Book Description
We consider a large scale multi-item periodic-review stochastic inventory system with Minimum Order Quantity (MOQ) requirements on the total order size of all items. The item demands are stochastic and non-stationary. The retailer must decide at each time period whether and how many of each item to order from the supplier. The goal is to maximize the reward generated by satisfying the end customers demands while minimizing the costs of holding unsold inventory. We introduce a parameterized heuristic, the w-policy, that relies on our detailed analysis of the problem.We demonstrate the scalability of this heuristic up to ten thousand items, which is unmatched in the literature on the non-stationary demand version of the MOQ problem. The results of our numerical study on large scale real world data sets exhibit the efficiency of the w-policy in these challenging settings. Moreover, our numerical study shows that the w-policy performs at least as well as the state-of-the-art method (S,T) on the simplified problem when the demand is considered stationary, while being more robust in a real supply chain environment.Our results therefore indicate that the w-policy is more robust, has a larger scope of applicability, and is significantly more scalable than state of the art methods for multi-item stochastic inventory control with MOQ requirements.
