| This thesis proposes a multi-objective, mixed integer, non-linear programming model of cellular manufacturing systems (CMS) design to maximize the system reliability and minimize the total system cost simultaneously. The model involves multiple machine types, multiple machines for each machine type, multiple part types, and alternative process routes for each part type. Each process route consists of a sequence of operations. System reliability associated with machines along process routes can be improved by increasing the number of parallel machines subject to acceptable cost. Assuming machine reliability to follow a lognormal distribution, the CMS design problem is to optimally decide the number of each machine type, assign machines to cells, and select, for each part type, the process route with the highest overall system reliability while minimizing the total cost. The total cost consists of the variable cost of manufacturing operations, the inter-cell material handling cost, the penalty cost of machine under-utilization, and machine annuity cost. Genetic algorithm (GA) is proposed as the solution procedure, and is applied to solve this practical-sized CMS design problem. It is shown that, with its characteristics of random selection, crossover, and mutation, GA is capable of finding a heuristic solution within a reasonable amount of computational time. |
