Uma abordagem de otimização para a roteirização e programação de navios: um estudo de caso na indústria petrolífera

• Autor(a) principal: Rodrigues, Vinícius Picanço
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFSCAR
• Texto Completo: https://repositorio.ufscar.br/handle/ufscar/3761
• topic: Problema de roteirização de veículos; Programação de veículos; Transporte marítimo; Petróleo; Métodos heurísticos; Coleta e entrega; Vehicle routing and scheduling (VRP); Pickup and delivery; Maritime transportation; Oil; MIP-Heuristics; ENGENHARIAS::ENGENHARIA DE PRODUCAO

This work studies the ship routing and scheduling problem in oil transportation from offshore platforms to inland terminals. It is motivated by a real situation in a Brazilian oil company. Brazil is one of the world's greatest oil producers and has around 80% of its oil explored in offshore mode. Thus, transportation costs play an important role in achieving operational excellence, and the recent growth trends for oil exploration in Brazil has transformed its operations and demanded agile and effective decision support systems for addressing the oil sector dynamism. This work's goal consists in developing and applying an optimization-based approach using a mixed integer linear programming model in real decision-making situations, along with a solution method based on mathematical programming (MIP-heuristics) in order to solve the model, such as relax-and-fix. The proposed model is inspired in a problem formulation for pickup and delivery with time windows (PDPTW) and heterogeneous fleet, where costs incurred for fuel consumption and fleet contracts is the objective function to be minimized. The pickup and delivery pairs are predetermined and the model's main decision refers to ship allocation to these pairs compounding a route. Furthermore, some additional constraints are modeled and proposed, such as terminal access and platform mooring limitation according to ship types, as well as product blend incompatibility. The model was implemented in a modeling language along with an optimizarion software. Computational experiments with the model and the heuristics are presented for different data sets supplied by the case study company. These experiments show the potential benefits of this approach for finding good solutions for the problem as well as the dificulty in finding solutions for realistic instances due to its NP-hard characteristics.