NDT Technical Authority When the Auditor Arrives: Bridges and Infrastructure

What an audit of an NDT programme is actually testing

Owners, certifying bodies and client quality organisations do not audit welds. They audit the evidence that welds were examined properly. The question at every step is the same: can you demonstrate, from records that existed at the time, that this examination was performed by a person qualified for it, to a procedure that had been approved, with equipment whose calibration was current, and that whatever was found was dispositioned by someone with the authority to disposition it.

That reframing explains why technically excellent inspection organisations still fail audits. The examinations were competent; the chain that proves it has a break in it. A report with no revision reference to the drawing it examined. A procedure with an approval signature but no evidence that anything was technically reviewed. A calibration block with no identification and no traceability. None of those means a defect was missed, and all of them are findings.

On bridge and infrastructure work the stakes are structural and political at once. Owner agencies hold records for the life of the asset, and a finding raised during fabrication propagates into every subsequent review of that structure. The cheapest moment to fix a broken chain is before anyone outside the organisation has looked at it.

The written practice is the auditable document, not the recommended practice it cites

ASNT SNT-TC-1A is a recommended practice. It provides guidelines an employer uses to establish its own written practice, which is where the actual, binding requirements live: the methods and levels used, the training hours required in each method, the experience required, the examination content and grading, the certifying authority, the recertification interval and the requirements for vision examination. When an auditor asks a question about personnel qualification, the correct reference is your written practice — and the finding, when it comes, is a difference between your written practice and your records.

ANSI/ASNT CP-189 behaves differently. It is a standard rather than a recommended practice, it sets minimum requirements that the employer cannot relax, and it imposes requirements on the Level III itself. Some owner-agency specifications invoke CP-189 explicitly, and some invoke ISO 9712, which is a third-party certification scheme rather than an employer-based one. A programme that certifies technicians under an employer written practice against a specification that requires independent ISO 9712 certification has a non-conformance that no amount of training documentation will resolve.

The practical consequence is that the written practice has to be written against the specifications you actually work to, reviewed when those specifications change, and then followed exactly. Where the workforce is being built up rather than merely documented, the training and experience side of the practice has to be planned as deliberately as the paperwork — which is where structured NDT training and certification to SNT-TC-1A and ISO 9712 belongs in the programme, rather than being reconstructed from timesheets after an auditor asks.

AWS D1.5 and fracture-critical members: what changes

Bridge fabrication is governed by AWS D1.5, not AWS D1.1, and the two are not interchangeable. D1.5 carries its own ultrasonic testing provisions with their own scanning patterns and evaluation approach, its own radiographic requirements, and its own acceptance criteria. Applying a D1.1 ultrasonic procedure to a bridge weld is a straightforward non-conformance that appears in audits more often than it should, usually because a fabricator's procedure library was built for building steel and extended to bridge work without technical review.

Fracture-critical members bring the fracture control plan, and the plan is where the record burden multiplies. Welders and welding operators must be qualified specifically for fracture-critical work. Procedures require the Engineer's approval. Filler metals must be controlled and supported by impact toughness data. Heat input is restricted, with preheat and interpass temperatures recorded rather than assumed. Examination acceptance tightens. Repair procedures are constrained, and repeated repair of the same location is itself a reportable condition.

Every one of those obligations produces a document that an auditor can ask for by name, and owner agencies generally require those documents to be retained for the life of the structure. That retention requirement is what makes fracture-critical records different in kind from ordinary shop records: they do not expire with the contract, and a gap discovered years later still belongs to the fabricator who created it.

What the auditor samples, and in what order

The pattern is consistent. The auditor picks a member from the drawing or the NDE map, confirms the drawing revision, and asks for the traveler. From the traveler they take the heat number and ask to reconcile it to a mill test report. They take the weld and ask for the WPS in use, then the PQR that supports it, then the qualification record of the welder who deposited it, then the continuity record proving that qualification was maintained.

Then they turn to the examination. Which NDT procedure applied, who approved it and on what technical basis, which technician performed it, what level that technician holds, what training and experience supports that level, when their vision examination was last performed, what instrument was used and when its linearity and decibel accuracy were last verified, and what reference block was used and how it is traceable.

Finally they close the loop on the result. If the report records an indication, they want the disposition: the evaluation, the repair procedure if repaired, the re-examination that confirmed the repair, and the acceptance. An unclosed loop is the single most damaging finding available, because it suggests the programme records what it finds but does not track what it does about it. Where the volume of records makes that traceability fragile, holding drawings, travelers, personnel records and examination data in one linked system rather than in parallel filing removes most of the failure modes without changing a single procedure.

The findings that recur on bridge and infrastructure work

Expired vision examinations, first and most damaging, because the exposure is a population of reports rather than a single document. Welder continuity records not maintained, so a qualification that appears current on its face was actually void for part of the contract. Procedures approved by a Level III with no evidence that a technical review took place — a signature on a cover sheet is not a review, and auditors have learned to ask what was reviewed. Reference blocks and calibration standards without identification or traceability.

Then the reporting failures. Reports referencing superseded drawing revisions. Reports signed by personnel not certified to the level the procedure requires, usually because a Level II was unavailable and a Level I completed the paperwork. Indications recorded with no traceable disposition. Radiographs whose density falls outside the specified single-film window, or whose image quality indicator sensitivity does not demonstrate the required sensitivity — reviewers do re-measure these, and a densitometer settles the argument in seconds.

The last category is the most preventable: interpretation drift. Over a long contract, acceptance decisions gradually loosen without anyone deciding to loosen them, because each individual call was defensible and nobody re-examined the aggregate. Periodic independent review of inspection data — a genuine re-read of a sample by someone outside the production pressure — is the only reliable detection method, and it produces a record that itself demonstrates programme control to an auditor.

What the Level III signs, and what it does not

The Level III's authority is technical and bounded. Within the written practice, the Level III establishes and approves NDT procedures, provides the technical basis for those procedures, prepares and administers qualification examinations, verifies training and experience, endorses personnel qualification, performs technical review of examination data, and represents the organisation on technical questions during an audit. Those are real signatures with real consequences, and they are the ones Atlantis provides.

What sits outside that authority matters just as much, and stating it plainly is itself an audit defence. Atlantis is not the API inspector of record for any asset, and does not act as a PSM auditor. It does not sell API 510, 570 or 653 inspector certification training — those certifications are administered by API and their training market is separate from what Atlantis provides. On bridge work, the Level III does not accept a weld in place of the Engineer of Record, does not sign the fabricator's certificate of compliance for the structure, and does not provide anything requiring a professional engineer's seal.

The boundary is also an employer-certification boundary. Under an employer-based framework, an outside Level III agency administers and recommends; the employer certifies. Where the written practice names the employer's certifying authority and an external signature appears in that place instead, the finding writes itself. Getting that architecture right at the outset is a routine part of setting up ASNT Level III consulting support, and it is far easier to establish than to unpick.

Gap review before the audit versus remediation after a finding

A gap review is the same exercise as an audit, run by someone on your side. It samples the same way — a member, a weld, a report, backwards through the chain — and it produces the same list. The difference is entirely in what happens next. A missing training record can still be reconstructed from the instructor's roster while the instructor still works there. A procedure with a thin technical basis can be revised and re-approved before it has been applied to another two hundred joints. An interpretation drift can be corrected quietly with a re-read of the affected sample.

Remediation after a finding is a different discipline and a more expensive one. It demands a root cause, not a fix. It demands an extent of condition: how many other reports, members, welders or procedures does this same failure touch, and how do you know. It demands corrective action with objective evidence that the action worked, and it usually demands re-examination of some population of work — sometimes with the structure fabricated, coated, shipped or erected.

Extent of condition is where the cost sits. One expired vision examination is a trivial administrative lapse until you calculate that the technician signed four hundred reports in the lapsed window, and then decide, with the owner watching, which of those four hundred welds get re-examined. Organisations that have been through that once tend to schedule gap reviews thereafter.

Building an evidence chain that holds without heroics

The programmes that pass audits comfortably are not the ones with the thickest manuals. They are the ones where each record is created at the moment the activity happens, by the person who performed it, referencing the identifiers that let it be reconnected later — drawing revision, joint number, heat number, procedure revision, technician identifier, instrument serial. Reconstructing that after the fact is always visible to an experienced auditor, and it converts a documentation gap into a credibility problem.

Three habits do most of the work. Keep a live personnel matrix showing every technician, method, level, certification date, recertification due date and vision examination expiry, and treat an approaching expiry as a production issue rather than an HR one. Keep procedures under revision control with the technical basis attached to the approval, so that the question "what was reviewed" has an answer. And close every indication in the record, including the ones that turned out to be nothing, because an auditor cannot distinguish an evaluated-and-rejected indication from an ignored one unless you wrote it down.

Atlantis supplies the technical authority end of that: written practice development, NDT procedure development and qualification, personnel qualification within the practice, independent review of examination data, and technical representation during the audit itself. If an audit is scheduled — a client audit, a certifying body, a class society or an owner agency — the useful first step is a gap review against the specification actually invoked on the contract. Send us the specification and the audit date and we will scope the review to what will actually be sampled.

Fracture-critical member traceability across inspection cycles

AWS D1.5's fracture-critical designation is not a one-time label. Once a member is designated FCM, every inspection cycle for the life of the structure has to be traceable back to that designation, and the traceability chain is exactly what an owner-agency audit tests hardest. It is not enough that the current inspection report lists the member as fracture-critical; the auditor wants to see that the designation has been carried forward consistently since the original fabrication drawings, through every subsequent inspection contractor, without a member quietly dropping off the FCM list because a later inspector working from an undated drawing set did not recognise it.

The failure mode this produces is specific: a bridge changes inspection contractors over a twenty-year service life, the FCM list is rebuilt from whatever drawing set the new contractor was handed, and a member that was fracture-critical on the original fabrication set is absent from the rebuilt list because the drawing set they received was an as-maintained revision that never carried the FCM annotation forward. The member gets inspected to non-fracture-critical acceptance criteria for one or more cycles before anyone notices the gap.

The control that actually closes this is a standing FCM register held independently of any single contractor's working files — a document whose authority does not reset when the inspection contract changes hands. Atlantis reviews inspection programmes specifically for this kind of drift: whether the FCM designation on the current cycle's welds matches the original fabrication record, not merely the previous cycle's report.

Request a consultation