Drivetrain fatigue strength characteristics of model‐predictive control for wind turbines

Abstract

AbstractWind turbines play a crucial role in the revolution towards renewable resources. They need to be economically competitive to be sustainable. This still requires to lower the cost of energy (COE). To this end, nonlinear model predictive control (NMPC) is used within this paper. As known from literature, NMPC significantly improves the energy extracting performance as well as the mitigation of tower loads of wind turbines. As it is shown in this paper, the drivetrain fatigue strength drops disproportionally in parallel, which either increases the demands on the turbine design or decreases the lifetime of drivetrain components. Without additional boundary conditions to the underlying optimal control problems (OCPs), the application of energy‐maximizing NMPC might so even raise the COE. Only penalizing axial torque oscillations by quadratic terms decreases energy‐extracting performance below the level of classical wind turbine controllers. This makes more sophisticated conditions necessary. In this paper, the increase of the drivetrain damage by NMPC is analyzed, and appropriate boundary conditions are derived, to balance the two contradicting objectives of energy maximization and drivetrain load mitigation. An NMPC approach based on indirect methods is used, to obtain a solution of the OCPs very efficiently. It applies the Hamilton equations and Pontryagin's maximum principle. Its accuracy and efficiency to solve OCPs was presented over the last decades.