Optimal Core Acquisition and Remanufacturing Decisions With Discrete Core Quality Grades

The acquisition of used products (cores) is essential for leveraging the advantages of remanufacturing. Due to limited historical information about cores, it is difficult for the remanufacturers to accurately assess the quality of cores before disassembly and inspection. In this article, we investigate the joint acquisition and remanufacturing problem with a discrete empirical distribution. Under both deterministic and stochastic demands, two integer programming models are developed to obtain operational rules for making optimal decisions. Since the discrete empirical distribution is assumed to be the nominal distribution or asymptotic distribution of the true quality, the error caused by the difference between the estimated discrete empirical distribution and the true distribution cannot be neglected. Thus, we evaluate the error using the Phi-divergence measure, and the corresponding effects of this difference on the resultant acquisition and remanufacturing rules are analyzed. Through numerical examples, we observe that optimal acquisition and remanufacturing decisions are negatively affected by the difference between the estimated distribution and the true distribution, even if the estimated distribution can pass a hypothesis test. Furthermore, to overcome suboptimal outcomes, a multisource information fusion approach is proposed, which can utilize quality information collected from multiple sources to make accurate acquisition and remanufacturing decisions.