AI-assisted Computed Tomography inspection enters production for GE9X turbine blades
After four years of development, our AI-assisted CT inspection system is now certified for production use on GE9X high-pressure turbine blades. Dr. Moreau explains what it catches that humans cannot.
Turbine blade inspection has always been a bottleneck. A single GE9X blade takes a trained inspector 45 minutes to evaluate under computed tomography. At our volume, that translates to roughly 12,000 inspection-hours a year — each one requiring sustained expert attention on millimetre-scale anomalies.
Our new system, co-developed with two European universities, cuts that to 4 minutes per blade. More importantly, it reduces the false-negative rate on hairline internal cracks by 63% compared to human inspection — not because the AI is smarter than our inspectors, but because it does not tire at hour seven.
The system does not replace our inspectors. It flags candidate anomalies and ranks them by severity. Human experts make every release-to-service decision. But it frees our best people to spend their time on the genuinely ambiguous cases — the ones where judgment, not pattern recognition, is the point.
Regulatory approval took three years, two major audits, and one painful redesign. Worth every hour. By Q4 2026 we expect to extend the system to GEnx and LEAP blades.