Why would an intelligent utility care about advancements in solar forecasting? Today, thanks to work at the University of California San Diego (UCSD), there are at least two answers to that question.

First, utilities with utility-scale solar generation already use forecasting, employing satellite imagery to predict expected weather conditions to anticipate the output of their solar facilities many hours ahead. But satellite imagery covers large areas and doesn't provide the local granularity for very short-term forecasting that might affect decisions about balancing supply and load. 

Second, due to advancements at UCSD in solar forecasting for its own microgrid, a technique is being developed that could help an islanded microgrid with a significant amount of variable, renewable energy sources balance its power supply and load. The intelligent utility in this case is San Diego Gas & Electric, which is supportive of UCSD's microgrid for several reasons. One very practical reason: if SDG&E needs to shed load in an emergency, UCSD can island itself and maintain its vital functions, which include hospitals and, for instance, a collection of ice cores that represent the history of global climate. To do so, forecasting the output of its solar photovoltaic arrays can help it balance power supply and load.

(You may recall we discussed aspects of UCSD's microgrid and how it supports SDG&E in "The Future Grid: A Seamless Ebb and Flow of Supply and Demand?")

That's the "why." Now, the "how."

Enter Jan Kleissl, UCSD professor of environmental engineering, who has combined his research interest in meteorology with engineering. In seeking to advance solar forecasting for relatively small locations such as the UCSD campus, which is 1,200 acres or roughly two square miles, he noted to me that one advantage is that sunshine and clouds are visible and measurable.

Where existing satellite imagery provides only a macro picture of, in this case, the UCSD campus, Kleissl has worked on a camera system that looks up from a single point on the ground with a fisheye lens to capture images of the sun and clouds overhead. The camera feeds a computer system with software that can discern variables such as the clouds' opacity, direction and speed and kick out forecasts of 10- to 30-minutes ahead. That's enough for the UCSD microgrid's master controller to take action, shedding load and perhaps stepping up its power output from a central, natural gas-fired generator to balance the system. Load-shedding in this example might include, say, throttling back the air conditioning in campus buildings.

The camera equipment itself is all drawn from basic gear available to anyone, but the secret sauce here is the software programs that process the imagery and translate it into meaningful, short-term predictions that allow the campus microgrid's master controller to do its balancing act, Kleissl told me.

Like nearly all such technical advancements, this one is proceeding through various phases from idea to proof of concept to real-time demonstration. With funding from the U.S. Department of Energy, the California Energy Commission, the California Public Utilities Commission and Sanyo, Kleissl told me he and his UCSD colleagues are only weeks away from the new system's first real-time demonstration.

And prime-time? UCSD has a devotion to lab-to-market commercialization and companies and other organizations keep a close eye on advancements there in their efforts to license cutting-edge technology that could reap benefits and profits. Kleissl said that more work is needed on the new system's hardware and software, but that is expected. He predicted that a year or two may bring commercialization.

How unique is this work to UCSD?

"We're probably the first ones to be serious about using cameras for localized, solar forecasting," Kleissl told me.

Though similar work is taking place in France and Greece, he said, "we're ahead."

I mentioned that science journalists are trained to avoid the word "breakthrough." Kleissl agreed.

"This isn't a `breakthrough,'" he acknowledged. "As a realist, I would call it an `incremental advancement.'"

Phil Carson
Intelligent Utility Daily