Machine cell/part family formation considering processing times and machine capacities: a simulated annealing approach

Binary machine-part matrices have been widely used to identify machine groups and part families. The methods based on binary machine-part matrices mostly focus on the reduction of setup times and material handling costs. However, some other objectives such as the maximization of within-cell utilization and minimization of workload imbalance may not be achieved without considering other important factors such as processing times, lot sizes and machine capacities. Ignoring the processing times may violate the capacity constraints, and thus lead to an infeasible solution. This paper proposes a generalized grouping efficiency considering processing times and lot sizes. A simulated annealing algorithm is developed to solve the grouping problem and a neural network approach is used to provide a seed solution. Our computational experience indicates that the proposed algorithm is able to find a near optimum solution with less number of duplicated machines and better workload balance as compared to the approach reported in the literature.