Knowledge that Transforms

We propose the use of robust optimization (RO) as a powerful methodology for multiperiod stochastic operations management problems. In particular, we study a two-echelon multiperiod supply chain problem, known as the retailer-supplier flexible commitment (RSFC) problem with uncertain demand that is only known to reside in some uncertainty set. We adopt a min-max criterion, whereby the cost function is minimized against the worst case demand occurrence. To solve the min-max RSFC problem we employ a recent extension of the RO method adapted to dynamic decision problems and known as the affinely adjustable robust counterpart (AARC) methodology. The AARC solution is tested by a large simulation study and found to provide excellent results.

Decentralized supply chains are traditionally order focused: stage managers focus on meeting replenishment orders of downstream stages; market demand information is relayed up through these orders, and therefore is subject to costly delay and distortion. This paper shows that sharing real-time sales data across all stages and a change in focus to meeting customer demand can mitigate performance impairment caused by the order focus. We show that a change of managerial focus in a decentralized chain can be made by measuring stages’ performance based on their respective echelon stocks, which only depends on how well they respond to the market demand. A demand-focused measurement scheme can be made incentive compatible with the knowledge of the demand distribution; a heuristic scheme independent of the demand distribution can be used to achieve near incentive compatibility.