Multi-server, Finite Waiting Room, M/G/c/K Optimization Models

Abstract

Optimization models for multi-server queueing systems are very appropriate capacity planning and design tools for manufacturing, telecommunication, transportation, and facility modeling applications. A design methodology based on a two-moment approximation algorithm is utilized for optimizing the number of servers, arrival and service rates, and amount of waiting room in these M/G/c/K multi-server systems where service in the system is from a general distribution. A property describing the relationship between K and ρ bounds the region in which optimal results should be found for K. Extensive computational results are provided to evaluate the efficacy and usefulness of this optimization methodology on a number of different system problems.