Ferramenta computacional baseada em heurísticas para otimização do mix de afretamento de embarcações de apoio marítimo

Abstract

Companies operating in the oil sector, carrying out exploration and production activities in a maritime environment, present the need to maintain their fleets of vessels dedicated to supplying offshore oil and natural gas exploration units. Due to this demand, it is necessary to establish the appropriate size of the fleet of maritime support vessels, as well as the definition of the optimal chartering policy for these vessels. Therefore, this master thesis proposes the development of an optimization tool to assist in identifying the optimal chartering policy for maritime vessels, using computational techniques based on heuristics methods such as Local Search, Simulated Annealing and Genetic Algorithm, with the aim of maintain full service to maritime units at a minimum cost. To achieve this objective, a conceptual optimization model is proposed, listing the main variables to be considered for the construction of the computational model. The definition of the computational model presents an approach for the solution encoding, as well as the definition of the search heuristics and the objective function. Finally, the computational tool is proposed, with the definition of the computational architecture, in addition to the specification of the parameters necessary for its operation. During the experiment phase, a tuning step of the main parameters used by the computational tool was carried out, aiming to improve the solutions generated during the optimization process. After the tuning step of parameters, the elaboration of optimization scenarios is presented, with these scenarios being defined based on the contract modalities available for use in the optimization process. For each new optimization scenario, new contract modalities are considered, with the aim of analyzing the performance of the heuristics as the complexity of the problem for defining the contracting mix increases. After analyzing the solutions, the performance achieved by the Local Search and Simulated Annealing heuristics stands out. As new contract modalities and durations are added to the optimization scenarios, increasing the complexity of resolving the charter mix, Simulated Annealing proved to be more effective in the experiments carried out, achieving better compositions of charter, in addition to providing solutions in less computational time compared to other heuristics available in the computational tool. [resumo fornecido pelo autor]