How Fracsun Used App Orchid to Predict Solar Farm Soiling Loss With 95% Accuracy

“We went from idea to proof of concept in three months — with one developer.” — Catlin Mattheis, CEO, Fracsun

When you think about solar farms, you probably picture clean panels glinting in the sun, silently turning light into power. But here’s something most people don’t see — dust, pollen, and grime quietly cutting into that efficiency day after day. That invisible buildup, known as soiling, can reduce power output by as much as 10–15%, costing millions across large portfolios.

The Challenge

Fracsun knows this better than anyone. They’ve spent years:

• Measuring soiling across 21GW of solar assets in 30 countries
• Building one of the world’s most trusted databases of soiling loss data

Still, one question kept coming up: What if you could predict soiling — anywhere — without installing a single sensor? That question sparked a breakthrough.

Building a Model That Learns From the Environment

Partnering with App Orchid, Fracsun set out to build a machine learning model capable of:

• Predicting soiling loss anywhere in the U.S.
• Running entirely on data — no physical sensors required
• Delivering results with unprecedented accuracy

“App Orchid moved incredibly fast,” says Catlin. “From concept to working proof of concept took about three months — and it only required one developer on our side.”

Using Fracsun’s massive database as a foundation, App Orchid’s AI combined:

• Weather and particulate matter data
• Environmental context from its Knowledge Graph
• Advanced machine learning to simulate how soiling builds up over time

Proving 95% Accuracy — Without Sensors

When the results came in, the team was floored.

Fracsun compared the model’s predictions against actual field measurements from its sensors — and they matched with roughly 95% accuracy. That level of precision gave the team the confidence to move from prototype to production in under six months.

“There were moments where we’d overlay the real-world data and the model’s output,” says Catlin, “and they were almost identical.” 

Scaling Smarter Decisions for Every Solar Operator

For large-scale solar farms, Fracsun’s hardware has always been a game-changer — but smaller operators often couldn’t justify the cost of installing physical sensors.

Now, with this new AI-powered model, every operator can access the same level of insight, regardless of project size.

• Project developers can eliminate uncertainty in production forecasts.
• O&M teams can clean panels only when it makes financial sense, cutting costs by up to 50%.
• Origination teams can identify ideal locations for development with minimal soiling impact.

“Essentially,” says Mattheis, “we’ve made the kind of predictive intelligence once reserved for big projects accessible to everyone.”

The Takeaway

Fracsun and App Orchid didn’t just build a model — they built a smarter way to power solar innovation. By combining years of field data with advanced AI, they proved that accuracy, efficiency, and scalability don’t have to be competing priorities.

Sometimes, all it takes is the right partnership — and the right question: What if you could predict ___________?

‍