A Simulation-Optimization Approach for Optimizing Service Provision of Multi-service Queues
Abstract
Abstract – Queueing systems in the real world can involve multiple types of services provided, such as doctors with different specializations in hospitals, highway toll booths handling cash or RFID payment, and the provision of several fuels in various dispensers in gasoline stations. These types of queues diverge from the common queue types in queueing theory, where it is assumed that only one service type is provided. This study investigates the scenario where a queueing system is to be designed to optimize the system performance with respect to relevant metrics, in particular, the average sojourn time of all customers that entered the system. Specifically, the study tackles the problem of determining which services to offer in a queueing system with a fixed number of servers and a fixed service capacity (i.e. number of services provided) per server. In order to provide a mathematically tractable solution, the combinatorial optimization problem is formulated as an integer program that is solved using the Particle Swarm metaheuristic. Results show improvements of up to 6.9342% in the identified performance upon the implementation of the optimal configuration of the queueing system. Sensitivity analysis shows the robustness of the methodology for varying mean values of the arrival distribution, allowing for a wider range of applicability in the real world.
Keywords: queueing optimization, multi-service queues, discrete-events simulation