Estimating Expected Completion Times with Probabilistic Job Routing

A common problem in production environments is the need to estimate the remaining time in system for work-in-progress jobs. Simulation can be used to obtain the estimates. However, when the future path of a job is uncertain (due to stochastic events such as rework), using simulation to estimate the remaining cycle time of a job at step k can be imprecise; traditional confidence intervals on the estimated remaining cycle times may be too large to be of practical significance. We propose a response surface methodology-based approach to estimating conditional confidence intervals on the remaining cycle times as jobs progress through the system and more information is obtained on them. This method will provide more useful and accurate estimates of remaining cycle times at various stages of the process flow. Further, we outline two different simulation approaches for estimating the response surfaces used to generate the confidence intervals