The NDT Written Practice That Has to Survive a Nuclear Audit
A nuclear written practice is the employer's own controlled document that defines how NDT personnel are trained, examined, certified and recertified. Under 10 CFR 50 Appendix B and ASME Section XI it must name the edition of SNT-TC-1A or CP-189 it follows, identify the Level III who owns it, and match every certificate in the file.
Nuclear work makes the written practice unusually exposed. 10 CFR 50 Appendix B, Criterion IX treats nondestructive examination as a special process, so the qualification of the people performing it is a licensing-adjacent control rather than an internal preference. ASME Section XI IWA-2300 then fixes which edition of SNT-TC-1A applies through the Code edition the plant's inservice inspection program is committed to, which is rarely the newest edition ASNT has published. Section XI Appendix VIII adds performance demonstration for ultrasonic examination of specific components, and that qualification is separate from, and additional to, certification under the practice. Utilities and their audit consortia sample certification files against the practice rather than against the standard, so a clause the practice never wrote is a clause you cannot claim credit for. Most findings are not technical failures. They are a document that stopped describing what the organization actually does.
Source: Written against ASNT SNT-TC-1A and ANSI/ASNT CP-189 (personnel qualification and certification); 10 CFR 50 Appendix B, Criteria II, V, VII, IX, XVII and XVIII; 10 CFR 50.55a; 10 CFR Part 21; ASME Boiler and Pressure Vessel Code Section XI (IWA-2300, Appendices VII and VIII) and Section V Article 1; and ASME NQA-1 Requirement 2 (indoctrination and training) and Requirement 18 (audits).
| Practice clause | What the document must state | What the auditor pulls | Recurring finding |
|---|---|---|---|
| Reference standard and edition | The exact standard and edition year, and the Code edition that invokes it | The certificate, the practice cover page and the ISI program commitment | Certificate cites a newer SNT-TC-1A edition than the Code edition permits |
| Level III of record | Name, method scope, certification basis, and who signs in their absence | The Level III's own certificate and the signature on a sampled technician file | Signatures on file from a Level III who left before the examination date |
| Training hours and education substitution | Hours per method per level, and exactly which substitutions are allowed | Course rosters, instructor qualification, sign-in sheets | Hours claimed with no attendance record, or an instructor with no stated qualification |
| Experience credit | How hours are logged, supervised and attributed when methods overlap | Experience logs laid against timesheets and job tickets | The same clock hours claimed in full against two or three methods |
| Examination content and grading | General, specific and practical composition, pass marks, question bank control | Answer keys, question bank custody, graded papers | Answer keys stored where trainees can reach them, and no evidence of bank rotation |
| Vision examination | Near-vision acuity and color differentiation requirements, frequency, examiner | Signed eye examination records against certification dates | Vision record dated after the certification it was supposed to support |
| Recertification and interrupted service | Interval per level, the evidence accepted, and what breaks continuity | Recertification packets and any layoff or reassignment records | Recertification by continued satisfactory performance with nothing on file evidencing it |
The document is what is being audited, not the certificate
Every NDT certificate in a nuclear supply chain is downstream of one controlled document: the employer's written practice. SNT-TC-1A is a recommended practice rather than a specification. It tells the employer what to decide, not what to do, and the latitude it leaves is deliberate. CP-189 closes much of that latitude and behaves like a requirement standard. Either way the certificate is only as defensible as the practice behind it, and an auditor who finds a well-kept certification file sitting behind a vague practice will write the finding against the practice.
That ordering matters more in nuclear than anywhere else, because 10 CFR 50 Appendix B, Criterion IX treats nondestructive examination as a special process. A special process is one whose result cannot be fully verified by later inspection of the product, so control has to be exercised through qualified personnel and qualified procedures instead. The written practice is where qualified personnel is defined. If it is not defined precisely there, nothing downstream can carry the weight.
A useful internal test: could a stranger pick up your practice and reproduce a certification decision you made three years ago? Reference edition, education substitution, training hours, examination composition and grading, vision requirements, experience credit, recertification trigger. If any of those live only in the Level III's judgement, the practice is incomplete no matter how sound the judgement actually was.
The clauses a nuclear practice cannot leave blank
Start with scope and reference. The practice must name the standard and edition it follows and, in nuclear, the Code edition that invoked it. It must list the methods and, where relevant, the techniques and limitations under which people will be certified. It must name the Level III of record by name and by method, state that person's own certification basis, and say who exercises the authority when they are unavailable. A practice that describes the Level III only as a role, with no name and no delegation clause, cannot be sampled.
Then the mechanics. Training hours by method and level, and precisely which education or experience substitutions are permitted, written as rules rather than as guidance. Examination composition: general, specific and practical, the number of questions, the pass mark for each part and for the composite, and how the practical is scored on real specimens with documented flaws. Question bank custody and rotation. Who may administer the examination, and the prohibition on that person having trained the candidate where a client demands that separation.
Finally the parts that get skipped. Vision examination content, frequency and who is qualified to administer it. Experience logging and supervision. Recertification interval by level and the evidence that supports it. Interrupted service: what counts as a break, how long a break voids continuity, and what re-entry requires. Technique-limited certification, where the certificate must carry the restriction on its face rather than relying on a memo. Records retention and the retention period, which in nuclear is set by contract and by Criterion XVII rather than by convenience.
Who owns the practice, and what a contracted Level III can actually do
Certification is employer-based. That is the single most misunderstood point in the whole system, and it is misunderstood in both directions. Technicians assume a certificate travels with them between employers; it does not. Employers assume that hiring an outside agency Level III transfers the liability; it does not do that either. The employer certifies, the employer keeps the records, and the employer is the party audited.
What a contracted Level III legitimately provides is technical authority: writing and approving the practice and the procedures, establishing and maintaining examinations, administering and grading them, conducting internal reviews, and representing the technical position during a client audit. In nuclear that arrangement is normal and is understood by every utility audit team. What they will probe is the contract behind it: response time commitments, custody of examination material, access to records, separation from training delivery, and above all what happens when that individual is no longer available.
The succession clause is the one most practices omit. If the Level III of record leaves, every procedure they approved remains valid, but nothing new can be approved, no examinations can be graded, and no certifications can be issued until the practice is revised to name a replacement. Organizations discover this in the middle of an outage. Naming an alternate in the practice from the beginning costs nothing and prevents a stoppage.
What the nuclear auditor samples, in the order they sample it
The pattern is consistent whether the auditor is a utility, a joint audit team acting for several utilities, or a client's own quality organization working to ASME NQA-1 Requirement 18 and Appendix B Criterion XVIII. They ask for the practice, check its revision status and approval signature, then ask for a list of currently certified personnel. From that list they pick a small number of files, usually including the newest certification, one recertification, and one person certified in more methods than average.
Then they trace backwards. Every claim on the certificate has to be provable from a record: training attendance, instructor qualification, examination papers with the grader's signature, the practical specimen and its documented flaw sheet, the experience log, the vision examination, and the Level III's own certification in the method being granted. Where a substitution was applied, the practice clause permitting it has to be shown. Where the certificate is technique-limited, the limitation has to appear on the certificate itself and not only in a procedure.
The Level III's own file gets the same treatment, and it is frequently the weakest one in the drawer. It is common to find a meticulously maintained set of Level II files behind a Level III whose method certifications, basic examination evidence and vision records have not been touched in years. Auditors know this and go there early.
Where ASME Section XI adds requirements the practice never mentions
Section XI IWA-2300 is the hinge between the certification system and the inservice inspection program. It invokes SNT-TC-1A with a fixed edition and adds requirements of its own, including annual vision examination and constraints on how certification is granted and maintained. A practice written for a general industrial client and then reused for nuclear work will usually be silent on all of it, because nothing in the base standard prompts the author to ask.
Appendix VIII sits alongside the practice, not inside it. Performance demonstration qualifies procedure, equipment and personnel together, on specimens containing flaws whose locations the candidate does not know, administered through a qualification body. It applies to defined component and examination categories, and passing it does not certify anyone under your practice. Not holding it does not invalidate the certificate either. The two systems answer different questions, and the practice should state which is which so that a field supervisor never treats one as the other.
There is a procurement dimension as well. NDT services bought into a safety-related scope are usually procured as a basic component or dedicated commercially, which means the critical characteristics of your certification system become verifiable attributes inside someone else's dedication package. That is a strong reason to write the practice so its clauses can be pointed at directly rather than paraphrased by a third party.
The experience arithmetic that fails more audits than anything else
Experience hours are where certification files most often come apart, and the mechanism is arithmetic rather than dishonesty. A technician works a two thousand hour year. Across that year they perform PT, MT and UT. The log shows two thousand hours of PT experience, two thousand of MT and two thousand of UT, because each entry was recorded against the day worked rather than the time actually spent in the method. An auditor who lays the log alongside timesheets and job tickets sees the same clock hours claimed three times.
The correct treatment is to apportion. Hours credited to a method are hours actually spent performing or directly supporting that method under qualified supervision, and the sum across methods cannot exceed hours worked. The practice has to state that rule explicitly, state who supervises, and state how the log is verified before certification is granted. Where a candidate genuinely falls short in one method, the honest answer is a limited certification or a delayed one.
A related trap is minimum hours per method regardless of the total. A technician with six thousand hours in UT and twenty hours in PT is not close to a PT Level II simply because the aggregate looks impressive. The minimums are per method, and in nuclear they will be checked per method, on paper, against dates.
The drift: how a compliant practice quietly stops being one
Drift is rarely a decision. It is a sequence of small, individually reasonable actions. A client asks for a technique the practice does not list, and it gets performed under the nearest method heading. A new instrument arrives and the technique sheet is updated but the procedure is not. A technician is recertified on the strength of continued satisfactory performance because everyone knows the work is good, and nothing is written down evidencing it. A revision is issued and certificates already in circulation still quote the superseded clause.
The second driver is turnover in the quality function rather than in the technical one. The person who understood why the practice was written the way it was leaves, and the successor treats the document as a template to be kept rather than a control to be exercised. Reviews continue to be signed, because signing is easy, and nothing is compared against what the organization actually does.
The counter is a documented annual review with a defined comparison set: current personnel list against current certificates, current certificates against the current practice revision, current procedures against the practice, and the last twelve months of client and internal findings against the clauses they touched. That review takes about a day and it is the cheapest audit insurance in the building.
Revision control and the day the practice changes
A revision to the practice is a configuration change, and it has downstream effects that need to be worked out before the revision is issued rather than after. If training hours increase, does the change apply prospectively only, and is that stated? If the reference edition moves, are existing certificates reissued, annotated, or left to expire under the old edition, and does the file explain which? If a method or technique is added, who is qualified to examine in it on day one?
Appendix B, Criterion V and Criterion XVII expect instructions and records to be controlled, which in practice means every issued revision is identifiable, superseded copies are withdrawn or marked, and the distribution is recorded. Uncontrolled copies riding in a truck or sitting in a field office are a routine finding and an entirely avoidable one.
The cleanest approach is to attach a transition statement to every revision: what changed, why, who is affected, what actions close the gap, and by when. It converts a document change into a managed change, and it gives the next auditor the answer before they have had to ask the question.
When a certification gap becomes a Part 21 question
This is the consequence that makes nuclear different in kind rather than in degree. If it emerges that a person who examined a safety-related component was not validly certified under the practice at the time of the examination, the question is no longer a quality finding. It becomes an evaluation of whether a defect or a failure to comply exists that could create a substantial safety hazard, with the reporting obligations of 10 CFR Part 21 attached to it.
The practical consequences arrive quickly. The examinations performed by that individual across the affected period have to be identified, their significance assessed, and in many cases re-examination scheduled inside an outage window that was never planned for it. The cost of that dwarfs the cost of the certification control that would have prevented it.
This is the reason to write the practice for reconstruction rather than for approval. A practice that reads well and passes its own review is worth less than one where every certificate can be rebuilt from records without depending on anybody's memory. Atlantis develops and reviews written practices to that standard, supplies the Level III of record where an employer needs contracted technical authority, and represents the technical position during client and joint audits. Consultations and scoping discussions are arranged on request through info@atlantisndt.com.
Which edition of SNT-TC-1A does a nuclear written practice have to follow?
Not necessarily the newest one. ASME Section XI IWA-2300 invokes a specific edition of SNT-TC-1A, and the edition that binds you is the one carried by the Code edition and addenda your inservice inspection program is committed to under 10 CFR 50.55a. A practice that quietly moved to the latest ASNT revision can end up out of step with the plant it serves, and that mismatch shows up on the certificate before it shows up anywhere else.
Can an outside agency Level III certify our technicians for us?
An outside agency Level III can be named in your practice, can write and approve procedures, and can prepare, administer and grade examinations. What they cannot do is become the certifying party. Certification under SNT-TC-1A and CP-189 is employer-based: your company signs, your company owns the records, and your company answers the audit. Contracted authority is normal in nuclear supply chains, but the contract has to define examination security, record custody, response time and succession.
Does certification under the written practice satisfy ASME Section XI Appendix VIII?
No. Appendix VIII performance demonstration qualifies the procedure, the equipment and the personnel as a system, run on flawed specimens through an administering body, and it applies to specified components and examination volumes. A technician can hold a valid UT Level II certificate under your practice and still be unqualified to perform an Appendix VIII scope. The practice should say so explicitly rather than leaving the two to be confused in the field.
Is CP-189 stricter than SNT-TC-1A for nuclear supply work?
In the ways that matter to an auditor, yes. SNT-TC-1A is a recommended practice that tells the employer what to decide; CP-189 behaves as a requirement standard, removes much of that latitude, and requires the NDT Level III to hold ASNT Level III certification in the applicable method. Some clients specify CP-189 outright. Where they do, a practice written in SNT-TC-1A language will not close the gap by reference alone.
What makes a written practice drift out of compliance without anyone noticing?
Almost always a revision that was never pushed downstream. The practice is revised, approved and filed, but certificates issued before the revision still cite the old clause, the examination bank was never rebuilt to the new content, and nobody re-examined the population affected by a changed training-hour requirement. Nothing failed technically. The document simply stopped describing the organization, and the file no longer reconstructs the decision that was made.
How often should the written practice itself be reviewed?
Annually is the defensible minimum for nuclear supply, plus a trigger review whenever the Code edition changes, a client imposes CP-189 or a supplement, the Level III of record changes, or a method or technique is added. The review has to be recorded as a review, with a signature and a date, even where nothing changed. An audit reads an unchanged document with no review record as an unmanaged one.