During a turnaround, most of the NDE on site is not performed by your people. UT crews, RT contractors, eddy current vendors, and PAUT specialists arrive with their own instruments, their own software, and their own reporting habits. The findings they generate become the basis for your repair decisions and your regulatory record, but the data starts its life in someone else's system.
If you do nothing about that, the outage ends with your equipment history scattered across four vendor formats, three portals, and a stack of PDFs named by job number instead of equipment tag. The reports exist. The record does not.
The data contract comes before the crew
The fix is contractual and procedural, and it has to happen before mobilization, because no vendor restructures their reporting in the middle of an outage.
- Equipment identity is yours, not theirs. Vendors organize by job and report number. Require every result to carry your equipment tag and, where relevant, your CML or weld identifiers. A report you cannot file against an asset is a report you will re-interpret later, badly.
- Structured data, not just signed PDFs. The signed report matters for the record, but thickness readings, indication tables, and tube results should also arrive as data: readings keyed to CMLs, indications keyed to welds or tubes. A PDF satisfies the auditor; the data serves the next engineer.
- Radiography in DICONDE, with the images. If you accept only the reader sheet, you own an interpretation without its evidence. DICONDE files keyed to weld identifiers mean the next examiner, or the next dispute, can go back to the image.
- Turnaround, in shifts not weeks. Repair decisions wait on results. Agree the reporting cadence, preliminary results by end of shift, signed report within 24 hours, and hold to it.
- Certifications attached, not asserted. Technician certs and procedure qualifications belong in the record alongside the results they authorize.
Put these five items in the service agreement and the kickoff meeting. They cost the vendor little and cost you everything when absent.
During the outage: receive, don't collect
The operational difference between a good outage and a bad one is whether results are received into the record as they are produced or collected after demobilization.
Received means: the UT crew's readings import against the CMLs on the vessel's record the day they are taken, the RT images land keyed to the weld map, and the turnaround team can see, per system, which examinations are complete, which are outstanding, and which produced findings that need a repair decision. The status board is the record.
Collected means: someone spends the two weeks after startup emailing three vendors for final reports, matching job numbers to equipment tags from memory, and discovering that one crew's technician left the company along with the answer to what "location 4" meant.
The reconciliation habit
One discipline pays for itself daily: a short end-of-shift reconciliation where the NDE coordinator confirms that everything examined on shift is in the system, keyed to the right asset, with anomalies flagged. Fifteen minutes per shift during the outage replaces weeks of forensic work after it, and it catches identity errors while the crew that made them is still on site and able to fix them.
Demobilization is the deadline
Whatever is not in your system when the vendor demobilizes may never be. Make final data delivery a condition of the last invoice: all signed reports, all structured results, all images, keyed to your identifiers, confirmed received. It is a mundane clause that protects years of equipment history.
intelliNDT is built to be the receiving side of this arrangement, importing vendor UT, tube, and DICONDE radiography results against your equipment and CML structure so subcontracted work lands in your record, not around it. If your last outage ended with a vendor-report chase, the data contract above is where to start, whatever system you run.
