Post turnaround baselines: resetting the record

The pressure at the end of a turnaround is to start the unit. Every hour of delay has a price, the crews are demobilizing, and the close-out report can be finished later. The inspection records written in those last days are also the ones the program will depend on for the next run: what is the new nominal thickness of that replaced shell course, what rate applies to the repaired nozzle, what is the next internal inspection date, and on what basis. A turnaround that leaves the integrity record unchanged, or updated as prose in a report, has spent its inspection budget and kept none of the return. This paper covers what has to be reset, how the standards set the next interval, and what a clean baseline handover looks like.

Why the post-turnaround record is the most valuable one you will write

The turnaround is the only time in a five-to-ten-year cycle when the vessel is open, the piping is accessible, the insulation is off, and the inspectors are on site in numbers. The readings taken then are the densest and most reliable the program will have until the next event. They are also the readings that set the next inspection interval, which sets the next turnaround scope, which sets the budget.

Three kinds of change need to land on the record before startup:

• Design basis changes. A replaced component has a new nominal thickness, possibly a new material or schedule, and possibly a new tMin. A rerate changes MAWP. A change of service changes the damage mechanisms expected.
• Condition changes. New baseline readings at the CMLs, findings closed or deferred, damage mechanisms confirmed or newly identified.
• Derived values. Corrosion rates, remaining life, and next inspection dates, all of which flow from the two above.

If the first two are recorded and the third is not recomputed, the program is running on stale intervals. If the third is recomputed from stale inputs, it is wrong with confidence. Both failures are common where the close-out is a report rather than a record.

What changes when a component is replaced or repaired?

Replacement and repair are different events for the record, and the distinction decides how the corrosion rate behaves afterward.

Replacement (a new shell course, a new nozzle, a new piping spool, a retubed bundle) means the metal at that location is new. The record needs:

• The new nominal thickness, material specification, and for piping the OD and schedule, from the material certificates.
• The new tMin, recomputed for the new geometry and material under the governing code (ASME B31.3 for process piping, ASME VIII Division 1 for vessels, API 653 for tanks).
• A baseline thickness measurement at each CML on the new component, taken at installation, with the instrument and technician recorded.
• The documentation of the repair or alteration under the applicable code and jurisdiction, including any authorized inspector involvement, pressure test, and nameplate changes.

Repair without replacement (a weld build-up, a patch, a re-rolled tube, a lining renewal) leaves most of the original metal in place. The record keeps its history, adds the repair as an event with its own documentation, and takes a new reading at the repaired location. Where the repair is a temporary one under the applicable repair code, the record has to carry the date by which a permanent repair is required.

In both cases the finding that drove the work is closed on the record with the disposition, and any finding deferred to a future event is recorded with its justification. The thickness data model that supports this is described in the glossary under CML and tMin.

How do you reset corrosion rates and remaining life after a repair?

Corrosion rate is where post-turnaround records go quietly wrong. API 510 and API 570 both define a long-term rate (from the initial or a reference reading to the latest) and a short-term rate (from the previous reading to the latest), with remaining life computed as the difference between the current thickness and the required thickness, divided by the rate. The governing rate for the interval is the higher of the two, or a different rate the inspector justifies.

After a replacement, the previous readings describe metal that no longer exists. The reset is:

• Record the installation reading as the new baseline for both rate calculations on that CML. The short-term rate is undefined until the next reading.
• Do not carry the old rate forward by default. It may be the best estimate of what the new metal will see, and the inspector may adopt it as an assumed rate for interval setting until data exists, but that is a decision to record, not an inheritance.
• Keep the pre-replacement readings on the asset for trend and mechanism analysis, with a visible discontinuity so nobody computes a rate across the replacement.

After a repair without replacement, the rate calculation continues, but the reading at the repaired location has to be taken at the same CML position and the effect of the repair on the local thickness noted, or the next survey will show a spurious thickness increase and a negative rate.

Remaining life is then (t minus tmin) divided by the governing rate for each CML, and the component's remaining life is the minimum across its CMLs. Where the turnaround readings give a rate that makes the remaining life shorter than the planned run length, the answer is a fitness-for-service assessment or a scope change, and it has to be decided before startup. The engineering functions behind these calculations are covered on the intelliINTEGRITY page.

How do API 510 and API 570 set the next interval?

The interval is derived from the remaining life, so it is only as good as the rate reset above.

API 510 (pressure vessels). The internal or on-stream inspection interval is limited to half the remaining life or ten years, whichever is less. Where the remaining life is short (under four years is the usual threshold), the interval is the full remaining life up to a two-year maximum. External inspections are due at least every five years or at the internal interval if that is shorter. An RBI assessment under API 580 and 581 can set a different interval where the assessment supports it, and the assessment must be updated when the turnaround changes its inputs: new mechanisms, new rates, changed service.

API 570 (process piping). Piping is classified by consequence. Class 1 (highest consequence, for example flammable services that could auto-ignite or toxic services) has the shortest thickness and visual intervals, Class 2 and Class 3 longer, and Class 4 optional. In every class the thickness measurement interval is also capped at half the remaining life. Injection points and soil-to-air interfaces have their own shorter intervals. A piping circuit whose turnaround readings show an increased rate can have its interval shortened regardless of class, and a circuit that was reclassified because the service changed needs its class recorded before the interval is set.

In both standards the interval is a record with a basis: the remaining life it was derived from, the rate that drove that remaining life, and the person who set it. An interval without a basis is a date on a calendar.

Which findings feed the RBI and damage mechanism records?

The turnaround confirms or refutes what the RBI assumed. A semi-quantitative RBI under API 580 and 581 scores probability of failure against consequence, and the probability side is driven by the damage mechanisms credited to the equipment and the measured rates. After the event:

• Confirmed mechanisms (the thinning was where the model predicted, at the rate predicted) validate the assessment and its interval.
• New mechanisms (cracking found where only thinning was credited, under-insulation corrosion on a circuit thought dry) have to be added as damage mechanism records, and the probability rescored.
• Rates outside the assumed band change the probability of failure and therefore the risk rank.
• Repairs and replacements reset the condition inputs for the affected components.

Where damage mechanisms are prose in a report, none of this happens until someone rereads the report. Where they are records on the asset, the RBI rescoring is a recalculation. The same applies to the CUI program: an insulation removal that found nothing is evidence too, and the frequency assessment should see it.

What does a clean baseline handover look like?

The test of a good close-out is that the inspection planner for the next cycle can open the asset and see the unit as it is at startup, without the turnaround report. That means, per component:

• Nominal thickness, material, tMin with its method and code, and MAWP, current as installed.
• Baseline CML readings dated to the event with instrument and technician.
• Corrosion rates and remaining life recomputed, with the governing rate identified and any assumed rate flagged as such.
• Next inspection dates with their basis (half-life cap, RBI assessment, or class interval).
• Findings closed with disposition; deferrals with justification, basis, and date.
• Damage mechanism records updated.
• Repair and alteration documentation attached to the component, not to the event.

Reports are the byproduct. The record is the deliverable.

How intelliSPEC handles it

CML points carry the fields the reset needs: nominal thickness, tMin and its method, design pressure, material, OD and schedule, allowable stress, joint efficiency, corrosion allowance, latest thickness, and three corrosion rates. The tMin function routes by equipment type to ASME B31.3, ASME VIII Division 1, or API 653 and returns the design code and formula with the number, so a replaced spool gets a tMin with its basis attached. MAWP is computed alongside. The corrosion summary computes short-term, long-term, and maximum rates, takes the maximum as governing, and computes remaining life as (t minus tmin) divided by the rate; the API 510 and 570 interval function caps the interval at half the remaining life.

Damage mechanisms are first-class records read by the API 580/581 semi-quantitative RBI scoring, so a mechanism confirmed or added at the turnaround changes the risk rank without a re-keying step. API 579 Part 4 Level 1 screening (RSF = t/tmin) is available for a thinning finding, ASME B31G remaining strength results are provisional until an engineer verifies them and never auto-derate the equipment, and API 579 Part 5 Level 1 is deliberately not implemented pending licensed-engineer verification. Readings from UT gauges enter over Bluetooth or keyboard wedge without retyping, and the thickness work itself is captured through intelliNDT. A completed record pins the field configuration it was executed with, so the baseline written at this event renders exactly as executed when the next planner opens it.