At a glance
- Challenge: a utility-scale array showing a small, unexplained production shortfall the monitoring system could not localize.
- Method: every-module radiometric thermal scan flown with the Matrice 300 RTK and H20T.
- Outcome: the loss traced to a specific string, with a severity-ranked defect list handed to the O&M crew for a targeted repair.
This is a representative walkthrough of UAV Imaging's solar thermal inspection methodology, drawn from how these jobs are run in the field rather than a single named client. It illustrates what the deliverable looks like and why aerial thermal finds what string-level monitoring alone misses.
The problem: a production dip with no address
Utility-scale arrays are monitored at the inverter or combiner level. That is enough to tell an operator that production is a percent or two below model — but on an array of tens of thousands of modules, “down 2%” is a symptom without a location. Sending a crew to walk every row with a handheld meter is slow and, in the meantime, the array keeps underproducing. The question the operator actually needs answered is: which modules, and how bad?
The method: every-module radiometric thermal
A defective module runs hotter than its healthy neighbours because energy that should be leaving as electricity is dissipated as heat. Radiometric thermal imaging makes that temperature difference measurable from the air. The flight:
- Aircraft and payload: DJI Matrice 300 RTK carrying the H20T radiometric thermal sensor.
- Conditions: flown in high irradiance (clear sky, sun high) so healthy and faulty modules separate cleanly on temperature; overcast washes the signal out.
- Coverage: automated grid mission imaging every module across the array at a consistent altitude and angle for comparable readings.
- Analysis: thermal frames reviewed for the signatures that matter — single hot cells, hot-spot patterns, whole-module heating, and the diagonal or block patterns that point to a string- or combiner-level fault rather than a lone module.
What the scan found
The array-wide production dip resolved into a pattern: a run of modules along one string reading consistently warmer than the surrounding array — the thermal signature of a string carrying a fault rather than a scatter of unrelated single-module defects. Instead of “the array is down 2%,” the operator now had a specific string, a location on the site map, and a severity ranking for each affected module.
The deliverable
- Geo-referenced thermal orthomosaic of the array
- Defect list keyed to module and string location, each classified by anomaly type and severity
- Representative thermal and visual image pairs for the flagged modules
- A prioritized action list the O&M team can dispatch against directly
Why aerial thermal earns its place
Inverter data tells you that something is wrong; an every-module thermal scan tells you where and how bad. For a utility-scale operator, closing that gap quickly is the difference between weeks of quiet underproduction and a same-week targeted repair. One flight covers an array that would take days to walk, and the severity-ranked output means the crew fixes the modules that actually move production first.