Forecasting is one of the main challenges facing renewable energies. “Wind and solar power depend on the weather and climate, which are changeable, so they generate electricity intermittently,” explains Marie Berthelot, head of the EDF R&D project on the Impact of Meteorology on Balancing Supply and Demand. How then can we predict the availability of the necessary electricity so we can meet the demand from customers? This is where generation prediction software comes in. Work first began in 2008 for wind farms, and 2013 for photovoltaics. The principle is as follows: “We needed a statistical tool which was capable of relating weather to electricity generation,” says Marie. “We can now supply generation forecasts up to 14 days ahead, based on weather and climate data such as solar radiation, temperature, wind, cloud cover and the diurnal solar cycle – for example the sunshine varies between dawn and midday.” “We can provide generation forecasts up to 14 days ahead” Meteorology: a relatively new specialismIn order to achieve accurate forecasts, researchers rely on a variety of documentation, including models from Météo France, data from the European Centre for Medium-Range Weather Forecasts (ECMWF), and daily breakdowns for each form of renewable generation supplied by French transmission system operator RTE. In order to fine-tune the predictions, especially at local level, sensors are also available in parks, and cameras observe cloud cover. Marie comments, “On the third day following our prediction, we are accurate at national level; there is obviously a – relatively small – margin of error, because meteorology is a relatively new specialism. We are therefore able to amend our data at all times.This ability to predict is welcome, particularly for the French distribution network ERDF, as it is a means of avoiding waste since this sort of energy is both intermittent and ‘unavoidable’: it will be lost if not used when it comes on-stream. “This energy is therefore our basic production from which to supply the grid, which underpins the other sources, such as large hydroelectric plants, which are easier to regulate,” states Marie. This software could prove invaluable at a time when renewables are powering ahead: photovoltaic and wind generation now represent 4,000 and 8,000 MW of installed generation capacity respectively, compared to just 30 and 1,000 MW in 2008.To go furtherHow wind forecasting can predict wind turbine output?