Inteligência computacional aplicada na otimização de parques eólicos offshore

Abstract

With the growing share of wind energy in the Brazilian and global electricity matrix, it becomes increasingly important to understand the factors that influence the performance of wind farms. The optimization of the turbine layout is essential to maximize energy generation and reduce the losses caused by the wake effect, which decreases wind speed downstream and increases turbulence, affecting the extracted power and the lifespan of the equipment. Among the models proposed to represent this phenomenon, the Jensen model is widely used due to its simplicity and effectiveness, and it is adopted in this work as the basis for wake modeling. Given the complex and nonlinear nature of the wind farm layout optimization problem, it is necessary to employ advanced computational techniques capable of efficiently exploring the search space. In this study, three global optimization metaheuristics are evaluated: BA, GWO and MVO. The main objective is to compare the performance of these techniques in generating more efficient offshore wind farm layouts, using exclusively the Jensen model to estimate energy losses. The methodology was applied to two real offshore wind farms in the environmental licensing process by IBAMA, named Pedra Grande and Camocim, both located on the northeastern coast of Brazil. This application made it possible to analyze the specific characteristics of each farm, as well as to evaluate in detail the efficiency of the algorithms regarding solution quality, convergence and energy gain obtained after the optimization of the layouts. The results highlight marked differences in the behavior of the three metaheuristics. The study reinforces the importance of properly choosing the metaheuristic for the layout optimization problem and demonstrates that it is possible to obtain improvements in the estimation of the energy generated by the wind farm.