See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/346631624

Role of the Soil-Water Characteristic Curve in Unsaturated Soil Mechanics

Conference Paper · August 2018

CITATIONS READS
6 110

1 author:

Delwyn D. Fredlund
University of Saskatchewan
610 PUBLICATIONS   23,880 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Slope Stability Research View project

Unsaturated soils View project

All content following this page was uploaded by Delwyn D. Fredlund on 04 December 2020.

The user has requested enhancement of the downloaded file.

Role of the Soil-Water Characteristic Curve in Unsaturated Soil Mechanics
Delwyn G. Fredlund, Golder Associates Ltd., Saskatoon, Saskatchewan, Canada
ABSTRACT: The soil-water characteristic curve, SWCC, has been referred to as the key to the
implementation of unsaturated soil mechanics into geotechnical engineering practice.
Methodologies proposed within soil physics are re-examined in this paper and applied for the
geotechnical engineering discipline. Each of the volume-mass variables needs to be taken into
consideration when estimating unsaturated soil property functions in geotechnical engineering.
The shrinkage curve can be used in conjunction with the (gravimetric) water content versus soil
suction relationship (i.e., w-SWCC) to provide a more complete understanding of unsaturated soil
behaviour. In this way it is possible to separate the effects of overall volume change from changes
in the degree of saturation during the drying of a soil. The estimation of various unsaturated soil
property functions, USPFs, requires the use of the saturated soil properties along with
mathematical algorithms related to one or more of the volume-mass SWCCs. This paper provides
a state-of-the-art synthesis related to the estimation of hydraulic USPFs for geotechnical
engineering applications.

1 INTRODUCTION needed to embrace a wide range of degrees of

saturation. The degree of saturation of the
Saturated soil mechanics defines the
soil ranged from saturation in the capillary
engineering description of a two-phase
zone to a discontinuous water phase in the
system comprised of solids particles and an
dry soil zone.
incompressible pore fluid (water). The Equilibrium
engineering application of saturated soil negative pore-
Degree of
water pressure
mechanics started around the 1930s Ground surface line saturation, S

(Terzaghi, 1943; Taylor, 1948). Saturated Cracked, fissured zone Dry zone
soil behavior was related to changes in the
effective stress state of the soil (i.e., the
difference between total stresses, x, y. z, Vadose or unsaturated zone Two-phase zone
Soil-water
characteristic
and pore-water pressure, uw). Saturated soil curve, SWCC

mechanics principles were applied to

commonly encountered geotechnical Capillary zone
Phreatic surface
engineering problems (e.g., seepage, volume
0% 100 %
change and shear strength related problems).
Success related to the application of saturated Figure 1: Definition of the vadose or unsaturated soil
soil mechanics was strongly related to the zone
ability to measure saturated soil property
constants which could be related to stress Figure 1 defines the subdivisions of the
state and used to describe various physical vadose (or unsaturated zone) above the
processes. phreatic surface. The zone above the phreatic
Unsaturated soil mechanics emerged later line has been referred to as the vadose zone
and involved the engineering description of a in various disciplines. The vadose zone is
multi-phase system that had two fluid phases defined as “that part of the earth between the
(i.e., water and air) as well as the unique land surface and the water table (i.e.,
behavior of the contractile skin (i.e., the air- atmospheric pressure)”, (United States
water interphase) and the soil solids. The Geological Survey). The definition for the
description of unsaturated soil behavior vadose zone is the same as that commonly

1
 used for the “unsaturated soil zone” in time. Ground surface water balance can be
unsaturated soil mechanics. This is an assessed based on weather station and soils
important definition to retain because of the information data. Published results
manner in which the soil-water characteristic associated with field case histories would
curve is measured in the laboratory and used suggest that, in general, the water balance
in engineering practice. Disciplines such as components near ground surface can be
hydrology define the unsaturated zone as adequately predicted for geotechnical
starting where air becomes present in the engineering purposes (Wilson et al., 1994;
pore-water (Freeze and Cherry, 1979); Tran et al., 2014; Fredlund et al., 2012).
however, this is not the definition commonly An unsaturated soil can have three distinct
used in unsaturated soil mechanics. zones of saturation. The wide potential
The ground surface is subjected to a variation in degree of saturation has led to the
moisture flux that is continually changing in need to define the soil properties in terms of
response to weather conditions. The ground nonlinear “unsaturated soil property
surface forms a new and complex type of functions, USPFs”. The soil properties are
boundary condition within soil mechanics. mathematical functions that make the
Figure 2 shows the components that combine engineering analysis (e.g., numerical
to give rise to net infiltration or percolation. modeling) the solution of one or more
The relative magnitudes of the upward nonlinear partial differential equations.
moisture flux (i.e., evaporation and evapo- Figure 3 shows what can be referred to as the
transpiration) and downward moisture flux three pillars of unsaturated soil mechanics;
(i.e., precipitation) perturb the equilibrium namely, i.) engineering protocols, ii.)
(negative) pore-water pressure profile giving laboratory testing procedures and iii.)
rise to a trumpet shape for soil suction numerical modeling techniques. Laboratory
variations over time. testing procedures need to either directly or
Evaporative fluxes Precipitation
indirectly provide information on the
physical soil properties while numerical
modeling techniques simulate physical
Rainfall and snowfall
Actual
evaporation Evapo-transpiration processes.
Bare ground Vegetative surface
Runoff
Consulting
Consulting Developing theories
Engineering that lead to prudent
Net moisture loss Net moisture gain
Engineering
Engineering
Protocols
Protocols engineering protocols
Protocols
Negative pore-
water pressures
Unsaturated
Water table Soil Mechanics
Laboratory
Laboratory Numerical
Numerical
Figure 2: Moisture flux components associated with Testing
Testing Modeling
Modeling
the calculation of net moisture flux at ground surface
Measurement of Solving nonlinear partial
Figures 2 shows how negative pore-water unsaturated soil properties differential equations, PDFs
Pillars for implementing saturated-unsaturated
pressures in the vadose zone are altered by Figuresoil
3: Pillars related
mechanics to engineering
into the implementation
practiceof
imposed weather conditions. Negative pore- unsaturated soil mechanics
water pressures can also vary due to
fluctuations in the groundwater table. There are three classes of physical
Common to numerous geotechnical processes that have historically formed the
engineering problems (e.g., soil cover core of saturated and unsaturated soil
designs) is the assessment of the water mechanics. The physical processes relate to:
balance at the ground surface as a function of i.) flow and storage of water in a porous

2
medium, ii.) shear strength of a particulate played by the “shrinkage curve”, SC, in
material, and iii.) volume change refining the estimation of unsaturated soil
(i.e., compression, distortion and expansion) property functions, and iii.) illustrate how the
of soils. An important part of saturated soil SWCC should be analyzed for the
mechanics has revolved around the computation of unsaturated soil property
laboratory measurement of appropriate functions. The application of the SWCC in
saturated soil property constants. Bringing this paper is limited to the consideration of
unsaturated soil mechanics into routine water flow in geotechnical engineering
geotechnical engineering practice has problems. The manuscript is written in the
involved the estimation of unsaturated soil form of a state-of-the-art and state-of-the-
property function, USPFs, based on the practice paper based on years of research into
laboratory measurement of soil-water unsaturated soil behaviour as well as years of
characteristic curves, SWCCs. Estimation experience in putting unsaturated soil
procedures for USPFs have been proposed mechanics into routine engineering practice.
for all application areas of unsaturated soil The paper constitutes a generalized template
mechanics. While the above-mentioned for the analysis of water flow and storage in
processes form the core of soil mechanics, unsaturated soils.
the geotechnical engineer needs to be aware This paper illustrates the pathway that can
that numerous other physical processes also be taken from the start of addressing an
play an important role when considering unsaturated seepage problem to the end of
near-ground-surface phenomena (e.g., heat quantifying the required unsaturated soil
flow, air flow, phase changes, chemical property functions, USPFs. An assumption is
transport). Engineering “protocols” for the made in this paper that the soil under
practice of unsaturated soil mechanics have consideration will be tested in the laboratory
emerged over time and are being applied to measure the entire soil-water characteristic
worldwide in a relatively consistent manner. curve, SWCC. If the soil undergoes volume
The practice of unsaturated soil mechanics change as soil suction is increased, then the
has been closely related to the use of the shrinkage curve, SC, must also be obtained
SWCC and for this reason it is prudent to through measurement or estimation.
summarize, as far as is possible, aspects that
appear to “define” the discipline of The following assumptions and conditions
unsaturated soil mechanics. are imposed.
The determination (i.e., measurement 1.) The entire SWCC extends over a soil
and/or estimation) of the soil-water suction range from about 0.1 kPa to
characteristic curve, SWCC, along with 1,000,000 kPa, and it is defined using
saturated soil properties has provided a laboratory measurements.
means of moving unsaturated soil mechanics 2.) The drying SWCC is measured over
into routine engineering practice. the entire soil suction range.
Consequently, a thorough understanding of 3.) An assessment is made as to whether
the SWCC becomes pivotal to the practice of or not there is significant volume
unsaturated soil mechanics. change as soil suction is increased.
The primary objectives of this paper are to: 4.) The shrinkage curve for the soil needs
i.) describe the primary role played by the to be measured when changes in
“soil-water characteristic curve”, SWCC in overall volume need to be separated
estimating unsaturated soil property from changes in degree of saturation
functions, ii.) describe the secondary role as soil suction is increased.

3
 5.) Procedures to address the effects of suctions for all soil types. A single pathway
hysteresis should be assessed in light is followed in the determination of USPFs for
of the engineering problem being seepage problems in geotechnical
addressed. engineering practice.
6.) All analytical steps are explained in 2 HISTORICAL CONTEXT FOR THE
going from the laboratory
DEVELOPMENT OF
measurements of the gravimetric UNSATURATED SOIL MECHANICS
water content SWCC and shrinkage
curve, (i.e., w-SWCC and SC) to the The study of unsaturated soil behavior has
calculation of the unsaturated soil historically emerged on two fronts; one
property functions, USPFs (i.e., within soil physics (and related agricultural
permeability function and water disciplines), and the other within soil
storage function). mechanics. Figure 4 illustrates how the
7.) Assumptions associated with all physics of unsaturated soil behavior found its
aspects of the analysis are explained expression in different applications areas
along with the significance of each within soil physics and soil mechanics.
assumption.
Developments in Unsaturated Soil
The scope of this paper is limited to the Mechanics
consideration of one class of geotechnical
engineering problems; namely, the flow of
Soil Physics Geotechnical Engineering
water through a saturated-unsaturated soil
system. The soil continuum is assumed to not
Water Flow Shear Volume Change
have significant secondary structure such as Strength (Volume-mass
changes)
fractures, cracks and fissures. The effects of
hysteresis and complex stress paths are Measurement of Measurement of Measurement of
SWCC & estimation of SWCC & estimation SWCC & SC &
addressed in a surficial manner due to lack of hydraulic properties of shear strength estimation of volume-
space. mass properties

There are other techniques that have been Figure 4: Historical developments for unsaturated
used to obtain an estimation of the soil-water soil behaviour
characteristic curve for a soil such as; i.)
database mining, (M. Fredlund; 1997), and Some of the early contributors in soil
ii.) calculations based on the grain-size physics are as follows: Haines (1927);
distribution curves (M. Fredlund et al., 2002). Richards (1931); Edelefsen and Anderson
There are engineering situations where these (1943); Childs and Collis-George (1950);
methodologies can be used; however, only Klute (1952); Burdine (1952); Gardner
methodologies based on laboratory (1961); Brooks and Corey (1964); Topp and
measurements of the SWCC and SC are Miller (1966); van Genuchten (1980); and
given consideration in this paper. Mualem (1976). Some of the early research
Numerous empirical equations have been contributors in geotechnical engineering can
proposed to characterize or best-fit SWCCs. be listed as follows: Croney (1952); Jennings
There is no attempt in this paper to compare and Knight (1957); Bishop, Alpan, Donald
the proposed empirical equations or address and Blight (1960); Aitchison (1961); Bishop
the limitations associated with various and Blight (1963); Jennings (1969); Barden
SWCC equations. Rather, an attempt is made (1965); Lytton and Woodburn (1973);
to illustrate the use of a set of mathematical Fredlund and Morgenstern (1976); Escario
equations that cover the entire range of soil (1980); and Ho and Fredlund (1982). The
above-mentioned list is by no means
4
complete; however, the list of references Several research conferences were held
illustrates the different time periods over in the 1960s and there was an attempt to
which basic unsaturated soils research was transfer research formulations related to
conducted. physical processes studied in soil physics into
the emerging field of unsaturated soil
2.1 Soil Physics Research in Unsaturated
mechanics. The proceedings of the 1961
Soil Behavior
London, England conference was titled,
Extensive research studies took place within “Pore Pressure and Suction in Soils”. In 1965
soil physics in the early 1900s. These studies a symposium-in-print titled “Moisture
mainly focused on moisture movement Equilibria and Moisture Changes in Soils
through soils in the vadose zone. Soil suction beneath Covered Areas” was edited by G.
was introduced as the energy head driving Aitchison (Aitchison, 1965) and presented to
water flow. Diffusive type models were the First International Conference on
proposed within soil physics with the Expansive Soils at Texas A & M, College
coefficient of diffusion combining the ease of Station, TX, in 1965. These conferences
water flow (i.e., coefficient of permeability) mainly focused on moisture movement and
with water storage capacity (i.e., water swelling clay problems encountered in
storage). The diffusive type formulation geotechnical engineering. A series of
required the input of a single diffusive soil research studies were undertaken at Imperial
property (Richards, 1931). College, London, in the 1950s and 1960s that
Little attention was given to overall identified some of the key aspects of
volume change of the soil; in fact, unsaturated soil behaviour and set the course
formulations generally assumed that the soil for subsequent research studies in many
structure was rigid. The earliest research into countries around the world. The research
unsaturated soil behavior was undertaken in studies at Imperial College were mainly
soil physics and mainly applied in under the supervision of Professor Allan W.
agriculture-related applications. The water Bishop and one of the primary contributing
storage capacity of near-ground-surface soils researchers was Geoffrey E. Blight in whose
was of interest from the standpoint of plant honour this lecture is given.
growth (Buckman and Brady, 1960). The A number of difficulties became apparent
amount of water storage in the soil was as mathematical formulations were
measured in terms of volumetric water transferred from soil physics into unsaturated
content, θw, and presented as a function of the soil mechanics. Geotechnical engineers were
negative pore-water pressure (or suction) in accustomed to using soil mechanics’
the soil. principles for water flow through soils. For
2.2 Geotechnical Engineering Research in example, hydraulic head (i.e., Y + uw/w
Unsaturated Soil Behavior where Y = elevation head, uw = pore-water
pressure, and w = unit weight of water) was
Geotechnical engineers were interested in used as the driving potential for saturated
water flow through unsaturated soils as well soils and the desire was to maintain a similar
as the shear strength and volume change formulation for water flow above the phreatic
behaviour of unsaturated soils. Shear strength surface. There was opposition to the use soil
and volume change problems were common suction and volumetric water content as
in geotechnical engineering practice and driving potentials for water flow in the
there was a desire for improved engineered unsaturated soil region.
solutions.

5
 Geotechnical engineers also viewed 2.3 Defining Soil Behaviour in Terms of the
water flow problems in terms of steady state Stress State of an Unsaturated Soil
and transient type analyses. As a result, the Probably the first research document
hydraulic properties for unsaturated soils mentioning the need for two independent
were viewed in terms of two independent soil stress state variables when describing
property functions; namely, i.) the water
physical processes in unsaturated soils was
permeability function, kw, and ii.) the water published by Biot (1941). Biot derived the
storage function, m2w. These two material theory of consolidation in terms of a partial
property functions served different roles differential equation for an unsaturated soil.
when considering solutions of interest in The derivation gave consideration to a pore
geotechnical engineering. Each of the fluid of water which contained air bubbles.
hydraulic property functions involved Even for this special case of an unsaturated
different mathematical operations for their soil, the use of two independent stress state
assessment even though both properties were variables was proposed. Dakshanamurthy et
closely related to the soil-water characteristic
al., (1984) showed that the 1941 Biot theory
curve, SWCC. of consolidation derivation could also be
One of the practical engineering
applied to an unsaturated soil with
problems facing geotechnical engineers was continuous air and water phases. Coleman
the prediction of heave in swelling soils. The (1962) also suggested the use of independent
need to predict total heave in swelling soils stress state variables when considering the
provided the primary impetus for a series of
volume change behavior of an unsaturated
international research conferences from 1964 soil.
to 1992. The conferences were directed Bishop (1959) proposed an effective
towards better understanding expansive soil stress equation that related the total normal
behavior. Formulations of moisture stresses to the matric suction through use of
movement in soil physics were based on the
an empirical  soil parameter. An extensive
assumption that the elemental volume under
experimental study by Bishop and Blight
consideration was rigid and therefore, did not
(1963) on the shear strength of several
undergo volume change. Consequently, it
different soils showed that the difference
was important to re-derive a more fitting
between the shear strength of a saturated soil
partial differential equation for use in
and an unsaturated soil was related to the
modeling moisture flow and volume change
degree of saturation of the soil at failure.
for geotechnical engineering applications.
Figure 5 shows a plot of the test results from
Studies in soil physics gave little or no
four different soils that were tested. These
consideration to shear strength and volume
early results show that the researchers were
change problems. The primary problems of
aware that the unsaturated soil shear strength
interest within soil physics were related to the
of a soil was related to the degree of
water storage and its depletion near the
saturation of the soils; however, the
ground surface. There was little research
relationship was not linear, suggesting that
consideration given to shear strength and
the relationship might be somewhat more
volume change of unsaturated soils.
complex. It was concluded that it is the
Consideration of shear strength and volume
“stress paths of two components, (σ – ua) and
change problems in unsaturated soils meant
(ua – uw) which have to be taken into
that the stress state variables for an
account.” The difference between ua and uw
unsaturated soil needed to be proposed,
was referred to as the matric suction.
verified and agreed upon (Fredlund, 2006).

6
 100
20% kaolin. The results were presented in
80
Vaich moraine terms of two independent stress components,
Degree of saturation, S (%)

Talybont clay
Selset clay
(σ – ua) and (ua – uw). Several other studies
60 Mangla shale have been undertaken related to the study of
changes in volume-mass behavior of
40 unsaturated soils. In general, the results were
presented in terms of the stress paths
20
followed in each of the tests (Barden et al.,
0
1969). In each of the above cases the
0 0.2 0.4
χ factor
0.6 0.8 1 measured constitutive surfaces proved to be
Figure 5: Relationship between shear strength stress path dependent (Pham and Fredlund,
parameters measured on several saturated and 2011).
unsaturated soils (modified from Bishop and Blight, Fredlund and Morgenstern (1977)
1963) presented a theoretical stress equilibrium
analysis justifying the components of an
In 1965, Blight presented results on the
unsaturated soil. The analysis was presented
volume change behavior of an unsaturated
within the context of continuum mechanics
soil. The behavior of the unsaturated soil was
principles and concluded that two
illustrated in terms of three-dimensional
diagrams with the horizontal axes comprised independent stress tensors consisting of (σ –
of two independent stress variables and the ua) and (ua – uw) were best suited for the
interpretation and application of unsaturated
ordinate being volumetric strain (Figure 6).
soil behavior. Research to-date would appear
Volumetric strain
Volumetric strain to indicate that two independent stress
Plane of volume
Plane of volume change
tensors form an adequate stress state
change duein to suction
due to suction
inunconfined
unconfinedsoil soil
description for physical processes involving
Expansive
soil
Expansive
soil Collapsing
Collapsing soil
unsaturated soils. It is possible that
soil
Volume change
Plane of volume for
change
Compression at
Compression
constant water at simplifications representing the stress state of
constant water
a due
saturated
saturated soil
soil
to suction in
Compression
content
content
an unsaturated soil may prove to be adequate
Compression curves
for saturated
curves for
soilsoil
saturated
for describing constitutive behaviour in some
cases but in general, two independent stress
Compression
Compression
curves for
curves
unsaturatedfor
soil
tensors would appear to be the more rigorous
unsaturated soil and generally acceptable description.
Figure 6: Void ratio versus stress state variables when
following stress paths for a stable-structured soil (after
Blight, 1961) 2.4 Discoveries in Soil Physics related to
the Soil-Water Characteristic curve,
Stress paths corresponding to various SWCC
physical processes were illustrated and it was
noted that understanding the collapse The soil-water characteristic curve, SWCC,
phenomena, in particular, turned out to prove has been central to water movement
challenging. Consideration of other stress modeling from the early studies in soil
paths revealed further stress path dependence physics (Klute, 1965). In 1986, Klute
(Fredlund and Morgenstern, 1976). Matyas identified a series of “findings” related to the
and Radhakrishna (1968) studied volume SWCC that form important reference points
change and degree of saturation changes for its usage in both soil physics and
while performing isotropic and Ko triaxial geotechnical engineering. Some of these
tests on a mixture of 80% flint powder and “finding” are paraphrased below. Words in

7
quotation marks are taken directly from Klute wetting, and iv.) there are an infinite set
(1986). of scanning curves inside the drying
1.) The relationship between soil water and wetting bounding curves. Figure 7
content and soil suction (i.e., herein shows a typical set of (initial and main)
referred to as the soil-water drying and wetting SWCCs. The
characteristic curve, SWCC, but also compilation of SWCCs defines the
referred to as the water retention curve, boundaries for the relationships
WRC, in the research literature) relates between the amount of water in the soil
a “capacity factor” (i.e., the amount of and soil suction.
water in the soil), and “the energy
100.
state”, (i.e., suction stress state) for the Initial drying curve
So.

Degree of saturation, S (%)

soil-water. Main drying
curve
2.) The SWCC “is a fundamental part of
Wetting to
the characterization of the hydraulic drying
properties of a soil”. reversal
Reversal
3.) The “energy per unit volume”, is Reversal point
Drying to wetting
point
equivalent to force per unit area or reversal
SSrr.. Main wetting
pressure (i.e., soil suction). curve
4.) Water content can “be expressed on a
0.1 1.0 10. 100.
weight, volume, or degree of saturation Soil suction, kPa
basis”. “For analysis of water flow in Figure 7: Family of drying and wetting curves of
soil profiles, the volume basis is most degree of saturation versus soil suction for a rigid
useful”. While the volume basis for structured soil (modified from Klute, 1986)
water content has been used in soil
physics, other designations for the 8.) The SWCC for “rigid structure soils
amount of water need to be given show constant water content up to the
consideration for geotechnical air-entry value” for the soil.
engineering. 9.) If the soil structure deforms (i.e.,
5.) The SWCC is “primarily dependent shrinking or swelling), the water
upon the – particle-size distribution of content can decrease as soil suction
the soil and the structure” (Croney et increases without reaching the entry of
al., 1958). air, (i.e., no change in degree of
6.) The SWCC is hysteretic (i.e., water saturation).
content at a given suction on the 10.) In some cases, only the drying SWCC
wetting curve is less than that along the is required for modeling water flow. In
drying curve), (Haines, 1927; Topp and other cases, only the wetting curve is
Miller, 1966). required. Modeling flow across the
7.) The branches of the SWCC are defined ground surface requires the
as follows: i.) the initial drying curve incorporation of hysteretic behavior
starting at a degree of saturation of (Mualem, 1976).
100%, ii.) the main wetting curve 11.) The SWCC is a constitutive material
measured after the soil has been dried property that relates a stress state (e.g.,
to near residual water content matric suction and total suction) to the
conditions, iii.) the main drying curve amount of water in a soil.
which may have 10 to 20% entrapped 12.) Water flow occurs in response to
air due to incomplete saturation upon gradients of total hydraulic head (i.e.,

8
 pore-water pressure head plus There has been fruitful and overlapping
elevation head). research undertaken within soil physics and
13.) SWCCs should be measured on geotechnical engineering. The most
“undisturbed core samples”. In some significant area of over-lapping research is
cases, remolded and disturbed material related to the use of the soil-water
may be all that is available for the characteristic curve, SWCC, for the
measurement of the SWCC. estimation of unsaturated soil property
It should be noted that there is an functions, USPFs. Much of the remainder of
arbitrary division between what is referred to this paper is devoted to describing the
as the “low suction range” and the “high application of the SWCC in solving typical
suction range”. The arbitrary division occurs geotechnical engineering problems. It has
at around 1500 kPa; an arbitrary division that been observed that the application of the
is established mainly on the basis of the SWCC requires careful consideration of the
highest air-entry ceramic disk that can be assumptions made as part of the estimation
manufactured. Consequently, soil suction is procedures used in calculating the USPFs.
defined in terms of matric suction from zero The basic measurement of water content
to 1500 kPa, while soil suction is defined in versus soil suction, (i.e., soil-water
terms of total suction in the range from 1500 characteristic curve, SWCC, or water
to 1,000,000 kPa. retention curve, WRC) originated within the
The arbitrary division between the low soil physics discipline. The emergence of
and high suction ranges results in two unsaturated soil mechanics witnessed an
different components of soil suction being attempt to transfer the experience and
used when measuring the SWCC of a soil. technology developed in soil physics into
The change in the soil suction components geotechnical engineering. The soil physics
along the SWCC would appear to mainly be discipline historically presented unsaturated
related to the influence of osmotic suction. soil behaviour in terms of a plot of volumetric
However, the inconsistency in the use of two water content versus soil suction. The use of
suction components does not appear to volumetric water content to designate the
created significant difficulties in applications amount of water in the soil appears to have
in both soil physics and geotechnical been influenced by agriculture related issues
engineering (Fredlund, 2015). (e.g., water storage for plant growth). Overall
volume changes related to suction changes
2.5 Historical contributions from Soil
were of secondary interest and were not taken
Physics’ Research for use of the SWCC
into consideration in soil physics
The historical context for unsaturated soil applications.
mechanics reveals that key complimentary
3.0 APPARATUSES FOR MEASURING
findings have emerged within the disciplines
THE DRYING SWCC
of soil physics and geotechnical engineering.
Geotechnical engineers have benefited much Laboratory tests in soil physics focused
from the research undertaken in soil physics. mainly on the measurement of the drying
At the same time, lessons have been learned soil-water characteristic curve. Most
about the importance of carefully examining laboratory test equipment for the agriculture-
assumptions associated with mathematical related disciplines was designed to
formulations (i.e., definition of material simultaneously measure the SWCC on
properties and physical processes) when several soil specimens that were placed on a
moving from one discipline to another. single high air-entry ceramic disk (Fredlund
and Rahardjo, 1993; SoilMoisture

9
Equipment Corporation, 1983). Following Interpretation of SWCCs

the establishment of equilibrium suction

conditions (using the axis-translation
No Volume Change with Soil Volume Change with Soil Suction
technique), each soil specimen was removed Suction Changes Changes
from the pressure plate and its water content Measure:
Measure:
was measured. Exceptions to the w-SWCC w-SWCC
Shrinkage curve, e vs. w
conventional apparatuses were the Tempe
Compute Volume-Mass relations:
cell and the volumetric pressure plate cell S e = w Gs
Compute Volume-Mass relations:
e vs. Suction; k = f(e)
which tested a single soil specimen. Θ vs. Suction
Θ vs. Suction
- Water storage; m2w = f(Θ)
The requirements for measuring the - Water storage; m2w = f(Θ)
w vs. Suction
S vs. Suction
SWCC in geotechnical engineering are - Air-entry value
- Air-entry value
- Residual conditions
different from those in soil physics. The - Residual conditions
- Permeability function;
- Permeability function;
- k = f(w)
amount of volume change experienced by the - k = f(S)
soil specimen as soil suction is increased is Figure 8: The influence of overall volume change on
important to the interpretation of the the volume-mass versus suction relations
laboratory result (Figure 8). The amount of
volume change that occurs as soil suction is
increased is one of the first factors requiring
a decision when measuring the SWCC for
geotechnical engineering applications. It is
assumed that the specific gravity of the soil is
known prior to analyzing the SWCCs.
The use of the basic SWCC results (i.e.,
gravimetric water content versus soil suction)
for the estimation of the unsaturated soil
property functions is sufficient provided
there is limited volume change as soil suction
is increased. However, overall volume
change as soil suction is increased can
significantly affect the calculation of
Figure 9: Front and side views of GCTS Pressure Plate
unsaturated soil property functions. Various cell (Courtesy of GCTS, AZ)
volume-mass versus soil suction relations
need to be computed when the soil undergoes A single soil specimen is tested in the
volume change as soil suction is changed. more recent pressure plate devices without
Apparatuses developed in soil physics disturbing the soil between suction
have been historically used to measure the applications. The pressure plate apparatus is
SWCC in geotechnical engineering used to establish matric suction in the lower
applications. However, over time, various suction range (i.e., suctions less than 1500
apparatuses have been developed that are kPa). Vapor pressure equilibrium conditions
better suited to meet the requirements for have been used to determine water content
geotechnical engineering applications. conditions for the high suction range (i.e.,
The equipment developed by GCTS total suctions greater than 1500 kPa).
(Figure 9) is typical of a number of Ko type Figure 10 shows a typical set of test results
apparatuses that better satisfy the needs measured on a silt soil (Pham 2000). It is
within geotechnical engineering. possible to measure both the drying and the
wetting SWCC; however, it is of primary

10
importance to measure the drying curve in
geotechnical engineering applications in 3.1 Steps to Estimating Unsaturated Soil
order to obtain the most accurate Property Functions, USPFs
interpretation of the SWCC for the
calculation of the unsaturated soil property A dataset for an example soil is generated to
functions. It is also possible to directly illustrate the use of the soil suction stress
measure overall volume change of the soil variable (i.e., matric suction and total
specimen under Ko conditions as long as the suction) to define the water flow and water
soil specimen does not separate from the storage constitutive properties for an
confining metal ring. unsaturated soil. It is known that unsaturated
soil properties are also affected by total
applied stresses; however, the assumption is
made that soil suction is the dominant stress
variable defining hydraulic soil properties.
This assumption appears to be adequate for
most geotechnical engineering applications.
The steps involved in estimating
unsaturated soil property functions, USPFs,
through use of the SWCC are listed below.
Subsequent sections in this paper address
details pertaining to each of the steps in the
analysis. The steps of the analysis are
outlined for the case where the soil undergoes
Figure 10: Drying and wetting SWCCs measured on some volume change as suction is increased.
a silt soil using the GCTS Pressure Plate apparatus Shortcuts in the analysis are identified for
(Pham 2002) situations where the soil does not undergo
significant volume change as suction is
An accurate interpretation of the SWCC increased. The effect of volume change can
involves the separation of desaturation of a generally be ignored when dealing with sand
soil specimen from the effects of overall and coarse-grained soils.
volume change (i.e., changes in void ratio). It Step 1: Take note of the assumptions and
is possible to develop triaxial testing limitations associated with the analysis for
equipment that can simultaneously measure the estimation of unsaturated soil properties.
both volume change and water content i.) Consideration is only given to the
change; however, it is more economical and analysis of the drying (or desorption)
expedient to independently measure the measured data (i.e., gravimetric water
shrinkage curve, SC, for the soil (i.e., void content versus soil suction).
ratio versus gravimetric water content), and ii.) A decision should be made at the start
then use this information for the of the testing program regarding
interpretation of the gravimetric water whether or not volume changes related
content SWCC. This paper focuses on a to suction changes needs to be taken
detailed analysis of the main drying SWCC. into account.
The analysis for the main wetting SWCC is iii.) The specific gravity, Gs, of the soil
largely outside the scope of this paper must have been measured.
although the wetting curve is given some iv.) The following aspects of the analysis
consideration. apply for soils that behave in a
unimodal manner. Bimodal behavior is

11
 considered to be outside the scope of soil mass is also measured. Other
this paper. methods have also been used to
v.) Terminology: The term “soil suction” measure the volume of the soil
or “suction” refers to matric suction in specimens during drying (Liu and
the range of suctions up to 1500 kPa. Buzzi, O., 2014; Liu et al., 2016).
The term “soil suction” or “suction” ii.) In some cases it is possible to estimate
refers to total suction in the range of the shrinkage curve with sufficient
suctions between 1500 kPa and one accuracy.
million kPa. Matric and total suctions iii.) Obtain the fitting parameters for the
are plotted using a continuous shrinkage curve. The M. Fredlund
logarithmic scale throughout the (2000) equation is used in this paper.
respective suction ranges. iv.) The shrinkage curve does not need to
Step 2: Measure gravimetric water be measured when the soil does not
content versus soil suction (w-SWCC) over undergo volume change upon suction
“Shrinkage curve” provides the relationship between
the entire suction range. changes.
instantaneous void ratio and gravimetric water content under
unconfined loading conditions
i.) Determine the initial volume-mass
properties of the wetted w-SWCC soil
specimen. Use of a digital micrometer for
diameter and thickness
ii.) Combine and plot the data from a low measurements
suction value to a high suction value
(i.e., starting at 0.01 or 0.1 kPa to a Specimen size:
12 mm thick
limiting value of 1,000,000 kPa) 37 mm diameter
iii.) Obtain the fitting parameters for an
equation that fits the data over the
entire range of suction values. The Figure 11: Measurement of the “Shrinkage Curve”,
Fredlund-Xing (1994) equation is used SC using micrometer calipers
in this paper; however, other suitable Step 4: Calculate and plot the void ratio
equations can be used provided it fits versus suction relationship.
the data over the entire suction range. It i.) The combination of the SWCC and the
is sometimes difficult to get a close fit shrinkage curve allow the separation
of the data points in cases where the between volume change and
soil undergoes large volume changes desaturation associated with suction
during drying. In such cases, it is increases.
possible to use a bimodal form of the ii.) The void ratio plot is not required when
Fredlund-Xing equation to fit the w- the soil does not undergo significant
SWCC data points. volume change as suction changes.
Step 3: Measure (or estimate) the Step 5: Calculate and plot the degree of
shrinkage characteristics of the soil. saturation SWCC, (S-SWCC)
i.) The soil specimen is commonly i.) Obtain the fitting parameters for the
prepared as a saturated paste, placed degree of saturation SWCC.
within a ring and allowed to slowly dry. ii.) Calculate the true Air-Entry Value,
It is also possible to test undisturbed AEV, from the S-SWCC.
soil specimens which are initially iii.) Calculate the Residual Point (i.e.,
saturated. Measurements of the residual suction and residual degree of
specimen volume are taken using saturation).
micrometer calipers (Figure 11). The

12
 Step 6: Calculate and plot the volumetric calculated to be 96.08% and the saturated
water content SWCC. coefficient of permeability is 2.0 x 10-6 m/s.
i.) The volumetric water content SWCC is Pressure plate apparatuses with high air-
computed from the gravimetric water entry ceramic disks have become the most
content SWCC and the shrinkage common means of measuring the SWCC in
curve. the suction range up to 1500 kPa. Pressure
Step 7: Calculate the Unsaturated Soil plate apparatuses apply matric suctions using
Property Functions, USPFs. the axis-translation technique.
i.) Calculate the permeability function Vapor pressure equalization
with respect to void ratio. methodologies are used to establish total
ii.) Calculate the relative permeability suction environments in the range in excess
function with respect to soil suction of 1500 kPa. The suggested methodologies
starting from the true air-entry value. used in soil physics (Klute, 1965) have
iii.) Calculate the water storage function essentially become the accepted procedures
from the volumetric water content used in geotechnical engineering except for a
SWCC. few exceptions (Fredlund and Rahardjo,
iv.) Calculate other USPFs (e.g., shear 1993). The methodology involves wetting the
strength function, thermal property soil to zero suction at the beginning of the test
functions). and then applying a small suction in the range
Each of the above-mentioned steps is between 0.1 and 1.0 kPa to establish the
more clearly detailed in the following initial water content corresponding to the
sections. start of the w-SWCC.
Figure 12a shows a typical dataset for the
basic drying w-SWCC. The diamond-shaped
3.2 Measurement of the Gravimetric Water symbols indicate matric suction data
Content SWCC obtained from a pressure plate apparatus. The
The classification properties of the soil, along square symbols indicate total suction data
with the specific gravity, Gs, should be obtained by using a vapor pressure
determined prior to commencing the w- equalization procedure (e.g., WP4-T chilled-
SWCC test. The initial volume-mass mirror apparatus; Decagon; 2009).
properties (e.g., water content, void ratio, and The entire dataset from a fraction of 1 kPa
degree of saturation) are required as part of to one million kPa can be best-fit with a
the analysis of volume-mass versus soil mathematical function that extends over the
suction data. A record should be kept of the entire range. Data for the drying SWCC
initial state of the soil specimen (i.e., slurry, generally takes on the form of a sigmoidal
compacted, or undisturbed). The soil mathematical function (Figure 12b) provided
specimen is initially allowed free access to the soil specimen does not undergo excessive
water and the soil moves towards saturated volume change. It should be noted that any
conditions. mathematical equation can be used for the
Data for an “artificial clayey silt” soil is best-fit provided the data can be fit in the low
used throughout this paper to illustrate the suction range while ending at zero water
steps involved in analyzing laboratory content at one million kPa.
measurements. The specific gravity of the
soil is 2.68, the initial gravimetric water
content is 31.5%, and the initial void ratio is
0.879. The initial degree of saturation is

13
 35 35

30 30
Matric suction Matric suction
Gravimetric water content (%)

Gravimetric water content (%)

25 Total suction 25 Total suction
Matric suction Best-fit data using
Best-fit line
20
Fredlund-Xing (1994)
20 af = 75.37 kPa
Basic data from equation
Pressure Plate and nf = 1.634
15 Vapour Pressure 15
Equalization tests
Specific gravity = 2.68 mf = 0.716
Specific gravity = 2.680 Initial volume-mass variables Residual suction =
10 Water content = 31.5%
Initial volume-mass variables 10
Total suction 1000. kPa
Water content = 31.50 %
5
5

0
0.1 1 10 100 1000 10000 100000 1000000 0
0.1 1 10 100 1000 10000 100000 1000000
Soil suction (kPa)
Soil suction (kPa)
Figure 12a: Matric and total suction data for the w- Figure 12b: Soil suction data with best-fit Fredlund-
SWCC Xing (1994) SWCC function

3.2.1 Plotting and Best-Fitting the Following is the proposed Fredlund-Xing

gravimetric water content SWCC (w- (1994) equation applied to the gravimetric
SWCC) water content versus soil suction data for the
The Fredlund and Xing (1994) equation example soil.
(with the applied correction factor for zero
water content at one million kPa of suction),
is used in this paper to illustrate the steps 𝑤𝑠 𝐶()
𝑤() = 𝑛 𝑚𝑓 [1]
involved in analyzing gravimetric water (ln(exp(1)+(/𝑎𝑓 ) 𝑓 ))
content versus soil suction data.
The best-fit regression analysis using the where: w() = water content at any soil
Fredlund and Xing (1994) equation yields 4 suction; ; af = fitting parameter near the
fitting parameters that provide a inflection point on the w-SWCC; nf =
mathematical representation of the w-SWCC. fitting parameter related to the maximum
The residual suction value can be rate of gravimetric water content change;
approximated using the empirical mf = fitting parameter related to the
construction procedure shown in Figure 12b. curvature near residual gravimetric water
The initial fitting of the gravimetric water content conditions; r = suction near
content SWCC should not be confused with residual conditions of the soil, and C() =
later best-fitting of the degree of saturation correction factor directing the w-SWCC
SWCC (S-SWCC). It should be noted that the towards a suction of 106 kPa at zero water
fitting parameters for w-SWCC can have content, written as:
slightly different meaning from those ln (1 +   r )
C ( ) = 1 −
obtained for the degree of saturation SWCC, 6
ln (1 + 10  r )
(S-SWCC).
Other empirical equations such as that Inserting the correction factor, C(), into
proposed by Pham and Fredlund (2011) can Eq. [1] yields the following form for the
be used at this stage to provide a closer fit of Fredlund-Xing (1994) equation.
measured data throughout the entire range of
suction values. The important guideline is
w ( ) =
(
ws 1 − ln (1 +   r ) ln (1 + 106  r ) ) [2]
that the fitted equation should closely adhere
(ln (exp (1) + ( a ) ))
mf
nf
to the measured (w-SWCC) laboratory data. f

14
 There is close agreement between the gravimetric water content change can be due
Fredlund-Xing (1994) equation and the to two processes (Fredlund and Rahardjo,
laboratory data shown in Figure 12b; 1993).
however, this is not always the case. The
fitting parameters are: residual suction, r = S f  e + e f S
w =
1000 kPa; af = 75.37 kPa; nf = 1.634 and mf Gs [4]
= 0.716. The w-SWCC shows that the
mathematical function starts to bend where the subscript ‘f’ refers to the “final”
downward in the vicinity of 20 kPa; however, volume-mass states. The separation of the
it should be noted that this bend in the curve volume change and degree of saturation
does not indicate the air-entry value of the change can be accomplished through use of a
soil (except in the case where there is no shrinkage curve.
volume change as soil suction is increased).
The “true” air-entry value will be later 3.3 Measurement of the Shrinkage Curve
computed from the degree of saturation Soil suction increases as a soil specimen dries
SWCC. from an initially wet condition (i.e., near zero
When there is considerable overall suction) to a suction of one million kPa.
volume change in the low suction range, it Measurements of changes in mass and
might not be possible to obtain a close fit of volume as the soil dries allows for the
the measured data points using the Fredlund- determination of the shrinkage curve, SC.
Xing (1994) equation or possibly any other Specimens for the shrinkage curve test
commonly proposed fitting equation. It is should have similar initial volume-mass
important to have a close fit of the data on the properties (i.e., water content and void ratio)
w-SWCC when the data is combined with the to those used for the w-SWCC laboratory
shrinkage curve data for the calculation of test. Figure 13 illustrates how the soil
other volume-mass SWCCs. Consideration specimens for the w-SWCC test and the
might be given to fitting the w-SWCC data shrinkage test can be prepared to give similar
using bi-modal forms of the equation or the initial (wet) conditions.
proposed Pham-Fredlund (2008) equation.
Changes in gravimetric water content ~ 70 mm

versus suction does not allow for the ~ 30 mm

separation of the two independent processes
that occur as the soil dries; namely, water
SC
content changes associated with volume Specimen ~ 10 mm
changes and water content changes
~ 30 mm
associated with changes in degree of
saturation. The following basic volume-mass SWCC Specimen
equation illustrates the relationship amongst
the volume-mass properties of a soil. Figure 13: Establishing similar initial volume-mass
conditions for the w-SWCC and SC tests
S e = Gs w
[3] Historical differences in the
where: S = degree of saturation, and e = interpretation of SWCC data centers around
void ratio. an assumption related to the rigidity of the
soil structure. The assumption is generally
Incremental differentiation of the basic made in soil physics that the soil structure is
volume-mass relationship illustrates that the rigid. Invoking the assumption of a rigid soil

15
structure can introduce significant errors can be calculated from the shrinkage limit of
when calculating subsequent unsaturated soil the soil. The bsh parameter is also closely
properties functions for geotechnical related to the ash parameter through the above
engineering applications. It is important in equation. An incompressible soil has a high
geotechnical engineering to quantify the csh value (e.g., csh = 50). In general, the
volume change characteristics of the soil with shrinkage curve is easy to estimate or
respect to changes in soil suction. measure.
The measurement of the shrinkage curve 3.4 Void Ratio versus Soil Suction
for a soil provides data on the relationship Relationship
between gravimetric water content and
volume change (i.e., void ratio change) as The volume change associated with suction
soil suction is increased from essentially a change can be calculated by combining the
zero value to one million kPa as shown in empirical equation for the shrinkage curve,
Figure 14. The shrinkage curve allows for the SC, with the gravimetric water content
calculation of all volume-mass variables with SWCC (w-SWCC) (Fredlund and Zhang,
respect to soil suction. 2013). The shrinkage curve relates void ratio
changes to gravimetric water content changes
1.6
as the soil dries and the w-SWCC relates
1.4
S = 20% S = 40% S = 60% S = 80% gravimetric water content to soil suction.
1.2 Drying from
initial saturated
Substituting the w-SWCC into the shrinkage
1.0
curve, SC, equation yields an equation for the
Void ratio

conditions
0.8
Line of saturation
ash = 0.501 void ratio (and overall volume changes), as
0.6
bsh = 0.181 soil suction increases during drying. The
0.4 csh = 5.618
Shrinkage curve
So = 96.08 %
resulting equation for void ratio versus soil
0.2
eo = 0.879 suction is shown in Eq. [6] and graphically
0.0
0 10 20 30 40 50 60 presented in Figure 15. All variables shown
Gravimetric water content, %
in Eq. [6] have been previously defined.
Figure 14: Variables associated with a laboratory
shrinkage curve test 1 csh
 
csh

A shrinkage curve equation proposed by e ( ) = ash  
s (
  w 1 − ln (1 +   ) ln (1 + 106  )
r r ) 
 + 1


((
  bsh ln exp (1) + ( a f ) ))  [6]
mf
M. Fredlund et al., (2002) can be used to best- n f 
 
  
fit the void ratio versus gravimetric water
content drying curve. 1.0

1 csh
  w csh  0.8

e ( w ) = ash    + 1 Volume change = 20 %

  bsh   [5] 0.6
 
Void ratio

where: ash = minimum void ratio upon 0.4

Gs = 2.680
complete drying; bsh = variable related to the 0.2 Initial water content = 31.50 %
Fredlund and
Zhang, (2013)
slope of the drying curve calculated as: bsh = Initial void ratio = 0.879
Initial degree of saturation = 96.08%
(ash ×So)/Gs, and csh = sharpness of curvature 0.0
0.1 1 10 100 1000 10000 100000 1000000
as the soil desaturates. Soil suction (kPa)

Figure 15: Void ratio versus soil suction showing

The best-fit parameters for the shrinkage volume change
curve for the example soil are: ash = 0.501, bsh
= 0.181, and csh = 5.618. The initial degree of The void ratio plot shows that volume
saturation, So, was 96.08%. The ash parameter change for the clayey silt starts at a suction

16
near to 20 kPa and continues to occur up to a The water storage function for a soil is
suction of about 400 kPa. The degree of calculated as the change in volumetric water
saturation plot later shows the separation of content with respect to a change in suction. It
the desaturation process from the volume is important to first compute the volumetric
change process as the soil dries. Desaturation water content versus suction relationship.
of the soil takes place when the applied The “instantaneous” volumetric water
suction exceeds the air-entry value of the soil. content of the soil, i, can be calculated based
There may also be a zone of applied matric on the w-SWCC and the shrinkage curve, SC,
suctions where both volume change and as shown in equation [7].
desaturation are occurring simultaneously.
Gs w ( )
3.5 Designation of Water Content of a Soil i ( ) = [7]
1 + e ( )
The amount of water in a soil can be
quantified in terms of: i.) gravimetric water
where w() = gravimetric water content
content, w, ii.) volumetric water content, w, written as a function of soil suction (w-
or iii.) degree of saturation, S. It is important
SWCC), and e() = void ratio as a function
that volumetric water content be defined with
of soil suction, .
respect to the “instantaneous” total volume
when the soil changes volume as soil suction The Fredlund-Xing (1994) w-SWCC
is changed. The amount of water in a soil has equation (i.e., equation [2]), can be
been historically quantified in the soil substituted into equation [7] along with the
physics discipline as the volume of water M. Fredlund (2000) shrinkage curve equation
referenced to the initial total volume of the (i.e., equation [5]) to give an equation for
soil. In geotechnical engineering, the volumetric water content written in terms of
volumetric water content should be defined basic laboratory data (i.e., equation [8]).
as the volume of water referenced to the
“instantaneous” total volume of the soil when
overall volume change occurs. It is i ( ) =
(
Gs ws 1 − ln (1 +   r ) ln (1 + 106  r ) )
  csh
 
1 csh

noteworthy that in either profession, water 


(
1 + a   ws 1 − ln (1 +   r ) ln (1 + 10  r )
 )  
(ln (exp (1) + ( a ) ))
mf
6
 
content is commonly measured as
nf

 sh    + 1  [8]
(( )) 
f

  bsh ln exp (1) + ( a f )

mf
  nf
 
 
gravimetric water content in the laboratory     

and the differences in the terminology
revolve around the manner in which the data It should be noted that other proposed
is reduced and applied. functional forms for w-SWCC and SC could
During early developments in also be substituted into equation [7].
unsaturated soil mechanics there appeared to The θi-SWCC for the sample soil data is
be little concern with regard to the manner in plotted as a continuous function in Figure 16.
which the amount of water in a soil was The calculation of the water storage function,
designated. Differences in the designation of m2w, is determined from the θi-SWCC and is
water content later became of increased shown later in this paper.
importance as geotechnical engineers
increasingly used the soil-water
characteristic curve for the estimation of
unsaturated soil property functions, USPFs.
3.6 Volumetric Water Content versus Soil
Suction Relationship

17
 All variables shown in equation [10] are
defined in terms of the original best-fit
variables associated with the w-SWCC and
the SC. Figure 17 shows the computed degree
of saturation SWCC data points calculated at
arbitrarily selected soil suction values (i.e.,
equally spaced points on a semi-logarithmic
plot).
100

Degree of saturation

Degree of saturation (%)

80
Figure 16: Volumetric water content, i, versus soil calculated at selected
suction 60 suction values

𝑆() = G w ( )
𝐺𝑠 𝑤()
3.7 Degree of Saturation versus Suction
S ( ) =𝑒(𝑤(s))
40

Relationship Based on w-SWCC and e ( w ( ) )

20
SC Data
The degree of saturation versus suction 0
0.1 1 10 100 1000 10000 100000 1000000
relationship can be computed by satisfying Soil suction (kPa)

the basic volume-mass relationship (i.e., Eq.

Figure 17: Data points for the degree of saturation
[3]) along with the w-SWCC and the SC SWCC calculated from w-SWCC and the SC
relationships as shown in equation [9].
3.7.1 Fitting of the Degree of Saturation
Gs w ( )
S ( ) = [9] SWCC (S-SWCC)
e ( w ( ) )
The calculated degree of saturation versus
where w() = Fredlund-Xing (1994) suction data points can once again be best-fit
equation for the w-SWCC or any other using the Fredlund-Xing (1994) equation.
equation that fits the laboratory data, and The best-fit provides a new set of fitting
parameters for further analysis of the degree
e() = void ratio written as a function of soil
of saturation SWCC. The Fredlund-Xing
suction, .
(1994) equation used to fit the degree of
Substituting the w-SWCC equation and saturation SWCC has the same form as
the SC equation into equation [9] allows the previously shown in Eq. [2] for the w-SWCC
calculation of degree of saturation versus data points. However, the fitting parameters
suction data points. In other words, the S- can have a slightly different meaning in the
SWCC data points are based on the original case where volume change occurs as soil
w-SWCC and SC laboratory data sets. The suction is increased.
resulting S-SWCC equation has the The new fitting parameters are
following form. designated as afs, nfs, mfs and rs. These fitting
parameters have the same general meaning as
S ( ) =
(
Gs ws 1 − ln (1 +   r ) ln (1 + 106  r ) ) those defined for the w-SWCC in the case
 
csh 1 csh
 when the soil does not undergo volume
(
m f   w 1 − ln (1 +   ) ln (1 + 10  )
) 
(( ))
6
 [10]
ash ln exp (1) + ( a f )
nf
 s r r
 + 1 change as soil suction is increased. The ‘s’

(( ( ) ( f) )) 
mf
+ 
nf
 subscript means the fitting parameters refer
  sh
b ln exp 1 a 
 
 
to the S-SWCC.

18
 S ( ) =
(
S0 1 − ln (1 +   rs ) ln (1 + 106  rs ) ) The S-SWCC shows that there is a
( (
ln exp (1) + ( a fs ) ))
[11]
m fs
n fs
gradual downward bend in the S-SWCC
relationship near a suction value of 100 kPa
where: S(ψ) = degree of saturation at any soil indicating an approximate value for the air-
suction, So = initial degree of saturation entry of the soil. While an approximate value
which is generally quite close to 100%; afs = can be estimated from the S-SWCC, it is also
fitting parameter near the inflection point on possible to determine a more precise and
the S-SWCC; nfs = fitting parameter related unique air-entry value.
to the maximum rate of degree of saturation 3.8 Analysis of the Degree of Saturation
change; mfs = fitting parameter related to the SWCC to obtain the Air-Entry Value,
curvature near residual degree of saturation AEV
conditions, and rs = suction near residual
conditions of the soil. An analytical procedure was described by
Zhang and Fredlund (2015) whereby a
The correction factor directing the S- unique and reproducible value can be
SWCC towards a suction of 106 kPa at zero computed and designated as the “true” air-
water content is included in Eq. [11]. entry value of the soil. It should be noted that
The residual soil suction can first be the calculated air-entry value is the result of
estimated using the empirical procedure an empirical construction that removes the
previously shown in Figure 12b. The residual curvature on the degree of saturation versus
logarithm of suction plot. The fitting
suction, rs, is estimated to be 2000 kPa. The
parameters for the S-SWCC can be used to
starting degree of saturation is the same as
compute the “true” air-entry value for the
previously calculated from the SC (i.e., So =
soil; however, the logarithm suction scale
96.08 %).
must first be transposed to an equivalent
The next step involves best-fitting the
arithmetic scale.
degree of saturation soil-water characteristic
The empirical construction associated
curve (S-SWCC) with the Fredlund and Xing
with the determination of the air-entry value
(1994) equation (Figure 18). A close fit is
requires the calculation of the point of
generally possible over the entire soil suction
inflection on the suction scale. A transformed
range as long as the S-SWCC is unimodal in
suction scale can be used for the
character. The new fitting parameters
differentiation step because it is difficult to
calculated for the S-SWCC are as follows: afs
calculate the correct inflection point directly
= 261.9 kPa, nfs = 1.922, and mfs = 0.519.
on the plot of a semi-logarithm relationship
100 for degree of saturation and suction. The
S = 96.08 % logarithmic suction scale can be converted to
Soo= 96.9 %
an arithmetic scale, , by using the following
Degree of saturation (%)

80
aafsfs==252.0
261.9
kPakPa
Objective: Calculate nnfsfs==1.841
1.922
60 the “true” air-entry mmfs ==0.5170
fs 0.519 scale transformation.
value from S-SWCC rs = 2000 kPa

 = log10 ( )
40
Degree of saturation [12]
Best-Fit F-X Equation
20

The transformed degree of saturation

0
0.1 1 10 100 1000 10000 100000 1000000 SWCC takes on the following mathematical
Soil suction (kPa)

Figure 18: Degree of saturation versus soil suction form shown in equation [13] and the plot of
with best-fit of the Fredlund-Xing (1994) equation

19
the transformed degree of saturation SWCC through the inflection point using the first
is graphically shown in Figure 19. derivative of the S-SWCC as the slope. The
line of tangency can be extended to cross a
SS ( ) =
(
S 0 1 − ln (1 + 10  rs ) ln (1 + 106  rs ) ) line passed through the initial degree of
( ln (exp (1) + (10 a fs ) )) [13]
m fs
 n fs
saturation. Details of the analysis are shown
on Figure 20.
where: SS(𝜉) = degree of saturation as a
function of the transformed suction. 100 Air-Entry Value,
Air-Entry Value
AEV
= 152.0=kPa
147 kPa
Determine the

Degree of saturation (%)

80
slope at the Intersection point
The remaining fitting parameters for the inflection point is the AEV
transformed degree of saturation equation are 60
i = 2.69 + Inflection point
,
the same as calculated by equation [11]. SS(i ) = 68.49 %
40 SS’(i ) = -0.532
So = 96.08 %
100 20 Tangency line through
inflection point
Log 10 Soil suction

So = 96.08 % Transformed x-scale

Degree of saturation (%)

80 0
S
aofs==96.9 %
261.9 kPa -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
Determine the afs = 252.0 kPa
n = 1.922 Log10 Soil suction (kPa)
60 inflection point on + nfsfs= 1.841
mfsfs= =0.517
m 0.519 Figure 20: Definition of terms used on the
the transformed rs = 2000 kPa substitution equation for the calculation of the “true”
40 suction scale
air-entry value
Transformed x-scale
 = log10 ( )
20
Log 10 Soil suction
Following are the definitions of variables
0
used on the transformed suction plot for the
-1.0 0.0 1.0 2.0 3.0
Log10 Soil suction (kPa)
4.0 5.0 6.0
computation of the “true” air-entry value for
Figure 19: Degree of saturation versus Log10 soil the soil.
suction So = initial degree of saturation at the start of
the S-SWCC test;  = any x-coordinate along
The transformed scale for the S-SWCC the transformed scale; i = x-coordinate at the
can now be differentiated with respect to inflection point on the transformed scale (i.e.,
transformed suction to find the point transformed soil suction); SS(i) = degree of
corresponding to the maximum slope on the saturation at the inflection point; SS’(i) =
degree of saturation graph. The first first derivative of the transformed equation at
derivative can be obtained using the inflection point, and TL(i) = equation for
Mathematica or other comparable software the line of tangency passing through the
(e.g., Mathcad or MATLAB). It is also inflection point.
possible to use Mathematica to write the
second derivative of equation [13]. The 3.8.1 Derivation of the “true” Air-Entry
second derivative can be set to zero, giving Value Equation
rise to the transformed suction value
corresponding to the inflection point on the The “true” air-entry value, AEV, corresponds
degree of saturation function. to the intersection point between the
The degree of saturation at the inflection horizontal line through the initial degree of
point can be computed by inserting the saturation and the line of tangency through
transformed suction at the inflection point the inflection point (Zhang et al., 2015). The
into equation [13]. The coordinates of the line of tangency through the inflection point
inflection point are i for the transformed is designated as TL(), and can be written as
suction and SS(i) for the degree of follows:
saturation. A tangent line can be drawn

20
 S0 − SS (i )
TL ( ) = SS ' (i )( − i ) + SS (i ) i +
SS '(i )
[14]  aev = 10 [19]
The variable TL(), can be set to the
Equation [19] can be used to compute the
initial degree of saturation in order to
“true” air-entry value of the soil using: i.)
calculate the air-entry value. Equation [14] is
coordinates of the inflection point on the
then solved for the suction on the
transformed suction scale plot, ii.) line of
transformed suction scale equal to the
tangency through the inflection point, and
designated degree of saturation (i.e., the  iii.) degree of saturation at the start of the
value corresponding to the air-entry value, SWCC test. The end result of the above
AEV). The derivation steps involved in derivation is an empirical procedure that
going from the line of tangency to the provides a way to calculate a unique value for
determination of the air-entry value are as the “true” air-entry value of a soil. For the
follows: artificial clayey silt soil, the “true” air-entry
value is computed to be 147 kPa. The “true”
S0 = SS ' (i )  aev − SS ' (i ) i + SS (i ) air-entry value is used in the integration
[15]
process when calculating the permeability
where: aev = x-coordinate or the air-entry function.
value on the transformed suction scale at the The information presented thus far is
intersection of the tangency line through the related to the preparation of the w-SWCC and
inflection point and a line through the initial the SC data for calculating the unsaturated
degree of saturation. soil property functions required for
Equation [15] can be solved for the air- undertaking unsaturated soil mechanics
entry value on the transformed suction scale, simulations; in this case, saturated-
aev. unsaturated seepage modeling. The
permeability function must be calculated
SS ' (i ) i + SS ( 0 ) − SS (i ) when performing a steady-state seepage
 aev =
SS ' (i ) [16] analysis and the water storage function must
also be calculated when solving an unsteady-
where: SS(0) = degree of saturation at the state or transient seepage problems.
start of the SWCC test (i.e., equal to So). 3.9 Application of the SWCC for
Equation [16] can be rearranged to the Determination of USPFs
following form.
Various forms of the volume-mass soil-water
S0 − SS (i ) characteristic curves are ready to be used for
 aev = i +
SS ' (i ) [17] the estimation of the unsaturated soil
property functions, USPFs, for saturated-
Equation [17] can now be converted from unsaturated seepage problems. The partial
the transformed suction scale to the original differential equation accounting for two-
soil suction scale using transform equation dimensional unsaturated seepage can be
[12] (i.e.,  = Log10()). derived in a manner similar to the procedure
historically used in saturated soil mechanics.
S0 − SS (i ) Let us assume that the major and minor
log10 ( aev ) = i + coefficients of permeability occur in the x-
SS ' (i ) [18]
and y-directions, respectively. The saturated-
Therefore, unsaturated transient seepage equation can be

21
written as follows (Fredlund and Rahardjo, of this paper is to describe in detail the
1993). estimation procedures that appear to be most
acceptable when solving geotechnical
 hw    hw  hw engineering problems.
 +  kwy  = m2  w g
w
 kwx [20]
x  x  y  y  t The permeability function and the water
storage function should be calculated as
where: kwx and kwy = major and minor independent soil property functions rather
permeability functions; m2w = water storage than having the two soil properties combined
function; hw = hydraulic head in the water and used as a single “diffusivity” type
phase, (i.e., elevation head plus water variable. It is also important that the
pressure head); g = acceleration due to estimation of the unsaturated soil property
gravity, and t = time. functions be based on w-SWCC and SC test
If the soil is anisotropic, the saturated results on the same soil. Both soil property
coefficients of permeability are different in functions are highly nonlinear and dependent
two orthogonal directions. Let us assume that on different volume-mass versus soil suction
the major and minor coefficients of relations. This methodology is particularly
permeability occur in the x- and y-directions. important when the soil undergoes volume
Anisotropic soils are commonly assumed to change as soil suction changes.
have the same air-entry values in both The degree of saturation and void ratio
directions and as a result the unsaturated are the two main volume-mass variables that
permeability functions will have the same influence the estimation of the permeability
functional characteristics in the x- and y- function, whereas in saturated soil
directions. Expanding equation [20] and mechanics, void ratio is the only factor that
assuming that kwx is equal to kwy (i.e., equal to influences the coefficient of permeability.
kw) results in the following form for Techniques used in soil physics for the
saturated-unsaturated transient seepage. estimation of the unsaturated permeability
function are most commonly based on the
 2 hw k w hw 2h k h h
kw + + k w 2w + w w = m2w  w g w [21] assumption that the soil structure is rigid and
x 2
x x y y y t
therefore no volume change occurs during
the drying process (Zhang et al., 2015). Only
Equations [20] and [21] show that two
independent soil property functions are changes in the degree of saturation are
assumed to result in changes in the
required in order to solve transient seepage
unsaturated coefficients of permeability.
problems common to geotechnical
Several estimation procedures have been
engineering. The required soil properties are
proposed for estimating the unsaturated
the coefficients of permeability function, kw,
coefficient of permeability functions, all
and the water storage function, m2w. Both soil
based on the assumption that the soil does not
properties are nonlinear functions of suction.
undergo volume change during the drying
It has become generally accepted engineering
process.
practice (as well as in other related
engineering and agricultural disciplines), to 3.9.1 The Water Storage Function
estimate the unsaturated soil property The water storage property, m2w, is defined as
functions on the basis of the volume-mass the slope of the (instantaneous) volumetric
SWCCs and the saturated soil properties. At water content versus soil suction relationship.
the same time, there has been some variance The water storage function is required
with respect to how the estimation whenever an unsteady-state seepage analysis
procedures are applied. One of the objectives is performed. The water storage modulus,

22
m2w, can be obtained through the separation of changes in the coefficient of
differentiation of any equation that fits the permeability resulting from volume change
volumetric water content versus suction (or void ratio change) from changes that
relationship, (w-SWCC) (Figure 21). occur as a result of changes in degree of
saturation. In each case, there are existing
d w d w theories that can be used to estimate the two
m2w = or m2w = [22]
d ( ua − u w ) d coefficient of permeability functions.
where (ua – uw) = matric suction, and  = The term, “relative permeability
total suction in high suction range (i.e., function” is used when referring to the effect
suctions > 1500 kPa) and matric suction in of changes in degree of saturation in a
low suction range (i.e., suctions < 1500 kPa). normalized manner. The “relative
permeability function” is set equal to 1.0 at
0.0010
the air-entry value for the soil (i.e., krw(S)).
d w The “relative permeability function” can also
0.0008
m2w =
d be defined in terms of soil suction, krw(),
Water storage, d/d

0.0006
Largest water storage estimated from the degree of saturation
0.0004
at inflection point on SWCC.
volumetric water

k w ( )
content SWCC
krw ( ) =
0.0002

0.0000
kref ( ) [23]
where kref()
0.1 1 10 100 1000 10000 100000 1000000
Soil suction (kPa) = reference saturated
Figure 21 Water storage function for the artificial coefficient of permeability as a function of
clayey silt soil
soil suction. In other words, kref() is the
The water storage of a soil is relatively saturated coefficient of permeability
small as soil suction tends towards zero, corresponding to the void ratio at a
becoming equal to the coefficient of volume designated soil suction; kw() = coefficient of
change, mv, for the saturated soil (i.e., relative permeability as a function of soil suction;
to a change in effective stress). A maximum krw() = relative coefficient of permeability
water storage value is reached in the vicinity for changes in the degree of saturation.
of the inflection point along the w-SWCC, The term “reference saturated
then tending towards a low value beyond permeability function” is used when referring
residual suction conditions. It is preferable to the effect of changes in void ratio or
for the water storage function and the overall volume change, (i.e., kref()). The
permeability function to be applied as reference saturated permeability function is
independent mathematical relations for developed based on a saturated permeability
numerical modeling purposes because of function along with the relationship of void
their uniquely different nonlinear ratio to soil suction. The saturated
characteristics. permeability function with respect to void
ratio change, ksw(e), can be studied in a
3.10 The Permeability Function
dimensionless manner. The “dimensionless
A revised methodology is required for the saturated permeability function”, ksd(), can
estimation of the coefficient of permeability be written as follows.
function when soils undergo volume change
as suction changes during the drying process. k sw ( e )
The estimation procedure for calculating the k sd ( e ) =
k sr [24]
permeability function is based on the

23
 C e3
where: ksw(e) = saturated coefficient of k sw ( e ) =
1+ e [25]
permeability as a function of void ratio, and
where: C = constant representing all factors
ksr = saturated coefficient of permeability at
(other than void ratio) affecting the
the reference state (i.e., a reference void
calculation of the saturated coefficient of
ratio).
permeability.
It is possible to calculate the “actual
The constant, C, can be used as a fitting
permeability functions” once the reference
parameter provided the void ratio of the soil
saturated permeability function for volume
is known. The coefficient of permeability can
change and the relative permeability function
be measured using a one-dimensional
for degree of saturation change are known.
consolidation test or a permeameter test. The
The two permeability functions can be
Kozeny-Carman relationship is used to
combined for solving practical seepage
quantify the changes in the coefficients of
problems.
permeability over the range of possible void
3.10.1 Estimation of Coefficient of ratio changes. Incorporating Eq. [24] and Eq.
Permeability Function with respect to [25], coefficients of permeability due to
Void Ratio Changes increases or decreases in void ratio are then
Estimation models for the saturated estimated as a proportionality as shown in the
coefficient of permeability of a porous following equation.
material are mainly dependent upon the size ksw ( e )  e3   er3 
of the pores and the tortuosity of the flow k sd ( e ) = =    [26]
path (Chapius, 2012). These two factors can k sr 1+ e   1 + er 
be treated in an independent manner by
separating the effects of changes in void ratio where ksr = saturated coefficient of
from the effects of changes in degree of permeability at a known reference void ratio,
saturation. ksw = saturated coefficient of permeability at
Kozeny (1927) developed an estimation another selected void ratio of e; er = void ratio
model for the coefficient of permeability as a reference point at which the permeability
based on applying Poiseuille’s law to laminar is known, and e = void ratio at which the
flow through straight circular pipes. The permeability is to be calculated.
model was later modified by Carman (1937) A single permeability measurement can
and is generally referred to as the Kozeny- be used along with equation [26] to compute
Carman model. The model took several the “actual saturated coefficient of
factors into consideration in calculating the permeability” with respect to void ratio. The
coefficient of permeability; however, it is the dimensionless saturated permeability
form in which void ratio changes are function, ksd(e), has a value of 1.0 at its
characterized that is relevant to the reference state (i.e., the initial void ratio
development of a permeability estimation corresponding to saturation in the w-SWCC
model based on the soil-water characteristic test). In other words, ksd(er) = 1.0. Figure 15
curve, SWCC. The effect of changes in void shows the maximum void ratio (i.e., emax =
ratio is shown in equation [25] where all 0.871) and minimum void ratio (i.e., emin =
other factors represented in the Kozeny- 0.500) that form the limits for the artificial
Carman equation are treated as a single clayey silt soil being analyzed.
constant (Taylor, 1948). Figure 22 shows plots of the
“dimensionless saturated coefficient of
permeability” function along with the “actual

24
saturated coefficient of permeability” 1.0E+00

function for the artificial clayey silt soil being 1.0E-01

Normalized coefficient of permeability

Coefficient of permeability, m/s

analyzed. 1.0E-02
versus suction
Actual coefficient of permeability versus
The “actual saturated coefficient of 1.0E-03
suction

permeability” function can also be written as G = 2.680

Gs = 2.68
s water content = 31.5%
Initial
1.0E-04 e = 0.879
a function of soil suction through use of the Initial
o void ratio = 0.871
Initial degree of saturation = 96.9%
So = 96.08 % Permeability function
basic laboratory test results; namely, the w- 1.0E-05
for changing void
SWCC and the shrinkage curve data. The 1.0E-06 ratio with suction
“actual saturated coefficient of permeability” 1.0E-07
function (i.e., ksw(e)) written as a function of 0.1 1 10 100 1000 10000 100000 1000000
Soil suction (kPa)
soil suction can be referred to as the reference
Figure 23 Changes in the permeability function as a
saturated coefficient of permeability function function of soil suction due to void ratio changes
(i.e., kref()).
1.00 1.0E-06
The “reference saturated coefficient of
permeability” function for void ratio change
Normalized coefficient of permeability

Actual Coefficient of permeability, m/s

Using Kozeny- can be written in terms of soil suction, kref(),
Carman as shown in equation [27].
approximation
Normalized coefficient of
0.10 permeability versus void ratio 1.0E-07 x
 1

   csh 
csh
Actual coefficient of 
permeability versus void ratio
   ws 1 − ln (1 +   r ) ln (1 + 10  r ) 
C  ash  
6
(  + 1 
  )
( ( ))
G
Gss= = 2.680
   ( )  
mf
2.68
( ) + 
nf

Initial water content = 31.5%
e = 0.879   
b ln exp 1 a   

sh f
o void ratio = 0.871
Initial  
S
Initial=degree
o 96.08 % = 96.9%
of saturation
k ws ( ) =  1
 [27]
0.01 1.0E-08  
csh
 sh
c

0 0.2 0.4 0.6 0.8 1   w 1 − ln (1 +   ) ln (1 + 106  ) 


1 + ash 
s r (r
 + 1
 )
( ( ))
Void ratio, e
 
bsh ln exp (1) + ( a f )
mf
nf

Figure 22 Dimensionless and actual saturated   
 
 
coefficients of permeability versus void ratio
Kozeny-Carman equation has been found
Figure 23 shows a plot of the reference to be better suited for sand soils than for clay
coefficient of permeability versus soil soils (Taylor, 1948). However, it should be
suction. The upper line is the reference noted that only an approximation of the effect
coefficient of permeability function in a of void ratio changes is required in most
dimensionless form reflecting the effect of cases because the major changes in the
void ratio change. The lower plot in coefficient of permeability are due to changes
Figure 23 shows the reference saturated in the degree of saturation. The following
permeability function assuming that the Somogyi (1980) equation can also be better
saturated coefficient of permeability used for suited for computing the effect of void ratio
non-dimensionalization was 1.0 x 10-6 m/s at changes on the saturated coefficient of
zero suction. The lower permeability permeability.
function shows how the permeability
function can be scaled up or down depending
on the saturated coefficient of permeability 𝑘𝑠𝑤 (𝑒) = 𝐴𝑒 𝐵
[28]
used for non-dimensionalization.
The relationship between void ratio and where: ksw(e) = saturated coefficient of
soil suction is mathematically described in permeability, A = one of the fitting
equation [6]. parameters for the void ratio versus
coefficient of permeability measurements,
and B = second exponential fitting parameter

25
for the void ratio versus coefficient of permeability function took the following
permeability measurements. form when using the degree of saturation
The use of the Somogyi (1980) equation SWCC.
requires that there be a series of
measurements of void ratio versus coefficient b
S ( e y ) − S ( )
 S ' ( e y ) dy
of permeability, possibly from a one- ln ( ) ey
krw ( ) =
S ( e y ) − S ( aev )
dimensional laboratory consolidation test. b [29]
 S ' ( e y ) dy
3.10.2 Estimation of coefficient of ln (
 ) e y
aev
permeability function with respect to
where: b = upper limit of integration (i.e.,
changes in degree of saturation
ln (1,000,000 ) ), y = dummy variable of
The permeability function that takes changes integration representing the natural logarithm
in degree of saturation into consideration can of suction, S’ = first derivative of the soil-
be formulated independent of void ratio water characteristic curve equation, and e y =
changes and referred to as the relative natural number raised to the dummy variable
coefficient of permeability function. Various power.
forms of integration along the S-SWCC have Equation [29] shows that integration
been used for the estimation of the coefficient commences at the air-entry value and
of permeability function with respect to continues to an upper limit of 1,000,000 kPa.
changes in the degree of saturation. The The relative coefficient of permeability,
individual permeability functions taking void krw(), is the ratio of the unsaturated
ratio changes and degree of saturation coefficient of permeability at soil suctions in
changes into consideration can then be excess of the air-entry value for the soil to the
combined for solving seepage problems reference saturated coefficient of
where the soil undergoes volume change and
permeability, kref(), as shown in Equation
desaturation as drying occurs.
[23]. The reference saturated coefficient of
Childs and Collis-George (1950)
permeability, kref(), corresponds to the
proposed a model for estimating the
relevant void ratio at the soil suction under
coefficient of permeability based on a
consideration.
random variation in pore sizes. The
The integration along the SWCC should
permeability function was derived based on
take place from the air-entry value of the soil
Poiseuille’s equation and the assumption was
to at least residual water content conditions.
made that the overall volume change of the
It is commonly assumed that the coefficient
soil was negligible as soil suction increased.
of permeability of a soil is essentially zero
The model was later improved by Marshall
when its water content is below the residual
(1958) and further modified by Kunze et al.,
water content. Kunze et al., (1968) concluded
(1968).
that the accuracy of the prediction is
Fredlund, Xing and Huang (1994) used
significantly improved when the soil-water
the Fredlund and Xing (1994) soil-water
characteristic curve extended at least to
characteristic curve equation along with the
residual conditions. Fredlund et al., (1994)
Childs and Collis-George (1950) physical
continued the integration process beyond
model to compute a water permeability
residual condition to near zero water content.
function. The procedure involved starting at
The authors were aware that there was no
saturated soil conditions and integrating
confirmation of the accuracy of the lower
numerically along the volumetric water
portion of the permeability function.
content SWCC to completely dry conditions.
However, the intent was to provide a
The Fredlund et al., (1994) relative

26
continuous permeability function over the procedure consists of a series of discrete data
entire possible suction range. The assumption points that can be best-fit using the Fredlund
was made that liquid flow of water would and Xing (1994) SWCC equation or the
tend towards zero as vapor flow commenced Gardner (1958) equation. In so doing, the
at some point near residual conditions. The permeability function becomes a closed-
relationship between liquid and vapor flow is form, continuous function.
dealt with later in this manuscript. To avoid
numerical difficulties, integration was 3.10.3 Combining the Effects of Volume
performed over the soil suction range from Change and Degree of Saturation
6
 aev to 10 kPa on an arithmetic scale. The
Change on the Permeability
Functions
“relative permeability function” for the
artificial clayey silt soil data used in this The overall permeability function for a soil is
paper are presented in Figure 24. the product of the coefficient of permeability
with respect to volume change (i.e., equation
1.0E+00
[27]) and the coefficient of permeability with
1.0E-01
respect to changes in degree of saturation
Relative coefficient of permeability,kr

1.0E-02
AEV
1.0E-03 (i.e., equations [29]).
1.0E-04

kw ( ) = krw ( )  kref ( )
1.0E-05
1.0E-06 Relative coefficient of permeability of
1.0E-07 drying curve [30]
1.0E-08 where kw() = coefficient of permeability at
1.0E-09
Integration along S-SWCC starting at AEV a particular suction , krw() = relative
1.0E-10
1.0E-11
coefficient of permeability as the soil
0.1 1 10 100 1000 10000 100000 1000000 desaturated at suctions beyond the air-entry
Soil suction, ψ (kPa)
value, and kref() = reference saturated
Figure 24 Permeability functions for the artificial coefficient of permeability.
clayey silt as soil suction is increased beyond “true” The reference saturated coefficient of the
air-entry value of the soil.
permeability, kref() refers to the coefficient
The “relative permeability” function of permeability of a saturated soil at a
starts at 1.0 corresponding to suctions up to particular void ratio when the overall porous
the “true” air-entry value. The logarithm of skeleton subjected to a suction of . The
the coefficient of permeability then decreases relative coefficient of permeability is 1.0 at
almost linearly with the logarithm of soil the saturated state. When soil suction is less
suction beyond the air-entry value. than the air-entry value, the soil remains
The relative permeability function saturated and changes in permeability are
calculated using the Fredlund et al., (1994) related to changes in void ratio.
integration procedure is shown as a series of When suction exceeds the air-entry value,
data points corresponding to the suctions at desaturation starts and the relative coefficient
which the coefficient of permeability was of permeability decreases from 1.0 to a value
calculated. The lower limit for the coefficient approaching zero as the soil dries. The
of permeability function can be set to either coefficient of permeability kw() for soil in
2.0 x 10-14 m/s or a coefficient of an unsaturated state is always smaller than
permeability value corresponding to a soil the reference saturated coefficient of
suction of 10,000 kPa, whichever is larger. permeability. The overall coefficient of
The permeability function calculated permeability kw() is equal to the product of
using the Fredlund et al., (1994) integration the reference permeability associated with

27
desaturation and the saturated coefficient of flow models gave similar flow patterns when
permeability at the suction corresponding to plotted versus soil suction for various soil
a particular void ratio. types (Figure 25). The results showed that the
Changes in the degree of saturation and maximum vapor flow occur near the residual
void ratio of a soil are the two main factors state of the soil and remains essentially
that result in changes in the coefficient of constant up to and beyond a suction of 10,000
permeability for a soil. Equation [30] has two kPa.
components; namely the degree of saturation
effect on the relative coefficient of
permeability, krw() and the void ratio effect
on the reference saturated coefficient of
permeability kref(). Decreases in the degree
of saturation significantly change the
tortuosity of the flow path within the porous
media, and as a result, the coefficient of
permeability is changed. In an unsaturated
soil, the degree of saturation and void ratio
combine to govern the overall coefficient of
permeability for a soil that undergoes volume Figure 25. Vapor permeability for three soil types
change as soil suction changes during a (after Lai et al, 1976).
drying process.
The average vapor permeability for the
3.10.4 Lower Limit for the Liquid
three soil types shown was 2.0 x 10-14 m/s. It
Coefficient of Permeability
is suggested that the vapor permeability be
Laboratory measurements and proposed used as a lower limit for the water coefficient
permeability models show that the of permeability. The water coefficient of
coefficient of permeability decreases permeability function then becomes a
logarithmically as soil suction increases continuous function over the entire suction
beyond the air-entry value. It is also range. The lack of a lower limit for water
suggested that at some point the water coefficient of permeability can give rise to
coefficient of permeability becomes so small numerical convergence issues when
that more moisture can be moved in the vapor modeling saturated-unsaturated seepage
phase than in the liquid phase (Ebrahimi- problems.
Birang et al., 2004). Tran et al., (2014) It is also possible that the permeability
showed that there was a “shut-off” suction function may not reach 2.0 x 10-14 m/s before
where the liquid flow of water essentially soil suction reaches 10,000 kPa. In this case,
ceases. it is suggested that the permeability at a
There is limited research on the transition suction of 10,000 kPa be used as the lower
point where liquid and vapor flow becomes limit of permeability.
essentially equal. Ebrahimi-Birang et al.,
4 HYSTERESIS ASSOCIATED WITH
(2004), suggested that there should be a
DRYING AND WETTING
lower limit for the water coefficient of
permeability and that the lower limit should Figure 10 showed that soils have hysteresis
be related to the vapor diffusion value. Lai et with the main drying and wetting curves
al., (1976) summarized various tortuosity forming boundaries for the water content
models associated with vapor flow. All vapor versus soil suction relationship. However, it
is only the drying S-SWCC (or the desorption
28
curve) that has been generally used to curve and an estimate is made regarding the
estimate the permeability function. Another magnitude of the hysteresis loop.
permeability function can be calculated The fitting parameters for the drying
corresponding to the wetting (or adsorption) curve are the same as for the wetting curve
S-SWCC. with the exception that the afs fitting
Pham (2002, 2005) analyzed the drying parameter must be reduced by a magnitude
and wetting curves for 34 datasets for a dependent upon the size of the hysteresis
variety of soils reported in the literature. The loop. The afs fitting parameter for the wetting
difference between the hysteresis loops at the curve can be calculated based on equation
inflection points was used as the primary [31] (Fredlund et al., 2011).
indicator of the magnitude of the hysteresis
loop. The laboratory measurements of the  = 100 ( log ( ad ) − log ( aw ) )
drying and wetting SWCCs showed that the [31]
bounding drying curve tended to be
approximately congruent, (i.e., parallel on a where:  = percent shift between the drying
semi-log plot), to the bounding wetting curve and wetting hysteresis loops, ad = suction
(Pham et al., 2003, 2005). The distance corresponding to the afs fitting parameter on
between the main drying and wetting curves the drying SWCC, and aw = suction
varied between 0.15 and 0.35 of a log cycle corresponding to the afs fitting parameter on
for sands (i.e., 15 to 35% of a log cycle). The the wetting SWCC.
bounding curve spacing for well-graded Equation [31] can be rearranged and
clayey silt soils varied between 0.35 and 0.60 solved for the suction on the wetting curve
of a log cycle. On average, the approximate that corresponds to the afs fitting parameter.
spacing between the drying and wetting
SWCCs was about 25% of a log cycle for log ( aw ) = log ( ad ) −  100 [32]
sands and 50% of a log cycle for well-graded
clayey silt. The overall average shift between Let us make the assumption that the shift
the drying and wetting bounding curves was between the drying and wetting SWCCs is
approximately 35%. 35%. The afs fitting parameter changes from
Pham (2002) measured the drying and 261.9 kPa for the drying curve to 117.0 kPa
wetting bounding curves for a sand soil and a for the wetting curve for the arbitrary soil
processed silt soil. Results for the processed being analyzed in this paper. The other fitting
silt soil were shown in Figure 10. A total of parameters remain the same and the wetting
three specimens were tested with each test SWCC can be calculated and plotted as
Hysteresis of the Degree of Saturation SWCC
showing essentially the same drying and shown in Figure 26.
wetting SWCCs. The drying curves were 100

measured up to residual suction conditions

Drying S-SWCC
Degree of saturation (%)

80 Drying S-SWCC
and showed wetting curves that were
essentially congruent with respect to the adad== 275.0
262 kPa
kPa
60 aw = 122.8 kPa
drying SWCC. It is not always practical to aw = 117 kPa
Wetting Wetting
S-SWCC S-SWCC

measure both the drying and the wetting 40

SWCCs when solving practical engineering 20

Wetting hysteresis loop
seepage problems. However, it would appear shifted by 35%

to be reasonable to estimate the wetting 0

0.1 1 10 100 1000 10000 100000 1000000
SWCC based on the assumption that the Soil suction (kPa)

drying curve is congruent with the wetting Figure 26 Drying and wetting hysteresis loops for the
January 10, 2017 49

artificial clayey silt soil

29
 The proposed protocols are meant to provide
The wetting SWCC can now be used to the geotechnical engineer with a thorough
compute a permeability function understanding of present theories for
corresponding to a wetting process. applying unsaturated soil mechanics in
It is recognized that the suggested engineering practice.
procedure for handling hysteresis is
approximate. The outlined procedure is
1.0E+00
meant to illustrate reasonable assumptions

Relative coefficient of permeability, kr

that can be made to accommodate the 1.0E-01

hysteretic behavior of soils. 1.0E-02

1.0E+00
1.0E-03
1.0E-01
Drying S-SWCC
Relative coefficient of permeability, kr

1.0E-02
1.0E-04
1.0E-03 Relative permeability for drying
1.0E-04 Wetting S-SWCC curve
1.0E-05
Relative permeability for wetting
1.0E-05
curve
1.0E-06 1.0E-06
1.0E-07 0 20 40 60 80 100
1.0E-08 Degree of saturation, S (%)
1.0E-09 Relative coefficient of permeability of drying curve Figure 28 Cross-plot of the drying and wetting
1.0E-10
1.0E-11
Relative coefficient of permeability of wetting curve relative permeability functions with the degree of
1.0E-12 saturation for a plot of relative permeability versus
0.1 1 10 100 1000 10000 100000 1000000
Soil suction, ψ (kPa)
degree of saturation for the artificial clayey silt soil
Figure 27 Drying and wetting permeability functions
for the artificial clayey silt soil The use of estimation procedures in
geotechnical engineering practice has found
Figure 27 shows the drying and wetting increasing acceptance over the past couple of
permeability functions for the artificial decades. The acceptance is mainly due to: i.)
clayey silt soil. Some computer codes can reduced costs associated with indirectly
take hysteresis effects into consideration by estimating the permeability function, and ii.)
using the appropriate drying and wetting realization that estimation techniques
permeability functions. Bashir et al., (2016) provide adequate information for most
suggested cross-plotting the drying and engineering design purposes. Success in
wetting relative permeability functions applying unsaturated soil mechanics in
versus the degree of saturation and thereby engineering practice has been closely related
producing essentially a unique permeability to the use of the soil-water characteristic
relationship that can be used in numerical curve for estimating the water coefficient of
modeling. The SWCC versus degree of permeability and the water storage functions.
saturation plot is unique as long as there is 5 CONCLUSIONS
congruency between the drying and wetting
SWCCs (Figure 28). The engineering protocols being used in
There are numerous assumptions that geotechnical engineering are closely tied to
have been made during the analysis of the the earlier research undertaken in soil
data from the measurement of the drying physics. The use of the pressure plate
curve soil-water characteristic curve and the apparatuses focused on measuring the drying
shrinkage curve. These assumptions are SWCC and the procedures and protocols
recognized and the intent is to provide the developed for engineering applications have
geotechnical engineer with the best possible been built on earlier findings in soil physics.
protocols for interpreting and applying data The early research in soil physics has been
associated with unsaturated soil behaviour. valuable in geotechnical engineering

30
applications but it has been necessary to surface hydrology for improved methods to
carefully review the assumptions associated estimate ground surface boundary
with use of the SWCCs. conditions, actual evaporation, runoff, and
Unsaturated soil mechanics can be infiltration. These topics are also of
applied in geotechnical engineering practice; importance to soil physicists and the ongoing
however, the manner in which it is applied sharing of information is important for
differs from saturated soil mechanics. The progress in both areas.
required soil properties for analysis purposes 7 ACKNOWLEDGEMENTS
take the form of nonlinear functions. The
costs associated with directly measuring the The authors wishes to acknowledge the
nonlinear functions in the laboratory are significant contribution of Feixia Zhang in
prohibitive. This paper has focused on the assembling the theoretical components
assessment of the material properties involved with this study as well as
associated with unsaturated seepage confirming the reliability of the regression
modeling. analyses. This paper reports on a small
In recent years there has been worldwide amount of the extensive confirmatory
research focused on the indirect laboratory testing that she has undertaken and
determination of the unsaturated soil reported in other research papers.
property functions through use of the soil-
water characteristic curve, SWCC. In 8 REFERENCES
particular, the drying SWCC can be used in Aitchison, G.D. (1961). Relationship of moisture and
conjunction with the measurement of the effective stress functions in unsaturated soils,
shrinkage curve, SC, to provide increased Pore Pressure and Suction in Soils Conference,
accuracy in the estimations of unsaturated Butterworths, London, England, pp. 47-52.
Aitchison, G.D. (1965) Moisture equilibria and
soil property functions. The use of the moisture changes in soils beneath covered areas,
shrinkage curve in conjunction with the A Symposium-in-Print, (Editor), Australia:
SWCC allows for the separation of effects of Butterworths,
volume change from the effects of changes in Barden, L. (1965). Consolidation of compacted and
degree of saturation of the soil. The SWCC unsaturated clays, Geotechnique, Vol. 15, No. 3,
pp. 267-286.
and SC data allow for a more rigorous Barden, L., Madedor, A.O. and Sides, G.R. (1969).
analysis of laboratory measurements for a Volume change characteristics of unsaturated
wide range of soil conditions encountered in clay, ASCE Journal of Soil Mechanics and
geotechnical engineering. Foundation Division, Vol. 95, SM1, pp. 33-52.
The concept of indirectly estimating Bashir, R., Sharma, J., and Stefaniuk, H. (2016).
Effect of hysteresis of soil-water characteristic
unsaturated soil property functions, (e.g., curve on infiltration under different climatic
coefficient of permeability and water conditions, Canadian Geotechnical Journal, Vol.
storage) has become acceptable as part of 53, pp. 273-284, dx.doi.org/10.1139/cgj-2015-
prudent geotechnical engineering practice. 0004.
Biot, M.A. (1941). General theory of three-
6 RECOMMENDATIONS dimensional consolidation, Journal of Applied
Physics, Vol. 12, No. 2, pp. 155-164.
There are numerous important issues that still Bishop, A.W. (1959). The principle of effective
require further research. Unsaturated soils are stress, (Invited Lecture in Oslo, Norway in 1955.
near to the ground surface and as such are Published in Teknisk Ukeblad, Vol. 106, No. 39,
heterogeneous and randomly cracked with pp. 859-863.
Bishop, A. W. and Blight, G.E. (1963). Some aspects
secondary structure. There is need for further
of effective stress in saturated and unsaturated
integration of geotechnical engineering with soils, Geotechnique, Vol. 13, No. 3, pp. 177-197.
other disciplines such as meteorology and

31
Bishop, A.W., Alpan, I., Blight, G.E., and Donald, Canadian Geotechnical Conference, Quebec,
I.B. (1960). Factors controlling the shear strength QC, Vol. 1, pp. 12-19.
of partly saturated soils, ASCE Research Edlefsen, N.E., and Anderson, A.B.C. (1943).
Conference on the Shear Strength of Cohesive Thermodynamics of soil moisture, Hilgardia,
Soils, University of Colorado, Boulder, pp. 503- Vol. 15, pp. 31-298.
532. Escario, V. (1980). Suction controlled penetration
Blight, G.E. (1961). Strength and consolidation and shear tests, Proceedings of the Fourth
characteristics of compacted soils, PhD Thesis, International Conference on Expansive Soils,
University of London, London. Denver, CO, ASCE, Vol.2, pp. 781-797.
Blight, G.E. (1965). A study of effective stress for Ebrahimi-Birang, N., Gitirana Jr., G.F.N., Fredlund,
volume change, In Moisture Equilibria and D.G., Fredlund, M.D., and Samarasekera, L.
Moisture Changes in Soils beneath Covered (2004). A lower limit for the water permeability
Areas, A Symposium-In-Print, G.D. Aitchison, coefficient, Proceedings of the Fifty-seventh
(Editor), Australia: Butterworths, pp. 259-269. Canadian Geotechnical Conference, Quebec,
Brooks, R.H. and Corey, A.T. (1964). Hydraulic QC, Vol. 1, pp. 12-19.
properties of porous media, Colorado State Fredlund, D.G. (2006). Unsaturated soil mechanics in
University, Fort Collins, CO, Hydrology Paper engineering practice, Terzaghi Lecture, Vol. 132,
No. 3, 27 p, (March). No. 3, pp. 286-321.
Buckman, H.O. and Brady, N.C. (1960). The nature Fredlund, D.G. (2015). Relationship between the
and properties of soils, 6th Edition, MacMillan, laboratory SWCC and the field stress state,
New York, N.Y. Proceedings of the Sixth Asia-Pacific Conference
Burdine, N.T. (1952). Relative permeability on Unsaturated Soils, Quilin, China, October, pp
calculations from pore-size distribution data, 23-26.
Transactions AIME. Fredlund, D.G. and Morgenstern, N.R. (1976).
Chapuis, R.P. (2012). Influence of element size in Constitutive relations for volume change in
numerical studies of seepage: Small-scale details, unsaturated soils, Canadian Geotechnical
Geotechnical News, 30(1); pp. 32-35. Journal, Vol. 13, No. 3, pp. 261-276.
Childs, E.C. and Collis-George, N. (1950). The Fredlund, D.G. and Morgenstern, N.R. (1977). Stress
permeability of porous materials, Proceedings of state variables for unsaturated soils, ASCE
the Royal Society, Vol. 201A, pp. 392-405. Journal of Geotechnical Engineering Division,
Coleman, J.D. (1962). Stress/strain relations for GT5, Vol. 103, pp. 447-466.
partly saturated soils, Geotechnique Fredlund, D.G., and Rahardjo, H., (1993). Soil
(Correspondence), Vol. 12, No. 4, pp. 348-350. mechanics for unsaturated soils, John Wiley &
Croney, D. (1952). The movement and distribution of Sons, New York, N.Y.
water in soils, Geotechnique, Vol. 3, pp. 1-16. Fredlund, D.G., Rahardjo, H., and Fredlund, M.D.
Croney, D., Coleman, J.D., and Black, W.P.M. (2012). Unsaturated soil mechanics in
(1958). Movement and distribution of water in engineering practice, John Wiley & Sons, New
soil in relation to highway design and York, N.Y.
performance, In Water and Its Conduction in Fredlund, D.G., Sheng, D., and Zhao, J. (2011).
Soils, Highway Research Board, Special Report, Estimation of soil suction from the soil-water
Washington, DC, No. 40, pp. 226-252. characteristic curve, Canadian Geotechnical
Dakshanamurthy, V., Fredlund, D.G., and Rahardjo, Journal, Vol. 48, No. 2, pp. 186-198.
H. (1984). Coupled three-dimensional Fredlund, D.G., and Xing, A. (1994). Equations for
consolidation theory of unsaturated porous media, the soil-water characteristic curve, Canadian
Proceedings of the Fifth International Geotechnical Journal, Vol. 31, No. 3, pp. 521-
Conference on Expansive Soils, Adelaide, South 532.
Australia, Institute of Engineers, Australia, pp. Fredlund, D.G., Xing, A., and Huang, S.Y. (1994).
99-103. Predicting the permeability function for
Decagon Services, Inc. (2009). The WP4-T chilled- unsaturated soils using the soil-water
mirror device, Promotional material on the WP4- characteristic curve, Canadian Geotechnical
T, Pullman, WA, USA. Journal, Vol. 31, No.4, pp. 533-546.
Ebrahimi-Birang, N., Gitirana Jr., G.F.N., Fredlund, Fredlund, D. G., and Zhang, F. (2013). Combination
D.G., Fredlund, M.D., and Samarasekera, L. of shrinkage curve and soil-water characteristic
(2004). A lower limit for the water permeability curves for soils that undergo volume change as
coefficient, Proceedings of the Fifty-seventh soil suction is increased. Proceedings of the 18th
International Conference on Soil Mechanics and

32
 Geotechnical Engineering, Paris, France, Sept 2- Klute, A. (1986). Water Retention: Laboratory
6. Methods, In Methods of Soil Analysis, Part 1 –
Fredlund, M.D. (1997). Design of a knowledge-based Physical and Mineralogical Methods, A. Klute,
system for unsaturated soil properties, M.Sc., Editor, American Society of Agronomy, Madison,
thesis, University of Saskatchewan, Saskatoon, WI, pp. 635-662.
SK. Kunze, R.J., Uehara, G., and Graham, K. (1968).
Fredlund, M.D. (2002). The role of unsaturated soil Factors important in the calculation of hydraulic
property functions in the practice of unsaturated conductivity, Soil Science Society of America
soil mechanics, PhD, thesis, University of Proceedings, Vol. 32, pp. 760-765.
Saskatchewan, Saskatoon, SK., 292 p. Lai, S.H., Tiedje, J.M., and Erickson, A.E. (1976). In
Fredlund, M.D., Fredlund, D.G., and Wilson, G.W. situ measurement of gas diffusion coefficient in
(2002). Use of the grain-size distribution for the soils, Soil Science Society of America Journal,
estimation of the soil-water characteristic curve, Vol. 40, pp. 3-6.
Canadian Geotechnical Journal, Vol. 39, No. 5, Liu, X., and Buzzi, O. (2014). Use of hand-spray
pp. 1103-1117. plaster as a coating for soil bulk volume
Fredlund, M.D., Wilson, G.W., and Fredlund, D.G. measurement, ASTM Geotechnical Testing
(2002). Representation and estimation of the Journal, Vol. 37, No. 3, doi:
shrinkage curve, Proceedings of the Third 10.1520/GTJ20130091, May.
International Conference on Unsaturated Soils, Liu, X., Buzzi, O., Yuan, S., Mendes, J., and Fityus,
UNSAT 2002, Recife, Brazil, pp. 145-149. S. (2016). Multi-scale characterization of the
Freeze, R.A. and Cherry, J.A. (1979). Groundwater, retention and shrinkage behaviour of four
Englewood Cliffs, NJ: Prentice-Hall, 604p. Australian clayey soils. Canadian Geotechnical
Gardner, W.R. 1958. Some steady state solutions of Journal, doi: 10.1139/cgj-2015-0145.
the unsaturated moisture flow equation with Lytton, R.L. and Woodburn, J.A. (1973). Design and
application to evaporation from a water-table, Performance of Mat Foundations on Expansive
Soil Science Journal, Vol. 85, No. 4. pp. 228-232. Soils, Proceedings of the International
Gardner, W.R. (1961). Soil Suction and Water Conference on Expansive Soils, Haifa, Israel, Vol.
Movement, Pore Pressure and Suction in Soils, 1, pp. 301-307.
London: Butterworths, pp. 137-140. Mualem, Y. (1976). A new model for predicting
GCTS (2011) Operating Instructions for SWCC hydraulic conductivity of unsaturated porous
Device SWC-150, Version 1.1, Tempe, AZ. media, Water Resources Research, Vol. 12, pp.
Haines, W.B. (1927). A Further Contribution to the 513-522.
Theory of Capillary Phenomena in Soils, Journal Mualem, Y. (1976). Hysteretical models for
of Agricultural Science, Vol. 17, pp. 264-290. prediction of the hydraulic conductivity of
Ho, D.Y.F. and Fredlund, D.G. (1982). A Multi- unsaturated porous media, Water Resources
Stage Triaxial Test for Unsaturated Soils, ASTM Research, Vol. 12, No. 6, pp. 1248-1254.
Geotechnical Testing Journal, Vol. 5, No. ½, pp. Pham, H.Q. (2002). An engineering model of
18-25. hysteresis for the soil-water characteristic curve,
Jennings, J.E. (1969). The Prediction of Amount and M.Sc. Thesis, University of Saskatchewan,
Rate of Heave Likely to be Experienced in Saskatoon, SK, 183 p.
Engineering Construction on Expansive Soils, Pham, H.Q. (2005). A volume-mass constitutive
Proceedings of the Second International model for unsaturated soils, PhD Thesis,
Conference on Expansive Soils, Texas A & M, University of Saskatchewan, Saskatoon, SK.
College Station, pp. 99-109. Pham, H.Q., Fredlund, D.G., and Barbour, S.L.
Jennings, J.E. and Knight, K. (1957). The Prediction (2003). A practical hysteresis model for the soil-
of Total Heave from the Double Oedometer Test, water characteristic curve for soils with negligible
Proceedings of the Symposium on Expansive volume change, Geotechnique, Vol. 53, No. 2, pp.
Clays, South African Institute of Civil Engineers, 293-298.
Johannesburg, Vol. 7, No. 9, pp. 13-19. Pham, H.Q., Fredlund, D.G., and Barbour, S.L. 2005.
Klute, A. (1965). Laboratory Measurement of A study on the hysteresis models for soil-water
Hydraulic Conductivity of Unsaturated Soil, In characteristic curve, Canadian Geotechnical
Methods of Soil Analysis, C.A. Black, D.D. Journal Vol. 42, No. 6, pp. 1548-1568.
Evans, J.L. White, L.E. Ensminger and F.E. Clark Pham, H. Q., and Fredlund, D.G. (2008). Equation for
(Editors), Monograph 9, Part 1, American Society the entire soil-water characteristic curve on
of Agronomy, Madison, WI, pp. 253-261. volume change soils, Canadian Geotechnical
Journal, Vol. 45, pp. 443-453.

33
 Matyas, E.L., and Radhakrishna, H.S. (1968). increased. Proceedings of AP-UNSAT 2015,
Volume change characteristics of partially Guilin, China, Oct 23 –26, pp. 383-392.
saturated soils, Geotechnique, Vol. 18, No. 4, pp.
432-448.
Richards, L.A. (1931). Capillary conduction of
liquids through porous medium, Journal of
Physics, Vol. 1, pp. 318-333.
Soil Moisture Equipment Corp. (1985). Commercial
Publications. P.O. Box 30025, Santa Barbara,
CA.
Somogyi, F., (1980). Large strain consolidation of
fine-grained slurries, Proceedings of the
Canadian Society for Civil Engineering Annual
Conference, Winnipeg, MB.
Taylor, D. (1948). Fundamentals of Soil Mechanics,
John Wiley & Sons, New York, N.Y.
Terzaghi, K. (1943). Theoretical Soil Mechanics,
John Wiley & Sons, New York, N.Y., 510 p.
Topp, G.C. and Miller, E.E. (1966). Hysteretic
moisture characteristics and hydraulic
conductivities for glass-bead media, Proceedings
of Soil Science of America, Vol. 30, pp. 156-162.
Tran, D.T.Q., D.G., Chan, D.H. and Fredlund,
(2014). Assessment of soil suction at evaporation-
rate reduction point for saturated-unsaturated soil
surfaces, Proceedings of the Geo-Congress on
Geo-Characterization and Modeling for
Sustainability, Atlanta, Georgia, February 23-26.
van Genuchten, M.T. (1980). A closed-form
equation for predicting the hydraulic
conductivity of unsaturated soils, Journal of Soil
Science Society of America, Vol. 44, pp. 892-
898.
Wilson, G.W., Fredlund, D.G. and Barbour, S.L.
(1994). Coupled Soil-Atmospheric Modeling for
Soil Evaporation. Canadian Geotechnical
Journal, Vol. 31, No. 2, pp. 151-161.
Zapata, C.E. (1999). Uncertainty in soil-water
characteristic curves and impact on unsaturated
shear strength predictions. Ph.D. Dissertation,
Arizona State University, Tempe, AZ.
Zhang, F., and Fredlund, D. G. (2015). Examination
of the estimation of relative permeability for
unsaturated soils. Canadian Geotechnical
Journal. 52 (12): 2077-2087, 10.1139/cgj-2015-
0043.
Zhang, F., Fredlund, D. G., Fredlund, M. D, and
Wilson, G. W. (2015). Role of air-entry value
and choice of SWCC in the prediction of the
unsaturated permeability. Proceedings of the 68th
Canadian Geotechnical Conference, Quebec
City, September 20-23.
Zhang, F., Fredlund, D. G., and Wilson, G. W.
(2015). Hydraulic properties for soils that
undergo volume change as soil suction is

34

View publication stats