Ultrasonic Procedure Development and Qualification for Aerospace Manufacturing and MRO

What the procedure has to prove before anyone scans a part

A procedure is a claim about detection. It says that a defined instrument, transducer, wedge, couplant, calibration and scan pattern, executed by a person of a stated qualification level, will find a defined class of discontinuity in a defined material and geometry, and will call it against a defined acceptance criterion. Everything else in the document exists to make that claim checkable by someone who was not there. If the claim cannot be traced from a page number to a block to a recorded amplitude, the document is a work instruction, not a procedure.

That distinction matters more in aerospace than almost anywhere else because the acceptance criterion usually comes from outside your building. The OEM component maintenance manual, the engineering order, the prime's supplier quality requirements or an AMS material specification sets the reject level, and your procedure has to reference it exactly — document, revision and paragraph — rather than restating the numbers in your own words. Restated numbers go stale the moment the OEM issues a revision, and the auditor who finds a superseded limit copied into your procedure will treat every report written against it as suspect.

The second thing the procedure must prove is coverage. An acceptance criterion applied to a region the beam never reached is worse than no inspection, because it produces a signed record of conformity. Coverage is arithmetic — beam width at depth, index increment, overlap percentage, skip distance, the part of the section that is only reachable on second leg — and it belongs in the procedure with the numbers shown, not asserted.

Essential variables: the list that decides whether you have one procedure or six

Essential variables are the parameters whose change invalidates the qualification. Aerospace procedures fail here in a predictable way: the writer lists the variables but gives no permitted range, so every value in the document becomes a fixed value by default. If the procedure says '5 MHz, 0.25 in element' and nothing more, a technician who picks up a 3.5 MHz probe because the 5 MHz is out for calibration has just worked outside the procedure, and the report is uncoverable after the fact.

The variables that catch people are the ones that feel like consumables. Cable length and construction change the loading on the transducer and therefore the amplitude a DAC curve was built at. A change of couplant from a glycerin-based gel to a water-based one changes transmission loss on a rough surface. A second-source wedge in a different plastic changes the incident velocity and therefore the refracted angle. None of those get a purchase order that says 'engineering change', which is exactly why they need to be named in the procedure with a permitted range and a verification step.

Get the list right and you often end up with fewer documents rather than more. A well-scoped procedure with honest ranges can cover a part family; six copies of the same procedure with one number changed in each is a configuration-control liability, because the day the referencing OEM manual revises, someone has to find and update all six.

Why the alloy changes the geometry: the refracted-angle trap

A shear wedge is marked with the angle it produces in carbon steel. Aluminium has a lower shear velocity than steel, so the same wedge refracts at a different angle in an aluminium airframe part — a nominal 45-degree steel wedge lands nearer 43 degrees in aluminium. That is not a rounding error. It moves the beam index point, changes the skip distance and shifts the depth the beam crosses the expected crack plane, so a technique sheet whose trigonometry was worked in steel is wrong from the first scan.

The fix is procedural, not clever: require the actual refracted angle and beam index to be verified on a reference block of the same alloy and heat treatment as the part, record the measured angle, and use that measured value in the coverage arithmetic. Titanium and nickel alloys need the same treatment for the same reason, with the added problem that coarse or texture-banded microstructure produces attenuation and beam skew that no wedge marking anticipates.

Auditors ask about this obliquely. The question is usually 'what block did you calibrate on?', and the finding is written when the block turns out to be carbon steel while the parts are 7075-T6 or Ti-6Al-4V. Transfer correction between block and part, measured and recorded, is the evidence that closes it.

Damage mechanisms drive technique selection, not the other way round

In manufacturing, the discontinuities you are hunting are inherent to the process: inclusion stringers and segregation in billet, forging laps and bursts, porosity and lack of fusion in weldments, delamination and disbond in bonded and composite structure. These are mostly volumetric or planar-but-favourably-oriented, they are distributed rather than located, and the inspection is a full-volume sweep against a sensitivity class. The procedure's hardest job is proving the noise floor is low enough that a class-A reflector stands clear of grain scatter.

In MRO the population is entirely different: fatigue initiating at a fastener hole, a fillet radius or a previously repaired area; stress corrosion in a highly loaded fitting; corrosion thinning under a lap joint or in a hidden faying surface. These are located — you know where to look — but they are tight, oriented and shadowed. That argues for angle beam with a defined skip plan, bolt-hole techniques, or a decision that ultrasonics is the wrong method and eddy current owns the fastener-hole scope instead. A procedure that does not state why UT was selected over eddy current for a surface-breaking crack at a hole is answering the wrong question.

Corrosion thickness work has its own arithmetic trap. A pulse-echo thickness reading on a painted skin measures to the first back-wall it can resolve, and depending on gate and mode it can read through paint, read the paint as metal, or lock onto a corrosion product interface and report metal that is no longer load-bearing. Procedures should state whether coatings are removed, what mode is used, and how a suspect reading is confirmed.

The qualification demonstration: what a defensible one looks like

A demonstration exists to convert a technical opinion into evidence. The minimum credible package is: specimens that match the part in alloy, thickness, surface condition and geometry; discontinuities representative in type, size and orientation of what the part actually develops; the procedure at the revision being qualified; a technician certified at the level the procedure requires performing the scan; and recorded raw data with the instrument settings captured, not transcribed.

Where the demonstration usually goes thin is representativeness. A flat plate with side-drilled holes proves the instrument works. It does not prove that a 43-degree beam on a curved, shot-peened, single-side-accessible fitting will find a tight fatigue crack at the radius. Where the geometry is difficult, either build a mock-up with a real or seeded defect, or narrow the procedure's stated scope to what you can actually demonstrate. Narrowing the scope is a legitimate engineering answer; claiming coverage you never showed is not.

Keep the demonstration file as a controlled record with the procedure, not in an engineer's drive. When the procedure is revised, decide explicitly whether the change touched an essential variable, record that decision, and re-demonstrate if it did. The written 'no impact on qualification' rationale, signed by the Level III, is often the single most useful page in the folder during an audit.

NAS 410 personnel constraints belong inside the procedure

Aerospace procedures are read alongside the personnel file. NAS 410 sets the education, training and experience floors for each level in each method, requires annual near-vision and colour differentiation checks where the method demands them, and constrains who may interpret, who may accept and who may approve. A procedure that says only 'to be performed by qualified personnel' pushes that decision to the shop floor and creates an audit finding that is really a written-practice finding.

State it explicitly: the level required to perform the scan, the level required to interpret and evaluate indications, the level required to accept, and the Level III who approved the technique. Where a Level II is limited to a specific technique or part family by their certification, the procedure should be written so those limits can actually be honoured — which sometimes means splitting a procedure so that limited-scope technicians can work within it.

This is also where the three roles get conflated in general industry and must not be in aerospace. The technician performs and evaluates within their certification. The Level III approves the procedure and the technique qualification, and certifies personnel against the written practice. Nobody in the NDT chain is authorising the airworthiness release; that sits with the certificated repair station or the design authority, on the strength of the examination record you produced.

Findings that recur when aerospace UT procedures are audited

Four findings appear again and again. First, the procedure cites an OEM manual or specification without a revision, so there is no way to tell which acceptance limits were in force when the report was signed. Second, the calibration block is not traceable — no material certification, no heat treatment record, no evidence its surface condition matches production. Third, transfer correction between the block and the part is neither measured nor recorded, which means the stated sensitivity is an assumption. Fourth, the technique sheet's coverage arithmetic does not close: the index increment and beam width leave gaps that nobody has computed.

A fifth, quieter finding is revision drift between the procedure and the technique sheets derived from it. The procedure gets revised to correct an evaluation level; the twenty technique sheets in the shop are not reissued, and technicians are working to the old number. Configuration control over derived documents is a paperwork problem with a technical consequence, and it is the easiest one for an auditor to demonstrate because they only need two documents side by side.

The last recurring finding is scope creep by silence. A procedure written for one alloy and thickness range is used on a new programme because nothing in it explicitly forbids that. Adding a clear limitations clause — materials, thickness range, geometry, surface condition, temperature — costs a paragraph and prevents the finding entirely.

How an Atlantis procedure development engagement runs

We start from the parts, not from a template. That means the part families in scope, the damage mechanisms they actually present, the referencing OEM and specification documents at their current revision, the equipment you already own, and the certification profile of the people who will run the scan. A procedure written for equipment you do not have and people you have not certified is a document that guarantees a nonconformance later.

We then write the procedure and the technique sheets, specify the reference standards and transfer-correction method, design the qualification demonstration, witness or perform it, and record the evidence in a form an auditor can follow without a guide. Where the right answer is that ultrasonics is not the correct method for part of the scope, we say so and write the alternative into the method selection rationale — that rationale is itself an audit-useful document.

Atlantis supplies NDT technical authority: procedure development and qualification, written practice work, personnel certification within that practice, independent review of inspection data, and representation on technical questions when an auditor or customer challenges a record. Affordable, accessible and fully customisable — request a consultation or a scoped quote at info@atlantisndt.com and we will tell you honestly which parts of your existing procedure set are already sound.

Request a consultation