Making the NDT Evidence Chain Hold Before the Auditor Pulls On It

The auditor does not read your manual, the auditor pulls a thread

Every experienced auditor works the same way, whatever the badge on the certificate. They take a finished item, pick a weld on it, and ask for everything that touches that weld. The drawing at the revision it was welded to. The welding procedure specification and the qualification record behind it. The welder, and that welder's continuity. The parent material heats and their certificates, with evidence of identification transfer. The examination request, the technique sheet, the raw data, the interpretation and the certification of the person who made it. Any repair, and the re-examination of that repair. The heat treatment chart, and the test report.

That vertical slice is far more revealing than a horizontal review of any single register. Registers are usually tidy. Threads are where systems fail, because a thread crosses departmental boundaries and every boundary is a place where a document can be superseded, misfiled or never created. The auditor knows this. The shop, working department by department, often does not, because nobody inside the shop ever traverses the whole chain in a single sitting.

So the preparation that works is the preparation that copies the method. Pick three welds from three different jobs, ideally jobs run by different project teams, and follow each one end to end without helping yourself. Note every point where you had to ask someone for an explanation rather than produce a document. Those are your findings, and finding them yourself is worth an enormous amount more than finding them in a closing meeting. This is the core of what ASNT Level III consulting delivers before an audit.

Four different audits, four different appetites

A fabrication shop typically faces four kinds of audit and they are not interchangeable. The joint review conducted for a Certificate of Authorization brings a team leader from the certifying body together with the Authorized Inspection Agency, and it centres on whether the documented quality control system is genuinely implemented, usually demonstrated on a real item taken through the system in front of the team. A second-party client audit is narrower and sharper, driven by the client's own specification and by whatever went wrong on their last order.

A class society or notified body survey follows its own scheme rules and tends to concentrate on personnel qualification, procedure approval and traceability of materials. A regulatory or statutory inspection is concerned with legal compliance in the jurisdiction of manufacture and is largely indifferent to the contract. The same shop can pass one and fail another in the same month, because each is sampling a different thing.

The practical consequence is that preparation must be aimed. Preparing for a client audit by rehearsing the certification scheme requirements wastes the effort. Read the audit scope in advance, identify what this particular auditor is empowered to sample, and rehearse that. Where the audit will reach into examination records, the technical authority for those records needs to be in the room and needs to have seen the files beforehand, not for the first time on the day.

The written practice is the document that fails most often

The written practice is the shop's own law on personnel qualification, built on SNT-TC-1A or CP-189 and invoking a specific edition. It states education and experience prerequisites, training hours, the structure and grading of general, specific and practical examinations, vision requirements and their frequency, the certifying authority, the recertification interval, and how interrupted service and transferred certifications are handled. It is short, it is easy to write ambitiously, and it is the single most common source of findings.

The mechanism of failure is always the same. The practice was adapted from a template and promises more than the shop performs. Forty hours of training where the file shows sixteen. A composite grade recorded with no individual examination scores behind it. A practical examination described in detail and evidenced by a single signature. A vision requirement stated as annual and last performed two years ago. None of this reflects incompetent examiners. It reflects a document that was never reconciled with the filing cabinet.

There are two honest remedies and both are legitimate. Bring the records up to the practice, or amend the practice so it describes what the shop genuinely does and can sustain, then meet it without exception. Auditors respect the second far more than shops expect, because a modest practice fully met is a working system while an ambitious practice half met is not. Where the shop is genuinely short on qualified people, structured NDT training and certification to SNT-TC-1A and ISO 9712 closes the underlying gap rather than the paperwork symptom.

Welding records and the interface with the Authorized Inspector

Welding documentation belongs to the welding engineer, not to the Level III, and the boundary matters when an audit starts assigning responsibility. The welding procedure specification and its supporting qualification record are prepared and signed on the welding side and accepted by the Authorized Inspector where the code of construction requires it. The examination that supports a procedure qualification, however, is examination work, and its adequacy is a question for the Level III.

The findings here are predictable. A production weld thickness outside the range the qualification supports, which happens whenever a shop qualifies on a convenient coupon and then welds heavier sections. A change in an essential variable that nobody recognised as essential. A welder qualification lapsed under the six month continuity rule and never renewed, discovered when the auditor compares the welder identification on the joint to the register. Filler metal classification substituted for availability without a supporting qualification.

Keep the interface clean in both directions. The Level III should not sign welding procedure specifications, and the welding engineer should not sign examination procedures or certify examination personnel. Audits go badly when one person signs everything, because the auditor immediately asks how independent the technical review can possibly be. Where the shop is small enough that roles overlap by necessity, document the arrangement deliberately rather than letting it emerge by default.

The numbers auditors actually check in examination records

Radiography is checked numerically because it can be. Density through the area of interest must be at least 1.8 for single film viewing with X-ray, at least 2.0 with gamma, and no more than 4.0, with variation held inside minus 15 and plus 30 percent of the density measured through the image quality indicator body. The indicator itself must be the right one for the thickness, correctly placed, with the essential hole or wire visible. Geometric unsharpness is bounded by material thickness under Section V, and location markers must permit the film to be placed back on the weld unambiguously.

Ultrasonic records are checked for traceability more than for numbers. Which instrument, by serial number. Which search unit, which cable, which calibration block, and is that block traceable. Are the screen height and amplitude control linearity checks on file and do they cover the instrument that was actually used on the job. Was the scan plan capable of covering the required volume, and does the recorded coverage match it. Penetrant and magnetic particle work is checked on dwell times, temperature range, field adequacy demonstrated with an artificial flaw shim rather than a field indicator alone, and lighting measured rather than asserted.

None of these are difficult requirements. They fail because the records were completed at the end of a shift from memory, or because a form was designed without a field for the number the code requires. A form that has no box for densitometer calibration date will reliably produce reports with no densitometer calibration date. Fixing the form fixes the finding, and where a shop has moved its examination records into inspection management software the required fields can simply be made mandatory at entry.

The findings that recur in pressure vessel and fabrication shops

The same short list appears year after year. Vision records expired or missing for one or two examiners. The Level III of record named in the practice with no current agreement on file. Procedures present but never demonstrated on representative geometry where the method requires demonstration. Reader sheets signed by a Level II whose certification lapsed between the date the film was shot and the date it was interpreted. Radiographic technique sheets that do not match the geometry visible on the film.

Then the process failures. Repairs excavated and re-examined over the excavation only, rather than over the full extent the code of construction requires. Nonconformance reports closed with a disposition but no extent-of-condition check across other items built the same week. Subcontracted examination laboratories used routinely with no evaluation on file, no review of their written practice, and no right of access written into the subcontract. Nameplates stamped before the final documentation review was complete, which converts a paperwork gap into a certification problem.

Every item on that list is preventable at negligible cost and expensive to fix under audit conditions. The reason they persist is that each sits at a boundary between two people's responsibilities, and boundaries are exactly what a vertical thread audit crosses. An external technical review that follows the thread rather than the org chart will find them, and finding them in March is a very different experience from finding them in a closing meeting in September. Independent report validation on a sample of completed jobs is the cheapest way to see what an auditor would see.

What the Level III signs, and what the Level III does not sign

The Level III signs the written practice, examination procedures, procedure qualification and demonstration records, the certification of Level I and Level II personnel within the practice, technical evaluations and deviation justifications, interpretations and reader sheets where they personally interpreted the data, and independent review reports. That is a substantial and clearly bounded set of responsibilities, and each signature carries a technical judgement the signer must be able to defend under questioning.

The boundaries are just as important. Atlantis supplies examination technical authority. It is not the Authorized Inspector and does not perform the Authorized Inspector's acceptance, does not sign the manufacturer's data report or the certificate of conformity for the vessel, and does not hold or lend a Certificate of Authorization. It is not a process safety management auditor and does not perform compliance audits under that programme. It is not the inspector of record for in-service equipment under API 510, 570 or 653, and it does not sell inspector certification training for those programmes.

Being explicit about this at the engagement stage prevents a specific and damaging failure mode: a shop assuming that because a Level III has been engaged, someone else's statutory role has been covered. It has not. What has been covered is the examination system, its procedures, its people and its data, which is the part of the audit where fabrication shops most often lose points. Clarify the scope in writing at the start, and the audit conversation stays on ground you control.

Gap review before the audit, remediation after the finding

A gap review is a rehearsal you run on your own terms. Typically two to three weeks: sample several completed jobs, run the vertical thread on each, review the written practice against the personnel files, review procedures against Section V and against the code of construction, examine a sample of radiographs and ultrasonic records for technical adequacy rather than for presence, and examine the nonconformance and repair history for closure quality. The output is a ranked list of gaps with the evidence and the fix, and everything on it is additive: procedures written, records completed, people certified, forms corrected.

Remediation after a recorded finding is a different animal because the clock belongs to someone else. A finding demands root cause rather than correction, containment of the immediate item, an extent-of-condition review across everything else potentially affected, a corrective action with an owner and a date, and objective evidence that the action worked. If the finding touches the certification scheme it can suspend shipment or hold a data report, and the commercial consequence usually dwarfs the technical work.

The arithmetic strongly favours rehearsal. A gap review costs a few weeks of technical attention and disturbs nothing. Remediation costs the same technical attention plus schedule, plus a corrective action programme, plus a follow-up verification, plus a client who now watches every subsequent shipment more closely. If a joint review, a client audit or a class survey is on the calendar within the next two quarters, the useful time to start is now. Book a consultation and begin with the thread audit, because that is the test the auditor will run.

Request a consultation