Nature and Phase Properties of Unsaturated Soil - Unsaturated Soil Mechanics in Engineering Practice - page 34

a calcium montmorillonite. Buildings founded on shallow
footings have been known to experience 50-150 mm of
movement in the years subsequent to construction.
The above soil profile is typical of conditions which are
found in many parts of the world. In each case, the natural
water contents are relatively low and the pore-water pres-
sures are highly negative. Throughout each season, and from
season to season, the soil expands and contracts in response
to changes in the net moisture flux at ground surface. It is the
climatic and human-imposed surface flux conditions at the
ground surface that strongly affect the potential expansion
and contraction of the underlying soil.

of the voids (i.e., the inverse of the distribution of the solid
particles) that forms the basis for numerous estimations tech-
niques that have been proposed for estimating the SWCC for
an unsaturated soil (M.D. Fredlund et al., 1997a).
Information from the grain-size distribution curve has
been used for the estimation of the SWCC of a soil,
as shown in Fig. 2.5. The grain-size distribution curve
provides an indication of the air-entry value and the rate of
desaturation of the soil. These estimation procedures are
referred to as pedo-transfer functions. Details pertaining to
several pedo-transfer estimation techniques for the SWCC
are presented later in Chapter 5.

2.2.2 Equation for Grain-Size Distribution Curve
A mathematical equation to represent the entire range of
measured grain-size particles provides a basis for further
analyses. A grain-size distribution equation provides a flexi-
ble means of searching databases for similar soils. The grain-
size equation provides a continuous mathematical function
that can be used for estimating the SWCC. Most grain-size
distribution curves have a unimodal distribution of particle
sizes while other soils have a bimodal distribution of parti-
cle sizes. Both types of distributions can be described by a
mathematical equation.
Gardner (1956) used a two-parameter, lognormal distri-
bution to fit grain-size distribution data. The two-parameter
fit of the grain-size distribution was performed using a geo-
metric mean parameter, x g , and a geometric standard devi-
ation, σ g . The method of fitting lognormal equations to the
grain-size distribution has some limitations in engineering
practice. Hagen et al., (1987) presented a computerized, iter-
ative procedure that required two sieve sizes to determine
the parameters for a standard, two-parameter lognormal dis-
tribution. The lognormal distribution, however, often fails to
provide a close fit of the actual grain-size distribution near
the extremities of the curve (Gardner, 1956; Hagen et al.,
1987). Wagner and Ding (1994) later attempted to improve
upon the log normal equation through use of three and four
lognormal parameters.
Campbell (1985) presented a classification diagram based
on the assumption that the particle-size distribution was
approximately lognormal. The particle-size distribution was
approximated using a Gaussian distribution function. Any
combination of sand, silt, and clay could be represented by
a geometric (or log) mean particle diameter and a geometric
standard deviation. A modified U.S. Department of Agricul-
ture (USDA) textural classification chart was suggested by
Shirizi and Boersma (1984).
The use of a lognormal-type equation has a limitation in
that the grain-size distribution is assumed to be symmetric.
However, grain-size distribution curves are often nonsym-
metric and can be better fit using other forms of equations.
A method is required to more accurately represent soils that
are bimodal or gap graded.

2.2 SOIL CLASSIFICATION

There are two laboratory tests that are routinely used in the
classification of soils: the grain-size distribution curve and
the Atterberg limits. These tests are performed on recon-
stituted soil samples. However, when unsaturated soils are
involved, the classification tests take on additional signifi-
cance and meaning.
The grain-size distribution curve provides information on
the distribution of the solid particles or the percentage of
each particle size. The distribution of the solid particles
bears a relationship to the distribution of pore sizes or void
spaces. Information on the pore-size distribution can be used
for the estimation of the water content-soil suction relation-
ship for the soil (i.e., SWCC). Consequently, the grain-size
distribution becomes of increased value for understanding
unsaturated soil behavior.
There are three plasticity characteristics of a soil which
are referred to as Atterberg limits: liquid limit, plastic limit,
and shrinkage limit. Independent of the shrinkage limit is
the “shrinkage curve,” which is measured by allowing soil
to dry while measurements are made of the volume and mass
of the specimen. These measurements allow the calculation
of void ratio (or specific volume) and water content as the
soil dries.
Some key characteristics of unsaturated soil behavior can
be visualized along the shrinkage curve. The shrinkage curve
provides an indication of the air-entry value of the soil as
well as residual moisture conditions. The shrinkage curve is
generally measured on an initially slurry soil, but it is also
possible to measure the shrinkage curve starting at unsatu-
rated initial conditions.
Both of the commonly performed soil classification tests
provide information on the unsaturated soil behavior of a
soil and are further discussed in the following sections.

2.2.1 Grain-Size Distribution Curves
The grain-size distribution curve provides information on
the distribution of the sizes of the solid particles in a soil
and can be used to introduce concepts related to unsaturated
soil behavior (M.D. Fredlund, 2000). It is the distribution