The National Center for Atmospheric Research (NCAR) says it and a national team of scientists have developed a new system that could help save the solar energy industry hundreds of millions of dollars through improved forecasts of the atmosphere.
The new forecasting system, known as Sun4Cast, has been in development for three years by the NCAR in collaboration with government labs, universities, utilities, and commercial firms across the U.S. Funded by the U.S. Department of Energy SunShot Initiative, the system greatly improves predictions of clouds and other atmospheric conditions that influence the amount of energy generated by solar arrays, says NCAR.
After testing Sun4Cast at multiple sites, the research team has determined that the system can be up to 50% more accurate than current solar power forecasts. This improved accuracy will enable utilities to deploy solar energy more reliably and inexpensively, reducing the need to purchase energy on the spot market.
As a result, utilities across the U.S. may be able to save an estimated $455 million through 2040 as they use more solar energy, according to an analysis by NCAR economist Jeffrey Lazo.
NCAR, which does not provide operational forecasts, is making the technology available so it can be adapted by utilities or private forecasting companies. The research is being highlighted in more than 20 peer-reviewed papers.
“These results can help enable the nation’s expanding use of solar energy,” says Sue Ellen Haupt, director of NCAR’s Weather Systems and Assessment Program, who led the research team. “More accurate predictions are vital for making solar energy more reliable and cost-effective.”
NCAR says the work builds on its experience in highly detailed atmospheric prediction, including the design of an advanced wind energy forecasting system, and utility Xcel Energy is already beginning to use the new solar forecasting system to forecast conditions at several of its main solar facilities.
“Our previous experience with the National Center for Atmospheric Research in developing a wind forecasting system has saved millions of dollars and has been highly beneficial for our customers,” says Drake Bartlett, senior trading analyst for Xcel Energy – Colorado. “It is our sincere hope and belief that we will see positive atmospheric forecasting results for predicting solar generation as well, again to the benefit of our Xcel Energy customers.”
Using a combination of advanced computer models, atmospheric observations, and artificial intelligence techniques, Sun4Cast provides zero- to six-hour nowcasts of solar irradiance and the resulting power production for specific solar facilities at 15-minute intervals, according to NCAR. In addition, forecasts extend out to 72 hours, allowing utility officials to make decisions in advance for balancing solar with other sources of energy.
NCAR says solar irradiance is notoriously difficult to predict. It is affected not just by the locations and types of clouds, but also by a myriad of other atmospheric conditions, such as the amount of dust and other particles in the air, relative humidity and air pollution. Further complicating the forecast, freshly fallen snow, nearby steep mountainsides, or even passing cumulus clouds can reflect sunlight in a way that can increase the amount of energy produced by solar panels.
To design a system to forecast solar energy output, NCAR and its partners drew on an array of observing instruments, including satellites, radars and sky imagers; specialized software; and mathematical and artificial intelligence techniques. Central to Sun4Cast is a new computer model of the atmosphere that simulates solar irradiance based on meteorological conditions. Called WRF-Solar, the model is derived from the NCAR-based Weather Research and Forecasting (WRF) model, which is widely used by meteorological agencies worldwide.
The team tested the system in geographically diverse areas, including Long Island, New York; the Colorado mountains; and coastal California.
“We have to provide utilities with confidence that the system maintains a high degree of accuracy year-round in very different types of terrain,” says NCAR Program Manager for Renewable Energy.
NCAR says that in addition to aiding the solar power industry, the work can also improve weather forecasting, in general, because of improved cloud prediction.
NCAR’s numerous partners on the project in the public and private sectors included the following:
Government labs: the National Renewable Energy Laboratory, the Brookhaven National Laboratory, the National Oceanic and Atmospheric Administration’s (NOAA) Earth System Research Laboratory, and other NOAA facilities.
Universities: Pennsylvania State University, Colorado State University, the University of Hawaii, and the University of Washington.
Utilities: The Long Island Power Authority, the New York Power Authority, the Public Service Company of Colorado, the Sacramento Municipal Utility District (SMUD), Southern California Edison, and Hawaiian Electric Co.
Independent system operators: New York ISO, Xcel Energy, SMUD, California ISO, and Hawaiian Electric.
Commercial forecast providers: Schneider Electric, Atmospheric and Environmental Research, Global Weather Corp., MDA Information Systems, and Solar Consulting Services.
Computing time was provided by the New York State Department of Economic Development’s division of science, technology and innovation on an IBM Blue Gene supercomputer at Brookhaven National Laboratory. Researchers also performed computing at the NCAR-Wyoming Supercomputing Center and the DOE National Energy Research Scientific Computing Center.