Posters

Abstract

Investigations on de-trending and turbine loads Since de-trending of measured wind speed time series for load simulations is allowed according to the most recent edition of IEC61400-1, it is increasingly applied in the wind industry. The manner in which it is applied usually reduces the simulated fatigue loads significantly by decreasing the simulated turbulence. As a turbine manufacturer, we are often asked by our customers to apply de-trending in order to increase suitability and profitability of our turbines. Since we take the responsibility for assuring the structural integrity of our turbines, we need to make sure that this assurance is not violated by the application of de-trending. Therefore we decided to prove the basic concept of this method in order to find out if we can use it or not. The documented investigations do not aim to compare specific de-trending methods. The aim is to challenge the basic concept of de-trending with a strong focus on loads. Artificial wind speed ramps and de-trending Load reference: 10 minute turbulent wind fields that contain a certain linear trend are taken as the input for aero elastic simulations. De-trending and loads: The same wind fields without a linear trend serve as a perfectly de-trended counterpart and deliver de-trended loads. 10 minute trends in 1h simulations Load reference: 1h turbulent wind fields without a trend are taken as the input for aeroelastic simulations (1h loads). De-trending and loads: Decomposition of the 1h wind fields into 10 minute segments, de-trending of the segments, statistical properties of the de-trended segments deliver the input for aeroelastic simulations resulting in de-trended 10 minute loads. Results and Conclusion Comparison of the reference loads and de-trended loads gives an impression of the trend-induced fatigue that is neglected by the application of de-trending. At the same time, the overestimation by not applying de-trending is investigated. Our conclusion is that we cannot apply de-trending without significant underestimation of the fatigue loads. If we don't apply de-trending, we overestimate the loads, but only with a reasonable order of magnitude of around 3%.