Séminaire conjoint CRT-Chaire de recherche du Canada en distributique
TITRE : An Alternative Modeling and Solution Approach for Multi-Facility Location-Allocation Problems with Large Scale Demand Data
CONFÉRENCIER : Alper Murat, Wayne State University
DATE et ENDROIT : 20 janvier 2006, 10h30, salle 3521, Pavillon André-Aisenstadt, Campus de l’Université de Montréal
RESPONSABLE : Gilbert Laporte (343-6143)
RÉSUMÉ : We consider a single-echelon, fixed-charge, planar facility location-allocation problem with large number of customers. Whereas traditional modeling approach for this problem uses discrete demand representation, we represent the customer demand information with a continuous demand density function. In addition, we project the location decisions of the location-allocation problem into allocation decision space. First, we show how our model could be cast as an optimal control problem with an infinite number of state and control variables and present the analytical properties characterizing the optimal solutions. Next, we describe a gradient based solution approach based on these properties. Finally, through numerical results, we demonstrate the efficiency of this approach for very large instances of this problem.