Documentos de Académico
Documentos de Profesional
Documentos de Cultura
5.1 INTRODUCTION
Although soils are composed of discrete soil particles and particle groups, a soil mass is almost
always treated as a continuum for engineering analysis and design. Nonetheless, the specific
values of properties such as strength, permeability. and compressibility depend on the size and
shape of the particles, their arrangements, and the forces between them. Thus, to understand
a property requires knowledge of these factors. Furthermore, new theories of particulate
mechanics and computational methods based on these theories are now becoming available.
With these theories and methods it may ultimately be possible to predict the mechanical
behavior of soil masses in terms of the characteristics of the particles themselves, although at-
taining this goal appears somewhat far off. Particle arrangements in soils remained largely un-
known until suitable optical, X-ray diffraction, and electron microscope techniques made direct
observations possible starting in the mid-1950s. Interest then centered mainly on clay particle
arrangements and their relationships to mechanical properties. In the late 1960s, knowledge
expanded rapidly, sparked by improved techniques of sample preparation and the de-
velopment of the scanning electron microscope. In the early 1970s attention was directed also
at particle arrangements in cohesionless soils. From this work came a realization that
characterization of the properties of sands and gravels cannot be done in terms of density or
relative density alone, as had previously been thought. Particle arrangements and stress
history must be considered in these materials as well. In the 1970s and 1980s, micromechanics
theories were developed that aimed to relate microstructure to macroscopic behavior. Various
homogenization tech niques that incorporate small-scale features such as inhomogeneity and
microfractures into continuum models became available (Mura, 1987; Nemat-Nasser and Hori,
1999). Increased computational speeds allowed simulation of an assembly of individual soil
particles by modeling particle contact behavior, and this led to the development of numerical
methods such as the discrete/distinct element method and contact dynamics (Cundall and
Strack, 1979; Moreau, 1994; Cundall, 2001). In the carly developments, simulations were
limited to an assembly of two-dimensional circular disks. However, it is now possible to
perform Simulations with various three-dimensional particleshapes, complex contact models,
and pore fluid interactions. These "digital"-type studies offer possibilities for systematic
investigation of soil fabric effects on mechanical properties in comparison to "laboratory""-
type studies, which contain inherent errors associated with measuring soil fabrics of different
specimens.
Furthermore, mechanical responses under the stress paths that are difficult to apply in the
laboratory can be investigated using distinct element methods. Other innovations in the past
two decades have led to improved material measurement techniques and their interpretation
using computers. These include the environmental scanning electron microscopy (ESEM),
nanoindentation and probing, complex digital image analysis, magnetic resonance imaging
(MRI), X-Ray tomography, and laser-aided tomography. Some of them have been used to
characterize the microscopic properties of soils (Oda and Iwashita, 1999). The more
established methods for studying and, where possible, quantifying the arrangements of par-
ticles, particle groups, and voids in different soils are described and illustrated in this chapter.
Some ele ments and applications of the newer methods are introduced in later chapters.
Glass balls allowed to fall freely form an anisotropic assembly, with the balls
tending to arrange themselves in chains (Kallstenius and Bergau, 1961). The
number of balls per unit area in contact with a vertical plane can be different
from the number in contact with a horizontal plane. The same behavior is
observed for sand pluviated through air and water. Spontaneous segregation
and stratification has been observed when granular mixtures of particles of two
different predominant sizes are dumped into a pile (Makse et al., 1997;
Fineberg, 1997). When a mixture of sizes is poured into a pile, the larger
particles tend to accumulate near the base. Makse and co-workers (1997)
experiments produced the interesting additional result that if the large grains in
a binary mixture have a greater angle of repose than the small grains, then the
mixture stratifies into alternating layers of small and large grains. If the small
grains have a larger angle of repose than the large grains, then segregation
with- out stratification results. This type of behavior is relevant to such
geoengineering problems as the stability of dumped mine waste piles,
geological formations susceptible to static liquefaction, and the processing and
transport of granular materials.
Particle Packings in Granular Soils
Particle sizes in soil vary, and as a result, smaller particles can occupy pore
spaces between larger particles.
This results in a lendency toward higher densities and lower void ratios than for
uniform spheres. On the other hand, irregular particle shapes produce a
tendency toward lower densities and higher porosities and void ratios. The net
result is that the range of porosities and void ratios in real soils with single-grain
fabrics may not be much different trom that for unitorm spheres shown by the
values in Table 5.1, that is, porosity in the range of 26 to 48 percent and void
ratio in the range of 0.35 to 0.91. This is illustrated by the data in Table 5.2. The
lower values of porosity and density and higher unit weight for silty sand and
gravel can be attributed to silt filling the large voids between the gravel particles.
Many studies have shown that a given cohesionless soil can have different
fabrics at the same void ratio or relative density. Characterization of this fabric
can be done in terms of grain shape factors, grain orientations, and interparticle
contact orientations (Lafeber, 1966; Oda. 1972a: Mahmood and Mitchell, 1974;
Mitchell et al., 1976). More recently, application of image analysis techniques
(Section 5.8) has led to better understanding and quantification of fabric
features.
The orientation of grains in a sand deposit can be described in terms of the
inclination of the particle axes to a set of reference axes. For example, the
orientation of the particle shown in Fig. 5.8 can be described by the angles a
and B. In most studies, however, a thin section is studied to give the
orientations of apparent long axes. The long axes of particles are referred to a
single horizontal reference axis by an angle 6. The spatial orientation of the thin
section it self with respect to the sample and to the field deposit is also an
essential part of the fabric description. Orientations of long axes for a large
number of grains can be expressed by a histogram or rose diagram. A
frequency histogram for a sand having a mean axial ratio equal to 1.65 and
placed by tapping the side of a vertical, cylindrical mold is shown in Fig. 5.9.
The orientation of each grain was assigned to one of the 15 intervals between 0'
and 180'. The V-section refers to a thin section from a vertical plane (oriented
parallel to the cylinder axis). The H-section refers to orientations in the
horizontal plane. Orientations of long axes in the vertical plane for two samples
of well-graded crushed basalt [mean (length)/(width) = 1.64] are shown by the
rose diagrams in Figs. 5.10 and 5.11. In this study, the orientations of at least
400 grains were measured for each sample, and the orientation of each was
assigned to