A hybrid spanning tree-based genetic/simulated annealing algorithm for a closed-loop logistics network design problem

Abstract

Affecting the efficiency and responsiveness of the logistics, supply chain network design, in particular, closed-loop logistics has attracted more attention in recent years. The problem in this paper is to minimise the total cost of the cyclic logistic network model by determining the best locations of plants, distribution centres (DCs), and dismantlers. Due to the NP-hard nature of the problem, it is inevitable to appeal to metaheuristic procedures to achieve satisfactory solutions for large-size problems. To tackle such an NP-hard problem, this paper proposes a simulated annealing (SA), a modified genetic algorithm (MGA), and a hybrid algorithm, applying these two algorithms based on a revised spanning tree and determinant encoding representation. To evaluate the proposed algorithms, we compare them with the genetic algorithm (GA) taken from the recent literature. Finally, it is shown that the proposed hybrid algorithm outperforms other algorithms.