What a defensible NDT report has to carry, element by element

Defensible means a stranger can reconstruct the examination from the paper

A report is defensible when a competent reviewer who was not there can read it and rebuild what happened: which component, which area, under which procedure, by whom, with what equipment, using what technique, against which criteria, reaching what conclusion. If any link in that chain is missing, the reviewer has to take the technician's word for it — and a report that requires trust instead of supplying evidence is the working definition of indefensible.

ASME BPVC Section V puts the burden plainly. T-190 places records and documentation under the referencing code section and makes the Manufacturer, fabricator or installer responsible for producing them. Section V then prescribes, article by article, what each examination record must contain. Those lists are not stylistic preferences. They are the minimum content a report has to carry before it can be said to comply, and more on the code sits on ASME BPVC Section V.

The nine elements below are what a validation tests, in the order it tests them. They are also, in practice, the list a client's reviewer runs down before rejecting a report — which is why the same items recur in the report format problems clients reject. Nothing on the list is exotic, nothing on it takes long to write at the time, and nothing on it is optional once the referencing code has been invoked on the front page.

Identification: the component and the area actually examined

The report must identify the component unambiguously — equipment number, drawing reference, heat or spool number, weld identifier — and then identify the specific area examined within it. "Weld 12" is an identifier. "Weld 12, full circumference, from the outside surface, 100% of the weld volume plus one inch either side of the toes" is an examined area. Only the second lets a reviewer determine whether the coverage matched what the code required.

Section V reflects this in the ultrasonic record requirements, which call for identification and location of the weld or volume scanned and the surface from which the examination was conducted. Radiography does it through a dimensional map of marker placement and, on the review form, a listing of each radiograph location. Both are answering the same question: which piece of metal does this result apply to, and how would somebody else find it again next year?

The failure mode here is silent. A report that names the component but not the extent examined will pass a filing check and fail a validation, because the disposition on the front page claims something the body never established. Where partial coverage was contracted, or physically forced by access restrictions, the report must say so on its face rather than letting a blanket acceptance statement imply full coverage that nobody ever achieved.

Procedure number and revision, because the revision is what fails

Naming the procedure is routine. Naming the revision in force on the examination date is where reports come apart. Section V asks for procedure identification and revision in the penetrant and magnetic particle record lists, and for the procedure in the ultrasonic record list. If a report cites "UT-01" with no revision, and UT-01 has been through four revisions in six years, nobody can determine which set of parameters the technician was working to on the day.

The second failure is a procedure that does not cover what was examined: thickness outside the qualified range, a configuration the procedure never addressed, a material group it was never demonstrated on. Section V requires procedure demonstration under T-150 where the referencing code calls for it, and requires that demonstration to be documented as the referencing code specifies. A report resting on a procedure never demonstrated for the configuration is resting on nothing at all.

Where a validation finds procedure gaps that are systemic rather than job-specific, the remedy is upstream — see NDT technical procedure development. One report with a missing revision number is a documentation defect. Fifty reports citing the same undemonstrated procedure is a programme defect, and fixing that retroactively across an archive costs far more than writing and demonstrating the procedure correctly once.

Personnel: name, level, method, and the certifying authority

A defensible report names the individual, their certification level, the method they are certified in, and who certified them. Section V's record lists ask for examination personnel identity and, where the referencing code section requires it, qualification level. That is the floor. A validation goes further and asks for the certification record itself, because initials on a report prove attendance rather than qualification, and the two are routinely conflated by everyone except the person about to reject the report.

Under the ASME model, certification is issued by the employer against a written practice built on ASNT SNT-TC-1A or ANSI/ASNT CP-189, with the 2023 edition of Section V carrying those requirements in T-120(e). The certifying authority is therefore a company, and the evidence is that company's file: training hours, experience hours, examination scores, the Level III who certified, and the vision records. Under ISO 9712 the authority is an independent certification body and the evidence is a certificate.

Currency is checked on the examination date. The 2020 edition of SNT-TC-1A puts recertification at five-year intervals for all levels, requires near-vision acuity examination annually at Jaeger Number 2 or equivalent at not less than 12 in. (30.5 cm), and requires colour contrast differentiation at initial certification and at five-year intervals afterwards. A certificate in date with a vision record that lapsed two months before the scan is a real and frequent finding.

Equipment and calibration with an actual traceability chain

The report must name the equipment specifically enough to identify the unit, and establish that it was in calibration when it was used. Section V's ultrasonic calibration records require the instrument identification including the manufacturer's serial number, the search unit identification including serial number, frequency and size, the beam angles used, the couplant brand name or type, and the search unit cable type and length. Serial numbers, not model names, and not a stock photo of a flaw detector.

Traceability means the calibration certificate leads somewhere: a calibration body, a reference standard, and a date range that contains the examination. A sticker reading "calibrated 03/25" is a claim. A certificate naming the reference standard, the performing laboratory and the due date is evidence. For magnetic particle work, Section V requires yoke lifting power to be verified before first use each day the yoke is used and whenever it has been damaged or repaired — at least 10 lb (4.5 kg) for alternating current and 40 lb (18 kg) for direct current or permanent magnet yokes, at the maximum pole spacing that will be used.

Calibration currency is the element most often lost to time rather than to negligence. Certificates live with the equipment owner, equipment gets sold on, contractors dissolve, and three years later nobody can produce the paper for a probe that was demonstrably fine. Programmes that track certification and calibration expiry as live data rather than as scanned PDFs in a shared folder — the pattern behind an NDT-configured ERP — stop generating this class of finding entirely.

Technique detail, method by method

Ultrasonics. Section V's examination record list asks for the procedure, the examination system, personnel identity and level, the calibration sheet identity, identification and location of the weld or volume scanned, the surface from which the examination was conducted, a map or record of indications detected or areas cleared, the date and time, the couplant brand name or type, identification of the basic calibration block if used, the surface condition, the frequencies, and any special equipment. Rejectable indications are recorded with identity, maximum amplitude, location and extent.

Radiography. The technique documentation asks for the identification required by the code, a dimensional map of marker placement, the number of exposures, X-ray voltage or isotope type, source size, base material and weld thickness with reinforcement, source-to-object distance, the distance from the source side of the object to the film, film manufacturer and type designation, the number of films in each cassette, and whether the exposure and the viewing were single or double wall. The manufacturer's review form then adds a listing of each radiograph location, the evaluation and disposition, the representative who performed final acceptance, and the date of that evaluation.

Radiographic image quality is separately provable. Density limits are 1.8 minimum for single-film viewing with an X-ray source, 2.0 minimum for a gamma source, and 4.0 maximum for single or composite viewing, with density in the area of interest held within -15% and +30% of the density through the IQI. IQI designation follows the thickness table, and film-side placement — permitted only where source-side placement is impractical — requires a lead F marker and an IQI one designation better. Method selection sits behind all of it: see RT versus UT for weld inspection.

Magnetic particle records call for procedure identification and revision, the equipment and type of current, whether the particles were visible or fluorescent and wet or dry, personnel identity and level, a map or record of indications, material and thickness, the lighting equipment, and the date and time the examinations were performed. Penetrant records call for procedure identification and revision, whether the penetrant was visible or fluorescent, the number or letter designation of each penetrant, remover, emulsifier and developer used, personnel identity and level, a map or record of indications, material and thickness, lighting equipment, and the date. Dwell comes from the penetration time table with a two-hour maximum, and a penetrant allowed to dry sends the examination back to the cleaning step.

Acceptance criteria: cite the code and cite the edition

The most consequential single line on an NDT report is the one naming the acceptance criteria, and it is routinely wrong in one specific way: reports cite Section V. Section V prescribes how examinations are performed and documented. It does not set accept or reject limits. Those come from the construction or in-service code the component was built or is maintained to, and citing the wrong document means the criteria applied may never have been the contracted ones.

Edition matters as much as the code. Codes are reissued, and limits, methods and referenced standards move between editions. A report that cites a code without an edition leaves a reviewer unable to determine which limit was applied, and leaves the contractor unable to prove they applied the right one. Where a contract fixes an edition — most do — the report should cite that edition rather than whichever one is current at the time of writing.

A validation checks three things here: that the cited criteria are the ones the contract invoked, that the edition matches, and that the criteria were applied as written rather than as remembered. The third is where interpretation errors surface. An indication evaluated against a length limit taken from a code that measures the same feature differently produces a confident, well-formatted report reaching the wrong disposition. Both reports look equally complete to anybody who is not checking the limit itself.

Results against criteria, and a disposition somebody signed

Results have to be recorded against the criteria, not merely recorded. A report listing four indications with lengths and locations, followed by the word ACCEPT, has skipped the step where each indication is measured against a stated limit. The reviewer then has to do that arithmetic themselves, and any disagreement about the outcome becomes a disagreement about the report. The fix costs nothing at the time and is close to impossible to retrofit years later.

Section V requires rejectable indications to be recorded — for penetrant and magnetic particle work, at minimum the type of indication, linear or rounded, with its location and extent by length, diameter or alignment; for ultrasonics, the identity, maximum amplitude, location and extent of the reflector. Areas cleared matter too: the ultrasonic record list asks for a map or record of indications detected or areas cleared, which is how a reviewer distinguishes "nothing found" from "never examined".

The disposition then has to be explicit and attributable. Accept, reject, repair, or accept with a named concession — signed by somebody with the authority to make that call. Radiographic practice makes this concrete by requiring identification of the Manufacturer's representative who performed final acceptance of the radiographs and the date of that evaluation. An unsigned or unattributed disposition is a recommendation, and recommendations do not close work orders or release retention.

Completeness: the referencing code adds requirements of its own

Section V is one half of the requirement. The referencing code supplies the other half, and a report can satisfy every Section V record item and still be incomplete. In-service inspection codes add previous condition, corrosion rate context, and the inspector's own certification. Class rules under IACS unified requirements add qualification expectations for the responsible supervisor and the operators, including shipbuilding knowledge that a generic NDT certificate does not evidence on its own.

Retention is the quiet element. The referencing code sets how long records are kept and by whom, and validations frequently fail not because a record never existed but because nobody kept it. Where a report is being examined for due diligence or a claim, the retention regime is itself evidence: a programme that cannot produce five-year-old certification files has already told you something material about the reports it was producing five years ago.

The practical test for completeness is adversarial. Hand the report to somebody whose commercial interest is to reject it, and watch what they reach for first. That is the same list an auditor works from — set out in what auditors ask inspection companies — and a report built to survive it needs no defending afterwards. Where individual reports fail the test, each gets its own verdict; where the system producing them fails, the scope becomes a multi-report validation programme.

Request a consultation