Field guide

Inspection · Aug 20, 2026 · 2 min read

Finding the faults you can't see: thermal drone inspection of solar arrays

Most solar faults are invisible until the output drops. A radiometric thermal flight finds them in an afternoon — and tells your O&M crew exactly which module to walk to.

A solar array fails quietly. A cracked cell, a corroded connector, a string that has dropped offline — none of it is visible from the access road, and as DroneDeploy's Conner Jones points out, most faults aren't noticed until production numbers sag. By then you've been paying for the loss for weeks.

Why walking it doesn't scale

Jones gives the scale: a 5 MW farm spans around 30–40 acres. A manual inspection means crews walking every row over days or weeks, often in desert heat and glare, around live electrical equipment and trip and fall hazards, with a handheld camera and a clipboard. The work is slow, the coverage is uneven, and the result is hard to compare with last year's.

What a thermal flight sees

A drone with a radiometric thermal camera — one that records an actual temperature for every pixel — flies the array from above in a fraction of the time. Electrical faults show up as heat: a single hot cell, a hot diode, a warm module in a cool string, or a whole string running hotter than its neighbours because it's disconnected or bypassed. A paired RGB camera captures the same module in visible light, which is how cracks, delamination, soiling, and physical damage get confirmed and how a hot spot gets separated from a bird dropping. Jones lists the same catalogue of anomalies: cell hot spots, cracks and delamination, soiling, corroded or disconnected wiring, degraded mounting hardware, and generally underperforming panels. His article also describes pairing the aerial pass with ground robots that run between rows for eye-level detail — complementary, and a sign of where automated inspection is heading.

From images to work orders

Thermal coverage is only useful if someone can act on it. The workflow Jones describes — and the one we deliver — organises every capture by location and date, lets the team tag each anomaly with a type and a priority, and hands maintenance a geolocated list: this row, this module, this issue. Over repeated inspections the same data shows patterns by zone or by panel type, which is how you catch a bad batch or a recurring connector problem rather than a series of isolated repairs. Jones recommends at least annual inspection and notes that with automation the cycle can tighten without labour costs rising in step.

What a facility manager receives

  • A thermal orthomosaic of the whole array and an RGB ortho to match.
  • An anomaly list with type, severity, and location down to the row and module, exportable to your CMMS.
  • Thermal/RGB image pairs for every finding so technicians see the problem before they leave the truck.
  • A dated record in your portal, so the next inspection is a comparison, not a fresh start.

Sources

  1. 1
    Automate the Inspection of Solar Farms

    Conner Jones · DroneDeploy blog · 1 May 2026

Written by the AerialWorx team. Figures attributed above come from the cited sources; everything else is our own explanation and field experience.