Civil Engineering Reference
In-Depth Information
27.4 Purposes of modelling
It would be very convenient to be able to construct and test a scale model that
reproduced all the significant features of the behaviour of a proposed construction.
Unfortunately, however, this is not generally possible for a variety of reasons. The
principal difficulties are rather like those associated with ground investigations and
laboratory testing (see Chapters 7 and 17) and are due to test samples not being fully
representative of the soil in the ground. It is also difficult to model geological his-
tory and complex construction sequences. Instead, geotechnical models are usually
constructed and tested to meet specific objectives.
The principal purposes and categories of geotechical modelling were discussed by
Taylor (1987) and these are as follows.
(a) Mechanistic studies
The basic methodology of engineering design is that engineers imagine all the possi-
ble ways in which a proposed construction may fail or distort and they then carry
out analyses that demonstrate that it will perform satisfactorily in any of these ways.
Sometimes major failures occur when the construction finds some other way to fail or
distort. For example, in the upper bound and limit equilibrium methods described in
Chapters 19 and 20 it is necessary to define compatible mechanisms and the solutions
depend on the mechanisms chosen. For relatively simple cases it is usually possible to
choose the critical mechanisms from previous experience, but in novel and complex
cases they may not be so obvious. In these cases relatively simple model tests may be
carried out simply to observe qualitatively the way in which the structure distorts and
fails, thus indicating the most appropriate analyses.
(b) Validation of numerical analyses
Design of geotechnical structures often requires complex numerical analyses using
finite element, or similar, methods with non-linear and inelastic soil behaviour (see
Chapter 13). These analyses are highly complex and before they are applied in design
studies they should be tested against exact analytical solutions or against observations
of the real events. Observations from relatively simple model tests can be used to test
numerical analyses. The models should be similar to the proposed construction but,
since the models are used only to calibrate the analyses, they need not reproduce all
the details of the prototype.
(c) Parametric studies
Another important procedure in design studies involves examining alternative con-
struction details and investigating the consequences of different design assumptions.
Furthermore, standard design codes and charts rely on studies of many different alter-
natives. Normally parametric studies are carried out using analytical or numerical
methods, but model studies have a role to play in parametric studies, either on their
own or together with other methods.
