Analysis of Approximate Planning Horizons

Manufacturers typically plan a production schedule on the basis of actual orders received together with orders that are forecasted for the next few weeks, on the assumption that there will be no new orders until the end of the planning horizon. A key issue is to determine a length for this planning horizon. A short planning horizon typically results in important future information being unavailable to the decision maker, and therefore schedules are more costly than if more information were available. On the other hand, forecasting demand and scheduling under a longer planning horizon is also costly and difficult to perform accurately. Therefore, a balance is required between these extremes. This proposal aims to study how the length of a planning horizon affects the quality of a schedule. Various scheduling environments will be considered, and methodologies developed for establishing a suitable length planning horizon that produces a schedule of the desired accuracy.