A demand-driven method for scheduling optimal smooth production levels

This paper is concerned with demand-driven production scheduling in a commercial environment where smoothed production plans generation over a rolling horizon is desirable as new observations of demand are received through time. Demands are assumed to be normally distributed and dependent on the previous observed levels. The method of chance constraint of Charnes and Cooper is extended to multi-product production planning with variable workforce, back-ordered inventory, and nonstationary stochastic demand process. Bayesian procedures for revising the chance constraints and several variants of linear-programming-based production planning models are presented. In all cases the proposed methodology ensures that demands are satisfied, at a given level of confidence, while achieving smooth production.