Research and design of test apparatus for laboratory measurement of lateral and vertical swell pressures

Material Information

Research and design of test apparatus for laboratory measurement of lateral and vertical swell pressures
Colby, Craig Allan
Publication Date:
Physical Description:
xix, 396 leaves : illustrations ; 29 cm


Subjects / Keywords:
Swelling soils -- Testing ( lcsh )
Swelling soils -- Testing ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references ([278]-288).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Department of Civil Engineering.
Statement of Responsibility:
by Craig Allan Colby.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
22927309 ( OCLC )
LD1190.E53 1990m .C64 ( lcc )

Full Text
B.S. Civil Engineering, University of Colorado at Boulder, 1980
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Craig Allan Colby
Department of Civil Engineering
\ >

This thesis for the Master of Science degree by
Craig Allan Colby
has been approved for the
Department of
Civil Engineering

Colby, Craig Allan (M.S., Civil Engineering)
Research and Design of Test Apparatus for Laboratory
Measurement of Lateral and Vertical Swell Pressures
Thesis directed by Professor Nien-Yin Chang
Although much has been learned about the behavior of expansive
soils in the last 30 years, extensive damage to many structures
founded on or in these soils is still very common, and typically
requires costly mitigation and repair.
To date, expansive soil research and testing have focussed on
vertical swelling potential and vertical swelling pressure. As
a result, the effects of lateral swelling are not well known and
the consideration of these effects has not become part of routine
geotechnical practice. A primary reason for this is the common
assumption that swelling is isotropic. Studies which have been
performed to date have not conclusively proven this. In
addition, most problems in the field may appear to be an
exclusive result of vertical swelling. Further, no standard
equipment exists for actually measuring lateral swelling
Research to date indicates that lateral swell pressures may not
be the same as vertical pressures and are typically quite
substantial. This could be of significant importance in the
design of drilled piers, basement walls, or buried structures in
expansive clays. At present, it is unclear how the effects of

lateral swelling would be incorporated into geotechnical design.
This research was conducted for the purpose of designing a new
laboratory test apparatus capable of simultaneously measuring
lateral and vertical swelling behavior. A design is presented
that will allow testing of remolded and undisturbed specimens
under a wide range of controlled stress or strain conditions.
The design is based on information obtained from extensive review
of the literature and upon results of laboratory testing
performed using a standard one-dimensional Bishop-type
consolidometer, an Anteus-type back pressure consolidometer with
and without back pressure, and a standard triaxial cell fitted
with a 0.5 inch thick aluminum cell chamber. Considerable
insight to the difficulties in conducting accurate swelling tests
was obtained from the laboratory testing program and is discussed
in the accompanying report. The new equipment design attempts
to incorporate the advantages of the various types of standard
equipment and overcome many of the difficulties of swell testing.
The form and content of this abstract are approved. I recommend
its publication.
Faculty meitf^er in charge of ^thesis

Ten years ago I would have never dreamed that I would, or
even could obtain a master's degree in Civil Engineering.
However, because of the undying devotion of my family and the
steadfast support of my close friends and unending guidance of
my teachers and colleagues, achievement of this important goal
has become a reality. Although words will never repay the debt
which I owe so many, I would at least like to take this
opportunity to express my sincerest thanks and appreciation.
To Dr. N.Y. Chang who supervised this work, I express my
deepest appreciation for his constant belief in me, his
encouragement and enthusiasm. His patience was also greatly
appreciated. I would also like to thank Dr. Fu-Hua Chen and Dr.
Gary Brierly for sharing their wisdom and experience in the
classroom and for reviewing this research.
Special thanks to Deb Vizzard for typing this manuscript,
Cynthia Herleth and Rudy Surya for assisting with drafting. Many
thanks to A.G. Wassenaar Incorporated, and Geocon Incorporated,
for their support, and the use of their facilities, equipment,
and resources.
Most importantly, I would like to express my deepest thanks
and love to my entire family. Their love, constant

encouragement, and unending patience have been paramount to my
success and will never be forgotten.
Finally, words cannot begin to express my gratitude to my
wife Joan and our son Kyle for their friendship, constant
support, and eternal love as they have made this endeavor
possible and mean more than they will every know. I only hope
that someday I can do the same for them. I love you both.

1. INTRODUCTION ........................................1
1.1 Problem ...........................................1
1.2 Purpose ........................................4
1.3 Scope of Study ....................................5
2.1 General ...........................................7
2.1.1 Type of Clay Mineral.......................8
2.1.2 Initial Dry Density .........................16
2.1.3 Initial Moisture Content .................... 20
2.1.4 Surcharge and/or Confining Pressure ... 22
2.1.5 Soil Structure and Fabric ...................22
2.1.6 Initial Stress and Boundary Conditions
in Soil ....................................25
2.1.7 Availability of Free Water or Vapor to
the Soil ...................................25
2.1.8 Type and Concentration of Ions with the
Layers of Adsorbed and Absorbed Water . 26
2.1.9 Thickness of Soil Stratum ...................28
2.1.10 Curing Time .................................28
2.1.11 Time Allowed for Swelling to Occur .... 29
2.1.12 Prevailing Climatic Conditions ......... 29
2.1.13 Number of Cycles of Wetting and Drying . . 30
2.1.14 Temperature ................................. 30
2.2 Summary ..........................................32

MEASUREMENT .......................................34
3.1 General .........................................34
3.2 Free-Swell Reload Method .................... 34
3.3 Variable Surcharge-Swell Method ................ 37
3.4 Constant Volume Method ......................... 39
3.5 Comparison of Methodologies .................... 41
4.1 General .........................................44
4.2 General Approaches and Results ................. 44
4.3 Summary .........................................65
5. TESTING PROGRAM .....................................72
5.1 General .........................................72
5.2 Soil Properties .................................72
5.3 Sample Preparation ............................. 76
5.3.1 General ...................................76
5.3.2 Standard Oedometer Cell ...................80
5.3.3 Backpressure Consolidometer ............... 86
5.3.4 Triaxial Cell .............................86
5.4 Equipment Calibration .......................... 96
5.4.1 General ...................................96
5.4.2 Standard Oedometer Cell ...................96
5.4.3 Backpressure Consolidometer ............... 97
5.4.4 Triaxial Cell ..............................100

5.5 Test Procedures ..................................102
5.5.1 Standard Oedometer Cell ..............102
5.5.2 Back Pressure Consolidometer ................ 124
5.5.3 Triaxial Cell ................................149
6. TEST RESULTS AND DISCUSSION..............................183
6.1 General ..........................................183
6.2 Standard Oedometer Cell...........................184
6.2.1 Test No. 1 184
6.2.2 Test No. 1A .................................189
6.3 Backpressure Consolidometer ..................... 194
6.3.1 Test No. 2...................................194
6.3.2 Test No. 3...................................198
6.3.3 Test No. 3A ..................................202
6.4 Triaxial Cell ....................................205
6.4.1 Test No. 4................................... 205
6.5 Summary of Test Results .................215
6.6 Lessons Learned ..................................224
6.7 Primary Advantages and Disadvantages of Each
Test Apparatus..................................232
7.1 General ..........................................240
7.2 Design Criteria ..................................240
7.3 Description of Proposed Test Apparatus .... 244
7.4 Equipment Assembly .............................. 264

RESEARCH ..............................................272
8.1 Summary ..........................................272
8.2 Conclusions ......................................274
8.3 Recommendations for Future Research ............. 276
BIBLIOGRAPHY ............................................ 279
A Summary of Test Data From Standard
One-Dimensional Oedometer Tests ................ 289
B Summary of Test Data From Back Pressure
Consolidometer Tests ........................... 295
C Summary of Test Data from Triaxial Test . . 308
D Suggested Detailed Test Procedure for
Standard One-Dimensional Constant Volume
Swell Test in Bishop-Type Consolidometer . . 317
E Suggested Detailed Test Procedure for
One-Dimensional Swell Test in Anteus-Type
Back Pressure Consolidometer ................... 331
F Suggested Detailed Test Procedure for
Three-Dimensional Constant Volume Swell
Test in Triaxial Cell............................357

1 Lateral Swell Pressure Ring Tests Using Variable
Surcharge Swell Method ............................ 53
2 Comparison of Lateral and Vertical Swell Pressures
Measured By Previous Researchers .................... 66
3 Comparison of Potential Lateral Swell Pressure
With Typical At-Rest Lateral Pressures .............. 69
4 Results of Index Property Tests For Soil Used
In Laboratory Swell Tests ........................... 77
5 Equipment Testing Schedule .......................... 183
6 Summary of Laboratory Swell Test Results ............ 217

1 Three Dimensional View, Schematic Representation
and Photograph of the Basic Tetrahedral and
Octahedral Molecular Units Which Make Up All
Clay Minerals..........................................10
2 Typical Schematic Representation of Kaolinite ... 14
3 Typical Schematic Representation of Illite ......... 15
4 Typical Schematic Representation of Montmorillonite 17
5 Relationship Between Initial Dry Density and Volume
Change for Samples with Constant Initial Moisture
6 Relationship Between Initial Dry Density and
Swelling Pressure for Samples with Constant
Initial Moisture Content .............................. 19
7 Relationship Between Initial Moisture Content and
Volume Change for Samples with Constant Initial
Dry Density........................................21
8 Relationship Between Surcharge Pressure and Volume
Change for Samples with Constant Initial Dry
Density and Moisture Content ......................... 23
9 Relationship Between Number of Cycles of Wetting
and Drying and Volume Change.......................31
10 Effects of Moisture Content Increase on Volume ... 33
Change and Swell Pressure
11 Swell-Reload Method to Determine Swell Pressure . . 36
12 Variable Surcharge-Swell Method to Determine
Swell Pressure ........................................38
13 Constant Volume Method to Determine Swell Pressure . 40
14 Controlled Strain Method to Determine Swelling
15 Swelling for Anisotropic Conditions .................. 47

16 Comparison of Potential Lateral Swell Pressures
with Typical At-Rest Lateral Pressures ................ 70
17 Site Plan Location Where Soil Obtained for
Laboratory Testing .................................... 73
18 Simplified Subsurface Profile of Area Where Soil
Obtained for Laboratory Testing ....................... 75
19 Typical Gradation Curve for Soil Used in Laboratory
Swell Tests............................................78
20 Laboratory Moisture Density Curves for Soil
Used in Laboratory Swell Tests ........................ 79
21 Photograph of Disassembled Proctor Mold .............. 81
22 Photograph of Standard Oedometer Ring on Base of
Proctor Mold ..........................................82
23 Photograph of Assembled Proctor Mold ................. 83
24 Photograph of Standard Oedometer Ring at
Bottom of Assembled Proctor Mold .................. 84
25 Photograph of 10-Pound Proctor Hammer Used to
Compact Test Specimens ............................... 85
26 Photograph of a Soil Specimen after being Remolded
in the Standard Oedometer Ring ........................87
27 Photograph of Specimen in Teflon Lined Ring .... 88
28 Photograph of Partly Disassembled Brass Mold .... 89
29 Photograph of Fully Assembled Brass Mold ............ 90
30 Photograph of Brass Mold Side View ...............91
31 Photograph of Brass Mold Top View...................92
32 Photograph of Triaxial Specimen after being
Remolded in the Brass Mold ............................94
33 Photograph of Remolded Triaxial Specimen Cut in
Half ..................................................95
34 Axial Compression Versus Applied Axial Pressure
Calibration of Standard Oedometer Cell ............... 98

35 Axial Compression Versus Applied Axial Pressure
Calibration of Back Pressure Consolidometer .... 99
36 Volume Change Versus Applied Cell Pressure
Calibration of Triaxial Cell ..........................101
37 Axial Compression Versus Applied Axial Pressure
Calibration for Triaxial Cell..........................103
38 Bishop-Type Consolidometer Device with Standard
Fixed Ring Oedometer Cell..............................104
39 Bishop-Type Loading Frames (Consolidometers) With
and Without Fixed Ring Oedometer Cell In Place . 105
40 Disassembled Standard Fixed Ring Oedometer Cell . 107
41 Base of Standard Oedometer Cell........................108
42 Top Cap with Attached Porous Stone, Steel
Calibration Disk, Bottom Porous Stone and Stainless
Steel Confining Ring from Standard Oedometer Cell . 109
43 Locking Ring and Associated Thumbscrews from
Standard Oedometer Cell Top View ....................110
44 Locking Ring and Thumbscrews from Standard
Oedometer Cell-Side View ............................. Ill
45 Lucite Cell Wall Attached to Base of Standard
Oedometer Cell ........................................112
46 Bottom Porous Stone Placed on Base of Standard
Oedometer Cell ........................................113
47 Stainless Steel Confining Ring Placed on Bottom
Porous Stone in Standard Oedometer Cell .............. 114
48 Stainless Steel Confining Ring Secured to the
Base of the Standard Oedometer Cell Using the
Locking Ring and Thumbscrews 115
49 Standard Oedometer Cell with Steel Calibration Disk
in Place Prior to Assembly of Top Porous Stone and
50 Fully Assembled Standard Oedometer Cell Top View 117
51 Fully Assembled Standard Oedometer Cell Side View 118

52 Close Up of Oedometer Cell Platform in Bishop-Type
Loading Frame ....................................... 119
53 Close Up of Standard Oedometer Cell Placed on
Platform in Loading Frame and Ready for Testing . 120
54 Front View of Standard Oedometer Cell in
Bishop-Type Loading Frame Ready for Testing .... 121
55 Rear View of Bishop-Type Loading Frame Showing
Moment Arm and Weights on Hangar.....................123
56 Schematic of Back Pressure Consolidometer ............ 125
57 Typical Anteus-Type, One-Dimensional, Back
Pressure Consolidometer ............................. 126
58 Back Pressure Consolidometer with Pressure
Transducer and 5-Way Valve Attached ................. 127
59 Disassembled Back Pressure Consolidometer ........... 128
60 Top View of Back Pressure Consolidometer Cell Base . 129
61 Pressure Dome Portion of Back Pressure Consolidometer
with Movement Pin Detached ............................130
62 Bottom View of Pressure Dome with Rolling Diaphragm
In Place and Fully Retracted ..........................131
63 Side View of Pressure Dome with Top Cap Connected
to Plate on Bottom of Rolling Diaphragm .............. 132
64 Top View of Cylindrical Lucite Cell Wall From Back
Pressure Consolidometer .............................. 133
65 Side View of Pressure Dome with Lucite Cell Wall
Attached ..............................................134
66 Oblique View of Teflon-Lined Confining Ring with
Soil Specimen Used in Back Pressure Consolidometer . 135
67 Bottom Porous Stone in its Retainer Connected to
Base of Back Pressure Consolidometer Cell..............136
68 Teflon-Lined Confining Ring with Soil Specimen
Placed on Bottom Porous Stone on Base of Back
Pressure Consolidometer Cell ......................... 137

69 Lucite Cell Wall Placed Around Teflon-Lined
Confining Ring on Base of Back Pressure
Consolidometer Cell....................................138
70 Pressure Dome Attached to Base of Back
Pressure Consolidometer Cell ........................ 139
71 Fully Assembled Back Pressure Consolidometer with
Fluid Reservoir........................................140
72 Close Up View of Dial Gauge Attached to Top of
Pressure Dome on Back Pressure Consolidometer to
Measure Movement of Needle Pin ....................... 141
73 Back Pressure Consolidometer Connected to Pressure
Control Panel ........................................ 142
74 Back Pressure Consolidometer Connected to Pressure
Control Panel with 5-Way Valve and Pressure
Transducer Attached .................................. 143
75 Schematic of Triaxial Cell Testing System ............ 150
76 Disassembled Triaxial Cell ............................151
77 Bottom View of Aluminum Triaxial Cell Chamber . . 152
78 Top View of Triaxial Cell Base ........................153
79 Oblique View of Triaxial Cell Base.....................154
80 Side View of Triaxial Cell Base with Soil Specimen
and Surrounding Latex Membrane In Place After
81 Close Up View of Soil Specimen and Surrounding
Latex Membrane on Triaxial Cell Base ...........156
82 Aluminum Cell Chamber Placed on Triaxial Cell Base . 157
83 Aluminum Cell Chamber on Triaxial Cell Base with
Locking Ring In Place .................................158
84 Fully Assembled Triaxial Cell With Steel End
Plates Secured to Insure a Tight Seal Between
the Cell Chamber and Cell Base and Reduce Axial

85 Side View of Fully Assembled Triaxial Cell with Steel
End Plates Secured .....................................160
86 Instron Loading Frame Used with Triaxial Cell . . 161
87 Triaxial Cell Placed in Instron Loading Frame with
Dial Gauges attached and Ready for Testing to
Proceed ................................................162
88 Close Up of Triaxial Cell Set Up for Testing In
Instron Loading Frame ................................ 163
89 Close Up of Dial Gauges Placed Between Crosshead
of Load Frame and Top End Plate on Triaxial Cell . . 164
90 Control Panel on Right Side of Instron Load Frame . 165
91 Close Up of Gear Box Shown in Figure 90...........166
92 Close Up View of Calibration and Scale Controls
for the Chart Recorder on the Left Side of the
Instron Loading Frame ................................ 167
93 Gear Boxes which Control the Chart Recorder
94 Typical Pressure Control Panel ...................... 169
95 Rear of Pressure Control Panel .....................170
96 Air Compressor used to Supply a Relatively Constant
Flow of Pressurized Air to the Pressure Control
97 Digital Multimeter used to Convert Electrical
Signal from Pressure Transducer and Display
it in Millivolts....................................172
98 Locations Where Silicone Grease and Large O-Ring
are placed on Triaxial Cell Base ......................177
99 Standard Oedometer Cell Test Results - Test No. 1 . 185
100 Standard Oedometer Cell Test Results - Test No. 1A . 190
101 Back Pressure Consolidometer Test Results Test
No. 2

102 Back Pressure Consolidometer Test Results Test
No. 3..................................................199
103 Back Pressure Consolidometer Test Results Test
No. 3A ................................................203
104 Triaxial Cell Test Results Test No. 4...............206
105 Relationship Between Lateral and Axial Swell
Pressure Measured in Triaxial Cell ................... 216
106 Comparison of Dry Densities Used in Swell Tests 1
through 4 to Relationship of Dry Density and Swell
Pressure Determined by Chen on Similar Soil .... 219
107 Summary of Vertical Swell Pressure Development
with Time for All Tests................................221
108 Summary of Vertical Swell Pressure Development
with Time for All Tests After Adjustment for
Surcharge and Initiation of Swelling ................. 222
109 Sectional View of Proposed Swell Consolidometer . . 245
110 Exploded Sectional View of Proposed Cell Base,
Lucite Cell walls, Top Ring and Tension Bolts . . 246
111 Sectional and Top Views of Proposed Cell Base . . . 247
112 Sectional and Bottom Views of Proposed Top Ring . . 248
113 Sectional and Top Views of Proposed Top Ring .... 249
114 Sectional, Top, and Bottom Views of Proposed
Bottom Pedestal ...................................... 250
115 Sectional, Top, and Bottom Views of Proposed Ram
and Top Cap............................................251
116 Exploded Sectional View of Proposed Bottom
Pedestal, Porous Stones, Ram and Top Cap.............252
117 Exploded View of Proposed Pressure Dome................253
118 Sectional and Bottom Views of Main Housing
Portion of Proposed Pressure Dome ................. 254
119 Sectional, Top and Bottom Views of Retaining
Flange Portion of Proposed Pressure Dome ............ 255

120 Sectional, Top and Bottom Views of Upper Pressure
Plate Portion of Proposed Pressure Dome..............256
121 Sectional, Top and Bottom Views of Lower Pressure
Plate Portion of Proposed Pressure Dome..............257
122 Oblique View of Future Stand and Plumbing System
for Proposed Swell-Consolidometer ................. 265
123 Side View of Future Stand and Plumbing System for
Proposed Swell-Consolidometer ........................ 266
124 Schematic of Future Plumbing and Electrical
Systems for Proposed Swell-Consolidometer ............ 267

1.1 Problem
The study of expansive soils has been going on for many
years. Although much has been learned about the behavior of
these soils, extensive damage to structures founded on such soils
each year is still very common, and often results in very
expensive mitigation and repairs not to mention staggering legal
costs. Unfortunately, even after mitigation measures are taken
and repairs are complete, swelling often continues with
additional damage occurring in the future.
In 1973 Jones and Holtz [20] estimated that each year
expansive soils were responsible for at least 2.3 billion dollars
in damage to houses, buildings, roads and pipelines in the United
States. In 1988, Chen [10] cited a 1978 National Science
Foundation Study that projected annual losses due to expansive
soils would exceed 4.5 billion dollars by the year 2000. Chen
also emphatically stated that with present day knowledge of
expansive soils economic design and construction of structures
to withstand movement caused by expansive soils is not yet always
possible. Hence, there is still a great need to continue studying
the mechanisms and behavior of swelling soils.
To date, research and testing of expansive soils has
concentrated primarily on the aspects of vertical swelling
potential and vertical swelling pressure. It has been reasoned
[19] that the vertical and horizontal swelling characteristics

of a soil should be isotropic. This is based on the premise that
under conditions of zero volume change the swelling pressure
originates from the pressures developed in water as it is
absorbed into the soil. Being a liquid, water exerts isotropic
pressure, and therefore, the swelling pressure should also be
isotropic. Consequently, extensive study of horizontal swelling
has not been accomplished, even though studies to date have
suggested that horizontal and vertical swelling characteristics
are not the same under various boundary conditions, states of
initial stress, moisture content, dry density, method of sample
preparation, and methods of test [7,10,14,24,39,46],
Based upon the author's professional experience within
the last 8 years (particularly the last 5), the effect of lateral
swelling pressures are typically ignored in the design of most
"routine" structures. However, lateral swelling pressures can
be of significant importance in the design of retaining walls,
basement walls, drilled piers, buried conduits, etc. For
example, in the Denver, Colorado area many instances of bowing
or collapsing basement walls have recently been reported [10].
Although many factors enter into the causes for this type of wall
distress, a recent laboratory study by Chen [10] on expansive
soil backfill suggests that horizontal swell pressures may exceed
vertical swell pressures by as much as 3.6 times until complete
saturation of the backfill is reached. As saturation is reached
the horizontal swell pressure decreases to nearly the same

magnitude as the vertical swell pressure. Results of similar
studies by others [21,25] have not substantiated these results
but have shown that lateral swelling pressures can be
significantly higher than typical at-rest or even passive earth
To date, several standard laboratory test procedures have
been developed to measure vertical swell potential and/or
vertical swell pressure. The most common methods all utilize a
standard one-dimensional consolidometer and flooding of the
sample [1,2,7,9,10,14,15,24,26,28,34,40,41,42,46,51], Several
attempts [10,19,21,22,24,25,26,27,28,39,42,43,45,47] have been
made to measure lateral or (horizontal) swell potential/pressure,
some with promise and others with significant limitations. To
date, however, no standard testing equipment or methods have been
Several investigators have attempted to modify the
standard triaxial cell to enable these measurements
[19,24,39,42,47]. Recent testing by the Army Corps of Engineers
Waterways Experiment Station [19] utilized a double walled
version of a standard triaxial device to measure horizontal and
vertical swelling pressures simultaneously. According to the
Corps of Engineers, this testing was an extension and
modification to the triaxial test method previously used by
Snethen and Haliburton in 1973 to measure horizontal and vertical
swell pressures.

The concept of using a modified triaxial type cell for
measurement of vertical and horizontal swell pressures appears
quite feasible based on the recent work by the Army Corps of
Engineers. The primary advantages of using such equipment are:
1) Sidewall friction which occurs in one-dimensional
consolidometers is virtually eliminated.
2) The soil specimen can be subjected to back pressure
and a pressure gradient created across the sample to
promote sample saturation.
3) Lateral and vertical swell pressures can be measured
4) Pore pressures can be measured.
5) Nonuniform stress or strain effects can be reduced.
6) Different stress paths can be studied.
Major disadvantages include:
1) Sample setup time is longer.
2) More equipment is required to measure pressures.
3) Typically is not well suited for routine testing on
a production basis, at least at the time of this
1.2 Purpose
It is the purpose of this research to ultimately design
a new apparatus that can simultaneously measure vertical and
horizontal swelling characteristics of relatively undisturbed or
remolded soils in the laboratory. The objective is to develop

an apparatus that can measure swelling pressures developed under
constant volume conditions, or controlled lateral and vertical
strain, as well as having the capability of measuring expansion
potential under variable surcharges, or applied pressures. It
is also hoped that the time required for conducting the various
swelling tests can be reduced significantly by applying a
backpressure to the water used to inundate or wet the specimen
being tested.
It is desirable to develop an apparatus that is
economical, accurate and easy to use so that standardized testing
procedures can be established and the equipment used on a routine
basis in the private consulting sector. In this day of
computers, advanced electronics and servo-control systems, the
equipment should ultimately be designed to be fully automated
using a small desktop computer, thereby reducing laboratory time
to take readings, and adjust pressures, and reduce the time
required for manual preparation of laboratory data.
1.3 Scone of Study
The scope of this study consisted of the following basic
1) Performing a series of swell-pressure tests using
three commonly available types of laboratory equipment
to gain experience and insight into advantages and
disadvantages of each.

2) Review of the literature to determine various methods
of swell pressure measurement used by previous
researchers and various equipment developed for such
3) Based on information gathered in Steps 1 and 2 above,
a modified triaxial cell was designed to accomplish
the objectives of the research and incorporate as many
of the advantages of existing equipment as
Results of each step of the research are discussed in
detail below. The laboratory testing program was accomplished
prior to extensive literature review because of job relocation
to California. Nevertheless, the results of the literature
review are presented first for the sake of clarity and

2.1 General
Extensive study has been made over the last 30 to 35
years to determine the various factors which influence swelling
soil behavior. Excellent summaries of results obtained from many
of these studies are presented by Bowles [6], Chen [10], Parcher
and Liu [39], and Seed, Woodward, and Lundgren [46] among others.
It is not the intent of this paper to present a complete
review of the studies accomplished to date. However, based upon
a review of the above referenced work, a list of the principal
factors which affect the swelling characteristics of soil are
listed below:
1) Type of clay mineral.
2) Initial dry density.
3) Initial moisture content.
4) Surcharge and/or confining pressure.
5) Soil structure and fabric.
6) Initial stress and boundary conditions in soil.
7) Availability of free water or vapor to the soil.
8) Type and concentration of ions within the adsorbed and
absorbed water.
9) Thickness of soil stratum.
10) Curing time (in the case of remolded soils).

11) Time allowed for swelling to occur.
12) Prevailing climatic conditions.
13) Number of cycles of wetting and drying.
14) Temperature.
The primary effects each of these factors have on
swelling soil behavior is discussed briefly below. Their inter-
relationship is complex and no attempt is made here to fully
explain each.
2.1.1 Type of Clay Mineral
As noted by Parcher and Liu [39], the mineralogical
composition of soil is paramount in its ability to exhibit
swelling characteristics. Bowles [6] describes clay minerals as
being predominantly crystalline in structure and comprised mainly
of silicates of aluminum and/or iron and magnesium. Some also
contain alkalies and/or alkaline earths. Most clay minerals have
sheet or layered structures although some have fibrous or tubular
Clay minerals are typically very small (less than 1
micron). Because of their molecular structure and extremely
small size they are very electrochemically active and exhibit an
extraordinary affinity for water. This affinity for water
results in a plastic behavior which is not exhibited by other
minerals although they may be just as small. Quartz is an
example of such a non-plastic mineral.

Two basic molecular units known as silica tetrahedra and
octahedral units make up all clay minerals. A three dimensional
drawing, schematic representation and photograph of a model for
each of these two basic units are shown on Figure 1. Silica
tetrahedra are comprised of four oxygen atoms surrounding a
silica atom such that the four oxygen atoms form the points of
a tetrahedron. The octahedral units consist of six hydroxyl
molecules in the shape of an octahedron typically surrounding an
aluminum or magnesium atom. Sometimes an iron, titanium, nickel,
chromium, or lithium atom takes the place of the usual aluminum
or magnesium atom. The octahedral unit is termed brucite if the
metallic atom is primarily magnesium and gibbsite if it is
primarily aluminum. The silica unit has a height of
approximately 4.6 angstroms and the octahedral units are
approximately 5.05 angstroms high [6].
These basic units typically form in thin sheets or less
commonly bundles. The sheetlike units are stacked like pages of
a book to form clusters. Different clay minerals are created
depending upon the way in which these basic sheets are stacked,
the different metallic ions present in the basic units, and the
type of bonding that subsequently occurs. A net negative charge
is produced on the exterior of each of these clusters for all the
possible combinations of the basic units which form clay

Figure 1: Three Dimensional View, Schematic Representation and
Photograph of the Basic Tetrahedral and Octahedral
Molecular Units Which Make Up All Clay Minerals
(Adapted from Bowles [6], 1979, p.161)

In an effort to balance this charge deficiency, water
which is dipolar is commonly attracted to the surface of the clay
particles. Consequently, clay particles are almost always
hydrated, i.e. surrounded by layers of water molecules called
adsorbed water. This layer is commonly about two water molecules
thick and is referred to as the diffuse double layer, the diffuse
layer, or just the double layer. Engineering properties of clay
soils ultimately depend upon the nature of this double layer of
adsorbed water. Water molecules are adsorbed to the surface of
clay minerals as a result of dipolar attraction of water
molecules to the negatively charged clay surface, hydrogen
bonding between the hydrogen in the water and the oxygens or
hydroxyls on the surface of clay, and the attraction of similarly
hydrated cations within the water to the clay surface. The
attraction of wate^r to the clay surface is very strong near the
surface and diminishes with distance from that surface.
The double layer of adsorbed water may be lost at
temperatures higher then 60C to 100C (140F to 212F). Some
of the adsorbed water may also be lost as a result of air drying.
The loss of the double layer will reduce the natural plasticity
of the clay and alter its engineering behavior. If dehydrated
at low temperatures the plasticity properties can usually be
recovered by mixing the clay with sufficient water and allowing
it to cure for at least 24 to 48 hours. If dehydration occurs
at high temperatures the loss of plasticity properties is

generally permanent.
Clay minerals have sufficient attractive potential to
hydrogen ions that a layer of water up to about 400 angstroms
thick can be absorbed around the clay mineral in addition to the
adsorbed double layer. It is this absorption of water that
causes clay sheets to spread apart, i.e. swell. If the clay
sheets are restrained from spreading, swell pressures are
Because different clays have crystal structures of
different sizes, specific surface, and different charge
deficiencies, the activity or ability of each clay to absorb
water and swell also varies substantially. From a geotechnical
engineering standpoint kaolinite, illite and montmorillonite are
the three basic groups of clay minerals of primary importance.
Kaolinite is the most stable of the three minerals and
consists of alternating sheets of silica tetrahedra and alumina
octahedral (gibbsite) units. The tips of the silica units are
embedded in the octahedral units and form a single layer
approximately 7 angstroms thick. It is therefore called a 1:1
mineral. A kaolinite cluster consists of a stacking of 70 to 100
or more of these 7A sheets like pages of a book. Strong hydrogen
bonds and van der Waals forces are present between each of the
7A sheets. The bonding combination which results has
considerable strength and stability with little tendency for
water to be absorbed between the individual layers. Kaolinite

has a typical specific surface on the order of 15 m/g and an
activity of 0.3-0.5 [6], The common schematic representation of
this mineral is shown in Figure 2.
Illite is of intermediate stability and is composed of
an octahedral sheet of gibbsite sandwiched between two sheets of
silica tetrahedra resulting in a layer approximately 10 A thick.
This produces a 1:2 mineral. Layers are stacked to form crystals
approximately 250 to 280 A thick. The layers are bonded together
relatively strongly by potassium ions which just fit into the
hexagonal spaces in the bottom of the silica sheets. Although
relatively strong, this bonding is weaker than the bonding in
kaolinite. In addition, some of the silica atoms are filled with
aluminum atoms. Illite has a typical specific surface on the
order of 80 m/g and an activity of 0.5-1.3 [6], Its common
schematic representation is presented on Figure 3.
Montmorillonite is the most unstable of the three basic
clay groups. It is composed of one octahedral sheet of gibbsite
sandwiched between two sheets of silica tetrahedra to form a
single layer approximately 9.6 A thick. Like illite it is also
a 1:2 mineral. However, the bonding between the layers is
primarily the result of van der Waals forces and is therefore
very weak relative to hydrogen bonding in kaolinite or ion
bonding in illite. Crystals are commonly about the same
thickness as the individual layers. Because of the weak bonding
between layers and because the octahedral sheet has a net

double layer of water
with Mg, K, Ca ions
V)| cl
c o<
.y, O
o. o
jQ o
+ o
o O
o O
o h- ii
Ca, K, or Mg
Figure 2: Typical Schematic Representation of Kaolinite (after
Bowles [6], 1979, p.162)

Figure 3
ff) OO (K) (K^
\________________/ ^fixed ions
Typical Schematic Representation of Illite (after
Bowles [6], 1979, p.165)

negative charge, water and exchangeable ions can enter and
separate the layers readily. Montraorillonite has a typical
specific surface on the order of 800 m/g and an activity of 1.5-
7 [6], A schematic representation of this mineral is presented
on Figure 4.
Referring to the information presented above it should
be noted that the specific surface (i.e. ratio of surface area
to volume of mass) for the least active clay or Kaolinite, is
much, much smaller than that for the most active clay,
montmorillonite. Specific surface accounts for the large
difference in potential moisture contents of these clays, as the
higher the specific surface, the more water it takes to cover the
surface of the clay mineral, and the more faces available to
attract and absorb water. Hence, montmorillonite has the ability
to absorb much greater amounts of water (i.e. swell) than
Kaolinite and Illite.
2.1.2 Initial Dry Density
In general, the denser an expansive soil, the greater the
swell potential and the greater the swell pressure [10,29]. As
shown in Figure 5, tests conducted by Chen [10] for samples at
consistent moisture content under a given surcharge, but varying
initial dry density indicate that as density increases, the swell
potential (percent volume) increases linearly. Likewise, as
reproduced in Figure 6, the log of swelling pressure versus dry

Figure 4: Typical Schematic Representation of Montmorillonite
(after Bowles [6], 1979, p.166)

Figure 5: Relationship Between Initial Dry Density and Volume
Change for Samples with Constant Initial Moisture
Content (Adapted From Chen [10], 1988, p.85)

Figure 6: Relationship Between Initial Dry Density and Swelling
Pressure for Samples with Constant Initial Moisture
Content (Adapted from Chen [10] 1988, P.86)

density is also linear with swell pressure increasing as the dry
density increases. Tests by Komornik and Zeitlen [29] show the
same trend but do not necessarily indicate the trend to be
2.1.3 Initial Moisture Content
Research has shown [10,29,34] that for a given value of
initial dry density (or void ratio), the swell potential i.e.
percent volume change increases with decreasing moisture content
(see Figure 7) Swell pressure is also affected by initial
moisture content, however, results in the literature are often
contradictory. Chen [10] reported that samples compacted to the
same initial dry density at different moisture contents yielded
approximately the same value of swelling pressure and
hypothesizes that swell pressure is an intrinsic soil property
which varies only with initial dry density for a given clay
mineral. Komornik and Zeitlen [29] presented results that show
for samples with the same initial dry density, swelling pressure
decreases somewhat with decreasing initial moisture content.
Kassiff and Shalom [23] present results showing remolded samples
of nearly the same initial dry density, but different initial
moisture contents to have lower final swelling pressures for
higher initial moisture contents. Similarly, Komornik and David
[28] found that from statistical analysis of over 200 disturbed
and undisturbed clay samples, no correlation existed between

Figure 7: Relationship Between Initial Moisture Content and
Volume Change for Samples with Constant Initial Dry
Density (Adapted from Chen [10], 1988, p.80)

swell pressure and moisture content alone. Initial dry density
was also a factor.
2.1.4 Surcharge and/or Confining Pressure
Nearly all the studies in the literature present results
which show that for conditions of equal initial dry density and
moisture content, as the surcharge or confining pressure of an
expansive soil is increased, the swell potential (percent swell)
is reduced. Figure 8 presents the results of work by Chen [10]
regarding this aspect. The swelling pressure which develops in
turn depends upon the amount of swell which may occur. However,
for a given dry density, it has been shown that their exists a
single value of swell pressure when no volume change is allowed
to occur regardless of initial surcharge pressure.
2.1.5 Soil Structure and Fabric
Parcher and Liu [39] state that
In naturally sedimented soils the
arrangement of the grains will be affected by
the nature and sizes of the particles and by the
environmental conditions prevailing during their
deposition. During compaction the arrangement
of grains in the soil structure will be
influenced by the energy of compaction effort,
by the manner in which compaction is
accomplished, and by the moisture content during
compaction. These same factors also produce
measurable effects on the swelling that may be
realized by the soil under given conditions.

Figure 8: Relationship Between Surcharge Pressure and Volume
Change for Samples with Constant Initial Dry Density
and Moisture Content (Adapted from Chen [10], 1988,

Komornik and Zeitlen [29] also reference work by Seed et.
al. that apparently indicate methods of compaction (i.e. dynamic
or kneading versus static) produced marked influence on the
swelling characteristics of an expansive clay. However, in their
work Komornik and Zeitlen found that for soils having a
relatively high plasticity which are compacted to a low relative
density (say less than 90 percent of maximum density as
determined by ASTM Test Method D1557) the resulting structure
of the soils was essentially the same for dynamic and static
Komornik and Livneh [27] showed that samples remolded
using dynamic compaction above optimum moisture content resulted
in anisotropic soil structure which measurably affected the
vertical swell potential depending upon direction of sampling
with respect to direction of compaction.
Chen [10] points out that permeability of a soil deposit
determines the rate of progress of water into the soil, be it by
gravitational flow or by diffusion. The rate of water flow into
the expansive soil then directly effects the rate at which heave
may occur. Holtz and Kovacs [18] and Bowles [6] point out that
permeability and soil strength is strongly influenced by the soil
macrostructure. Hence the rate of swell and amount of swell
and/or swell pressure can also be strongly affected by the soil

2.1.6 Initial Stress and Boundary Conditions in Soil
As previously mentioned, surcharge or confining pressure
restricts the amount of swell that may occur in a natural soil
deposit. Similarly, the nature of existing in situ stresses will
affect how the soil will react when it increases in moisture
content. Initial stresses may change as a result of loading or
unloading the soil by structures, fill and excavation. Drying
of the soil can also increase the effective stresses within the
soil. Depth of the water table or location of non-expansive
layers may also affect the degree of swelling that may occur.
Proximity to structures can also affect stress conditions as
noted by Komornik and Zeitlen [30],
2.1.7 Availability of Free Water or Vapor to the Soil
Work by Chen [10], Kassiff and Shalom [23] and Komornik,
Livneh and Smucha [31] have indicated that there exists a
limiting amount of swell or swell pressure development which will
occur for discrete amounts of moisture made available to a
sample. Hence, if only a small amount of water is available to
a soil, full swelling or full development of swell pressure may
not be realized as opposed to if an unlimited supply of moisture
is available. In general, the more moisture made available to
an expansive soil, the closer the sample will come to reaching
its maximum swelling potential or swell pressure for a given
condition of confining pressure, initial dry density and initial

moisture content. It is also important to note, however, that
Chen [10], Kassiff and Shalom [23], and Komornik et. al. [31]
have found that very small increases in moisture content, even
on the order of 1 percent can result in substantial even
detrimental amounts of swelling and/or swell pressure to develop.
2.1.8 Type and Concentration of Ions within the Lavers of
Adsorbed and Absorbed Water
As previously mentioned water is attracted to the surface
of clay minerals. This attractive force is very strong at the
surface of the clay but diminishes with the square of the
distance from the surface. This attraction is a result of the
surface of the clay having an unbalanced negative charge. This
unbalanced negative charge is caused by isomorphous substitution
in the crystal lattice and imperfections, such as "broken" edges
of the crystal. Depending upon the amount of unbalanced negative
charge, positively charged cations within the pure water may be
attracted to the clay surface to varying degrees to satisfy the
unbalanced charge. These ions then cause different degrees of
ionic bonding to occur between the basic unit layers. However,
these cations are exchangeable; that is they can be replaced by
one or more cations that satisfy the valence deficiency of the
clay crystal. Calcium (Ca+) and magnesium (Mg+) are the
predominant exchangeable cations in soils. Potassium and sodium
are less common.
The depositional environment as well as

subsequent weathering and leaching will govern what ions at
present in a particular soil deposit. The relative ability of
the different cations to replace others are depicted below. The
further to the left, the more easily the cation may be replaced
by a cation shown to its right.
Li < Na < NH4 < K < Mg < Rb < Ca < Co < AL
For example Ca will more easily replace Na or Mg than Na or Mg
will replace Ca.
The exchangability of cations is dependent primarily on
the valence of the cation but also the size. Higher valence
cations easily replace lower valence cations and for ions of the
same valence, larger ions can more readily replace smaller ions.
There size relative to "holes" or spaces in the crystal lattice
are also important. For example potassium ions are almost the
perfect size to fit in the hexagonal spaces created by the silica
sheets. This lends itself to a strong bond between the clay
crystal and the monovalent cation. In general, the greater the
replacement ability of cations, the stronger the ionic bonds
between the clay crystals and cations, and the more stable the
clay structure becomes, as it is less easy for water to enter
between the crystal layers. This is the basic concept used for
chemical stabilization of expansive soils. Another effect of
cation absorption within the double layer of water is an osmotic
effect. The cations create a concentrated salt solution. When
"pure" water or water of a lower salt concentration comes into

contact with the soil, the external water is absorbed by the salt
solution in an effort to dilute the solution, and create an
equilibrium condition. Hence, additional swelling can result
from high ionic concentrations within the adsorbed pore water
(double layer).
2.1.9 Thickness of Soil Stratum
The thickness of the clay stratum or clay specimen
effects the time required for the soil to become fully saturated
and hence attain its full swelling potential. It is primarily
a question of permeability. The thicker the stratum the longer
it takes water to travel through it.
In addition, the thicker the stratum or specimen, the
greater the absolute change in thickness when allowed to swell.
The relative or percent volume increase is the same regardless
of the thickness. The swell pressure that can develop is also
independent of the stratum or specimen thickness.
2.1.10 Curing Time
Because of the fine grained nature of expansive soils,
permeability is low and uniform distribution of moisture within
a specimen is difficult and takes significant time to achieve.
For an unsaturated soil to be at equilibrium the soil moisture
must be uniformly distributed. If it is not erratic measurements
of swell potential and/or swell may occur. Parcher and Liu [39]

reported on work by E.S. Barber which indicated swell pressures
for remolded specimens decreased measurably with increasing
curing time. Lateral stresses induced by compaction also
dissipate with time as reported by Komornik and Zeitlen [29].
Failure to account for this effect leads to erroneously high
values of initial lateral swelling pressures.
2.1.11 Time Allowed for Swelling to Occur
Chen [10] indicates that the time required for a soil to
reach its maximum swell potential varies considerably and depends
upon the initial dry density, permeability, thickness of specimen
(or deposit) and availability of water. As confirmed by Komornik
and Zeitlen [29], primary and secondary swelling occurs and time
for each can be determined by plotting elapsed time with percent
swell or swell pressure, similar to consolidation testing. If
sufficient time is not allowed for a material to become fully
saturated, full swelling may not be realized. Allowing swelling
to occur for an overnight period as is done in many laboratories
may not be sufficient to allow full swelling to be realized.
Depending on the initial conditions, soaking samples in water for
a period of one day to several weeks may be necessary.
2.1.12 Prevailing Climatic Conditions
Climatic conditions affects precipitation, evaporation
and transpiration and thereby, strongly influences the movement

of moisture within a soil deposit. The depth and degree of soil
desiccation is also affected by climatic conditions and can have
a significant impact on soil permeability. As previously noted,
permeability has a strong impact on swelling behavior.
2.1.13 Number of Cycles of Wetting and Drying
Chen [10] presents results of tests conducted by himself
and also sites work by T. Chu and C.H. Mou and also by M. Popescu
that indicate that the percent expansion of soils tend towards
an equilibrium value after several cycles of wetting and drying.
The results of Chen's tests are reproduced in Figure 9. The
reason for this phenomenon is theorized by Chen to be a result
of the dry density initially decreasing to a critical value where
swelling and subsequent shrinkage equalize.
2.1.14 Temperature
Chen [10] emphasizes that heaving of expansive soil may
take place without the presence of free water. The reason for
this is vapor transfer through the soil. Water vapor in the soil
will migrate from an area of higher (warmer) temperature to lower
(cooler) temperature, to equalize the thermal energy gradient
between the two. Upon reaching the cooler area such as beneath
buildings, sidewalks and pavements, condensation can take place
and provide sufficient moisture increase for significant swelling
to occur. Results of tests conducted by Kassiff and Shalom [23]

Figure 9:
Relationship Between Number of Cycles of Wetting and
Drying and Volume Change (Adapted from Chen [10],
1988, P.43)

confirms that only a very small percentage of moisture increase
is necessary to cause swelling or swell pressure to develop
rapidly. A summary of their results are reproduced in Figure 10.
2.2 Summary
From even the general nature of the summary presented
above, it may be concluded that the possible combinations and
permutations of the factors affecting swelling soil behavior are
enumerable. As such, it is extremely difficult to test these
soils in the laboratory and be able to directly compare results,
much less be able to create an accurate model of the usually
highly variable and constantly changing in situ conditions
encountered in practice. Nevertheless, attempts to do so are
necessary in order to obtain a general range of behavior that may
be anticipated for typical conditions expected during the life
of a structure so that an attempt can be made at rationally
deriving an engineered design. Hence, recognition of existing
and future conditions, accurate and sufficient testing, and
general experience are all necessary to successfully deal with
expansive soil problems.

Moiilur* cont*nl ahtr t*11; *
Moiiturt content gfttr tit ; X
Figure 10: Effects of Moisture Content Increase on Volume Change
and Swell Pressure (Source: Kassiff and Shalom [23],
1971, Geotechnique. p.250)

3.1 General
Since the late 1950's and early 1960's, several different
methods have been devised to measure swelling pressure exerted
by expansive soils. Three basic methods have been developed
which are still commonly used today and are listed below:
1) Free Swell-Reload Method (a.k.a. Pre-swelled Method)
2) Variable surcharge Swell Method (a.k.a. Different
Pressures Method).
3) Constant Volume Swell Method.
All three of the methods listed above commonly utilize
a cylindrical specimen confined in a ring which is loaded in a
typical consolidometer device.
3.2 Free-Swell Reload Method
The basic technique of this method involves placing a
small "seating load" on a sample (usually 100, 144 or 200 psf),
allowing it to consolidate, then flooding the sample with water
and allowing the sample to swell fully. Time for full swelling
may be determined by plotting percent swell versus log time,
similar to percent consolidation versus log time plots, and
estimating the time at which primary swelling is complete.
However, more often than not, an arbitrary time period, such as

24 hours, is allowed for swelling. Figure 11 demonstrates this
concept. The time required for primary swelling to occur may
range from several hours to several days and depends upon the
soil type, sample thickness, sample diameter, initial dry density
and initial moisture content. In the Denver-Metro area, 24 hours
is commonly used.
After primary swelling is completed, the specimen is
loaded in increments as in a standard consolidation test, until
it compresses to the void ratio obtained just prior to flooding.
The pressure at which the specimen reaches the same void ratio
as the pre-flooding value is deemed the swelling pressure (see
Figure 11). Katti, Lai, Kulkarni, and Fotedar [24,25] termed
this pressure the "apparent swelling pressure". Over time this
test method has been modified by many researches and consultants
so that the initial surcharge load applied to the sample prior
to flooding consists of a load equivalent to the anticipated
overburden and/or any structural loadings from foundations [9].
The sample is then flooded and allowed to swell fully. The
remainder of the test procedure is essentially unchanged. With
this modification researchers believed that the in-situ stress
history would be represented more realistically, as described by
Brackley in 1975 [7]. Separate research by Chen [9] and Brackley
[7] in 1975 concluded that swelling pressure measured using the
same method was dependent upon the initial density of the
specimen. Chen also concluded that the measured swell pressure

Figure 11:
Swell-Reload Method to Determine Swell Pressure

is independent of the magnitude of the initial preload or
surcharge pressure, moisture content and initial degree of
saturation. However, Chen's conclusions with regard to swell
pressure being independent of initial surcharge seems to
contradict with the conclusion that swell pressure depends on the
initial dry density of the soil. As surcharge pressures are
increased prior to the addition of water, the soil compresses and
becomes increasingly dense. Hence, the swell pressure should
also increase.
3.3 Variable Surcharge-Swell Method
This method, also referred to as the "Different Pressures
Method" by El-Sohby and Mazen [14] consists of subjecting
specimens having identical initial dry density and moisture
content to different initial surcharge pressures, allowing them
to compress and then flooding them with water. The resulting
volume change (i.e. percent swell) is measured. By plotting the
resulting percent swell versus initial surcharge pressure, the
pressure corresponding to zero volume change can be extrapolated
or interpolated and is deemed the swelling pressure (see Figure
Brackley [7] found that for one South African soil the
relationship between percent swell versus logarithm of applied
surcharge pressure is linear, making the determination of swell
pressure theoretically obtainable by testing only two identical

Figure 12:
Variable Surcharge Swell Method to Determine Swell Pressure

specimens at different initial surcharge pressures and
extrapolating (or interpolating) a straight line between them and
obtaining the pressure corresponding to zero volume change.
Other investigators including El-Sholby and Mazen [14], Chen
[9,10], and Porter and Nelson [40], have presented results
indicating this relationship is not always linear. This method
would seem to simulate potential in situ stress paths better than
the swell-reload method, but has the problem of causing an
initial increase in soil density prior to flooding the specimen
with water. On the other hand, it has the advantage of not
creating the hysteresis effects of swelling then reloading which
occur in the swell-reload method.
3.4 Constant Volume Method
This method consists of flooding a sample and restraining
it from increasing in volume by use of a locking-type device,
proving ring, or by continually adding surcharge load or applying
pressure to maintain the specimen at a constant height (volume
is constant if laterally restrained and height is maintained
constant). The pressure at which no further tendency to swell
occurs is deemed the "true" swelling pressure. The stress path
for such a test is shown in Figure 13.
A variation of the constant volume method as described
by Porter and Nelson [40], allows the swelling pressure to be
determined for a given amount of swell (i.e. strain). By first

1 0
1 0
Figure 13: Constant Volume Method to Determine Swell Pressure

allowing full swell pressure to develop under constant volume
conditions, the constant volume swell pressure is determined.
Then, by allowing discrete increments of strain (swelling) to
occur the resulting pressure can also be determined. This
technique is termed strain controlled testing and is depicted in
Figure 14.
3.5 Comparison of Methodologies
Unfortunately, data found in the literature suggest that
the three basic methods of swell pressure measurement do not
yield the same value of swelling pressure. Studies by El-Sohby
and Mazen [14] and Brackley [7] indicate that the free swell-
reload method always yields swell pressure higher than those
determined using the variable surcharge-swell method. Escario
[12], and Porter and Nelson [40] found that the free swell-reload
method resulted in higher values of swell pressure than did the
constant volume method. Brackley [7] and Noble [34] determined
that the constant volume and variable surcharge-swell methods
gave swelling pressure values that were nearly the same but
considerably lower than obtained using the free swell-reload
method. Chen [9,10], however, concluded from his research that
the free swell-reload procedure produced swell pressures which
were virtually the same as those determined by
extrapolating/interpolating results from variable surcharge-swell

1% INCREMENTS--------"
1 0
14: Controlled Strain Method to Determine Swelling Pressures

In all the methods described above, the specimen is
confined laterally and vertical strain of the soil is varied or
maintained constant. Measurements of horizontal swelling
pressure are generally not made except for research purposes.

4.1 General
As previously described, the measurement of expansive
swell pressure most commonly used is based upon measuring
pressures exerted in the vertical direction while restraining the
soil laterally. This appears to be a result of the assumption
that swelling pressure is an isotropic property of the soil, as
well as the fact that most problems with swelling soils seem to
have been attributed to vertical movements.
4.2 General Approaches and Results
In 1965 Parcher and Liu [39] at Oklahoma State University
conducted some of the first research on lateral swelling to
investigate this assumption using samples of natural and
compacted clay. In general, the results of the study found that
under conditions of free vertical and lateral strain, horizontal
swelling was greater than vertical swelling. In their study
lateral swelling measurement was enabled by essentially modifying
a triaxial cell.
The loading ram or top cap was made the Same diameter as
the specimen to separate effects of vertical and lateral strains.
The specimen was placed in a thin latex membrane and placed
inside the modified cell which was completely filled with water.

Vertical swelling was measured by a dial gauge placed on the top
cap. Lateral swelling was obtained by measuring the volume of
water displaced from the cell as the sample strained laterally.
In addition, the quantity of water absorbed by the soil specimen
was determined by measuring the volume of water drawn from a
reservoir using a micrometer surface gauge. The test apparatus
utilized by Parcher and Liu would appear to be capable of
producing very accurate results. However, the apparatus as used
was not capable of measuring swell pressures.
In 1965, in Haifa, Israel, Komornik and Zeitlen [26] also
began the study of lateral swelling pressure. They developed a
modified odometer ring which could measure small lateral strains
using electrical wire strain gauges, while utilizing the variable
surcharge-swell method for vertical strains.
Typical results indicated that for conditions of
essentially zero volume change, horizontal swelling pressures
were less than the vertical pressures. However, as vertical
surcharge pressures were decreased below the pressure required
for zero vertical strain, vertical strain and the ratio of
lateral swell pressure to vertical surcharge increased. For free
vertical swell the ratio of lateral swelling pressure, ffSL, to
vertical surcharge, CTV, was approximately 13. The ratio dSL/CTsv
at essentially zero volume change was approximately 0.72 to 0.80.
The horizontal swelling pressures were substantial, ranging from
approximately 1330 psf to 4809 psf for conditions of free

vertical swell and zero vertical swelling, respectively. These
lateral pressures are considerably higher than typical at-rest
earth pressure or even passive pressures.
In 1967 at the Israel Institute of Technology Komornik
and Livneh [27] extended the 1965 work of Komornik and Zeitlen
[26] by using the same soil and the same type of swelling
pressure rings. The intent of this study was to determine the
effect of soil anisotropy on swelling characteristics of
compacted clay. Their work was largely based on work done by
Seed et. al. in 1962, which was reported to show that particle
rearrangement into parallel type structure does occur under
dynamic or kneading type compaction at high degrees of
saturation, or high moisture contents. By dynamically compacting
specimens to a constant density at 95 percent saturation
anisotropic soil structure was created.
The swell pressure rings were extruded into the compacted
specimens in directions parallel and perpendicular to the
direction of compaction. Test results showed that using the
variable surcharge-swell method, at nearly zero lateral strain,
and a given percentage of vertical strain, the lateral swell
pressure was lower for samples having their central axis oriented
parallel to the direction of compaction as compared to those
specimens oriented with the central axis oriented perpendicular
to the direction of compaction (refer to Figure 15). However,
for conditions of zero vertical swell, the vertical swell

Figure 15: Swelling for Anisotropic Conditions

pressure was the same, regardless of axis orientation. What is
of considerable interest, however, is that even at moisture
contents 5 percent above optimum and 92 percent relative
compaction (using AASHTO Test Standard T180) lateral pressures
measured ranged from 3683 to 6241 psf for 3 percent and zero
percent vertical swell, respectively. These pressures are much
greater than conventional at-rest or passive pressures typically
used for design. By extrapolating to obtain swell pressures for
conditions of constant volume these test results also indicate
sl/sv = 0-85.
Additional study of lateral swelling pressure was
conducted in 1969 in India by Katti, Lai, Fotedar and Kulkarni
[24,25] Their research was conducted using both small and large
scale laboratory tests of compacted expansive clays. Small scale
testing included the modification of a triaxial cell to
simultaneously measure vertical and lateral swelling pressures.
Modification to the cell included making the top cap/loading ram
the same diameter as the specimen to separate vertical and
lateral behavior. The cell was also fitted with proving rings
to measure vertical pressures and Bishop's pore pressure
apparatus was used to measure lateral pressures. Lateral
swelling pressures were taken to be the pressure developed within
the confined fluid surrounding the specimen as the specimen
attempted to swell laterally.
Large scale tests included compacting soil into a "rigid"

box approximately 3 feet deep, 4.5 feet wide, by 10 feet high.
Free vertical swell was allowed while lateral swelling was
essentially prevented and resulting lateral swell pressures
measured at several discrete depths using proving rings with
plungers attached to metal plates.
Results of triaxial tests using the constant volume
method indicated the ratio of CSL/asv ranged from 1.00 to 1.48 and
was dependent upon the initial void ratio (or dry density) .
These results are contrary to the Osl/osv ratios for constant
volume obtained by previous researchers [26,29],
The large scale test results indicated that the lateral
swell pressure increased from zero at the surface to a constant
value at depths of 4 to 5 feet below the surface. The constant
lateral swell pressure attained was approximately 6,000 psf. The
large scale tests also indicated that the percent vertical swell
decreased to zero between depths of 2.5 and 3.5 feet. Hence,
constant volume conditions would be expected below say 3.5 feet
even if the surface was able to swell freely.
These results may be somewhat misleading considering that
the ratio of least sample dimension to depth of the specimen was
approximately 0.33. With such dimensions the friction effects
of the box on the sides of the specimen were likely quite high,
thus not allowing truly representative vertical swelling. In
comparison, ASTM recommends a minimum diameter to height ratio
of 4 to reduce the effects of ring friction in standard oedometer

Several tests were also conducted in the direct shear
device to measure the vertical swell pressure using the swell-
reload method. Comparison of results obtained for two different
soils at equal initial void ratios indicated that the vertical
swell pressure measured using the swell reload method were 1.49
to 2.50 times larger than the vertical swell pressure obtained
using the constant volume method in the triaxial cell.
Other significant results of this research were that the
lateral swelling pressure measured in the triaxial cell increased
to a peak value at a saturation less than 100 percent, then
decreased to an equilibrium value as the percent saturation
increased. The peak lateral swell pressure was typically
observed to occur between 70 and 95 percent saturation depending
upon the initial molding void ratio. In general, the peak
lateral swelling pressure occurred at lower degrees of saturation
for increasing initial void ratio (i.e. as the initial density
decreased, the degree of saturation for peak lateral swelling
also decreased). Additionally, the numerical difference between
the peak lateral swell pressure and equilibrium swell pressure
increased for increasing dry density (decreasing void ratio) but
the ratio of aSL peak/aSL equilibrium decreased with increasing
density and ranged from 1.86 to 1.41. Finally, the relationship
between log of lateral swell pressure versus initial void ratio
is nearly a straight line for a given soil type.

In 1970 Komornik and Zeitlen [29] presented results of
additional testing using the lateral swelling pressure ring as
previously described and the same montmorillonite clay previously
tested. Approximately 90 tests were conducted using statically
compacted specimens after preliminary testing suggested
negligible difference in swelling characteristics between
statically and dynamically compacted specimens. The negligible
difference in swelling characteristics appeared to be contrary
to the results obtained by Komornik and Livneh [27] but was
attributed to the high plasticity and relatively low densities
of the compacted specimens, which did not permit significant
anisotropic structure to be developed.
Vertical and lateral swelling characteristics were
determined at relative compactions of 75, 81 and 87 percent at
moisture contents of 4 percent below, 1 percent above and 6
percent above optimum moisture content (based on ASTM Test
Standard D1557). The variable surcharge-vertical swell method
was utilized with simultaneous measurement of lateral swell
pressure under conditions of approximately zero lateral strain.
Test results indicated that lateral swell pressures
increased with increasing dry density for samples of equal
moisture content and percent vertical swell. At constant dry
density and percent vertical swell, the lateral swell pressure
decreased with increasing moisture content but only slightly.
The lateral swelling pressure ranged from a low of approximately

818 psf at 75 percent relative compaction, at 6 percent above
optimum moisture content and 4 percent vertical swell to a high
of 4911 psf at 87 percent relative compaction, at 4 percent below
optimum moisture and zero percent vertical swell. The high value
is greater than the low value by a factor of six, indicating a
wide range of potential lateral pressure. Even so, the low value
of 818 psf is substantially higher than conventional at-rest
earth pressures.
For comparison, consider a typical expansive clay which
has an effective angle of friction, 0 equal to 25 degrees. The
lateral earth pressure coefficient for at rest conditions, 1^,
can be roughly approximated by 1-Sin(25) or 0.6. The lateral
earth pressure at rest, ah, is assumed to be equal to KoCTy. For
a soil having a total unit weight of 130 pcf, the lateral
pressure at the bottom of a 10 foot wall (assuming it is rigid
and at rest conditions apply) is approximately CTh = 0.6(130)(10)
= 780 psf, which is less than the 818 psf obtained for very wet,
poorly compacted swelling soil backfill.
Interpolating the results of the variable surcharge-swell
tests to determine the vertical swell pressure at zero percent
swell (constant volume) and corresponding lateral swell
pressures, ratios of lateral swell pressure to vertical swell
pressure, oSL/osv were as shown below:

Table 1
Lateral Swell Pressure Ring Tests Using
Variable Surcharge-Swell Method
Relative Percent Compaction
_______per ASTM D1557_______
At Constant
.86 1.45
1.08 1.17
0.74 0.80
Lateral swell pressures exceeded the applied vertical
stress for vertical swelling greater to or equal to zero percent
for relative compaction of 75 percent. For relative compaction
of 81 percent lateral swell pressure is also greater than applied
vertical pressure for vertical swell greater than about one
percent. For relative compaction of 87 percent, lateral swell
pressure is greater than applied vertical pressure for vertical
swell greater than one percent for moisture contents 6 percent
above optimum and for vertical swell greater than approximately
2.5 percent for moisture contents 4 percent below optimum. Based
on these results, for a given moisture content and zero volume
change conditions, there may be a density below which lateral
swelling pressures exceed vertical swelling pressures and above
that density, lateral swell pressures are less than vertical
swell pressures. In general though, the higher the applied
vertical stress, the higher the lateral swell pressure. In other
words, the more vertical restraint, the more lateral pressure
that can develop.

Komornik and Zeitlen concluded that lateral swell
pressures can exceed applied vertical stress (i.e. overburden
stress) except when vertical swell is close to zero for higher
density materials. Hence, lateral swelling pressure would be the
major principal stress and crSL/av exceeds 4 for conditions of
near free vertical swell and restrained lateral movement.
In 1975 Ranganatham and Pandian [42] described an
apparatus also designed to measure lateral and vertical swell
pressures. As described, the apparatus essentially consists of
an oedometer ring modified to be similar to a small triaxial
cell. The basic concept consists of an oedometer ring split into
6 equal sections and glued to the inside of a typical rubber
membrane used for triaxial testing. A specimen is placed in the
ring and the membrane attached and sealed to upper and lower
retaining rings. A small anulus is created between the ring and
surrounding cell which is closed at the top and bottom. The
annulus is completely filled with de-aired water. Vertical
pressure is applied with a vertical ram similar to a normal
consolidometer while lateral pressures are measured using a null
pressure device. As the specimen tends to swell the split ring
expands and tries to displaces water from the annulus. Water
displacement is not allowed and the pressure created is measured
using Bishops null pressure device. The system would appear to
have merit but no test results were presented. Calibration test
results indicated that pressures applied within the ring filled

with fluid was the same as those measured in the anulus fluid
using the Bishop device. However, potential disadvantages
include interference of the upper ring if some lateral swelling
and vertical swelling were to occur or were desired. Also, the
problem of ring friction is still present.
An attempt was made to measure lateral swell pressure in
large scale field tests by Robertson and Wagener in 1975 [43].
Two pits approximately 5m x 5m in plan dimension (16.4' x 16.4')
and approximately 2.2 m deep (7.1') were backfilled using hand
and mechanical tampers. Total pressure load cells oriented
horizontally and vertically were carefully placed approximately
1 meter (3.28 feet) below the ground surface. Using vertical and
horizontal sand drains installed in the pits, the backfill soils
were saturated over a period of months, and resulting lateral
swell pressures measured. Laboratory tests were also conducted
or the same soil using standard odometers and the swell-reload
method to measure "apparent" vertical swell pressures. Results
indicated lateral swell pressures measured in the field were 0.9
to 1.17 times the "apparent" vertical swell pressures measured
in the laboratory. The authors indirectly concluded that the
horizontal and vertical swell pressures were essentially the
same. It is interesting to note that the lateral swell pressures
measured ranged from 1,300 to 2,150 psf, again substantially
higher than at-rest pressures typically assumed. Results of
lateral swell pressure development with time were inconclusive

as one test pit which presumably was not fully saturated
indicated that lateral swell pressures decreased with time while
the other test pit which was determined to be fully saturated
indicated lateral swell pressure generally increased to a maximum
value with time.
Laboratory tests results presented in an unpublished Term
Project by Walter Schultz at the University of Colorado at Denver
in 1978 [45] indicated CTsl/^sv as measured under supposedly
constant volume conditions in a standard triaxial cell ranged
from approximately 0.4 to 0.8. However, equipment problems and
limited data render this data as low quality. Nevertheless, the
results are not altogether different from those obtained by other
researchers as noted previously.
In 1980, Joshi and Katti [21] expounded on previous
studies regarding lateral swell pressures conducted in 1969 by
Katti, Kulkarni, and Fotedar [24] and Katti, Lai, Fotedar and
Kulkarni [25]. The 1969 studies presented results from swell
tests made in large scale wall tests and in modified triaxial
cell tests. In the 1980 studies a 1' x 1' x 1.5' deep square
steel container was constructed and set up as a typical odometer
press often used in consulting laboratories in the United States.
An opening on one face was made in the bottom of the container
and an adjustable proving ring was attached to measure lateral
swell pressures. By adjusting the proving ring, conditions of
zero lateral movement could be maintained. Vertical swell

pressures were measured using the variable surcharge-swell
method. Results of the tests indicated several trends:
1) Lateral swell pressure, <7SL, increased as the applied
vertical surcharge pressure, (7V, increased.
2) Lateral swell pressure increased rapidly upon wetting
to a peak value then leveled off to an equilibrium
value less than the peak. The ratio (aSL)peak/(CTSL)
equilibrium was approximately 1.22.
3) Time required to reach equilibrium lateral swell
pressure increased from 15 20 days for a vertical
surcharge of 1,000- psf to 45 60 days for an 8,000
psf vertical surcharge.
4) The ratio cjsl/ctv ranged from 10 for CTV = 100 psf to
1.19 for ov = 10,000 psf. Note: the vertical swell
pressure for zero volume change, CTSV, could not be
determined from the data presented, but the authors
indicate a "swell pressure" for the soil on the order
of 5,000 to 6,000 psf.
It appears that the authors believe that the swell
pressure for an expansive soil is unique and only varies
laterally and vertically as a consequence of boundary and
existing stress conditions.
In 1980 Didier, Kastner and Bourdeau [11] described a new
cell developed capable of rapidly measuring hydraulic
conductivity of feebly permeable swelling soils. The cell can

also be utilized as a swell/consolidometer capable of measuring
vertical swelling pressure under conditions of constant volume
or controlled vertical strain. Although not capable of measuring
lateral swelling pressures two important applications were
utilized in the equipment.
Pore pressure and/or swelling pressure measurements were
made using a displacement transducer to measure very small
deflections, of a thin metal plate acting as a sensitive
diaphragm. This is the basic principle behind pressure
transducers. By using an appropriate pressure transducer swell
pressures developed during swelling can be measured
simultaneously while maintaining almost zero volume change. The
second important feature of this equipment is the use of thin
stainless steel porous discs. The use of these porous discs
instead of porous stones would surely reduce the amount of
compression obtained within the apparatus.
Using thin-walled odometer rings instrumented with strain
gauges (lateral swell pressure (LSP) rings) very similar to that
developed by Komornik and Zeitlen in 1965 [26] Ofer [35] in 1980
conducted additional lateral swell pressure tests while
developing an apparatus capable of measuring lateral swell
pressure in situ.
Using the variable surcharge-swell method and allowing
compacted specimens to swell freely under a surcharge of 397 psf,
lateral swell pressures were measured for essentially zero

lateral strain. Results of the tests were consistent with those
of the other researchers noted previously in that as the initial
dry density of the specimen increased so did the lateral swelling
pressure, aSL. In fact lateral swell pressures ranged from 940
to 3655 psf for dry densities of 72 and 97 pcf respectively. The
clay used for testing was a montmorillonite clay.
The in situ swelling pressure probe (ISP Probe) developed
by Ofer consists of a hollow cylinder with a cutting edge
instrumented with strain gauges in a thin walled section of the
cylinder which is located between wetting rings. The apparatus
is placed in a slightly undersized pre-drilled hole to obtain
good contact between the surrounding soil and the thin walled
portion of the probe where the strain gauges are attached. Water
is introduced to the soil surrounding the probe through the
wetting rings. An electrical cable from the probe is connected
to a digital strain indicator at the ground surface. As the soil
swells, small strains in the thin walled portion of the probe are
recorded and converted to swell pressure measurements using a
standard calibration.
Laboratory testing of the ISP Probe using the same
compacted soil as used in the lateral swell pressure rings
yielded lateral swell pressure values on the order of 12,000 psf
for an initial dry density of approximately 90 pcf. These
results are considerably higher than those obtained using LSP
rings. However, it would appear this may be due to the method

of testing the probe. Soil was compacted in a 15.7 inch diameter
steel pipe and the 3.54 inch diameter probe placed in a 3.46 inch
diameter hole pre-drilled through the compacted soil. The
relatively small diameter of the surrounding pipe may have
contributed a much greater degree of confinement and side
friction than did the swell pressure rings, thus not enabling the
same degree of vertical swelling to occur. Regardless of these
potential problems, however, results indicate that under a wide
range of conditions, lateral swell pressures can be extremely
high. One other interesting result of the ISP probe tests was
that after seven days the measured lateral swell pressures
reached a peak value, but did not decrease as did those in the
tests by Joshi and Katti [21].
Some additional results of lateral swell pressure testing
is presented by Shiming [47] 1984. Similar to Parcher and Liu
[39] and, Katti, et. al. [24,25], Shiming modified a triaxial
cell to separately measure lateral and vertical expansion by
making the top loading cap/ram the same diameter as the soil
specimen. Lateral expansion is obtained by measuring the volume
of water forced out of the triaxial cell as the sample expands.
Lateral swell pressure is measured by not allowing water to be
forced out of the cell and measuring the pressure required to
accomplish this. A standard pressure gauge and burette system
were utilized for this purpose. Axial deformation was measured
using a typical dial gauge attached to the loading cap/ram.

Vertical load was applied by hanging weights to a load frame
suspended by the loading cap/ram.
Unfortunately, details of the procedures used to
determine swell pressures in Shiming's study of monoaxial versus
triaxial swelling pressure were not provided, making it
impossible to fully compare results with those of other
investigators. Swell pressure results which were presented by
Shiming were for undisturbed specimens. In general, the test
results for "identical" undisturbed specimens in the triaxial
cell indicated that vertical swelling pressure for triaxial tests
were less than or equal to the vertical swell pressure measured
in standard one-dimensional odometer tests. No reasons for this
difference were given, nor were the details of the test methods.
Results of the triaxial swell pressure tests indicated the
lateral swell pressure, aSL, to be less than the vertical swell
pressure, asv, with the ratio of oSL/asv ranging from 0.45 to 1.00.
However, considerable scatter of the data is evident.
In addition, it was observed that by plotting results of
swelling with time, primary swelling for the 2 1/2 high by 2 3/4
inch diameter triaxial specimens was achieved in 10 to 48 hours.
However, no details of the hydraulic gradient used to saturate
the specimens were presented.
In 1987, Johnson and the Army Corps of Engineers [19]
conducted studies to determine the feasibility of measuring
lateral and vertical swell pressures using a triaxial apparatus

which was also capable of conducting consolidation and shear
strength tests on the same specimen.
In that study, a standard triaxial cell was modified to
have a double walled cell chamber, pneumatically applied vertical
and horizonal pressure, pressure transducer and gauges to
measured applied pneumatic pressure and a linear variable
differential transformer (LVDT) to measure axial strain. Lateral
swelling was controlled by applying pressure to a burette filled
with water connected to the cell fluid (same principle as
Bishop's null pressure device). Calibration testing of a single
walled cell found that considerable volume change within the cell
occurred as lateral pressures increased. To maintain the
specimen at constant volume during swelling, considerable effort
and constant attention was necessary to continuously increase the
confining cell pressure in the fluid to account for volume
changes in the cell. By creating the double walled cell,
identical pressures were applied to opposite sides of the inner
cell wall, thus substantially reducing the volume change
occurring within the inner cell.
A disadvantage to the apparatus designed was that the
lateral and vertical swell pressures could not be completely
separated because the loading ram was not the same diameter as
the specimen. Hence, change in the horizontal confining pressure
also changed the vertical pressure applied to the specimen. In
essence, the apparatus tested could not measure applied lateral

pressures in excess of vertical pressures.
Tests were performed on undisturbed samples. X-ray
diffraction tests confirmed soil anisotropy and hence it was
hypothesized that vertical swell pressures would exceed lateral
swell pressures. However, test results for constant volume
conditions indicated that lateral swell pressures, CTSL, were
equal to vertical swell pressures, CTSV. The test results
presented, however, are somewhat confusing. For example, in one
test, lateral consolidation occurred at approximately 23 psi, but
no vertical movement occurred. The isotropic pressure was then
reduced to 21 psi. Lateral swelling occurred initially but
consolidation occurred approximately 13 to 14 hours later. Upon
reducing the isotropic pressure to 21 psi, vertical swelling also
occurred immediately but stopped. Approximately 13 to 14 hours
later vertical consolidation occurred. It is not clear why
delayed vertical consolidation occurred at 21 psi when no
consolidation occurred at 23 psi. It is suspected that the
vertical swell pressure was reduced as the vertical deflection
increased by a total of 0.002 inches between wetting and reducing
the isotropic pressure to 21 psi.
Based upon the Corps' test results, triaxial measurement
of lateral and vertical swell pressures appears feasible, but
measurements are extremely sensitive to deformation allowed.
However, a difference of say 2 psi (288 psf) in swell pressure
measurement is very small when considering swell pressures on the

order of 6,000 to 30,000 psf (i.e., 1 to 5 percent of total
value). On the other hand, 300 psf could be quite significant
when designing a retaining wall.
In his 1988 book, Chen [10] describes results of a large
scale laboratory test patterned after those by Katti, et al [25]
and Joshi and Katti [21]. In the tests a rigid rectangular box
approximately 15 inches by 15 inches in plan dimension and 12
inches in height was instrumented with load cells. The load
cells were placed on the surface of compacted soil placed in the
box and at three discrete depths along one wall of the box. The
load cells were made to be adjustable so that the compacted soil
was maintained at constant volume.
Conventional constant volume tests conducted in an
odometer indicated a vertical swell pressure, (Jsv, equal to 2189
psf. In the large scale model the lateral swelling pressures at
all depths were initially greater than the vertical swelling
pressure. With time the lateral pressures reached a peak value
then decreased to an equilibrium value equal to or slightly
greater than the constant volume vertical swell pressure
obtained. The maximum ratio of Osl/Osv was approximately 3.6.
In one test an equilibrium lateral swell pressure equal to the
vertical swell pressure was reached at 1728 psf. In a second
test, a constant equilibrium lateral swell pressure was not
obtained after 90 days, but lateral swell pressures were trending
towards the equilibrium vertical swell pressure.

Peak lateral swell pressures on the order of 4896 psf
were obtained. The peak lateral swell pressure occurred
somewhere between 20 hours and 7 days, after wetting was begun,
a considerable range. Based upon Chen's results, it appears that
there is a limiting depth below which In this test it occurred between 35 to 50 percent of the wall
height. This is consistent with tests by Katti, et. al. [25],
4.3 Summary
As can be seen from the information summarized in Table
2 below, there is relatively limited data concerning the
relationship between lateral and vertical swelling pressures.
Many of the test results presented are not directly comparable
because of the different, test methods and procedures used to
obtain them and/or the lack of test method descriptions. In
general the existing data suggests the following trends:
1) As the initial dry density of an expansive soil
increases, the lateral swelling pressure as measured
under constant volume conditions also increases.
There is evidence that void ratio, e, versus log of
true lateral swell pressure, aSL, is a linear
2) True lateral swell pressure increases as the vertical
restraint increases (vertical strain is restricted) .

Comparison of Lateral and Vertical Swell Pressures
Measured By Previous Researchers
Invest i rators Test Eauinment Samples Tested Remolded or Undisturbed Lateral Swell Pressure Measured __ ... <7c> (psf) At Constant Volume CTr, /d, Comments
Parrher N Liu (1905) Mollified Triaxinl Cell Remo 1 Vert leal Strati
Komornik 6 Zeitlen (1965) Thin walled Oedometer Rings w/Strain Gauges Remolded 1330 - 5320 o QO o Limited Testing
Komornik 6. Zeitlen (1965) Same as 1965 Remolded 818 - 6911 0.76 1.65 Extensive Testing
Komornik 6 Livneh (1969) Same as 1965 Remolded 3683 - 6261 0.85 Samples of Compacted Soil Tested Parallel and Normal to Direction of Compaction
Katti, Kulkarni & Fotedar(1969) 1. Modified Triaxial Cell 2. Large Scale Box Remolded 1200 - 6660 1.0 1.67
Katti, Lai, Fotedar, Kulkarni (1969) 1. Modified Triaxial Cell 2. Standard Oedometer Remolded 6000 psf 1.0 1.67
Ranganathau 6> Pandian (1975) Modified Oedometer Cell (Psuedo Triaxial Cell) No Tests -
Robertson 6 Wagener (1975) In Situ Earth Pressure Cells Remolded 1316 - 2150 0.9 1.17
Joshi & Kntti (1980) Large Scale Box (2 Types) Remolded 1000 - 12000 Could Not Determine si/av 1.19 to 10
Ofer (1980) 1. Thin walled Oedometer Rings w/Strain Gauges 2. In Situ Probe w/Thin walled Section Instru- mented w/Strain Gauges Remolded 960 3655 Ring 7 12008 Probe aSL/av For Ring 2.37 9.2 aSL/ov For Probe 7 33
Shiming (1986) 1. Modified Triaxial Cell 2. Standard Oedometer Undisturbed Remolded 260 - 1600 0.65 1.00
Johnson (1987) Modified Triaxial Cell Undisturbed 3000 1.00
Chen (1988) Large Scale Box Remolded 1728 - 6896 1.0 3.6
(Si.mi.lar to Katti,
and Josh! & Kntti)

3) The ratio of lateral swelling pressure, CTSL, to
vertical swelling pressure, (Jsv, typically ranges from
0.7 to 1.5 for remolded specimens for all methods of
pressure determination reported.
4) There is not sufficient data to determine if the
lateral swell pressure increases to a peak value and
remains at that level or whether it decreases to an
equilibrium value with time.
5) The lateral swell pressure, aSL, can exceed applied
vertical stress (overburden stress), CTV, and thereby
become the major principal stress.
6) The lateral swell pressure measured in tests on
remolded specimens ranged from approximately 800 psf
to as high as 12,000 psf. Typical low end values
averaged approximately 1,500 psf. These lateral
pressures are substantial and should be considered in
design. For comparative purposes, an example
calculation of wall pressures is presented below for
a ten foot high concrete basement wall.
Assume the wall is sufficiently restrained by top and
bottom structural floor systems to prevent rotation so that at-
rest pressures would prevail. Use data presented by Komornik and
Zeitlen in 1970 [29]. In accordance with ASTM Test designation
D1557 the maximum dry density of the backfill soil is 99.8 pcf
with an optimum moisture content of 24 percent. Assume the

backfill is compacted to a relative compaction of 87 percent at
a moisture content of 25 percent, one percent above optimum. The
total unit weight of the backfill in this state is approximately
109 pcf. By using results of lateral swell pressure ring tests
for these soil conditions, the data presented in Table 3 and
Figure 16, is obtained.
As can be seen from the above example the effect of high
lateral swelling pressures can be substantial when designing
basement or retaining walls. It is also easy to see that the
high lateral pressure against grade beams and drilled shafts in
expansive soils can also be quite large. This has at least two
potential effects. The increased normal force against the wall
or shaft increases the total shear force that can be developed
along the wall or shaft; i.e. F = N tan <5, where F is the shear
force along the wall or shaft, N is the lateral swelling force
normal to the wall/shaft and S is the angle of adhesion between
the wall/shaft and soil. Hence, as N increases, F increases
proportionally, and uplift forces can become quite large.
The second effect concerns drilled shafts. As pointed
out by Amir and Sokolov [3] swelling around drilled shafts is
typically not uniform at least initially. The development of
large lateral pressures on one side of the pier has the potential
of creating large bending moments in the shaft which are not
commonly designed for, at least not in the Denver-Metro area.

Comparison of Potential Lateral Swell Pressure
With Typical At-Rest Lateral Pressures
* Calculated
At-Rest Lateral
Calculated Corresponding (7H=K0CTV
Depth (ft) v (psf) CJSL Measured (psf) Ko500 Kc=.741 sl/h Average
1 109 1760 55 81 25.9
2 218 2067 109 162 15.3
3 321 2271 161 238 11.4
4 436 2456 218 323 9.1
5 545 2558 273 404 7.6
6 654 2701 327 485 6.7
7 763 2742 382 565 5.8
8 872 2824 436 646 5.2
9 981 2885 491 727 4.7
10 1090 2988 545 808 4.4
* K0 = 1 -Sin0
H = KoaV
ffy = so oH (1-Sin0) ( for 0 15 K0 0.741
0 30 K0 0.500

Figure 16:
Comparison of Potential Lateral Swell Pressures with
Typical at-Rest Lateral Pressures

These are only two examples of the need for considering
lateral swelling pressures in design. A rational method of
incorporating lateral swelling pressures into design is also
needed. Before this can be accomplished, however, a standard
method of defining and estimating lateral swell pressure is
necessary. Hence, the purpose of this study is to assimilate
existing data and experience in hopes of designing an improved
apparatus that can be used to measure lateral swell pressure both
for research and eventually as a standard testing tool in private

5.1 General
Phase I of this research consisted of conducting swell
pressure tests using three different types of standard laboratory
equipment. The purpose of this was to gain experience with the
advantages and disadvantages of the various test procedures and
the associated equipment. Using this experience Phase II of the
research consisted of designing an apparatus to meet the
objectives of the research. If time permitted, Phase III of this
research was to manufacture the prototype equipment and confirm
it to be fully functional by conducting several swell tests and
comparing the results to those obtained in Phase I, and then
convert the test equipment to a fully automated system.
5.2 Soil Properties
In order for test results to be comparable, the soil used
for Phase I testing was obtained from one specific location and
depth. The soil selected consisted of a claystone bedrock
material (i.e. a clay shale) obtained from within the Roxborough
Village Subdivision located just west of Rampart Range Road, and
south of Titan Road, in Douglas County, Colorado. Refer to the
Vicinity Map in Figure 17. This site is at the foot of the
eastern slope of the geological uplift known as the Dakota

Figure 17: Site Plan Location Where Soil Obtained for
Laboratory Testing

Test borings drilled in the immediate area generally
indicate the subsoils consist of approximately zero to 9.0 feet
of compacted, sandy to very sandy clay fill overlying zero to
22.0 feet of stiff to very stiff, sandy to very sandy clay and
highly weathered claystone bedrock. Firm to very hard claystone,
and claystone-sandstone bedrock was encountered at depths of zero
to 31.0 feet. A typical subsurface profile of the area is shown
in Figure 18.
The specific area selected for obtaining the soil is
located in the southeastern corner of the Subdivision. The soil
profile in this specific area consists of approximately 2.0 feet
of compacted fill consisting of sandy to very sandy clay, over
1.5 feet of stiff, native sandy to very sandy clay. Hard
claystone bedrock is at a depth of about 3.5 feet.
Approximately 200 pounds of olive and gray colored
claystone was collected from the bottom 2 feet of an 8 foot deep
excavation. Results of laboratory testing indicate in-situ
moisture contents of the claystone in the general area range from
approximately 12.8 to 18.6 percent, with in-situ dry densities
ranging from 104.3 to 119.4 pounds per cubic foot. Gypsum
crystals were observed in many of the fractures in the bedrock.
Several index property tests were conducted on the
claystone to further classify and characterize the material.
Moisture density relationships were also obtained in accordance
with ASTM test procedures D698 and D1557 using representative

-#4=100% -#200=88%
Pl=49 -#4=100% -#200=98%
Figure 18: Simplified Subsurface Profile of Area Where Soil Obtained for Laboratory Testing

portions of the processed material. Results of these tests are
summarized in Table 4 and Figures 19 and 20.
As seen in Figure 20, available in-situ dry density
values ranged from approximately 2 pcf lower to 16 pcf higher
than the maximum dry density obtained for the D1557 moisture-
density curve. The average in-situ dry density is approximately
3 pcf higher than the curve value. To insure a highly swelling
condition, a target value of 117.0 pounds per cubic foot dry
density at 10.0 percent moisture content was selected for use in
the remolded laboratory swell tests.
5.3 Sample Preparation
5.3.1 General
The claystone was prepared for remolding by first
allowing it to air-dry for several days. It was then
mechanically processed using a small rock crusher set to an
opening of approximately 1/4 inch. After mechanically processing
the air-dried material it was screened through a No. 4 sieve, and
thoroughly mixed in a large, clean steel trough using a shovel.
As previously mentioned, the target dry density and
moisture content for the remolded soil test specimens was 117.0
pcf and 10.0 percent, respectively. Approximately five pounds
of processed soil was prepared at a time. An initial moisture
content was obtained and additional water added to the soil to

Table 4
Results of Index Property Tests
For Soil Used In Laboratory Swell Tests
Test Results
Maximum Dry Density (ASTM D698) 98.5 pcf
(ASTM D1557) 112.0 pcf
Optimum Moisture Content (ASTM D698) 22.8 %
(ASTM D1557) 15.5 %
Liquid Limit (LL) 68
Plasticity Index (PI) 49
Specific Gravity (G ) 2.76
Percent Passing No. 8 Sieve 100
Percent Passing No. 200 Sieve 98
Percent Finer Than 5/x 73
Percent Finer Than 2/i 60
USCS Designation CH
AASHTO Designation A-7-6 (54)
Activity (PI/% 2/x) 0.82

6 5 4 SIN. I.5IN. 3/4 IN. 3/6 IN. 4 10 20 40 60 100 200

> 0
Ui so

CH CLAYSTONE, silty 68 19 k9 2.76
olive to gray

Figure 19: Typical Gradation Curve for Soil Used in Laboratory Swell Tests

Dry Density, (pcf)
Moisture Content, (%)
Figure 20: Laboratory Moisture Density Curves for Soil Used in
Laboratory Swell Tests

bring it to an initial moisture content of 10.0 percent. Water
was added in small amounts using a spray bottle and the soil was
thoroughly mixed by hand after each incremental addition of
water. When the addition of water was complete the soil was
placed in a double plastic bag and allowed to cure for at least
36 hours.
When preparing the five pounds of soil used for Test Nos.
1A and 3, it was observed that upon sample molding the existing
moisture content was only about 8 percent after curing for
approximately 48 hours. Hence, water was added to bring the soil
to 10 percent moisture but no time for curing was allowed at this
5.3.2 Standard Pedometer Cell
Specimens for the Bishop Consolidometer were prepared by
carefully centering the 2.5 inch diameter steel confining ring
in the bottom of a standard Proctor mold as shown in Figures 21
through 24, then placing 470.6 grams of the pre-moistened soil
into the mold and gently leveling the surface. This soil was
then compacted using 60 blows of a 10 lb. Proctor hammer, dropped
18 inches. The hammer used is shown in Figure 25. The soil and
ring were then removed from the mold and the soil around the ring
trimmed flush to the top and bottom of the ring using a thin
bladed knife and straight edge. Shavings from the trimming
operations were saved, weighed and dried for determination of