A GRASP + ILP-based metaheuristic for the capacitated location-routing problem

  • Authors:

  • Claudio Contardo;Jean-François Cordeau;Bernard Gendron

  • Affiliations:

  • Département de management et technologie, ESG UQÀM, Montréal, Canada H2X 3X2 and Groupe d'études et de recherche en analyse des décisions (GERAD), Montréal, Canada H3 ...;Canada Research Chair in Logistics and Transportation, HEC Montréal, Montréal, Canada H3T 2A7 and Centre interuniversitaire de recherche sur les réseaux d'entreprise la logistique e ...;Département d'informatique et de recherche opérationnelle, Université de Montréal, Montréal, Canada H3C 3J7 and Centre interuniversitaire de recherche sur les réseaux ...

  • Venue:
  • Journal of Heuristics
  • Year:
  • 2014

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper we present a three-phase heuristic for the Capacitated Location-Routing Problem. In the first stage, we apply a GRASP followed by local search procedures to construct a bundle of solutions. In the second stage, an integer-linear program (ILP) is solved taking as input the different routes belonging to the solutions of the bundle, with the objective of constructing a new solution as a combination of these routes. In the third and final stage, the same ILP is iteratively solved by column generation to improve the solutions found during the first two stages. The last two stages are based on a new model, the location-reallocation model, which generalizes the capacitated facility location problem and the reallocation model by simultaneously locating facilities and reallocating customers to routes assigned to these facilities. Extensive computational experiments show that our method is competitive with the other heuristics found in the literature, yielding the tightest average gaps on several sets of instances and being able to improve the best known feasible solutions for some of them.