ASNT Level III Technical Authority for Shipyard NDT Audits

Who is auditing a shipyard's NDT, and against what

Three parties audit NDT in a shipbuilding or repair yard, and they do not want the same things. The class society surveyor — ABS, DNV, Lloyd's Register, Bureau Veritas, RINA, ClassNK, KR or CCS — audits against the society's rules and the approval basis for that vessel. The owner's or client's representative audits against the building specification, which is usually stricter and always more specific. A quality system auditor, where the yard holds ISO 3834-2 or a client-mandated system approval, audits the process rather than the product.

The technical standards behind all of them are broadly common. ISO 17635 governs how a method is selected and how the specified weld quality level maps to a testing level and an acceptance level. ISO 17636 covers radiography, ISO 17640 with ISO 11666 covers ultrasonics, ISO 13588 with ISO 19285 covers phased array, ISO 17637 covers visual testing, and ISO 5817 defines the quality levels that everything else is anchored to. Class guidance such as DNV-CG-0051 and the ABS materials and welding rules sits on top, adding extent, weld category and reporting requirements.

The recurring surprise for yards used to being audited on workmanship is how little of the audit is about workmanship. A surveyor who wants to know whether the welding is good looks at the welding. A surveyor conducting an audit wants to know whether the yard could prove the welding is good to a third party in two years, with the original crew gone. Those are different questions and the second one is answered entirely on paper.

The evidence chain, and where it usually breaks

The audit method is vertical, not horizontal. The surveyor selects one weld from a block and follows it down: drawing revision, approved welding procedure and its qualification range, welder identity and qualification validity on the welding date, the method and extent required by the quality level, the NDT procedure revision in force on the test date, the operator's certificate and vision record, the equipment verification and reference block, the report, the acceptance standard cited, and the disposition of anything rejectable. Nine or ten links, each requiring a dated document with a signature.

The breaks cluster in three places. Procedure revision control is the first: the yard can produce the current revision but not the one in force on the date of test, so the report cannot be tied to the rules it was performed under. Personnel scope is the second: the certificate is valid but its sector or product scope does not cover what was examined, or the annual vision record expired mid-contract. Equipment is the third: instrument verification is current, but no transfer correction was determined for the actual plate thickness, surface condition and curvature, so the sensitivity used cannot be shown to be the sensitivity required.

None of these is discovered by looking at welds. All of them are discovered in about forty minutes with a filing system, which is why a pre-audit exercise that reproduces the surveyor's own sampling method is worth more than any amount of general readiness. This is the core of what our ASNT Level III consulting work does in yards: pull ten weld numbers at random and try to close the chain on every one, before someone else does it with authority.

Subcontracted NDT: the yard owns what it did not perform

In most yards the majority of NDT is subcontracted, and in almost all of them the audit lands on the yard. The society's position is straightforward: the yard is responsible for the product, therefore the yard is responsible for demonstrating that examinations performed on its behalf met the approved requirements. A subcontractor's report on subcontractor letterhead, filed without review, does not discharge that responsibility. What auditors look for is evidence that the yard technically accepted the work — a review signature, a records check, something showing the yard read the report rather than filed it.

There is a second layer where the service is performed for survey purposes. Thickness measurement of hull structure, tightness testing and in-water survey fall under the IACS procedural requirements for service suppliers, which put the firm itself inside a class approval regime with a defined scope, named qualified operators, controlled equipment and periodic audit. Auditors check the approval certificate's validity, and then — the part yards forget — whether the scope on the certificate covers the service that was actually performed.

Managing subcontractor evidence is where most yards lose control, because reports arrive in volume from several firms in different formats, and the operator certificates behind them expire at different times. The practical fix is to hold subcontractor personnel certificates, equipment records and reports in the same controlled repository as the yard's own, so that validity on the date of test can be answered without emailing a subcontractor. Yards that run this properly usually end up managing it inside an inspection data management system rather than in project folders.

Personnel certification: ISO 9712 against employer-based schemes

This is the most common structural mismatch in international shipbuilding. ISO 9712 certification is issued by a third-party certification body accredited to ISO/IEC 17024, covers a specific method, level, sector and scope, and belongs to the individual. Employer-based certification under SNT-TC-1A or CP-189 is issued by the employer under a written practice, with a Level III responsible for the examinations, and belongs to the employment relationship. Both are legitimate; they are not interchangeable, and a specification that names one does not accept the other by implication.

Yards drawing on a mixed labour market — a US-based subcontractor on a European newbuild, or an offshore repair crew mobilised into a yard whose class approval was written around ISO 9712 — hit this repeatedly. The technicians are competent. The certificates are real. They are simply the wrong scheme for the contract, and the finding is written against the yard for accepting them. The place to catch it is the subcontract, at approval stage, not on the deck of a block during a survey.

Within an employer-based practice, the Level III's role is defined and limited: writing and maintaining the written practice, preparing and grading examinations, certifying personnel within that practice, and approving procedures. Where the contract requires ISO 9712, personnel must go to an accredited body, and the yard's route is to plan that lead time into mobilisation. Where a yard is building its own certification capability under an employer practice, the training and examination structure has to be designed against the written practice from the outset, which is the work our NDT training and certification programmes are set up to do.

Extent, sampling plans and the extension rule

Class rules and the building specification define how much NDT is required and where, usually by weld category: erection and block joints, crossing joints, shell and deck plating, higher-stress regions, and areas identified in the approved plan. The percentages are not the interesting part. The interesting part is what happens when a weld fails.

Almost every regime requires the examination to be extended when a rejectable indication is found — additional length either side of the defective location, with further extension if the additional lengths also fail. In practice the extension is usually performed, because the technicians know the rule. What is missing is the evidence: the extension welds are examined under a new report number with no reference to the rejection that triggered them, so an auditor reading the records cannot demonstrate that the rule was applied. It reads as a random additional test.

The fix is a reporting convention rather than a technical change. Every extension report should carry the identifier of the rejection that caused it, and the weld map or NDT log should show the extension against the original location. It costs nothing at the time and it removes an entire category of finding. The same convention makes the sampling plan auditable in the other direction — you can show that the specified percentage was achieved by category, which is a question surveyors increasingly ask on repair work where the plan evolves with the survey.

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

Scope discipline protects everyone, and audits expose it quickly. An ASNT Level III supplying technical authority to a yard signs the written practice, the NDT procedures and their qualification records, technique sheets, the examination records of personnel certified within the employer's practice, and technical reviews of examination data where the yard's acceptance signature is required. Those signatures are meaningful because they are backed by method qualification and demonstrable independence from production pressure.

The list of what the Level III does not sign is equally important. Not the class survey report — that belongs to the surveyor. Not an ISO 9712 certificate, which only an accredited certification body may issue. Not the yard's welding coordination responsibilities under ISO 14731, which sit with the yard's designated welding coordinator. Atlantis is not a PSM auditor, is not the API inspector of record on any asset, and does not deliver API 510, 570 or 653 inspector certification training; those are separate roles held by separate bodies and we say so in writing at proposal stage.

Audit representation follows the same boundary. The Level III attends to answer technical questions about method, procedure, sensitivity, acceptance level and disposition — to give the auditor a competent counterpart rather than a project manager guessing. It is not a negotiating role. Where the yard's position is technically wrong, the useful thing an independent Level III does is say so before the surveyor has to, because a finding the yard raises itself is a corrective action and a finding the surveyor raises is a condition of class.

Findings that recur in newbuild and repair yard audits

The same list appears across yards on three continents. Annual vision examinations expired mid-contract, which is the single most frequent NDT personnel finding anywhere. Certificates valid but with the wrong sector or product scope. Superseded procedure revisions not retrievable. Ultrasonic sensitivity set on a reference block of different material, thickness or curvature than the production item, with no transfer correction recorded. Subcontractor reports filed without any yard acceptance signature. Extension examinations after a rejection with no traceable link to the rejection.

Newbuild adds its own. Welder qualification continuity confirmations lapsed at the six-month interval while the welder continued producing class work. Consumables used outside the approved grade or stored outside the bake-out regime. Welding procedures approved for a range that does not quite cover the production thickness, usually at the top end where the shell plating thickens. And the mapping error that costs nothing to fix and is visible from the procedure alone: a testing level applied that is below the level ISO 17635 requires for the specified ISO 5817 quality level.

Repair yards add records problems of a different shape. Thickness measurement reports where the readings cannot be tied to the class report form locations. Substantial corrosion determinations made without the diminution allowance recorded alongside. Radiographs from an earlier campaign that cannot be produced because the film archive moved. On repair work the vessel's history is the product, and the audit reflects that. Where the historical data is contested, an independent review of the inspection reports and their underlying data is usually the fastest way to establish what can be defended and what has to be re-shot.

Gap review before the audit versus remediation after a finding

A gap review is a rehearsal. It uses the surveyor's own method — vertical sampling from randomly selected weld identifiers — and it runs while production continues normally, which means findings can be closed on a sensible timescale and in a sensible order. The output is a prioritised list with an owner and a date against each item, separated into what must be fixed before the audit, what can be shown as an open corrective action, and what is a longer-term system change. Nothing is at stake commercially, so people tell the truth.

Remediation after a finding is a different exercise entirely, and the difference is not effort but constraint. The yard must contain the issue by identifying all product potentially affected, which on a systemic records failure can mean an entire block. It must correct the immediate non-conformity, establish root cause, define corrective action, and demonstrate effectiveness — all under the surveyor's visibility, frequently with product on hold and sometimes with re-examination at the yard's cost. The audit finding also stays visible in the survey record.

One boundary has to be stated plainly because the pressure to cross it is real. Records are never reconstructed after a finding. Where a certificate was not valid on the date of test, or a procedure revision cannot be produced, or a calibration record does not exist, the correct response is to document what actually exists and re-examine where the evidence is absent. Back-dating or recreating records converts a quality finding into a integrity failure, and no surveyor, owner or insurer treats those as the same thing. If you are preparing for a class or owner audit and want the chain sampled the way the surveyor will sample it, start with a scoped gap review rather than a general readiness exercise.

Request a consultation