Combining unsteady blade pressure measurements and a free-wake vortex model to investigate the cycle-to-cycle variations in wind turbine aerodynamic blade loads in yaw

Abstract

Prediction of the unsteady aerodynamic flow phenomenon on wind turbines is challenging

and still subject to considerable uncertainty. Under yawed rotor conditions, the wind turbine blades

are subjected to unsteady flow conditions as a result of the blade advancing and retreating effect and

the development of a skewed vortical wake created downstream of the rotor plane. Blade surface

pressure measurements conducted on the NREL Phase VI rotor in yawed conditions have shown that

dynamic stall causes the wind turbine blades to experience significant cycle-to-cycle variations in

aerodynamic loading. These effects were observed even though the rotor was subjected to a fixed

speed and a uniform and steady wind flow. This phenomenon is not normally predicted by existing

dynamic stall models integrated in wind turbine design codes. This paper couples blade pressure

measurements from the NREL Phase VI rotor to a free-wake vortex model to derive the angle of

attack time series at the different blade sections over multiple rotor rotations and three different yaw

angles. Through the adopted approach it was possible to investigate how the rotor self-induced

aerodynamic load fluctuations influence the unsteady variations in the blade angles of attack and

induced velocities. The hysteresis loops for the normal and tangential load coefficients plotted against

the angle of attack were plotted over multiple rotor revolutions. Although cycle-to-cycle variations

in the angles of attack at the different blade radial locations and azimuth positions are found to be

relatively small, the corresponding variations in the normal and tangential load coefficients may be

significant. Following a statistical analysis, it was concluded that the load coefficients follow a normal

distribution at the majority of blade azimuth angles and radial locations. The results of this study

provide further insight on how existing engineering models for dynamic stall may be improved

through the integration of stochastic models to be able to account for the cycle-to-cycle variability in

the unsteady wind turbine blade loads under yawed conditions.