Models for machine-part grouping in cellular manufacturing

For cellular manufacturing strategies to succeed, the productive system first has to be divided into highly independent cells. This means that a partition of the machines into machine groups, a partition of the parts into part families, and a matching between the machine groups and the part families have to be simultaneously determined. Mathematically, this question can be expressed as the problem of finding a near block diagonal permutation of the machine-part incidence matrix. Research on such grouping problems has primarily concentrated on the design of heuristics. Different grouping efficiency criteria have been proposed to express the quality of the groupings proposed by the heuristics. This paper is concerned with mathematical programming approaches to the formation of production cells. Existing models are reviewed and their features are briefiy discussed. An alternative model is proposed, which allows for the formulation of various constraints and grouping efficiency criteria. Finally, some test problems are used to support the claim that this model may be adequate for the solution to optimality of the cell formation problem.