Operational Design of a Cellular Manufacturing System

Cellular Manufacturing (CM), which contains the flexibility of Job-Shop and at the same time has a higher rate of production as flow lines, is proving to be a useful substitute for the production carried out in batches. In spite of the fact that there are so many benefits associated with CM but designing CM, for real world problems, is a very complex job. Since the main task in designing a CM is grouping of machines into cells and parts into corresponding families, therefore, most of the research carried out so far has considered the Cellular Manufacturing System (CMS) design as a Machine-Part grouping problem only and focus on the operational aspects of the design has been very little. Once the Machine-Part grouping stage is over, scheduling of the system is supposed to be the next stage in completing the operational design of a CMS. This is the stage where important production related information; such as processing sequence and processing time is taken into consideration. Scheduling is very essential as it enhances productivity and maximizes the usefulness of a given manufacturing system by utilizing the available resources in an optimized manner. Therefore, alongside Machine-Part grouping, scheduling is of paramount importance too, as it ensures proper utilization of resources. In order to carryout a complete operational design of CMS, a two stage methodology has been developed in this research. First, the problem of Machine-Part grouping (CMS design) is solved, and then sequencing and scheduling of parts on machines is carried out. Since each cell is like a Job-Shop, therefore the scheduling part of the problem is solved using a similar approach as in case of a Job-Shop scheduling problem (JSSP). Separate hybrid tools, for solving Machine-Part grouping problem and Job-Shop Scheduling Problem (JSSP), has been developed by combining Genetic Algorithms (GA) with Local Search Heuristics (LSH). Each tool’s effectiveness has been verified, separately, by solving a number of benchmark problems from literature. Finally, the two tools are combined in such a manner that the output of the Machine-Part grouping serves as an input to the tool developed for the scheduling of Job-Shop. Final outcome of the program is a cellular arrangement of the system (machine groups and corresponding part families) and detailed information about the sequencing and scheduling of the system. The development of two effective hybrid GA based tools, for Machine-Part grouping and Job-Shop Scheduling, and their combination are the main contributions of this research.