| It is common practice in service systems to have customers who cannot be served immediately upon arrival wait in queue until system resources become available to the customer. A customer's waiting experience typically affects his evaluation of the service provided. For service providers, making delay announcements is a relatively inexpensive way of reducing customer uncertainty about delays, thereby improving customer satisfaction with the service provided. Our work focuses on applying queuing theory and computer simulation to develop effective ways to accurately predict customer delay in customer service systems, in real time. Primarily, these real-time delay predictions are intended to help service providers make delay announcements. But, they may also be used by service providers to better manage their systems. For instance, recognizing that customer delay is longer than planned at a service facility, the service provider may elect to provide additional service at that facility in order to reduce customer delay. Our general approach is to consider large heavily-loaded queueing models that mimic the operations of a real-life service system such as call center or a hospital emergency department. We are particularly concerned with the practical appeal of our delay prediction procedures. That is why we incorporate important real-life phenomena such as customer abandonment, time-varying arrival rates, and a time-varying number of servers. We also consider general arrival, service, and abandonment-time distributions (exponential and non-exponential), which are commonly observed in practice. We use heavy traffic limits and computer simulation to quantify the accuracy of the alternative delay predictors proposed, and to compare them with delay predictors commonly used in practice.Keywords: Real-time delay prediction delay announcements, many-server queues simulation heavy-traffic call centers customer abandonment time-varying arrival rates nonstationary queues time-varying number of servers... |
