Optimal tolerance design of mechanical assemblies for economical manufacturing in the presence of alternative machines — a genetic algorithm-based hybrid methodology

Abstract

Traditional practice to tolerance design has been based on a sequential approach to design and manufacturing considerations. Integrated tolerance design involving simultaneous selection of design and manufacturing tolerances was introduced in the last decade. Choice of manufacturing processes (or machines) from among the alternatives, frequently encountered in different stages of realization of individual dimensions, is an important issue in product development. The problem becomes more complicated if the available alternative manufacturing processes (or machines) have a non-overlapping precision range. The optimal tolerance design problem formulated with these issues considering both the cases of overlapping and non-overlapping precision range, based on different stack-up conditions, such as the worst case, root sum square, and Spotts’ modified criteria is the focus of this study. The resulting optimization problem involving a combinatorial, non-linear, non-convex search space cannot be effectively solved for a global solution using conventional optimization techniques. Genetic algorithm, a non-traditional optimization technique, is useful to solve such a complex problem to obtain a near-global solution. This solution may further be improved by making subsequent use of conventional techniques appropriately. In this research a hybrid approach has been proposed based on this philosophy. In the first step, a near-global solution is obtained using the genetic algorithm; in the second step the optimal solution obtained in the first step is refined using the MATLAB-optimization-toolbox. The study has been planned to obtain a better insight into solution of the tolerance design problem, and has been demonstrated with the help of a numerical example.