Title: The server allocation problem for Markovian queueing networks

Authors: Anderson Ribeiro Duarte

Addresses: Statistics Department, Universidade Federal de Ouro Preto, Ouro Preto, MG, Brazil

Abstract: Manufacturing systems often need to improve productivity to meet customer demand drastically. In the design phase of these systems, one of the most significant problems is how to allocate resources (servers) to design efficient queueing systems. Queueing networks with multiple servers, arbitrary acyclic topologies with Markovian arrivals, and services are considered in this paper. We present an approach for optimal server allocation in several topologies (series, merge, and split). The methodology uses a multi-objective heuristic strategy using the simulated annealing algorithm. Server utilisation performance is maximised simultaneously with the minimisation of the expected overall time of customers in the queueing network. Results from computational experiments show the efficacy of the proposal.

Keywords: multi-objective optimisation; simulated annealing; productivity; expected overall time.

DOI: 10.1504/IJSOM.2024.138931

International Journal of Services and Operations Management, 2024 Vol.48 No.2, pp.256 - 271

Received: 28 Oct 2021
Accepted: 12 Mar 2022
Published online: 04 Jun 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article