What a rejected NDT report actually means, and what to do in the first week

The first 48 hours: preserve, do not react

Before you write anything, freeze the evidence. Copy the raw acquisition data — film, A-scan files, PAUT or TFM data sets, photographs, field sheets — to a location nobody edits. Export the report exactly as issued, with its transmittal record and date. Pull the procedure revision that was in force on the examination date, not the current one. Pull the technician's certification file as it stood on that date. This takes an afternoon, and it is the difference between a defensible position and a reconstruction.

Then stop. The rejection notice is a claim, not a finding. Nothing in it obliges you to agree, to re-inspect, or to explain yourself within the hour. A short acknowledgement — received, under technical review, response by a stated date — buys the days you need and costs nothing. Contractors lose more ground in the first reply than in the entire dispute that follows, because the first reply is written by whoever happened to open the email.

The reason to preserve before responding is that some rejections escalate. A report that is merely rejected today becomes a claim document in six months if the weld it covers fails, or if payment is withheld and you pursue it. What you can prove about that examination is fixed at the moment you preserve the data and only degrades afterwards, which is why the same discipline applies whether the record ends up in a dispute file or a routine re-issue.

Six causes of rejection, and why they are not interchangeable

Rejections arrive in one envelope and come from six different places. A genuine technical deficiency means the examination itself was wrong: an indication missed, a call made outside the technique's capability, a scan plan that never covered the volume the report claims. A documentation gap means the examination was right and the record of it is incomplete. Those two look identical on the rejection notice and have nothing else in common.

The third is an acceptance-criteria disagreement, where both parties read the data the same way and apply different criteria to it. The fourth is procedural: the technique was sound but the procedure was never formally approved by the client, or the revision cited on the report is not the revision that was approved. The fifth is a challenge to personnel qualification — level, method, expiry, or a scope limitation printed on the certificate. The sixth is a commercial position wearing technical clothes.

Treating these as one problem produces the standard contractor error: offering a free re-inspection to close a dispute that a corrected header block would have closed. It also produces the opposite error, arguing code for three weeks over a report that genuinely missed a planar flaw. Sorting first is not delay. It is the step that tells you which of six answers to give, and the sorting itself takes hours.

Reading the rejection against four documents, in order

Read the contract first, not the code. The contract states which code and edition is invoked, which acceptance criteria apply, whether client procedure approval is a condition precedent, and which certification scheme personnel must hold. Rejections dissolve at this step when the client is applying a criterion from a code the contract never invoked, or demanding an approval the contract never required. Establishing that costs an hour and settles the matter without a technical argument.

Read the referencing code second, in the edition the contract invoked. Editions move acceptance criteria, records requirements and qualification rules. ASME BPVC Section V sets general records requirements in Article 1 and adds method-specific items in each examination article — the ultrasonic article requires the instrument identification, search unit identification, beam angles, couplant and search unit cables to be recorded, and a report missing those is deficient however sound the interpretation was.

Read the approved procedure third, at the revision in force on the examination date, and the report itself fourth. By the time you reach the report you already know which of the six causes you are testing, and the read becomes a targeted check rather than a general worry. Reversing the order — starting with the report and working outward — reliably produces the wrong answer at higher cost, because the report is the only one of the four documents you already know by heart.

Genuine technical deficiency: what an independent re-read actually checks

An independent re-read does not start with the conclusion. It starts with whether the data could have supported any conclusion. Was the volume covered? Does the scan plan account for the geometry, the access and the flaw orientations credible for that joint? Was sensitivity set so the smallest reportable reflector at the far surface still produced a usable signal? A report can be internally consistent and still rest on data that was never capable of finding what the client says was missed.

Then the re-read examines the calls. For radiography that means the image itself: transmitted density within the required range, the correct IQI present with the essential hole or wire resolved, and density variation across the area of interest inside the code's limits. ASME Section V requires a minimum transmitted density of 1.8 for single-film viewing with an X-ray source and 2.0 with a gamma source, 1.3 per film for composite viewing, and 4.0 maximum. A film outside that range is not evidence of anything.

For ultrasonics the re-read goes to the A-scan or the encoded data set where one exists. Calibration confirmation is checkable against the code: system calibration is to be confirmed before and at the completion of each examination or series of similar examinations, and when examination personnel are changed. A missing end-of-shift calibration check does not prove the data was wrong. It means nobody can now prove it was right, and in a rejection that distinction decides who carries the cost.

Documentation gaps: the cheapest rejection to close

A documentation rejection costs a crew mobilisation when it is misdiagnosed and an hour of typing when it is not. The recurring omissions are unglamorous: instrument model and serial, couplant, surface condition and preparation, temperature where the method is temperature-sensitive, procedure revision, calibration block identity, technician level and certification number, the Level III review signature, and a clear statement of the extent examined against the extent required. None of them touch the interpretation and all of them make a report rejectable.

Close them by correcting the document and re-transmitting under a new revision number with a note stating what changed and why. Do not re-issue silently under the same number. That single habit converts a trivial administrative correction into an integrity question, and it is the first thing anyone looks for when a report set is examined years later. Every re-issue should leave a trail that makes the change visible rather than one that conceals it.

The structural fix sits upstream. Reports that keep getting rejected for content come from a template that was never mapped against the records clause of the invoked code and the client's own data sheet. We cover the recurring pattern in what clients actually reject in NDT report formats, and the same mapping belongs inside technical procedure development — a procedure that does not tell the technician what to record produces reports that omit it.

Acceptance-criteria disagreement: whose code, which edition

This is the rejection where both sides are competent and both are looking at the same indication. The client's engineer rejects a cluster you accepted. Neither of you misread the data. You applied different criteria, or the same criteria from different editions, or a workmanship standard where a fitness-for-service assessment was contractually available. Nothing about the examination is in question, and a re-inspection would change nothing.

Establish four things before arguing: which code and edition the contract invoked, which clause each side is applying, whether the criterion is a workmanship or a fitness-for-purpose criterion, and whether the client's engineer has authority to impose a criterion beyond the contract. Structural and pressure work diverge sharply here, and edition drift shows up in the citation itself — AWS D1.1:2020 moved its inspection requirements into Clause 8, so a rejection citing the old Clause 6 numbering tells you the objector is reading an earlier edition than the one you were contracted to.

Never concede a criterion informally. An acceptance criterion conceded on one weld becomes the criterion for the remaining scope, and there may be hundreds of welds behind it. Where the disagreement is genuine, escalate it as a written technical query between the Level III of record and the client's engineering authority, so it is decided as a code question, which is what it is. Where you do not hold a Level III in that method, outside Level III support can carry and sign the position.

Procedure approval and personnel qualification: the two binary attacks

These two rejections are attractive to a client because neither requires a technical argument. Either the procedure was approved or it was not. Either the technician held certification for that method, level and scope on that date or he did not. They are documentary, they are binary, and they are the two grounds most likely to be raised when a client wants a reason and does not want a debate about ultrasonics.

A procedure-approval rejection is usually recoverable. Check the transmittal record, check whether the contract made approval a condition precedent to examination, and check whether the revision cited on the report differs from the approved revision in substance or only in a header. A revision that changed a distribution list does not invalidate an examination. A revision that changed the scan plan, the sensitivity setting or the coverage does, and that difference is the whole question.

Qualification challenges are decided entirely by the file. Under an employer-based scheme the written practice governs, and SNT-TC-1A places the employer's Level III behind it — so your own document is the standard you are judged against. ISO 9712 certification comes from an independent body and travels with the individual between employers. Contracts that specify one scheme and receive records under the other generate rejections that are entirely real. Certificates current today but lapsed on the examination date are the version that hurts most, and there is no argument against it.

The commercial rejection wearing technical clothes

Some rejections are not about the report. They arrive when a schedule has slipped, when an invoice is due, when the client's own scope has changed, or when someone upstream needs a reason for a delay they caused. They are recognisable by what they lack rather than what they say: no specific component, no specific clause, no specific criterion, and no named technical author on the objection.

The response is a written request for particulars, phrased neutrally and addressed to the technical authority rather than the commercial one. Ask which report number, which weld or component, which clause of which edition, and which criterion is said to be unmet. State that the technical review will be completed within a defined period once particulars are received. This is not aggressive. It is the ordinary handling of a technical query, and it moves the burden without accusing anybody of anything.

Two outcomes follow, and both are good. Particulars arrive, and you now have a real rejection to sort into one of the other five categories. Or particulars do not arrive, and the correspondence shows a rejection that was never substantiated — precisely the record you want if the matter later moves to a payment claim or an adjudication. Either way you have converted open-ended commercial pressure into a bounded technical question with a date on it.

Writing the response: structure that concedes nothing and hides nothing

A good technical response has four parts and no adjectives. First, what was examined and under what: scope, procedure revision, code and edition, personnel and certification, dates. Second, the objection exactly as the client stated it, quoted rather than paraphrased. Third, the finding of the review, stated as a conclusion with the evidence it rests on attached. Fourth, the proposed disposition — what you will do, and what you will not.

Write it so a non-specialist can follow it and a specialist cannot fault it. That means citing the clause and edition rather than asserting compliance, attaching the evidence rather than referring to it, and stating openly any limitation of the original examination. A response that volunteers a limitation the client had not spotted is stronger, not weaker, because it demonstrates that a real review took place rather than a defensive one.

Where the review finds the client was right, use the same structure. Identify the deficiency, bound the affected scope precisely — which welds, which components, which dates, which shift — and propose corrective action with its verification. Validation that always agreed with whoever commissioned it would be worthless, and the value of the exercise is a defensible answer either way. A contractor who arrives with a self-identified, bounded problem and a plan keeps the relationship. One who is shown it by the client's consultant does not.

What re-inspection actually costs you beyond the crew day

The visible cost of a conceded re-inspection is a crew, an access package and a schedule slot. The invisible costs are larger. A re-inspection performed as a gesture is a documented admission that the first examination was inadequate, and it sits in the project record where the next dispute will find it. It also sets the precedent for every subsequent rejection on that contract, and there will be subsequent rejections.

It can make matters worse technically. Conditions change: surfaces corrode, coatings go on, components move, welds are ground. A re-inspection months later may not reproduce the original result even when the original result was correct, and the discrepancy will be read against you rather than against the interval. The one honest reason to re-inspect is that the review found the original examination could not have supported the conclusion it reached.

Which is the entire argument for sorting the rejection before answering it. Four of the six causes close without anyone returning to site, and two of those close the same week. If you want an independent technical read of a rejected report before you commit to a position, that is what a validation review is for — the data and the record examined against the invoked code, with a written finding you can put in front of your client whichever way it lands.

Request a consultation