How an inspection record holds up when it is read by people outside the industry

Records are judged on process, not on interpretation

Outside the industry nobody evaluates your interpretation. An adjuster, a purchaser's counsel, an arbitrator or a tribunal has no way to assess whether a reflector at a given depth was correctly characterised, and no appetite to try. What they assess is whether the record shows a controlled process: known people, known equipment, known procedure, known date, unbroken from acquisition through to the issued report. That is the test, and it is not the test practitioners prepare for.

This inverts the engineer's instinct. Engineers defend the call. Disputes are decided on the scaffolding around the call. A technically excellent interpretation recorded on an undated field sheet, transcribed by somebody who was not present, into a report signed by a name with no certification level beside it, is worth less in a dispute than a merely adequate interpretation with a complete provenance chain behind it.

The practical consequence is that record quality has to be built at acquisition, because it cannot be added later. Everything described on this page is cheap on the day and impossible afterwards. That is also why the discipline is structural rather than personal — organisations that treat inspection records as controlled documents rather than files in a folder tree survive scrutiny better, which is the whole premise of independent report validation before a record is ever tested.

Personnel qualification: the attack that needs no expert

The first document an opposing expert requests is the certification file for every person who touched the examination. It is requested first because it is binary: on the date of the examination, did this individual hold certification in this method, at this level, for this scope, under the scheme the contract required? No knowledge of ultrasonics is needed to answer it, and once answered it cannot be argued with.

Failures here are almost never misconduct. They are administrative. A certificate that expired eleven days before the examination and was renewed a month after. An annual vision examination performed late. A technician certified for thickness measurement signing weld examination reports. A Level II performing an evaluation the written practice reserves to a Level III. Each is discoverable from a spreadsheet in an afternoon, and each is fatal to the record it supports regardless of how good the underlying work was.

The scheme matters as much as the level. Employer-based certification under SNT-TC-1A rests on the employer's own written practice, authorised by its Level III — so in a dispute your written practice is read as the standard you set for yourself, and you are judged against it. ISO 9712 certification is issued by an independent body and is portable between employers. Contracts and class rules specify one or the other, and scope limitations printed on a certificate are read literally, a point we cover in method and scope limitations.

Calibration traceability: an unbroken chain or nothing

Traceability has a formal definition, and it is stricter than the industry's casual use of the word. Metrological traceability is the property of a measurement result whereby the result can be related to a reference through a documented unbroken chain of calibrations, each contributing to the measurement uncertainty. Every link must be documented and every link must carry an uncertainty. A chain with one undocumented link is not a chain, and calling it traceable does not make it one.

This produces a specific failure that appears constantly and is fatal when tested: a calibration certificate reporting only pass or fail, with no stated uncertainty and no identification of the reference standard used. That document proves somebody checked the instrument. It does not establish traceability. An ISO/IEC 17025 accredited calibration additionally carries the accreditation of the laboratory that produced it — a third party has assessed that lab's people, methods, equipment and quality system — whereas a bare claim of NIST traceability tells you neither who made the measurement nor how.

The chain includes far more than the instrument. Reference and calibration blocks, step wedges, thickness standards, densitometer calibration strips, hardness test blocks and temperature devices each need their own certificates covering the examination date. A thickness survey where the instrument certificate is immaculate and the step wedge has never been calibrated has a broken chain, and an opposing expert will find the step wedge, because that is exactly where they look second.

Contemporaneous versus reconstructed: why the gap is so expensive

A contemporaneous record is made at the time of the examination by the person performing it. A reconstructed record is made afterwards from memory, from a summary, or from another document. The first is evidence of what happened. The second is testimony about what somebody believes happened, in documentary form, and it is weighed that way by anyone with experience of records.

The distinction becomes visible under examination in ways practitioners do not anticipate. Field sheets carrying a single handwriting and a single pen across an entire shift. Thickness tables where readings end in zero far more often than chance allows. Timestamps clustered at the end of a week for work performed across it. A report dated before the calibration certificate it relies on. None of these proves anything alone, and collectively they establish that the record was assembled rather than captured.

The expensive version is the record reconstructed after an incident. Once a failure has occurred, everything written about the preceding inspections is produced by people who know the outcome, and its evidential weight collapses accordingly. The record you rely on in a dispute is the record that existed the day before the incident. There is no way to improve it afterwards, and every attempt to tidy it makes the position materially worse than leaving it thin.

Chain of custody for inspection data

Chain of custody answers one narrow question: is the data in the report the data that came off the asset? For NDT it spans acquisition, transfer, storage, processing and reporting, with each handover identified by person and date. Radiographic film has a physical custody trail that people understand instinctively. Encoded ultrasonic data has a digital one, and digital custody is where nearly every programme is weakest.

The common failure is casual rather than improper. Raw files copied to a technician's laptop, sent from a personal email account, renamed during transfer, edited inside the acquisition software with no original retained anywhere. Nothing wrong has occurred and the chain is nevertheless unprovable. The remedy is unremarkable: retain the original acquisition file unmodified, record checksums or preserve original file metadata, log every transfer, and perform all analysis on copies.

Custody also covers the report itself. A report re-issued under the same number with different content, with no revision history behind it, raises an integrity question that is very hard to answer three years later even when the change was innocent and trivial. Controlled revision numbering with a stated reason for each change removes the question entirely, costs nothing, and is the single cheapest improvement available to most inspection organisations.

What the record must say about what was not examined

The most damaging omission in a disputed record is rarely a wrong call. It is silence about coverage. A report stating that ultrasonic examination was performed on welds W-101 to W-140 with no recordable indications, which does not state that eleven of those welds were only partially accessible, has made a claim broader than the data behind it. That gap is found immediately and it discredits the whole report.

Codes require this to be recorded. The ultrasonic reporting requirements in ASME Section V include a record of inaccessible weld areas alongside recordable and unacceptable indications, with the general records requirements in Article 1 sitting behind every method article. A limitation stated at the time is a professional record. The same limitation raised for the first time in a dispute, to explain why something was missed, reads as an excuse regardless of whether it is true.

Coverage documentation is the cheapest insurance in inspection. A weld register marked examined, partially examined and not examined against each item. A coverage map for volumetric scanning. A stated reason for each limitation. It takes minutes on the day. It is the entire difference between a bounded scope and an unbounded one when the component that failed turns out to be the one nobody could reach.

The expert witness role, and its limits

A technical expert in a dispute is retained to give an opinion within a defined discipline, on a defined question, from defined material. In the United States, expert testimony in federal court is governed by Rule 702, amended with effect from 1 December 2023 to make explicit that the proponent must demonstrate it is more likely than not that the testimony meets the rule's requirements, and that the opinion reflects a reliable application of the expert's principles and methods to the facts of the case.

The practical effect for an NDT expert is a narrower lane than many expect. An opinion must not run beyond what the underlying data supports. 'The recorded data does not support the conclusion stated in the report' is an opinion within the discipline. 'The contractor was negligent' is not — that is a legal conclusion wearing technical language. An expert who overstates is exposed on exactly that point, and the overstatement damages the parts of the opinion that were sound.

This is why the useful engagement usually happens well before an expert report is required. A validation review conducted while positions are still open tells a party what its records actually support, which is the information needed to decide whether to defend, settle or correct. Nothing on this page is legal advice; it describes the technical dimension only. Where an independent opinion on the method itself is needed, outside Level III support is a separate and deliberately narrow engagement.

The report that is technically right and evidentially weak

Most disputed reports are not wrong. They are thin. The interpretation was competent, the technician was skilled, the call was correct, and the record captures none of the surrounding facts that would let anyone else verify it years later. That report loses ground it never needed to lose, and it loses it to opponents who never had to engage with the technical question at all.

The signature of a thin report is that everything specific to the day is missing. No surface condition. No temperature. No couplant. No calibration block identity. No instrument serial number. No procedure revision. No statement of coverage. Every one of those was a fact the technician knew and did not write down, and every one of them is a question an opposing expert will ask and nobody will now be able to answer.

The fix is a report template mapped against the records clause of the invoked code rather than against habit, and a procedure that instructs the technician to capture each field at acquisition rather than from memory at the end of the shift. That mapping is ordinary work and it belongs in technical procedure development. It converts a record that satisfies a client this month into one that survives examination in four years.

Preserving records once a dispute is foreseeable

The duty to preserve arises when litigation is reasonably anticipated, not when proceedings are issued. From that point, routine destruction under a retention schedule has to stop for anything relevant. In US federal practice the loss of electronically stored information that should have been preserved is dealt with under Rule 37(e), and the most severe measures — including an instruction that the lost information was unfavourable to the party that lost it — turn on a finding that the party acted with intent to deprive another of its use.

For inspection organisations this bites on data destroyed for entirely ordinary reasons. Acquisition files purged when a laptop is reimaged. Film disposed of under a retention policy. An email archive rotated on schedule. A phone replaced. None of it is misconduct and all of it is damage. When a dispute becomes foreseeable, the preservation instruction has to reach the technicians and the IT function on the same day, not after the first request for documents arrives.

Preservation is broader than the report. It covers raw acquisition files, field notes, photographs, calibration certificates current at the time, procedure revisions, personnel qualification files as they stood, transmittals, and the correspondence surrounding the examination. Retention decisions taken calmly in advance are far cheaper than preservation decisions taken under a hold notice, which is one of the practical reasons to fix retention before anyone needs it.

What a validation review produces for a dispute

A records-focused validation review produces a written finding on three questions: what the record establishes, what it does not establish, and where it is vulnerable. It works from the same material an opposing expert would receive, in the same order, and it reports the weaknesses on your own side first, because those are the ones you need to know about and the ones nobody internal will volunteer.

It is deliberately capable of concluding that the record is sound. A finding that the examination was performed to the invoked code, by personnel qualified on the date, on equipment with a traceable calibration chain, with coverage documented and raw data retained, is a valuable output in its own right. It converts an anxious position into a defensible one and lets a party hold a line it would otherwise have paid to settle.

The output is written so a non-specialist can follow the reasoning and a specialist cannot fault the basis. Where a deficiency exists it is stated with its scope bounded — which welds, which dates, which shift — rather than described in general terms that invite the widest possible reading. If you are holding an inspection record that is about to be read by people outside the industry, commission that review early, while there is still a decision to inform.

Request a consultation