Characterization and simulation of inhomogeneous and non-stationary turbulent wind fields for assessment of wind turbine reliability
MetadataShow full item record
Wind turbines are designed to withstand turbulent inflows that are modeled as stationary Gaussian fields, a general assumption frequently made in wind engineering practice that is partly done out of convenience and partly out of our inadequate understanding of the wind itself. While the reliability of wind turbines continues improve, it also continues to be unsatisfactory and necessitates a re-examination of our assumptions about the wind and how they relate to structural reliability. Using measurements from the 200 m tower at Texas Tech University, the statistical characteristics of wind fields are analyzed and many are shown to deviate from assumed behavior in certain conditions. In particular, wind fields in the stable boundary layer, especially during low-level jet events, are often inhomogeneous in nature, with turbulent winds described by different probability distributions at different heights, while conditions during thunderstorm events are seen to be both inhomogeneous and non-stationary. In order to investigate aerodynamic loads and fatigue on wind turbines due to inhomogeneous inflows, a method for the simulation of non-Gaussian fields is presented based on translations of Gaussian fields. The method allows the simulation of inhomogeneous, non-Gaussian fields that match skewness and kurtosis of local turbulence probability distributions and is combined with conditional simulation to assimilate a subset of measured data points within the field. The effects of the inhomogeneous turbulence as well as direction shear measured during low-level jet events are investigated via the wind turbine simulation software FAST for a 1.5 MW and 5 MW wind turbine and compared the International Technical Commission’s IEC 61400-1 wind turbine design guidelines. Generally the fatigue damage resulting from non-Gaussian inflows are found to be lower than that induced by Gaussian inflows for cases with realistic turbulence intensities seen during LLJ events, however at higher turbulence intensities, small increases (~1%) in fatigue were noted for inhomogeneous inflows and inflows using constant positive skewness and kurtosis. At the same time, however, these inflows are found to be correlated with increased peak loading. Wind fields simulated with direction shear, are shown to substantially increase fatigue (11-60 %) on wind turbine components compared to those simulated without direction shear.