Soils and model tests

Abstract Two main uses of model testing in soil mechanics are considered. The first is to examine, usually on only a qualitative basis, at a reduced scale the assumptions that have been made in theoretical analyses of prototype problems, the object being to develop analysis and models side by side with a view to improvement of the former. An outline is then given of current methods of analysis based only on failure conditions and of the use of new methods developed at Cambridge of investigating the behaviour of soils in models at all stress levels. Thereafter attention is confined to the second use of models in which attempts are made to satisfy the conditions of similarity so that prototype behaviour can be predicted from model data. These conditions are derived. Simulation of soils by other media is discussed but is not considered to be a rewarding procedure. It is then shown how the Cambridge approach to the mechanical behaviour of soils has enabled the formulation of a useful law in model testing for work-hardening soils, and similar granular media, at all working loads. This law states the conditions under which the same soil may be used in model and prototype. Finally the potentialities of centrifugal model testing are discussed; they provide an opportunity of satisfying simultaneously this law and the conditions of similitude.