Localização de concentradoresaplicada ao transporte público

Abstract

One of the biggest problems that large cities have faced nowadays is traffic congestion. An alternative to solve this problem would be greater use of public transport like trains, subways, stick-row and so on. However, due to the high cost to install direct connections between all pairs of origin-destination points, modeled the urban transport networkas a network-type hub-and-spoke. In this work a model based on the formulation proposed by Contreras et al. (2009) is used to model public transport. Due to the characteristics of the formulation used in this work, the Benders decomposition method (Benders, 1962) is applied to solve the problem. Computational results show that the classic version of Benders can only solve small problems. Therefore, three variants of Benders decomposition scheme that possesses a selection of cuts that tend to increase the convergence of the method are presented. In addition, a new technique to improve the generation of cuts, namely -optimal, is proposed. Computational results show that this new technique has the best performance by accelerating the convergence ofthe method.