Optimization of the spatial disposition of an array of floating offshore wind turbines using a Python genetic algorithm

Electricity has a social value as well as a monetary one, which is constantly increasing. It is a strategic asset that has become indispensable for years. For this reason, the energy production sector is fundamental, as is sustainable production, given the ongoing climate change. The aim of this study is to optimize the spatial arrangement of an off-shore floating wind turbine array in such a way as to increase the efficiency and maximize the energy yield in relation to the cost of the initial investment. The problem was tackled by combining two different analyses: a fluid dynamics one and an optimization one. The first one allowed to simulate the aerodynamic interactions between the turbines and the wind speed field, using analytical relations such as the Jensen model and the Bastankhah Gaussian model for the calculation of the wakes, accompanied by models for the calculation of the blockage effect, models for the turbulence, to simulate the ground effect or to model the rotors. The use of such models avoided using complex fluid dynamics simulation methods such as CFD, allowing to reduce the computational costs and enabling the use of heuristic optimization algorithms. The second one, indeed, allowed to find the optimal arrangement of the turbines by means of a genetic algorithm that selected the best individual from generation to generation, that is, the disposition with the best performance. To perform these calculations, two different Python libraries were used: PyWake for the aerodynamic calculations, and Pymoo for the evolutionary algorithm. The investigation did not limit itself to an analysis of producibility but also considered economic aspects of the investment. The results show the existence of an optimal number of wind turbines to be inserted in a specific domain. They also show that for the case study, the arrangement is strongly influenced by the dominant wind direction. The algorithm proved to be stable and convergent. This thesis contributes to provide solutions for the design of wind farms by suggesting a synthesis between accuracy of results and computational intensity.