A UT gauge produces a number in about two seconds. Turning that number into something an inspector will stand behind five years later, in front of an auditor or an incident review, takes a workflow. Most sites still lose the argument in the gap between the gauge and the record.
Where readings lose their standing
A thickness reading defends itself only if you can answer four questions about it: where exactly was it taken, with what instrument and calibration, by whom, and what was the reading before it. Each question fails in a familiar way.
- Location fails on CML identity. If the condition monitoring location lives on a marked-up isometric in a filing cabinet, two technicians can measure two different spots and file both against CML 14. Corrosion rates computed from that pair are fiction.
- Instrument context fails on transcription. A reading copied from the gauge screen into a notebook, then from the notebook into a spreadsheet, arrives with no serial number, no calibration block, no couplant note, and at least one chance of a typo per hop.
- Identity fails on batch entry. When a data clerk keys in forty readings a week after the survey, the technician who took them never sees what was filed under their name.
- History fails on file format. Prior readings scattered across contractor spreadsheets mean the person at the CML cannot see whether today's number is a continuation or an anomaly.
The workflow that closes the gap
The fix is not a better gauge. It is capturing the reading against a durable CML identity at the moment it is taken.
- The technician opens the CML on a device, sees its location photo, its grid position, and its last three readings before coupling the probe.
- The reading lands directly against that CML, stamped with technician, timestamp, instrument, and calibration reference. No transcription hop exists.
- Validation runs at the point of capture. A reading thicker than nominal, or implying an absurd short-term rate, gets challenged while the technician is still on the platform and can re-shoot.
- Sync happens the same shift, offline capture included, so the corrosion engineer works from the survey, not from a two-week-old copy of it.
That fourth step matters more than it sounds. On sites without cell coverage, an app that quietly fails offline pushes technicians back to paper, and the whole chain collapses to step zero.
What this buys you downstream
Once readings carry their own context, the downstream work stops being archaeology. Remaining-life calculations run on trustworthy pairs. Rate outliers point to real corrosion or a mislocated CML, and you can tell which. When the auditor pulls a CML at random, the record shows the reading, the instrument, the calibration, the technician, and the trend, with nothing reconstructed after the fact.
The test for your current process is simple: pick one CML and try to produce all four answers for its latest reading in under five minutes. If you cannot, the problem is the workflow, not the people.
This is the workflow intelliNDT was built around: CML identity, capture-time validation, and offline sync that assembles the record as the survey happens. Bring one of your own thickness routes to a demo and run it through.
