Evaluating Wind Power Prediction Uncertainty Using Scanning Doppler Wind Lidar

Wind energy provides an enormous clean electricity resource to help mitigate anthropogenic climate change and stimulate the economy. However, currently, a lack of measurements and understanding of wind condition in the turbine’s rotor layer (approximately 40 -200m) results in poor understanding of wind resources and corresponding power prediction biases. This power prediction uncertainty impedes progress toward optimized project designs and the associated reduced capital and financing costs. During the VERTical Enhanced miXing (VERTEX) campaign, scanning Doppler Wind Lidar and other remote sensing instruments were used to collect data about the wind and atmospheric conditions at varying heights in front of an active turbine. This work develops a new method to reconstruct vertical wind profiles that are known to be in front of the turbine using scanning Doppler Wind Lidar and Optimal Interpolation analysis. These profiles are classified based on the goodness-of-fit to several mathematical expressions and relative speed minimum and maximum criteria. Microwave radiometer measurements are used to classify atmospheric stability conditions and understand relationships with classified wind profiles. Finally, a systematic method incorporating a range of power predicators is used to quantify the relationship between turbine performance uncertainty and site-specific atmospheric conditions. Results demonstrate a general reduction in power prediction uncertainty when using Rotor Equivalent Wind Speed to predict power rather than only hub-height wind speed. However, the magnitude of this improvement increases as classified wind profiles deviate from expected, near power-law, shapes with low shear. Collectively, results demonstrate that remote sensing measurements and the wind profile and stability classification tools introduced in this work advance understanding of how site-specific atmospheric conditions influence turbine performance as well as help determine the most important atmospheric-related power predictors needed to reduce uncertainty. Future work will test the relative value of other atmospheric variables as power predictors and use results to create site-specific power prediction models.