A prototype WRF-based wind power forecasting tool for Argentina

SAULO, C.; WAIMANN, C.; PINSON, PIERRE

Montreal

Conferencia; World Weather Open Science Conference; 2014

Wind power forecasting has become an important service for the appropriate management of wind power generation, particularly in those countries where there is a significant penetration of wind energy into their electricity markets. Although Argentina is at an early stage in the exploitation of wind energy (there is only 218 MW of installed capacity as of 2013), there are near-term plans to further deploy renewable energy capacities, with more than 1000 MW in wind power projects to be executed. This research is part of a series of studies whose objectives are the creation and validation of dynamic-statistical methodologies for wind power forecasting in Rawson Wind Park, located in northeastern Patagonia region and consisting of 43 wind turbines. To accomplish this, our first task was to convert WRF highresolution wind speed forecasts to wind power predictions trough the implementation of adaptive nonlinear regression models. This relies on a nonparametric methodology applied to find the relationship between wind speed forecasts at different heights and wind power observations from Rawson Wind Park. The techniques employed are inspired by those already developed and applied in Denmark. Then, wind power predictions are combined with an autoregressive linear model in order to better capture short-term local wind power dynamics.A simple reference model based on linear  combination of WRF wind speed forecasts, theoretical power, and an autoregressive term, has been used to quantify the improvement of our proposed methodology for time horizons up to 72 hours ahead. Results show errors slightly larger than those found for Denmark.