Concentrating solar power (CSP) plants, when combined with thermal energy storage systems, can supply electricity to the grid at any time. The plants have a bright future in countries with sufficient sunlight, where they could potentially round out solar photovoltaic energy production.

However, grid operators need models to predict CSP production in order to effectively manage the energy supplied to the grid and improve plant operation and energy-storage strategies. Such models do exist for PV plants, but there is currently no equivalent for CSP plants.

Liten researchers used a direct solar irradiance prediction tool developed at the institute to come up with a flexible model that can actually learn from its mistakes (gaps between forecasts and actual measurements) and become more accurate over time. The model can be used to predict short-term electricity production to stabilize the grid and maintain a baseline production level. Medium-term predictions can be used to determine plant operating strategies. The model was tested successfully at the CEA facility in Cadarache, in the south of France, where it is being used to improve the operation of complex energy systems combining production, storage, and desalination.