Performance of WRF Wind Speed Forecast in La Rumorosa Wind Farm, Sierra De Juarez, Baja California, Mexico

In recent years wind energy production has shown a rapid growth in the world and in Mexico, this has required the study of instantaneous energy production. Parque Eolico La Rumorosa Wind Farm is composed of five Gamesa G87 wind turbines of 2 MW each, managed by Comisión Estatal de Energía de Baja California, in Sierra de Juárez, B. C. Mexico. Located approximately 14 km south of the border with United States, at 32.49"N, 116.09"W, and 1300 meters msl. 10-minute wind velocity data was obtained and analysed from over-the-turbine cup anemometers. As wind is variable in space as well as in time, usually wind resource is evaluated using mean wind speed values, but this approach should be used with caution because error in wind speed increases the error of power prediction with a factor to the cube of speed, hence mean speed values are not useful for daily wind power dispatch operations or to justify new large scale wind projects, therefore, it is of great interest to produce forecast to reduce uncertainty. Short-term forecasts (ranging from 1 h up to 72 h) are used in power system planning for unit commitment and dispatch, and for electricity trading in certain electricity markets where wind power and storage can be traded or hedge. Short term wind power forecasts can be used for optimization of the management of a power system and optimal trading of wind production in an electricity market. In this work the wind speed forecast from the WRF prediction model is evaluated and therefore the behavior of wind turbines. The effect mesoscale and synoptic processes is observed from wind farm data and then tested if WRF could replicate those events, used to perform wind power forecasts for a wind farm. Simulation results were compared with field observations that are measurements of wind speed and direction obtained from the instruments in the wind turbine nacelle. To initialize simulations and observe meteorological events, NAM and GFS data was used and WRF downscaling for the case study.