It is a well known experimental fact that for soils the stress-strain laws are not linearly elastic for the entire range of interest and that their actual behaviour show great variety. Drastic idealizations are therefore needed in order to develop mathematical models for practical applications and, indeed, several proposals, aimed at certain classes of applications, can be found in the literature: hypoelastic models, variable moduli models, Cam Clay model, cap models, isotropic and kinematic hardening models, etc. Yet, none of them can completely describe the behaviour of such complex materials as soils. Another different approach, focused on this paper, is to use neural network concepts to predict the constitutive relationships under specific stress paths; herein, specifically, we selected the residual soil found in the city of Rio de Janeiro, Brazil, and investigated its stress-strain characteristics using laboratory results from direct shear tests. The choice of the neural network, as well as the relationships thus obtained. are presented in the next sections.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £82 +P&P)