A Novel Tolerance Design Approach to Manufacturing and Quality Loss Cost Optimization in Mechanical Assemblies

Abstract

Tolerance design has a profound impact on the manufacturability, ease of assembly, performance, quality and cost of mechanical components. However, assigning appropriate tolerances is still far from being a trivial engineering task. The optimization of dimensional and geometric tolerances for components that control functional dimensions in mechanical assemblies is receiving continuous attention from researchers. In this paper a novel tolerance design approach is introduced. The basic concept of the proposed approach is the transformation of an unconstrained optimization problem of multiple decision variables to a constrained optimization problem of a single variable in order to improve the convexity of the set of feasible solutions. Through the development of a Quasi-Newton algorithm, our objective is the optimization of both the manufacturing cost of each component and the quality loss cost of the mechanical assembly. A step-by-step mathematical formulation of the approach is presented by using the widely referenced example of the overrunning clutch assembly. The experimental results obtained for this application example are directly compared with alternative tolerance synthesis techniques that are found in technical literature, illustrating that the proposed approach is reasonable and effective.