CX-102095: Learned Productivity Under Variable Solar Conditions

Award Number: DE-EE0009352
CX(s) Applied: A9
Solar Energy Technologies Office
Location(s): WI
Office(s): Golden Field Office

The U.S. Department of Energy (DOE) is proposing to provide funding to the University of Wisconsin – Madison to develop a model that accurately predicts receiver thermal production of concentrating solar power (CSP) plants during variable solar conditions. Historical data would be collected including ground station measurements, ground-to-sky camera images and processing information, local heliostat photovoltaic module output, and historical forecasts from satellite or statistical models. This data would be used in machine learning and deep learning to replicate operating patterns and create a model that identifies optimal operation schedules for CSP plant operators. After determining viability of this method of operation for use in operator decision-making, methodological details would be published for broader use. Project activities would occur at the University of Wisconsin in Madison, BrightSource Energy in Israel, and the National Renewable Energy Laboratory in Golden, CO.