Montline.

AI Applications · Energy and Facility Inspection

Surface Defect Screening on Wind Turbine Blades

A blade is expensive to reach and more expensive to replace. Screening from the air turns a rope access job into an imaging job — but what comes out is a ranked list of things worth looking at closely, not a verdict on each one.

3 of 10Method note

Overview

  • Imaging geometry decides what is detectable: sun angle relative to the blade, rotor position, and whether the background is sky, sea or terrain.
  • Surface condition varies hugely between units — soiling, repairs, leading-edge erosion and ice all look different, and the assessment has to be per asset.
  • Comparing against the previous survey separates what is new from what has been there for years; absolute condition is much harder to judge from one pass.
  • Severity classification and any decision to stop a turbine remain a human judgement made by qualified inspectors.
Detection output is a screening signal. Severity grading remains an inspection decision made by qualified personnel.

Why blades are surveyed from the air

A wind turbine blade is one of the most expensive single components in an operating asset, and reaching the outer two thirds of it traditionally means rope access or a lift — both of which cost money, take time and put people into a genuinely hazardous position. An aircraft does not remove the need for judgement; it removes the cost of getting close enough to form one.

What this changes economically is the threshold for looking. When each inspection requires a rope team, inspections are scheduled rarely and defects are found late. When imaging is cheap and repeatable, the asset can be surveyed on a predictable cadence and changes become visible between surveys rather than appearing all at once.

What actually determines whether something is detectable

The limiting factor is rarely the model. It is imaging geometry. Sun angle relative to the blade surface changes how much a surface feature casts a shadow or reflects; rotor position determines which part of the blade is presented to the camera; and the background — sky, sea, terrain or another turbine — changes the contrast available to work with. A defect that is obvious in one frame can be invisible in the next, shot seconds later from a slightly different angle.

This is why the capture plan matters more than the detection threshold. Specifying approach direction, stand-off distance, lighting window and overlap is what makes two surveys comparable. Without a repeatable plan, each data set is a one-off and none of them can be differenced against each other.

Surface condition adds another layer. Soiling, insect residue, repair patches, leading-edge erosion and ice each alter the appearance of an otherwise sound surface. The practical consequence is that a model trained on clean blades behaves differently on a coastal site with salt film, and differently again in winter.

Turning images into something an inspector can act on

The output is not a diagnosis. It is a ranked list: locations, cropped imagery, and enough context for a qualified inspector to decide whether the item is real, how it compares with the previous survey, and whether it warrants anything further. Anything that presents itself as a final verdict has skipped a step that matters.

Comparison against the previous survey is where most of the practical value sits. Separating what is new from what has been there for years is far easier and far more defensible than trying to grade absolute condition from a single pass. The link between imagery and the asset register is what makes that comparison possible at scale — each candidate has to belong to a known blade on a known unit.

Candidates also need to reach the system that closes them out. That loop is described in defect review and work orders: an item that stays a floating image in a folder will be looked at once and then forgotten.

Limits we state up front

Severity grading and any decision to stop a turbine are inspection decisions made by qualified personnel, not by a screening pass. Internal structure — delamination, bond line condition, lightning protection continuity — is not something surface imagery can see.

Weather is a hard constraint. Blade inspections are typically flown below the wind limits applicable to the aircraft, which are lower than the limits at which the turbine itself still generates. This is a planning issue rather than a technical one, but it decides how much of the year is actually flyable.

Related pages

More in this area

All topics in Energy and Facility Inspection → Discuss this requirement with us →

Tell us what the system needs to do

We will tell you what is achievable, what it takes, and what we would need from you.

Discuss Your Project