Machine-part grouping in group technology manufacturing systems

Abstract

In this thesis, the machine-part grouping problem in group technology (GT) manufacturing systems is considered. Many methods have been proposed to find mutually independent machine cells. Most of them usually assume that each part has only one fixed process routing which indicates the sequence of facilities used to process each part. Grouping problem with single routing is referred to the simple GT problem(SGTP) in this thesis. However, this assumption may not be realistic in GT environment since each operation of a particular part may be performed on alternative machines. Grouping problem with multiple process routings is called the generalized GT problem(GGTP) in this thesis. The main concern of this thesis is to develop some methods for solving GGTP. The objective is to find the machine-part cells which attain the fewest intercellular moves of parts in a GT manufacturing systems. First, a similarity coefficient method is proposed. Considering that the existing similarity coefficient methods which use the similarity coefficient defined between routings cause the critical problem in terms of the computational complexity, new generalized similarity coefficient defined between machines is suggested. The new coefficient includes existing machine similarity coefficient for solving SGTP as a special case and can handle both SGTP and GGTP. A reduction algorithm using the generalized similarity coefficient is developed. The algorithm resolves the machine chaining problem by incorporating multiple clustering criteria. Second, reduced p-median models using integer programming are proposed. By introducing the new generalized similarity coefficient, new models include much smaller number of binary variables and constraints than the original p-median model given by Kusiak(1987). To heighten the solution quality in terms of the number of exceptional elements, two-stage approach reassigning the bottleneck or idle machines is established. Third, a part-based assignment model...