Trusting third-party NDT results in a pulp and paper mill
Mill NDT data arrives from a boiler OEM's service arm, an insurer-driven inspection contractor, a regional rope-access crew and your own people, in four formats with four naming conventions. Deficiency tracking for pulp and paper has to normalize identity and units at intake, because a recovery boiler thickness trend built from mismatched conventions is worse than no trend at all.
The hardest input is the co-extruded composite tube in the lower furnace: a 304L outer layer metallurgically bonded to a carbon steel base such as SA-210. A UT thickness reading through that wall can report total wall or base metal only, and the two differ by roughly the clad thickness. One contractor calibrates on a carbon steel step wedge and reports total wall; the next uses a clad reference and reports base only. Both write a number to three decimals. Trend them together and you manufacture either a phantom loss of about 0.060 inch between outages or a phantom recovery, and someone will make a tube replacement decision on it. The same failure repeats on tube map identity, where floor tubes numbered from the north wall in one report and from the south in another quietly relocate every reading. Intake validation, not report review, is where this gets caught.
Source: Sources: OSHA 29 CFR 1910.119, including the Appendix A list of highly hazardous chemicals covering chlorine dioxide and the mechanical integrity requirements at 1910.119(j), particularly 1910.119(j)(5) on equipment deficiencies; Black Liquor Recovery Boiler Advisory Committee recommended good practices for recovery boiler inspection and emergency shutdown procedure; National Board Inspection Code NB-23, Parts 2 and 3; ASME Boiler and Pressure Vessel Code Sections I, V and VIII; ASME B31.1; TAPPI technical information papers covering digester and Yankee dryer inspection; ASNT SNT-TC-1A; API 579-1/ASME FFS-1 where fitness-for-service is used.
| Deliverable | What typically arrives | Ambiguity that breaks the trend | Intake rule that closes it |
|---|---|---|---|
| Recovery boiler pressure part survey | OEM spreadsheet keyed to their own drawing coordinates | Tube identity does not map to your CML register, nor to the previous contractor's numbering | Reject rows that do not resolve to a registered location; require the mapping table before the crew demobilizes |
| Composite waterwall thickness readings | Values to three decimals with no measurement basis stated | Total wall and base metal differ by roughly the clad thickness on the same tube | Mandatory basis field, velocity, reference block and tube construction from the asset record; no basis, no import |
| Digester weld WFMT report | Narrative and photographs, indications described qualitatively | No acceptance criteria named, and no distinction between pre-grind and post-grind examination | Require procedure and revision, acceptance document, examination stage, and a disposition for each indication |
| Yankee dryer grinding record | Depth removed, issued by the grinding contractor on its own paperwork | Cumulative removal is never reconciled against measured thickness or the minimum shell limit | Post grinding and survey into one derived margin held on the asset; block closure until that margin updates |
| Piping and vessel CML sweep | Blanks where scaffold was never built | A blank reads as no change, so an unreachable location appears stable for years | Import as not examined for access reasons, raising its own deficiency with a named access owner |
| Corrected report issued after the outage | An emailed revision carrying the same filename | The original stays live and decisions get made on withdrawn data | Versioned examination records with explicit supersession, and an alert on every deficiency that cited the old revision |
Four organizations, four conventions, one trend line
A mill of any size buys its inspection from several places at once. The recovery boiler OEM's service arm does the pressure part survey because they hold the design drawings and the tube maps. The property insurer's engineering arm, or a contractor working to the insurer's scope, covers what the policy requires before the boiler goes back into service. A regional NDT contractor handles digester welds, evaporator bodies, liquor piping and the bleach plant. And a small in-house team covers whatever is left and gets asked, in the week after the outage, to make sense of all of it.
None of these organizations shares a naming convention with the others. The OEM identifies a tube by its own drawing coordinate. The insurer's report identifies it by elevation and wall. The regional contractor writes floor, row three, tube twelve, counting from the north. Your maintenance system knows the whole boiler as a single asset number. The individual readings are usually perfectly good work. The identity that is supposed to carry them across years is what fails, and it fails silently, because a mislabeled reading still plots on a chart.
This is the real content of the complaint that contractor data cannot be trusted. In most mills, the data is not wrong. It is unjoinable. What you need is therefore less a report repository than an identity service with a gate on the front: a controlled register of condition monitoring locations, and a rule that no reading enters the trend until it resolves to exactly one of them.
The composite tube reading that is not what it says
Lower furnace waterwalls in a modern recovery boiler are commonly co-extruded composite tubes, with a stainless outer layer, typically 304L, metallurgically bonded to a carbon steel pressure-retaining base such as SA-210. That construction exists because the lower furnace runs reducing and sulfidic and plain carbon steel does not survive it for long. It also means a wall thickness number is meaningless until you know which wall was measured, and no report format in common use makes that field mandatory.
A conventional thickness gauge calibrated on a carbon steel step wedge and running a carbon velocity will read to whatever strong reflector it resolves first. Depending on technique, gating and instrument, that may be the clad interface or the outside surface. Two contractors can honestly report 0.230 and 0.180 on the same tube and both be describing something real about different boundaries. Trend those in one column and you have invented roughly sixty thousandths of loss, or recovery, across a single outage, and someone downstream will act on it.
The intake rule is unglamorous and completely effective. The schema requires a measurement basis field with a controlled vocabulary of total wall, base metal or clad remaining, plus the velocity used, the reference block, and the tube construction as recorded on the asset. Rows without a basis are rejected outright, not imported and flagged. Flagged rows get imported by somebody in a hurry at the end of a long shift. Rejected rows get a phone call to the contractor while their crew is still on site and their instrument settings are still recoverable.
Grading indications on digester welds
Continuous and batch digester shells crack. The mechanism is stress corrosion cracking in weld heat affected zones, driven by the liquor chemistry and the operating temperature, and the industry's response since the cracking episodes of the 1980s has been periodic wet fluorescent magnetic particle examination of internal welds, usually with grinding to remove indications and re-examination afterward to confirm removal.
Here the data quality problem is not units, it is judgment. One contractor reports extensive linear indications on weld fourteen and supplies a photograph. Another reports a count per linear foot with individual lengths and the depth at which each was removed. A third reports only what remains after repair, which is nothing, and files a clean report. None of them states the acceptance criterion applied, because in many mills nobody has written one. Trending the word extensive against seventeen indications at a maximum length of eight millimetres, all removed at 2.4 millimetres of grind depth, is not trending. It is a category error with a chart on top.
A deficiency module can force the issue without dictating the engineering. Require the examination record to name the procedure and its revision, the acceptance criteria document, the technique used including whether the examination was performed before grinding, after grinding or both, and the disposition of each indication. Then require the mill's own Level III to have approved the contractor's procedure. When those fields are mandatory, the second contractor either supplies them or reveals that they have been working to their own house practice for six years, which is itself the finding you were looking for.
The Yankee dryer arithmetic that three parties each hold a third of
A tissue machine Yankee dryer is a large cast iron or steel pressure shell that is ground periodically to restore the surface finish the sheet requires. Every grind removes material permanently. The safe operating limit is a minimum shell thickness for the operating pressure, and the shell is separately examined for cracking, particularly in the head-to-shell region and around the journals, where a failure is not a leak but an energetic rupture.
The arithmetic that matters is cumulative, and it is routinely split across three suppliers. The grinding contractor knows how much material was taken off. The NDT contractor knows the measured thickness at the last survey. The OEM, or a consulting engineer, holds the minimum thickness calculation and the assumptions behind it. If any one of those three records lives only in an email thread, the mill does not actually know its remaining margin, and it will not find out until someone assembles all three under time pressure. Dryers have been run close to their limit through nothing worse than three tidy, separate filing systems.
An inspection and deficiency record that handles this properly treats the dryer as an asset carrying a running derived value: measured thickness, cumulative removal since the last survey, the calculated limit, and the resulting margin. It refuses to accept a grinding record without a nominated responsible person, and it refuses to close a survey deficiency until the margin has been recalculated. This is exactly the class of cross-supplier arithmetic that a shared record wins outright and a folder structure loses every time.
Intake rules that make third-party data admissible
The practical work of trusting contractor data all happens at one moment: when a file crosses from their system into yours. Define a required schema and enforce it there. At minimum that means the asset and condition monitoring location resolved against your register; examination method and technique; procedure identifier and revision; acceptance criteria reference; technician identity with certification level, method and expiry as at the examination date; instrument serial number with calibration due date; couplant and reference block for ultrasonic work; measurement basis and nominal reference for thickness; the reading itself; and the date and shift.
Then decide, explicitly, what an absent reading means. This single rule prevents more bad decisions than any other. If scaffold was not built and a location was not examined, that must import as not examined for access reasons, never as a blank that a trend engine treats as unchanged, and never as a carried-forward repeat of the previous value. It is entirely ordinary for a mill to discover that a location has been reported as stable for three outages because nobody could physically reach it and the software had no way to say so.
Finally, hold the contractor's own quality documents against the same record: their written practice, the Level III approval of the procedure you accepted, and their equipment calibration certificates with traceability. When you can produce all of that from the deficiency the examination supports, in one screen, the conversation with an insurer's engineer or a jurisdictional inspector changes character completely. You stop defending your process and start discussing the component.
What the regulator and the insurer each require of the same record
Bleach plants that generate and store chlorine dioxide bring the mill inside OSHA's process safety management standard, because chlorine dioxide appears on the Appendix A list of highly hazardous chemicals in 29 CFR 1910.119. That pulls the covered equipment into the mechanical integrity requirements at 1910.119(j), and specifically into 1910.119(j)(5), which requires that equipment deficiencies outside acceptable limits be corrected before further use, or corrected in a safe and timely manner when necessary means are taken to assure safe operation in the interim.
Read that clause slowly, because it is a deficiency tracking requirement written into federal law. It obliges you to have acceptable limits defined in advance, to detect departures from them, and to record either the correction or the interim measures that justify continued operation. A deficiency record on PSM-covered equipment that says monitor, with no acceptable limit stated and no interim safeguard identified, is a citation waiting to be written and, more importantly, is a decision nobody has actually made.
The insurer's requirements run in parallel and are often stricter in practice than anything a regulator asks. Recovery boiler inspection expectations follow the recommended good practices published by the Black Liquor Recovery Boiler Advisory Committee, and the mill's ability to demonstrate a complete, current pressure part inspection record is what allows a boiler to return to service after an outage. Meanwhile jurisdictional pressure equipment repairs run through the National Board Inspection Code, where an authorized inspector signs and the repair organization's certificate matters. Three audiences, one record, and none of them accepts a spreadsheet with merged cells.
The outage compresses judgment into hours
A recovery boiler outage is short and fully scheduled from the first hour. Thickness data lands in volume in the first days, and decisions about tube replacement have to be made while the welders, the scaffold and the tube stock are still on site. Nobody has a week to reconcile naming conventions and measurement bases afterward. The reconciliation must have happened at intake, automatically, or it will not happen at all, and the mill will run another year on data it privately distrusts.
What a good module provides during that window is not analysis. It is triage: a queue of rows that failed validation, ordered by consequence, each carrying the specific reason and the contractor to call. Ten rejected rows on the primary air port elevation, where composite tube cracking concentrates, matter more than four hundred clean rows on an upper wall. If the exception queue is the first screen your inspection coordinator opens each morning of the outage, contractor data quality stops being an annual complaint and becomes a small, daily, solvable task.
The same window is where new deficiencies are created faster than anyone can type them. Capture has to work on a tablet inside a boiler with no signal, with the asset register and open findings available offline, and it has to reconcile without creating duplicates when the device reconnects in the trailer. A system that requires connectivity produces paper, and paper produces exactly the reconciliation problem the whole exercise was meant to eliminate.
How to evaluate this module with your own data
Run the evaluation with two real reports from two different contractors on the same asset, from two different years. Hand them to the vendor and ask them to import both and produce a single trend. Watch carefully what the system does with the conflicts. If it imports everything cleanly and returns a smooth chart, that is the failure mode, not the success case. What you want is for it to stop and tell you that the two reports disagree about tube numbering and measurement basis, and to name the specific rows.
Then ask for the reverse trace. Given a single reading three years old, produce the technician, their certification status on that date, the procedure revision, the instrument and its calibration record, and the access status of the neighbouring locations that day. This is the same trace an insurer's engineer or a jurisdictional inspector will request after an event, and it is the only real proof that the intake discipline was enforced rather than merely described in a sales deck.
Finally, test the boring parts, because that is where mills actually lose data. What happens when a contractor sends the same file twice? What happens when a corrected version arrives a week after the outage with the same filename? Duplicate handling and supersession look like clerical concerns until the day a corrected report silently fails to replace the original and a tube replacement decision gets made on withdrawn numbers.
Why do two contractors report different thicknesses on the same tube?
Usually because they measured different things and neither said so. On a co-extruded composite tube the instrument may resolve the clad interface or the outside surface depending on velocity setting, technique and instrument gating, so total wall and base metal both get reported as thickness. Add spot location differences of an inch or two on a corroding surface, and two honest crews produce a spread wide enough to trigger a tube replacement that nothing physical justifies.
What should the system do with a location that could not be accessed?
Import it as an explicit not-examined status with a reason code, never as a blank and never as a repeat of the previous value. Then raise a deficiency against the access itself, with an owner and a target outage, because the scaffold decision is what needs correcting. Mills routinely carry apparently stable thickness on locations nobody has physically reached in three outages, and the trend chart gives no hint that anything is wrong.
How should acceptance criteria for digester weld indications be set?
They should be written by the mill, approved by a Level III, and referenced by every contractor procedure the mill accepts. Caustic cracking in weld heat affected zones is graded very differently by different crews when nobody has defined what constitutes a reportable indication, what depth of grind-out is permitted before engineering review, and what triggers a fitness-for-service assessment. Without that document, contractor reports are opinions in different dialects and cannot be compared across years.
Is API 510, 570 or 653 inspector training part of this offer?
No. Atlantis provides NDT training to ASNT SNT-TC-1A and ISO 9712 across UT, RT, MT, PT, ET, VT, PAUT and TOFD, ASNT Level III consulting, inspection management and reporting software, digital twins, 3D laser scanning and independent report validation. API inspector certification is administered by API under its own examination and body-of-knowledge programme, and a mill needing that credential should route it to an authorized provider.
Does this replace the mill's CMMS or maintenance system?
No, and it should not try. The CMMS owns work orders, labour, spares and cost. The inspection record owns condition monitoring locations, examinations, findings and evidence, which the CMMS was never designed to model. The integration that matters is narrow: a deficiency generates a work request, and work order completion returns a reference, without either system pretending to be the master of the other's data. Two masters is how mills end up with three thickness histories.
How do you get contractors to actually send data in your format?
Put the schema in the purchase order and reject at intake rather than complain at review. Most contractors will comply once rejection is automatic and happens while their crew is still on site, because a phone call during the outage costs them an hour and a re-mobilization costs them a week. Providing an import template and a short mapping session before the outage removes almost all of the friction, and it converts an annual argument into a one-off setup task.
Built for any business that runs on operations
Most companies do not fail at their craft. They lose time, margin and goodwill in the gaps between the tools they use to run the place — a quoting spreadsheet that does not talk to the job sheet, a job sheet that does not reach accounts, and a compliance folder nobody can search when a client asks. Atlantis closes those gaps by putting the whole operation on one platform, so information is entered once and everything downstream stays in step.
What you can run on it
- Sales and CRM — leads, quotes, follow-ups and the pipeline that tells you what next month looks like.
- Projects and job costing — plan the work, track the hours and materials against it, and see the margin while the job is still live rather than at final account.
- Field and service teams — dispatch, schedules, mobile capture that works with no signal, and sign-off from site.
- Inventory and purchasing — stock, suppliers, reorder points and goods receipt, joined to the jobs that consume them.
- People — records, qualifications and licences with renewal reminders, timesheets, leave and payroll.
- Quality and documents — procedures and forms under revision control, with the audit trail an inspection or accreditation body actually asks for.
- Accounts — invoicing, expenses, multi-currency and the reporting your accountant stops chasing you for.
Affordable, accessible, fully customizable — and we mean each word
Affordable because the whole suite is included rather than sold to you a module at a time, and because implementation is done by people who have run operations rather than by a chain of subcontractors. Accessible because it runs in a browser and on a phone, works for a small team on day one, and does not need a specialist on staff to keep it alive. Fully customizable because your process is the thing that makes you competitive — the software should bend to it, not the other way round.
Industries we configure for
Service businesses and contractors, manufacturing and fabrication, trading and distribution, laboratories and testing houses, engineering consultancies, construction and facilities, and asset owners across energy, marine, aerospace and infrastructure. Inspection and testing is where we started, and it remains the sector we go deepest in — but the platform underneath is general-purpose, and most of what it does has nothing to do with inspection at all.
What happens when you get in touch
A short conversation, not a sales sequence. We ask how the business runs today and where it hurts, show you the platform doing that work, and send a written quote shaped to your region, your team size and the scope you actually need. No obligation, nothing to install first, and no pressure to decide on the call. Reach out and tell us what you are trying to fix.
Related: business management platform · inspection management software · choosing the right category of software · modules · by industry · asset integrity platform. Book a free consultation.