Event Details

Dave Worthington will present a seminar to the Management Science Department

Abstract: A recent Masters project investigated the extent to which established queueing theory models could be used to approximate the behaviour of a queueing system in which the service received by patients discharged from hospital consisted of decreasing numbers of home visits per day over the patient’s period of recovery.

We found that a discrete time Mbatch(t)/Geom1/S’(t) model provides good quality approximate results when compared to a more faithful simulation model of the real system if:

• the mean service time and batch size distribution are chosen for the Mbatch(t)/Geom1/S’(t) model so that the mean and variance of the total amount of work associated with a patient matched (via the Law of Total Variance) that of the simulated model;
• S’(t) is chosen to match the service capacity of the simulated model.

This finding has led to the idea of using discrete time Mbatch(t)/Geom1/S’(t) models to provide approximate analytical results for M(t)/G/S(t) queueing systems. Early empirical results are very promising; and the approach is computationally much cheaper than existing analytical approaches.