Managing Uncertainty of Product Data

Abstract

Uncertainty is inherent to product design since a design process starts with needs defined semantically and ends with precise product and process technical specifications. Managing uncertainty of product and process data is essential to conceptual design since data are still imprecise and to concurrent engineering since it guaranties data consistency at any time and an optimal paralleling of development and production activities [1–2]. After noting the poverty of tools and techniques in conceptual design of mechanical systems, the design process is presented as a process of topological and dimensional variability narrowing. Notably, a propagation of the reduction of design variable domains must occur towards the performance domains as soon as possible and conversely, so as to respect the principles of data consistency and simultaneous engineering. Then, one briefly shows that probabilistic simulations and fuzzy inference techniques can respond in a static way to the sole management of the dimensional variability. A state-of-the-art of constraint programming (CP) techniques especially over continuous domains is performed. One believes that these latter techniques are the most promising ones, in the long term, for supporting a breakthrough to conceptual design and concurrent engineering practices.