Comparison of wind turbine fatigue loads between standard site assessment practice and well-resolved LES

Abstract

In this paper, a comparison of blade root and tower base fatigue loads computed using turbulent wind fields from different models is conducted. First, a well-resolved LES wind field is generated using AMR-Wind and is assumed to be the closest representation of real turbulent wind. From this wind field, key characteristics typically available from a site assessment, such as hub-height wind speed, hub-height turbulence intensity, and wind shear are extracted. These values are then used to generate wind fields based on the Kaimal and Mann spectral models. All other input parameters required for these models are taken from the IEC 61400-1, Edition 4 standard. The subsequent fatigue simulations show that the LES generally predicts higher loads than the Kaimal and Mann models for the analyzed wind speed of 7 m/s. An exception is the blade edgewise moment, for which the choice of wind field has no significant effect on the fatigue load. These differences can be attributed to the absence of large-scale turbulent motions in the spectral models compared to LES. This study highlights the importance of using site-specific integral length scales when generating artificial spectral turbulence.