VTechWorks staff will be away for the Thanksgiving holiday beginning at noon on Wednesday, November 22, through Friday, November 24, and will not be replying to requests during this time. Thank you for your patience, and happy holidays!
NCAR's Recent Advances in Wind Power Forecasting
Delle Monache, Luca
MetadataShow full item record
The National Center for Atmospheric Research (NCAR) has been developing and enhancing a Wind Power Forecasting System in partnership with Xcel Energy that integrates high resolution and publically available modeling capabilities with artificial intelligence methods. This forecasting system has recently been updated to include specific technologies for uncertainty quantification, short-term detection of wind power ramps by blending the Variational Doppler Radar Analysis System with an Expert System, enhanced wind-to-power conversion techniques, and prediction of icing and snow conditions. The system ingests external, publicly available weather model data and observations. In order to provide information specific to Xcel Energy's region, high resolution numerical weather prediction (NWP) simulations assimilate specific local weather observations. The weather observations range from routine meteorological surface and upper air data to data from the wind farms, including wind speed data from the Nacelles. Finally, to optimize estimates of short-term changes in wind power requires nowcasting technologies such as the Variational Doppler Radar Analysis System (VDRAS) that is blended with an observation-based Expert System. The strength of the wind power forecasting system lies in blending the various components to produce a power forecast that can be used by Xcel Energy grid operators and energy traders. That blending is accomplished with the Dynamical Integrated Forecast (DICast) System. The wind speeds predicted by DICast must then be translated into power using NCAR's empirical power conversion algorithms. A quantile approach is used to quality control the observations. The uncertainty of these predictions can then be quantified using an Analog Ensemble (AnEn) approach. Finally, warnings of potential icing and heavy snow are provided.