Abstract
A computational design tool is presented which permits the optimal allocation of tolerances for mechanical and electrical components and assemblies. The basic approach involves the coupling of a nontraditional optimization method with a Monte Carlo based tolerance analysis. The objective is to determine the tolerance range value to assign to each nominal dimension which minimizes the production cost of the component or assembly while simultaneously meeting all critical dimensional and functional constraints imposed upon the design. A discretization of possible tolerance range values is performed and a global search is conducted by a genetic algorithm. Both the fine and course grain performance of the combined algorithm is demonstrated on a series of test problems ranging from a simple assembly of blocks to several real mechanical design problems. Solutions generated on problems taken from the literature indicate superior performance to existing techniques. Extensions which would allow for a complete optimal dimensional management environment are examined.
