Modelo e algoritmos para um problema integrado de planejamento, sequenciamento e alocação de pátios

• Ano de defesa: 2016
• Autor(a) principal: Gustavo Campos Menezes
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Logística; Branch and Price; Sequenciamento; Planejamento Integrado; Otimização matemática; Algoritmos de computador; Computação
• Link de acesso: http://hdl.handle.net/1843/ESBF-AEHQM3

Organizations are continuously searching for quality, productivity and cost reduction. Proper planning, supported by decision support systems and by an integrated view of operations is an important strategy to guarantee survival in global, dynamic and highly competitive markets. This thesis investigates an integrated production planning,scheduling and yard allocation problem. The problem is to dene the amount and destination of each input or output order in a bulk cargo terminal, establishing a set of feasible routes to guarantee that products are stored and shipped on schedule, minimizing operational costs. In this research, this problem will be addressed based on the operations in a bulk port terminal. The main objective consists in developing an integrated formulation for the production planning, scheduling and yard allocation problem, as well as designing algorithms which are able to provide quality solutions for large instances of the problem. The contributions of this research are the following ones: an integrated mathematical programming model, an exact algorithm based on the useof Column Generation and Branch and Bound, and an heuristic algorithm based on a hierarchical approach. The problems were solved through a combination of heuristics and exact methods based on two classical graph optimization problems: vertex coloring and the maximum weight independent set. The computational results show that the exact and heuristic approaches have been able to produce exact solutions for smalland medium-size instances but is compatible with real cases and that it oers strong bounds for large instances for which optimization packages are not able to provide solutions.