estudo de casos para comparar os resultados obtidos com o programa implementado.Um objetivo adicional deste trabalho é expor à comunidade naval a utilidade e efetividade dos algoritmos genéticos nos problemas de projeto multiobjetivo. O enfoque desta pesquisa está na metodologia implementada e na sua recomendação para o projeto preliminar de navios. Palavras chaves: Algoritmos Genéticos. Navio Tanque (Projeto). OtimizaçãoMatemática. ABSTRACTThis paper shows the rational process of selecting the optimal dimensions and forms coefficients of tanker ships using the technique of genetic algorithm. It is proposed a procedure to balance designs in weight and useful space and assesses their feasibility so that weight and volume limited tankers can be balanced. The elaboration of a procedure and its implementation in an efficient program is so important in the preliminary dimension of a ship, so this paper is a divulgation of the genetic algorithm as robust technique for the preliminary ship design. Two objective attributes are used to evaluate each design: Total Cost and Mean Oil Outflow. This attributes are conflicting between them, that is, a low cost correspond a high outflow and vice versa. The Mean Oil Outflow is evaluated with the probabilistic methodology proposed by the INTERTANKO to IMO. The assessment of mean oil outflow, by a non dimensional parameter, supposes the event of grounding and/or collision resulting in bottom and side damage respectively. In estimating the cost two models are used performing a comparison of their effects on the final values of principal dimension. In addition, for the balance procedure, five hull weight models are tested and analyzed the impact on the form and distribution of Pareto frontier.A genetic algorithm is implemented to search the optimal design parameters and identify non-dominated frontier of Pareto. The algorithm is tested with four test functions found in the technical literature. The test functions selected are convex, non convex, discontinuing and one with four restrictions. The implementation shows similar results with those showed in others papers. A Suezmax, an Aframax, and a Panamax ships are used as case study in order to compare the gotten results by the implemented program.One additional goal of this work is to expose to naval community the usefulness and effectiveness of the genetic algorithm in the Multi-Objective design. The emphasis of this paper is on methodology of design implemented and is suggested for preliminary ship design.

No abstract