Méthodologie d'analyse et d'optimisation des tolérances dans un contexte de conception intégrée: TOL-ANALYSES

Abstract

Real parts of a batch are theoretically all identical but physically all different. Their dimensions cannot be the same. A dimension is tolerable if its deviation remains close enough to the ideal value. This deviation is limited by two values, either bounds of an interval or the mean and the variance of a batch. Usually, we define two extreme bonds of dimensions, called tolerances that must be determined wisely. A tolerancing uses syntactic and grammatical information to provide a sense. The aim of this article is to propose and validate a methodology for assistance in the selection and verification of tolerances. This methodology is based on two complementary approaches: the method at worst case and the Monte Carlo method. An implementation computer showed the feasibility of the tool as well as the major contribution of the methodology proposed in aid to the specification, optimization and verification of tolerance, taking account of the conditions of manufacture. © AFM, EDP Sciences 2009.