Modelos de otimização e métodos de solução para o planejamento de redes logísticas

• Ano de defesa: 2022
• Autor(a) principal: Jalal Osorio, Aura Maria
• Orientador(a): Morabito, Reinaldo
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• 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: Logística; Localização; Política de estoque; Transporte; Incerteza; Otimização robusta
• Palavras-chave em Inglês: Logic-based Benders decomposition; Heurística Fix-and-Optimize; Logistics; Location; Inventory policy; Safety stock; Transportation; Fix-and-Optimize heuristic; Uncertainty; Robust optimization
• Palavras-chave em Espanhol: Logística; Localización; Política de inventario; Inventario de seguridad; Transporte; Incertidumbre; Optimización robusta
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA DE PRODUCAO::PESQUISA OPERACIONAL
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/16921

Logistics Network Planning (LNP) involves decisions such as facility location, demand allocation, inventory, and transportation management. These decisions differ in terms of periodicity and frequency over the planning horizon. However, the integration of these decisions has been receiving attention from academics and practitioners in the last years aiming to achieve an adequate service level and efficient performance, in terms of network logistics costs and competitive advantages. Nevertheless, there is still a lack of research in this area. Thus, in this work, we study integrated planning in logistics networks. Foremost, we carry out a systematic literature review to understand the main decisions in logistics planning, the integration approaches, and the solution methods. Then, we present a generic mathematical model for the integration of network design, inventory, and transportation planning. We integrate features and characteristics of the real-world application, such as demand variability, location-based lead times, storage capacity constraints in distribution centers (DCs), piecewise linear transportation costs, and a multi-period and multi-product context. The model determines the DC locals to rent; the selection of the capacity level at the DCs; the assignment of retailers to DCs; the cycle, safety stock, and anticipation inventory levels at DCs; the selection of the cost range/segment for transportation. In addition, we investigate solution methods exploring specific characteristics of the problem. A Logic-based Benders decomposition (LBBD) that enhances the master problem with a non-standard decomposition and a piecewise linear lower bound function of safety stock is proposed. Furthermore, we address the case of a pharmaceutical logistics network in Brazil to propose mathematical modeling for location and transportation planning with some characteristics such as safety measures in cargo transportation and tax issues. Particularly, we address the Tax of Circulation of Goods and Services (Imposto de Circulação de Mercadorias e Serviços - ICMS, in Portuguese), a relevant tax for supply chains in Brazil, but it is little explored in the literature. We also handle uncertainty in demand by proposing a robust counterpart of the mathematical model. We deal with instances based on real data, for which a general-purpose software provides poor-quality solutions. Therefore, we propose a Fix-and-Optimize heuristic to solve the models near optimality. We also present robustness analyses and practical insights about the problem. The results show the potential of the models and solution methods to address integrated problems in LNP. Therefore, by studying relevant practical features and suggesting effective solution methods, this thesis contributes to the literature on supply chain optimization and the development of tools to support decision-making in practice.