Metaheuristics for Sustainable Supply Chains

Abstract

Optimizing models of sustainable supply chains is a high-dimensional multi-objective optimization problem. The primary focus of this optimization is minimizing different kinds of costs. Costs can generally be grouped into economic, environmental, and social costs. Metaheuristics can be used for tackling this kind of task efficiently. In this thesis, a realistic model of a supply chain is implemented. Different metaheuristics are implemented or adapted for optimizing the above-mentioned model. The results are then compared. It was found that genetic algorithms performed the best out of the three compared stand-alone metaheuristics which also included simulated annealing and particle swarm optimization. The results obtained by the genetic algorithms were feasible solutions to the problem. Other stand-alone metaheuristics did not provide solutions of sufficient quality. Two hybrid methods were also used. The first one is a combination of the genetic algorithm and the particle swarm optimization. The second one is a combination of the genetic algorithm and simulated annealing. The simulated annealing hybrid did not improve on the initial solution provided by the genetic algorithm in the simulated annealing phase. It was found that the particle swarm hybrid improved the result of the genetic algorithm in the particle swarm phase. Based on the experiments in this thesis the implementation of the hybrid genetic algorithm combined with particle swarm optimization outperformed the implementation of the stand-alone genetic algorithm.