686 MODELS IN HYDRAULIC ENGINEERING
be employed to model significant changes in rock deformability associated
with such features or with differing rock types.
Deformations under hydrostatic and other loadings are determined
by suitably mounted transducers or dial gauges. Stresses are determined
from strains recorded by surface-bonded strain gauges or strain rosettes at
strategic locations on upstream and downstream faces of the model dam.
Self-weight loads are the most difficult to simulate at model scale.
One technique involves progressively cutting the model down in
stages following completion of all hydrostatic load tests. At each stage or
level in turn, the superincumbent self-weight load is represented through a
system of vertical springs acting on spreader plates on the model. An
alternative technique involves inversion of the model and its immersion in
mercury.
Construction details, e.g. joints, or ‘defects’ such as cracks can be
represented in a sophisticated model, and temperature effects can also be
studied if the scaling laws are further developed. Comprehensive reviews
of structural modelling techniques and their application to specific studies
have been presented by Rydzewski (1963) and Rocha, Serafim and
Azeveda (1961).
It may be noted that plaster–filler mixtures also lend themselves to
simulation of geomechanical problems involving rocks and jointed rock
masses. They have been applied to investigations of this nature associated
with major dam projects, as discussed by Oberti and Fumagalli (1963) and
by Fumagalli (1966).
16.2.3 Modelling of embankments
Application of physical modelling to the study of geotechnical problems
and embankments is severely constrained by the dominance of self-weight
loading and by the complexity and non-uniformity of the range of proto-
type materials, i.e. foundation soils, compacted earthfills and rockfills. The
approach to design for fill dams is in any event very different from that for
concrete dams, focusing on seepage, deformation and stability rather than
upon stress. Physical modelling has therefore been confined to limited
investigations of embankment slopes and of embankments on soft founda-
tions in terms of pore pressure changes, deformation and stability, self-
weight scaling being achieved by mounting the model on the rotating arm
of a large centrifuge. In the case of the Cambridge geotechnical centrifuge
as described by Schofield (1980), model packages of the order of 1000 kg,
could be subjected to accelerations of up to 125gat a radius (rotor arm
length) of 4 m.
Geotechnical models offer many attractions in principle, but in prac-
tice the problems are almost intractable. Natural soils must be used to con-
