Timing intermittent demand with time-varying order-up-to levels

Current intermittent demand inventory control models assume that the interval is memoryless: probability of observing a positive does not depend on time since last occurred. Contrarily, several forecasting contributions suggest intervals contain more distributional information. We find data M5 competition confirms this. Therefore, we propose an model explicitly uses full distributions sizes and thereby acknowledges occurrence may vary throughout interval. To exploit this information, also allow for time-varying order-up-to levels flexibly adjust inventories according to dynamic requirements. derive long-run average holding costs, non-stockout probability, order fill rate, volume rate. Inspired by analogy with multi-item models, greedy marginal-analysis heuristic optimize levels, which benchmark against optimal solution theoretical instances. In simulation study demonstrate method’s improved on-target service performance compared traditional solutions. furthermore show target can be achieved at significantly lower costs than fixed levels.