Modeling Queues with Simulation versus M/M/C Models
Abstract
This paper examines the performance of single-queue service systems using a combination of computer simulation and M/M/C queuing models. Our results show that the accuracy of M/M/C models is significantly affected by the assumptions supporting the models. Managers should therefore exercise caution in using the M/M/C models for designing queuing systems when the models’ assumptions are violated. Our results show that cost-centric and servicecentric firms should manage their queues differently. While cost-centric firms should target higher arrival load, single service session, and front-loaded arrival pattern for higher efficiency, service-centric firms should strive for lower arrival load, multiple short sessions and even arrival pattern for better service. In addition, both cost-centric and service-centric firms can consider pooling servers together and reducing the variability of inter-arrival and service times to improve both cost and service simultaneously