Abordagens de otimização para o problema de alocação dinâmica de veículos no contexto de transporte rodoviário de carga no Brasil

• Ano de defesa: 2017
• Autor(a) principal: Alvarez Cruz, Cesar Dario
• Orientador(a): Morabito Neto, Reinaldo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia de Produção - PPGEP
• Palavras-chave em Português: Alocação dinâmica de veículos; Relaxação lagrangiana; Decomposição de Dantzig Wolfe; Teração de colunas; Transporte rodoviário de carga; Logística
• Palavras-chave em Inglês: Dynamic vehicle allocation problem; Lagrangean relaxation; Dantzig Wolfe decomposition; Column generation; Road freight transportation; Logistics
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA DE PRODUCAO
• Link de acesso: https://repositorio.ufscar.br/handle/ufscar/9312

This work aims at treating the Dynamic Vehicle Allocation Problem (DVAP) in the context of the Brazilian Freight Transportation system. The problem consists of allocating empty vehicles to different terminals so as to attend the demand of freight transport during a predetermined planning horizon while maximizing the profit from these services. These type of decisions arise in customized freight transport services and in between-terminals operations of consolidation freight services. Given the size of the resulting models of real life problems confronted by third party logistics operators are large for using exact solution methods, heuristic methods have been used for giving good quality solution at the expense of optimality guarantee. In this context, the objective of this work is to contribute with solution methods that provide optimality guarantee or quality solution certificates for treating large-scale problems in reasonable computational times. The methods utilized are lagrangean relaxation, using subgradient optimization, and DantzigWolfe decomposition together with a lagrangian heuristic and factibilization method, respectively. Computational experiments are presented and analyzed for randomly generated instances and real-world instances from a brasilian freight operator. The latter method shows great potential for treating large-scale problems.