Closed Queuing Networks with Multiple-server Stations: New Approximations for Analysis

Closed Queuing Networks are used in modeling systems such as Flexible Manufacturing (FMS), CONWIP Material Control, Computer/Communication Systems, Biotech Manufacturing and Health Care. Mean Value Analysis (MVA) is often used to compute the performance measures for these models. For networks with multiple-server stations, the exact MVA algorithm becomes computationally complex. The Schweitzer-Bard (S-B) approximation for MVA is simple and computationally efficient. However it has only been developed for networks with single-server stations. We provide a simple extension to S-B MVA to enable the analysis of networks with multiple-server stations. We develop an approximation for networks with single customer classes and then extend it to networks with multiple customer classes and general service time distributions. Comparison with exact results and simulation experiments for a wide variety of network configurations demonstrates the accuracy of this approach.