Vanapalli, S. K„ Fredlund, D. G. & Pufahl, D. E. (1999). Geotechnique 49, No.

2, 143-159

The influence of soil structure and stress history on the soil-water

characteristics of a compacted till
S. K. VANAPALLI,* D. G. FREDLUND* AND D. E. PUFAHL*

The soil-water characteristic defines the rela¬ II existe plusieurs relations theoriques et empiri-
tionship between the soil suction and gravi¬ ques permettant de faire une maquette des
metric water content, w>, or the volumetric water proprietes d’un sol non sature en utilisant la
content, 0, or the degree of saturation, S. Theor¬ courbe caracteristique sol-eau et les proprietes
etical and empirical relationships are available conventionnelles des sols satures. Ces procedures
to model the unsaturated soil properties such sont interessantes pour les praticiens industriels
as the coefficient of permeability and shear car il est difficile, long et couteux de realiser des
strength using the soil-water characteristic and tests rigoureux en laboratoire sur sols non sa¬
the conventional saturated soil properties. These tures. Divers investigateurs utilisent normale-
procedures are attractive to engineering practi¬ ment une courbe caracteristique sol-eau sur une
tioners because rigorous laboratory tests on un¬ gamme de succions limitee, allant habituelle-
saturated soil are difficult and time-consuming ment de 0 a environ 1500 kPa, pour faire une
and, therefore, costly. Various investigators gen¬ maquette du comportement du sol non sature.
erally have used the soil-water characteristic Les sols, cependant, passent d’un etat sature a
over a limited suction range (usually 0 to about un etat sec dans une gamme de succions allant
1500 kPa) to model the unsaturated soil beha¬ de 0 a 1 000 000 kPa. Nous justifions l’extension
viour. Soils, however, change from a saturated to de la courbe caracteristique sol-eau jusqu’a
a dry condition over a range of suctions from 1 000 000 kPa et nous donnons une methode
0 to 1 000 000 kPa. A rationale for extending permettant d’evaluer l’etat de saturation resi-
the soil-water characteristic up to 1 000 000 kPa duelle a partir de la courbe etendue. Plusieurs
is provided in this paper, and a method for facteurs exercent une influence sur le comporte¬
estimating the residual state of saturation is ment de la courbe caracteristique sol-eau. Pour
presented. Soil-water characteristics were devel¬ utiliser avec succes la courbe caracteristique sol-
oped for the entire range of suctions from 0 to eau afin de predire les proprietes du sol non
1 000 000 kPa on statically compacted clayey till sature, il est important de bien comprendre les
specimens prepared at three different initial facteurs qui influencent les prises de mesure. Il
water contents. The influence of initial water est egalement important de simuler autant que
content, soil structure and stress history, as it possible les conditions in situ qui sont suscepti-
relates to the soil-water characteristic, has been bles de se produire sur le terrain pendant les
studied and is reported in this paper. The initial essais en laboratoire. Des courbes caracteris-
water contents selected for this study represent tiques sol-eau ont ete developpees pour toute la
the dry of optimum, optimum and wet of opti¬ gamme de succions de 0 a 1 000 000 kPa sur des
mum conditions with corresponding densities specimens d’argile a moraines compactes de
determined from the standard AASHTO test. maniere statique avec trois teneurs initiates en
The results indicate that initial moulding water eau. Nous montrons dans cet expose l’influence
content has considerable influence on the result¬ de la structure du sol (et de l’agregation) et de
ing structure (and aggregation), which in turn l’etat de contrainte sur la courbe caracteristique
influences the soil-water characteristics. In the sol-eau. Les teneurs en eau initiates choisies
low suction range (i.e. 0-1500 kPa), macrostruc¬ pour cette etude represented les conditions
ture governs the soil-water characteristics for seches de la valeur optimum, la valeur optimum
specimens compacted with dry of optimum ini¬ et les conditions bumides de la valeur optimum,
tial water contents and they exhibit a steeper avec des densites correspondantes determinees
slope. However, for the specimens compacted at par le test AASHTO. Les resultats montrent que
wet of optimum, microstructure governs the la teneur en eau de moulage initiale a une
soil-water characteristic behaviour. The soil- influence considerable sur la structure du sol (et
water characteristic of the specimens compacted l’agregation) qui en resulted.
at dry of optimum are influenced by the stress
history; however, stress history appears to have
no significant influence on the soil-water char¬
acteristics of the specimens compacted with wet

143
144 VANAPALLI, FREDLUND AND PUFAHL

of optimum conditions. The soil-water charac¬

teristic behaviour of specimens compacted with
optimum initial water contents lies in between
those of specimens compacted with water con¬
tents that are dry and wet of optimum. It
appears that soil-water characteristics are not
significantly influenced either by the soil struc¬
ture (aggregation) or the stress history for the
high suction ranges (i.e. 20000-1 000000 kPa).

KEYWORDS: fabric/structure of soils; soil-water

characteristic; stress history; suction; unsaturated
soils.

INTRODUCTION viour may be primarily related to the volume of

A theoretical framework for unsaturated soil mech¬ the separate phases (e.g. water content), or the
anics, which closely parallels that of saturated soil continuity and tortuosity of the liquid phase (e.g.
mechanics, has been developed and firmly en¬ coefficient of permeability, molecular diffusion) or
trenched over the past 15 years. The constitutive the air phase (e.g. coefficient of vapour or oxygen
equations for volume change, shear strength, and diffusion). In other cases it is the nature of the
flow through unsaturated soil are receiving general interphase contact area controlling stress transfers
acceptance in geotechnical engineering applications (e.g. shear strength, volume change) or interphase
(Fredlund & Rahardjo, 1993). Because experimen¬ mass transfers (e.g. chemical adsorption, volatiliza¬
tal studies on unsaturated soils are time-consuming tion) which controls behaviour (Barbour, 1999).
and costly, relationships between the soil-water These interphase relationships can be derived using
characteristic and saturated soil properties are now the soil-water characteristic data and can then be
being developed to predict/model the engineering used for estimating unsaturated soil properties.
behaviour of unsaturated soils. The soil-water characteristic and the saturated
The soil-water characteristic defines the rela¬ coefficient of permeability have been used in pre¬
tionship between the soil suction and either the dicting the relationship between suction and the
gravimetric water content, w, or the volumetric coefficient of permeability (Brooks & Corey, 1964;
water content, 0, or the degree of saturation, S. van Genuchten, 1980; Mualem, 1986; Fredlund et
Geotechnical engineers are generally more conver¬ al., 1994). It has also been shown that the soil-
sant with the term degree of saturation, S, than water characteristic and the saturated shear strength
with volumetric water content, 6. Therefore, the parameters can be used to predict the variation in
soil-water characteristic relationship should have shear strength with respect to suction (Vanapalli et
greater meaning if it is presented using degree of al., 1996; Fredlund et al., 1996).
saturation versus suction. The soil-water character¬ This paper provides an explanation of the differ¬
istic can be described as a measure of the water¬ ent phase relationships from a saturated condition
holding capacity (i.e. storage capacity) of the soil to a dry condition (i.e. for a suction range of 0-
as the water content changes when subjected to 1 000 000 kPa) using the soil-water characteristic.
various values of suction. A construction procedure is also presented to de¬
The soil-water characteristic is a conceptual fine the residual state of saturation in an unsatu¬
and interpretative tool by which the behaviour of rated soil using the entire soil-water characteristic.
unsaturated soils can be understood. As the soil The distinguishing features of the soil-water
moves from a saturated state to drier conditions, characteristic depend on several factors such as soil
the distribution of the soil, water, and air phases structure (and aggregation), initial moulding water
changes as the stress state changes. The relation¬ content, void ratio, type of soil, texture, mineral¬
ships between these phases take on different forms ogy, stress history, and method of compaction. Of
and influence the engineering behaviour of unsatu¬ the factors stated above, the stress history and
rated soils. For example, in some cases the beha- initial moulding water content seemingly have the
most influence on the soil structure (and aggrega¬
tion), which in turn dominates the nature of the
Manuscript received 7 March 1997; revised manuscript
accepted 17 August 1997. soil-water characteristic for fine-grained soils. Spe¬
Discussion on this paper closes 2 July 1999; for further cimens of a particular soil, in spite of having the
details see p. ii. same texture and mineralogy, can exhibit different
* University of Saskatchewan. soil-water characteristics if they are prepared at
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 145
different initial moulding water contents and pos¬ is small (i.e. O-lOOOkPa). However, it is also
sess different stress histories. As a result, the common to plot the soil-water characteristic beha¬
engineering behaviours of the specimens will also viour on a semi-logarithmic plot if the suction
differ. range used for testing is large. While many engi¬
Soil-water characteristics are commonly devel¬ neering applications are concerned with the lower
oped in the laboratory using pressure plate equip¬ end of the suction range (i.e. less than 500 kPa),
ment that is not capable of applying a confining other applications such as soil covers and soil
pressure or significant vertical stresses to the speci¬ liners may require consideration of rather large
men, whereas in the field the soil usually has a values of suction (i.e. in excess of 3000 kPa).
complex stress history. To obtain a reliable soil- A typical soil-water characteristic is illustrated
water characteristic, the soil structure and stress in Fig. 1 for the entire range of suction values (i.e.
history, which may occur in the field, should be 0-1 000000 kPa). The key features of the soil-
reasonably simulated in the laboratory. In this water characteristic are explained using this figure.
paper, a method for developing the soil-water
characteristic for fine-grained soils under differing
stress histories is proposed. The influences of stress Air-entry value of the soil
history and initial moulding water content on the Conceptually, the air-entry value represents the
soil-water characteristics of statically compacted differential pressure between the air and water that
till specimens have been evaluated. This evaluation is required to cause desaturation of the largest
has been accomplished by loading and unloading pores (i.e. ‘air entry’). The air-entry value of the
specimens in a conventional oedometer and then soil is obtained by extending the constant slope
using a pressure plate apparatus and vacuum desic¬ portion of the soil-water characteristic to intersect
cators to determine the suction-water content rela¬ the suction axis at 100% saturation (Fig. 1). The
tionship from 0 to 1 000 000 kPa. The influence of corresponding value of suction is taken as the air-
initial moulding water content on the structure entry suction value of the soil. If the ordinate in
(and aggregation) is evaluated by comparing the Fig. 1 is plotted as volumetric water content, the
soil-water characteristic data with data obtained slope of the portion of the curve up to the air-entry
from individually compacted specimens. The ma¬ value is equivalent to the coefficient of volume
trix suction of the individually compacted speci¬ compressibility, mv.
mens was determined using the axis-translation
technique on a null pressure plate apparatus.
Identifiable stages
There are three identifiable stages of desatura¬
KEY FEATURES OF SOIL-WATER tion (Fig. 1): namely, the boundary effect stage, the
CHARACTERISTIC transition stage (i.e. primary and secondary) and
Soil-water characteristics are plotted on an ar¬ the residual stage of unsaturation (Vanapalli, 1994).
ithmetic scale if the suction range used for testing Figure 1 also illustrates the variation in the wetted

100
'3
80 Air-entry valui

Soil .Transition zone

c particleÿÿÿ"
-2 60-
5
3 Water Air
o 40 --
Boundary effect zonal Transition zone Residual zone of unsaturation
eO!
H 1
O Residual state of saturation
20

0
1 10 100 1000 10000 100000 1000000
Suction: kPa

Fig. 1. Typical soil-water characteristic showing zones of desaturation

146 VANAPALLI, FREDLUND AND PUFAHL

area of contact for the different stages of the soil- saturation can be defined as the point where the
water characteristic. In the boundary effect stage, line extending from 1 000 000 kPa along the curve
almost all of the soil pores are filled with water. intersects the previous tangent line (Fig. 1).
The soil desaturates at the air-entry value in the
transition stage. In this stage, the flow of water is
in the liquid phase as the applied suction increases High values of suction
and the soil dries rapidly with increasing suction. Models based on soil -water characteristics that
The connectivity of the water in the voids or pores were developed for a limited range of suction (i.e.
continues to reduce with increased values of suc¬ 0-1500 kPa) may not be suitable for the prediction
tion, and eventually large increases in suction lead of unsaturated flow properties at low water con¬
to relatively small changes in the degree of satur¬ tents and high suctions. For example, in the predic¬
ation. tion of performance of soil covers for waste
disposal sites, estimates of actual evaporation are
required. These predictions require that the soil-
Residual state of saturation water characteristic be defined at suctions exceed¬
The residual state of saturation can be consid¬ ing 3000 kPa (Wilson et al., 1994). While the
ered to be the degree of saturation at which the
liquid phase becomes discontinuous. Consequently,
—
matrix suction, («a uw), component largely gov¬
erns the engineering behaviour of unsaturated soils
the residual state of saturation represents the de¬ which are in excess of a metre or so below the
gree of saturation value beyond which it becomes ground surface, the surface phenomenon of eva¬
increasingly difficult to remove water from a speci¬ poration is controlled by total suction (Wilson et
men by drainage. The point at which residual state al, 1994). Total suction, \p, is composed of matrix
of saturation is reached is not always clearly
defined.
—
suction, (t/a «w), and osmotic suction, xpQ- There
appears to be a common value of total suction at
Traditionally, soil-water characteristic has been which all soils approach zero water content. This
defined over a range of suctions that is limited suction value corresponding to zero water content
from 0 to 1500 kPa. A suction value of 1500 kPa is approximately 1000 000 kPa (Croney et al.,
has taken on significance as ‘residual suction’ 1958; Russam, 1958; Fredlund, 1964; Vanapalli,
because it corresponds to the wilting point for 1994). These observations are also supported by
many plants (van Genuchten, 1980). However, this thermodynamic principles (Richards, 1965). Thus,
arbitrary value may not actually correspond to a it is useful and meaningful to use the soil-water
residual state of saturation condition. Some authors characteristic for the entire range of suction values
have shown suction to increase asymptotically to (i.e. 0-1 000 000 kPa) in the prediction of unsatu¬
infinity along the soil-water characteristic as the rated soil properties.
degree of saturation approaches the residual state The data for the soil-water characteristic are
(i.e. approximately a constant value) (Nitao & generally obtained from a pressure plate apparatus
Bear, 1996). at low to moderate suctions (0-1500 kPa) and an
It is often necessary to define the residual state osmotic desiccator for the higher suction range
of saturation condition in order to obtain the fitting (i.e. 3500-300 000 kPa or higher). Luckner et al.
parameters in numerical models for predicting per¬ (1991) expressed some concerns about using a
meability (Brooks & Corey, 1964; van Genuchten, soil-water characteristic defined over the entire
1980) or shear strength (Vanapalli et al., 1996; range of suctions based on pressure plate and
Fredlund et al., 1996). Such equations offer com¬ desiccator tests, since these techniques are based
putational advantages and are useful in developing on different modes of water movement (i.e. liquid
closed form solutions. Several empirical procedures flow versus vapour migration). However, if the
are available to define the residual state of satura¬ soil-water characteristic is viewed from a phenom¬
tion using the soil -water characteristic (Brooks & enological point of view, the total suction repre¬
Corey, 1964; White et al, 1970). These procedures sents the total energy deficiency in the water
typically do not use the soil-water characteristic phase. Whether equilibrium with the applied en¬
data for the entire range of suction (i.e. 0- ergy state is obtained by liquid flow or by equili¬
1 000 000 kPa). A graphical procedure can be used bration with the vapour phase is not of concern for
to define the residual state of saturation when the the definition of the soil-water characteristic.
entire suction range is used (Fig. 1). This proce¬
dure is similar to the Casagrande construction for
finding the point of 1 00% consolidation on a Typical soil-water characteristics for various soils
deflection versus log time relationship. The proce¬ Typical soil-water characteristics for various
dure involves first drawing a tangent line through soils over the entire range of suctions are shown in
the inflection point on the straight line portion of Fig. 2. The gradation and plasticity properties of
the soil-water characteristic. The residual state of these soils are provided in Table 1.
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 147
100 T
Regina Clay

80
S?
« Indian Head
Till
° 60
1
3
ss
°CD 40 - Sand
2
CJ)
<D
Q
20
Silt
7
0
10 100 1000 10000 100000 1000000
Suction: kPa

Fig. 2. Typical soil-water characteristic for four Canadian soils

Table 1. Summary of the soil properties

Soil type Sand: % Silt: % Clay: % coL: % a)p: % Reference
Sand 98 2 Unpublished data, University of
Saskatchewan Laboratory (1994)
Silt 52-5 37-5 10 22 16-6 Huang (1994)
Indian Head Till 28 42 30 35-5 16-8 Vanapalli (1994)
Regina Clay 8 41 51 75-5 24-9 Fredlund (1964)

These curves illustrate that the key features of limit and plastic limit and grading properties are
the soil-water characteristic (such as the air-entry given in Table 1. The clay fraction is predomi¬
value and the residual state of saturation) are well nantly a calcium montmorillonite. The AASHTO
defined for most of the soils. A coarse-grained soil standard compacted maximum density is 1-80
such as a gravel or sand has large interconnected Mg/m3 at an optimum water content of 16-3%.
pores and shows a tendency to change in degree of The relative density of the soil solids is 2-73.
saturation at a fast rate as values of suction in¬ The soil was air-dried for several days, pulver¬
crease. The rate of drying decreases with an in¬ ized using a rubber mallet and passed through a
crease of fines. The water storage capacity of a 2 mm sieve. A prescribed amount of distilled water
soil that corrresponds to a particular value of was sprayed on the air-dried soil in several layers
suction is higher for a soil with a higher percent¬ and left overnight in tightly covered plastic bags in
age of fines. The air-entry value is also higher for a humidity-controlled room. The soil was then
soils which have more fines. Similarly, the residual thoroughly hand-mixed. To help prevent the forma¬
state of saturation also increases with the increase tion of soil-water clods, the mixed soil was again
in fines. The construction procedure for defining passed through a 2 mm sieve. The mixed soil was
the residual state of saturation appears to be suita¬ placed in plastic bags and kept in a humidity-
ble for most soils except for very fine-grained controlled room for at least 48 h.
soils, such as expansive soils and slurried soils All samples for the testing programmes were
(e.g. Regina Clay), which desaturate continuously statically compacted to 100 mm in diameter and
without exhibiting a distinct break (Fig. 2). 21 mm in height. The samples were prepared in a
single layer using a constant volume mould ,to
obtain the required initial conditions of water con¬
SOIL AND TESTING PROGRAMME tent and density. A specimen 63-5 mm in diameter
A sandy clay till obtained from Indian Head, was cut from the 100 mm dia. sample using stain¬
Saskatchewan, Canada was used in this testing less steel sharpened consolidation rings. These
programme. This soil is classified as a CL accord¬ specimens were then used to obtain data for the
ing to the Unified Classification System. The liquid soil-water characteristic from a pressure plate
148 VANAPALLI, FREDLUND AND PUFAHL

apparatus. In a pressure plate apparatus, the pre¬ moved from the oedometers and placed in the
pared specimen sits on a high air-entry ceramic pressure plate apparatus. These specimens are re¬
disk in a sealed air pressure chamber. Water in a ferred to as specimens with 0 kPa equivalent pres¬
compartment beneath the disk is maintained at zero sure in the paper.
water pressure while an applied air pressure in¬
duces a matrix suction under which the specimen
is allowed to come to equilibrium. Saturated specimens with simulated stress history
A conventional pressure plate apparatus does not
allow specimens to be loaded externally during
Testing programme, section 1 (soil-water testing. Nevertheless, it remains of interest to
characteristics) assess the importance of applied stress and stress
The testing programme was performed in two history for the features of the soil-water character¬
sections. Tests on specimens in section 1 were used istic. Therefore, it was decided to use specimens
to obtain data for the soil-water characteristics. which had a stress history (i.e. the specimens had
Four specimens with different stress histories (i.e. a known equivalent pressure). The void ratio versus
with equivalent pressures 0, 25, 100 and 200 kPa) stress relationship (both in loading and unloading
and different ‘initial’ water contents were prepared conditions) was determined through conventional
for testing. The meaning of the term equivalent odeometer testing.
pressure is detailed in the next section. The procedure used for inducing a predeter¬
mined equivalent pressure is explained using Fig.
3. This gives the void ratio versus stress relation¬
Saturated specimens with 0 kPa equivalent pressure ship for a specimen with a water content equal to
Compacted specimens 63 5 mm in diameter and 16-3% (i.e. representing optimum water content
21 mm in height, having been prepared with the conditions). This compacted specimen was placed
required initial water content and density, were in an oedometer, saturated under constant volume
sandwiched between filter paper and porous stones conditions and then loaded to 200 kPa (point A).
in consolidation rings and were loaded to 3-5 kPa The specimen was then allowed to swell under a
in a conventional oedometer. These specimens nominal pressure of 3-5 kPa (point B). While the
were submerged in distilled water, allowing access specimen had experienced a maximum prestress
to drainage at top and bottom for about 36 h. The pressure of 200 kPa, it had a void ratio correspond¬
degree of saturation of these specimens was ing to 100 kPa on the initial compression branch
checked using waxed trial samples which were after swelling under the applied pressure of 3-5 kPa
weighted in air and water. The degrees of satura¬ (point C). The equivalent pressure for this speci¬
tion of specimens were greater than 99% for all men is equal to 100 kPa.
the samples. Volume changes were not measured Using a similar procedure, specimens of differ¬
for these specimens. These specimens were re- ing equivalent pressures as shown in Fig. 4 were

0-56
'Soil specimen loaded to 200 kPa (A) and allowed to swell
0-54 under a nominal load of 3-5 kPa (B) represents
equivalent pressure condition of 100 kPa loading (C)

0-50 -Nominal load = 3-5 kPa Cc = 0-100

0}
Cs = 0-017
§ 0-48 B
2
? 0-46
I A
0-44 Initial void ratio, e = 0-52
Initial water content, w= 16-3%
0-42 Dry density = 1 -80 Mg/m3

0-40

0-38 '1 Dw
1 10 100 1000
Stress: kPa

Fig. 3. Void ratio versus the applied stress for an initial void ratio of
0-52
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 149
1 I i
Dry of optimum Optimum Wet of optimum
yd = 1-73 Mg/m3 Xd = F8U Mg/m3 yd = 1-77 Mg/m3
w= 13% w= 16-3% w = 19-2%
1 i i
Equivalent pressure (kPa) Equivalent pressure (kPa) Equivalent pressure (kPa)

1. 0 kPa 1. OkPa 1. 0 kPa

2. 25 kPa (40 kPa) 2. 25 kPa (25 kPa) 2. 25 kPa (25 kPa)
3. lOOkPa (160 kPa) 3. 100 kPa (200 kPa) 3. lOOkPa (210 kPa)
4. 200 kPa (320 kPa) 4. 200 kPa (400 kPa) 4. 200 kPa (420 kPa)

Note: The values in parentheses are prestress pressures.

Fig. 4. Testing programme for soil-water characteristics (section 1)

prepared and used for testing. The prestress pres¬ The chamber above the soil specimens was
sures are also shown in parentheses. To use this pressurized to the desired value of suction. The
procedure, the compression index, Cc, and the water from the specimen discharged under the
swelling index, Cs, values have to be known. These influence of the suction. The volume of water that
values were measured from a separate laboratory was discharged under these conditions was esti¬
test programme for the three different initial water mated from the known initial mass of the speci¬
content conditions and are summarized in Table 2. men. Equilibrium was assumed when water no
longer discharged from the pressure plate. Six to
seven days were typically required for the speci¬
Saturated specimens with 25 kPa equivalent mens to reach equilibrium with the applied suc¬
pressure tions. At this point, the specimens were removed
The procedure described earlier would not be and weighed to determine their water content. This
. suitable for preparing optimum and wet of opti¬ procedure was repeated at each desired value of
mum initial water content specimens with an suction. Soil-water characteristics were determined
equivalent pressure of 25 kPa, because these speci¬ for suctions ranging from 0 to 1500 kPa using
mens are in a state of constant volume conditions pressure plate apparatus. More details of the test
at these pressures (see Fig. 3, for example). Hence, procedure are available in ASTM (1993).
the specimens were loaded to 25 kPa in stages and When the tests on each specimen were termi¬
were allowed to remain for a period of 36 h and nated at a suction of 1500 kPa, the specimen was
were used for testing. removed and approximately half of it was used as
a final check of the water content. A small sample
of the remainder of the specimen, about 4-5 g,
Pressure plate tests for all specimens was used in the osmotic desiccator apparatus to
The specimens prepared using conventional oed- determine total suction which is described later in
ometer testing were kept in their consolidation rings this section.
and set on the ceramic disk of the pressure plate. A The calculations of degree of saturation in this
mass of 500 g was placed on top of the specimens study were made with reference to the void ratio
to ensure that good contact was maintained between of the specimen at the equivalent pressure (i.e. no
the ceramic disk and the soil specimen. This applied allowance was made for subsequent volume
mass resulted in a vertical pressure of 1-6 kPa on change). The changes in void ratio with respect to
the specimen. It was impractical to apply pressures suction were calculated for a series of tests and
as large as 3-5 kPa because of space limitations in were not found to be significant for the sandy clay
the pressure plate apparatus. till used in this project (Vanapalli, 1994). Statically

Table 2. Compression and swelling indices values for different

initial conditions
Initial condition Cc Cs
Dry of optimum (w = yd = 1-73 Mg/m3)
13%, 016 0016
Optimum (w = 16-3%, yd = 1-80 Mg/m3) 0-10 0-017
Wet of optimum (w = 19-2%, yd = 1-77 Mg/m3) 0-08 0-010
150 VANAPALLI, FREDLUND AND PUFAHL

compacted specimens of this material are relatively content, and since the matrix component increases
stiff and resistant to volume change with respect to exponentially with decreasing water content, the
changes in suction. Results from Sridharan et al. difference between total suction and matrix suction
(1971) also support this behaviour. However, for should become increasingly small at high suctions.
soils of high plasticity such as expansive clays, the
variation of void ratio with respect to suction may
be significant. Testing programme, section 2 (individually
compacted specimens)
Section 2 comprises three sets of tests to deter¬
Osmotic desiccator tests mine the matrix suction of individually compacted
An osmotic desiccator was used to complete the specimens of 100 mm in diameter and 21 mm in
soil-water characteristic for ranges of suction height. These tests were used to evaluate the role
greater than 1500 kPa. Salt solutions in glass desic¬ that soil structure or aggregation plays in determin¬
cators control the relative humidity and vapour ing soil suction. The specimens were prepared
pressure in the specimen. In this study, five aqu¬ based on the water content and density associated
eous solutions were selected, producing a range of with the AASHTO standard compaction tests. The
suction values from 3500 to 300000kPa (CRC matrix suction for the individually compacted spe¬
Handbook of Chemistry and Physics, 1995). The cimens was determined with a null pressure plate
salts used in this study and their associated relative using the axis-translation technique (Hilf, 1956).
humidities and suction values at a temperature of Each set of individually compacted specimens was
23°C are summarized in Table 3. A maximum prepared at the same dry density (i.e. constant void
value of 300 000 kPa of total suction was consid¬ ratio with varying initial water contents). Three
ered adequate to define the soil -water characteris¬ such sets were prepared and the results of these
tic. tests are summarized in Table 4.
Each sub-specimen taken from the pressure plate Specimens were prepared with initial water con¬
tests was placed in one of the glass desiccators tents ranging from 12-5% to 19-2%. Matrix suction
containing a liquid of specific salt concentration measurements for specimens with water contents
corresponding to a known value of total suction. lower than 12-5% could not be made due to the
This testing programme was undertaken in a room limitations of the air-entry value of the porous
where the temperature was controlled. The mass stone in the null pressure plate (i.e. 500 kPa). It
was measured using a sensitive electronic balance was difficult to prepare the specimens with water
(i.e. 0-001 g). After the specimen had reached contents greater than 19-5% because the specimens
equilibrium with the atmosphere in the desiccator, were very soft and difficult to handle. In some
the water content was determined. In this way, a cases a film of water developed on the surface of
complete soil-water characteristic for the entire the specimen, suggesting that positive pore water
range of suctions was determined for the soil pressures might have been developing in the speci¬
specimen associated with an initial water content mens at these higher water contents. Therefore, it
and predetermined stress history. was decided to restrict the study of matrix suction
While the pressure plate apparatus measures measurements for individually compacted speci¬
matrix suction only, the osmotic desiccator meas¬ mens to a water content range of 12-5% to 19-2%.
ures total suction. Since the osmotic component of
suction should only be weakly dependent on water
RESULTS AND DISCUSSION
Properties influencing the desaturation
Table 3. Summary of salt solutions, humidities and characteristics of fine-grained soils
equivalent total suction used in osmotic desiccator Soil-water characteristics with different initial
tests water contents and densities are shown in Fig. 5
Salt Relative Equivalent total (equivalent pressure OkPa). The drying character¬
humidity: % suction: MPa istics with respect to soil suction vary with the
initial moulding water contents. Fine-grained soils,
Lithium chloride 1 1-3 297-6
LiCl.HjO such as this sandy clay till, typically have two
Magnesium chloride 32-9 151-7 levels of structure: a macro-level structure and a
MgCl2.6H20 micro-level structure. The soil microstructure is
Magnesium nitrate 53-4 85-6 described as the elementary particle associations
Mg(N03)2.6H20 within the soil, whereas the arrangement of the soil
Sodium chloride 75-7 38 aggregates is referred to as the macrostructure
NaCl (Mitchell, 1976). Typically, both the macro- and
Potassium sulphate 96-8 4-4 micro-levels of structure are present in natural and
K2S04 compacted clayey soils. The resulting macrostruc-
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 151
Table 4. Variation of degree of saturation (water content) versus matrix suction
for individually compacted specimens (section 2)
Matrix Dry density Dry density Dry density
suction: 1-73 Mg/m3 1-80 Mg/m3 1-77 Mg/m3
kPa Void ratio, e = 0-58 Void ratio, e = 0-52 Void ratio; e = 0-545
74 87-2 (18 5)
92 84-7 (18 0)
138 79-8 (16-9)
148 78-7 (7(5-7)
186 76-8 (16-3)
228 68-1 (14-5)
232 67-8 (14-5)
258 66-5 (141)
368 61-0 (12-9)
372 60-5 (121)
39 98-5 (18-76)
62 92-4 (17-60)
140 86-2 (16-41)
160 86-3 (16-44)
220 78-6 (14-97)
300 71-0 (13-52)
82 90-4 (18-04)
126 85-3 (17-02)
246 78-2 (15-61)
283 72-6 (14-49)
376 65-2 (13-01)

100 tj. T
X' - Wet of optimum
V- Void ratio, e = 0-545
Initial water content, tv = 19-2% -
S* 80 V:
Oi
o’
1
1
OT
o 40
a>
2
60
-- Optimum
e = 0-52
tv =16-3%
Dry of optimum
/

e = 0-6, w = 13%

Pressure plate tests

/
SC

*\ —
V
Desiccator
N
tests
—
O)
a> Sis
Q
20

0
1 10 100 1000 10000 100000 1 000000
Suction: kPa

Fig. 5. Soil-water characteristics for specimens compacted at different

initial water contents

ture of specimens prepared at different initial water within the clods. The relatively low suction values
contents is different in spite of their identical associated with removing water from the large
mineralogy, texture and method of preparation. pores are significantly different from the large
The resistance to water discharge (i.e. desatura¬ suctions required to remove water from the micror
tion) is relatively low in the dry of optimum speci¬ scopic pore spaces between soil particles within
mens in comparison to optimum and wet of the clods of clay. As a result, the macrostructure
optimum specimens (Fig. 5). The specimens with controls the initial desaturation of compacted
initial water content dry of optimum contain rela¬ clayey specimens with initial water contents, which
tively large pore spaces which are located between are dry of optimum.
the clods of soil as compared to the pore spaces The pore spaces in a clayey soil compacted at
152 VANAPALLI, FREDLUND AND PUFAHL

an initial water content wet of optimum are not saturation with respect to suction for specimens at
generally interconnected or are in an occluded optimum and wet of optimum with different
state. These soils are more homogeneous and have equivalent pressures. In spite of the different
a higher storage capacity due to their different equivalent pressures, the soil-water characteristics
structure. They have no visible interclod pores and of specimens with wet of optimum initial water
offer more resistance to desaturation under an content conditions appear to be the same (i.e. the
applied suction in comparison to those specimens soil-water characteristics appear to be independent
compacted dry of optimum. In contrast to the of the stress history; Fig. 8). The drying character¬
specimens compacted dry of optimum, the micro- istics are governed by the microstructure for wet of
structure in the specimens compacted wet of opti¬ optimum water content specimens for the entire
mum controls and resists the desaturation (drying) suction range. The soil-water characteristic beha¬
characteristics of the soil. Hence, the slope of the viour of the specimens compacted at optimum
soil-water characteristic is relatively flatter for the water content condition lies between those of spe¬
wet of optimum specimen in comparison to the dry cimens tested with dry and wet of optimum initial
of optimum initial water content specimen in the conditions (Fig. 7). The air-entry value increases
lower suction range where the desaturation was with both an increase in the initial moulding water
attained by the liquid-phase drainage (i.e. 0- content and the equivalent pressures for all the
1500 kPa). The boundary between the occluded specimens tested.
pore space and the open pore conditions occurs at The variation of air-entry pressure with the ini¬
water contents approximately equal to the optimum tial void ratio for all the specimens tested with
water content (Marshall, 1979), and, hence, the different initial water contents is shown in Fig. 9.
specimen prepared at optimum water content con¬ The figure shows that for any particular void ratio,
dition lies between these two. the air-entry values of the specimens tested dry of
The soil-water characteristics developed for the optimum are always lower than those of the speci¬
specimens compacted dry of optimum and with mens tested at optimum and wet of optimum.
equivalent pressures of 0, 25, 100 and 200 kPa are Thus, the soil structure (and aggregation), which is
shown in Fig. 6. It is apparent that the air-entry a function of the initial moulding water content,
value of the specimens increases with increasing governs the air-entry value. The air-entry values of
equivalent pressure. In general, beyond the air- specimens wet of optimum do not appear to be
entry value of suction, the specimens subjected to strongly dependent on the applied stress history.
higher equivalent pressures have higher degrees of
saturation. The macrostructure appears to dominate
the soil-water characteristic features of the speci¬ Soil-water characteristic behaviour in the high
mens prepared dry of optimum in spite of the suction range
increase in the equivalent pressures. For all the initial conditions of water content
Figures 7 and 8 show the variation of degree of (i.e. dry of optimum, optimum and wet of opti-

100

90
Equivalent pressure = 200 kPa
Initial void ratio, e = 0-438
<6 >£— (Air-entry value = 36 kPa)
80 -V
.1 \l
2 25 kPa, e = 0-59
I
<n
75
01
ID
Q
70

60
- (6 kPa) -

0 kPa, e = 0-60
(3-5 kPa)
X S'
50 T
100 kPa, e = 0-543 \
(15 kPa) \
40
1 10 100 1000 10000
Suction: kPa

Fig. 6. Soil-water characteristics for specimens compacted dry of

optimum water content
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 153

--
100 <t
Equivalent pressure = 200 kPa
V Initial void ratio, e = 0-444
90 0 kPa, e - 0*517 value = 100 kP;
(25 kPa) ,
s? /
c
. 80 \
25 kPa, e = 0-514 \
2 (35 kPa) o' 3v
3 70
<9 in
\
A
ffi v.° \
2 60 100 kPa, e = 0-474 A.
O)
(80 kPa) V VD
Q

50 <ir

40
1 10 100 1000 10000
Suction: kPa

Fig. 7. Soil-water characteristics for specimens compacted at optimum

water content

100
A Equivalent pressure = 200 kPa
V c.'\\ initial void ra,i°- e = 043
' Air-entry value = 100 kPa
90
0j /
c 0 kPa, e = 0-545 V\«
B (80 kPa)
2
3 80 A.
s 100 kPa, e = 0-47 V
•s<D (90 kPa) A.
2 8\
cn V ?
o
Q 70 £ *\
0
\V
V>
60
1 10 100 1000 10000
Suction: kPa

Fig. 8. Soil-water characteristics for specimens compacted wet of

optimum water content

mum) and stress history, the soil-water character¬ The soil-water characteristic (testing programme
istic behaviour appears to be similar at higher - section 1) and statically compacted individual
suctions (i.e. 20 000-300 000 kPa) (Figs 10-12). In specimens (testing programme - section 2)
other words, the inter-aggregate structure appears The data from section 2 of the testing pro¬
to be the same for all the specimens at these gramme (Table 4) are used to explain the influence
higher suctions. Presumably, the water films at of initial water content on the soil structure (and
these suctions are so thin that all the water is aggregation), and thereby on the suction in the soil.
within the range of influence of the osmotic and The soil-water characteristics (testing programme,
adsorptive fields. section 1) described in the previous section are
154 VANAPALLI, FREDLUND AND PUFAHL

120

Wot of optimum initial conditions

100

ra
\
80
\
2 \
3 \
£ 60 \
Q.

£ Optimum *
!
<
40
A
\
» \ Dry of optimum
20

0
0-4 0-45 0-5 0-55 0-6 0-65
Initial void ratio

Fig. 9. Air-entry value versus initial void ratio from the soil-water
characteristics

100 *
- -2 i.
V, Wet of optimum
80 e = 0-54
5? V.,
Optimum, e = 0-514 .V
O 60
V
1 Dry of optimum 7
5
15 Initial void ratio, & \
(/)
e = 0-59 s
o 40
0) \ V
Cl)
\ tests
o> Pressure plate tests
V
20

0
1 10 100 1000 10000 100000 1000000
Suction: kPa

Fig. 10. Soil-water characteristics for specimens with an equivalent

pressure of 25 kPa

compared with the matrix suction versus degree of Dry of optimum initial water content conditions
saturation results of individually compacted speci¬ Figure 13 shows the best-fit soil- water charac¬
mens. The relationship between the degree of teristic for a specimen with dry of optimum initial
saturation and matrix suction obtained from indivi¬ water content conditions with an equivalent pres¬
dually compacted specimens is different from the sure of 0 kPa. The degree of saturation versus
soil-water characteristic, which is the variation of matrix suction results of individually .compacted
degree of saturation with suction for a single speci¬ specimens are also shown. The initial void ratios
men. Moreover, the resulting soil structure (and for all specimens are the same. Matrix suction and
aggregation) of the individually compacted speci¬ corresponding degrees of saturation greater than
mens at various initial water contents will be dif¬ 75% for individually compacted specimens lie
ferent from that of specimens with a corresponding above the best-fit soil-water characteristic. These
water content on the soil-water characteristic. specimens with degrees of saturation greater than
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 155
100

8?
80 zx
Dry of optimum
e = 0-543
Wet of optimum
e = 0-470

<6 \
o 60 Optimum
I e = 0-474 \
V
§ 40
&
SS
20
1 t-
&
Pressure plate tests Desiccator tests
0
1 10 100 1000 10000 100000 1 000000
Suction: kPa
Fig. 11. Soil- water characteristics for specimens with an equivalent pressure of
100 kPa

100

80 Optimum -
' *. x\t-
*
X
Wet of optimum
e = 0-430 _
e = 0-444
8?
(ri \
'\ \
I2 60 Dry of optimum
e = 0-438 \
3 A
s '\
:\
S 40 ST
CD
d>
o
20

Pressure plate tests Desiccator tests

o “i 4
1 10 100 1000 10000 100000 1 000000
Suction: kPa

Fig. 12. Soil-water characteristics for specimens with an equivalent pressure of

200 kPa

75% have water contents above 16%. The resulting degree of saturation of the specimens used to
soil structure (and aggregation) at this water con¬ determine the soil-water characteristic. At these
tent and higher should be similar to that of the water contents of approximately 14-5%, the indivir
specimens at optimum and wet of optimum condi¬ dually compacted specimens and the specimen
tions. used for the soil-water characteristic can be con¬
The matrix suction values of the individually sidered to be ‘identical’ due to their ‘similar’ soil
compacted specimens with degrees of saturation structure (and aggregation).
less than 68% and corresponding water content of It is apparent that the behaviour of the individu¬
14-5% are comparable to the values of suction and ally compacted specimens at higher saturations is
156 VANAPALLI, FREDLUND AND PUFAHL
100
Individually compacted specimens
a = 0-58, water content varying
90 A

Initial water content iiTV

individually compacted specimens'
c
80- greater than 16% -
5
3
CO
w
t X a
A

O 70
0)
4
£
o> i \o
0
60 -
Initial water content in
individually compacted specimens
/H
lower than 14-5% Soil-water characteristic
D
(dry of optimum specimen, a = 0-58)
50 £
1 10 100 1000
Matrix suction: kPa

Fig. 13. Comparison of soil-water characteristics for specimens com¬

pacted dry of optimum water content and compacted at the same initial
void ratio

different from the soil-water characteristic beha¬ Optimum initial water content conditions
viour. In these cases, the initial water contents of Figure 14 shows the best-fit soil-water charac¬
the specimens are different, and the resulting soil teristic with optimum initial water content condi¬
structure (and aggregation) is different. Over the tions (void ratio equal to 0-52 and initial water
selected range of water contents (i.e. from 12-5% content of 16-3%) and an equivalent pressure of
to 19-2%), the degrees of saturation of the indivi¬ zero. The degree of saturation versus matrix suc¬
dually compacted specimens varied from 58-8% to tion of individually compacted specimens is also
90-3%. These specimens are different from one shown for comparison. The soil-water characteris¬
another because they have different soil structures tic lies below the results of individually compacted
(resulting from the interparticle aggregations) specimens in the region of 0-150 kPa matrix suc¬
based on their initial water content. tion. The initial water content in individually corn-

100? 0 a*- Individually compacted

D
specimens e = 0-52,
water content varying
a
s. 90
<6 Similar initial conditions and
s structure
!
15
80
to

0)

£ A
Ol
o® 70 Water content lower than 14%
7
Soil-water characteristic,
(results in a different structure) initial water content 16-3%
=
(optimum)
60 4
1 10 100 1000
Matrix suction: kPa

Fig. 14. Comparison of soil-water characteristics for specimens com¬

pacted at optimum water content and those compacted at the same initial
void ratio
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 157
pacted specimens for this range of matrix suction ment with the soil -water characteristic also in¬
is higher than 16-3% (i.e. degree of saturation creases.
greater than 86%). With the increased water con¬ Figure 16 shows the degree of saturation versus
tent in the compacted specimens, the soil aggrega¬ matrix suction for all of the individually compacted
tions differ and result in a different soil structure. specimens with different initial water contents for
As the water content in the individually compacted the three void ratios (i.e. dry densities) tested. The
specimens decreases, the values of matrix suction figure shows that the results fall within a narrow
fall below the soil-water characteristic. This dis¬ band. The same data, when plotted as gravimetric
crepancy is expected because the specimens with water content versus matrix suction, show a clearer
lower water contents exhibit a different soil struc¬ relationship (Fig. 17). Olson & Langfelder (1965)
ture based on their aggregation, which in turn and Krahn & Fredlund (1972) reported similar
depends on the water content. The resulting struc¬ observations for a wider range of void ratios (i.e.
ture (or aggregation) is similar to the soil structure dry densities). Thus, for the void ratios used in this
expected for specimens with dry of optimum con¬ study (i.e. 0-52-0-58), it is the initial water content
ditions. At a degree of saturation around 86% (i.e. which governs the matrix suction of individually
at 16-3% water content), both the individually compacted specimens, and not the initial void ratio.
compacted specimens and the specimen used for
the soil -water characteristic show similar matrix
suction values. This correspondence is possible as SUMMARY AND CONCLUSIONS
both the structure and density in the two specimens Several empirical and theoretical procedures are
are approximately the same and they behave as available in the literature to model unsaturated soil
‘identical’ specimens. properties using the soil-water characteristic. The
literature indicates that the range of suction from
saturation to the dry condition in a soil varies from
Wet of optimum initial water content conditions 0 to 1 000 000 kPa irrespective of the type of soil.
Figure 15 gives the comparisons of soil-water Two key features of the soil-water characteristic
characteristics obtained using wet of optimum are the air-entry value and the residual state of
initial conditions (i.e. initial water content equal saturation. Residual state of saturation is used as
to 19-2%) and an equivalent pressure of zero one of the parameters in several models. A graphi¬
with those of individually compacted specimens. cal procedure is proposed in this paper to define
The individually compacted specimen results fall the residual state of saturation using the entire
below the soil-water characteristic. The individu¬ soil-water characteristic. Models that use the entire
ally compacted specimens were tested with water soil-water characteristic have a definite advantage
contents in the range of 13-19-2%. It can be for predicting unsaturated soil properties over the
seen that, as initial water contents in the indivi¬ entire range of suction and water contents.
dually compacted specimen increase, the agree- The initial moulding water content has a consid-

100 ?

S.
Co
90 — Soil-water characteristic,—
initial water content = 19-2%
c (wet of optimum)

5
3 80
o
a>
0 A

70 /
Individual compacted specimens
e = 0-545, water content varying »

A
60
1 10 100 1000
Matrix suction: kPa

Fig. 15. Comparison of soil-water characteristics for specimens com¬

pacted wet of optimum initial water content and compacted at the same
initial void ratio
158 VANAPALLI, FREDLUND AND PUFAHL
100 •ÿX

yD
80 -- Optimum, w- 16.3%; Matrix suction = A X°‘*
150 kPa; Initial void ratio, s = 0-52
%
A*
O
60 -- Dry o( optimum, w= 13%; Matrix suction
! = 368 kPa; Initial void ratio, e - 0-58
i
•s 40 « Dry of optimum
x Optimum
o>
Wet of optimum
D
20
Wet of optimum, w= 19-2%
Matrix suction = 68 kPa
Initial void ratio, e = 0-545
0
1 10 100 1000
Matrix suction: kPa

Fig. 16. Degree of saturation versus matrix suction values for compacted
specimens

20

19

18
y\
*e 17 V
Io 16

|15 « e = 0-58 (dry of optimum)

x e = 0-52 (Optimum) 4\D
14
e = 0-545 (Wet of optimum)
13

12
1 10 100 1000
Matrix suction. kPa
Fig. 17. Gravimetric water content versus matrix suction for compacted
specimens

erable influence on the resulting structure (and ture. The air-entry value and the residual state of
aggregation) of fine-grained soils such as the sandy saturation of the soil increase with the equivalent
clay till used in this research programme. The soil- pressure for specimens with dry of optimum initial
water characteristic of fine-grained soils is depen¬ water content conditions.
dent mainly on the structure (and aggregation) and Microstructure apparently governs the soil-water
the stress history, rather than the initial void ratio. characteristic behaviour of specimens compacted
Macrostructure governs the soil-water character¬ wet of optimum. The soil-water characteristic
istic behaviour for specimens compacted with ini¬ behaviour is not significantly influenced by the
tial water contents dry of optimum, particularly in equivalent pressure or the stress history in this
the low range of suction values. These specimens study (i.e. 0-200 kPa). Specimens wet of optimum
exhibit a steeper soil-water characteristic when have higher air-entry values and higher values of
compared with specimens compacted at optimum residual state of saturation than those compacted at
and wet of optimum water contents. The dry of optimum or dry of optimum water contents.
optimum specimens act more like a coarse-grained The soil-water characteristics appear to be ap¬
soil because of their highly aggregated macrostruc- proximately the same over suction ranging from
 SOIL-WATER CHARACTERISTICS OF A COMPACTED TILL 159
20 000 to 1 000 000 kPa (desiccator tests) for speci¬ tic. Can. Geotech. J. 33, No. 3, 440-448.
mens tested with different initial water content Hilf, J. W. (1956). An investigation of pore water pressure
conditions. The soil structure (and aggregation) in compacted cohesive soils. US Department of the
appears to have no influence on the soil-water Interior, Bureau of Reclamation, Design and Con¬
struction Division, Denver, CO, PhD dissertation,
characteristic behaviour in this range of suctions. Tech. Memo. 654.
The matrix suctions of individually compacted Huang, S.-Y. (1994). Evaluation and laboratory measure¬
specimens determined from null pressure plates ment of the coefficient of permeability in deformable,
were dependent on the initial water content rather unsaturated soils. PhD thesis, University of Saskatch¬
than the initial void ratio. At ‘identical’ conditions ewan, Canada.
(i.e. at similar densities, water contents and stress Krahn, J. & Fredlund, D. G. (1972). On total, matric and
state conditions), the matrix suctions corresponding osmotic suction. J. Soil Sci. 114, No. 5, 339-348.
to a particular degree of saturation were observed Luckner, L., van Genuchten, M. Th. & Nielsen, D. R.
to be the same as those given by the soil-water (1991). Reply to Nimmo, J. R. Comment on the
treatment of residual water content in ‘A consistent
characteristic. The conclusions of this study should set of parametric models for the two-phase flow of
be valid for all fine-grained soils. immiscible fluids in the subsurface’. Water Resources
More fundamental research, however, is recom¬ Res. 27, No. 4, 661-662.
mended to fully understand the soil-water charac¬ Marshall, R. J. (1979). Analysis of fundamentals of
teristic, particularly because it relates to the devel¬ compacted soils. General Report, Main Session, IV
opment of methods for estimating the properties of Proc. Sixth Pan-American Conf. Soil Mech. Found.
unsaturated soils. Additional experimental work Engng, Lima, Peru, 3, 328-348.
will be required on various soils throughout the Mitchell, J. K. (1976). Fundamentals of soil behaviour.
world to add credence to the results presented New York: Wiley.
Mualem, Y. (1986). Hydraulic conductivity of unsaturated
here. soils: prediction and formulas. In Methods of soils
analysis. Part 1. Physical and mineralogical methods
2nd edn, Agronomy (ed. A. Klute), pp. 799-823.
REFERENCES Madison, Wis: American Society of Agronomy.
ASTM (1993). Test method for capillary-moisture relation¬ Nitao, J. J. and Bear, J. (1996). Potentials and their role in
ships for coarse- and medium-texture soils by porous- transport in porous media. Water Resources Res. 32,
plate apparatus. ASTM D2325-68 (1981). In Annual No. 2, 225-250.
book of ASTM standards, Vol. 4.08, pp. 301-303. Olson, R. E. and Langfelder, L. J. (1965). Pore-water
Philadelphia: American Society of Testing Materials. pressures in unsaturated soils. J. Soil Mech. Found.
Barbour, S. L. (1999). The soil-water characteristic curve Div. ASCE 91, SM4, 127-160.
- a historical perspective and application to the beha¬ Richards, B. G (1965). Measurement of the free energy of
viour of unsaturated soils. Colloquium address pre¬ soil moisture by the psychrometric technique using
sented to the Canadian Geotechnical Conference thermistors. In Moisture equilibria and moisture
1995, Vancouver, BC, Canada. Can. Geotech. J. 36 changes in soils beneath covered areas, A Symp. in
(in press). Print, pp. 39-46. Butterworths.
Brooks, R. H. & Corey, A. T. (1964). Hydraulic proper¬ Russam, K. (1958). An investigation into the soil moisture
ties of porous media. Colorado State University (Fort conditions under roads in Trinidad, B. W. I. Geotech¬
Collins), Hydrology Paper, No. 3. nique 8, No. 1, 57-71.
CRC Handbook of Chemistry and Physics (1995). Section Sridharan, A., Altachaeffl, A. G. & Diamond, S. (1971).
15 - Practical laboratory data, 75th edn, p. 25. Boca Pore size distribution studies. J. Soil Mech. Found.
Raton: CRC Press. Div. ASCE 97, SM5, 771-787.
Croney, D., Coleman, J. D. & Black, W. P. M. (1958). The van Genuchten, M. Th. (1980). A closed-form equation
movement and distribution of water in soil in relation predicting the hydraulic conductivity of unsaturated
to highway design and performance. Road Research soils. Soil Sci. Soc. Am. J. 44, 892-898.
Laboratory, Research Note 3209, pp. 1-14. Vanapalli, S. K. (1994). Simple test procedures and their
Fredlund, D. G. (1964). Comparison of soil suction and interpretation in evaluating the shear strength of an
one-dimensional consolidation characteristics of a unsaturated soil. PhD thesis, University of Saskatch¬
highly plastic clay. MSc thesis, Department of Civil ewan, Canada.
Engineering, University of Alberta, Edmonton, Alber¬ Vanapalli, S. K., Fredlund, D. G., Pufahl, D. E. & Clifton,
ta, Canada. A. W. (1996). Model for the prediction of shear
Fredlund, D. G. & Rahardjo, H. (1993). Soil mechanics strength with respect to soil suction. Can. Geotech. J.
for unsaturated soils. New York: Wiley. 33, No. 3, 379-392.
Fredlund, D. G., Xing, A. & Huang, S. (1994). Predicting White, N. F., Duke, H. R. Sunada, D. K. & Corey, A. T.
the permeability function for unsaturated soils using (1970). Physics of desaturation in porous materials.
the soil-water characteristic. Can. Geotech. J. 31, No. J. Irrigation Drainage Div. Am. Soc. Civ. Engrs 96,
3, 533-546. No. IR2, 165-191.
Fredlund, D. G., Xing, A., Fredlund M. D. & Barbour, Wilson, G. W., Fredlund, D. G. & Barbour, S. L. (1994).
S. L. (1996). The relationship of the unsaturated soil Coupled soil-atmosphere modelling for soil evapora¬
shear strength function to the soil-water characteris- tion. Can. Geotech. J. 31, 151-161.