Abstract

The use of numerical analyses of boundary values problems has become very frequent in both research and geotechnical practice. There are numerous examples of the successful application of these techniques to even very complex situations. Surprisingly, however, simple geotechnical problems can be found in the literature for which the use of these techniques is problematic, even when the soil behaviour is reproduced by means of very simple constitutive models, such as those based on the theory of perfect plasticity. In this talk it will be shown that many of the problems are related to the formulation of classical failure criteria for geomaterials and that the adoption of a very recent reformulation of two such criteria, namely the Mohr-Coulomb and the Matsuoka-Nakai, allows to perform the analyses without any problem. As numerical analyses are often performed by considering a cross section of an infinite long structure, the link between failure conditions and plane strain plastic collapse will be also discussed and sources of potential problems highlighted.