A computational design tool will be developed which will allow for optimal tolerance allocation for mechanical and electrical components. The basic approach is to couple a nontraditional optimization method with a Monte Carlo based tolerance analysis code. The ability to allocate tolerances in an optimal fashion will allow an installed capital base to be utilized in the most productive fashion. A successful application of the technique will have a dramatic impact on reducing manufacturing cost and product cycle times as well as increasing productivity, quality and customer satisfaction. Current tolerance analysis used in industry does not provide the ability to allocate tolerances in an optimal fashion. Traditional nonlinear programming methods are not robust enough to solve the complex problem resulting from the consideration of a "real" component or assembly. The combination of the genetic algorithm with a tolerance analysis package offers an attractive alternative which will run efficiently on a parallel or distributed computing network. Initial consideration will be placed on discrete tolerance limit optimization. The applicability to modifying nominal dimensions and assembly sequence will also be investigated.
