Can we predict short term extreme conditions from 10-minute data only?

Abstract

Abstract In this study we compare 50-year return values of extreme estimates from a model to values estimated from data analysis. The outputs present a suite of extreme wind load conditions relevant for wind turbine design. The input parameters for the model consist of only statistical wind data, rather than high-frequency measurements that are less available. The model generally predicts lower extreme values than prescribed in the IEC wind turbine safety standard and is therefore of great interest to wind turbine manufacturers. This is the first time a systematic validation of the full suite of extreme models is performed in the whole wind speed range. The accuracy of the model predictions is estimated using a comprehensive data set from a complex terrain site. It is found that the mean absolute percentage error between the data analysis and the model outputs lies within the range of 8.1% - 65.8%.