Modeling and analysis of remanufacturing systems with stochastic return and quality variation

With rising costs of energy and raw materials, remanufacturing can help companies achieve sustainable manufacturing by recapturing residual value of used products. Of the special characteristics of remanufacturing, the uncertain quantity and quality of product returns limit the effectiveness of planning and control methods for traditional manufacturing systems. Hence we develop new models for stochastic remanufacturing systems with quality variation.;First, we study optimal policy for modular product reassembly within a remanufacturing setting where a firm receives product returns with variable quality and reassembles products of multiple classes to customer orders. Higher quality modules are allowed to substitute for lower quality modules during reassembly to provide the remanufacturing system with flexibility. We formulate the problem as a Markov decision process and characterize the structure of the optimal control policy. We show that the optimal reassembly decisions follow a state-dependent threshold policy. We also compare the optimal policy to the exhaustive assembly policy and show that there is great benefit in module substitution and threshold-based assembly control.;Second, we consider a reassemble-to-order (RATO) system with admission control on product returns. We propose a new admission control model on the used product returns and conduct performance analysis at different levels of the system. We model this problem in a multi-server system and translate it to a quasi-birth-and-death process to obtain the key performance measures of the system. We analyze the performance for a single quality grade problem and then extend the analysis to a multi-grade problem by developing a donor-beneficiary queue model. Through extensive numerical experiments, we provide insights into decision-making for admitting stochastic product returns.;Finally, we consider design of optimal admission policy on stochastic product returns. Based on the performance analysis results, we propose two optimal policies driven by different objectives. The first admission policy attempts to find optimal admission threshold levels in a RATO system which minimizes the expected cost with a reuse level constraint. The second admission policy determines the optimal admission level by examining the time value decay of returned products. We demonstrate the effectiveness of the admission policies and guidelines for implementation.