Searching for Production Robustness Through Simulation-Based Scheduling Optimization

This paper proposes a new way to consider the dynamics of production execution through discrete event simulation. The proposed method models and simulates a production schedule using spreadsheets supplying input information for a discrete event simulation model that includes randomness (perturbations or time uncertainties) to processing and setup times. This is a new method that allows one to preview, for instance, how robust (resilient) a given schedule really is in midst of real production environment, where resources fail, suppliers delay deliveries, products need reprocessing etc. The proposed approach allows one to more accurately estimate performance of a given schedule execution subject to undesired and unexpected events because it models times using probability distributions instead of deterministic ones, often used by production planners (schedulers) and/or scheduling software tools. This method is very different from traditional mathematical optimization and simulation models, since it simulates the schedule itself, not using dispatching rules nor arrival rates. A three-machine production schedule illustrates the proposed approach. Under the assumptions considered, a 5% increase in total processing in time will probably occur. This waste (loss) was not "seen" during the time the production planner created the schedule (using deterministic setup and processing times).