STUDY OF AN ASSEMBLY TOLERANCE ALLOCATION MODEL-BASED MONTE-CARLO SIMULATION

The traditional design methods of assembly tolerance allocation are us ually based on engineers' experience, or the worst on worst tolerance analysis (WOW) method, or the root sum square tolerance analysis (RSS) method. However, the above-mentioned methods, whilst used frequently in the analysis of a single-dimensional chain, are not suitable for th e analysis of geometrical tolerance and multi-dimensional chains. Also , the relationship between tolerance and manufacturing cost is not con sidered and a suitable tolerance allocation based on minimum manufactu ring cost can not be obtained. Some research works have applied linear or non-linear programming methods to optimize the tolerance allocatio n of each part in an assembly. However, the convergence of the solutio n is not ensured. The purpose of this study is to provide an integrate d approach, including tolerance design, manufacturing cost analysis an d multiple chains consideration, using the Monte Carlo method to optim ize the tolerance allocation with minimum cost. The Monte Carlo method , a statistical simulation method, was used to simulate the dimension variance of each part and each dimensional chain. The contribution per centage of each part on each dimensional chain was calculated. Toleran ce cost was chosen as an object function and the tolerance allocation model as a constraint condition, the optimum tolerances of each part b eing obtained by the iteration method. Results of computer simulation for several examples were compared with published data for demonstrati ng the feasibility of the proposed method. It can be concluded that th e tolerance-allocation model combined with a tolerance-cost relationsh ip can provide a very practical and useful approach for design enginee rs. (C) 1997 Elsevier Science S.A.