Are Sea State Measurements Required for Fatigue Load Monitoring of Offshore Wind Turbines?

Neural network algorithms have shown the capability to infer the actual wind turbine loading from standard signals commonly used for operational control purposes. Fatigue load monitoring done with this readily available data, can offer a robust and cost effective alternative to conventional maintenance-intensive mechanical stress measurement devices. The concept needs to be adopted to offshore wind turbines, where the exposure to the harsh environment with rather difficult accessibility makes the use particularly attractive. At such a site the impact of hydro-dynamically dominated loads might result in poor fatigue estimates, which is due to the lack of information on the surrounding sea state. In order to avoid the need of measuring-buoys, this work studies the employment of additional accelerometers mounted directly at the structure. Various potential placements and three sub-structure types are considered to account for the characteristic structural response caused by wave induced loading. The feasibility is demonstrated on generic data, gained from simulations. Recommended practices are deduced and applied to data from the AREVA M5000 turbine at "alpha ventus".