When You Uprate an Old Plant, You Invalidate Its Inspection Plan

Debottlenecking is the most cost-effective capacity in the industry. You take a plant that already exists, already has a licence to operate and already has a grid connection and a jetty, and you find the handful of constraints holding it below what the rest of the equipment could do. Replace a few exchangers, re-wheel a compressor, add a train, change the feed slate, and the asset produces materially more than it did last year for a fraction of what a new facility would cost.

It happens everywhere. A refinery lifts throughput and moves onto a heavier crude slate. A gas plant is uprated to push more through existing trains. A petrochemical complex re-rates a cracker or switches feedstock. An LNG receiving terminal increases send-out and raises its tank fill and cycling rates. A chemical plant finds another ten per cent in a reactor loop.

The project gets a great deal of engineering attention. Process simulation, hydraulics, metallurgy for the new items, relief system re-rating, constructability, hook-up and commissioning. All of it properly done.

What frequently gets no attention at all is the inspection plan for the equipment that was already there.

That equipment did not change. Its service did. And an inspection plan is not a property of equipment — it is a property of equipment in a service. Change the service and the plan silently stops being valid, usually without anyone producing a document that says so.

What actually changes when you uprate

Three things shift, and they compound.

The damage mechanisms change. A heavier, sourer or simply different feed can introduce mechanisms the original design never had to consider — sulphidation at temperatures that were previously below threshold, naphthenic acid corrosion where the acid number has risen, wet H₂S damage and the hydrogen-induced cracking family where the water phase has changed, amine or carbonate stress corrosion cracking in units running harder, chloride stress corrosion cracking in stainless where the chloride load has moved. On cold and cryogenic service the list is different but no shorter: more frequent thermal cycling changes the fatigue picture at nozzles and supports, and corrosion under insulation behaves quite unlike it does on hot lines, concentrating where cold surfaces draw and hold moisture. API RP 571 is the right reference for the mechanisms, API RP 970 for documenting the corrosion control basis, and the right exercise is a formal re-screen against the new operating case rather than a check of whether the existing list still looks reasonable.

The rates change. This is the one that quietly destroys an inspection interval. If the sulphur in the feed doubles, or the velocity in a line goes up because you are pushing 30% more through it, or an exchanger now runs forty degrees hotter, or a storage tank that used to turn over four times a year now cycles monthly, the degradation rate assumption underneath your next-inspection date no longer holds. The thickness readings you took eighteen months ago were honest measurements of a different plant. Projecting them forward at the old rate gives a remaining-life figure that is not conservative — it is simply wrong, and wrong in the unsafe direction.

The consequence changes. Higher throughput means a larger inventory and a bigger release if something fails. It also means more lost production per day of unplanned downtime. Both sides of the risk calculation move, and they move in the same direction.

So the inspection plan that was defensible last year now under-inspects the circuits most affected by the change — and very often over-inspects everything else, because nobody rebalanced it.

Four things to do before start-up, not after

Re-screen the damage mechanisms against the new operating case. Circuit by circuit, for everything in the affected envelope — which is wider than the project’s battle limits, because feed and product changes propagate upstream and downstream of the units being modified. The output is a revised corrosion study with a dated basis, not a marked-up copy of the old one.

Reset the integrity operating windows before the plant runs, not after the first excursion. API RP 584 is the reference and it is widely specified and rarely operationalised. An uprate is the moment integrity operating windows earn their keep: the limits on temperature, sulphur, acid number, chloride, water content, velocity and pH that keep the plant inside the envelope the inspection plan assumes. Set them during commissioning planning, give each one an owner and a documented response when it is exceeded, and you have converted a corrosion study into a working control. Leave them until after start-up and you will be setting them retrospectively, from excursion data, on an asset that has already run outside its envelope.

Re-run the risk assessment on the affected circuits — not the whole plant. This is where teams either do nothing because a full re-study is unaffordable, or lose a year doing a full re-study they did not need. API 580 and 581 do not require you to start again. They require the assessment to reflect current conditions. Scope the reassessment to the circuits where the damage mechanism, the rate or the consequence has actually moved, and carry the rest forward unchanged with a note explaining why. A targeted reassessment completed before start-up is worth more than a comprehensive one delivered in the second year of operation.

Re-baseline the measurement locations. New mechanisms and new flow regimes put metal loss in new places. Condition monitoring locations chosen for the old service may now be in the wrong position entirely — injection points, mixing tees, elbows downstream of a changed velocity, dead legs that were dormant and are now in intermittent flow. Where the degradation rate assumption has moved, the minimum thickness and the next-inspection calculation move with it: API 510 for vessels, API 570 for piping, API 653 for atmospheric and low-pressure tanks, and API 625 with API 620 Annex Q where the tankage is refrigerated or cryogenic. Plant-specific codes matter here too — NFPA 59A and EN 1473 set expectations around LNG facilities, API 2510 around LPG storage, and a re-rate can move you against them without anyone re-reading them. And where the change altered the demand rate on a protective function, the SIL verification done under IEC 61511 needs revisiting — a safety integrity level is a function of how often the function is called upon, and that has just changed.

The part everyone skips: writing down that you did it

Each of those four steps produces a decision. The decision matters less than the record of it.

Something will eventually go wrong on an uprated plant — a leak, a shutdown, an inspection finding worse than expected. When it does, the question will not be whether your corrosion study was perfect. It will be whether, at the point the plant’s service changed, somebody systematically asked what that change meant for integrity, and acted on the answer.

An organisation that can produce a dated reassessment, a revised set of operating windows with named owners, a re-baselined set of measurement locations and a clear statement of what was deliberately left unchanged and why, is in a completely different position from one holding a project close-out report and a set of as-built drawings. The engineering in both cases may have been identical. The defensibility is not.

This is also, unglamorously, the reason the record has to live somewhere other than the project team’s shared drive. Project teams disband. The integrity obligation does not.

Where ONE fits

ONE was built for exactly this problem with North Sea operators managing ageing offshore assets through repeated modification — the same discipline, in a different climate.

The Component Asset Register holds equipment and corrosion circuits at the level the reassessment actually works at, with the materials, service conditions and inspection history attached, so “which circuits are affected by this change” is a query rather than a workshop. Risk Based Assessment runs qualitative or quantitative assessments on your own probability, consequence and confidence scales and your own risk and inspection matrices — and supersedes rather than overwrites, so the pre-uprate basis stays intact and reviewable alongside the new one. That single behaviour is what makes a reassessment auditable years later. Workpacks and Inspections and Mobile ONE carry the revised scope to the technician and bring results back against the right component, including where connectivity is poor. Anomaly Management takes every finding to audited close-out, Performance Standards holds the safety-critical element verification trail, and Reporting and KPIs gives the engineer, the asset manager and the auditor three views of one dataset.

Apache North Sea reported roughly 40% less administration time and a full audit trail after moving off heritage systems onto ONE.

If you have an uprate in execution or in planning, the question worth asking this week is a short one: who owns the revised integrity basis, and when is it due? If the answer is nobody and after start-up, that is worth fixing while it is still cheap.

If you would like to talk through what an integrity basis review looks like on your estate, request a demo or get in touch.

Frequently Asked Questions

What is debottlenecking in a process plant?

Debottlenecking is increasing a plant’s capacity by removing the few constraints that limit it, rather than building new facilities. Typical changes include replacing exchangers, re-wheeling compressors, adding a train or changing the feed slate. It is usually the cheapest way to add capacity, but it changes the service conditions of the existing equipment.

Does a plant modification require a new RBI assessment?

API RP 580 expects a risk-based inspection assessment to be updated when process or operating conditions change. It does not require a full re-study. The reassessment can be scoped to the circuits where the damage mechanism, degradation rate or consequence of failure has actually changed, with the reasoning for the unchanged circuits recorded.

What are integrity operating windows?

Integrity operating windows are limits on process variables such as temperature, pressure, chemistry and flow rate that keep equipment inside the conditions its inspection plan assumes. API RP 584 sets out how to establish them. Each window should have an owner and a defined response when it is exceeded.

How does a feed change affect corrosion?

A heavier, sourer or otherwise different feed can introduce new damage mechanisms and speed up existing ones. Examples include sulphidation, naphthenic acid corrosion and wet H₂S cracking. API RP 571 describes these mechanisms, and a formal re-screen against the new operating case shows which circuits need their inspection plans revised.