As the noticeable buzzing sound coming from a panel of solar power inverters begins to fade, Rubin York cranes his head and looks skyward. Seconds later, only the whoosh of a strong breeze is audible as the inverters fall silent, signaling the loss of solar generation from sections of OUC’s Kenneth P. Ksionek Community Solar Farm. York turns to a small group of University of Central Florida engineering students visiting the site and shrugs. Clouds. Patches of big, puffy stratocumulus clouds moving in from the west are passing overhead. When sunlight breaks through a gap in them, the inverters begin to hum again, only to power down when another batch of clouds gets between the sun and the rows of photovoltaic panels.
One minute thousands of PV modules are basking in the glow of sunlight; the next, they’re darkened by shade. This is the ebb and flow of the life of a solar farm in the Sunshine State, a nickname that is only partly accurate.
Florida spends a lot of time under cloud cover, a weather pattern caused by warm waters hugging its coastlines, sea breezes coming from the east and west, high humidity and long rainy seasons. Orlando, the home of OUC – The Reliable One, sees 277 days of partly cloudy or cloudy weather a year, according to data from the National Oceanic and Atmospheric Administration.
Clouds, or more precisely the shadows they cast during daylight, are why York and fellow OUC Emerging Technologies Project Engineer Timothy Remo are showing five UCF seniors around this 13-MW solar field built atop a byproduct landfill overlooking the Stanton Energy Center in east Orlando. The students are part of an eight-member team working on phase two of a cloud-tracking technology that could help OUC forecast the arrival of clouds over solar fields and predict the impact they’re likely to have on power output.
OUC considers cloud data crucial to its ability to manage the fluctuations of daytime solar energy. Knowing how soon and for how long clouds will cover part or all of a solar farm tied to its grid gives the municipal utility time to ramp up other resources and seamlessly fill gaps in generation.
“We’re basically trying to figure out with this device if we can accurately forecast cloud development and movement,” explained Justin Kramer, supervisor of emerging technologies. “We also want to predict which solar arrays will be in shadows created by cloud cover and for how long. Give me over the next half hour what’s going to be shaded and what’s going to come out of the shade so I know the net effect on the grid.”
There’s a sense of urgency driving this project as OUC plans to increase its solar portfolio to 270.5 megawatts by 2025, making it the leading utility in the state in solar watts per customer. Related to OUC’s investment in clean energy is its commitment, announced in February, to achieve net-zero carbon emissions by 2050 and reach a goal of reducing carbon emissions by 50 percent from 2005 levels by 2030.
“As we see our penetration of PV going higher and higher, we realize we need to better understand weather and its possible impact on solar farms,” said Kramer. “My team met with dispatch and operations people, and we got feedback on what they are looking for to best operate the grid with an increased presence of solar. One of the things they asked for was the ability to view the sky to see whether clouds were going to be temporary or permanent. Satellite data is a little delayed and we needed something more real time.”
Kramer and his ET team found cameras that captured panoramic images of the sky and provided some cloud movement information, but they were expensive. Placing them at 22 weather stations, as Kramer wants to do, would have run a couple hundred thousand dollars.
So, Kramer approached Mark Steiner, professor of mechanical and aerospace engi-neering at UCF, with this initial challenge for class of 2019 engineering students: Build a camera-based, cloud-tracking device that gives altitude, density, direction and speed of clouds, and predicts how soon they’ll float over a solar farm.
And build the entire system on a budget of $1,000.
Thus began a UCF senior design project called “Combating Climate Change with Cloud Tracking.” Five students — two aspiring mechanical engineers and three future computer scientists — took on the first phase of the project, building a prototype system using off-the-shelf parts, open-source software and original algorithms. The students also built a website for assessing and storing data, wrapping up their contributions to the project in December 2019.
“It seemed like a rational challenge,” said Steiner. “It wasn’t simple. We felt we could do it and we did.”
The sky cam cloud-tracking prototype is a plastic box the size of a carry-on suitcase, with a surveillance-style fisheye camera mounted on the lid. Inside are a battery, motherboard and other electronic components. Its nondescript looks belie the box’s sophistication.
“We spent $15,000 and they created a prototype for us,” said Kramer, adding the sky cam cost $872 in materials and Steiner received a stipend as faculty adviser. “Had we gone out into the open market and had it designed, we probably would have spent hundreds of thousands of dollars. We were really pleased with the results. They exceeded our expectations.”
For the students who built the prototype and the current group charged with building four more sky cams with refined cloud-mapping and predictive technologies, the project offered an opportunity to make a positive difference in the world.
“I liked the impact that it could have,” said 2019 team member Monica Del Valle, who wrote algorithms for tracking the sun and predicting when clouds would obscure it. “If it worked, it would be pretty immediate and very significant. The environmental impact is what really drew me to the project.”
Now in its second phase of development, the cloud-tracking project challenges 2020 team members to improve on their predecessors’ work while designing mapping technology showing how much of a PV site would come under shadows, and use that data to predict the amount of power loss and for how long. Additionally, the team is to develop a “neural network” so the devices can talk to each other.
“Tracking clouds and shadows was not something everyone in the group was familiar with at first, but after doing a little research we all felt more comfortable in that respect,” said the current UCF team’s project manager, Tim O’Brien.
“They [OUC] want the maps integrated, so working across such a large region of land with multiple solar farms could pose some difficulty,” he said. “But we’re just going to work as hard as we can to get it done.”
Midway through the spring semester, UCF closed its campus and moved all classes on-line due to the coronavirus crisis. Still, O’Brien felt confident his group of computer science and mechanical engineering students would be able to meet its year-end deadline.
“We really want to be successful,” O’Brien said, adding he and his teammates feel a lot of pressure to meet the project’s goals. Assuming they do, Kramer expects to hand over the next phase of the project to UCF grad students in 2021, tasking them with improving user interface and data sharing capabilities. The final product could be patented and taken to a manufacturer for mass production, he said.
“I think it could become a revenue stream for OUC,” he said. “We haven’t seen anything on the market that compares with the cloud-tracking technology we’re creating with these UCF engineering students. Their work on this project could prove to be a game-changer in the future of solar energy.”