffness is determined from con- may not be back to their original locations after
tact interactions, particle packing arrangement, the cyclic loading. When the stress state reaches
and elastic stiffness of the solids. The soil stiff- the outer boundary of zone 2 (called the Y2 en-
ness values can be obtained from elastic wave velope), plastic strains start to develop. The ini-
velocity measurements, resonant column testing, tiation of plastic strains can be determined by
and very accurate local strain transducers. Cyclic examining the onset of permanent volumetric
loading produces only very small hysteresis by strain in drained conditions or residual excess
stick–slip motions at particle contacts and other pore pressures in undrained conditions after un-
mechanisms, producing very small energy dissi- loading. Hence the strain level that defines the Y2
pation less than 1 percent. The strains at which envelope is called volumetric threshold strain.23
l
the stress state reaches the outer boundary of The value of the volumetric threshold strain is
ria
zone 1 (called Y1 envelope) are usually described generally one order of magnitude higher than that
as elastic limit strains or elastic threshold strains. of the elastic threshold strains. The available ex-
This state is illustrated as state A in Fig. 11.112b. perimental data suggest that it ranges between
The elastic limit axial strain depends on soil type, 7 ⫻ 10⫺5 and 7 ⫻ 10⫺4 (the lower limit for un-
solid stiffness, and confining pressure as shown cemented normally consolidated sands and the
ate
in Table 11.7 for different geomaterials. Micro- upper limit for high plasticity clays and cemented
mechanics analysis by Santamarina et al. (2001) sands). At this strain level, the stiffness degrades
shows that it increases from less than 5 ⫻ 10⫺6 to 60 to 85 percent of the true elastic value
strain, for nonplastic soils at low confining pres- (Ishihara, 1996).
sure conditions, to greater than 5 ⫻ 10⫺4 strain
dM
at high confining pressure conditions or in soils
with high plasticity.
2. Zone 2 (Nonlinear Elastic Region) Soil parti- 23
Other definitions of the Y2 surface are available. For example, (a)
cles start to slide or roll relative to each other in perform undrained cyclic loading test and find the linear relationship
this zone. The stress–strain behavior becomes between max and p / max, where max is the maximum strain for each
cycle and p is the residual strain (Smith et al., 1992); (b) the strain
nonlinear, and the stiffness begins to decrease level when excess pore pressures start to accumulate in a sequence
from the true elastic value as the applied strains of undrained cyclic tests at different strain levels (Vucetic, 1994); (c)
hte
or stresses increase. However, a complete cyclic change in the direction of strain path in the vol–s space in drained
tests (Kuwano, 1999); and (d) change in the slope of the excess pore
loading (unloading and reloading) shows full re- pressure–vertical effective stress in undrained triaxial compression
covery of strains and therefore the zone is called test (Kuwano, 1999).
rig
Table 11.7 Elastic Limit Strain for Various Geomaterials from Triaxial Tests
l
tinct kink in the stress–strain relationship and rection in relation to the previous stress path direction
ria
plastic strains develop fully. This state is illus- (Atkinson et al., 1990). If the soil is allowed to age at
trated as state C in Figs. 11.112a and 11.112d. a fixed effective stress point, the Y1 and Y2 envelopes
The yield envelope expands or shrinks depending may grow in size.
on the plastic increments; in general, the yield
envelope expands if positive plastic volumetric
ate
strain (contraction) develops, whereas it shrinks 11.18 LINEAR ELASTIC STIFFNESS
if negative plastic volumetric strain (dilation)
develops. The sizes of Y1 and Y2 surfaces may Knowledge of soil stiffness in the linear elastic region
change with the enlargement or shrinkage of Y3 is important for evaluating soil response under dy-
surface. If the stress state reaches the critical
dM namic loadings such as earthquakes, mechanical vibra-
state, the soil is considered to have reached fail- tion, and vehicle vibration. It also provides indirect
ure. information regarding the state and natural soil struc-
ture, and, therefore, stiffness values can be used to as-
Examples of experimentally determined boundaries sess the quality of soil samples (i.e., the degree of soil
are shown in Fig. 11.12b for Bothkennar clay and Fig. disturbance). The linear elastic stiffness of soils is eval-
11.113 for Ham River sand. These zones are not fixed uated from measurements of elastic wave velocities or
in space when the stress state moves inside the Y3 yield use of local displacement transducers. Theoretical
hte
(11.64)
q = σa – σr
-0.3 24
In practice, Gmax and pare often normalized by pa (reference pres-
sure such as atmospheric pressure) so that the equation appears to
Figure 11.113 Y1, Y2, and Y3 envelopes for Ham River sand be dimensionally consistent. However, there is no physical meaning
(Jardine et al., 2001). to this.
Angular-grain crushed 3,270 (2.973 ⫺ e)2 0.5 0.6–1.3 RC Hardin and Richart
quartz 1⫹e (1963)
l
Several sands 9,000 (2.17 ⫺ e)2 0.4 0.6–0.9 RC Iwasaki et al. (1978)
ria
1⫹e
Toyoura sand 8,000 (2.17 ⫺ e)2 0.5 0.6–0.8 Cyclic TX Kokusho (1980)
1⫹e
ate
Several cohesionless 4,500– 1 0.5 NA RC Hardin and Blandford
and cohesive soils 140,000 0.3 ⫹ 0.7e2 (1989)
Ticino sand 7,100 (2.27 ⫺ e)2 0.43 0.6–0.9 RC and Lo Presi et al. (1993)
dM 1⫹e TS
Clays
Reconstituted NC 3,270 (2.973 ⫺ e)2 0.5 0.5–1.5 RC Hardin and Black
kaoline 1⫹e (1968)
Several undisturbed 3,270 (2.973 ⫺ e)2 0.5 0.5–1.7 RC Hardin and Black
NC clays 1⫹e (1968)
hte
Several undisturbed 893–1,726 (2.973 ⫺ e)2 0.46–0.61 0.4–1.1 RC Kim and Novak
silts and clays 1⫹e (1981)
1⫹e (1982)
Undisturbed Italian 4,400–8,100 e⫺1.3(average 0.40–0.58 0.6–1.8 RC and Jamiolkowski et al.
clays from e⫺x: x BE (1995)b
⫽ 1.11–1.43)
Co
Several soft clays 5,000 e⫺1.5 0.5 1–5 SCPT Shibuya and Tanaka
(1996)c
Several soft clays 18,000– 1 0.5 1–6 SCPT Shibuya et al. (1997)c
30,000 (1 ⫹ e)2.4
a
RC: resonant column test, TX: triaxial test, TS: torsional shear test, BE: bender element test, SCPT: seismic cone test.
b
From anisotropic stress condition.
c
Using v instead of p.
After Yimsiri (2001).
Coordination
Packing Number Shear Modulus
Simple cubic 6 Gmax
Gg
⫽
3
2 冉冊 1/3
(1 ⫺ g)1 / 3 p
2 ⫺ g Gg 冉冊 1/3
冉冊 冉冊
l
Face-centered cubic 12 Gmax 1 3
1/3
(4 ⫺ 3g) p
1/3
ria
⫽
Gg 2 2 (2 ⫺ g)(1 ⫺ g)2/3
Gg
ate
After Santamarina and Cascante (1996).
late the global elastic stiffness to microscopic proper- F(e). Several expressions are available for the void ra-
ties such as particle stiffness and Poisson’s ratio, tio function as listed in Table 11.8. These functions are
dM
number of contacts, void ratio, and contact force di- empirical and apply for specific ranges of void ratios
rections (see Table 11.9). These solutions suggest that and, therefore, should be used with caution.
the pressure p and Gmax could be normalized by the Equation (11.64) is derived assuming isotropic stress
shear modulus of the particle itself (Gg). conditions. Anisotropic stress conditions as well as an-
It is noted from Table 11.8 that the values of the isotropic soil fabric give stiffness values that depend
exponent n range from 0.4 to 0.6. As shown in Table on the direction of loading. The shear modulus is a
hte
11.9, however, classical contact mechanics solutions function of the principal effective stresses in the direc-
using the Hertz–Mindlin contact theory predict n ⫽ tions of wave propagation and particle motion and is
–13 . This is because the soil particles are assumed to be relatively independent of the out-of-plane principal
smooth elastic spheres. If the contacts are considered stress. This is shown in Fig. 11.114, in which the var-
to be an interaction of rough surfaces, the modification iations of measured shear wave velocities propagating
of theory leads to increases in the exponent to values in three different directions (Vsxy, Vsyz, and Vszx) are
rig
that are closer to the experimental observations given shown as the vertical effective stress z was increased
in Table 11.8 (Yimsiri and Soga, 2000).
By comparing Eq. (11.64) with the micromechanical
model listed at the bottom of Table 11.9, it is possible
σz Change in Vertical Effective Stress
to relate the void ratio function F(e) to number of con-
py
tacts per particle (i.e., coordination number) and A to 400 Direction of Wave Propagation
Vs-zx
S-wave Velocity, Vs (m/s)
the elastic properties of particle itself. From the anal- Particle Motion
ysis of uniform grain fabrics, the coordination number Vs-xy
Cn can be related to the porosity n by Eq. (5.1) or to
Co
Vs-xy
the void ratio e by the following equation. (Chang et Vs-yz
300
al., 1991).
Vs-yz
Vs-zx
Cn ⫽ 13.28 ⫺ 8e (11.65)
with the horizontal effective stresses x and y being σz Change in Vertical Effective Stress
held constant (Stokoe et al., 1995). The shear wave 600 Direction of Wave Propagation
Vp-zz
Vsxy, which propagates and has the particle motion in
l
where i is the effective normal stress in the direction
ria
Figure 11.115 Variation of P-wave velocities in different di-
of wave propagation, j is the effective normal stress
rections as a function of anisotropic stresses (Stokoe et al.,
in the direction of particle motion, and AG, rG, sG, and 1995).
k are material constants. Experimental evidence sug-
gests that rG ⬇ sG. Hence, an alternative equation that
ate
relates the stiffness to the mean state of stress on the and Tatsuoka, 1998). This leads to the following em-
plane of particle motion is also available: pirical equation for small strain Young’s modulus:
Gij(max) ⫽ AGFG
dM冉 i ⫹ j
2 冊 nG
OCRk (11.67)
Ei(max) ⫽ AEF(e)i nE (11.69)
tuted and undisturbed London clay. It can be argued 11.117 indicate that Poisson’s ratio vh (i.e., horizontal
that the void ratio function is a redundant factor since expansion by vertical load) increases with vertical ef-
the void ratio is a unique function of present effective fective stress and decreases with increase in horizontal
stress, stress history (OCR), and soil compressibility. stress. The following empirical equations are proposed
However, this argument should be restricted to recon- by Horque and Tatsuoka (1998) for Poisson’s ratios:
stituted clays and not applied to natural clays.
rig
5000
Ev SLB Sand Ev Toyoura Sand
Eh SLB Sand E
Ehv Toyoura
Toyoura sand
Sand
Ev/(F(e)Pa) or Eh/(F(e)Pa)
4000
3000
l
2000
ria
For Ev (or Eh) measurement, σv/Pa (or σh/Pa) is
varied between 1.0 and 2.0 at each σv (or σh)
1000
1.0 1.5 2.0 Figure 11.117 Poission’s ratio as a function of anisotropic
σv/Pa or σh/Pa stresses (Hoque and Tatsuoka, 1998).
ate
(a) Ev versus σv and Eh versus σh
1.2
dM
Ev/(AEF(e)σv)
1.0
0.8
hte
Figure 11.116 Vertical and horizontal Young’s modulus as Figure 11.118 Stiffness anisotropy of undisturbed London
a function of anisotropic stresses for Toyoura sand (Hoque clay under isotropic stress conditions (Jovicic and Coop,
and Tatsuoka, 1998). 1998).
py
For sands, most studies show that the ratio Ghh /Gvh
that, for a given confining pressure, the values of Ghh is greater than 1 (e.g., Lo Presti and O’Neill, 1991;
are larger than those of Gvh ⬇ Ghv. Hence, the soil is Stokoe et al., 1991; Bellotti et al., 1996). However,
inherently stiffer horizontally than vertically due to its reported values for the ratio of Ev /Eh are inconclusive;
Co
soil fabric. some sands are stiffer in the vertical direction (Hoque
The reported data on clay under isotropic stress con- and Tatsuoka, 1998), whereas the others are stiffer in
ditions consistently show that Ghh is approximately 50 the horizontal direction (Stokoe et al., 1991). Aniso-
percent larger than Gvh, indicating inherent anisotropic tropic properties are related to fabric (contact) aniso-
characteristics caused by orientation of platy clays tropy, and therefore the mixed results obtained may be
(Pennington et al., 1997; Jovicic and Coop, 1998). The due to the differences in sample preparation proce-
ratios of Ghh /Gvh for six Italian clays measured in one- dures.
dimensional consolidation tests were between 1.3 and The experimental data show that the small strain
2.0, and the ratio increased with overconsolidation ra- stiffness is rather insensitive to the strain rate and num-
tio (Jamiolkowski et al., 1995). ber of loading cycles as long as the loading is within
the true elastic range but that the elastic limit strain analysis. For instance, assume that the true elastic axial
increases with strain rate (Shibuya et al., 1992; Tat- stiffness of a soil is 100 MPa. Considering that the
suoka et al., 1997). Resonant column tests on clays and elastic threshold axial strain is of the order of 10⫺5,
sands show that the small strain shear modulus is in- the axial stress increment required to reach to this
dependent of frequency in the range of 0.05 to 2500 strain level is only 1 kPa. Hence, errors in stiffness of
Hz (e.g., Hardin and Richart, 1963; Hardin and Drnev- 100 percent result in small differences in the asso-
ich, 1972; Stokoe et al., 1995). ciated stress increments (a few kilopascals). Typical
Although conservation of energy may be an issue strain levels under working loads are usually in an in-
for true elastic response, experimental evidence indi- termediate level between linear elastic and plastic de-
cates that energy is dissipated even at this strain level formation, and, therefore, the knowledge of nonlinear
and damping values are typically 0.35 to 1 percent for (zone 2) and irreversible (zone 3) deformation char-
l
sands and 1.0 to 1.5 percent for clays. Similar to the acteristics is more important for evaluating ground
ria
small strain stiffness, the damping at very small strain movements accurately.
also depends on confining pressure and the following Stiffness degradation from small strains to interme-
empirical form is proposed (Hardin, 1965): diate strains has been recognized in resonant column
testing since the 1960s when the soil was subjected to
⫽ Bpm (11.72) cyclic loading (Hardin and Drnevich, 1972). Nowa-
ate
days, detailed characterization of deformation proper-
where B and m are material constants. The reported ties at intermediate strain levels is possible with the
values of the exponent m range from ⫺0.05 to ⫺0.22 use of local strain measurement systems, as described
(Santamarina and Cascante, 1996; Stokoe et al., 1999). previously.
Although the particles in contact are not moving rel- The shear modulus decreases and the damping in-
ative to each other, some microscopic proportion of the
dM
contact area can slide or slip, which is known as the
creases as the shear strain increases because of struc-
tural breakdown that results in a decreasing proportion
stick–slip frictional contact loss. Micromechanical of elastic deformation and an increasing proportion of
analysis considering the energy dissipation by this plastic strain with increasing shear strain. The shear
behavior gives m ⫽ ⫺–23 . Santamarina and Cascante modulus degradation curves of Ticino sand, obtained
(1996) attribute the difference to other attenuation by monotonic and cyclic loadings using various testing
mechanisms available in soils. These include chemical apparatus (triaxial compression, torsional shear, and
interaction of adsorbed layers at contacts, wave scat-
hte
Cyclic Triaxial
loop) (Zytynski et al., 1978). To be thermomechani- Cyclic Torsional shear
cally consistent, theoretical models for the pressure- 80
py
Resonant Column
dependent stiffness of soils are available (e.g.,
Houlsby, 1985; Hueckel et al., 1992; Borja et al., 1997; 60
Einav and Puzrin, 2004). They show that, if both shear
and bulk moduli are to be mean pressure dependent,
Co
l
11.10 shows the stiffness degradation of sands and A summary of normalized shear modulus degradation
ria
clays subjected to increase in shear stress at different curves for a variety of soils are shown in Fig. 11.121
confining pressures. The effect of stress path directions (Kokusho, 1987). The curve for modulus degradation
on the stiffness degradation curve is shown in Fig. with increasing strain may be somewhat flatter for
11.120 (Atkinson et al., 1990). Triaxial tests were per- gravels than that for sands and clays. The curves tend
formed on reconstituted overconsolidated London clay to move to the right as the confining pressure in-
ate
specimens in such a way as to maintain a constant creases; it is possible that the degradation curve at very
mean pressure. Different stiffness degradation curves high confining pressure (in the megapascal range) may
were obtained even though they were sheared along lie beyond the bands given in Fig. 11.121 (Laird and
the same stress path (OX in Fig. 11.120a). This is be- Stokoe, 1993).
dM Sands and Gravels
The following relationship can be used for the dynamic
shear modulus of sands and gravels at different strain
Deviator stress q (kPa)
X
levels (Seed et al., 1984):
100 D
0
C O A
G
pa
⫽ 22.1K2
p
pa 冉冊 1/2
(11.73)
hte
40
a factor from about 1.35 to 2.5 for gravels than for
sands. Values of K2 vary with relative density and shear
(D➝)O➝X
30
Shear Modulus G (MPa)
py
(A➝)O➝X
1.0
Shear Modulus Ratio G/Gmax
20
Co
(C➝)O➝X
10 Clay, 100 kPa
0.5
(B➝)O➝X Sand, 50 kPa
strain approximately as shown in Fig. 11.122 and with Dobry (1991) based on the results of a review of avail-
void ratio and shear strain as shown in Fig. 11.123. able cyclic load data from 16 different studies. The
Equation (11.73) assumes that the exponent is –12 . Ex- influences of various compositional and environmental
perimental evidence suggests that the exponent in- factors on shear modulus and damping ratio of nor-
creases with strain level as shown in Fig. 11.124 and mally consolidated and moderately overconsolidated
reaches 0.8 to 0.9 at a strain level of 1 percent (Jovicic clays are listed in Table 11.10.
and Coop, 1998; Yamashita et al., 2000). Vucetic and Dobry (1991) hypothesized that increas-
Values of the damping ratio for sands and gravels ing plasticity influences the degradation curves in the
are about the same, and they are only slightly influ- following manner. Increasing plasticity index reflects
enced by grain size and density. The ranges of values decreasing particle size and increasing specific surface
as a function of cyclic shear strain are shown in Fig. area. The number of interparticle contacts becomes
l
11.125. The damping value decreases with increasing large, and interparticle electrical and chemical bonding
ria
number of loading cycles and confining pressure, and and repulsive forces become large relative to the par-
much of the decrease occurs in the first 10 cycles (Sto- ticle weights in comparison with sands. The many
koe et al., 1999). bonds within the microstructure act as a system of rel-
atively flexible linear springs that can resist larger
ate
Clays shear strains (up to 0.1 percent before they are broken)
Although the variation of shear moduli and damping than is the case for sands, wherein particle elasticity is
ratio with shear strain is relatively independent of com- practically the only source of linear behavior, and in-
position for sands and gravels, the same is not the case terparticle sliding at contacts may start at strains as low
for cohesive soils. Curves of the type shown in Figs. as percent with the onset of plastic deformations.
11.121 and 11.125 are displaced to the right for clays
dM To these ideas might be added the fact that the thin,
with increasing plasticity, as shown by Fig. 11.126. platy morphology of most clay particles make them
These relationships were developed by Vucetic and able to deform elastically to considerably greater levels
hte
rig
py
Co
Figure 11.122 Shear modulus factor K2 for sands as a function of relative density and shear
strain (Seed et al., 1984).
l
ria
ate
dM
Figure 11.123 Shear modulus factor K2 for sands as a function of void ratio and shear strain
(Seed et al., 1984).
hte
rig
py
Co
l
lightly overconsolidated clays because the initial stress
ria
state is either already on or near the yield envelope.
The most important mechanical feature of soil in the
plastic state is dilatancy, in which there is coupling
between shear and volumetric deformations. That is,
dense sands and heavily overconsolidated clays exhibit
ate
volume dilation in drained conditions and negative ex-
cess pore pressure generation in undrained conditions,
whereas loose sands and normally consolidated and
lightly overconsolidated clays exhibit volume contrac-
dM tion in drained conditions and positive excess pore
pressure generation in undrained conditions. The rule
that governs the generation of plastic volumetric strain
associated with plastic deviator strain is called the di-
latancy (or flow) rule. Some examples of this for dense
sands were already presented in Eqs. (11.34) and
(11.35), in which the degree of dilatancy [dy/dx in Eq.
(11.34) and in Eq. (11.35)] is related to the applied
hte
Figure 11.126 Normalized modulus and damping ratio as a principal stress ratio (or the mobilized friction angle)
function of cyclic shear strain showing the influence of soil and the internal friction angle. These observations are
composition as measured by plasticity index (from Vucetic important because the incorporation of stress–dilatancy
and Dobry, 1991). Reprinted with permission of ASCE. into plasticity theory can lead to a useful form of con-
stitutive modeling for soils.
The development of plastic strains is often charac-
rig
intergrain contacts than could be accommodated with- essential aspects of soil behavior observed during plas-
out sliding. tic deformation are summarized here.
l
Overconsolidation, Increases with OCR Not affected Not affected
ria
OCR
Plasticity index, PI Increases with PI if Increases with PI Decreases with PI
OCR ⬎ 1; stays
about constant if
OCR ⫽ 1
ate
Cyclic strain, c — Decreases with c Increases with c
Strain rate, ˙ Increases with ˙ G increases with ˙ ; Stays constant or may
(frequency of cyclic G/ Gmax probably not increase with ˙
loading) affected if G and
Gmax are measured at
same ˙
Number of loading
dM Decreases after N Decreases after N cycles Not significant for
cycles, N cycles of large c but of large c (Gmax moderate c and N
recovers later with measured before N
time cycles)
From Dobry and Vucetic (1987).
hte
0.4 rule.
l
0 0.2 0.4 0.6 0.8 1.0
of the arrows is the deviator plastic strain increment
ria
(σa+ σr)/2σp
d ps (or d p), whereas the horizontal component is the
(a) volumetric plastic strain increment dpv.25 Similarly, the
plastic strain vectors measured in Winnipeg clay are
0.6 Intact State shown in Fig. 11.130 (Graham et al., 1983b). The vec-
ate
tor of the plastic strain increment appears to be a func-
Destructured State
tion of the current stress state. This observation leads
(σa – σr)/2σp
(b) dpv q
⫽M⫺ ⫹ 0 (11.75)
ds
p
p
hte
2. The yield envelope increases in size with increas- havior when the dilation angle is negative and q/p is
ing preconsolidation pressure pc, which is often less than M ⫹ 0, whereas the soil exhibits dilative
associated with the generation of plastic volu- behavior when the dilation angle is positive and q/p
metric strain. The size increases as the soil is is more than M ⫹ 0. Figure 11.131 shows the stress–
more densely packed along the normal consoli- dilatancy relationship for the data presented in Fig.
py
dation line. A mathematical form that describes 11.130. The data follow a similar trend to Eq. (11.75).
the change in pc with generation of plastic strains Other stress–dilatancy rules that are used to derive
is called the hardening rule. constitutive models for clays are available.
3. The shape of the yield envelope is often an in- Experimental evidence suggests that the stress–
Co
clined ellipse in the p –q plane. The inclination dilatancy relationship for sand depends on confining
is related to the anisotropic consolidation history pressure and density as well as soil fabric, compared
as well as the anisotropic fabrics. Some yield en- to a simpler form used in clays such as Eq. (11.75).
velopes of sands are shown in Fig. 11.129 (Yas- Rowe (1962) derived the following stress–dilatancy
ufuku et al., 1991). The yield envelopes were
determined by applying different stress paths and
connecting the stress state when the plastic 25
In triaxial condition, dpv ⫽ dpa ⫹ 2drp, dsp ⫽ (–23 )(dpa ⫺ dpr ), and
strains initiate for a given stress path. The shape d p ⫽ dpa ⫺ dpr , where dpa is the axial plastic strain and dpr is the
of the yield envelopes resembles a tear drop, and radial plastic strain.
26
Note that Taylor’s expression was for the peak state only. This
the inclinations of the yield envelopes are clearly equation is applied to all stress state conditions under plastic defor-
affected by the initial anisotropic stress condi- mation for both loose and dense cases.
l
-0.4 -0.4 -0.4
ria
(a) (b) (c)
Figure 11.129 Yield surfaces of sands with different initial stress histories. Initial states (a)
compression, (b) isotropic, and (c) extension (Yasufuku et al., 1991).
ate
dε p Cam-clay
1.4
(Roscoe and Schofield, 1963)
dεsp
dε pv/dε ps = M - q/p
dM 1.2
dε vp Modified Cam-clay
Stress Ratio q/p'
0.6
0.4
0.2
hte
0.2
Cam-clay
0 0
0 0.2 0.4 0.6 0.8 1.0 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
p
Plastic Strain Ratio (-dε pv/dε s)
p/σp
rig
Figure 11.130 Plastic strain vectors at yielding of natural Figure 11.131 Stress dilatancy relations of natural Winnipeg
Winnipeg clay (Graham et al., 1983b). clay (Wood, 1991).
py
rule for sand in triaxial loading based on his experi- respectively. Equations (11.76) and (11.77) have a sim-
mental data as well as theoretical analysis: ilar form to Eq. (11.75), in which the dilation depends
on stress ratio and material constants.27 However,
a
r 冉 冊 冉
⫽
⫺2dpr
da
tan2
c
⫹ 冊 Rowe (1962) noted that the material constant c used
in Eqs. (11.76) and (11.77) is influenced by the density.
Co
p
4 2
Different initial anisotropic stress states give different
in triaxial compression (11.76)
r
a
⫽ 冉 冊 冉
⫺dap
2drp
tan2
c
4
⫹
2 冊 27
Equations (11.76) and (11.77) can be rewritten in terms of p, q,
d pv, and d p (Pradhan and Tatsuoka, 1989):
where dpa and dpr are the axial and radial strain
increments, c is the ‘‘characteristic friction angle’’
q
⫽ 冋
3 (K ⫹ 2)(⫺d pv / d p) ⫺ 2(K ⫺ 1)
p 2 (1 ⫺ K)(⫺d vp / d p) ⫹ (2K ⫹ 1) 册 for d pa ⬍ 0
and a and r are the axial and radial effective stresses, where K ⫽ (1 ⫹ sin c) / (1 ⫺ sin c).
l
stress increments. For example, for an isotropic elastic grains as well as pore fluid. The particle contact prop-
ria
model, erties will also be modified. A change in temperature,
therefore, causes either a change in void ratio or a
change in effective stress (or a combination of both)
dev dp G in a saturated clay, as described in Section 10.12. In
⫽3 (11.78)
dse
dq K this section some effects of temperature on shear re-
ate
sistance of soils are considered.
A change in temperature can cause a strength in-
where G is the shear modulus, K is the bulk modulus crease or a strength decrease depending on the circum-
and dev and dse are the elastic volumetric and devia- stances (e.g., temperature variation during initial
toric strains, respectively. consolidation or during shearing in drained or un-
The physical mechanisms of elastic deformation and drained conditions), as illustrated by Fig. 11.133.
dM
plastic deformation are fundamentally different, that is,
stress increment dependent versus stress dependent.
The higher the consolidation temperature, the
greater the shear strength at any given test temperature
Because of this, the same stress increment may give because of the greater decrease in void ratio at the
very different strain increments. Careful selection of higher consolidation temperatures.28 In Fig. 11.133, Tc
elastic and/or plastic models is therefore necessary in represents the temperature at consolidation and Ts the
ground deformation analysis. temperature of shear for consolidated undrained direct
hte
Case (b)
decrease by about 10 percent for a temperature in-
Case (c)
1.5 crease from 0 to 40C. For overconsolidated clays, the
in Fig.11.129 undrained shear strength is less influenced by temper-
1.0 ature (Marques et al., 2004). The relative insensitivity
Compression
of overconsolidated clay to temperature may be due to
py
-4 -3 -2 -1 0 1 2 3 4
? pressure, and hence the size of the yield envelope, de-
Strain Increment
ratio -dεvp/dγ p
creases with increase in temperature, as illustrated in
-0.5
Fig. 10.46 and Fig. 11.134 for natural clay specimens
?
tested between 10 and 50. Hence, the weakening of
Extension
-1.0
l
ria
ate
Figure 11.133 Effect of consolidation and test temperatures on the strength of alluvial clay
in direct shear (Noble and Demirel, 1969).
dM
hte
rig
py
soil structure by increase in temperature is apparent. sion of soil volume because thermal expansion of wa-
On the other hand, the critical state friction angle is ter is much larger than that of soil particles. This
found to be independent of temperature (Hueckel and results in generation of positive excess pore pressure
Baldi, 1990; Graham et al., 2001; Marques et al., and, as a consequence, undrained stiffness and shear
2004). strength decrease as shown in Fig. 11.16.
Drainage conditions during heating prior to shear are If drainage is allowed, the expanding water is free
important, as illustrated in Fig. 11.133. If drainage is to drain and hence the volume change of the soil is
prevented, the expansion of water controls the expan- governed by the expansion of soil particles and the
change in particle contact conditions. Normally con- a dense state (below critical state) can only be achieved
solidated clays often result in decrease in void ratio, by unloading, and, therefore, the preconsolidation pres-
and hence the initial stiffness generally increases with sure can be used to characterize the peak strength and
temperature. However, it has been reported that the de- deformation. For sands, on the other hand, the differ-
crease in void ratio in normally consolidated clays can- ence in strength and deformation behavior of normally
not be solely accounted for the increase in stiffness consolidated dense sand and overconsolidated sand is
(Tsuchida et al., 1991; Kuntiwattanakul et al., 1995). noted even when they are at the same void ratio and
This observation is similar to the aging effect discussed confining pressure. This is because of possible differ-
in Chapter 12. Hence, it can be considered that tem- ent soil fabrics. The critical friction angle of cohesion-
perature is one of the driving forces in time-dependent less soils contains contributions from particle crushing,
deformation of soils, and the rate process theory de- particle rearrangement by rolling, as well as from in-
l
scribed in the next chapter conveniently explains much terparticle sliding. The critical state concept can be
ria
of the observed temperature–time–effective stress be- used to characterize the density effect on peak strength
havior of soils. for normally consolidated sand. Rearrangement and
rolling are unimportant when the clay content is high
enough to prevent granular particle interference. Ide-
ally, the critical state strength or friction angle should
ate
11.22 CONCLUDING COMMENTS
be used for design of simple geotechnical structures.
Limit equilibrium and plasticity analyses, as done, for Otherwise, a careful selection of safety factor is needed
example, in studies of slope stability, lateral pressure, when the peak strength or peak friction angle is used.
and bearing capacity, depend on accurate representa- However, whether it is possible to find the true critical
tion of soil strength. So also does soil resistance
dM state from conventional triaxial and torsional shear
against failure due to earthquakes or other cyclic load- tests is questionable, especially for sands.
ings. The stresses and deformations under subfailure Because of the great diversity of soil types and the
loading conditions depend on stress–strain properties. range of environmental conditions to which they may
The factors responsible for and influencing strength be subjected, evaluations of deformation and strength,
have been identified and analyzed. their characterization for analyses, and prediction of
The strength of most uncemented soils is provided future behavior will continue as major components of
by interparticle sliding, dilatancy, particle rearrange- any project. In the majority of geotechnical engineer-
hte
ments, particle crushing, and true cohesion. Frictional ing projects and problems, correct site characterization
resistance is developed by adhesion between contact- and property evaluation are the two most critical ele-
ing asperities on opposing particle surfaces. Values of ments. If they are not done reasonably and reliably,
true friction angle () range from less than 4 for then there cannot be understanding or confidence from
sodium montmorillonite to more than 30 for feldspar subsequent soil mechanics analyses, no matter how so-
and calcite. In the absence of cementation, true cohe- phisticated they may be or how powerful the computer
rig
sion in soils is small. Results from discrete particle that provides the numerical solutions.
simulations indicate that the deviatoric load applied to
a particle assembly is transferred exclusively by the
normal contact forces in the strong force networks. The
QUESTIONS AND PROBLEMS
interparticle friction therefore acts as a kinematic con-
py
straint of the strong force network and not as the direct 1. Based on the descriptions given in Section 11.3
source of macroscopic resistance to shear. and 11.6, summarize microscopic interpretation of
The residual friction angle depends on gradation, overconsolidation, compaction, dilation, peak fric-
mineralogical composition, and effective stress. The tion angle, and critical state friction angle.
Co
l
a. Using the Hvorslev parameters of hc ⫽ 0.1 and
tests?
ria
e ⫽ 15, plot the failure envelope on the –
c. Draw the effective stress path of a drained tri- plane.
axial compression test on the p –q plane. The b. Plot shear strength f / v as a function of OCR
initial effective isotropic confining pressure is and compare the results to the data shown in
100 kPa. Find the drained strength and void ra- Fig. 11.65.
ate
tio at critical state.
d. Sketch possible stress–axial strain and axial 7. Why does a sample with shear bands give different
strain–void ratio curves of the drained triaxial strengths depending on sample size?
compression test considered in part (c). Con- 8. Find a case study that describes the importance of
sider two different initial void ratios: (i) e ⫽ knowing the residual friction angle of clay. Ex-
0.4 and (ii) e ⫽ 0.7. Comment on the results.
dM
e. Repeat the calculations of parts (c) and (d)
plain (a) the geologic and hydrogeologic condi-
tions, (b) the possible peak, critical, and residual
when the initial confining pressure is 1 MPa. friction angles, and (c) microscopic interpretation
Comment on the results. of decrease in friction angle at residual state.
4. Using the critical state of the sand defined in 9. Consider two saturated samples of the same soil
Question 3, plot void ratio versus peak friction an- having exactly the same water content, density,
gle at three different confining pressures: (i) 5 kPa, temperature, and structure are initially at equilib-
hte
(ii) 500 kPa, and (iii) 5 MPa. To develop the plot, rium under the same effective stress states. Com-
try (i) Eq. (11.37) or (ii) Fig. 11.56. Comment on pare and explain differences in strength, if any,
the results by discussing the relative importance of that you would expect if
confining pressure and void ratio on friction angle a. One is loaded in triaxial compression and the
of soils. other in plane strain.
rig
5. A clay was isotropically normally consolidated b. One is tested in triaxial compression and the
and the isotropic compression line was found to other is tested in plane stress.
be e ⫽ 1.5 ⫺ 0.35 ln p. The clay was then un- c. One is tested as is and the other is tested after
loaded isotropically and the slope of unloading heating with (i) no drainage allowed and (ii)
py
and (iii) location C, which is located some dis- at a depth of 20 m. Consider both fracturing in (i)
tance away from the embankment. undrained conditions assuming that the injected
fluid has not permeated into the ground and (ii)
drained conditions assuming the injection is in a
steady state seepage state.
14. Convert some of the stiffness degradation curves
plotted in Figs. 11.10 and 11.119 to shear stress
versus logarithm of strain. Identify the shear
stresses required to reach the boundaries of differ-
ent zones described in Section 11.17. Discuss
which zones are important for what type of geo-
l
technical activities.
ria
11. Find a paper that describes the effects of soil fabric
on liquefaction resistance of sands. Give the mi- 15. Give physical microscopic explanations of differ-
croscopic interpretation of why a sample with a ent stiffness degradation curves presented in Fig.
certain soil fabric generates more excess pore pres- 11.120. Why can the multisurface concept pre-
sures than others. sented in Section 11.17 be used to model this com-
ate
12. Provide physical explanations of how and why the plex behavior?
following factors can affect the cyclic resistance 16. Discuss the differences between elastic and plastic
ratio (CRR) of sands: deformations of soils as microscopic behavior and
a. Confining pressure macroscopic behavior.
b. Initial K0 stress condition
dM
c. Static shear stress along the sloping ground
17. The data showing volume reduction with increas-
ing temperature at a given pressure were presented
d. Shear modes (triaxial compression and exten- in Fig. 10.44 (Campanella and Mitchell, 1968). If
sion, simple shear, etc.) we consider the normal compression curve at
e. Sample preparation and soil fabric 76.5F to be the reference state, the compression
f. Silt fines and clay fines curves at the other temperatures can be interpreted
to have exhibited temperature-induced creep be-
hte
13. Water is injected into overconsolidated clay with havior and hence reached the quasioverconsoli-
an OCR of approximately 3. Using the correlations dated state. Can the data presented in Fig. 11.133
and data presented throughout the book, estimate be explained in such a way using the Hvoslev
the injection pressure required to fracture the clay strength concept for overconsolidated clays?
rig
py
Co
l
Deformation
ria
ate
12.1 INTRODUCTION dM Time-dependent deformations and stress relaxation
are important in geotechnical problems wherein long-
Virtually every soil ‘‘lives’’ in that all of its properties term behavior is of interest. These include long-term
undergo changes with time–some insignificant, but settlement of structures on compressible ground, de-
others very important. Time-dependent chemical, formations of earth structures, movements of natural
geomicrobiological, and mechanical processes may and excavated slopes, squeezing of soft ground around
result in compositional and structural changes that tunnels, and time- and stress-dependent changes in soil
lead to softening, stiffening, strength loss, strength properties. The time-dependent deformation response
hte
gain, or altered conductivity properties. The need to of a soil may assume a variety of forms owing to the
predict what the properties and behavior will be complex interplays among soil structure, stress history,
months to hundreds or thousands of years from now drainage conditions, and changes in temperature, pres-
based on what we know today is a major challenge in sure, and biochemical environment with time. Time-
geoengineering. Some time-dependent changes and dependent deformations and stress relaxations usually
rig
their effects as they relate to soil formation, composi- follow logical and often predictable patterns, at least
tion, weathering, postdepositional changes in sedi- for simple stress and deformation states such as uni-
ments, the evolution of soil structure, and the like are axial and triaxial compression, and they are described
considered in earlier chapters of this book. Emphasis in this chapter. Incorporation of the observed behav-
in this chapter is on how time under stress changes the ior into simple constitutive models for analytical de-
py
structural, deformation, and strength properties of scription of time-dependent deformations and stress
soils, what can be learned from knowledge of these changes is also considered.
changes, and their quantification for predictive pur- Time-dependent deformation and stress phenomena
poses. in soils are important not only because of the imme-
Co
When soil is subjected to a constant load, it deforms diate direct application of the results to analyses of
over time, and this is usually called creep. The inverse practical problems, but also because the results can be
phenomenon, usually termed stress relaxation, is a used to obtain fundamental information about soil
drop in stress over time after a soil is subjected to a structure, interparticle bonding, and the mechanisms
particular constant strain level. Creep and relaxation controlling the strength and deformation behavior.
are two consequences of the same phenomenon, that Both microscale and macroscale phenomena are dis-
is, time-dependent changes in structure. The rate and cussed because understanding of microscale processes
magnitude of these time-dependent deformations are can provide a rational basis for prediction of macro-
determined by these changes. scale responses.
465
12.2 GENERAL CHARACTERISTICS of the soil. The most active clays usually exhibit
the greatest time-dependent responses (i.e.,
1. As noted in the previous section soils exhibit smectite ⬎ illite ⬎ kaolinite). This is because
both creep1 and stress relaxation (Fig. 12.1). the smaller the particle size, the greater is the
Creep is the development of time-dependent specific surface, and the greater the water ad-
shear and/or volumetric strains that proceed sorption. Thus, under a given consolidation
at a rate controlled by the viscouslike resistance stress or deviatoric stress, the more active and
of soil structure. Stress relaxation is a time- plastic clays (smectites) will be at higher water
dependent decrease in stress at constant defor- content and lower density than the inactive clays
mation. The relationship between creep strain (kaolinites). Normally consolidated soils exhibit
and the logarithm of time may be linear, con- larger magnitude of creep than overconsolidated
l
cave upward, or concave downward as shown soils. However, the basic form of behavior is
ria
by the examples in Fig. 12.2. essentially the same for all soils, that is, undis-
2. The magnitude of these effects increases with turbed and remolded clay, wet clay, dry clay,
increasing plasticity, activity, and water content normally and overconsolidated soil, and wet and
dry sand.
ate
3. An increase in deviatoric stress level results in
an increased rate of creep as shown in Fig. 12.1.
Some soils may fail under a sustained creep
stress significantly less (as little as 50 percent)
than the peak stress measured in a shear test,
dM wherein a sample is loaded to failure in a few
minutes or hours. This is termed creep rupture,
and an early illustration of its importance was
the development of slope failures in the Cucar-
acha clay shale, which began some years after
the excavation of the Panama Canal (Casa-
grande and Wilson, 1951).
4. The creep response shown by the upper curve
hte
constant stress and (b) stress relaxation under constant strain. about 1.0 kg/cm2 are compatible with the pre-
dictions of rate process theory, discussed in Sec-
tion 12.4. At deviator stress approaching the
1
The term creep is used herein to refer to time-dependent shear strength of the material, the strain rates become
strains and / or volumetric strains that develop at a rate controlled by
the viscous resistance of the soil structure. Secondary compression very large and signal the onset of failure.
refers to the special case of volumetric strain that follows primary 6. A characteristic relationship between strain rate
consolidation. The rate of secondary compression is controlled by and time exists for most soils, as shown, for
the viscous resistance of the soil structure, whereas, the rate of pri-
mary consolidation is controlled by hydrodynamic lag, that is, how example, in Fig. 12.4 for drained triaxial com-
fast water can escape from the soil. pression creep of London clay (Bishop, 1966)
l
ria
ate
dM
hte
rig
py
Figure 12.2 Sustained stress creep curves illustrating different forms of strain vs. logarithm
Co
of time behavior.
and Fig. 12.5 for undrained triaxial compression logarithm of time. The slope of this relationship
creep of soft Osaka clay (Murayama and Shi- is essentially independent of the creep stress;
bata, 1958). At any stress level (shown as a per- increases in stress level shift the line vertically
centage of the strength before creep in Fig. 12.4 upward. The slope of the log strain rate versus
and in kg/cm2 in Fig. 12.5), the logarithm of the log time line for drained creep is approximately
strain rate decreases linearly with increase in the ⫺1. Undrained creep often results in a slope be-
l
ria
ate
dM
hte
rig
Figure 12.3 Variation of creep strain rate with deviator stress for undrained creep of re-
molded illite.
tween ⫺0.8 and ⫺1 for this relationship. The ditions are also susceptible to creep rupture due
py
onset of failure under higher stresses is signaled to softening associated with the increase in wa-
by a reversal in slope, as shown by the topmost ter content by dilation and swelling.
curve in Fig. 12.5. 8. Although stress relaxation has been less studied
7. Pore pressure may increase, decrease, or remain than creep, it appears that equally regular pat-
Co
constant during creep, depending on the volume terns of deformation behavior are observed, for
change tendencies of the soil structure and example, Larcerda and Houston (1973).
whether or not drainage occurs during the de- 9. Deformation under sustained stress ordinarily
formation process. In general, saturated soft produces an increase in stiffness under the ac-
sensitive clays under undrained conditions are tion of subsequent stress increase, as shown
most susceptible to strength loss during creep schematically in Fig. 12.6. This reflects the
due to reduction in effective stress caused by time-dependent structural readjustment or ‘‘ag-
increase in pore water pressure with time. Heav- ing’’ that follows changes in stress state. It is
ily overconsolidated clays under drained con- analogous to the quasi-preconsolidation effect
l
ria
ate
dM
Figure 12.4 Strain rate vs. time relationships during drained creep of London clay (data
from Bishop, 1966).
hte
⫽ 冉 冊
˙ 1
˙ 1(ref)
(12.1)
virgin compression line and the value of the pre- where Cc is the virgin compression index, Cr is
consolidation pressure, p, determined in the the recompression index and p(ref) is the pre-
laboratory are influenced by the rate of loading consolidation pressure at a reference strain rate
during one-dimensional consolidation (Graham ˙ 1(ref). In this equation, the rate effect increases
et al., 1983a; Leroueil et al., 1985). Thus, esti- with the value of ⫽ Ce /(Cc ⫺ Cr). The var-
py
mations of the overconsolidation ratio of clay iation of preconsolidation pressure with strain
deposits in the field are dependent on the load- rate is shown in Fig. 12.8 (Soga and Mitchell,
ing rates and paths used in laboratory tests for 1996). The data define straight lines, and the
determination of the preconsolidation pressure. slope of the lines gives the parameter . In gen-
Co
If it is assumed that the relationship between eral, the value of ranges between 0.011 and
strain and logarithm of time during compression 0.094. Leroueil and Marques (1996) report val-
is linear over the time ranges of interest and that ues between 0.029 and 0.059 for inorganic
the secondary compression index Ce is constant clays.
regardless of load, the rate-dependent precon- 11. The undrained strength of saturated clay in-
solidation pressure p at ˙ 1 can be related to the creases with increase in rate of strain, as shown
axial strain rate as follows (Silvestri et al, 1986; in Figs. 12.9 and 12.10. The magnitude of the
Soga and Mitchell, 1996; Leroueil and Marques, effect is about 10 percent for each order of mag-
1996): nitude increase in the strain rate. The strain rate
l
ria
ate
dM
Figure 12.5 Strain rate vs. time relationships during undrained creep of Osaka alluvial clay
(Murayama and Shibata, 1958).
qƒ
qƒ(ref)
⫽ 冉 冊
˙ 1
˙ 1(ref)
(12.2)
12.3 TIME-DEPENDENT
DEFORMATION–STRUCTURE INTERACTION
Figure 12.6 Effect of sustained loading on (a) stress–strain In reality, completely smooth curves of the type shown
and strength behavior and (b) one-dimensional compression in the preceding figures for strain and strain rate as a
behavior. function of time may not exist at all. Rather, as dis-
l
ria
ate
dM
hte
rig
py
Co
l
ria
ate
Figure 12.8 Strain rate dependence on preconsolidation pressure determined from one-
dimensional constant strain rate tests (Soga and Mitchell, 1996).
dM
hte
rig
py
Figure 12.9 Effect of strain rate on undrained strength (Kulhawy and Mayne 1990). Re-
printed with permission from EPRI.
Co
cussed by Ter-Stepanian (1992), a ‘‘jump-like structure surmounting energy barriers, (2) mutual displacement
reorganization’’ may occur, reflecting a stochastic char- of particles as a result of bond failures, but without
acter for the deformation, as shown in Fig. 12.11 for rearrangement, (3) the structural level of soil defor-
creep of an undisturbed diatomaceous, lacustrine, ov- mation involving mutual rearrangements of particles,
erconsolidated clay. Ter-Stepanian (1992) suggests that and (4) deformation at the aggregate level. Behavior at
there are four levels of deformation: (1) the molecular levels 3 and 4 is discussed below; that at levels 1 and
level, which consists of displacement of flow units by 2 is treated in more detail in Section 12.4.
l
ria
ate
Figure 12.10 Strain rate dependence on undrained shear strength determined using constant
strain rate CU tests (Soga and Mitchell, 1996).
atomaceous, lacustrine, overconsolidated clay (from Ter- considers sliding at interparticle contacts to be visco-
Stepanian, 1992). frictional. The rate at which sliding of two particles
relative to each other occurs depends on the ratio of
shear to normal force at their contact. The relationship
between rate and force is formulated in terms of rate
py
Time-Dependent Process of Particle Rearrangement process theory (see Section 12.4), and the mechanistic
Creep can lead to rearrangement of particles into more representations of the contact normal and shear forces
stable configurations. Forces at interparticle contacts are shown in Fig. 12.12. The time-dependent compo-
have both normal and tangential components, even if nent in the tangential forces model is given as a ‘‘sinh-
Co
the macroscopic applied stress is isotropic. If, during dashpot’’.2 The average magnitudes of both normal and
the creep process, there is an increase in the proportion
of applied deviator stress that is carried by interparticle 2
Kuhn (1987) used the following equation for rate of sliding at a
normal forces relative to interparticle tangential forces, contact:
then the creep rate will decrease. Hence, the rate at
which deformation level 3 occurs need not be uniform Ẋ ⫽
2kT
h 冉 冊 冉
exp ⫺
F
RT
sinh
ƒt
2kTn1 ƒn 冊
owing to the particulate nature of soils. Instead it will
where n1 is the number of bonds per unit of normal force, ƒt is the
reflect a series of structural readjustments as particles tangential force and ƒn is the normal force. The others are parameters
move up, over, and around each other, thus leading to related to rate process theory as described in Section 12.4.
l
ria
ate
dM
Figure 12.12 Normal and tangential interparticle force mod-
els according to Kuhn (1987).
creep rate.
A numerical analysis of an irregular packing of cir-
cular disks using the sinh-dashpot representation gives
creep behavior comparable to that of many soils as
shown in Fig. 12.13 (Kuhn and Mitchell, 1993). The
py
60
modulus–time response are evident. During primary
consolidation, values of the shear modulus increase
rapidly at the beginning and begin to level off as the
40
excess pore pressure dissipates. After the end of pri-
l
mary consolidation, the modulus increases linearly
ria
夹 with the logarithm of time during secondary compres-
20
After
sion.
5 days After The expected change in shear modulus due to void
Before
夹
290 s Test
䊉
ratio change during secondary compression can be es-
0夹 䊉
timated using the following empirical formula for
ate
䊉
0 100 200 300 400
Sieve Size (μm) shear modulus as a function of void ratio and confining
pressure (Hardin and Black, 1968):
Figure 12.14 Changes in particle size distribution of sand
before loading and after two different load durations (from (2.97 ⫺ e)2 0.5
Leung, et al., 1996). G⫽A p (12.3)
dM 1⫹e
producing fines. The fines fill the voids between larger ondary time-dependent increase in shear modulus. This
particles and crushed particles progressively rear- aging effect has been recorded for a variety of mate-
ranged themselves with time. rials, ranging from clean sands to natural clays (Afifi
and Richart, 1973; Kokusho, 1987; Mesri et al., 1990,
Aging—Time-Dependent Strengthening of Soil and many others). Further discussion of aging phenom-
rig
schaeffl (1964) showed that this increase in preconso- with time after load application in one-dimensional
lidation pressure cannot be accounted for in terms of compression were measured by Bowman and Soga
the void ratio decrease during the sustained compres- (2003). Resin was used to fix sand particles after var-
sion period. Time-dependent changes of these types are ious loading times. Pluviation of the sand produced a
Co
a consequence of ‘‘aging’’ effects, which alter the horizontal preferred particle orientation of soil grains,
structural state of the soil. The fabric obtained by creep and increased vertical loading resulted in a greater ori-
may be different from that caused by increase in stress, entation of particle long axes parallel to the horizontal,
even though both samples arrive at the same void ratio. which is in agreement with the findings of Oda (1972a,
Leroueil et al. (1996) report a similar result for an ar- b, c), Mitchell et al. (1976), and Jang and Frost (1998).
tificially sedimented clay from Quebec, as shown in Over time, however, the loading of sand caused parti-
Fig. 12.15a. They also measured the shear wave ve- cle long axes to rotate toward the vertical direction
locities after different times during the tests using (i.e., more isotropic fabric).
bender elements and computed the small strain elastic Experimental evidence (Bowman and Soga, 2003)
shear modulus. Figure 12.15b shows the change in showed that large voids became larger, whereas small
shear modulus with void ratio. voids became smaller, and particles group or cluster
A A
2.6 Normal
Consolidation Line Stiffness Change
During the Primary
Consolidation
Between B and C
2.4 B B
Quasi-
preconsolidation
Void Ratio e
Pressure
l
2.2 C C
ria
Creep for
120 days
D D
Increase in Stiffness
2.0 Destructuring E during Creep (C-D) E
ate
State
F F
1.8
4 6 8 10 20 0.5 1 2 5
dM Vertical Effective Stress σv (kPa) Small-strain Stiffness G0 (MPa)
(a ) (b )
Figure 12.15 (a) Compression curve and (b) variation of the maximum shear modulus G0
with void ratio for artificially sedimented Jonquiere clay (from Leroueil et al., 1996).
together with time. Based on these particulate level a strengthening of physicochemical bonds between
hte
findings, it appears that the movements of particles particles. To illustrate this, Nakagawa et al. (1995) ex-
lead to interlocking zones of greater local density. The amined the physicochemical interactions between clays
interlocked state may be regarded as the final state of and pore fluid using a special consolidometer in which
any one particle under a particular applied load, due to the sample resistivity and pore fluid conductivity could
kinematic restraint. The result, with time, is a stiffer, be measured. Shear wave velocities were obtained us-
rig
more efficient, load-bearing structure, with areas of rel- ing bender elements to determine changes in the stiff-
atively large voids and neighboring areas of tightly ness characteristics of the clay during consolidation.
packed particles. The increase in stiffness is achieved Kaolinite clay mixed with saltwater was used for the
by shear connections obtained by the clustering. Then, experiment, and changes in shear wave velocities and
when load is applied, the increased stiffness and electrical properties were monitored during the tests.
py
strength of the granular structure provides greater re- The test results showed that the pore fluid compo-
sistance to the load and the observed aging effect is sition and ion mobility changed with time. At each
seen. The numerical analysis in Kuhn and Mitchell load increment, as the effective stress increased with
(1993) led to a similar hypothesis for how a more pore pressure dissipation, the shear wave velocities,
Co
‘‘braced’’ structure develops with time. For load appli- and therefore the shear modulus, generally increased
cation in a direction different to that during the aging with time as shown in Fig. 12.17. It may be seen, how-
period, however, the strengthening effect of aging may ever, that in some cases, the shear wave velocities at
be less, as the load-bearing particle column direction the beginning of primary consolidation decreased
differs from the load direction. slightly from the velocities obtained immediately be-
fore application of the incremental load, probably as a
Time-Dependent Changes in Physicochemical result of soil structure breakdown. During the subse-
Interaction of Clay and Pore Fluid quent secondary compression stage, the shear wave ve-
A portion of the shear modulus increase during sec- locity again increased. As was the case for the results
ondary compression of clays is believed to result from in Fig.12.16, the increases in shear wave velocity dur-
50 Ball Kaolinite
40
IG = ΔG per log time
= 6.2 MPa
30 Possible Change in G
by Void Ratio Decrease
Only Estimated Using
20 Eq. (12.3)
10
l
Primary Consolidation Secondary Compression
ria
0
10 1 10 2 10 3 10 4
Time (min)
Sample Height Change (mm)
0.5
ate
1.0
1.5 dM
Initial Consolidation Pressure = 70 kPa
Initial Void Ratio e 0 = 1.1
2.0
10 0 10 1 10 2 10 3 10 4
Time (min)
ter, increasing the electrical conductivity. With time, layed deformation of micropores in the clay aggregates
the conductivity decreased, suggesting that the released (Berry and Poskitt, 1972; Matsuo and Kamon, 1977;
ions are accumulating near particle surfaces. Some of Sills, 1995). The mobility of water in the micropores
these released ions are expelled from the specimen as is restricted due to small pore size and physicochem-
consolidation progressed as shown in Fig. 12.18b. A ical interactions close to the clay particle surfaces. Ak-
slow equilibrium under a new state of effective stress agi (1994) did compression tests on specially prepared
is hypothesized to develop that involves both small clay containing primarily Ca in the micropores and Na
particle rearrangements, associated with decrease in in the macropores. Concentrations of the two ions in
void ratio during secondary compression, and devel- the expelled water at different times after the start of
opment of increased contact strength as a result of pre- consolidation were consistent with this hypothesis.
Concept of Activation
The basis of rate process theory is that atoms, mole-
cules, and/or particles participating in a time-
dependent flow or deformation process, termed flow
units, are constrained from movement relative to each
l
other by energy barriers separating adjacent equilib-
ria
rium positions, as shown schematically by Fig. 12.19.
The displacement of flow units to new positions re-
quires the acquisition of an activation energy
F of
sufficient magnitude to surmount the barrier. The po-
ate
tential energy of a flow unit may be the same following
the activation process, or higher or lower than it was
initially. These conditions are shown by analogy with
the rotation of three blocks in Fig. 12.20. In each case,
an energy barrier must be crossed. The assumption of
dM a steady-state condition is implicit in most applications
to soils concerning the at-rest barrier height between
successive equilibrium positions.
The magnitude of the activation energy depends on
the material and the type of process. For example, val-
ues of
F for viscous flow of water, chemical reac-
tions, and solid-state diffusion of atoms in silicates are
about 12 to 17, 40 to 400, and 100 to 150 kJ/mol of
hte
l
ria
ate
dM
Figure 12.20 Examples of activated processes: (a) steady-state, (b) increased stability, and
(c) decreased stability.
hte
rig
py
Co
the average thermal energy per flow unit is kT, where The net frequency of activation in the direction of
k is Boltzmann’s constant (1.38 ⫻ 10⫺23 J K⫺1) and T the force then becomes
冉 冊 冉 冊
is the absolute temperature (K). Even in a material at
rest, thermal vibrations occur at a frequency given by kT
F ƒ
kT/h, where h is Planck’s constant (6.624 ⫻ 10⫺34 J ( →) ⫺ ( ←) ⫽ 2 exp ⫺ sinh
h RT 2kT
s⫺1). The actual thermal energies are divided among
the flow units according to a Boltzmann distribution. (12.8)
It may be shown that the probability of a given unit
becoming activated, or the proportion of flow units that
are activated during any one oscillation is given by Strain Rate Equation
At any instant, some of the activated flow units may
冉 冊
l
F successfully cross the barrier; others may fall back into
ria
p(
F) ⫽ exp ⫺ (12.4) their original positions. For each unit that is successful
NkT
in crossing the barrier, there will be a displacement .
The component of in a given direction times the
where N is Avogadro’s number (6.02 ⫻ 1023), and Nk
number of successful jumps per unit time gives the rate
is equal to R, the universal gas constant (8.3144 J K⫺1
of movement per unit time. If this rate of movement
ate
mol⫺1). The frequency of activation then is
is expressed on a per unit length basis, then the strain
冉 冊
rate ˙ is obtained.
kT ⫺
F Let X ⫽ F (proportion of successful barrier cross-
⫽ exp (12.5)
h NkT ings and ) such that
dM
In the absence of directional potentials, energy bar-
riers are crossed with equal frequency in all directions,
˙ ⫽ X[( →) ⫺ ( ←)] (12.9)
and no consequences of thermal activations are ob- Then from Eq. (12.8)
served unless the temperature is sufficiently high that
softening, melting, or evaporation occurs. If, however,
a directed potential, such as a shear stress, is applied,
then the barrier heights become distorted as shown by
˙ ⫽ 2X
kT
h 冉 冊 冉 冊
exp ⫺
F
RT
sinh
ƒ
2kT
(12.10)
hte
curve B in Fig. 12.21. If ƒ represents the force acting The parameter X may be both time and structure de-
on a flow unit, then the barrier height is reduced by an pendent.
amount (ƒ /2) in the direction of the force and in- If (ƒ /2kT) ⬍ 1, then sinh(ƒ /2kT) ⬇ (ƒ /2kT), and
creased by a like amount in the opposite direction, the rate is directly proportional to ƒ. This is the case
where represents the distance between successive for ordinary Newtonian fluid flow and diffusion where
equilibrium positions.3 Minimums in the energy curve
rig
冉 冊
For most solid deformation problems, however,
kT
F/N ⫺ ƒ /2 (ƒ /2kT) ⬎ 1 (Mitchell et al., 1968), so
→⫽ exp ⫺ (12.6)
h kT
冉 冊 冉 冊
Co
ƒ 1 ƒ
sinh ⬇ exp (12.12)
and in the opposite direction, the increased barrier 2kT 2 2kT
height decreases the activation frequency to
and
←⫽
kT
h
exp ⫺
kT 冉
F/N ⫹ ƒ /2
冊 (12.7)
˙ ⫽ X
kT
冉 冊 冉 冊
exp ⫺
F
exp
ƒN
(12.13)
h RT 2RT
l
uniformly among S flow units per unit area, then
ria
Soil Deformation as a Rate Process
Although there does not yet appear to be a rigorous
ƒ⫽ (12.17)
proof of the correctness of the detailed statistical me- S
chanics formulation of rate process theory, even for
ate
simple chemical reactions, the real behavior of many Displacement of a flow unit requires that interatomic
systems has been substantially in accord with it. Dif- or intermolecular forces be overcome so that it can be
ferent parts of Eq. (12.13) have been tested separately moved. Let it be assumed that the number of flow units
(Mitchell et al., 1968). It was found that the tempera- and the number of interparticle bonds are equal.
ture dependence of creep rate and the stress depend- If D represents the deviator stress under triaxial
ence of the experimental activation energy [Eq.
dM stress conditions, the value of ƒ on the plane of max-
(12.14)] were in accord with predictions. These results imum shear stress is
do not prove the correctness of the theory; they do,
however, support the concept that soil deformation is D
ƒ⫽ (12.18)
a thermally activated process. 2S
冉 冊 冉 冊
hte
is termed the experimental activation energy. For all creep rates and evaluations of the influences of differ-
conditions constant except T, and assuming that ent factors are made under conditions of equal struc-
X(kT/h) ⬇ constant ⫽ A, ture. The time dependency of creep rate and the
冉 冊
possible time dependencies of the parameters in Eq.
Co
soils at different temperatures but at the same structure known, /S is calculated as a measure of the number
are given by Mitchell et al. (1968, 1969). of interparticle bonds.4
Determination of Number of Bonds For stresses
large enough to justify approximating the hyperbolic Activation Energies for Soil Creep
sine function by a simple exponential in the creep rate Activation energies for the creep of several soils and
equation and small enough to avoid tertiary creep, the other materials are given in Table 12.1. The free energy
logarithm of strain rate varies directly with the deviator of activation for creep of soils is in the range of about
stress. For this case, Eq. (12.19) can be written 80 to 180 kJ/mol. Four features of the values for soils
in Table 12.1 are significant:
˙ ⫽ K(t) exp(D) (12.21)
1. The activation energies are relatively large, much
l
higher than for viscous flow of water.
where
ria
2. Variations in water content (including complete
冉 冊
drying), adsorbed cation type, consolidation pres-
kT
F sure, void ratio, and pore fluid have no significant
K(t) ⫽ X exp ⫺ (12.22)
h RT effect on the required activation energy.
3. The values for sand and clay are about the same.
ate
⫽ (12.23) 4. Clays in suspension with insufficient solids to
4SkT
form a continuous structure deform with an ac-
tivation energy equal to that of water.
Parameter is a constant for a given value of ef-
fective consolidation pressure and is given by the slope
dM
of the relationship between log strain rate and stress. 4
A procedure for evaluation of from the results of a test at a
It is evaluated using strain rates at the same time after succession of stress levels on a single sample is given by Mitchell
the start of creep tests at several stress intensities. With et al. (1969).
Activation Energy
Material (kJ/ mol)a Reference
1. Remolded illite, saturated, water contents of 105–165 Mitchell, et al. (1969)
30 to 43%
2. Dried illite: samples air-dried from 155 Mitchell, et al. (1969)
rig
l
sumption for is incorrect, calculated values of S will very large number of bonds developed by drying is
ria
still be in the same correct relative proportion as long responsible for the high dry strength of clay.
as remains constant during deformation. Overconsolidated Clay Samples of undisturbed
Normally Consolidated Clay Results of creep tests San Francisco Bay mud were prepared to overcon-
at different stress intensities for different consolidation solidation ratios of 1, 2, 4, and 8 following the stress
ate
pressures enable computation of S as a function of con- paths shown in the upper part of Fig. 12.25. The sam-
solidation pressure. Values obtained for undisturbed ple represented by the triangular data point was re-
San Francisco Bay mud are shown in Fig. 12.23. The molded after consolidation and unloading to point d,
open point is for remolded bay mud. An undisturbed where it had a water content of 52.3 percent. The un-
specimen was consolidated to 400 kPa, rebounded to drained compressive strength as a function of consol-
dM idation pressure is shown in the middle section of Fig.
12.25, and the number of bonds, deduced from the
creep tests, is shown in the lower part of the figure.
The effect of overconsolidation is to increase the num-
ber of interparticle bonds over the values for normally
consolidated clay. Some of the bonds formed during
consolidation are retained after removal of much of the
consolidation pressure.
hte
l
ria
ate
Figure 12.23 Number of interparticle bonds as a function of consolidation pressure for
normally consolidated San Francisco Bay mud.
consolidated clays.
7. Overconsolidation leads to more bonds than in
normally consolidated clay at the same effective
consolidation pressure.
8. Strength depends only on the number of bonds.
9. Remolding at constant water content causes a
decrease in the effective consolidation pressure,
which means also a decrease in the number of
bonds.
Figure 12.24 Number of bonds as a function of water con- 10. There are about 100 times as many bonds in dry
tent for illite. clay as in wet clay.
l
ria
ate
dM
hte
rig
py
Figure 12.25 Consolidation pressure, strength, and bond numbers for San Francisco Bay
mud.
Co
Although it may be possible to explain these results up to four hydrogen bonds with its neighbors. If the
in more than one way, the following interpretation ac- single-bond interpretation is also correct for soils, then
counts well for them. The energy
F activates a mole consistency in Eq. (12.10) requires that shear force ƒ
of flow units. The movement of each flow unit may pertain to the force per bond. On this basis, parameter
involve rupture of single bonds or the simultaneous S indicates the number of single bonds per unit area.
rupture of several bonds. Shear of dilute montmoril- In the event activation of a flow unit requires simul-
lonite–water pastes involves breaking single bonds taneous rupture of n bonds, then S represents 1/nth of
(Ripple and Day, 1966). For viscous flow of water, the the total bonds in the system.
activation energy is approximately that for a single hy- That the activation energy for deformation of soil is
drogen bond rupture per flow unit displacement, even well into the chemical reaction range (40 to 400 kJ/
though each water molecule may form simultaneously mol) does not prove that bonding is of the primary
l
ria
ate
Figure 12.26 Strength as a function of number of bonds for
San Francisco Bay mud.
Figure 12.28 Composite relationship between shear strength
dM and number of interparticle bonds (from Matsui and Ito,
1977). Reprinted with permission from The Japanese Society
valence type because simultaneous rupture of several of SMFE.
weaker bonds could yield values of the magnitude ob-
served. On the other hand, the facts that (1) the acti-
vation energy is much greater than for flow of water,
(2) it is the same for wet and dry soils, and (3) it is
essentially the same for different adsorbed cations and
hte
rig
py
Co
Figure 12.27 Strength as a function of number of bonds for dry Antioch River sand.
l
sult from slow diffusion of oxygen ions in and around For example, for a compressive strength of 3 kg/cm2
ria
interparticle contacts. The important minerals in both (⬇ 300 kPa) there are 8 ⫻ 1010 bonds/cm2 of shear
sand and clay are silicates, and their surface layers surface. Oxygen atoms on the surface of a silicate min-
consist of oxygen atoms held together by silicon at- eral have a diameter of 0.28 nm. Allowing an area 0.30
oms. Water in some form is adsorbed onto these sur- nm on a side for each oxygen gives 0.09 nm2, or 9 ⫻
10⫺16 cm2, per bonded oxygen for a total area of 9 ⫻
ate
faces. The water structure consists of oxygens held
together by hydrogen. It is not too different from that 10⫺16 ⫻ 8 ⫻ 1010 ⫽ 7.2 ⫻ 10⫺5 cm2 /cm2 of soil cross
of the silicate layer in minerals. Thus, a distinct bound- section.
ary between particle surface and water may not be dis-
cernable. Under these conditions, a more or less Hypothesis for Bonding, Effective Stress, and
continuous solid structure containing water molecules
dM Strength
that propagates through interparticle contacts can be Normal effective stresses and shear stresses can be
visualized. transmitted only at interparticle contacts in most soils.5
An individual flow unit could be an atom, a group The predominant effects of the long-range physico-
of atoms or molecules, or a particle. The preceding chemical forces of interaction are to control the initial
arguments are based on the interpretation that individ- soil fabric and to alter the forces transmitted at contact
ual atoms are the flow units. This is consistent with points from what they would be due to applied stresses
both the relative and actual values of S that have been alone.
hte
determined for different soils. Furthermore, by using a Interparticle contacts are effectively solid, and it is
formulation of the rate process equation that enabled likely that both adsorbed water and cations in the con-
calculation of the flow unit volume from creep test tact zone participate in the structure. An interparticle
data, Andersland and Douglas (1970) obtained a value contact may contain many bonds that may be strong,
of about 1.7 Å3, which is of the same order as that of approaching the primary valence type. The number of
rig
individual atoms. On the other hand, Keedwell (1984) bonds at any contact depends on the compressive force
defined flow units between quartz sand particles as transmitted at the contact, and the Terzaghi–Bowden
consisting of six O2⫺ ions and six Si4⫹ ions and be- and Tabor adhesion theory of friction presented in Sec-
tween two montmorillonite clay particles as consisting tion 11.4, can account for strength. The macroscopic
of four H2O molecules. strength is directly proportional to the number of
py
could give a 100-fold increase in the number of inter- during virgin compression, an overconsolidated soil at
particle contacts, as would have to be the case accord- a given effective stress has a greater number of bonds
ing to Fig. 12.27. A more plausible interpretation is and higher strength than a normally consolidated soil.
that drying, while causing some increase in the number This effect is more pronounced in clays than in sands
of interparticle contacts during shrinkage, causes because the larger and bulky sand grains tend to re-
mainly an increase in the number of bonds per contact
because of increased effective stress.
5
At any value of effective stress, the value of S is Pure sodium montmorillonite may be an exception since a part of
the normal stress can be carried by physicochemical forces of inter-
about the same for both sand and clay. The number of action. The true effective stress may be less than the apparent effec-
interparticle contacts should be vastly different; how- tive stress by R ⫺ A as discussed in Chapter 7.
cover their original shapes when unloaded, thus rup- From the relationships in Section 12.5, the following
turing most of the bonds in excess of those needed to relationship between bonds per unit area and effective
resist the lower stress. The strength of the interparticle stress is suggested.
contacts can vary over a wide range, depending on the
number of bonds per contact. S ⫽ a ⫹ bƒ (12.28)
The unique relationship between strength and num-
ber of bonds for all soils, as indicted by Fig. 12.28, where a and b are constants and ƒ is the effective
reflects the fact that the minerals comprising most soils normal stress on the shear plane. Thus, Eq. (12.27)
are silicates, and they all have similar surface struc- becomes
tures.
In the absence of chemical cementation, interparticle
⫽
2a
F 2akT ˙
⫹ ln ⫹
2b
F 2bkT ˙
⫹ 冉 ƒ 冊
l
bonds may form in response to interparticle contact ln
N B N B
ria
forces generated by either applied stresses, physico-
chemical forces of interaction, or both. Any bonds ex- (12.29)
isting in the absence of applied effective stress, that is,
when ⫽ 0, are responsible for true cohesion. There Equation (12.29) is of the same form as the Cou-
should be no difference between friction and cohesion lomb equation for strength:
ate
in terms of the shearing process. Complete failure in
shear involves simultaneous rupture or slipping of all ⫽ c ⫹ ƒ tan (12.30)
bonds along the shear plane.
By analogy,
2a
F 2akT ˙
dM
12.6 SHEARING RESISTANCE AS A RATE
PROCESS
c⫽
N
⫹
ln
B
(12.31)
2b
F 2bkT ˙
Deformation at large strain can approach a steady-state tan ⫽ ⫹ ln (12.32)
N B
condition where there is little further structural change
with time (such as at critical state). In this case, Eq.
These equations state that both cohesion and friction
(12.19) can be used to describe the shearing resistance
hte
冉 冊 冉 冊
py
冉 冊
increase in water content. This follows directly from
kT
F Eq. (12.29) because
ln ˙ ⫽ ln X ⫺ ⫹ (12.26)
h RT 2SkT
d 2akT 2bkT 2kT
⫽ ⫹ ƒ ⫽ (a ⫹ bƒ)
By assuming X(kT/h) is a constant equal to B d ln(˙ /B)
(Mitchell, 1964), Eq. (12.26) can be rearranged to give (12.33)
⫽
2S
N
F ⫹
2SkT
ln
˙
B 冉冊 (12.27) that is, d /d ln (˙ /B) is proportional to the number of
bonds, which decreases with increasing water content.
l
ria
ate
dM
hte
Figure 12.29 Effect of rate of shear on shearing resistance of remolded clays as determined
by the laboratory vane apparatus (prepared from the data of Karlsson, 1963).
rig
This interpretation of the data in Figs. 12.9 and 12.29 12.7 CREEP AND STRESS RELAXATION
assumes that the effective stress was unaffected by
changes in the strain rate, which may not necessarily Although the designation of a part of the strain versus
time relationship as steady state or secondary creep
py
that the term (˙ /B) is less than one (Mitchell, 1964). tured state is likely to persist only during deformation
Thus the quantity ln(˙ /B) in Eq. (12.29) is negative, at a constant rate, that is, at failure. This state is often
and an increase in temperature should give a decrease called ‘‘steady state,’’ in which the soil is deforming
in strength, all other factors being constant. That this continuously at constant volume under constant shear
is the case is demonstrated by Fig. 12.30, which shows and confining stresses (Castro, 1975; Castro and
deviator stress as a function of temperature for samples Poulos, 1977).
of San Francisco Bay mud compared under conditions Otherwise, bond making and bond breaking occur
of equal mean effective stress and structure. Other ex- at different rates as a result of different internal time-
amples of the influence of temperature on strength are and strain-dependent phenomena, which might include
shown in Figs. 11.6 and 11.133. thixotropic hardening, viscous flows of water and ad-
l
ria
ate
Figure 12.30 Influence of temperature on the shearing re-
sistance of San Francisco Bay mud. Comparison is for sam-
ples at equal mean effective stress and at the same structure.
dM
sorbed films, chemical, and biological transformations,
and the like. Furthermore, distortions of the soil struc-
ture and relative movements between particles cause
changes in the ratio of tangential to normal forces at
interparticle contacts that may be responsible for large
changes in creep rate. Because of these time depend-
encies some of the parameters in Eq. (12.19) may be
hte
range of stresses, a nearly linear relationship is found active the clay, the more important are stress relaxation
between logarithm of strain rate and stress, also as pre- and creep, as illustrated by Figs. 4.22 and 4.23, where
dicted by Eq. (12.10) for the case where the argument creep rates, approximated by steady-state values, are
of the hyperbolic sine is greater than 1. At stresses related to clay type, clay content, and plasticity. Time-
approaching the strength of the material, the strain rate dependent deformations are more important at high
becomes very large and signals the onset of failure. water contents than at low. Deviatoric creep and sec-
Other examples of the relationships between logarithm ondary compression are greater in normally consoli-
of strain rate and creep stress corresponding to differ- dated than overconsolidated soils.
ent times after the application of the creep stress are Although the magnitude of creep strains and strain
given in Fig. 12.31 for drained tests on London clay rates may be small in sand or dry soil, the form of the
l
usually referred to as deviatoric creep or shear creep.
ria
Time-dependent deformations under constant stress re-
ferred to as volumetric creep. Secondary compression
is a special case of volumetric creep.
Deviatoric creep is often accompanied by volumetric
ate
creep. The ratio of volumetric to deviatoric creep fol-
0.6
D
0.4 90 min
hte
0.2 20 min
C
2 min
Figure 12.32 Variation of creep strain rate with deviator 0
stress for undrained creep of normally consolidated San Fran- A B 0.5 1.0 1.5 2.0 2.5 3.0 3.5
cisco Bay mud. Deviator strain (%)
(a)
rig
crease the creep rate even though the basic mechanism 0.6
of creep is the same for dry and wet materials (Losert
et al., 2000). Takei et al. (2001) showed that the de- 0.4
velopment of creep strains due to time-dependent
Co
lows a plastic dilatancy rule. Walker (1969) inves- history, with some samples contracting or dilating
tigated the time-dependent change of these two com- (Lade and Liu, 1998; Ahn-Dan et al., 2001). Some
ponents from incremental drained triaxial creep tests dense sand samples contract initially but then dilate
on normally consolidated kaolinite. The increase in with time (Bowman and Soga, 2003). Further discus-
shear strains with increase in volumetric strains at dif- sion of the creep behavior of sands in relation to me-
ferent times is shown in Fig. 12.33a. At the beginning chanical aging phenomena is given in Section 12.11.
of the triaxial test, the deviator stress was instantane- The fundamental process of creep strain develop-
ously increased from 344 to 377 kPa and kept constant. ment is therefore similar to that of time-independent
After an immediate increase in shear strains at constant plastic strains, and the same framework of soil plastic-
volume (AB in Fig. 12.33a), section BD corresponds ity can possibly be used. It can be argued whether
to primary consolidation that is controlled by the dis- it is necessary to separate the deformation into
l
sipation of pore pressures. After point D, creep oc- time-dependent and independent components. Rate-
ria
curred, and the ratio of volumetric to deviatoric strains independent behavior can be considered as the limiting
was independent of time. This ratio decreased with in- case of rate-dependent behavior at a very slow rate of
creasing stress ratio as shown in Fig. 12.33b. This ob- loading.
Volumetric-deviatoric creep coupling implies that
servation led to the time-dependent flow rule, which is
rapid application of a stress or a strain invariably re-
ate
similar to the dilatancy rule described in Section 11.20.
sults in rapid change of pore water pressures in a sat-
Sand deforms with time in a similar manner. Under urated soil under undrained conditions. For a constant
progressive deviatoric creep, the volumetric creep re- total minor principal stress, the magnitude of the pore
sponse is highly dependent on density, the stress level, pressure change depends on the volume change ten-
and the stress path before creep. The rate of both vol- dencies of the soil when subjected to shear distortions.
umetric and deviatoric creep increases with confining
dM These tendencies are, in turn, controlled by the void
pressure, particularly after particle crushing becomes ratio, structure, and effective stress, and can be quan-
important at high stresses (Yamamuro and Lade, 1993). tified in terms of the pore pressure parameter A as dis-
For dense sand under high deviator stress, dilative cussed in Chapters 8 and 10. An example showing pore
creep is observed (Murayama et al., 1984; Mejia et al., pressure increase with time for consolidated undrained
1988). The volumetric response of dense sand and creep tests on illite at several stress intensities is shown
gravel with time is a highly complex function of stress in Fig. 12.34. Figure 12.35 shows a slow decrease in
hte
rig
py
Co
Figure 12.34 Pore pressure development with time during undrained creep of illite.
l
ria
ate
Figure 12.35 Normalized pore pressure vs. time relationships during creep of kaolinite.
there are chemical, biological, or mineralogical Samples were normally consolidated both isotropically
changes during the creep period. This is illustrated by and under K0 conditions to the same vertical effective
the stress paths shown schematically in Fig. 12.37, stress. Samples consolidated isotropically were tested
where the pre- and postcreep strengths fall on the same in triaxial compression. Coefficient K0 consolidation
failure envelope. was used for both K0 triaxial and plane strain tests.
The results shown in Fig. 12.40 indicate that the pre-
rig
stresses corresponding to specific values of strain de- preventing strain in one horizontal direction and/or the
crease at higher temperature. These effects are illus- intermediate principal stress were not factors of major
trated by the data shown in Figs. 12.38 and 12.39. importance for this soil under the test conditions used.
Co
Effects of Test Type, Stress System, and Stress Path Interaction Between Consolidation and Creep
Most measurements of time-dependent deformation Experimental evidence suggests that creep occurs dur-
and stress relaxation in soils have been done on sam- ing primary consolidation (Leroueil et al., 1985; Imai
ples consolidated isotropically and tested in triaxial and Tang, 1992). Following the initial large change
compression or by measurement of secondary com- following load application, the pore pressure may ei-
pression in oedometer tests. However, most soils in ther dissipate, with accompanying volume change if
nature have been subjected to an anisotropic stress his- drainage is allowed, or change slowly during creep or
tory, and deformation conditions conform more to stress relaxation, if drainage is prevented. The devel-
plane strain than triaxial compression in many cases. opment of complete effective stress and void ratio
Some investigations of these factors have been made. equilibrium may take a long time. One illustration of
40
1
l
ria
30
Effective Stress State
After 1,000 min Creep
20
After 20,000 min Creep
ate
Effective Stress Path
10 of Undrained Creep
0
dM 0 10 20 30 40 50 60 70 80
Mean Pressure p (kPa)
(a)
400
Total Stress Path of
Possible Critical
Triaxial Compression Test
State Line
320 3
Deviator Stress q (kPa)
1
rig
240
Effective Stress State
After 1,000 min Creep
160
After 20,000 min Creep
py
0
Co
(b)
Figure 12.36 Measured stress paths of undrained creep tests of San Francisco Bay mud.
Initial confining pressure: (a) 49 kPa and (b) 392 kPa (from Arulanandan et al., 1971).
l
ria
ate
Figure 12.37 Effects of undrained creep on the strength of normally consolidated clay.
dM
hte
rig
Figure 12.38 Creep curves for Osaka clay tested at different temperatures—undrained tri-
axial compression (Murayama, 1969).
py
this is given by Fig. 10.5, where it is shown that the of the other tests was influenced by a pore pressure
relationship between void ratio and effective stress is that contained a contribution from the prior consoli-
dependent on the time for compression under any dation history.
Co
given stress. Another is given by Fig. 12.41, which The magnitude and rate of pore pressure develop-
shows pore pressures during undrained creep of San ment if drainage is prevented following primary con-
Francisco Bay mud. In each sample, consolidation un- solidation depend on the time allowed for secondary
der an effective confining pressure of 100 kPa was al- compression prior to the prevention of further drain-
lowed for 1800 min prior to the cessation of drainage age. This is illustrated by the data in Fig. 12.42, which
and the start of a creep test. The consolidation period show pore pressure as a function of time for samples
was greater than that required for 100 percent primary that have undergone different amounts of secondary
consolidation. The curve marked 0 percent stress level compression.
refers to a specimen maintained undrained but not sub- In summary, creep deformation depends on the ef-
ject to a deviator stress. This curve indicates that each fective stress path followed and any changes in stress
l
ria
ate
Figure 12.39 Influence of temperature on the initial and final stresses in stress relaxation
tests on Osaka clay—undrained triaxial compression (Murayama, 1969).
dM
hte
rig
py
Co
Figure 12.40 Creep curves for isotropically and K0-consolidated samples of undisturbed
Haney clay tested in triaxial and plane strain compression (from Campanella and Vaid, 1974).
Reproduced with permission from the National Research Council of Canada.
l
ria
Figure 12.41 Pore pressure development during undrained creep of San Francisco Bay mud
ate
after consolidation at 100 kPa for 1800 min (from Holzer et al., 1973). Reproduced with
permission from the National Research Council of Canada.
dM
hte
rig
Figure 12.42 Pore pressure development under undrained conditions following different
periods of secondary compression (from Holzer et al., 1973). Reproduced with permission
from the National Research Council of Canada.
py
with time. Furthermore, time-dependent volumetric re- is necessary to reconcile the time-dependent deforma-
sponse is governed both by the rate of volumetric creep tions observed in the field and laboratory.
Co
(σ1 – σ3)/2σ1c
Belfast Clay 4 m
l
The undrained shear strength and apparent precon- σ1c = σv0
ria
solidation pressure of soils decrease with decreasing 0.3
R
strain rate or increasing duration of testing. Preconsol-
idation pressures obtained from one-dimensional con-
solidation tests and undrained shear strengths obtained 16%/h
from triaxial tests are just two points on a soil’s yield 0.2
ate
1%/h
envelope in stress space. For a given metastable soil 0.25%/h
structure, the degree of rate dependency of preconsol-
idation pressure is similar to that of undrained shear Winnipeg Clay 11.5 m
0.1 σ1c > σv0
strength (Soga and Mitchell, 1996). If the apparent pre-
consolidation pressure depends on the strain rate at
dM CAU Triaxial Compression Tests
which the soil is deforming, then the same analogy can Relaxation Tests (R)
be expanded to the assumption that the size of the en- 0
0 4 8 12 16 20
tire yield envelope is also strain rate dependent (Tav-
enas and Leroueil, 1977). Figure 12.44 shows a family Axial Strain
of strength envelopes corresponding to constant strain Effective Stress σv (kPa)
rates7 obtained from drained and undrained creep tests 0 50 100 150 200 250
on stiff plastic Mascouche clay from Quebec (Leroueil 0
. .
hte
εv2
rig
εv1
at the maximum ratio of 1 / 3. The data were obtained
by various undrained tests, and a unique failure line
can be observed. Figure 12.45b shows the undrained 15 .
εv1
stress paths and the critical state line of reconstituted
.
py
mixtures of sand and clay with plasticity indices rang- SP1 test εv2
ing from 10 to 30 (Nakase and Kamei, 1986). A unique 20 SP2 test .
εv3
critical state line can be observed although the rates of .
εv1 = 2.70 & 10–6 s–1
. .
εv1
shearing are different. The change in undrained shear εv2 = 1.05 & 10–7 s–1 .
.
εv3 = 1.34 & 10–5 s–1 εv1
strength with strain rate results from a difference in 25
Co
l
ria
ate
dM
Figure 12.44 Influence of strain rate on the yield surface of Masouche clay (from Leroueil
and Marques, 1996).
hte
Excess Pore Pressure Generation in Normally stress relaxation tests in which the axial strain is kept
Consolidated Clays constant.
Excess pore pressure development depends primarily
on the collapse of soil structure. Accordingly, strain is Overconsolidated Clays
rig
the primary factor controlling pore pressure generation. Rate dependency of undrained shear strength decreases
This is shown in Fig. 12.46 by the undrained stress– with increasing overconsolidation, since there is no
strain–pore pressure response of normally consolidated contraction or collapse tendency observed during creep
natural Olga clay (Lefebvre and LeBouef, 1987). The of heavily overconsolidated clays. Sheahan et al.
natural clay specimens were normally consolidated un- (1996) prepared reconstituted specimens of Boston
py
der consolidation pressures larger than the field over- blue clay at different overconsolidation ratios and
burden pressure and then sheared at different strain sheared them at different strain rates in undrained con-
rates. Although the deviator stress at any strain in- ditions. Figure 12.47 shows that the undrained stress
creases with increasing strain rate, the pore pressure path and the strength were much more strain rate de-
Co
versus strain curves are about the same at all strain pendent for lightly overconsolidated clay (OCR ⫽ 1
rates. At a given deviator stress, the pore pressure gen- and 2) than for more heavily overconsolidated clay
eration was larger at slower strain rates as a result of (OCR ⫽ 4 and 8). The results also show that the
more creep under slow loading. This is consistent with strength failure envelope is independent of strain rate
the observation made in connection with undrained as discussed earlier.
creep of clays as discussed in Section 12.7. Strain- The strain rate effects on stress–strain–pore pressure
driven pore pressure generation was also suggested by response of overconsolidated structured Olga clays are
Larcerda and Houston (1973) who showed that pore shown in Fig. 12.48 (Lefebvre and LeBouef, 1987).
pressure does not change significantly during triaxial The natural samples were reconsolidated to the field
0.4
Const. Stress Creep
Const. Rate of Strain Shear
Const. Rate of Loading Shear
0.3
Aged Samples
Const. Load Creep
(σ1 – σ3)/2σ1c
Step Creep
0.2 Thixotropic Hardened
l
0.1
ria
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
(σ1 + σ3)/2σ1c
ate
(a)
1.0 1.0
0.8 dM 0.8
e
e
Lin
Lin
0.6 0.6
ate
ate
St
St
e
al-
in
al-
M-15. e
-L in M-10
itic
itic
0.4 0.4 -L .
K0 ε(%/min) ε(%/min)
Cr
Cr
. K0
7&10-1: ε. 1 .
q/σvc
q/σvc
7&10-1: ε. 1
0.2 7&10 : ε2
-2 0.2 7&10-2: ε2
. .
hte
7&10-3: ε3 7&10-3: ε3
0 0
Cr
Cr
–0.2 –0.2
itic
itic
al-
al-
St
St
ate
ate
rig
–0.4 –0.4
Lin
Lin
e
–0.6 –0.6
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
p/σvc p/σvc
py
M15 Soil (Plasticity Index = 15) (b) M10 Soil (Plasticity Index = 10)
Figure 12.45 Strain rate independent failure line: (a) Haney clay (from Vaid and Campa-
nella, 1977) and (b) reconstituted mixtures of sand and clay (from Nakase and Kamei, 1986).
Co
overburden pressure. The deformation is brittle, with generation becomes rate dependent. This is due to local
strain softening indicating development of localized drainage within the specimens as the deformation be-
shear failure planes. Up to the peak stress, the response comes localized. As the time to failure increased, there
follows what has been described previously, that is the is more opportunity for local drainage toward the di-
stress–strain response is rate dependent and the pore lating shear band and the measured pore pressure may
pressure generation is strain dependent but indepen- not represent the overall behavior of the specimens.
dent of rate. However, after the peak, the pore pressure The difference in softening due to swelling at the fail-
100 50
Deviator Stress q (kPa)
40
80
30
Axial Strain Rate
60 0.1 %/hr
20 0.5 %/hr
Axial Strain Rate 2.5 %/hr
l
40 0.1 %/hr 10 12.3 %/hr
ria
0.5 %/hr
2.6 %/hr 0
1 2 3 4 5 6 7 8
20 12.3 %/hr Axial Strain (%)
20
Excess Pore Pressure Δu (kPa)
0
1 2 3 4 5 6 7 8
ate
Axial Strain Rate
140 Axial Strain (%) 0.1 %/hr
120 10 0.5 %/hr
2.6 %/hr
100 12.3 %/hr
80 Axial Strain Rate
0.1 %/hr 0
60 1 2 3 4 5 6 7 8
0.5 %/hr Axial Strain (%)
40
20
dM 2.6 %/hr
12.3 %/hr Figure 12.48 Stress–strain and pore pressure–strain curves
0 for overconsolidated Olga clay (from Lefebvre and LeBouef,
1 2 3 4 5 6 7 8
Axial Strain (%) 1987).
0.4
Axial Strain Rate = 5.0 %/hr Similar rate-dependent stress–strain behavior is ob-
Axial Strain Rate = 49 %/hr
served in sands (Lade et al., 1997), but the effects are
OCR=1
quite small in many cases (Tatsuoka et al., 1997; Di
py
0.3 Large
OCR=8 OCR=4
OCR=2
Rate
Effect
Benedetto et al., 2002). An example of time depend-
Negligible OCR=1 ency observed for drained plane strain compression
0.2 Rate Effect
tests of Hostun sand is shown in Fig. 12.49 (Matsushita
et al., 1999). The stress–strain curves for three differ-
Co
0.1
OCR=2
ent strain rates (1.25 ⫻ 10⫺1, 1.25 ⫻ 10⫺2, and 1.25 ⫻
OCR=4
10⫺3 %/min) are very similar, indicating very small
0.0
0.2 0.4 0.6 0.8
rate effects when the specimens are sheared at a con-
OCR=8 (σa + σr)/2σvm stant strain rate. On the other hand, the change from
one rate to another temporarily increases or decreases
-0.1
the resistance to shear. The influence of acceleration
Figure 12.47 Rate dependency stress path and strength of rather than the rate is reflected by the significant creep
overconsolidated Boston blue clay (from Sheahan et al., deformation and stress relaxation of this rate-
1996). insensitive material as shown the figure. This is differ-
CRS at
0.125%/min
0.125%/min
200
5.0
Creep Stress Relaxation
4.5 CRS at 0.00125%/min
Creep
4.0
CRS at 0.125%/min 100
l
3.5 Creep
ria
Accidental Pressure Drop Followed
CRS at 0.125%/min by Relaxation Stage
3.0
0 1 2 3 4 5 6 7 8
Shear Strain γ = εa – εr (%) 0
10-3 10-2 10-1 100
Axial Strain, εa or
ate
Figure 12.49 Creep and stress relaxation of Hostun sand Single Amplitude of Cyclic Axial Strain, (εa)SA (%)
(from Matsushita et al., 1999).
Figure 12.50 Clay stiffness degradation curves at three
strain rates (from Shibuya et al., 1996).
ent from the observations made for clays as shown in
dM
Fig. 12.43 in which a unique stress–strain–strain rate
relationship was observed. Hence, the modeling of
stress–strain–rate behavior of sands appears to be
more complicated than that of clays, and further in-
vestigation is needed, as time-dependent behavior of
sands can be of significance in geotechnical construc-
tion as discussed further in Section 12.10.
hte
G
G( ) ⫽ (12.34)
log ˙ G( , ˙ ref)
l
and machine foundations are in the range of 10 to 100 nounced at larger strain, that is, as plastic deformations
ria
Hz. Similarly to monotonic loading, the effect of load- become more significant. Hence, it is possible that the
ing frequency on shear modulus degradation is small fundamental mechanisms of rate dependency are dif-
in clean, coarse-grained soils (Bolton and Wilson, ferent at small elastic strain levels than at larger plastic
1989; Stokoe et al., 1995), but the effect becomes more strains.
Small strain damping shows more complex fre-
ate
significant in fine-grained soils (Stokoe et al., 1995;
d’Onofrio et al., 1999; Matesic and Vucetic, 2003; quency dependency, as shown in Fig. 12.53 (Shibuya
Meng and Rix, 2004). An example of frequency effects et al., 1995; Meng and Rix, 2004). At a frequency of
on a shear modulus degradation curve for a clay ob- more than 10 Hz, the damping ratio increases with
tained from cyclic loading is shown in Fig. 12.50 along increased frequency, possibly due to pore fluid viscos-
with the monotonic data. Figure 12.52 shows the effect
dM ity effects. As the applied frequency decreases, the
of frequency on shear modulus of several soils at very damping ratio decreases. However, at a frequency less
small shear strains (less than 10⫺3 percent) measured than 0.1 Hz, the damping ratio starts to increase with
by torsional shear and resonant column apparatuses decreasing frequency. This may result from creep of
(Meng and Rix, 2004). The effect is 10 percent in- the soil (Shibuya et al., 1995).
crease per log cycle at most.
At a given frequency of cyclic loading, the strain
12.9 MODELING OF STRESS–STRAIN–TIME
rate applied to a soil increases with applied shear strain
hte
BEHAVIOR
as shown by the equation below:
Constitutive models are needed for the solution of geo-
˙ ⫽ 4ƒ c (12.35) technical problems requiring the determination of de-
rig
250
Vallencca Clay
200
Shear Modulus G (MPa)
py
100
Kaolin
Kaolin
Sandy Lean Clay
Figure 12.52 Rate dependency of cyclic small strain stiffness of a sandy elastic silt (from
Meng and Rix, 2004).
6
Clayey Subgrades
5
Augusta Clay
4 Sandy Lean Clay
Damping (%)
Pisa Clay
Vallencca Clay Fat Clay
3
Sandy Silty Clay
2
l
1 Sandy Elastic Silt
ria
Kaolin
ate
Figure 12.53 Effect of strain rate of damping ratio of soils (from Shibuya et al., 1995 and
Meng and Rix, 2004).
metric and deviatoric deformations in systems that are of 0.7 to 1.3 for triaxial creep tests; lower values are
many times undergoing consolidation as well as sec- reported for undrained conditions than for drained con-
ondary compression or creep, it is not surprising that ditions. For the development shown here, the stress
development of general models that can be readily im- intensity D is taken as the deviator stress (1 ⫺ 3). A
plemented in engineering practice is a challenging un- shear stress or stress level could also be used.
dertaking. The same data plotted in the form of Figs. 12.3,
rig
Nonetheless, some progress has been made in estab- 12.31, and 12.32 can be expressed by
lishing functional forms and relationships that can be
applied for simple analyses and comparisons, and one
of these is developed in this section. A complete re-
view and development of all recent theories and pro-
ln 冋 册 ˙
˙ (t,D0)
⫽ D (12.38)
py
et al. (1996).
in which ˙ (t,D0) is a fictitious value of strain rate at
General Stress–Strain–Time Function
D ⫽ 0, a function of time after start of creep, and
Strain Rate Relationships between axial strain rate is the slope of the linear part of the log strain rate
˙ and time t of the type shown in Figs. 12.4 and 12.5 versus stress plot. From Eqs. (12.37) and (12.39)
can be expressed by
ln
˙
˙ (t1,D)
⫽ ⫺m ln 冉冊 t
t1
(12.36) ln ˙ (t1,D) ⫺ m ln 冉冊t
t1
⫽ ln ˙ (t,D0) ⫹ D (12.40)
or For D ⫽ 0,
ln ˙ (t,D0) ⫽ ln ˙ (t1,D0) ⫺ m ln 冉冊 t
t1
(12.41)
vided the variation of strength with water content is
known. Since normal strength tests are considerably
simpler and less time consuming than creep tests, the
in which ˙ (t1,D0) is the value of strain rate obtained by uniqueness of the quantity Dmax can be useful because
projecting the straight-line portion of the relationship the results of a limited number of tests can be used to
between log strain rate and deviator stress at unit time predict behavior over a range of conditions. A further
to a value of D ⫽ 0. Designation of this value by A generalization of Eq. (12.43) then is
冉冊
and substitution of Eq. (12.41) into Eq. (12.39) gives m
t
冉冊
˙ ⫽ A exp(D) 1 (12.44)
t t
ln ˙ ⫽ ln A ⫹ D ⫺ m ln (12.42)
t1
l
where
ria
which may be written
D
冉冊 m ⫽ Dmax D⫽ (12.45)
D
t1 Dmax
˙ ⫽ Ae (12.43)
t
Strain A general relationship between strain and
ate
This simple three-parameter equation has been time is obtained by integration of Eq. (12.43). Two
found suitable for the description of the creep rate be- solutions are obtained, depending on the value of m.
havior of a wide variety of soils. The parameter A is If ⫽ 1 at t ⫽ t1 ⫽ 1, then
shown in Fig. 12.54. Since it reflects an order of mag-
nitude for the creep rate under a given set of condi- A
⫽ 1 ⫹ exp (D)(t1⫺m ⫺ 1) when m ⫽ 1
dM
tions, it is in a sense a soil property. A minimum of
two creep tests are needed to establish the values of A,
1⫺m
(12.46)
, and m for a soil. If identical specimens are tested
using different creep stress intensities, a plot of log and
strain rate versus log time yields the value of m, and
a plot of log strain rate versus stress for different values ⫽ 1 ⫹ A exp (D)ln t when m ⫽ 1 (12.47)
of time can be used to find and A from the slope
hte
and the intercept at unit time, respectively. Creep curve shapes corresponding to these relation-
The parameter has units of reciprocal stress. If ships are shown in Fig. 12.55. These curves encompass
stress is expressed as the ratio of creep stress to the variety of shapes shown in Fig. 12.2. A similar
strength at the beginning of creep, D/Dmax, then the equation to Eq. (12.46) was developed by Mesri et al.
dimensionless quantity Dmax should be used. For a (1981) from Eq. (12.43). The initial time-independent
rig
given soil and test type, values of Dmax do not vary strain was neglected, and the resulting equation is
greatly for different water contents, as the change in
with water content is compensated by a change in
Dmax. Thus the strain rate versus time behavior for any
stress at any water content can be predicted from the
⫽
1⫺m
At1
exp(D)
t
t1 冉冊 1⫺m
(12.48)
py
results of creep tests at any other water content, pro- It may be seen in Fig. 12.56 that this equation de-
scribes the uniaxial creep behavior of three clays very
well. Data for both drained and undrained creep are
shown.
Co
l
ria
ate
Figure 12.55 Creep curve shapes predicted by the general stress–strain–time function of
Eqs. (12.46) and (12.47).
D
⫽
D
D0 D0
dM
⫽ 1 ⫺ s log
t
t0
(t ⬎ t0) (12.49) t0 ⫽
h0
˙
(12.52)
where s is the slope of the stress relaxation curve, and where h0 is the strain rate to give a delay time of t0 ⫽
the zero subscript refers to conditions at the start of 1 min before stresses begin to relax. The data presented
stress relaxation. Also by Lacerda and Houston (1973) indicate that the values
of # and h0 increase with increasing plasticity of the
hte
# soil.
s⫽ (12.50)
D
Constitutive Models
where Different rheological models have been proposed for
the mathematical description of the stress–strain–time
rig
l
ria
ate
dM
hte
rig
py
Co
Figure 12.56 Correspondence between creep strain predicted by Eq. (12.48) and measured
values. Diagrams are from Mesri et al. (1981), which were based on analyses by Semple
(1973).
l
ria
Figure 12.57 Influence of prior strain rate on stress relaxa- Loss of strength during creep is particularly important
tion. in soft clays deformed under undrained conditions and
heavily overconsolidated clays in drained shear. Both
of these conditions are pertinent to certain types of
engineering problems: the former in connection with
ate
stability of soft clays immediately after construction,
and the latter in connection with problems of long-term
stability.
The loss of strength as a result of creep may be
explained in terms of the following principles of be-
havior:
dM
1. If a significant portion of the strength of a soil is
due to cementation, and creep deformations
cause failure of cemented bonds, then strength
will be lost.
2. In the absence of chemical or mineralogical
hte
mation.
considered. None appears to exist that has the gener- 4. Water content changes cause strength changes.
ality and simplicity of the three-parameter creep Eqs.
(12.43), (12.46), and (12.47). These processes are illustrated by the stress paths
Both plasticity and creep are controlled by the mo- and effective stress envelope shown schematically in
py
tension of the Cam-clay model to take into account Goldstein and Ter-Stepanian (1957), and Vialov and
time-dependent volumetric and deviatoric deforma- Skibitsky (1957). This may be explained as follows.
tions (Kavazanjian and Mitchell, 1980; Borja and Ka- Shear deformations cause dilation and the development
vazanjian, 1985; Kaliakin and Dafalias, 1990; Borja, of negative pore pressures, which do not develop uni-
1992; Al-Shamrani and Sture, 1998; Hashiguchi and formly throughout the sample but concentrate along
Okayasu, 2000) and on an elasto-viscoplastic equation planes where the greatest shearing stresses and strains
developed using flow surface theory (Sekiguchi, 1977, develop. With time during sustained loading, water mi-
1984; Matsui and Abe, 1985, 1986, 1988; Matsui et grates into zones of high negative pore pressures lead-
al., 1989; Yin and Graham, 1999) and overstress theory ing to softening and strength decrease relative to the
(Adachi and Oka, 1982; Katona, 1984: Kutter and strength in ‘‘normal’’ undrained strength tests. This
Sathialingham, 1992; Rocchi et al., 2003). leads to shear band formation.
l
starting from these conditions leads to strength G, tests in the laboratory may allow establishment of a
ria
which is less than the original value at B. It is evident stress level versus time-to-failure relationship.
also that if the negative pore water pressure is large For soils subject to failure during undrained creep,
enough, and the sustained load is applied long enough, the time to failure is usually a negative exponential
then point E could reach the failure envelope. This function of the stress, for stresses greater than some
appears to have been the conditions that developed in limiting value below which no failure develops even
ate
several cuts in heavily overconsolidated brown London after very long times.8 The relationship between devi-
clay, which failed some 40 to 70 years after excava- ator stress, normalized to the pretest major principal
tions were made (Skempton, 1977). effective stress, and time to failure for Haney clay is
shown in Fig. 12.60. These and similar data define cer-
Time to Failure dM
The time to failure of soils susceptible to strength loss
under sustained stresses depends on the rates at which 8
This critical stress below which creep rupture does not occur has
pore pressures develop and at which water can migrate been termed the upper yield, the lower yield being the stress below
into or out of the critical shear zone. These rates are, which deformations are elastic (Murayama and Shibata 1958, 1964).
hte
rig
py
Co
Figure 12.59 Stress paths for normal undrained shear and drained creep of heavily over-
consolidated clay.
l
tionship is unique, as may be seen in Fig. 12.62,
ria
which shows that
C
tƒ ⫽ (12.53)
˙ min
ate
Figure 12.60 Time to rupture as a function of creep stress Values of the constant C accurate to about 0.2
for Haney clay (Campanella and Vaid, 1972). log cycles are given in Table 12.2.
3. The strain at failure is a constant independent of
stress level, as shown in Fig. 12.63. The failure
dM
hte
rig
py
Co
Figure 12.61 Creep rate behavior of K0-consolidated, undisturbed Haney clay under axi-
symmetric loading (Campanella and Vaid, 1972).
C ⫽ (1 ⫺ m)ƒ (12.57)
l
Predictions of the time to failure under a given stress
ria
may be made in the following way. Strain at failure
can be determined by either a creep rupture test or by
a normal shear or compression test. If a normal
strength test is used, then the rate of strain must be
ate
slow enough to allow pore pressure equalization or
drainage, depending on the conditions of interest, and
the stress history and stress system should simulate
those in the field. Parameter m can be established from
Figure 12.62 Relationship between time to failure and min-
a creep test, and then C can be computed from Eq.
imum creep rate (from Campanella and Vaid, 1974). Repro-
dM
duced with permission from the National Research Council (12.57). Values of A and are established from creep
of Canada. tests at two stress intensities. Then, for t1 ⫽ 1,
C ⫽ ˙ mintƒ ⫽ A exp(D)t1⫺m
ƒ (12.58)
strain is taken as the strain corresponding to the and corresponding values of D and tƒ can be calculated
minimum strain rate. For the case of undrained using Eq. (12.58) rewritten as
hte
rectly from the fact that the strain at the point of min- ing conditions. For example, Sekiguchi (1977) de-
imum strain rate is a constant independent of stress or veloped a viscoplastic model that gives excellent
strain rate. The general stress–strain rate–time function representations of strain rate effects on undrained
[see Eq. (12.43)] describes the strain rate–time behav- stress–strain behavior, stress relaxation, and creep rup-
ior until ˙ min is reached. For t1 ⫽ 1 and ⫽ 0 at t ⫽
py
Creep
Rate C ⫽ (˙min tƒ)
Test Parameter, (0.2 log
Soil Typea m cycles)
Undisturbed Haney ICU 0.7 1.2
clay, N.C.b
Undisturbed Haney ACU 0.4 0.2
clay, N.C.b
l
Undisturbed Haney ACU-PS 0.5 0.3
ria
clay, N.C.b
Undisturbed Seattle ICU 0.5 0.6
clay, O.C.c
Undisturbed U 0.8 1.6
Tonegawa loamc
ate
Undisturbed ICU 0.75 2.8
Redwood City
clay, N.C.c
Undisturbed Bangkok ICU 0.70 1.4
mudc
Undisturbed Osaka 1.0 0.07
dM
clayc
a
ICU, isotropic consolidated, undrained triaxial; ACU, K0 consoli-
dated, undrained triaxial; ACU-PS, K0 consolidated, plane strain; and
U, compression test.
b
Data from Campanella and Vaid (1974).
c
Data from Singh and Mitchell (1969).
hte
rig
py
Co
Figure 12.63 Axial strain at minimum strain rate as a function of creep stress for undis-
turbed Haney clay (from Campanella and Vaid, 1974). Reproduced with permission from
the National Research Council of Canada.
l
ACU-PSc 0.5 0.5 0.25 0.30 data in the figure have been supplemented by values
ria
a
Isotropic consolidated, undrained triaxial. of
G/G for several sands compiled by Jamiolkowski
b
Anisotropic, consolidated, undrained triaxial. (1996). Mesri et al. (1990) report that NG for sands
c
Anisotropic consolidated, undrained, plane strain. varies between 0.01 and 0.03 and increases as the soil
Data from Campanella and Vaid (1974). becomes finer. Jamiolkowski and Manassero (1995)
give values of 0.01 to 0.03 for silica sands, 0.039 for
ate
sand with 50 percent mica, and 0.05 to 0.12 for car-
bonate sand. Experimental results show that the rate of
shear, swelling, chemical and biological changes, and increase in stiffness with time for very loose carbonate
the like. Until recently it has not been appreciated that sand increases as the stress level increases (Howie et
cohesionless soils exhibit this behavior as well. Much al., 2002). Isotropic stress state resulted in a slower
recent field evidence of the changing properties of
dM
granular soils over time is now available and these data
rate of increase in stiffness.
There is only limited field data that shows evidence
suggest that recently disturbed or deposited granular of aging effects on stiffness. Troncoso and Garces
soils gain stiffness and strength over time at constant (2000) measured shear wave velocities using downhole
effective stress—a phenomenon called aging. The ev-
idence includes the time-dependent increase in stiff-
ness and strength of densified sands as measured by ΔG : Modulus Increase in Every 10-fold Time Increase
0.30
hte
cone penetration resistance (Mitchell and Solymer, G1000 : Modulus at 1,000 min
1984; Thomann and Hryciw, 1992; Ng et al., 1998)
ΔG/G1000 = 0.03PI 0.5
and the setup of displacement piles in granular mate-
rials (Astedt et al., 1992; York et al., 1994; Chow et 0.25
al., 1998; Jardine and Standing, 1999; Axelsson,
Modulus Increase Ratio ΔG/G1000
Sand
Jamiolkowski (1996)
confining stress, and this is considered to be the con- a = Ticino Sand (Silica)
sequence of aging. This behavior can be quantified by e b = Hokksund Sand (Silica)
0.05 c = Messina Sand and Gravel (Silica)
a coefficient of shear modulus increase with time using d d = Glauconite Sand
the following formula (Anderson and Stokoe, 1978): c (Quartz/Glauconite)
e = Quiou Sand (Carbonate)
a, b f = Kenya Sand (Carbonate)
IG ⫽
G/log(t2 /t1) (12.60) 0.00
0 20 40 60 80 100
Plasticity Index PI
NG ⫽ IG /G1000 (12.61)
Figure 12.64 Modulus increase ratio for clays (from Ko-
where IG is the coefficient of shear modulus increase kusho, 1987), supplemented by the data for sands (from
with time, t1 is a reference time after primary consol- Jamiolkowski, 1996).
wave propagation tests in low-plasticity silts with fines and seepage blanket. Due to large depths of the loose
contents from 50 to 99 percent at four abandoned tail- sand deposit requiring densification, a two-stage den-
ing dams in Chile. The shear modulus normalized by sification program was performed. The upper 25 m of
the vertical effective stress is plotted against the age sand (and a 5- to 10-m-thick sand pad placed by hy-
of the deposit in Fig. 12.65. The age of the deposits is draulic filling of the river) was densified using vibro-
expressed as the time since deposition. Although the compaction. Deposits between depths of 25 to 40 m
soil properties vary to some degree at the four sites,9 were densified by deep blasting.
very significant increase in stiffness at small strains can During the blasting operations, it was observed that
be observed after 10 to 40 years of aging. The degree the sand exhibited both sensitivity—that is, strength
to which secondary compression could have contrib- loss on disturbance—and aging effects. A typical ex-
uted to this increase is not known. ample of the initial decrease in penetration resistance
l
after blasting densification and subsequent increase
ria
Time-Dependent Behavior after Ground with time is shown in Fig. 12.66. Initially after im-
Improvement provement, there was in some cases a decrease in pen-
Stiffness and strength of sand increase with time after etration resistance, despite the fact that surface
disturbance and densification by mechanical processes settlements ranging from 0.3 to 1.1 m were measured.
With time (measured up to 124 days after improve-
ate
such as blasting and vibrocompaction. Up to 50 per-
cent or more increase in strength has been observed ment), however, the cone penetration resistance was
over 6 months (Mitchell and Solymer, 1984; Thomann
and Hryciw, 1992; Charlie et al., 1992; Ng et al., 1998;
Ashford, et al., 2004) as measured by cone penetration
testing. dM
The Jebba Dam project on the Niger River, Nigeria,
was an early well-documented field case where aging
effects in sands were both significant and widespread
(Mitchell and Solymer, 1984). The project involved the
treatment of foundation soils beneath a 42-m-high dam
hte
rig
py
Co
9
The four sites identified by Troncoso and Garces (2000) are called
Barahona, Cauquenes, La Cocinera, and Veta del Agua and the aging
times between abandonment and testing were 28, 19, 5, and 2 years,
respectively. The tailing deposits at Barahona had a liquid limit of
41 percent and a plastic limit of 14 percent, whereas those at the Figure 12.66 Effect of time on the cone penetration resis-
other three sites had liquid limits of 23 to 29 percent and plastic tance of sand following blast densification at the Jebba Dam
limits of 2 to 6 percent. site.
l
variability in the magnitude of aging effects throughout
ria
the site. Because of the greater density increase caused
by vibrocompaction than by blast densification, no in-
itial decrease in the penetration resistance was ob-
served at the end of the compaction process.
Charlie et al. (1992) found a greater rate of aging
ate
after densification by blasting for sands in hotter cli-
mates than in cooler climates and suggested a corre-
lation between the rate of aging and mean annual air Figure 12.68 Effect of time on the cone penetration resis-
temperature for available field data as shown in Fig. tance of hydraulic fill sand after placement at the Jebba Dam
12.69. In the figure, the increase in the CPT tip resis-
dM site.
tance (qc) with time is expressed by the following
equation:
qc (N weeks)
⫽ 1 ⫹ K log N (12.62)
qc (1 week) 1.0 Mitchell and
Solymer (1984)
Empirical Constant K
of sands, similar to diagenesis in locked sands and with time (Axelsson, 2000). Evidence suggests that
young rocks (Dusseault and Morgenstern, 1979; Bar- piles in silts and find sands set up more than those in
ton, 1993) in which grain overgrowth has been ob- coarse sands and gravels (York et al., 1994). Both
served. However, others have questioned whether driven and jacked piles exhibit setup, whereas bored
significant chemical reactions can occur over the short piles do not. Hence, the stress–strain state achieved
time of observations. In addition, there is some evi- during the construction processes of pile driving have
dence of aging in dry sands wherein chemical proc- an influence on this time-dependent behavior and var-
esses would be anticipated to be very slow. ious mechanisms have been suggested to explain this
(Astedt et al., 1992; Chow et al., 1998; Bowman,
Setup of Displacement Piles 2002). Unfortunately, at present, there is no conclusive
evidence to confirm any of the proposed hypotheses.
l
Much field data indicates that the load-carrying capac- Despite the many field examples and laboratory
ria
ity of a pile driven into sand may increase dramatically studies on aging effects, there is still uncertainty about
over several months, long after pore pressures have the mechanism(s) responsible for the phenomenon.
dissipated (e.g., Chow et al., 1998; Jardine and Stand- Understanding the mechanism(s) that cause aging is of
ing, 1999). The amount of increase is highly variable, direct practical importance in the design and evaluation
ranging from 20 to 170 percent per log cycle of time of ground improvement, driven pile capacity, and sta-
ate
as shown in Fig. 12.70 (Chow et al., 1998; Bowman, bility problems where strength and deformation prop-
2002). Most of the increase in capacity occurs along erties and their potential changes with time are
the shaft of the pile as the radial stress at rest increases important. Mechanical, chemical, and biological fac-
tors have been hypothesized for the cause of aging.
dM Biological processes have so far been little studied;
however, mechanical and chemical phenomena have
been investigated in more detail, and some current un-
derstanding is summarized below.
l
disappear quickly when the soil is subjected to new
ria
stress state. 0.05 Leighton Buzzard
Uniform Silica Sand
When a constant shear stress is applied to loose
sand, large creep accompanied by volumetric contrac- 0.00
tion is observed (Bopp and Lade, 1997). Higher con- 10 1 10 2 10 3 10 4 10 5
tact forces due to loose assemblies contribute to
ate
0.05 Montpellier Natural Sand
increased particle crushing, contributing to contraction
behavior. Hence, decrease in volume by soil crushing 0.00
(1998). This implies that sands at a high relative den- Time (s)
sity will set up more as more interlock between par-
Figure 12.71 Dilative creep observed in triaxial creep tests
ticles may occur (Bowman, 2002). The laboratory
of dense fine sand (by Bowman and Soga, 2003).
observation of initial contraction followed by dilation
conveniently explains the field data of dynamic com-
paction where the greater initial losses and eventual
rig
gains in stiffness and strength of sands are found close ing. Terzaghi originally referred to a ‘‘bond strength’’
to the point of application where larger shear stresses in connection with the presence of a quasi-
are applied to give dilation (Dowding and Hryciw, preconsolidation pressure in the field (Schmertmann,
1986; Thomann and Hryciw, 1992; Charlie et al., 1991). Generally, this mechanism has been thought of
as type of cementation, which would increase the co-
py
1992).
Increased strength and stiffness due to mechanical hesion of a soil without affecting its friction angle.
aging occurs predominantly in the direction of previ- Denisov and Reltov (1961) showed that quartz sand
ously applied stress during creep (Howie et al., 2002). grains adhered to a glass plate over time. They placed
No increase was observed when the sand was loaded individual sand grains on a vibrating quartz or glass
Co
in a direction orthogonal to that of the applied shear plate and measured the force necessary to move the
stress during creep (Losert et al., 2000). grains as shown in Fig. 12.72. The dry grains were
allowed to sit on the plate for varying times and then
the plate was submerged, also for varying times, before
12.13 CHEMICAL PROCESSES OF AGING vibrating began. It was found that the force required
to move the sand grains continued to increase up to
Chemical processes are a possible cause of aging. His- about 15 days of immersion in water. The cementating
torically, the most widespread theory used to explain agent was thought to be silica-acid gel, which has an
aging effects in sand has involved interparticle bond- amorphous structure and would form a precipitate at
l
different times up to 2 years. Strength and stiffness
ria
increases were observed in all cases, and a typical plot
䉭 䊊 of load–displacement curves at various times is shown
f
––– 3.0 in Fig. 12.73. The effects of aging were greater for the
f0 (1) Without Soaking
(2) 42-hour Soaking submerged sand than for the dry specimens. Scanning
electron micrographs of the aged specimens in distilled
ate
(3) 6-day Soaking 䊊
(4) 14-day Soaking
䉭
water and seawater showed precipitates on and in be-
2.0
䉭 4 䉭 tween sand grains. For the river sand in distilled water,
䊊 &
3
䊊
the precipitates were composed of calcium (the soluble
fraction of the sand) and possibly silica. For the river
1.0 & sand in seawater, the precipitates were composed of
2
&
dM 1
& sodium chloride.
However, there are several reported cases in which
cementation was an unlikely mechanism of aging, at
0 least in the short term. For example, dry granular soils
10 min 2h 20h
Time can show an increase in stiffness and strength with
time (Human, 1992; Joshi et al., 1995; Losert et al.,
Figure 12.72 Results of vibrating plate experiment from 2000). Cementation in dry sand is unlikely, as moisture
hte
Figure 12.73 Effect of aging on the penetration resistance of River sand (from Joshi et al.,
1995).
l
is approximately 3 percent. Mesri et al. (1990) argue in the field there will be impurities, biological activity,
ria
that this large strain would destroy any cementation, and heterogeneity of void ratio and fabric. Further-
and therefore another less brittle mechanism must be more, the introduction of air and other gases during
responsible for the increase in stiffness. ground improvement may have consequences that have
In summary, experimental evidence indicates that so far not been fully evaluated. Arching associated
mechanical aging behavior is enhanced by shear stress with dissipation of blast gases and the redistribution of
ate
dM
hte
rig
py
Co
Figure 12.74 Effect of aging on stress–strain relationship of Ham River sand (from Dara-
mola, 1980).
stresses through the soil skeleton may also play a role analyses and predictions can be made for large and
(Baxter and Mitchell, 2004). The boundary conditions complex geotechnical structures.
associated with penetration testing in rigid-wall cylin-
ders in the laboratory may prevent detection of time-
dependent increases in penetration resistance that are QUESTIONS AND PROBLEMS
measured under the free-field conditions in the field.
1. Find an article about a problem, project, or issue
that involves some aspect of the long-term behavior
of a soil as an important component. The article
may be from a technical journal or magazine or
12.14 CONCLUDING COMMENTS elsewhere. The only requirement is that it involves
l
With exception of settlement rate predictions, most soil consideration of time-dependent ground behavior in
ria
mechanics analyses used in geotechnical engineering some way.
assume limit equilibrium and are based on the as- a. Prepare a one-page informative abstract of the
sumption of time-independent properties and defor- article.
mations. In reality, time-dependent deformations and b. Summarize the important geotechnical issues in
ate
stress changes that result from the time-dependent or the article and write down what you believe you
viscous rearrangement of the soil structure may be re- would need to know to understand them well
sponsible for a significant part of the total ground re- enough to solve the problem, resolve the issue,
sponse. advise a client, and so forth. Do not exceed two
Rate process theory has proven a particularly fruitful pages.
approach for the study of time-dependent phenomena
dM c. Identify topics, figures, equations, and other ma-
in soils at consistencies of most interest in engineering terial in Chapter 12, if any, that might be useful
problems, that is, at water contents from about the in addressing the problems.
plastic limit to the liquid limit. From an analysis of the
influences of stress and temperature on deformation 2. The figure below shows relationships between (1)
number of interparticle bonds and effective consol-
rates and other evidence, it has been possible to deduce
idation pressure and (2) compressive strength and
that interparticle contacts are essentially solid and that
number of interparticle bonds for three soils as de-
clay strength derives from interatomic bonding in these
hte
70
600
60
Sand, Illite, Bay Mud
400 50
l
ria
40
200
30
ate
20 40 80 100 200 400
0 Compressive Strength (kPa)
3. Equation (12.43) is a simple three-parameter equa- intensities are shown below, as is the variation of
tion for strain rate during constant stress creep of compressive strength with water content. A tem-
soils.
dM porary excavation is planned that will create a slope
with an average factor of safety of 1.5. The average
a. Show the meaning of , D, and m on a clearly water content of the clay in the vicinity of the cut
labeled sketch. is 50 percent. The excavation is planned to remain
b. Modify Eq. (12.43) and indicate the information open for a period of 4 months. Prepare a plot of
needed to permit prediction of creep rates for a strain rate versus time for an element of clay and
given soil at any value of water content and assess the probability of a creep rupture failure oc-
stress intensity from a knowledge of creep rates curring during this period.
hte
4. The results of triaxial compression creep tests on taken as the time corresponding to minimum
samples of overconsolidated Bay mud at three stress strain rate, ˙ min, prior to acceleration of defor-
mation and failure, and tests have shown that
py
XRupture
˙ mintƒ ⫽ constant
50
If a test embankment designed at a factor of
Water Content = 60% D = 100 kPa
40 safety of 1.05 based on shear strength deter-
Axial Strain (%)
Co
6. Would you expect that creep and stress relaxation tration tests) can increase significantly over time pe-
will be significant contributors to the stress– riods as short as weeks or months following depo-
deformation and long-term strength of soils on the sition and/or densification. Outline a test program
Moon? Why? that might be done to test the validity of one of
these causes.
7. List possible causes of sand aging wherein the stiff-
ness and strength (usually as determined by pene-
l
ria
ate
dM
hte
rig
py
Co
l
ria
ate
a area As specific surface area per unit weight of
a coefficient for harmonics solids
a cross-sectional area of a tube Å Angstrom unit ⫽ 1 ⫻ 10⫺10 m
a crystallographic axis direction or distance b coefficient of harmonics
a effective cluster contact area b crystallographic axis direction or distance
a volumetric air content
dM b intermediate stress parameter
a thermal diffusivity B parameter in rate process equation ⫽
ac effective area of interparticle contact X(kT/h)
am coefficient of compressibility with respect B Bishop’s pore water pressure coefficient
to changes in water content Bq grain breakage parameter
at coefficient of compressibility with respect Br Hardin’s relative breakage parameter
to changes in ( ⫺ ua) c cohesion
av coefficient of compressibility in one di- c cohesion intercept in total stress
hte
523
l
Cn coordination number E potential energy
ria
CR compression ratio E Young’s modulus
CRR cyclic resistance ratio E voltage, electrical potential
Cs compressibility of a solid E50 secant modulus at 50 percent of peak
Cs shape coefficient strength
Cs swelling index
ate
Emax small strain Young’s modulus
Cu coefficient of uniformity Er rebound modulus
Cu compressibility of soil skeleton by pore ESP exchangeable sodium percentage
pressure change E( ) distribution function for interparticle con-
CW compressibility of water tact plane normals
C, Ce coefficient of secondary compression
dM ƒ force acting on a flow unit
d diameter ƒ frequency
d distance ƒi fraction of particles between two sizes
d10 sieve size that 10% of the particles by ƒn normal force
weight pass through ƒt tangential force
d60 sieve size that 60% of the particles by F force of electrostatic attraction
weight pass through
F formation factor
dx incremental horizontal displacement at
F free energy
hte
peak
F freezing index
dy incremental vertical displacement at peak
F pressure-temperature parameter
D diameter of particle
F tensile strength
D dielectric constant, relative permittivity
D diffusion coefficient F, F0 Faraday constant ⫽ 96,500 coulombs
D deviator stress F partial molar free energy on adsorption
rig
D stress level ⫽ D/Dmax Fd free energy of the double layer per unit
D0 molecular diffusivity of water vapor in air area at a plate spacing of 2d
D0 self-diffusion coefficient
F free energy of activation
D50 sieve size that 50% of the particles by FE electrical force per unit length
FH hydraulic seepage force per unit length
py
l
i gradient Kso stress-optical material constant
ria
i unit vector K wavelengths of monochromatic radiation
ic chemical gradient l length
ie electrical gradient l material thickness
ih hydraulic gradient l total number of pore classes
ate
it thermal gradient L latent heat of fusion
I electrical current L length
I intensity Lij coupling coefficient or conductivity coef-
I1, I2, I3 stress invariants ficient
IG coefficient of shear modulus increase with LI liquidity index
time dM LIeq equivalent liquidity index
IR dilatancy index LL liquid limit
Iv void index Ls latent heat of fusion of water
Jc chemical flow rate m slope of relationship between log creep
JD chemical flow rate strain rate and log time
Ji flux of constituent i m total mass per unit total volume
Ji value of property i in clay-water system m total number of pore classes
Js flow rate of salt relative to fixed soil layer mc mass of clay
hte
l
Ne number of load cycles qi concentration of solids
ria
NG normalized shear modulus increase with qt heat flow rate
time qvap vapor flux density
Ns moles of water per unit volume of sedi- qw water flow rate
ment Q electrical charge
ate
Nw moles of salt per unit volume of sediment Q quantity of heat
OCR overconsolidation ratio r pore radius
p constant that accounts for the interaction r radius
of pores of various sizes rk ratio of horizontal to vertical hydraulic
p hydrostatic pressure conductivities
p matrix or osmotic pressure
dM rp pore size
p pressure rp tube radius
p partial pressure of water vapor in pore R coefficient of roundness
space R electrical resistance
p vertical consolidation pressure R gas constant ⫽ 1.98726 cal/ K-mole
p mean effective pressure 8.31470 joules/ K-mole
po present overburden pressure 82.0597 cm3 atm/ K-mole
pa atmospheric pressure R long-range repulsion pressure
hte
l
tm time for adsorption of a monolayer Vw volume of water
ria
T intercluster tortuosity w water content
T shear force wL, wl liquid limit
T temperature wP, wp plastic limit
T time factor W water content
ate
T0 initial temperature W width
Tc intracluster tortuosity W fluid volume
Tc temperature at consolidation W water transport
TFP freezing temperature W weight
Ts surface temperature x distance from the clay surface
Ts temperature of shear for consolidated un-
dM X distance
drained direct shear tests X friction coefficient
TV time factor Xi driving force
u excess pore pressure y potential function ⫽ ve /kT
u ionic mobility z direction of gravity
u midplane potential function z distance from drainage surface
u pore water pressure z electrolyte
u pore water pressure in the interparticle z ionic valence
hte
l
solid fraction of a contact area ! real relative permittivity
ria
relative retardation !ⴖ polarization loss, imaginary relative per-
p particle eccentricity distance mittivity
dielectric constant, permittivity compression index
porosity correction coefficient for frost depth pre-
ate
strain diction equation
˙ strain rate damping ratio
0 permittivity of vacuum, 8.85 ⫻ decay constant
10⫺12 C2 /(Nm2) pore size distribution index
1 axial strain separation distance between successive
˙ a vertical strain rate in one dimensional
dM positions in a structure
consolidation wave length of X ray
ƒ strain at failure wave length of light
˙ min minimum strain rate cs critical state compression index
rd volumetric strain that would occur if chemical potential
drainage were permitted coefficient of friction
s deviator strain dipole moment
˙ s deviator strain rate fusion parameter
hte
crit
friction angle at critical state liquid-to-solid contact angle
e, e Hvorslev friction parameter orientation angle
ƒ friction angle corrected for the work of volumetric water content
dilation m volumetric water content at full saturation
Co
l
1 major principal effective stress shear strength
ria
1c major principal stress during consolida- shear stress
tion surface tension
1ƒ major principal stress at failure swelling pressure or matric suction
1ƒƒ major principal effective stress at failure undrained shear strength
2 intermediate principal effective stress a apparent tortuosity factor
ate
3 minor principal total stress c applied shear stress
3 minor principal effective stress c contaminant film strength
3c minor principal stress during consolida- cy undrained cyclic shear stress
tion d drained shear strength
3ƒƒ minor principal effective stress at failure
dM ƒƒ shear stress at failure on failure plane
a axial effective stress i shear strength
ac axial consolidation stress i shear strength of contact
as interfacial tension between air and solid m shear strength of solid material in yielded
aw interfacial tension between air and water zone
c crushing strength of particles peak applied shear stress at peak
c tensile strength of cement initial static shear stress
e electrical conductivity ' mass flow factor
hte