Optimization of flexible manufacturing system buffer with a bottleneck activity using simulated annealing and genetic algorithm

The incorporation of the Flexible Manufacturing System (FMS) into the modern factory has expanded significantly over the last two decades. The literature over the last 15 years has defined the benefits and usage for the FMS including the optimization of layout, part selection, routing and scheduling for the FMS. One area that has received very little attention is that of defining the size and description of an input buffer for an FMS. Though the impact of optimizing buffer size can result in reduction of Work-In-Process and reduction in the buffer's initial capital cost. Literature on production buffers has focused primarily on size and location of output buffers for those sequential systems best described as a flowshop.; The objective of this study is to define the minimum number of pallets (buffers) which can satisfy the resource input needs and scheduling for competing multiple part types on a 5 station FMS. One of the part types is described as a having a lower rate of production than the other part types, otherwise noted as being a “bottleneck.” This study will utilize the Theory of Constraint principles to maximize the production of the bottleneck part type without adversely impacting the production requirements of the remaining non-bottleneck part types.; The results from two metaheuristic approaches (Simulated Annealing and Genetic Algorithm) are compared to the results from two traditional scheduling approaches (Longest Processing Time and Due Date) to determine which method provides the minimal makespan time for a given set of part types. The constraints of the current FMS system and the principles of the Theory Of Constraints are modeled in a Visual C++ program. A Lagrangian Relaxation objective function is used to provide the numerical results for each solution.