Abstract
A two stage procedure for economically identifying the cell composition of a cellular manufacturing system is presented. In the first stage an algorithm is presented that is capable of identifying orthogonal sets of bottleneck machines and exceptional parts as well as the part families and their respective machine cells. This algorithm requires a part-machine incidence matrix, and it is compared with three other procedures using three published data sets. In the second stage subcontracting and inter-cell transfer costs of exceptional parts as well as duplication costs of bottleneck machines are considered. Integer programs are presented for minimum cost treatment of the exceptional parts and/or bottleneck machines that were identified in the first stage. A numerical example for the second stage is presented.