What Makes an Inspection Record Credible in a Claim File

What a non-specialist can actually verify in an NDT record

An adjuster, an underwriter or a purchaser cannot judge whether an ultrasonic indication was correctly sized. They can judge everything around it. Whether the technician's certification covered the work date. Whether the instrument had a valid calibration. Whether the written procedure attached to the file is the revision that was current that day. Whether the summary numbers appear anywhere in the raw data. These are documentary checks, and they resolve most credibility questions before any technical judgement is required.

The value of separating those two layers is practical. Documentary findings are unanswerable, because a date either falls inside a window or it does not. Technical findings are arguable, because two qualified people can size the same indication differently. A validation that leads with documentary findings and reserves technical opinion for what genuinely requires it produces a report the file can use.

This is the ordering used across report validation, and the underlying standard a record has to meet is set out in what makes an NDT report defensible. The insurance context changes the audience, not the method.

Contemporaneous records versus reconstructed ones

A contemporaneous record is created while the work happens: field notes with running times, instrument screens captured on the day, thickness readings written down as they were taken, a scan file with its original timestamps. A reconstructed record is written afterwards from those inputs, or from memory when the inputs are gone. The first is evidence of the inspection. The second is testimony about it, presented in the format of evidence.

Reconstruction is not automatically improper. Inspection organisations legitimately produce a formal report days after fieldwork, drawn from field records. What matters is whether the field records exist and can be produced, and whether the formal report is consistent with them. The failure case is a report that has no underlying field record at all, and whose first physical existence postdates the loss.

Reviewers look for the seams. A polished report with no field notes behind it. Uniform dating across work that spanned multiple shifts. Instrument parameters transcribed from the procedure rather than recorded as used. Readings that cluster on round numbers in a way real measurement does not produce. A single one of these prompts a request for the underlying data. Several together define the shape of the finding.

Chain of custody for inspection data

Chain of custody for inspection data answers a narrow question: between the moment the data was captured and the moment it reached the file, who held it and what changed. Modern NDT produces files, so the question is concrete. Scan data, radiographic images and thickness logs carry creation and modification metadata. A file whose modification date sits after the loss, with no explanation, is a finding regardless of what it contains.

The practical failure is not tampering; it is transfer without record. Data moves from an instrument to a laptop to a contractor's server to a client's document system to a claim file, and nobody logs the hops. By the time anyone asks, the original capture files have been superseded by a converted copy, and the conversion is undocumented. A reviewer reports the gap without characterising it, because the technical record supports the observation and not an inference about intent.

Owners who care about this build it in advance rather than reconstruct it later. Certification records, calibration certificates, procedure revisions and inspection data held in a controlled system with an audit trail answer custody questions automatically. Facilities running that discipline through an inspection and asset ERP close this attack surface as a matter of routine rather than as a response to a claim.

Personnel currency: the first thing an opposing reviewer attacks

Personnel currency goes first because it is documentary and binary. The reviewer requests the technician's certification record, the employer's written practice under which it was issued, and the vision examination, then compares three dates against one date. There is no room for technical disagreement in that comparison, which is exactly why an opposing reviewer starts there rather than with interpretation.

The governing document is the employer's written practice, not the recommended practice behind it. ASNT SNT-TC-1A, 2020 edition, is a recommended practice that each employer adopts and tailors, and it recommends recertification at intervals not exceeding five years together with an annual near-vision acuity examination and demonstrated ability to distinguish the colours or grey shades the method uses. Whatever intervals the employer wrote into its own practice are the intervals it will be measured against.

The most common finding is not an unqualified technician. It is a qualified technician whose annual vision examination lapsed by a few weeks across the work date, or whose certification covers a related method rather than the one performed. Neither says the inspection was wrong. Both say the record cannot demonstrate compliance, and both are avoidable with a certification matrix that is checked before mobilisation rather than after a loss.

Calibration traceability: the second thing

Calibration is attacked second, for the same reason. It reduces to a chain of documents: the instrument serial number on the report, a calibration certificate covering that serial number and that date, and a reference block or standard whose own traceability is documented. A break anywhere in that chain is a statement of fact rather than an opinion, and it does not require the person making it to understand ultrasonics.

Instrument performance checks sit alongside the calibration certificate. ASME BPVC Section V, Article 4 requires screen height linearity under T-461.1 and amplitude control linearity under T-461.2, at intervals not exceeding three months for analogue instruments and one year for digital instruments, or prior to first use thereafter. A report that names an instrument without a current linearity check has omitted a requirement that appears in the code text, and the omission is checkable by anyone with the code open. The wider examination framework is summarised in ASME BPVC Section V.

Reference blocks are the piece most often forgotten. A calibration is only meaningful with respect to a specific block, and the block has to be identified, dimensioned appropriately for the material and geometry examined, and itself traceable. A report stating that the instrument was calibrated, without naming the block, has documented an activity rather than a standard.

What a validation can establish

A technical validation can establish that the personnel who performed the work held qualification covering the work date under a written practice that was in force. It can establish that the instrument was calibrated against an identified standard and had passed its required performance checks. It can establish that a written procedure existed, at a stated revision, and that the technique described in the report matches it.

It can establish internal consistency: whether every value on the summary sheet traces to a reading or an image in the underlying data, whether the coverage claimed matches the scan or shot record, and whether the conclusions drawn follow from the data presented. Where raw data survives, it can establish whether an independent reader reaches the same interpretation, which is a materially stronger statement than the original report alone.

It can also establish the negative precisely, which is often what the file needs. A finding that says the record supports process compliance but cannot support a conclusion about condition on a specific date is a useful, defensible statement. It tells the people making the coverage and liability decisions exactly how far the technical evidence reaches, and it stops the record being asked to carry weight it was never built to carry.

What a validation cannot establish

A validation does not determine coverage, interpret policy language, allocate liability, establish causation of a loss, or quantify damage. Those questions belong to the parties whose job they are, and a technical reviewer who drifts into them weakens the technical findings by association. The discipline is to answer the inspection question completely and stop.

There are technical limits too, and stating them plainly is part of the work. A validation cannot establish what the equipment contained on a date when nobody examined it. It cannot recover data that was never captured. It cannot convert a summary report into raw evidence. And where the technique used was incapable of detecting the feature now in dispute, the validation says the record is silent on that feature rather than treating silence as absence.

That last distinction carries real weight in claim files. An inspection that found nothing is not the same as an inspection that could not have found it. A reviewer who identifies a technique-capability limitation is describing what the record can support, not criticising the original technician, who examined for what the procedure asked for.

When the record holds up

Records survive review more often than the framing of a dispute suggests. The technician was certified, the vision examination was current, the instrument was calibrated against an identified block, the procedure revision matches, the field notes exist and the report reconciles to them. The reviewer writes that finding, and the record moves from unverified to independently checked.

That outcome has direct value to whoever commissioned the review. A record that has been examined by an independent Level III and found sound is a stronger position than an unexamined record, whichever side of the file it sits on. It also short-circuits the standard opening attack, because the personnel and calibration questions have already been asked and answered with documents attached.

Neither party can predict the outcome before the review runs, which is the argument for running it early. Records that look disorganised frequently survive intact because the fieldwork was rigorous and the write-up was brief. Records that look immaculate occasionally turn out to have no field notes behind them at all. The finding is the deliverable in both directions, and it is worth having before positions harden.

Building records that survive review before a loss happens

Everything an opposing reviewer attacks is cheap to fix in advance and expensive to fix afterwards. A certification matrix checked at mobilisation catches the lapsed vision examination. A calibration register tied to instrument serial numbers catches the expired certificate. A requirement that raw data be delivered with every report catches the missing field notes. None of this is technically demanding, and all of it is invisible until the day it matters.

Facilities operating under OSHA process safety management already have the skeleton in regulation. 29 CFR 1910.119(j)(4)(iv) requires each inspection and test to be documented with the date, the name of the person who performed it, the serial number or other identifier of the equipment, a description of the inspection or test performed, and the results. A record built to satisfy that sentence answers most credibility questions without further work. The programme context is covered in OSHA PSM mechanical integrity and NDT.

Organisations that want their inspection records to hold up under adversarial review commission a validation on a sample of current reports rather than waiting for a claim. Atlantis performs independent record review as a technical exercise, quoted on request and priced affordably against the record inventory rather than against the size of the underlying dispute. Send a description of the records and the question at contact.

Request a consultation