Two efficient hybrid metaheuristic methods for solving the load balance problem

Abstract

In this paper we consider a discrete Load Balance location problem (LOBA). We propose two efficient hybrid metaheuristic methods for solving the LOBA problem: a combination of reduced and standard variable neighborhood search methods (RVNS-VNS), and hybridization of genetic algorithm and VNS approach (GA-VNS). The proposed hybrid methods are first benchmarked and compared on existing test problems for the LOBA problem with up to 100 customers and potential suppliers. In order to test effectiveness of the proposed methods, we modify some large-scale instances from the literature with up to 402 customers and potential suppliers. Exhaustive computational experiments show that proposed hybrid methods quickly reach all known optimal solutions, and provide solutions on large-scale problem instances in short CPU times. Regarding solution quality and running times, we conclude that the proposed GA-VNS approach outperforms other considered methods for solving the LOBA problem.