A hybrid metaheuristic algorithm for the vehicle routing problem with stochastic demands

Andres Gutierrez (1, 2) , Laurence Dieulle (3) , Nacima Labadie (2) , Nubia Velasco (1)

1 UNIANDES - Universidad de los Andes [Bogota]
2 LOSI - Laboratoire d'Optimisation des Systèmes Industriels
3 LM2S - Laboratoire Modélisation et Sûreté des Systèmes

This article deals with the Vehicle Routing Problem with Stochastic Demands. To solve this problem, a hybrid metaheuristic combining a Memetic Algorithm and Greedy Randomized Adaptive Search Procedure is designed. The developed approach is tested on a 40 instances benchmark. The results are validated by comparing them to state of the art metaheuristics, they show that our method outperforms these metaheuristics in terms of quality and efficiency. A new testbed of 39 instances with up to 385 customers is also proposed and tested. This paper is the first one to deal with closer to real life size problems.

Keywords: Vehicle routing; Stochastic programming; Stochastic demands; Hybrid metaheuristic; Memetic algorithm; Greedy randomized adaptive search

Domains: Operations Research [cs.RO]; Data Structures and Algorithms [cs.DS]

Andres Gutierrez, Laurence Dieulle, Nacima Labadie, Nubia Velasco. A hybrid metaheuristic algorithm for the vehicle routing problem with stochastic demands. Computers and Operations Research, 2018, 99, pp.135-147. ⟨10.1016/j.cor.2018.06.012⟩. ⟨hal-02477121⟩