Robust Tolerance Design With the Integer Programming Approach

Abstract

The quality loss function incorporates the cost of tolerances, however, it does not consider the manufacturing cost and design constraints. In this paper, a stochastic integer programming (SIP) approach is presented for simultaneous selection of tolerances and manufacturing processes. A direct link between the minimum manufacturing cost and the required level of manufacturing yield is established through the process capability index Cpk. As the tolerances in SIP are discrete, the solution generated is acceptable for manufacturing. It is shown that the integer programming models are applicable in the quality loss function and six sigma design approaches. The SIP approach is illustrated with a classical example of nonlinear tolerance design. The comparison of the proposed SIP approach, the Taguchi method, and the conventional mathematical models in tolerance synthesis is presented.