Review of Wind Energy Forecasting Methods for Modeling Ramping Events

Wharton, S ; Lundquist, J K ; Marjanovic, N ; Williams, J L ; Rhodes, M ; Chow, T K ; Maxwell, R

关键词: ACCURACY;    BOUNDARY LAYERS;    DETECTION;    EDUCATIONAL FACILITIES;    FORECASTING;    LAWRENCE LIVERMORE NATIONAL LABORATORY;    POWER GENERATION;    REMOTE SENSING;    SIMULATION;    VELOCITY;    WIND POWER;    WIND TURBINE ARRAYS;    WIND TURBINES;   

DOI  :  10.2172/1022139
RP-ID  :  LLNL-TR-476934
PID  :  OSTI ID: 1022139
Others  :  TRN: US201118%%318

学科分类：再生能源与代替技术

美国|英语

来源: SciTech Connect

【 摘 要 】

Tall onshore wind turbines, with hub heights between 80 m and 100 m, can extract large amounts of energy from the atmosphere since they generally encounter higher wind speeds, but they face challenges given the complexity of boundary layer flows. This complexity of the lowest layers of the atmosphere, where wind turbines reside, has made conventional modeling efforts less than ideal. To meet the nation's goal of increasing wind power into the U.S. electrical grid, the accuracy of wind power forecasts must be improved. In this report, the Lawrence Livermore National Laboratory, in collaboration with the University of Colorado at Boulder, University of California at Berkeley, and Colorado School of Mines, evaluates innovative approaches to forecasting sudden changes in wind speed or 'ramping events' at an onshore, multimegawatt wind farm. The forecast simulations are compared to observations of wind speed and direction from tall meteorological towers and a remote-sensing Sound Detection and Ranging (SODAR) instrument. Ramping events, i.e., sudden increases or decreases in wind speed and hence, power generated by a turbine, are especially problematic for wind farm operators. Sudden changes in wind speed or direction can lead to large power generation differences across a wind farm and are very difficult to predict with current forecasting tools. Here, we quantify the ability of three models, mesoscale WRF, WRF-LES, and PF.WRF, which vary in sophistication and required user expertise, to predict three ramping events at a North American wind farm.