Comparison study of metaheuristics

Abstract

Many existing studies have used hypothetical data to evaluate the performance of various metaheuristics in solving delivery route optimization. As empirical data impose characteristics of a particular problem, it is necessary to evaluate whether the problem characteristics may influence to the performance of metaheuristics. This article therefore attempts to compare the performance of metaheuristics, that is, genetic algorithm, ant colony optimization (ACO), particle swarm optimization, and simulated annealing (SA), to solve an empirical delivery problem in Yogyakarta, Indonesia. Two cases are developed to capture different characteristics of empirical data. The first case introduces delivery problem of one logistics operator and 58 retailers; the second case presents collaborative strategy in delivery problem, involving two logistics operators and 142 retailers. Results indicate that ACO and SA perform better with respect to less distance traveled for both cases and higher truck utility and lower number of routes for the second case.