Solutions — 09
Thermography & Solar Inspection
A defective module looks exactly like a working one. Not in the thermal image.
A thermal camera shows what the eye does not: the module running hotter than its neighbours because a cell has stopped contributing. The pipe above which the snow melts first. The point in a façade where heat escapes. The drone flies the area, and the analysis assigns every finding to its module, section or component.
Detection is done by a model that knows the typical fault patterns: single hot cells, whole bypass-diode areas, shaded strings, soiling. It separates them from what merely looks like a fault — reflections, bird droppings, a passing cloud shadow.
Why an array loses yield without anyone noticing
Photovoltaics does not report a failure, it simply delivers less. An inverter shows the sum of a string; that three modules in that string are underperforming disappears in it. Only when a larger section fails does it become visible in the yield curve — and by then months have passed.
On a solar farm with several thousand modules, searching by hand is not an option. A technician with a handheld meter needs days, necessarily works by sampling, and finds the fault only where they look.
The same applies to heating networks. A leak in a buried pipe shows itself long before the pressure drops — as a heat trace at the surface. It is visible only in the thermal image and only at the right time.
How the flight and the analysis are done
Thermography is a question of conditions, not of cameras. For photovoltaics that means sufficient irradiance, little cloud, low wind, and an array under load. Fly under the wrong conditions and you produce imagery on which every finding stays questionable. We plan the appointment around that — not around when a drone happens to be free.
The flight follows a fixed grid at constant distance, so every module is captured at comparable resolution. Thermal and visible imagery are recorded in parallel, so every finding has its visible photograph beside it.
The analysis locates each finding: array, table, string, module position. A report saying "a hot module somewhere in the northern field" saves nobody any work.
What the report contains
Per finding: position, fault pattern, temperature difference to the surroundings, thermal image and photograph side by side, plus an assessment of whether this is an immediate case or something for the next maintenance round.
With repeated surveys the comparison to the previous year comes on top — the same analysis as in change detection, applied to thermal imagery.
What it delivers
For photovoltaics the survey pays for itself through recovered yield: one per cent of performance loss on a megawatt array is a four-figure sum over a year. On top of that come the records operators have to keep for insurers and financing anyway.
For heating networks it is the leak found early, before it becomes a winter excavation. For buildings, the proven weak point instead of the suspected one.
→
Benefits at a glance
- Every finding located to the module, not just "in the northern field"
- Thermal image and photograph side by side — every finding stays checkable
- Automatic separation of real fault patterns from reflections and soiling
- Appointment planned around measuring conditions, not around the calendar
→