Performance Analysis of Exponential Multi-Server Production Lines with Fluid Flow and Finite Buffers

This article presents an approximation method for fluid flow production lines with multi-server workstations and finite buffers. Each workstation consists of parallel identical servers, which are subject to operation-dependent failures with exponentially distributed uptimes and downtimes. The proposed method decomposes the production line into single-buffer subsystems, each described by a continuous state Markov process, the parameters of which are determined iteratively. The approximation method is appropriate for the analysis of longer production lines, able to accurately estimate performance characteristics (e.g., throughput and mean buffer content), and shown to perform well on a large test set.