Abstract
Wind power development is essential in the fight against global warming. However, it can negatively impact local ecosystems, particularly birds, as they collide with the wind turbines. This thesis introduces a novel approach to bird strike mitigation in wind turbines, inspired by the SKARV concept which is patented by SINTEF Energy Research. We realize the SKARV concept through uncertainty-aware nonlinear model predictive control (NMPC) to ensure that the wind turbine blades avoid predicted collisions with approaching birds. To achieve bird avoidance, a bird-blade distance constraint is defined to ensure a minimum safety distance between predicted bird trajectories and the turbine blades. Multi-stage NMPC is employed to account for uncertain bird trajectories. The uncertainties associated with the predicted bird trajectories is quantified through linear uncertainty propagation. The uncertainty is used to compute possible bird trajectory scenarios, which are optimized in parallel by the multi-stage NMPC. Additionally, a back-off NMPC scheme is tested, approximating multi-stage NMPC by increasing the safety distance based on the bird trajectory uncertainty rather than formulating multiple scenarios. The proposed uncertainty-aware approaches are validated in simulations against a non-uncertainty-aware nominal NMPC, simulated with linear, noisy bird trajectories and a nonlinear wind turbine model. Results indicate that the multi-stage NMPC is the most promising in avoiding bird strikes with an avoidance rate of 81.1%, closely followed by the back-off NMPC with an avoidance rate of 80.1%. The nominal NMPC responded slowly to birds that were initially predicted to be outside collision danger, leading to abrupt control input adjustments and a lower bird avoidance rate of 73.8%. These results show the importance of integrating uncertainty into the control scheme. The multi-stage NMPC resulted in an estimated annual energy production loss of 8.90*10^-4% versus 7.97*10-4% with the back-off NMPC and 7.39*10^-4% with the nominal NMPC. Despite the slightly increased energy loss with the uncertainty-aware approaches, a loss in energy to achieve bird strike avoidance may be economically justified, considering potential fines, harm to the ecosystem, and reputational damage from having excessive bird fatalities. Overall, the findings advocate adopting uncertainty-aware NMPC techniques to preserve avian populations around wind farms.