Floating offshore wind farm responses prediction under wake steering comparing machine learning models

Abstract

Wind energy is important in advancing towards sustainable and cleaner energy sources with wind turbines being key in converting wind's kinetic energy into electricity. Like their onshore and fixed-bottom offshore counterparts, floating offshore wind farms experience wake effects that impact their efficiency. However, due to the lower surface roughness offshore, wake recovery is slower compared to onshore wind farms. Therefore, wake mitigation strategies, such as yaw misalignment to divert wakes away from downwind turbines, are crucial for floating offshore wind farms. Investigating the potential impacts of yaw misalignment on wind turbines within a wind farm to optimize performance and efficiency is essential.

The first part of this work investigates the influence of yaw misalignments on a 5MW wind turbine mounted on spar buoy floating platform in a 2x3 wind farm setting. The thesis focuses on motion response, generated power, and blade root stress derived from uniaxial loads. For various yaw settings, both steady and turbulent wind conditions are analysed. Results indicate that power output, platform motion, and blade root stress are all sensitive to changes in yaw angle.

Predicting wind turbine performance is crucial for optimizing energy production as wind energy expands rapidly. The complex interactions among control systems, aerodynamics, structural dynamics, and hydrodynamics of floating offshore wind turbines make their design and optimization particularly challenging. However, the second part of this thesis introduces machine learning models for forecasting the dynamic performance of floating offshore wind turbines. The thesis findings show promising results using a backpropagation neural network, regression model, a gaussian process regression and an ensemble model highlighting models' effectiveness in accurately forecasting turbine performance and, thereby, enhancing the efficiency of wind energy production.