When to replace industrial equipment: estimating how long it has left

Estimating remaining useful life comes down to answering one simple question: on what date will this piece of equipment reach its allowable limit if the observed degradation carries on at the same rate? The formula itself is elementary. The hard part is assembling the conditions that make the result defensible, and knowing how to say when those conditions are not met.

What the calculation assumes

Take the most common case, loss of wall thickness through corrosion. The logic runs as follows: you know the current wall thickness at a point, you know the thickness below which the equipment can no longer be kept in service under its operating conditions, and you know the rate at which the gap is closing. The time remaining follows.

Each of these three terms deserves a close look, because each carries its own way of going wrong.

The current wall thickness is the least troublesome, provided it comes from a properly identified location. The whole exercise rests on measuring the same point over time, not a point that merely sits nearby. See tracking the degradation of a piece of equipment on this.

The minimum allowable wall thickness is not a figure you read off an inspection report. It is the result of a calculation that accounts for the service pressure and temperature, the material, the geometry and the rules that apply to the equipment. It belongs to the technical authority, the design office or the notified body. No software can produce it on its own, and any tool that claimed to guess it should be set aside without further discussion. Treating a standard, generic limit as though it applied to your specific vessel is one of the quietest ways to manufacture a reassuring but meaningless number.

The rate of degradation is the most fragile part. It assumes a sufficient history at the same location, under comparable conditions, and above all it assumes that the past informs the future, which is only true as long as the service conditions do not change. The moment the process, the product or the operating regime shifts, the historical rate stops being a guide and becomes a distraction.

The cases where you should refuse to calculate

• Fewer than three campaigns at the locationtwo values always give you a straight line. The slope obtained can be nothing more than a measurement artefact, and the deadline that flows from it hands out false assurance.
• Minimum allowable wall thickness unknownwith no limit there is no target, and therefore no life. Displaying a deadline by assuming a standard limit amounts to fabricating information.
• Change of service conditionsa change of product, temperature or operating regime makes the earlier history unrepresentative. The past rate no longer says anything about the rate to come.
• Non-uniform degradationlocalised corrosion, pitting or under-deposit attack does not behave like a general loss of thickness. The linear calculation simply does not apply.
• Points poorly matched between campaignsthis is the most frequent and the most insidious case. The slope is computed across different points and means nothing at all.

In each of these cases, the right answer is not a cautious estimate. It is the absence of an estimate, accompanied by a clear statement of what is missing. A dashboard that displays "insufficient data: two campaigns available" is more useful than one that displays a false date, because it points to the action that needs taking rather than papering over the gap.

How to present a defensible estimate

A useful estimate is not just a date. It comes with everything needed to judge it. Strip those elements away and you are asking people to trust a number on faith, which is precisely what an audit will not accept.

Table scrolls horizontally on small screens.

This presentation has a practical advantage: it makes discussion possible. An inspection manager can challenge a chosen rate, ask for a doubtful campaign to be set aside, or flag a change of service. A date shown on its own, without its assumptions, cannot be argued with. It is either endured or ignored, and neither of those is a decision.

What the figure is actually for

A remaining useful life estimate is not meant to trigger a replacement. Its first job is to set the schedule.

It tells you which items of equipment need to be examined at the next turnaround, and which can wait for the one after. It also lets you anticipate an order for a part with a long lead time, or prepare a justification file ahead of a regulatory deadline rather than after it. Aggregated across a fleet, it finally gives management a view of the investment to plan over several years, a very different conversation from the annual scramble for budget.

On that last point, see building the business case for an AI project in maintenance and our Maintenance Intelligence page.

What software does, and what it does not

An asset integrity tool brings three things to the calculation. It automatically matches campaigns by condition monitoring location, so the slope is computed on the same point rather than on a moving target. It applies the same method across the whole fleet rather than to the handful of items that happen to be closely watched. And it keeps the link back to the source reports, so any figure can be traced to the measurement it came from.

What it does not do: decide fitness-for-service, set the minimum allowable wall thickness, or replace the inspector's judgement on the nature of a defect. A serious tool produces an estimate when the data allows it and explicitly flags the cases where the data does not. A figure that cannot be defended in an audit has no practical value, however confident it looks on a screen.

For the overall approach, see asset integrity management. On the equipment families involved, see vessels and tanks and heat exchangers.