Rolling horizon production planning for probabilistic time-varying demands

Publisher Website
Google Scholar
Abstract
This paper introduces and tests a production planning procedure to be applied in a rolling horizon with probabilistic demands, based on the work of Bookbinder and Tan (1983). First, the case of no lead time for replenishment is considered and trend-seasonal demands are studied. We vary the set-up cost, order cycle and number of periods in the future for which demand forecasts are available. The procedure is compared to that of Silver (1978) in terms of cost performance, percentage of demand short/period and percentage of periods with stock-outs. Each approach appears to have merit. The proposed procedure generally yields a lower solution cost, while Silver's procedure usually has a lower percentage of demand short/period and a smaller percentage of periods with stockouts. Differences between the two procedures were relatively insensitive to which of three lot-sizing rules was employed within them. Finally, we further extended our procedure to consider non-zero lead times for replenishment and different demand patterns, particularly the normal distribution.