Blade inspection is a resolution problem, not a thermal problem. Structural damage is found with a long-focal-length visible camera at 30–60 m standoff; thermal earns its place on subsurface delamination, lightning-strike damage and nacelle electrical faults. A payload that does both removes an entire second flight.
Key takeaways
- Thermal and visual find different defects: thermal reveals subsurface delamination and lightning damage, visual finds erosion and surface cracks.
- Ground sample distance is the whole game — the smallest defect you must catch sets standoff distance and therefore flight time per turbine.
- The thermal pass has a window: shortly after the blade has been loaded or solar-heated, when subsurface defects still differ from sound laminate.
- Savings come from avoided rope access and shorter downtime per turbine, not from the imagery.
On this page
Which sensor finds which defect
Operators routinely buy a thermal payload for blade work and then discover that 80 % of the findings came from the visible channel. Both matter, but for different failure modes.
| Defect | Primary sensor | Thermal signature | Typical standoff |
|---|---|---|---|
| Leading-edge erosion | Visible zoom | None | 25–40 m |
| Surface crack, gelcoat | Visible zoom | None until through-thickness | 20–35 m |
| Subsurface delamination | Thermal | Warm patch, slow cooling after solar load | 30–50 m |
| Lightning strike damage | Both | Local hot or cold anomaly at receptor | 25–40 m |
| Water ingress in blade | Thermal | Cool region, high thermal mass | 30–50 m |
| Trailing-edge bond failure | Visible zoom | Occasional thermal line | 20–35 m |
| Nacelle bearing / generator fault | Thermal (radiometric) | Localised temperature rise | 40–80 m |
| Transformer and switchgear | Thermal (radiometric) | Hot connection, phase imbalance | 40–80 m |
Ground sample distance is the whole game
For blade surface defects the specification that decides success is GSD, not megapixels. A defect must span several pixels to be classifiable rather than merely detectable. As a working rule, target 0.5–1.0 mm/pixel for crack classification and 2–3 mm/pixel for erosion mapping.
GSD = (sensor pixel pitch × distance) ÷ focal length. A 1/2.8″ sensor with 2.9 µm pixels behind a 47 mm lens at 30 m gives roughly 1.85 mm/pixel — adequate for erosion, marginal for hairline cracks. Push to a longer focal length or close the standoff, and be honest about which one your airspace and wind conditions allow. The trade is unpacked in hybrid zoom on UAV cameras.
Thermal GSD is coarser by an order of magnitude and that is fine — thermal is looking for regions, not edges.
Timing the thermal pass
Subsurface thermal contrast on composite blades comes from differential heating and cooling, not from the defect generating heat. That makes timing decisive.
- Best window: 30–90 minutes after peak solar load begins to fall, when sound laminate and delaminated regions cool at different rates.
- Worst window: uniform overcast with no thermal loading history. Contrast approaches zero and the flight is wasted.
- Wind cools blades unevenly and rapidly; above roughly 8–10 m/s expect thermal contrast to wash out along with your image stability.
- A parked, feathered rotor is required for both channels. Inspecting a turning rotor is a different, much harder problem.
Recommended UAVThermal payloads
LX-6U — long-reach zoom pod. When the standoff has to stay large for wake and clearance reasons, this is the payload that still returns classifiable blade imagery.
OP-125A — multi-sensor pod with thermal, long zoom and laser ranging. One flight covers blade surface, blade subsurface and nacelle electrical.
AX-40T — stabilized gimbal for integrators building a dedicated blade-inspection aircraft around their own camera choice.
Related reading
Technology: thermal infrared optics and gimbal stabilization technology.
Field practice: wind turbine blade inspection with zoom and thermal, thermal lens focal length and what 0.01° pointing accuracy means at 1,000 m.
Adjacent missions: solar farm thermal inspection and power line inspection.
Cost per turbine and where the savings come from
Rope-access blade inspection typically consumes most of a day per turbine and requires the machine stopped throughout. Drone inspection of the same three blades runs 25–45 minutes including setup, so one crew covers six to ten turbines a day rather than one.
The saving that dominates the business case is not labour — it is lost generation. Every hour a turbine is parked for inspection is production that never happens. Compressing a full-day stop to under an hour, across a site of forty machines, is usually worth more than the entire inspection contract.
The counter-cost is data handling. A forty-turbine site produces thousands of high-resolution frames per campaign and somebody has to review them. Budget for the analysis workflow, not just the flying, or imagery accumulates unreviewed and the programme quietly stops.
Where this goes wrong
Blade inspection is unforgiving of small planning errors.
- Flying thermal on an overcast day. Without a solar loading cycle there is no subsurface contrast. The flight looks successful and finds nothing, which is worse than not flying.
- Standoff creep. Operators drift outward as wind picks up. GSD degrades linearly and hairline cracks silently drop below detectability — the survey still produces images, just not classifiable ones.
- Inconsistent blade positioning. If blade angles differ between campaigns, defect tracking over time becomes guesswork. Lock the rotor to the same position every visit.
- Ignoring the nacelle. Teams focused on blades routinely skip a two-minute radiometric pass over the nacelle and transformer, which is where the highest-value electrical findings usually are.
FAQ
Do I need thermal for blade inspection at all?
For surface damage, no — a long-focal-length visible camera does that better. Thermal adds subsurface delamination, water ingress and lightning-receptor damage that visible inspection misses entirely, plus nacelle electrical faults on the same flight. Most operators find the combination pays for itself; thermal alone does not.
What standoff distance should we fly?
25–40 m for visible defect classification and 30–50 m for thermal. Closer improves GSD but increases collision risk and rotor-wash interaction with the tower; further degrades resolution faster than most people expect because GSD scales linearly with distance.
Can we inspect while the turbine is running?
Not meaningfully. Blade tip speeds of 70–90 m/s defeat both image stability and any usable exposure time. Standard practice is a parked, feathered, locked rotor with the blades in a known position.