Production planning and inventory control for a two-product recovery system

The significance of product recovery through remanufacturing has been widely recognized and has compelled manufacturers to incorporate product recovery activities into normal manufacturing processes. Consequently, increasing attention has been paid to production and inventory management of the product recovery system where demand is satisfied through either manufacturing brand-new products or remanufacturing returned products into new ones. In this work, we investigate a recovery system with two product types and two return flows. A periodic-review inventory problem is addressed in the two-product recovery system and an approximate dynamic programming approach is proposed to obtain production and recovery decisions. A single-period problem is first solved and the optimal solution is characterized by a multilevel threshold policy. For the multi-period problem, we show that the threshold levels of each period are solely dependent on the gradients of the cost-to-go function at points of interest after approximation. The gradients are estimated by an infinitesimal perturbation analysis-based method and a backward induction approach is then applied to derive the threshold levels of each period. Numerical experiments are conducted under different scenarios and the threshold policy is shown to outperform two other heuristic policies.