Multi-stage supply chain network solution methods: hybrid metaheuristics and performance measurement
We study a three-stage supply chain network (SCN) design problem for a single-product system. The SCN is composed of suppliers that provide raw materials for the plants, plants that produce and send the finished products to distribution centres (DCs), and DCs that transport finished products to the customers. The use of different conveyances and step-fixed costs improves the applicability and the results but increases the complexity, generating an NP-hard problem. The overall objective of the problem is to minimise the total cost, which is composed of the opening and transportation costs in all three stages. Two hybrid metaheuristics – genetic algorithm–variable neighbourhood search (GA-VNS) and variable neighbourhood search–simulated annealing (VNS-SA) – are proposed to solve this NP-hard problem. In addition to the novelty of the proposed algorithms, we develop an innovative priority-based decoding method to design chromosomes and solutions related to the nature of the problem. A robust parameter and operator setting is implemented using the Taguchi experimental design method with several random test problems. The performance of these algorithms is evaluated and compared for different problem sizes. The experimental results indicate that the GA-VNS is robust and superior to the other competing methods.
Tavana, Madjid; Santos-Arteaga, Francisco J.; Mahmoodirad, Ali; Niroomand, Sadegh; and Sanei, Masound, "Multi-stage supply chain network solution methods: hybrid metaheuristics and performance measurement" (2017). Business Systems and Analytics Faculty Work. 115.