A network flow approach to solve clustering problems in group technology

Abstract

A fundamental group technology (GT) problem is the identification of part families and the decomposition of the manufacturing system into machine cells. The objective of the network flow methodology developed in this article is to measure the functional similarity between machines and then to group the machines into cells in such a way that all the parts in each family can be processed in a machine cell. The solution to the GT problem is obtained in terms of one complete-loop and several sub-loops identified by using a state-of-the-art relaxation method for solving the network flow problem. The concept of using a network flow model for cluster analysis is a new approach. It appears to have an excellent potential to provide computationally efficient optimal solutions when compared to the p-median model.