Metaheuristics for a Profitability Assessment of the Production of Hybrid Products in a Closed-Loop Supply Chain with Remanufacturin

A source of complexity inherent in remanufacturing environments is the unbalanced number of remanufacturable component returns due to durability issues. A possible solution to this problem is produce hybrid products, which consist of both new and returned components. Therefore, the focus of this study is the production of hybrid products, and the profitability assessment of producing hybrids in a hybrid manufacturing and remanufacturing system. To evaluate the system, three versions of a dynamic lot-sizing problem are modeled: 1) capacitated dynamic lot-sizing problem with returns and hybrid products (CLSPRH), 2) uncapacitated dynamic lot-sizing problem with returns and hybrid products (DLSPRH), and 3) uncapacitated dynamic lot-sizing problem at the component level without hybrids (DLSPR-C). These problems are mixed-integer nonlinear programming (MINLP) problems, and they are NP-hard. Hence, metaheuristic algorithms are designed and proposed to produce near optimal solutions. First, Simulated Annealing (SA), Genetic Algorithm (GA) and Variable Neighborhood Search (VNS) algorithms are designed to solve CLSPRH and DLSPRH. Then, an improved SA algorithm with a neighborhood list (SA_NL) and a GA based heuristic (GA_H) are designed to solve CLSPRH and DLSPRH, respectively. The effectiveness of the proposed algorithms is compared to each other, and to the linearized approximations of the problems. Over all instances considered, numerical results on CLSPRH indicate that SA_NL performs better than SA, VNS, GAOP, GATP and GAOPPB by 0.54%, 0.35%, 1.92%, 1.78%, and 2.92%, respectively. Comparison of SA_NL with the optimal values of the linearized approximation suggests that the average percentage error over all instances is 2.73%. On the other hand, computational experiments on DLSPRH suggest that GA_H outperforms SA and VNS algorithms by 2.53% and 2.19%, respectively. Comparison of GA_H solutions with the optimal values of the linear approximation indicates that GA_H is a very good candidate to solve the problems considered in this dissertation with an average error of 1.32%. In addition, production of hybrids is recommended for medium - high holding cost environments when remanufactured item demand rate is low. At high levels of remanufactured item demand, production of hybrid products is recommended for these environments when especially high holding costs are incurred.