Integrated uncertainty management in parametric design and tolerancing

Abstract

The key purpose of robust design and tolerancing approaches is the management of uncertainty. Against this background, it is not sur prising that there is a large overlap between the basic ideas and concepts in both fields. However, while sharing the same objective, the focus of the corresponding development phases is quite differ ent; that is (i) the determination of solutions that react insensitive, in other words robust, to n oise factors – Robust parametric design; and (ii) the limitation of the effects of manufacturing imprecision by the specification of optimal tolerances – Tolerancing. As a consequence, there also is a significant gap between both concepts. Focusing on the improvement of design solutions, robustness is often related to uncertainty of not known designs or manufacturing processes. Due to the complexity of a largely matured solution, tolerancing tasks are usually based on previously specified, key characteristics or behavior models that are supposed perfect. Therefore, an overview of robust design and tolerancing is used to highlight the deficiencies, and to formalize a new classification of tolerance analysis issues based on the type of uncertainty considered. The proposed framework is based on Dempster-Shafer evidence theory and allows to efficiently perform statistical tolerance analyses under model imprecision.