A closed-loop supply chain configuration considering environmental impacts: a self-adaptive NSGA-II algorithm

• Published in: Applied intelligence (Dordrecht, Netherlands) Vol. 52; no. 12; pp. 13478 - 13496
• Main Authors: Babaeinesami, Abdollah, Tohidi, Hamid, Ghasemi, Peiman, Goodarzian, Fariba, Tirkolaee, Erfan Babaee
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.09.2022; Springer Nature B.V
• Subjects: Adaptive algorithms; Algorithms; Artificial Intelligence; Big Data-Driven Large-Scale Group Decision Making Under Uncertainty; Computer Science; Configurations; Customers; Distribution centers; Distribution management; Environmental impact; Genetic algorithms; Machines; Manufacturing; Mechanical Engineering; Network design; Processes; Refurbishment; Sensitivity analysis; Sorting algorithms; Supply chain management; Supply chains; Taguchi methods; Environmental impacts; Self-adaptive NSGA-II algorithm; Mathematical modeling; Closed-loop supply chain network
• ISSN: 0924-669X; 1573-7497
• DOI: 10.1007/s10489-021-02944-9

Configuration of a supply chain network is a critical issue that contributes to choose the best combination for a set of facilities in order to attain an effective and efficient supply chain management (SCM). Designing a closed-loop distribution network of products is an important field in supply chain network design, which offers a potential factor for reducing costs and improving service quality. In this research, the question concerns a closed-loop supply chain (CLSC) network design considering suppliers, assembly centers, retailers, customers, collection centers, refurbishing centers, disassembly centers and disposal centers. It aims to design a distribution network based on customers’ needs in order to simultaneously minimize the total cost and total CO 2 emission. To tackle the complexity of the problem, a self-adaptive non-dominated sorting genetic algorithm II (NSGA-II) algorithm is designed, which is then evaluated against the ε-constraint method. Furthermore, the performance of the algorithm is then enhanced using the Taguchi design method to tune its parameters. The results indicate that the solution time of the self-adaptive NSGA-II approach performs better than the epsilon constraint method. In terms of the self-adaptive NSGA-II algorithm, the average number of Pareto solutions (NPS) for small and medium-sized problems is 6.2 and 11, respectively. The average mean ideal distance (MID) for small and medium-sized problems is 2.54 and 5.01, respectively. Finally, the average maximum spread (MS) for small and medium-sized problems is 3100.19 and 3692.446, respectively. The findings demonstrate that the proposed self-adaptive NSGA-II is capable of generating efficient Pareto solutions. Moreover, according to the results obtained from sensitivity analysis, it is revealed that with increasing the capacity of distribution centers, the amount of shortage of products decreases. Moreover, as the demand increases, the number of established retailers rises. The number of retailers is increasing to some extent to establish 7 retailers.