SNT-TC-1A and ASNT Level III Review of Blade Data

What the standard covers

The practice sets out how a company qualifies and certifies its NDT personnel across levels, and the Level III grade covers the interpretation and the sign-off of results. On a blade programme that is the grade behind the AI agreement figure and behind the review of C-scan data that turns a detection rate into a validated number.

What it demands

• Personnel qualified through a written practice aligned with SNT-TC-1A
• A Level III grade for data interpretation and sign-off
• Validation of the detection figure against a Level III re-review rather than a baseline
• A record tying each dataset to its qualification level

How it lands on a blade scan

The 99.2 % figure is what a Level III review makes defensible. Manual UT detection runs at 87 % at Level II, and the automated figure of 99.2 % is validated against an ASNT Level III re-review of all C-scan data rather than against the manual baseline alone. The AI layer is held to the same standard, agreeing with ASNT Level III review at 99.2 % on delamination, porosity and disbond classification, and trained on over 10,000 blade inspections across 10+ countries. So the qualification framework and the software figure point at the same reference, which is a qualified human reviewing the data.

How to apply it to a blade

Run these checks in order. By Step 5 the scan plan has the threshold, the geometry and the record it needs.

1. Fix the acceptance threshold. Start from the threshold the zone carries, not from the method you happen to own. IEC 61400-5 Section 11 asks for volumetric detection of indications 6 mm and larger in the laminate plane, and DNV-ST-0126 tightens the leading edge to 3 mm, so the threshold is set before the probe is chosen.
2. Read the geometry off the drawing. Measure the thickness and the curvature the probe has to travel. A spar cap up to 80 mm of unidirectional laminate and a root transition of changing radius are two different scan problems, and the radius decides how the array is coupled.
3. Choose the modality pair. Pick modules by defect geometry, not by habit. PAUT images planar separations that lie parallel to the plies, TOFD catches volumetric clusters whatever their orientation, and running both on a 256-channel system puts the two mechanisms in one pass.
4. Automate the scan path. Put the probe on an encoder-driven carriage and let the path, not the technician, set coverage. A 600 mm spar cap falls to one longitudinal pass at 500 mm/s on a linear array, where a single-element probe needs 20 parallel passes and runs about five times longer.
5. Lock the evidence chain. Write the coverage map, the daily calibration and the indication log to the blade serial while the scan is running. It is the difference between a 3-day type-certification audit and a 3-hour document review, and it is what ISO 9712 qualification is there to sign.

These are the working numbers from our own machine and campaign records, restated rather than re-pitched.