A turnaround changes the plant. Shells are weld-repaired, sections of pipe are replaced, bundles are retubed, valves are overhauled. Every one of those changes invalidates some piece of the inspection history that came before it, and the weeks after startup are when the record either gets reset correctly or starts quietly lying.
The failure mode is specific: a corrosion rate calculated across a replacement. A line section renewed during the outage reads full nominal thickness at the next inspection, the software or the spreadsheet compares it to the pre-replacement reading, and the computed rate goes negative or near zero. The remaining-life number that falls out of that calculation is fiction, and it is optimistic fiction, which is the dangerous kind.
What a baseline reset actually involves
Resetting the record is not one task, it is a sweep across everything the turnaround touched:
- Replaced components get new service-start data. New material, new nominal thickness, new installation date, and a break in the corrosion rate calculation. Future rates compute from the replacement baseline, not across it. The old data stays, clearly bounded, as the history of the retired component.
- Repaired components get their repair on the record. Weld repairs, overlays, and clamps documented with their procedures and NDE acceptance, because the next inspector needs to know that the thick spot at CML 7 is an overlay, not good news.
- CMLs are re-measured where work occurred. Post-repair and post-replacement thickness readings taken and recorded as the new time-zero. Without them, the next reading has nothing sound to trend against.
- CML layouts are updated where geometry changed. Rerouted lines and modified nozzles can orphan old measurement locations. Orphaned CMLs should be retired deliberately, not left to collect confusing data.
- Deferred findings enter the run-cycle plan. Everything found but not fixed carries forward with its justification and its revisit date, into the inspection plan rather than into a closeout appendix nobody reopens.
None of this is intellectually hard. It is bookkeeping under fatigue, performed by a team that has just worked six weeks of nights, which is exactly why it needs to be systematic rather than heroic.
Make the reset a byproduct of closeout
The reliable way to get the reset done is to stop treating it as a separate post-outage project. If repairs and replacements were recorded against equipment during execution, with their NDE results and sign-offs attached, then the reset is mostly a review pass: confirm each replacement triggered a rate break, confirm post-work thicknesses landed, confirm deferred items generated future inspection tasks.
If execution was recorded in daily reports and contractor PDFs, the reset becomes an archaeology project, and archaeology projects get abbreviated when the next unit's planning starts competing for the same engineers.
The verification pass
Before declaring the record reset, run three checks:
- Rate sanity. Query for corrosion rates that changed sign or dropped sharply after the outage. Each one is either a legitimate improvement or a rate calculated across a replacement. Prove which.
- Coverage. Every work order that touched pressure boundary should map to an updated equipment record. Work orders without record updates are the gaps.
- Forward plan. Every deferral and every new baseline should appear in the future inspection schedule. The turnaround is not closed until the run cycle knows what it inherited.
An hour of querying buys years of trustworthy trending.
The record is the deliverable
Startup gets the attention, but the durable product of a turnaround is a plant whose documented condition matches its actual condition. That match is what every remaining-life calculation, every interval extension, and every audit for the next five years will stand on.
intelliINTEGRITY handles rate breaks at replacement and carries deferred findings into the forward plan, drawing on repair records that intelliPROJECTS captured during execution. However you run the reset, the three verification checks above are worth an afternoon before you call the outage closed.
