14:45 〜 15:00
[MAG32-04] Evaluation of a Scanning Doppler LiDAR-based Wind Field Observation in LES
キーワード：Large-eddy simulation (LES), Scanning Doppler LiDAR, Mean Wind Speed, Turbulence
Doppler LiDARs are considered as promising alternative to meteorological masts for wind resource assessments for wind energy application. The current study models a single scanning LiDAR-based wind field measurements in the LES and quantify the effect of scan parameters, i.e, measurement range, azimuth and elevation angles and wind direction on the accuracy of two-parameter velocity volume processing (VVP) method for computing velocity vectors from radial wind speeds. The study also proposes a modification to the existing method for computing turbulence statistics based on Reynolds decomposition. To that end, simulations of the ABL are performed using the open source CFD software OpenFOAM v2206. We assume neutrally stratified ABL and model the flow fields using LES. Wind speed vectors is computed from multivariant fitting of radial wind speeds. The study assumes spatial and temporal homogeneity during a short scan cycle. Furthermore, because the elevation angle is very small (5° or less) in most measurement campaigns, vertical component is considered negligibly small.
The mean wind speeds computed from LiDAR measurements show good agreement with the original LES data. The error increases with the measurement range, but it decreases with azimuth range, with θrange=60° giving the most accurate mean wind speeds among the three azimuth range considered in this study. The wind direction did not particularly affect the accuracy of the mean wind speed estimation, though larger difference between wind direction and scan direction results in increased variation in the VVP fitting. The effect of elevation angle is investigated with lower elevation angle scan of 3.4° . Although stronger shear near the ground led to larger difference between the LiDAR and LES data, for higher points the effect of vertical shear on mean wind speeds is not significant. In terms of turbulence intensities, the two-parameter VVP significantly underestimates their values for all the case considered in this study. This is because a significant fraction of the fluctuating components is filtered out while fitting the data over the scan arc. The study therefore, proposes an improvement to the conventional VVP method, based on the Reynolds decomposition of wind speed components. Turbulence intensities estimated using this method show higher degree of variation, though the accuracy improved with increasing azimuth range.
Acknowledgements
Jay Prakash Goit acknowledges support from Amano Institute of Technology, Public Interest Foundation. The computations were performed on Wisteria/BDEC-01 supercomputer system of The University of Tokyo.
The mean wind speeds computed from LiDAR measurements show good agreement with the original LES data. The error increases with the measurement range, but it decreases with azimuth range, with θrange=60° giving the most accurate mean wind speeds among the three azimuth range considered in this study. The wind direction did not particularly affect the accuracy of the mean wind speed estimation, though larger difference between wind direction and scan direction results in increased variation in the VVP fitting. The effect of elevation angle is investigated with lower elevation angle scan of 3.4° . Although stronger shear near the ground led to larger difference between the LiDAR and LES data, for higher points the effect of vertical shear on mean wind speeds is not significant. In terms of turbulence intensities, the two-parameter VVP significantly underestimates their values for all the case considered in this study. This is because a significant fraction of the fluctuating components is filtered out while fitting the data over the scan arc. The study therefore, proposes an improvement to the conventional VVP method, based on the Reynolds decomposition of wind speed components. Turbulence intensities estimated using this method show higher degree of variation, though the accuracy improved with increasing azimuth range.
Acknowledgements
Jay Prakash Goit acknowledges support from Amano Institute of Technology, Public Interest Foundation. The computations were performed on Wisteria/BDEC-01 supercomputer system of The University of Tokyo.