Leading Edge Erosion: Drone vs UT

Leading edge erosion drives more in-service blade repairs than any other defect, and it’s accelerating as tip speeds climb. We’re not debating whether to inspect. Which method delivers real depth numbers, not just pictures?

TLDR Erosion’s there, the drone confirms it. Coating delamination? That’s where thermography shines. But when you need millimetre-accurate remaining laminate thickness (the number that makes the repair-vs-reprofile-vs-replace call), only automated ultrasonic gauging delivers.

The Erosion Problem in Numbers

Raindrops at 80-90 m/s hit like an industrial sandblaster. That’s not hyperbole. A single storm season strips away 0.5-1.0 mm of leading-edge coating, and once that coating’s gone, the glass-fibre laminate underneath erodes at 2-5 mm each year. AEP loss from leading-edge roughness runs 2-5% per affected blade. For a 5 MW offshore turbine bringing in €1.5M annually, that’s €30,000-75,000 in lost revenue (per turbine, every year).

Method 1: Drone Visual Inspection

A drone with a good camera can sweep a whole wind farm in a day. The images feed into AI that flags every visible defect. If you just want to know that erosion exists, drones are brilliant: cheap, fast, and clear on surface cracks and impact marks. But a photo won’t tell you if it’s 0.5 mm or 3 mm of laminate lost, and that’s the number that separates a cosmetic touch-up from a structural grind-and-fill. Cameras don’t give it.

Method 2: Active Thermography

Active thermography heats the blade and watches how it cools. Subsurface air pockets (delams, disbonds) change the thermal decay pattern, so it’s very good at finding hidden coating separation that a drone would miss. But it can’t give you erosion depth with any precision. A 1.5 mm scar and a 4 mm scar look the same on the thermal feed. For structural calls, that’s a problem.

Method 3: Automated Ultrasonic Thickness Gauging

Now, automated UT. A 5 MHz dual-element transducer sits on a crawler, scanning the leading edge and mapping remaining wall thickness to ±0.1 mm. That C-scan overlays the blade’s 3D model, pinpointing every spot below threshold, by how much. You get quantitative, repeatable data that tells you exactly what to do next (no guesswork). Unlike the other two, it doesn’t skip anything that matters for the erosion repair decision. Delams, porosity, thickness loss: it’s all there.

The Cost of Getting It Wrong

If you underestimate erosion depth, your cosmetic fix fails in six months (so you’re back, paying for access and repair again). Overestimating means grinding out healthy laminate and adding dead weight. One North Sea contractor found that 30% of blades slated for structural repair (based purely on drone images) actually had enough laminate left when they followed up with UT. Across a 50-turbine farm, those unnecessary structural jobs cost about €180,000.

Frequently Asked Questions

Which method gives a usable erosion depth number?

Only thickness gauging does. A 5 MHz dual-element transducer on a crawler maps remaining wall thickness to ±0.1 mm, so the C-scan shows every spot below threshold and by how much.

What can a drone sweep actually tell you?

A drone reads surface condition and flags visible defects fast, but a photo cannot separate a 0.5 mm scar from a 3 mm one, which is the number the repair decision turns on.

Where does thermography fit?

It finds hidden coating separation and disbonds that a camera misses, because subsurface air pockets change the thermal decay pattern, but it does not size erosion depth.

What does erosion cost a 5 MW offshore turbine?

Roughness-driven AEP loss runs 2-5 % per affected blade. On a 5 MW machine earning about €1.5M a year that lands at €30,000-75,000, every year.

How much can misread erosion cost a farm?

One North Sea contractor put 30 % of drone-flagged blades into structural repair that UT showed did not need it, and across a 50-turbine farm that over-scope came to about €180,000.

Bottom line: Drones for screening: fast, cheap, surface-only. Thermography for hidden disbonds. But the moment you are about to grind into structural laminate, you need an ultrasonic thickness number. That scan costs a tiny slice of what a wrong repair will eat.