DISCRETE-EVENT MODELING AND OPTIMIZATION OF UNRELIABLE PRODUCTION LINES WITH RANDOM RATES

We consider a serial production system with unreliable machines mainta ined by a limited number of repairmen, and finite storage between mach ines. Processing times may be random variables with exponential or gam ma distributions, or deterministic. We develop a continuous-flow model for such a system utilizing simulation and analysis. Random processin g times are approximated by sums of deterministic variables using a si mple probabilistic technique. The model observes a limited number of e vents which are sufficient to determine system performance and mean bu ffer levels. By appropriately reducing the rates of starved and blocke d machines and using analysis to compute the times of next event at ea ch machine and buffer, discrete part computations are avoided. It is d emonstrated that this approximate model is highly accurate and faster by a factor of 3 or more when compared to conventional simulators. The paper addresses also optimal repair allocation to maximize the expect ed throughput of the system. Two different approaches are proposed: pe rturbation analysis and experimental evaluation of various nonpreempti ve rules for assigning a repairman to failed machines.