Operação de redes de escoamento de petróleo utilizando algoritmo genético multiobjetivo.

Abstract

The optimization of the operation in oil pipeline networks is essential to guarantee the attendance of constraints and to maximize the efficiency of the operational objectives, in order to draw out its useful life while keeping the oil production in the requirements of quantity and quality with rationality of the costs of electric energy. The use of computational optimization models for solving this type of problem, allows to offer a more rational operation, diminishing the subjectivity, the risks of imperfections and the conflicts that generally occur when this operation are made decentralized and based solely in the experience acquired in field. Genetic algorithms have been used to optimize these systems, and it detaches for its robust nature, that allows associating its effectiveness and efficiency characteristics. Recently, has been developed various multi-objective genetic algorithms, which treat more reliability complex problems which need simultaneous optimization of diverse objectives. This work analyzes the incorporation of diverse techniques and in the Strength Pareto Evolutionary Algorithm 2, with the intention of increase its efficiency and becoming applicable to the problem of pump scheduling in oil pipeline network. The incorporated alternatives surround the use of the technique of seeding, of knowledge for repair and discarding of infeasible solutions, the integration of the concept of viability with the dominance criteria, the introduction of knowledge in the reproduction operators, among others. The experiments are based in a small part of the pumping network of “Unidade de Negócios de Exploração e Produção do Rio Grande do Norte e Ceará”, of PETROBRAS, and the results show an efficient and applicable methodology to the problem, which generates good alternatives for the pump scheduling, for a period of 24 hour, in a skillful time.