No Ply Goes Unchecked: How WorkWatch Enhances Defect Detection in Composite Layup

A quality inspector walks the length of a wind blade mold, checklist in hand, confirming two things about each fiberglass ply as it goes down. Is this the right ply, in the right spot on the layup plan? And is it actually sitting where it's supposed to be, within a tolerance measured in millimeters?

Both checks matter. Neither is hard to understand. But no QA team, however large, can watch every ply on every blade, on every shift, without ever blinking. Composite layup doesn't pause for an inspection. By the time a spot-check catches a problem, three more plies may already be stacked on top of it.

Coverage is the real constraint, one that only gets tighter as delivery timelines shrink. That's the gap WorkWatch is built to close.

The high cost of composite manufacturing defects

Composite manufacturing quality control used to be treated as a downstream activity. Build first, inspect after, catch problems in a final audit. A review of AI and robotics adoption across the offshore wind sector points to a shift: pairing advanced sensing with AI so quality verification happens as part of the production process itself, not as a separate step after the fact.

As cycle times keep shrinking to meet delivery commitments, and composite architectures keep getting more complex, early defect detection is becoming increasingly essential. When a defect does slip through, the cost isn't a quick fix. Industry data on blade failure puts the cost of an up-tower repair in the tens of thousands of dollars, a full blade replacement in the hundreds of thousands, and a catastrophic failure in the millions. It's the kind of bill that makes catching a problem in production worth the investment many times over.

Manual, sampling-based inspection was simply built for a slower version of this industry. No inspector can be in twelve places on a mold at once, and a checklist run once a shift only ever samples a fraction of what actually gets laid.

RFID tagging and automated fiber placement fall short for ply verification

Manufacturers have tried other ways to close this gap before landing on computer vision, and it's worth understanding why they fall short.

RFID tagging looks promising on paper: tag each ply, scan it into place, get a digital record automatically. In practice, it runs into scale problems fast. A single blade can carry 400 or more individual plies. Many of those plies aren't pre-cut, instead cut directly from rolls on the mold, making pre-tagging impossible. Even where tagging is feasible, the system only confirms a tagged ply is present, not where it landed or whether it's correctly oriented. And placement is exactly where the risk lives.

Automated fiber placement (AFP) is the aerospace and automotive standard for high-precision layup, laying fiber tows with sub-millimeter accuracy using robotic heads. It hasn't made the jump to wind blades, for a simple reason: AFP's deposition rate isn't built for surfaces the size of a turbine blade, which need thousands of square meters of layup covered. The gantry robots large enough to do it would be enormously expensive, and the ROI doesn't pencil out while skilled labor remains competitively priced.

WorkWatch inspects every ply in real time

WorkWatch provides the best of both worlds. There’s no new tagging step, no giant capital outlay, and placement is verified, not just presence. 

WorkWatch uses cameras positioned along the blade mold to run two checks continuously: does this ply's label match what the plan calls for, and is it physically placed within the tolerance lines laser-projected onto the mold? Both checks can happen as soon as the ply goes down, instead of waiting for a periodic walk-through. A mismatched label or an out-of-tolerance placement gets flagged in real time before more plies bury the problem.

If the case is ambiguous — a crane blocking the camera's view, an odd angle, a label the system can't quite resolve — WorkWatch defers to human expertise. It classifies the ply as unverified and flags it for review. An operator picks up a tablet, pulls up the ply in question and takes a photo to confirm what the camera couldn't. Nothing slips by.

Get full coverage at scale with WorkWatch

Put together, WorkWatch gives producers two things a sampling-based QA process can't. First, assurance that no ply has gone unchecked. Every layer gets the same scrutiny, automatically documented, instead of whatever fraction a spot-check happens to catch. Second, it frees up QA time for the work that actually needs a professional's eye: the edge cases, the close calls, the ply that's genuinely worth a second look.

The end result? An evolution from yesterday's sampling and hoping to the full-coverage solution today's market demands.

See how WorkWatch brings continuous, real-time visibility to your manufacturing floors. Sign up for a WorkWatch demo.

‍