Algorithms for determining capacities of individual buffers in assembly/disassembly systems

We deal with a buffer allocation problem of assembly/disassembly (AD) systems with a finite buffer capacity. The problem is to determine capacities of individual buffers for a given total available capacity of buffers with the objective of maximizing throughput rate. We first present an algorithm (BA-G) in which a conventional gradient search is used for finding an initial solution and a two-optimization procedure is used for improving the initial solution. Since this algorithm may require an excessive computation time, especially when a procedure for performance evaluation employed in the algorithm takes a long computation time, a new algorithm (BA-P) is developed in which a sophisticated procedure is used for finding an initial solution. In this procedure, capacities of individual buffers are determined based on the efficiency of their upstream and downstream machines. Results of computational experiments show that the latter algorithm gives the same solutions as those obtained by the former algorithm in much shorter time.