3 présentations

• 
  09h00 - 09h25

  Heuristics for the dynamic facility location problem with modular capacities

  • Allyson Silva, prés., Université Laval
  • Daniel Aloise, Polytechnique Montréal
  • Leandro C. Coelho, Université Laval
  • Caroline Rocha, Universidade Federal do Rio Grande do Norte

  We study the Dynamic Facility Location Problem with Modular Capacities. A linear relaxation based heuristic (LRH) and a hybrid evolutionary heuristic using a genetic algorithm and a variable neighborhood descent (GA+VND) are proposed to solve it. Experiments are performed to compare their performance to a state-of-the-art mixed integer programming (MIP) formulation for the problem from the literature solved by CPLEX. For the existing benchmark instances, the solution generated by LRH improved by VND finds solutions within 0.03% of the optimal ones in less than half of the computation time of the state-of-the-art MIP. In order to yield a better representation of real-life scenarios, we introduce a new set of instances for which GA+VND proved to be an effective approach to solve the problem, outperforming both CPLEX and LRH methods.

  Keywords: Location, modular capacity, hybrid metaheuristic, genetic algorithm, variable
  neighborhood descent.
• 
  09h25 - 09h50

  A general covering location model including time-dependent and stochastic features

  • Alfredo Marín, Universidad de Murcia
  • Luisa I. Martínez-Merino, prés., Concordia University
  • Antonio M. Rodríguez-Chía, University of Cádiz
  • Francisco Saldanha-da-Gama, Universidade de Lisboa

  This work presents a covering location problem including time-dependency and uncertainty. It generalizes most of the covering problems in the literature and its aim is to minimize the total expected operating cost ensuring that certain coverage constraints are satisfied. A formulation and a Lagrangian based heuristic are introduced.
  Keywords: Multi-period facility location, Covering, Lagrangian relaxation.
• 
  09h50 - 10h15

  The conditional p-dispersion problem

  • Marilène Cherkesly, prés., ESG UQÀM
  • Claudio Contardo, ESG UQÀM / GERAD

  We introduce the conditional p-dispersion problem, a variant of the p-dispersion problem, where initial facilities are given and p additional facilities need to be located to maximize the minimal distance between each pair of facilities. We propose an exact solution algorithm and conduct computational experiments to derive practical managerial insights.