One of the primary challenges in diagnostic health monitoring and control of wind turbines is compensating for the variable nature of wind loads. Given the sometimes large variations in wind speed, direction, and other operational variables (like wind shear), this paper proposes a data-driven, online rotor model identification approach. A 2 m diameter horizontal axis wind turbine rotor is first tested using experimental modal analysis techniques. Through the use of the Complex Mode Indication Function, the dominant natural frequencies and mode shapes of dynamic response of the rotor are estimated (including repeated and pseudo-repeated roots). The free dynamic response properties of the stationary rotor are compared to the forced response of the operational rotor while it is being subjected to wind and rotordynamic loads. It is demonstrated that both narrowband (rotordynamic) and broadband (wind driven) responses are amplified near resonant frequencies of the rotor. Blade loads in the flap direction of the rotor are also estimated through matrix inversion for a simulated set of rotor blade input forces and for the operational loading state of the wind turbine in a steady state condition. The analytical estimates are shown to be accurate at frequencies for which the ordinary coherence functions are near unity. The loads in operation are shown to be largest at points mid-way along the span of the blade and on one of the three blades suggesting this method could be used for usage monitoring. Based on these results, it is proposed that a measurement of upstream wind velocity will provide enhanced models for diagnostics and control by providing a leading indicator of disturbances in the loads.
Paint it black: Efficacy of increased wind turbine rotor blade visibility to reduce avian fatalities
