What a Reviewer Checks in a Pipeline Girth Weld Record

What a girth weld report review actually examines

The review starts with the record, not the weld. Four artefacts have to be present together: the NDT report itself, the interpreter's qualification file, the written procedure the shot was taken under, and the weld map that ties the number on the report to a position on the line. A package missing any one of the four cannot be closed on paper, and the gap is worth knowing before a crew is standing in an open ditch waiting for a disposition.

The second step is a blind re-read. A reviewer who reads the original interpretation first will anchor to it. Working the film or the AUT data cold, recording indications independently, and only then comparing against the original call is what makes the review worth commissioning. Where the two readings agree, the original report gains an independent corroboration it did not have before. Where they diverge, the divergence is specific enough to argue about.

This is the same discipline applied across every record type in report validation, and the general principles are set out in what makes an NDT report defensible. Pipeline records add one complication the others do not have: the interval between the work and the review is measured in decades, so every person who could explain an ambiguity has usually left.

Technician qualification on the date of the weld, not today

Qualification is judged as of the work date. A technician who holds current certification in 2026 tells you nothing about a radiograph shot in 1994. The reviewer needs the certification record, the employer's written practice under which it was issued, and the vision examination, all covering the day the film was exposed. 49 CFR 192.243(b) requires nondestructive testing to be performed by persons trained and qualified in the established procedures and with the equipment employed, and 49 CFR 195.234 imposes the same on hazardous liquid lines.

The binding document is the employer's written practice, not the recommended practice behind it. ASNT SNT-TC-1A, 2020 edition, is a recommended practice that an employer adopts and adapts; it recommends recertification at intervals not exceeding five years and an annual near-vision acuity examination, along with demonstrated ability to distinguish the colours or grey shades used in the method. A reviewer checks the employer's adopted intervals against the dates in the file, because those adopted intervals are what the employer bound itself to.

Where the file is thin, the practical remedy is an independent re-read by a currently qualified Level III rather than an argument about a lost certificate. Method-level competence is what the regulation is protecting, and it can be supplied now. Atlantis delivers this through outsourced ASNT Level III consulting, and runs NDT method training for operators rebuilding in-house capability after a records gap.

Workmanship versus alternative acceptance: which basis was used

A girth weld is accepted on one of two bases, and the report has to say which. 49 CFR 192.241(c) states that the acceptability of a weld that is nondestructively tested or visually inspected is determined according to the standards in section 9 or Appendix A of API Std 1104. 49 CFR 195.228(b) says the same for hazardous liquid pipelines. Section 9 is the workmanship route: an indication is judged against fixed dimensional limits. Appendix A is the alternative route, where an engineering critical assessment establishes a tolerable flaw size for the actual pipe, loading and toughness.

One rule cuts across both citations without exception. Appendix A of API Std 1104 may not be used to accept cracks. That sentence appears verbatim in both 192.241(c) and 195.228(b). A record showing an indication characterised as a crack and then dispositioned under an alternative criteria calculation fails on the regulation alone, before any re-read of the image. This is the single most consequential finding in girth weld record review, and it is fast to check.

When an alternative acceptance route was used, the review has to extend to the assessment behind it: the flaw dimensions assumed, the toughness input, and the loading case. That is fitness-for-service work under API 579, and a girth weld accepted on an ECA is only as good as the assessment that produced the criteria.

The standard edition trap

Reports cite the edition the contractor had on the shelf. Regulations cite the edition the agency incorporated. These are rarely the same document. At 49 CFR 192.7, PHMSA incorporates API Standard 1104, Welding of Pipelines and Related Facilities, 21st edition, September 2013, including Errata 1 through 5 issued between April 2014 and September 2018, Addendum 1 (2014) and Addendum 2 (2016). That incorporation is approved for use in 192.225(a), 192.227(a), 192.229(b) and (c), 192.241(c), and Appendix B to Part 192.

So a 2023 report citing a newer edition of API 1104 has applied criteria that PHMSA has not incorporated. That is a finding a reviewer records plainly: the acceptance may still be correct, and the criteria may even be more conservative, but the record does not demonstrate compliance with the incorporated text until someone maps one to the other. Operators discover this during audit far more often than during construction.

The reverse case matters too. A weld made in 1978 was accepted against whatever edition was in force then, and re-judging it against a 2013 text is a category error. The reviewer establishes the edition in force on the weld date, states it in the finding, and applies it. A validation that silently swaps editions produces a rejection that will not survive challenge.

Reading the repair record

Rejection is not the interesting part of a construction record. Disposition is. 49 CFR 192.243(f) requires the retained record to show the number rejected and the disposition of the rejects, which means every rejected weld has to end somewhere: repaired and re-tested, or cut out and replaced with a new weld carrying a new number. A reviewer reconciles the reject count against the disposition entries, and the arithmetic either closes or it does not.

Three patterns account for most failures. A weld number appears twice with no repair narrative between the two entries. A repair is recorded with no re-test result attached, so the repair was never proved. A cut-out is noted with no replacement weld number, leaving a gap in the sequence that nobody can account for thirty years later. None of these require reading a radiograph to find, which is why they are checked first.

When a girth weld resurfaces during an integrity dig, this is often the whole question. The weld in the ditch shows a repair cap. The construction record says the weld was accepted first time. Reconciling those two facts, or documenting that they cannot be reconciled, changes what the integrity engineer does next.

Traceability from the weld number back to the crew

Traceability runs in a chain: excavated weld to weld number, weld number to milepost or engineering station, station to the welder and welding procedure, procedure to the NDT technique used, technique to the interpreter, interpreter to the retained image or data file. A validation walks that chain in both directions and reports the first link that breaks. 49 CFR 192.243(f) fixes the geographic end of the chain by naming milepost, engineering station or geographic feature as the permitted identifiers.

Welder-level traceability has its own regulatory hook. 49 CFR 192.243(e) requires a sample of each welder or welding operator's work to be nondestructively tested for each day, and 49 CFR 195.234 requires at least 10 percent of the girth welds made by each welder and welding operator during each welding day. A record that reports coverage only as a segment percentage, with no welder identity attached, cannot demonstrate either requirement.

PHMSA's records language elsewhere is useful as a test even where it does not directly apply. At 49 CFR 192.607 the agency requires material records that are traceable, verifiable and complete, and it does not define the three words. Reviewers apply them to weld records anyway, because 192.243(f) makes those records life-of-pipeline documents: traceable to a specific weld, verifiable against a retained original, complete across the reject and disposition counts.

Radiographic records versus AUT records

A radiographic package is reviewable because film survives. The reviewer checks image quality indicator placement and sensitivity, density, weld and location identification burned into the image, and whether the coverage geometry could actually image the region the report calls acceptable. Interpretation is then re-performed against the acceptance basis in force. Old film that was stored badly is its own finding, and a package where the images are gone reduces to a paper audit that can confirm process but cannot confirm the call.

Automated ultrasonic records fail differently. AUT produces strip charts, zone assignments and calibration records rather than a picture a generalist can read. The reviewer checks the calibration block used and whether its reflectors match the pipe wall and bevel geometry, the zone map against the weld preparation, coupling monitoring across the scan, and whether any zone lost data. An AUT report with no coupling record is not reviewable, because there is no way to distinguish a clean zone from an unmonitored one.

The choice between the two methods drives what can be reconstructed years later, which is covered in RT versus UT for weld inspection. Where an operator's procedures cross into pressure equipment written to ASME rules, the examination requirements in ASME BPVC Section V apply alongside the pipeline standard rather than instead of it.

When the review confirms the original report

The common outcome is confirmation. The indication was called correctly, the acceptance basis was the right one, the qualification file covers the date, and the disposition arithmetic closes. Operators sometimes treat that as a wasted engagement. It is the opposite: the weld now carries an independent second reading, performed by a Level III with no commercial relationship to the original contractor, documented against the regulation. That is a stronger record than the operator held the day before.

A confirmed record does concrete work. It closes the dig without a cut-out. It supports the integrity management decision that follows. It ends the internal argument between the integrity group and the field crew that started when someone said the record looked thin. And when a regulator or an acquirer asks the same question two years later, the answer is already written and dated.

Validation is worth commissioning precisely because you cannot tell in advance which outcome you will get. A package that looks disorderly often survives review intact, and a clean-looking package sometimes turns out to cite an edition that was never in force. The finding is the product, in either direction.

Who does pipeline girth weld report review, and how the scope is set

The reviewer needs three things: Level III qualification in the method being reviewed, working knowledge of the API 1104 edition in force on the weld date, and independence from the contractor who produced the original report. Independence is not a formality. A review performed by the same organisation that shot the film is a re-read, and it carries no weight in an audit or a dispute where the original interpretation is the thing being questioned.

Scope is set from an inventory before any price is discussed. How many welds are in the package. Whether original film or raw AUT data exists, or only a summary sheet. Whether the personnel and procedure files are retrievable. Whether the weld map has to be reconstructed from as-built drawings. A dig closeout on one weld and a segment records audit are different pieces of work built from the same method.

Atlantis is affordable, accessible and fully customisable on this work, and quotes on request against your file inventory. Send the package list through contact and the response states what is reviewable from what you hold, what an independent re-read would add, and what has to be reconstructed before any conclusion is possible.

Request a consultation