Heuristic algorithms for buffer allocation in a production line with unreliable machines

Efficient heuristic algorithms are developed for the buffer allocation in a production line so as to maximize the throughput. The authors focus on the unbalanced production line in which the mean processing time of each machine is not identical. The optimal buffer allocation problem is formulated as a multi-dimensional non-linear search problem with a linear resource constraint such that total capacity available is finite. The concept of a standard and non-standard exchange vectors are introduced to develop two different versions of the algorithm. Since the objective function representing the throughput is not differentiable, the authors suggest the pseudo-gradient for obtaining improving directions. Numerical experiments show that the proposed algorithms work well even for the larger problems that could not have been handled as yet.