Third-party review versus re-inspection: the honest limits of each
A document review establishes whether the examination described was adequate for the flaw type expected and whether the results were correctly interpreted against the stated criteria. It cannot establish that the examination was performed as recorded, or what condition the material is in today. Re-inspection answers those two questions and nothing about the original report's defensibility. Most disputes need the review first.
The two purchases are priced very differently and that asymmetry drives the decision. A document review costs reviewer hours against a fixed evidence pack: nothing mobilises, no access is built, no production stops. Re-inspection costs mobilisation, permits, scaffold, insulation removal and reinstatement, coating removal and recoat, surface preparation, and where the asset must come down, production loss that dwarfs the entire inspection budget. The examination itself is often the smallest line on the invoice. That makes the review the cheap instrument that tells you whether the expensive one is necessary and where to point it. Buying them in the other order spends the large budget before the small one has said where the large one is needed. Where both are required, the sequence is review, target, re-inspect — and the review's output should be written as a re-inspection specification rather than as an opinion.
Source: ASME BPVC Section V Articles 1, 2, 4, 6 and 7 record and technique requirements; ASNT SNT-TC-1A (2020) certification currency and vision examination intervals; Atlantis NDT Level III validation and field inspection practice.
| The question blocking your decision | Document review | Re-inspection | Cheapest path that actually answers it |
|---|---|---|---|
| Did the contractor deliver what the contract required? | Answers it fully from the record | Answers a different question entirely | Review only |
| Were the personnel and equipment qualified on the day? | Answers it from dated certification and calibration records | Cannot answer it — the day has passed | Review only |
| Was the reject call correct on the recorded data? | Answers it where the raw data, radiographs or scan files survive | Produces new data, not a verdict on the old call | Review, with the raw data written into scope |
| Did the examination actually take place as recorded? | Finds correlates and implausibilities; cannot prove it | Cannot prove the past either, but tests the material now | Review to select a sample, then targeted re-inspection |
| Is the component fit to run for another cycle? | Cannot answer it at any price | Answers it | Re-inspection, scoped by the review |
| Was the volume the code required actually covered? | Answers whether it was recorded as covered, and where it was not | Answers what is there now in the volume with no data | Review first; re-inspect only the volume that has no data |
Two different questions, priced very differently
The buyer usually thinks they are choosing between two ways of getting the same answer. They are not. A document review answers whether the examination that is recorded was adequate and whether it was correctly interpreted. Re-inspection answers what is in this material now. Both are legitimate purchases, they cost different amounts, they take different amounts of time, and neither one substitutes for the other in either direction.
Getting this wrong is expensive both ways. Buying re-inspection when the dispute is about whether a contractor's report complied with the contract produces fresh data and settles nothing about the original invoice. Buying a review when the real question is whether a vessel is fit to run for another five years produces a competent opinion about paperwork and no information whatever about the condition of the steel it describes.
The decision is usually made under time pressure by somebody who is not an NDT specialist — a project manager, a claims handler, a lawyer, a buyer working through due diligence. The framing that works is simple: write down the decision you have to make, then ask which of the two questions above is actually blocking it. The answer falls out. Where the decision needs both, it needs them in a specific order, covered further down this page.
What a document review can establish
A review establishes whether the examination as recorded was capable of finding what the component was at risk of. That is a technique judgement: the method chosen against the expected flaw type and orientation, the coverage against the volume the code required, the sensitivity against the thickness, the criteria against the contract. All of it is testable from the paper by somebody holding the right Level III method scope for each method in the package.
It establishes whether the people and equipment were qualified on the day. Certification currency, vision records, written practice compliance, calibration certificates with a traceability chain — every one of these is a document with a date on it, and every one is checkable against the examination date without going anywhere near the asset. This class of finding is the most common and the hardest to argue with, because the dates either line up or they do not.
And where the raw data survives — radiographs, A-scans, encoded scan files, photographs — a review can re-read it. That converts the exercise from a compliance check into a genuine second interpretation: the same evidence, evaluated independently against the same criteria. When a validation has the data, the finding on interpretation carries the same weight as the finding on documentation, and a disagreement about a call becomes resolvable on the record. The full element list sits on what makes an NDT report defensible.
What a document review cannot establish
It cannot establish that the examination was actually performed as recorded. A report can name the right procedure, the right calibration block, the right couplant and the right coverage, and describe an examination that never took place, or one that covered half the length claimed. Nothing in a document review detects that on its own. This is the hard limit of the service, and any validator who implies otherwise is overselling what document work can do.
It cannot establish the current condition of the material. Even a perfect report describes a moment that has already passed. Corrosion has progressed since, a crack has grown, an operating excursion has happened. For a fitness-for-service decision or a remaining-life estimate, the age of the data matters more than its quality, and a validation confirming that a five-year-old report was excellent says precisely nothing about the component today.
It cannot recover evidence that never existed. Where a technique sheet was never written, a lighting check never recorded, or raw data never retained, the review reports the absence. Absence is a finding with real consequences — it is frequently enough to defeat a report in a dispute — but it is not the same as knowing what the examination would have shown. Honest scoping states these three limits before the engagement starts, not in the report's caveats.
The pencil-whipped report problem, stated plainly
The scenario that worries owners most is the report written without the examination being done, or done properly. Document review cannot detect this directly. What it can do is find the correlates: technique parameters that could not have produced the results claimed, timings that do not fit the coverage recorded, indications absent from a component type that reliably shows them, calibration records that do not move across a week of continuous scanning.
None of those is proof. Each of them is a reason to re-inspect a defined sample rather than the whole scope, and that is the practical value of running the review first: it converts a diffuse suspicion into a short list of specific welds or components where re-examination will actually resolve something. Re-inspecting everything answers the question too, at a cost nobody wants to carry and few can justify to a board.
Where a pattern spans a contractor's whole output rather than one job, the instrument shifts from validation to a programme audit and gap assessment, which examines whether the system could have produced compliant work at all. Findings there — no written practice, no procedure demonstrations, uncontrolled revisions — change the weight given to every report that contractor issued, including the ones nobody has questioned yet.
Cost and time: what actually drives each
A document review's cost is reviewer hours against a fixed evidence pack. Nothing mobilises, no access is built, no production stops, and the work runs in parallel with everything else on the project. Its timeline is set by how fast the records can be collected — usually the slowest part of the engagement, and usually the client's side of the work rather than the reviewer's. Reviewers can be reading while the remaining records are still arriving.
Re-inspection's cost is dominated by everything that is not the examination. Mobilisation and travel, permits, scaffold, insulation removal and reinstatement, coating removal and recoat, confined-space entry, cleaning to a scannable surface, and where the asset must come down, production loss that exceeds the entire inspection budget by orders of magnitude. The NDT itself is frequently the smallest line on the invoice by a wide margin.
That asymmetry is the whole reason the review exists. It is the cheap instrument that tells you whether the expensive instrument is necessary and where to point it. Buying them in the other order — re-inspect first, review later — spends the large budget before the small one has told you where the large one was needed. It also survives being questioned later, because the field scope has a written rationale standing behind it.
When re-inspection is unavoidable
Re-inspection is unavoidable when the decision depends on present condition. Fitness-for-service assessment, remaining-life calculation, a return-to-service decision after an operating excursion, or a run-or-repair call on a component with active degradation all need current data. No review of a historical report substitutes for that, however good the report was. The report can tell you where to look and what to expect; it cannot tell you what is there now.
It is also unavoidable where the review finds the technique was incapable. If a weld prone to lack of side-wall fusion was examined with a technique that could not orient to that flaw, the correct finding is that the volume was never effectively examined — and the only remedy is to examine it. The same applies where coverage fell short of what the code required: the missing volume has no data behind it, and no amount of document review creates data that was never taken.
And it is unavoidable where the evidence pack cannot be produced at all. A report with no procedure, no certification records and no calibration certificates cannot be validated into acceptability. Where the asset matters and the paper is gone, re-inspection is the only route to a defensible position, and the review's contribution is to have established that in days rather than after a month of document chasing across a dissolved contractor.
The hybrid: review first, then targeted re-inspection
The pattern that works on most real engagements is sequential. Review the reports, produce a verdict on each, and where the verdict is deficient in a way that changes the disposition — or cannot be established from the record at all — generate a targeted re-inspection scope naming specific components, specific methods and specific coverage. The review has then paid for itself by shrinking the field work down to what field work can actually settle.
Sampling is where the two stages connect. A review that finds a systemic technique defect across a contractor's output justifies re-inspecting a sample sized to test that finding rather than a sample sized by convention. If the sample confirms the defect, the scope expands with evidence behind it. If the sample clears, the documentary findings can be closed as documentation-only, and the owner has bought a defensible position for a fraction of a full re-inspection.
This sequence also survives scrutiny better than either half alone. An owner who re-inspected on a hunch has an expensive result and no rationale for the scope. An owner who reviewed and stopped has a rationale and no current data. An owner who reviewed, targeted, then re-inspected can show why each component was chosen — which is precisely the reasoning an auditor will ask for when the file is opened later.
Writing the scope so the two stages actually connect
Write the review scope so its output is a re-inspection specification, not an essay. That means requiring the reviewer to state, per report, the verdict, the specific gap, whether the gap is documentary or technical, and — where technical — the component, volume, method and coverage that would close it. A review that ends at the word "inadequate" has left the buyer to do the hardest part of the job themselves, usually without the qualification to do it.
Fix the acceptance criteria before either stage starts. Both the review and any re-inspection must be run against the code and edition the contract invoked, not against whatever is current today. Where the original contract is ambiguous on this point — many are — resolving it is the first deliverable, because a review run against the wrong criteria produces confident findings that a competent counterparty will dismantle in an afternoon.
Keep independence intact across both stages. The party that reviews the reports should not be the party that re-inspects if the re-inspection result will be used to defend the review's own finding, and where one firm does both, the roles and personnel are separated and the report states it plainly. Scope questions, and whether a review or a re-inspection fits the decision in front of you, start here — or with an outside Level III engagement where the programme rather than the report is the problem.
Can a document review prove the inspection was actually performed?
No, and no honest reviewer claims otherwise. A review can find indicators that the recorded examination is implausible — timings inconsistent with the coverage claimed, calibration records that never change across a week, parameters that could not have produced the results shown. Those indicators justify a targeted re-inspection sample. They are not proof, and the review report says so explicitly rather than implying more.
Which is faster, a document review or a re-inspection?
The review, by a wide margin, and the gap widens with asset complexity. A review needs the records assembled and reviewer hours; nothing mobilises and nothing stops. Re-inspection needs access, permits, scaffold, insulation and coating removal, surface preparation, and often a shutdown window months away. Collecting the records is usually the slowest step in a review, and it sits on the client's side.
If the review finds problems, do we have to re-inspect everything?
No. The point of reviewing first is to convert a general suspicion into a specific list. Documentary-only deficiencies need records, not scanning. Technical deficiencies need re-examination of the volume that was never effectively examined, which is a named subset rather than the whole scope. Sampling to test a systemic finding is legitimate and expands only where the sample confirms the finding.
Can re-inspection settle a payment dispute about the original report?
Not on its own. New data tells you the condition today; it does not establish whether the contractor delivered what the contract required. A weld that is sound now proves nothing about whether the original examination complied, and a flaw found now may have grown since the work was done. Contract compliance is a documents question, and documents questions are settled by review.
What if the supporting records were never kept at all?
Then the report cannot be validated into acceptability, and the review's value is establishing that quickly and cheaply. Where the asset matters, re-inspection becomes the route to a defensible position. Where it does not, the finding itself — that a whole class of work has no supporting evidence behind it — is usually enough to resolve the commercial question in the buyer's favour.
Should the same firm do both the review and the re-inspection?
Only with the roles separated and the separation stated on the face of both reports. A firm whose re-inspection result would vindicate its own review finding has an interest in the outcome, and a counterparty will say so in the first hour. Where the re-inspection is uncontested field work under a scope somebody else wrote, one firm is efficient. Where the finding will be contested, use two.