CHAPTER 4

Soil Composition and

l
Engineering Properties

ria
ate
4.1 INTRODUCTION ceeds 50 percent.1 The engineering properties of co-
hesionless soil are often determined by applied
The engineering properties of a soil depend on the confining pressure and looseness or denseness as in-
composite effects of several interacting factors. These dicated by the relation of the current void ratio to the
factors may be divided into two groups: compositional
dM lowest and highest possible values of void ratio for the
factors and environmental factors. Compositional fac- soil. The engineering properties of cohesive soil are
tors determine the potential range of values for any often characterized by stiffness and strength and by
property. They include: relating the current water content and past consolida-
tion history to the compositional characterization pro-
1. Types of minerals vided by the plasticity index. Some engineering
2. Amount of each mineral characteristics of coarse-grained and fine-grained soils
3. Types of adsorbed cations
hte

are listed and compared in Fig. 4.1. Detailed discussion

4. Shapes and size distribution of particles of the combined effects of compositional and environ-
5. Pore water composition mental factors on the three most important property
6. Type and amount of other constituents, such as classes for engineering problems, that is, conductivity,
organic matter, silica, alumina, and iron oxide volume change, and deformation and strength, is given
in Chapters 9, 10, and 11.
rig

The influences of compositional factors on engineering
properties can be studied using disturbed samples.
Environmental factors determine the actual value of
any property. They include:
py
Quantitative determination of soil behavior com-
pletely in terms of compositional and environmental
factors is impractical for several reasons:
1. Most natural soil compositions are complex, and

1. Water content determination of soil composition is difficult.

2. Density 2. Physical and chemical interactions occur between
3. Confining pressure different phases and constituents.
4. Temperature 3. The determination and expression of soil fabric
Co

5. Fabric in quantitatively useful ways is difficult.

6. Availability of water
Undisturbed samples, or in situ measurements, are re-
quired for the study of the effects of environmental
factors on properties.
Soils are classified as coarse grained, granular, and
cohesionless if the amount of gravel and sand exceeds
50 percent by weight or fine grained and cohesive if
the amount of fines (silt and clay-size material) ex-
4. Past geologic history and present in situ environ-
ment are difficult to simulate in the laboratory.
5. Physicochemical and mechanical theories for re-
lating composition and environment to properties
quantitatively are inadequate.
1
The terms cohesionless and cohesive must be used with care, as
even a few percent of clay mineral in a coarse-grained soil can impart
plastic characteristics.

83

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

84 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

“Granular Soils” “Fines”

Gravel Sand Silt Clay

75 mm 5 mm 0.07 mm 0.002 mm
3 in. 0.2 in. 0.003 in. 0.00008 in.

Apples to English peas Finer than Much finer than

English peas to baking flour baking flour baking flour

l
ria
Particles visible without magnification Particles not visible without magnification
Grain size measurable with sieves Grain size not measurable with sieves
Grain size measured by sedimentation rate

Grains do not form a coherent mass Grains stick together when mixed with water due

ate
even when wet – cohesionless to pore water suction and physicochemical pore
fluid-mineral interaction – cohesive

Nonplastic – there is no range of water Plastic – deforms without cracking over a range
content where the soil can be deformed of water content between the liquid limit and the
without cracking or crumbling. plastic limit
dM Liquid (pancake batter)
Plastic (modeling clay)
Liquid Limit (LL)
Plastic Limit (PL)
Semisolid (chocolate bar)
Shrinkage Limit
Solid (chalk)

Permeability is moderate to high (10 -6 to 10-1 m/s). Permeability is low to very low (<10-7 m/s).
hte

Water flows easily through the voids. Water flows slowly through the voids.
Drainage takes weeks to tens of years.

Drainage occurs rapidly except under dynamic
loading; e.g., earthquakes.
Only “drained” strength is important for conditions
other than earthquake loading or rapid landslides.
rig
Both “drained” and “undrained” strengths are important.
“Undrained” strength is low when preconsolidation
pressure is low.

Most important indicators of mechanical behavior Most important mechanical

are relative density, Dr , and applied confining behavior is “preconsolidation
pressure Behavior of silts pressure pp” and applied
varies from “sand- confining pressure
Dr = 0 to 20% Very loose like” to “clay-like” as
py

pp = 0 to 50 kPa Very soft

Dr = 20 to 40% Loose grain size decreases
Dr = 40 to 60% Med. dense pp = 50 to 100 kPa Soft
Dr = 60 to 80% Dense pp = 100 to 200 kPa Firm
Dr = 80 to 100% Very dense pp = 200 to 400 kPa Stiff
pp = 400 to 800 kPa Very stiff
Co

pp = 0.8 to 1.6 MPa Hard

Very loose _ Compressible
Liquefiable during earthquakes Very soft – Very highly compressible
φ ~30° Undrained shear strength <12.5 kPa
Very dense _ Very low compressibility
Stable during earthquakes Very dense _ Low compressibility
φ ~45° Undrained shear strength >100 kPa

Figure 4.1 Compositional and environmental factors contributing to engineering properties

(adapted from course notes by J. M. Duncan, 1994).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com


Nonetheless, compositional data are valuable for de-
velopment of an understanding of properties and for
establishment of qualitative to semiquantitative guide-
lines for how real soils behave. Accordingly, some
relationships between compositional factors and engi-
neering properties are summarized in this chapter.
4.2 APPROACHES TO THE STUDY OF
COMPOSITION AND PROPERTY
INTERRELATIONSHIPS
l
ria
Study of soil composition in relation to soil properties
may be approached in two ways. In the first, natural
soils are used, the composition and engineering prop-
erties are determined, and correlations are made. This
method has the advantage that measured properties are
ate
those of naturally occurring soils. Disadvantages, how-
ever, are that compositional analyses are difficult and
time consuming, and that in soils containing several
minerals or other constituents such as organic matter,
silica, alumina, and iron oxide the influence of any one
dM
constituent may be difficult to isolate.
In the second approach, the engineering properties
of synthetic soils are determined. Soils of known com-
position are prepared by blending different commer-
cially available clay minerals of relatively high purity
with each other and with silts and sands. Although this
approach is much easier, it has the disadvantages that
hte
the properties of the pure minerals may not be the
same as those of the minerals in the natural soil, and
important interactions among constituents may be
missed. Whether the influences of constituents such as
organic matter, oxides and cementation, and other
chemical effects can be studied successfully using this
rig
approach is uncertain.
Regardless of the approach used, there are at least
two difficulties. One is that often the variability in both
composition and properties in any one soil deposit may
be great, making the selection of representative sam-
py
ples difficult. Variations in composition and texture oc-
cur in sediments within distances as small as a few
centimeters. Residual soils, in particular, are likely to
be very nonhomogeneous.
Co
A second difficulty is that the different constituents
of a soil may not influence properties in direct or even
predictable proportion to the quantity present because
of physical and physicochemical interactions. As an
example of physical interactions, blending of equal
proportions of uniform sand and clay, each having a
compacted unit weight of 17 kN/m3, would not nec-
essarily yield a mixture also having a unit weight of
17 kN/m3 after compaction. The resulting unit weight
might be as high as 20 kN/m3 because the clay can
fill void spaces between sand particles.
Copyright © 2005 John Wiley & Sons
ENGINEERING PROPERTIES OF GRANULAR SOILS 85
Physicochemical interaction between clay minerals
is shown in Fig. 4.2. Mixtures of bentonite (sodium
montmorillonite) and kaolinite and of bentonite and a
commercial illite containing about 40 percent illite clay
mineral, with the rest mostly silt-sized nonclay, were
prepared, and the liquid limits were determined. The
dashed line in Fig. 4.2 shows the liquid limit values to
be expected if each mineral contributed in proportion
to the amount present. The data points and solid lines
show the actual measured values. Although the ben-
tonite–kaolinite mixtures gave values close to theoret-
ical, the liquid limit values for the bentonite–illite
mixtures were much less than predicted. This resulted
from excess salt in the illite that, when mixed with the
bentonite, prevented full interlayer expansion of the
montmorillonite particles in the presence of water.
4.3 ENGINEERING PROPERTIES OF
GRANULAR SOILS
The mechanical behavior of granular materials is gov-
erned primarily by their structure and the applied ef-
fective stresses. Structure depends on the arrangement
of particles, density, and anisotropy. Particle sizes,
shapes, and distributions, along with the arrangement
of grains and grain contacts comprise the soil fabric.
The packing characteristics of granular materials are
discussed further in Chapter 5.
Particle Size and Distribution
Figure 4.3 illustrates the tremendous range in particle
sizes that may be found in a soil, where different sizes
are shown to the same scale. The largest size shown
represents fine sand. It may be recalled that particles
finer than about 0.06 mm cannot be seen by the naked
eye. The orders of magnitude difference in particle
sizes found in any one soil is often better appreciated
from a representation such as that in Fig. 4.3 than by
the usual size distribution (or grading) curve where
particle diameters are shown to a logarithmic scale.
The origin of a cohesionless soil can be reflected by
its grading. Alluvial terrace deposits and aeolian de-
posits tend to be poorly graded or sorted. Glacial de-
posits such as Boulder clays and tills are often well
graded, containing a wide variety of particle sizes.
Small particles in a well-graded soil fit into the voids
between larger particles. Well-graded cohesionless
soils are relatively easy to compact to a high density
by vibration. The loss of fine fraction by internal ero-
sion can lead to large changes in engineering proper-
ties. Uniformly graded soils are usually used for
controlled drainage applications because they are not
susceptible to loss of fines by internal erosion and their
Retrieved from: www.knovel.com
86 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

l
ria
ate
dM
Figure 4.2 Interactions between clay minerals as indicated by liquid limit (data from Seed
et al., 1964).
hte

hydraulic conductivity can be maintained within defin-
able and narrow limits.
The slope of the grain size distribution curve is char-
acterized by the coefficient of uniformity Cu:
rig
minimum void ratios change by mixing sand and silt
in different proportions. At low silt contents, silt par-
ticles fit into the voids between larger sand particles,
so the void ratio of sand–silt mixtures decreases with
increase in silt content. However, at a certain silt con-

d60 tent, the silt fully occupies the voids, and the increase
Cu ⫽ (4.1) in silt content results in sand particles floating inside
d10
the silt matrix. Then, the void ratios increase with fur-
ther increase in silt content.
where d60 and d10 correspond to the sieve sizes that 60
py

The relative density, DR, a measure of the current

and 10 percent of the particles by weight pass through.
void ratio in relation to the maximum and minimum
A soil with Cu ⬎ 5 to 10 is considered well-graded.
void ratios, and applied effective stresses controls the
The possible range of packing of soil particles is
mechanical behavior of cohesionless soils. Relative
often related to the maximum and minimum void ratios
Co

density is defined by
(or minimum and maximum densities) reflecting the
loosest and densest states, respectively. Uniformly
graded soils tend to have a narrower range of possible emax ⫺ e
DR ⫽ ⫻ 100% (4.2)
densities compared to well-graded soils. Soils contain- emax ⫺ emin
ing angular particles tend to be less dense than soils
with rounded particles, as discussed later in this sec- in which emax, emin, and e are the maximum, minimum,
tion. However, angular and weak materials may crush and actual void ratios.
significantly more during compression, compaction, or The relative density correlates well with other prop-
deformation. Figure 4.4 shows how the maximum and erties of granular soils. As different standard test meth-

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

 ENGINEERING PROPERTIES OF GRANULAR SOILS 87

2.0
1.8 Maximum void ratio

1.6 Minimum void ratio

1.4

Void ratio
1.2
1.0
0.8
0.6
0.4

l
ria
0.2
0.0
0 10 20 30 40 50 60 70 80 90 100
Silt content (%)

ate
Figure 4.4 Maximum and minimum void ratios of Monterey
sand–silt mixtures (from Polito and Martin, 2001).

Morphology (large scale) Roundness Texture

(intermediate scale)
dM
hte

Roundness Texture Surface Texture

(intermediate scale) (small scale)

Figure 4.5 Scale-dependent particle shape characterization.

Figure 4.3 Different grain sizes in soil. The solid line gives the particle outline. Morphology de-
rig

scribes overall shape of the particle as given by the heavy

ods can give different limiting void ratios, the use of
the relative density is sometimes criticized, especially
py
dotted line. Texture reflects the smaller scale local features
of the particles as identified by light dotted circles. The ex-
amples are surface smoothness, roundness of edges and cor-
ners, and asperities.

when considered in relation to the random in situ var-

iations of the density of most sand and gravel deposits.
Nonetheless, if properly interpreted, relative density
can provide a very useful measure of cohesionless soil elliptical, elongated, and so forth. At smaller scales,
the texture, which reflects the local roughness features
Co

properties.
Particle Shape
Particle shape is an inherent soil characteristic that
plays a major role in mechanical behavior of soils.
Characterization of particle shape is scale dependent,
as shown in Fig. 4.5. At larger scales, that is, that of
the particle itself, the particle morphology might be
described as spherical, rounded, blocky, bulky, platy,
such as surface smoothness, roundness of edges and
corners, and asperities, is important.
With the exception of mica, most nonclay minerals
in soils occur as bulky particles.2 Most particles are
2
Quartz particles become flatter with decreasing size and may have
a platy morphology when subdivided to a fineness approaching clay
size (Krinsley and Smalley, 1973).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

88 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

not equidimensional, however, and are at least slightly tion of aspect ratio and roundness. A convenient way
elongate or tabular. A frequency histogram of particle to characterize particle shapes in more detail is by a
length-to-width ratio (L/W) for Monterey No. 0 sand Fourier mathematical technique. For instance, the (R,
is shown in Fig. 4.6. This well-sorted beach sand is ) Fourier method is in the following form:
composed mainly of quartz with some feldspar. The

冘 (a cos n ⫹ b sin n)
mean of all the particle measurements is an L/W ratio N
of 1.39. This distribution is typical of that for many R() ⫽ a0 ⫹ n n (4.3)
n⫽1
sands and silty sands.
Particle morphology in soil mechanics has histori-
cally been described using standard charts against
which individual grains may be compared. A typical
l
chart and some examples are shown in Fig. 4.7 (Krum-
ria
bein, 1941; Krumbein and Sloss, 1963; Powers, 1953).
Sphericity is defined as the ratio of the diameter of a
sphere of equal volume to the particle to the diameter
of the circumscribing sphere. Roundness is defined as
the ratio of the average radius of curvature of the cor-
ate
ners and edges of the particle to the radius of the max-
imum sphere that can be inscribed (Wadell, 1932).
Sphericity and roundness are measures of two very dif-
ferent morphological properties. Sphericity is most de-
pendent on elongation, whereas roundness is largely
dM
dependent on the sharpness of angular protrusions
from the particle. Different definitions of sphericity
and roundness are available, as shown in Table 4.1.
Due to the variety of definitions available, the quanti-
fication of particle shape requires accurate specifica-
tion of their definition.
In recent years, techniques for computer analysis of
hte
shape data by digital imaging have improved greatly,
and standard software applications include determina-
rig
py
Co
Figure 4.6 Grain shape distribution of Monterey No. 0 sand. Results are based on study of
277 particles, d50 ⫽ 0.43 mm, Cu ⫽ 1.4 (Mahmood, 1973).
Copyright © 2005 John Wiley & Sons
where R() is the radius at angle , N is the total num-
ber of harmonics, n is the harmonic number, and a and
b are coefficients giving the magnitude and phase for
each harmonic. The lower harmonic numbers give the
overall shape; for instance, the sphericity is expressed
by the first and second harmonics. The coefficient val-
ues for higher-order descriptors generally decay with
increasing descriptor or harmonic number, which ex-
presses smaller features (i.e., texture) (Meloy, 1977).
Other mathematical methods to curve-fit particle
shapes are listed in Table 4.1. Further discussion on
particle shape characterization is given by Barrett
(1980), Hawkins (1993), Santamarina et al. (2001), and
Bowman et al. (2001).
In an assembly of uniform size spherical particles,
the loosest stable arrangement is the simple cubic
packing giving a void ratio of 0.91. The densest pack-
ing is the tetrahedral arrangement giving a void ratio
of 0.34. Particle shape affects minimum and maximum
void ratios as shown in Fig. 4.8 (Youd, 1973). The
values increase as particles become more angular or
the roundness (defined as roundness 1 in Table 4.1)
Retrieved from: www.knovel.com
 ENGINEERING PROPERTIES OF GRANULAR SOILS 89

0.9

0.7

Sphericity
0.5

l
ria
0.3

ate
0.1 0.3 0.5 0.7 0.9
Roundness
(a)

dM
High Sphericity

Low Sphericity
hte

Very Angular Subangular Subrounded Rounded Well

Angular Rounded
(b)

Figure 4.7 Particle shape characterization: (a) Chart for visual estimation of roundness and
sphericity (from Krumbein and Sloss, 1963). (b) Examples of particle shape characterization
rig

(from Powers, 1953).

decreases. When R ⫽ 1, the particle is a sphere. As the following linear fit to the relationship is proposed
py

particles become more angular, R decreases to zero. (Santamarina and Cho, 2004);
Void ratios are also a function of particle size distri-
bution; the values decrease as the range of particle repose ⫽ 42 ⫺ 17R (4.4)
sizes increases (increase in the coefficient of unifor-
mity Cu).
Co

where R is the coefficient of roundness defined as

The friction angle increases with increase in particle
angularity, possibly as a result of an increase in coor-
dination number. For example, values of the angle of
repose3 are plotted against roundness in Fig. 4.9 and
3
Angle of repose can be determined by pouring soil in a graduated
cylinder filled with water. Tilt the cylinder more than 60 and bring
it back slowly to the vertical position. The angle of the residual sand
slope is the angle of repose. Further details of the method can be
found in Santamarina and Cho (2004).
roundness 1 in Table 4.1. Similar data relating friction
angle from drained triaxial tests and particle shape is
presented by Sukumaran and Ashmawy (2001).
Particle Stiffness
Soil mass deformation at very small strains originates
from the elastic deformations at points of contact be-
tween particles. Contact mechanics shows that the elas-
tic properties of particles control the deformations at
particle contacts (Johnson, 1985), and these deforma-

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

90 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

Table 4.1 Methods for Particle Shape Characterization

Method Definition
Morphology—Sphere
Sphericity 1 Diameter of a sphere of equal volume
Diameter of circumscribing sphere

Sphericity 2 Particle volume

l
Volume of circumscribing sphere

ria
Sphericity 3
Projection sphericity Area of particle outline
Area of a circle with diameter equal to the longest length of outline

ate
Inscribed circle sphericity Diameter of the largest inscribed circle
Diameter of the smallest inscribed circle

dM Morphology—Ellipse
Eccentricity p /Rap, where the ellipse is characterized by Rp ⫹ p cos 2 in polar coordinates

Elongation Smallest diameter

Diameter perpendicular to the smallest diameter

Slenderness Maximum dimension

Minimum dimension
hte

Texture—Roundness

Roundness 1 Average of radius of curvature of surface features, (兺ri)/N

Radius of the maximum sphere that can be inscribed, rmax
rig

Roundness 2 Radius of curvature of the most convex part

0.5 (longest diameter through the most convex part)

Roundness 3 Radius of curveture of the most convex part

py

Mean radius

Morphology—Texture
Co

Fourier method Eq. (4.3), first and second harmonics, characterize sphericity, whereas higher harmonics
(around 10th) characterizes roundness. Surface texture is characterized by much
higher harmonics.
Fourier descriptor More flexible than the Fourier method by using the complex plane (Bowman et al.,
method 2001). Lower harmonics give shape characteristics such as elongation, triangularity,
squareness, and asymmetry. Higher harmonics (larger than 8th) give textural features.
Fractal analysis Use as a measure of texture (Vallejo, 1995; Santamarina, et al. 2001).
From Hawkins (1993), Santamarina et al. (2001), and Bowman et al. (2001).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

 ENGINEERING PROPERTIES OF GRANULAR SOILS 91
tions in turn influence the stiffness of particle assem-
blages. Elastic properties of different minerals and
1.4 rocks are listed in Table 4.2. The modulus of a single
grain, which determines the particle contact stiffness,
is at least an order of magnitude greater than that of
Maximum Void ratio, emax

1.2
the particle assembly. Further details on the relation
An
between particle stiffness and particle assemblage stiff-
gu ness are given in Chapter 11.
1.0 lar
R=
0.2
0
Particle Strength
0.8 0.20
Sub

l
an The crushability of soil particles has large effects on
Sub gular

ria
rou
nde 0.25 the mechanical behavior of granular materials. At high
Rou d
0.6 nde 0.30 stresses, the compressibility of sand becomes large as
d
0.35 a result of particle crushing, and the shape of an e–log
0.49 p compression curve becomes similar to that of nor-
Minimum Void Ratio, emin

0.8
0.70 mally consolidated clay (Miura et al., 1984; Coop,

ate
Ang
ular
R=0
1990; Yasufuku et al., 1991). Under constant states of
.20
0.6 stress, the amount of particle breakage increases with
Ang
ular
R=0 time, contributing to creep of the soil (Lade et al.,
.17
0.20
1996). The amount of crushing in a soil mass depends
0.4 Suba both on the stiffness and strength of the individual
ngula
r 0.25 0.35 grains and how applied stresses are transmitted through
0.2
Subrounded
Rounded
dM 0.30 0.49
0.70 the assemblage of soil particles.
1 2 3 4 6 10 15 Particle strength or hardness is characterized by
Coefficient of Uniformity, Cu crushing at contacts or particle tensile splitting. There
is a statistical variation in grain strength for particles
Figure 4.8 Maximum and minimum void ratios of sands as of a specified material and of a given size (Moroto and
a function of roundness and the coefficient of uniformity Ishii, 1990; McDowell, 2001). Random variation in
(from Youd, 1973).
hte

grain strengths leads to distributions of particle sizes

when large stress is applied to a soil assembly. Table
4.3 lists the characteristic tensile strengths of some soil
particles. The values are smaller than the yield strength
of the material itself. The strength also depends on the
particle shape. For example, Hagerty et al. (1993) show
rig

that angular glass beads were more susceptible to

breakage than round glass beads.
py

50
Table 4.2 Elastic Properties of Geomaterials at
Angle of repose φrepose

Room Temperature
40
Young’s Shear Modulus Poisson’s
Co

30 Material Modulus (GPa) (GPa) Ratio

Quartz 76 29 0.31
20 Limestone 2–97 1.6–38 0.01–0.32
φrepose = 42 – 17R Basalt 25–183 3–27 0.09–0.35
Granite 10–86 7–70 0.00–0.30
10
0.0 0.2 0.4 0.6 0.8 1.0 Hematite 67–200 27–78 —
Roundness R Magnetite 31 19 —
Shale 0.4–68 5–30 0.01–0.34
Figure 4.9 Angle of repose as a function of roundness (from
Santamarina and Cho, 2004). After Santamarina et al. (2001).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

92 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

Table 4.3 Strength of Soil Particles

37% Tensilea Mean Strengthb

Sand Name Size (mm) Strength (MPa) (MPa) Reference
Quartz
Leighton Buzzard silica sand 1.18 — 29.8 Lee (1992)
2.0 — 24.7
3.36 — 20.5
Toyoura sand 0.2 147.4 136.6 Nakata et al. (2001)
Aio quartz sand 0.85 51.2 52.1 Nakata et al. (1999)
1.0 47.7 46.6

l
1.18 37.9 35.6

ria
1.4 46.7 42.4
1.7 39.6 38.5
Silica sand 0.5 147.4 132.5 McDowell (2001)
1 66.7 59.0

ate
2 41.7 37.3
Silica sand 0.28 110.9 147.3 Nakata et al. (2001)
0.66 72.9 73.1
1.55 31.0 29.7

dM Feldspar
Aio feldspar sand 0.85 20.9 24.6 Nakata et al. (1999)
1.0 24.3 22.8
1.18 18.1 18.2
1.4 23.1 21.4
1.7 18.9 18.3

Calcareous Sand
hte

Oolitic limestone particle 5 — 2.4 Lee (1992)

8 — 2.1
12 — 1.8
20 — 1.5
30 — 1.3
40 — 1.2
rig

50 — 1.1
Carboniferous limestone 5 — 14.9
particle 8 — 12.2 Lee (1992)
12 — 10.3
20 — 8.3
py

30 — 7.0
40 — 6.2
50 — 5.7
Quiou sand 1 109.3 96.19 McDowell and Amon (2000)
Co

2 41.4 36.20
4 4.2 3.87
8 0.73 0.63
16 0.61 0.54

Others
Masado decomposed granite
soil 1.55 24.2 22.1 Nakata et al. (2001)
Glass beads 0.93 365.8 339.6 Nakata et al. (2001)
Angular glass 0.93 62.1 60.0 Nakata et al. (2001)
a
Stress below which 37% of the particles do not fracture.
b
Force/d 2 at which particle of size d is crushed.

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

 ENGINEERING PROPERTIES OF GRANULAR SOILS 93
The breakage potential of a single soil particle in- 100
creases with its size as illustrated in Table 4.3. This is Maximum stress
because larger particles tend to contain more and larger 20.7 MPa
internal flaws and hence have lower tensile strength. 80 41.4 MPa

Percent Finer by Weight

Fig. 4.10 shows that oolitic limestone, carboniferous 62.1 MPa
limestone, and quartz sand exhibit near linear declines 103 MPa
60
in strength with increasing particle size on a log–log 345 MPa
plot (Lee, 1992). 517 MPa
The amount of particle crushing in an assemblage 40 689 MPa
of particles depends not only on particle strength, but
also on the distribution of contact forces and arrange-
Uncrushed

l
ment of different size particles. It can be argued that 20

ria
larger size particles are more likely to break because
the normal contact forces in a soil element increase
with particle size and the probability of a defect in a 0
0.01 0.1 1.0
given particle increases with its size as shown in Fig
Grain size (mm)
4.10 (Hardin, 1985). However, if a larger particle has

ate
contacts with neighboring particles (i.e., larger coor- Figure 4.11 Evolution of particle size distribution curve
dination number), the load on it is distributed, and the upon crushing (from Hagerty et al., 1993).
probability of facture is less than for a condition with
fewer contacts. Experimental evidences suggest that
fines increase as particles break by increase in applied
dM
pressure. For example, the evolution of particle size tact with many smaller particles. The very smallest
distribution curves for Ottawa sand in one-dimensional particles have a lower coordination number because
compression is shown in Fig. 4.11 (Hagerty et al., there are fewer smaller particles available for contact.
1993). Hence, the coordination number dominates over Hence, the largest particles in the aggregate become
size-dependent particle strength. Larger particles have protected by the surrounding newly formed smaller
higher coordination numbers because they are in con- particles, and smaller particles are more likely to break
hte

50 Leighton Buzzard Sand Rounded River Gravel

Angular River Gravel

Carboniferous Limestone
rig

10
Particle Strength (MPa)

5.0
py
Co

1.0 Oolitic Limestone

0.5

0.2
1 5 10 50 100
Average Particle Size (mm)

Figure 4.10 Relationship between tensile strength and particle size (from Lee, 1992).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

94 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

or move. Further details on particle breakage effects

on compression behavior of sands are given in Chap- 冉
eG VGS ⫽ 1 ⫺
C
冊
Ws
e
100 GSG w G
(4.5)
ter 10.
The volume of water plus volume of clay is given by

w WS C WS
VW ⫹ VC ⫽ ⫹ (4.6)
4.4 DOMINATING INFLUENCE OF THE CLAY 100 w 100 GSC w
PHASE
If clay and water completely fill the voids in the gran-
In general, the more clay in a soil, the higher the plas- ular phase, then
ticity, the greater the potential shrinkage and swell, the
冉 冊

l
lower the hydraulic conductivity, the higher the com- w Ws C W C Ws

ria
⫹ ⫽ 1⫺ e (4.7)
pressibility, the higher the cohesion, and the lower the 100 w 100 GSC w 100 GSG w G
internal angle of friction. Whereas surface forces and
their range of influence are small relative to the weight which simplifies to
and size of silt sand particles, the behavior of small

冉 冊

ate
and flaky clay mineral particles is strongly influenced w C C eG
by surface forces, as discussed in Chapter 6. Water is ⫹ ⫽ 1⫺ (4.8)
100 100GSC 100 GSG
strongly attracted to clay particle surfaces, also dis-
cussed in Chapter 6, and results in plasticity, whereas The void ratio of a granular material composed of
nonclay particles have much smaller specific surface bulky particles is of the order of 0.9 in its loosest pos-
and less affinity for water and do not develop signifi-
dM sible state. The specific gravity of the nonclay fraction
cant plasticity, even when in finely ground form. in most soils is about 2.67, and that of the clay fraction
If it is assumed as a first approximation that all of is about 2.75. Inserting these values in Eq. (4.8) gives
the water in a soil is associated with the clay phase,
the amount of clay required to fill the voids of the C ⫽ 48.4 ⫺ 1.42w (4.9)
granular phase and prevent direct contact between
granular particles can be estimated for any water con- This relationship indicates that for water contents typ-
tent. The weight and volume relationships for the dif-
hte

ically encountered in practice, say 15 to 40 percent,

ferent phases of a saturated soil are shown in Fig. 4.12.
In this figure W represents weight, V is volume, C is
the percent clay by weight, GSC is the specific gravity
of clay particles, w is the water content in percent, w
is the unit weight of water, and GSG is the specific
rig
only a maximum of about one-third of the soil solids
need be clay in order to dominate the behavior by pre-
venting direct interparticle contact of the granular par-
ticles. In fact, since there is a tendency for clay
particles to coat granular particles, the clay can signif-

gravity of the granular particles. The volume of voids
in the granular phase is eG VGS, where eG is the void
ratio of the granular phase and VGS is the volume of
granular solids, given by
py
Co
icantly influence properties. For example, just 1 or 2
percent of highly plastic clay present in gravel used as
a fill or aggregate may be sufficient to clog handling
and batching equipment.

Figure 4.12 Weight–volume relationships for a saturated clay-granular soil mixture.

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com


4.5 ATTERBERG LIMITS
Atterberg limits are extensively used for identification,
description, and classification of cohesive soils and as
a basis for preliminary assessment of their mechanical
properties. The potential usefulness of the Atterberg
limits in soil mechanics was first indicated by Terzaghi
(1925a) when he noted that ‘‘the results of the simpli-
fied soil tests (Atterberg limits) depend precisely on
the same physical factors which determine the resis-
tance and the permeability of soils (shape of particles,
l
effective size, uniformity) only in a far more complex
ria
manner.’’
Casagrande (1932b) developed a standard device for
determination of the liquid limit and noted that the
nonclay minerals quartz and feldspar did not develop
plastic mixtures with water, even when ground to sizes
ate
smaller than 2 m. Further studies led to the formation
of a soil classification system based on the Atterberg
limits for identification of cohesive soils (Casagrande,
1948). This system was adopted, with minor modifi-
cations, as a part of the Unified Classification System.
A plot of plasticity index as a function of liquid limit
dM
that is divided into different zones, as shown in Fig.
4.13, is termed the plasticity chart. This chart forms
an essential part of the Unified Soil Classification Sys-
tem.
hte
rig
py
Co
Figure 4.13 Plasticity chart.
Copyright © 2005 John Wiley & Sons
ATTERBERG LIMITS 95
Although both the liquid and plastic limits are easily
determined, and their qualitative correlations with soil
composition and physical properties are quite well es-
tablished, fundamental interpretations of the limits and
quantitative relationships between their values and
compositional factors are more complex.
Liquid Limit
The liquid limit test is a form of dynamic shear test.
Casagrande (1932b) deduced that the liquid limit cor-
responds approximately to the water content at which
a soil has an undrained shear strength of about 2.5 kPa.
Subsequent studies have indicated that the liquid limit
for all fine-grained soils corresponds to shearing resis-
tance of about 1.7 to 2.0 kPa and a pore water suction
of about 6 kPa (Russell and Mickle, 1970; Wroth and
Wood, 1978; Whyte, 1982).
Liquid limit values are determined using both the
Casagrande liquid limit device and the fall cone device.
Different standards adopt different devices and, there-
fore, correlations based on liquid limit should be used
with some caution. The variation of undrained shear
strength with water content can be obtained from a
series of fall cone tests and solutions are available us-
ing the theory of plasticity for various geometries used
in fall cones (Houlsby, 1982; Koumoto and Houlsby,
Retrieved from: www.knovel.com
96 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES
2001). Furthermore, with the aid of critical state soil
mechanics (see Chapter 11), some other engineering
properties, such as compressibility, can be deduced
(Wood, 1990).
Values of hydraulic conductivity at the liquid limit
for several clays are given in Table 4.4, from Nagaraj
et al. (1991). The striking aspect of these data is that,
although the water contents and void ratios at the liq-
uid limit for the different clays vary over a very wide
range, the hydraulic conductivity is very nearly the
same for all of them. This means that the effective pore
l
sizes controlling fluid flow must be about the same for
ria
all the clays at their liquid limit. Such a microfabric is
consistent with the cluster model for hydraulic con-
ductivity discussed in Chapter 9. In this model, the
individual clay particles associate into aggregates or
flocs, as shown schematically in Fig. 9.11. The size of
ate
voids between the clusters or aggregates controls the
flow rate according to either model.
The approximately equal strengths, pore water suc-
tions, and hydraulic conductivities for all clays at their
liquid limit can be explained by the concepts that (1)
dM
the aggregates or clusters are the basic units that in-
teract to develop the strength, that is, the aggregates
act somewhat like single particles, (2) the average ad-
sorbed water layer thickness is about the same on all
particle surfaces, and (3) the average size of interclus-
ter pores is the same for all clays. Concept 2 provides
the key to why different clays have different values of
hte
liquid limit. All clays have essentially the same surface
structures, that is, a layer of oxygen atoms in tetrahe-
dral coordination with silicon, or a layer of hydroxyls
in octahedral coordination with aluminum or magne-
sium. The forces of interaction between these surfaces
and adsorbed water should be about the same for the
rig
different clay minerals. Thus, the amount of water ad-
sorbed per unit area of surface that corresponds to a
pore water suction of 6 kPa should be about the same.
This means that the greater the specific surface, the
py
Table 4.4 Hydraulic Conductivity at Liquid Limit for Several Clays
Co
Liquid Limit,
Soil Type wL (%)
Bentonite 330
Bentonite ⫹ sand 215
Natural marine soil 106
Air-dried marine soil 84
Oven-dried marine soil 60
Brown soil 62
From Nagaraj et al. (1991).
Copyright © 2005 John Wiley & Sons
greater the total amount of water required to reduce
the strength to that at the liquid limit. The specific
surface areas of the different clay minerals (Table 3.6)
are consistent with the liquid limit values of different
clay minerals in Table 4.5. Additional support for this
concept is given by the following relationship found
for 19 British clays:
LL ⫽ 19 ⫹ 0.56As (20%) (4.10)
where LL is the liquid limit and As is the specific sur-
face in square meters per gram (Farrar and Coleman,
1967).
The effects of electrolyte concentration, cation va-
lence and size, and dielectric constant of the pore fluid
on the liquid limit of kaolinite and montmorillonite are
illustrated and discussed by Sridharan (2002). The ef-
fects are generally consistent with the above interpre-
tation and can be explained also through double-layer
(see Chapter 6) influences on swelling, flocculation
and deflocculation of clay particles, and shear strength.
Plastic Limit
The plastic limit has been interpreted as the water con-
tent below which the physical properties of the water
no longer correspond to those of free water (Terzaghi,
1925a) and as the lowest water content at which the
cohesion between particles or groups of particles is
sufficiently low to allow movement, but sufficiently
high to allow particles to maintain the molded posi-
tions (Yong and Warkentin, 1966). Whatever the struc-
tural status of the water and the nature of the
interparticle forces, the plastic limit is the lower
boundary of the range of water contents within which
the soil exhibits plastic behavior; that is, above the
plastic limit the soil can be deformed without volume
change or cracking and will retain its deformed shape;
below the plastic limit it cannot. Plastic limit values
Void Ratio at Hydraulic
Liquid Limit, Conductivity
eL (10⫺7 cm/s)
9.240 1.28
5.910 2.65
2.798 2.56
2.234 2.42
1.644 2.63
1.674 2.83
Retrieved from: www.knovel.com
 INFLUENCES OF EXCHANGEABLE CATIONS AND pH 97
Table 4.5 Atterberg Limit Values for the Clay Minerals

Liquid Plastic Shrinkage

Limit Limit Limit
Minerala (%) (%) (%)
Montmorillonite (1) 100–900 50–100 8.5–15
Nontronite (1)(2) 37–72 19–27
Illite (3) 60–120 35–60 15–17
Kaolinite (3) 30–110 25–40 25–29
Hydrated halloysite (1) 50–70 47–60

l
Dehydrated halloysite (3) 35–55 30–45

ria
Attapulgite (4) 160–230 100–120
Chlorite (5) 44–47 36–40
Allophane (undried) 200–250 130–140
a
(1) Various ionic forms. Highest values are for monovalent; lowest

ate
are for di- and trivalent. (2) All samples 10% clay, 90% sand and silt.
(3) Various ionic forms. Highest values are for di- and trivalent; lowest
are for monovalent. (4) Various ionic forms. (5) Some chlorites are
nonplastic.
Data Sources: Cornell University (1950), Samuels (1950), Lambe
and Martin (1955), Warkentin (1961), and Grim (1962).
dM
for different clay minerals are listed in Table 4.5. The plasticity index
Activity ⫽ (4.12)
undrained shear strength at the plastic limit is reported % ⬍ 2 m
to be in the ranges of 100 to 300 kPa with an average
value of 170 kPa (Sharma and Bora, 2003). For many clays, a plot of plasticity index versus clay
hte

content yields a straight line passing through the origin

Liquidity Index as shown for four clays in Fig. 4.14. The slope of the
The liquidity index (LI) is defined by line for each clay gives the activity. Approximate val-
ues for the activities of different clay minerals are
water content ⫺ plastic limit listed in Table 4.6.
LI ⫽ (4.11) The greater the activity, the more important the in-
rig

plasticity index fluence of the clay fraction on properties and the more
susceptible their values to changes in such factors as
wherein the plasticity index is given by PI ⫽ LL ⫺ type of exchangeable cations and pore fluid composi-
PL. The liquidity index is useful for expressing and tion. For example, the activity of Belle Fourche mont-
py

comparing the consistencies of different clays. It nor-
malizes the water content relative to the range of water
content over which a soil is plastic. It correlates well
with compressibility, strength, and sensitivity proper-
ties of fine-grained soils as illustrated in later chapters
Co
morillonite varies from 1.24 with magnesium as the
exchangeable cation to 7.09 for sodium saturation of
the exchange sites. On the other hand, the activity of
Anna kaolinite only varies from 0.30 to 0.41 for six
different cation forms (White, 1955).

of this book.

4.7 INFLUENCES OF EXCHANGEABLE

4.6 ACTIVITY
Both the type and amount of clay influence a soil’s
properties, and the Atterberg limits reflect both of these
factors. To separate them, the ratio of the plasticity
index to the clay size fraction (percentage by weight
of particles finer than 2 m), termed the activity, is
very useful (Skempton, 1953):
CATIONS AND pH
Cation type exerts a controlling influence on the
amount of swelling of expansive clay minerals in the
presence of water. For example, sodium and lithium
montmorillonite may undergo almost unrestricted
interlayer swelling provided water is available,
the confining pressure is small, and the electrolyte

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

98 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

l
ria
ate
dM
Figure 4.14 Relationship between plasticity index and clay fraction (from Skempton, 1953).
hte

Table 4.6 Activities of Various Clay Minerals sions ordinarily flocculate in the presence of divalent
and trivalent cations.
Mineral Activity pH influences interparticle repulsions because of its
effects on clay particle surface charge. Positive edge
Smectites 1–7 charges can exist in low pH environments. These ef-
Illite 0.5–1 fects are of greatest importance in kaolinite, lesser
rig

Kaolinite 0.5 importance in illite, and relatively unimportant in

Halloysite (2H2O) 0.5 smectite. In kaolinite, the pH may be the single most
Halloysite (4H2O) 0.5 important factor controlling the fabric of sediments
Attapulgite 0.5–1.2 formed from suspension.
Allophane 0.5–1.2 The influences of cations and pH are examined fur-
py

ther in Chapter 6.

concentration is low. On the other hand, divalent and 4.8 ENGINEERING PROPERTIES OF CLAY
Co

trivalent forms of montmorillonite do not expand be- MINERALS

yond a basal spacing of about 17 Å and form multi-
particle clusters or aggregates, regardless of other
environmental factors.
In soils composed mainly of nonexpansive clay min-
erals, adsorbed cation type is of the greatest impor-
tance in influencing the behavior of the material in
suspension and the nature of the fabric in sediments
that form. Monovalent cations, particularly sodium and
lithium, promote deflocculation, whereas clay suspen-
Different groups of clay minerals exhibit a wide range
of engineering properties. Within any one group, the
range of property values may also be great. It is a
function of particle size, degree of crystallinity, type
of adsorbed cations, pH, the presence of organic mat-
ter, and the type and amount of free electrolyte in the
pore water. In general, the importance of these factors
increases in the order kaolin ⬍ hydrous mica (illite) ⬍

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com


smectite. The chlorites exhibit characteristics in the ka-
olin–hydrous mica range. Vermiculites and attapulgite
have properties that usually fall in the hydrous mica–
smectite range.
Because of the influences of the above composi-
tional factors, only typical ranges of property values
are given in this section. Factors that determine the
actual values in any case are analyzed in more detail
in subsequent chapters.
Atterberg Limits
ENGINEERING PROPERTIES OF CLAY MINERALS 99
4. The type of adsorbed cation has a much greater
influence on the high plasticity minerals (e.g.,
montmorillonite) than on the low plasticity min-
erals (e.g., kaolinite).
5. Increasing cation valence decreases the liquid
limit values of the expansive clays but tends to
increase the liquid limit of the nonexpansive min-
erals.
6. Hydrated halloysite has an unusually high plastic
limit and low plasticity index.
7. The greater the plasticity the greater is the shrink-

l
age on drying (the lower the shrinkage limit).

ria
Plasticity values for different clay minerals are listed
in Table 4.5 in terms of ranges in the liquid, plastic,
and shrinkage limit values. Most of the values were Particle Size and Shape
determined using samples composed of particles finer Different clay minerals occur in different size ranges
than 2 m. Several general conclusions can be made (Table 3.6) because mineralogical composition is a ma-

ate
concerning the Atterberg limits of the clay minerals.
1. The liquid and plastic limit values for any one
clay mineral species may vary over a wide range.
2. For any clay mineral, the range in liquid limit
values is greater than the range in plastic limit
values.
dM
3. The variation in values of liquid limit among dif-
ferent clay mineral groups is much greater than
the variation in plastic limits.
hte
jor factor in determining particle size. There is some
concentration of different clay minerals in different
bands within the clay size range (less than 2 m), as
indicated in Table 4.7. The shapes of the most common
clay minerals are platy, except for halloysite, which
occurs as tubes (Fig. 3.21). Particles of kaolinite are
relatively large, thick, and stiff (Fig. 3.13). Smectites
are composed of small, very thin, and filmy particles
(Fig. 3.25). Illites are intermediate between kaolinite
and smectite (Fig. 3.29) and are often terraced and thin

Table 4.7 Mineral Composition of Different Particle Size

Ranges in Soils

Particle
Size Predominating Common Rare
rig

(m) Constituents Constituents Constituents

0.1 Montmoril- Mica inter- Illite (traces)
lonite mediates
Beidellite
py

0.1–0.2 Mica inter- Kaolinite Illite

mediates Montmo- Quartz
rillonite (traces)
0.2–2.0 Kaolinite Illite Quartz
Co

Mica inter- Montmo-

mediates rillonite
Micas Feldspar
Halloysite
2.0–11.0 Micas Quartz Halloysite
Illites Kaolinite (traces)
Feldspars Montmo-
rillonite
(traces)
From Soveri (1950).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

100 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

at the edges. Attapulgite, owing to its double silica

chain structure, occurs in lathlike particle shapes (Fig.
3.31).

Hydraulic Conductivity (Permeability)

Mineralogical composition, particle size and size dis-
tribution, void ratio, fabric, and pore fluid character-
istics all influence the hydraulic conductivity. This
property is considered in detail in Chapter 9. Over the
normal range of water contents (plastic limit to liquid
limit), the hydraulic conductivity of all the clay min-

l
erals is less than about 1 ⫻ 10⫺7 m/s and may range

ria
to values less than 1 ⫻ 10⫺12 m/s for some of the
monovalent ionic forms of smectite minerals at low
porosity. The usual measured range for natural clay
soils is about 1 ⫻ 10⫺8 to 1 ⫻ 10⫺10 m/s. For clay
Figure 4.15 Ranges in effective stress failure envelopes for
pure clay minerals and quartz (from Olson, 1974). Reprinted
with permission of ASCE.

ate
minerals compared at the same water content, the hy-
draulic conductivities are in the order smectite (mont-
morillonite) ⬍ attapulgite ⬍ illite ⬍ kaolinite.

Shear Strength
There are many ways to measure and express the shear
dM
strength of a soil, as described in most geotechnical
engineering textbooks. In most cases, a Mohr failure
envelope, where shear strength (usually peak, critical
state, or residual) is plotted as a function of the direct
effective stress on the failure plane, or a modified Mohr
diagram, in which maximum shear stress is plotted ver-
hte

sus the average of the major and minor principal ef-

fective stresses at failure, is used. A straight line is fit
to the resulting curve over the normal stress range of
interest and the shear strength  is given by an equation
of the form Figure 4.16 Strength envelopes for a range of soil types
(from Bishop, 1966).
rig

 ⫽ c ⫹  n tan  (4.13)

where  n is the effective normal stress on the shear

plane, c is the intercept for  n equals zero, often called grained the soil and the greater the amount of clay, the
the cohesion, and  is the slope, usually called the smaller the inclination of the failure envelope.
py

friction angle. From a number of studies [e.g., Hvorslev (1937,

Effective stress strength envelopes are useful for re- 1960), Gibson (1953), Trollope (1960), and Schmert-
lating strength to composition. Zones that encompass mann and Osterberg (1960), and Schmertmann
the effective stress failure envelopes, based on peak (1976)], it has been believed that the total strength of
Co

strength, for pure clay minerals and quartz are shown
in Fig. 4.15. The increase in shear strength with in-
crease in effective stress, that is, the friction angle, is
greatest for the nonclay mineral quartz, followed in
descending order by kaolinite, illite, and montmoril-
lonite. The ranges in the position of a failure envelope
for a given mineral result from differences in such fac-
tors as fabric, adsorbed cation, pH, and overconsoli-
dation ratio. A similar pattern of failure envelopes for
some natural soils is shown in Fig. 4.16. The finer
a clay is composed of two distinct parts: a cohesion
that depends only on void ratio (water content), and a
frictional contribution, dependent only on normal ef-
fective stress. Evaluation of these two parts was done
by measurement of the strength of two samples both
at the same void ratio or water content, but at different
levels of effective stress. This condition is obtained by
using one normally consolidated and one overconsol-
idated sample. The strength parameters determined in
this way, often termed the Hvorslev parameters or true

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com


cohesion and true friction, show increasing cohesion
and decreasing friction with increasing plasticity and
activity of the clay.
However, two samples of the same clay at the same
void ratio but different effective stresses are known to
have different structures, as discussed in Chapter 8.
Thus, they are not equivalent, and the strength tests
measure the effects of both effective stress and struc-
ture differences. Furthermore, tests over large ranges
of effective stress show that actual failure envelopes
are curved in the manner of Fig. 4.16 and that the
l
cohesion intercept is either zero or very small, except
ria
for cemented soils. Thus, a significant true cohesion,
if defined as strength in the absence of normal stress
on the failure plane, does not exist in the absence of
chemical bonding. These considerations are discussed
in more detail in Chapter 11.
ate
dM
hte
rig
py
Co
Figure 4.17 Residual friction angles for clay–quartz mixtures and natural soils (from Ken-
ney, 1967).
Copyright © 2005 John Wiley & Sons
ENGINEERING PROPERTIES OF CLAY MINERALS 101
Even the largest of the friction angle values for clay
minerals is significantly less than the residual value for
cohesionless soils, wherein values of drained friction
angle are generally in the range of 30 to 50. The
residual strengths of some quartz–clay mixtures are
shown in Fig. 4.17. If each mineral were an equally
important contributor to strength, then the curve for a
given mixture should be symmetrical about the 50 per-
cent point, as is the case for kaolinite and hydrous mica
with no salt in the pore water. In the other mixtures,
however, the clay phase begins to dominate at clay
contents less than 50 percent. This is because with ex-
pansive clay minerals (montmorillonite) or flocculated
fabrics (30 g salt/liter) the ratio of volume of wet clay
to volume of quartz is greater than the ratio of dry
volumes. It is further illustration of the dominating in-
fluence of the clay phase discussed earlier.
Retrieved from: www.knovel.com
102 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES
Compressibility
The compressibility of saturated specimens of clay
minerals increases in the order kaolinite ⬍ illite ⬍
smectite. The compression index Cc, which is defined
as the change in void ratio per 10-fold increase in con-
solidation pressure, is in the range of 0.19 to 0.28 for
kaolinite, 0.50 to 1.10 for illite, and 1.0 to 2.6 for
montmorillonite, for different ionic forms (Cornell
University, 1950). The more compressible the clay, the
more pronounced the influences of cation type and
electrolyte concentration on compressibility.
l
Compression index values for a number of different
ria
natural clays are shown in Fig. 4.18 as a function of
plasticity index (Kulhawy and Mayne, 1990). The val-
ues for pure clays plot generally within the defined
ranges in Fig. 4.18. The compression index for un-
ate
loading and reloading is about 20 percent of the value
for virgin compression.
As both compressibility and hydraulic conductivity
are strong functions of soil composition, the coefficient
of consolidation cv is also related to composition be-
cause cv is directly proportional to hydraulic conduc-
dM
tivity and inversely proportional to the coefficient of
compressibility.4 Values of cv determined in one study
4
The coefficient of compressibility av is the negative of the rate of
change of void ratio with effective stress.
hte
rig
py
Co
Figure 4.18 Compression and unload–reload indices as a function of plasticity index (from
Kulhawy and Mayne, 1990). Reprinted with permission from EPR1.
Copyright © 2005 John Wiley & Sons
(Cornell University, 1950) were in the ranges of 0.06
⫻ 10⫺8 to 0.3 ⫻ 10⫺8 m2 /s for montmorillonite, 0.3 ⫻
10⫺8 to 2.4 ⫻ 10⫺8 m2 /s for illite, and 12 ⫻ 10⫺8 to
90 ⫻ 10⫺8 m2 /s for kaolinite. Coefficients of consoli-
dation for kaolinite, illite, montmorillonite, halloysite,
and two-mineral mixtures of these clays ranged from
1 ⫻ 10⫺8 m2 /s for pure montmorillonite to 378 ⫻ 10⫺8
m2 /s for pure halloysite in another study (Kondner and
Vendrell, 1964). Individual minerals did not influence
the coefficient of consolidation in direct proportion to
the amounts present.
Approximate ranges of the coefficient of consoli-
dation for natural clays are given in Fig. 4.19. The
above values for pure clays and clay mineral mixtures
are within the same general ranges. One conclusion
that can be drawn from the comparability of compres-
sion index and coefficient of consolidation values for
natural clays with those for pure clays is that the clay
phase dominates the compression and consolidation
behavior, with the nonclay material playing a passive
role as relatively inert filler.
Swelling and Shrinkage
The actual amount of volume change of a soil in re-
sponse to a change in applied stress depends on the
environmental factors listed in Section 4.1 as well as
on the cation type, electrolyte type and concentration,
Retrieved from: www.knovel.com
 ENGINEERING PROPERTIES OF CLAY MINERALS 103

p.s.i.)
re - 1
p.s.i.)

ressu
re - 1

ge P
ressu

char
ge P

(Sur

.)
.s.i
l
char

1p
ria
e-
(Sur

sur
res
Figure 4.19 Coefficient of consolidation as a function of

eP
liquid limit (from NAVFAC, 1982).

rg
ha
c .

i.)
ur p.s
(S 4

ate
6.9
e-
sur
and pore fluid dielectric constant. However, the poten- Pres
e
harg
tial total amount of swell or shrinkage is determined (Surc
by the type and amount of clay. From a consideration
of the clay mineral structures and interlayer bonding
dM
(Chapter 3), it would be expected that smectite and
vermiculite should undergo greater volume changes on
wetting and drying than do kaolinite and hydrous mica. Figure 4.20 Four correlations between swelling potential
Experience indicates clearly that this is indeed the and plasticity index (from Chen, 1975).
case. In general, the swelling and shrinking properties
of the clay minerals follow the same pattern as their
plasticity properties, that is, the more plastic the min-
hte

eral, the more potential swell and shrinkage. Illustra-
tions of the influences of adsorbed cation type and pore
fluid composition are given in Chapter 10 and by Srid-
haran (2002).
Because of the many problems encountered in the
performance of structures founded on high volume
rig
sure during swelling is clearly shown. The tests by
Seed et al. (1962b) were done using artificial mixtures
of sand and clay minerals compacted at optimum water
content using Standard AASHTO compactive effort al-
lowed to swell under a surcharge pressure of 1 psi (7
kPa). The measurements by Holtz and Gibbs (1956)

change soils, numerous attempts have been made to
develop reliable methods for their identification. The
most successful of these are based on the determina-
tion of some factor that is related directly to the clay
mineral composition, such as shrinkage limit, plasticity
py
were made using both undisturbed and remolded sam-
ples allowed to swell from an air-dry state to saturation
under a surcharge of 1 psi (7 kPa).
The results of the tests on artificial sand–clay min-
eral mixtures obtained by Seed et al. (1962b) correlate

index, activity, and percentage finer than 1 m. well with compositional factors that reflect both the
Simple, unique correlations between swell or swell type and amount of clay, that is, the activity A, defined
pressure and these parameters that reflect only the type as PI/ C, and the percent clay size C (% ⬍ 2 m),
and amount of clay are not possible because of the according to
Co

strong dependence of the behavior on initial state

(moisture content, density, and structure) and the other
environmental factors. This is illustrated by Fig. 4.20,
which shows four different correlations between swell-
ing potential and plasticity index (Chen, 1975). The
two curves showing the Chen correlations were ob-
tained for different natural soils compacted to dry unit
weights between 100 and 110 pounds per cubic foot
(15.7 and 17.3 kN/m3) at water contents between 15
and 20 percent. The large influence of surcharge pres-
S ⫽ 3.6 ⫻ 10⫺5 A2.44C 3.44 (4.14)
where S is the percent swell for samples compacted
and tested as indicated above. A chart based on this
relationship is shown in Fig. 4.21. For compacted nat-
ural soils the swelling potential could be related to the
plasticity index with an accuracy of 35% according
to

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

104 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

In general, the greater the organic content and the

wetter and more plastic the clay, the more pronounced
is the time-dependent behavior. Both the type and
amount of clay are important, as indicated, for exam-
ple, by the variation of creep rate with clay content for
three different clay mineral–sand mixtures, as shown
in Fig. 4.22. In these tests, environmental factors were
held constant by preparing all specimens to the same
initial conditions (isotropic consolidation of saturated
samples to 200 kPa) and application of a creep stress
equal to 90 percent of the strength determined by a

l
normal strength test. The variation in creep rate for

ria
these specimens as a function of plasticity index is
shown in Fig. 4.23. The correlation is reasonably
unique because the plasticity index reflects both the
type and amount of clay.

ate
4.9 EFFECTS OF ORGANIC MATTER
Organic matter in soil may be responsible for high
Figure 4.21 Classification chart for swelling potential (mod- plasticity, high shrinkage, high compressibility, low
ified from Seed et al., 1962b).
dM
S ⫽ 2.16 ⫻ 10⫺3 (PI)2.44 (4.15)

Somewhat different relationships have been found to

better classify the swell potential of some soils, and no
hte

single relationship is suitable for all conditions. Thus,

while the above relationships and plots such as Figs.
4.20 and 4.21 illustrate the influences of compositional
factors and provide preliminary guidance about the po-
tential magnitude of swelling, reliable quantification of
swell and swell pressure in any case should be based
rig

on the results of tests on representative undisturbed

samples tested under appropriate conditions of con-
finement and water chemistry.
py

Time-Dependent Behavior
Different soil types undergo varying amounts of time-
dependent deformations and stress variations with
time, as exhibited by secondary compression, creep,
Co

and stress relaxation. The potential for these phenom-

ena depends on compositional factors, whereas the ac-
tual amount in any case depends on environmental
factors. For example, it is known that retaining walls
with wet clay backfills must be designed for at-rest
earth pressures because of stress relaxation along a po-
tential failure plane that results in increased pressure
on the wall. On the other hand, if dry clay is used, and
if it is maintained dry, then designs based on active
pressures are possible because time-dependent in- Figure 4.22 Effect of amount and type of clay on ‘‘steady-
creases in pressure will be negligible. state’’ creep rate (see Chapter 12).

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com


l
ria
ate
dM
hte
Figure 4.23 Relationship between clay content, plasticity in-
dex, and creep rate.
rig
hydraulic conductivity, and low strength. Soil organic
matter is complex both chemically and physically, and
many reactions and interactions between the soil and
the organic matter are possible (Oades, 1989). It may
occur in any of five groups: carbohydrates; proteins;
py
fats, resins, and waxes; hydrocarbons; and carbon. Cel-
lulose (C6H10O5) is the main organic constituent of
soil. In residual soils organic matter is most abundant
in the surface horizons. Organic particles may range
Co
down to 0.1 m in size. The specific properties of the
colloidal particles vary greatly depending upon parent
material, climate, and stage of decomposition.
The humic fraction is gel-like in properties and neg-
atively charged (Marshall, 1964). Organic particles can
strongly adsorb on mineral surfaces, and this adsorp-
tion modifies both the properties of the minerals and
the organic material itself. Soils containing significant
amounts of decomposed organic matter are usually
characterized by a dark gray to black color and an odor
of decomposition. At high moisture contents, decom-
Copyright © 2005 John Wiley & Sons
CONCLUDING COMMENTS 105
posed organic matter may behave as a reversible swell-
ing system. At some critical stage during drying,
however, this reversibility ceases, and this is often
manifested by a large decrease in the Atterberg limits.
This is recognized by the Unified Soil Classification
System, which defines an organic clay as a soil that
would classify as a clay (the Atterberg limits plot
above the A line shown in Fig. 4.13) except that the
liquid limit value after oven drying is less than 75 per-
cent of the liquid limit value before drying (ASTM,
1989).
Increasing the organic carbon content by only 1 or
2 percent may increase the limits by as much as an
increase of 10 to 20 percent in the amount of material
finer than 2 m or in the amount of montmorillonite
(Odell et al., 1960). The influences of organic matter
content on the classification properties of a soft clay
from Brazil are shown in Fig. 4.24.
The maximum compacted densities and compressive
strength as a function of organic content of both nat-
ural samples and mechanical mixtures of inorganic
soils and peat are shown in Figs. 4.25 and 4.26, re-
spectively. Both the compacted density and strength
decrease significantly with increased organic content
and the relationships for natural samples and the
mixtures are about the same. Increased organic content
also causes an increase in the optimum water content
for compaction.
The large increase in compressibility as a result of
high organic content in clay is illustrated by the data
in Fig. 4.27 for the clay whose classification properties
are shown in Fig. 4.24. In Fig. 4.27 CR is the com-
pression ratio, defined as CC /(1 ⫹ e0) expressed as a
percentage, and C is the secondary compression ratio,
defined as the change in void ratio per 10-fold increase
in time after the end of primary consolidation.
The effect of organic matter on the strength and
stiffness of soils depends largely on whether the or-
ganic matter is decomposed or consists of fibers that
can act as reinforcement. In the former case, both the
undrained strength and the stiffness, or modulus, are
usually reduced as a result of the higher water content
and plasticity contributed by the organic matter. In the
latter, the fibers can act as reinforcements, thereby in-
creasing the strength.
4.10 CONCLUDING COMMENTS
Knowledge of soil composition is a useful indicator of
the probable ranges of geotechnical properties and
their variability and sensitivity to changes in environ-
mental conditions. Although quantitative values of
properties for analysis and design cannot be derived
from compositional data alone, information on com-
Retrieved from: www.knovel.com
106 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

l
ria
ate
dM
Figure 4.24 Influence of organic content on classification properties of Juturnaiba organic
clay, Brazil (from Coutinho and Lacerda, 1987).

position can be helpful for explaining unusual behav- and more direct evaluation of their significance is
hte

ior, identification of expansive soils, selection of
sampling and sample handling procedures, choice of
soil stabilization methods, and prediction of probable
future behavior.
For example, if it is known that a soil to be used in
earthwork construction contains either hydrated halloy-
rig
needed. Examples of some soil types in which these
factors may be especially important are decomposed
granite, tropical residual soils, volcanic ash soils, col-
lapsing soils, loess, and carbonate sand, as discussed
in more detail by Mitchell and Coutinho (1991).

site, organic matter, or expansive minerals, then air-

drying laboratory samples prior to testing is likely to QUESTIONS AND PROBLEMS
result in erroneous data on mechanical properties and
1. Show that the loosest and densest packings of uni-
must be avoided. If a soil contains a large amount of
form size particles give void ratios of 0.91 and 0.34,
active clay minerals, then it can be anticipated that
py

respectively. What is the coordination numbers

properties will be sensitive to changes in chemical en-
(number of particle contacts for each particle) for
vironment. Compositional data on the soil and pore
each packing?
water are useful to estimate the dispersion and erosion
potential of a soil (Chapter 8) and the risk of instability 2. Explain why smaller particles are stronger than
Co

as a result of leaching and solutioning processes.
In many cases, the effects of composition on behav-
ior are reflected by information on particle size, shape,
and size distribution of the coarse fraction, and the
Atterberg limits of the fine fraction. On large projects
and whenever unusual behavior is encountered, how-
ever, compositional data are valuable aids for interpre-
tation of observations. Furthermore, the influences of
compositional and structural factors are not always ad-
equately reflected by the usual classification properties,
larger particles and why angular particles are more
susceptible to breakage than round particles.
3. Using Figs. 4.8 and 4.11, show how the maximum
and minimum void ratio changes with applied load
as particles progressively break and the coefficient
of uniformity Cu increases. Plot the data in e–log
v space and discuss the result.
4. Using Eq. (4.8), derive a relationship between C
(the percentage of clay) versus w (water content)

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

 QUESTIONS AND PROBLEMS 107

l
ria
ate
dM Figure 4.27 Effect of organic content on the compressibility
properties of Juturnaiba organic clay, Brazil (from Coutinho
and Lacerda, 1987).

Figure 4.25 Maximum dry density as a function of organic

for different values of eG (the void ratio of the gran-
content for a natural soil and soil–peat mixtures (from Frank-
lin et al., 1973). Reprinted with permission of ASCE.
ular phase). Discuss the sensitivity of eG on sand–
clay mixture packing. What happens if silt is mixed
hte

instead of clay?
5. Using the reported undrained shear strengths at liq-
uid limit and plastic limit, derive a relationship be-
tween the compression index Cc and plasticity index
PI. Assume that the ratio of undrained shear
strength to vertical effective stress, su /  v, is 0.3.
rig

Compare the result with the data presented in Fig.

4.18.
6. Assuming the thickness of adsorbed water layer is
100 Å, estimate the amount of free water per gram
py

of clay for the following conditions and discuss the

results:
a. Montmorillonite at its liquid limit with mono-
valent adsorbed cations (specific surface ⫽ 840
Co

m2 /g of dry clay), liquid limit ⫽ 900 percent

Figure 4.26 Unconfined compressive strength as a function
of organic content for a natural soil and soil–peat mixtures
(from Franklin et al., 1973). Reprinted with permission of
ASCE.
b. Montmorillonite at its plastic limit with mono-
valent adsorbed cations (specific surface ⫽ 840
m2 /g of dry clay), plastic limit ⫽ 100 percent
c. Montmorillonite at its liquid limit with divalent
adsorbed cations (specific surface ⫽ 50 m2 /g of
dry clay), liquid limit ⫽ 100 percent
d. Montmorillonite at its plastic limit with divalent
adsorbed cations (specific surface ⫽ 50 m2 /g of
dry clay), plastic limit ⫽ 50 percent

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com

108 4 SOIL COMPOSITION AND ENGINEERING PROPERTIES

e. Kaolinite at its liquid limit (specific surface ⫽
15 m2 /g of dry clay), liquid limit ⫽ 70 percent
f. Kaolinite at its plastic limit (specific surface ⫽
15 m2 /g of dry clay), plastic limit ⫽ 30 percent
7. By examining the data presented in Figs. 4.24 and
4.29, discuss why organic clays exhibit larger com-
pressibility compared to inorganic clays (see Fig.
4.18).
8. Assume that you are able to determine accurate,
reliable quantitative values for all details of the min-
soil—discuss the strengths and weaknesses of the
Unified Soil Classification System (USCS) in pro-
viding a clear and unambiguous picture of the prob-
able behavior of the following soil types. In
developing your answer, be specific concerning
what is measured and the terms of reference used
in the USCS and what is most important in deter-
mining any property being discussed. (Note: Some
of the information in Chapter 8 may be useful in
developing your answer to this question.)
a. Clean sand

l
eralogical, chemical, and biological constituents of b. Decomposed granite

ria
a given soil. All particle sizes, shapes, and distri- c. Calcareous sand
butions are also known. Speculate on your ability
d. Organic silt
to predict the volume change, strength, and per-
meability properties of this soil over a range of wa- e. Expansive clay
f. Glacial till

ate
ter contents. Give reasons for why you would have
low or high confidence in your predictions. g. Loess
9. In light of what is known about the dependence of h. Dispersive clay
engineering properties on soil composition—both i. Volcanic ash
of the particles and of the other phases present in a
dM j. Estuarine mud
hte
rig
py
Co

Copyright © 2005 John Wiley & Sons Retrieved from: www.knovel.com