THE FIGURE THAT ISN'T ON THE DATASHEET

Two questions decide most plant projects, and neither of them appears in the performance table.

How many hours a year will it actually run, and how long does it take to get it running again when it stops.

Everything else — efficiency, output quality, control sophistication — is multiplied by those two numbers. A plant that produces a slightly less impressive output for a lot of hours a year will beat a plant that produces an excellent output for fewer hours, every year, for the life of the asset. Performance figures describe a plant that is running. Availability describes whether it is.

THE ARITHMETIC THAT SETTLES ARGUMENTS

Take two designs with similar capital cost. One is expected to run 8,000 hours a year, the other 6,000. Those figures are illustrative, but the relationship they show is not.

The first produces a third more output than the second, from the same capital, before any difference in efficiency is considered. Over a twenty-year operating life, that is not a marginal difference between two business cases. It is the difference between a project that services its debt comfortably and one that does not.

Now note where those two numbers come from. Not from the reactor design. From how long the plant is deliberately stopped for maintenance, and how often it is stopped when nobody planned to stop it. Availability is a maintenance outcome, and maintenance is a design outcome.

WHY THIS IS NOT A TECHNOLOGY QUESTION

When a project underperforms on availability, the cause is usually not an exotic technical failure. It is one of four ordinary things.

Access. Can a person reach the component, with the tools required, in a reasonable time? A part that needs scaffolding, a crane, confined-space entry and a two-week production stop to replace is not an expensive part. It is an expensive decision that was made early, when the layout was drawn and nobody was thinking about year eight.

Spares. The difference between a two-day repair and a twelve-week repair is usually the presence of a part on a shelf. Most projects specify spares for consumables and forget the long-lead items — the pumps, valves, instruments and specialised components that take months to arrive and stop everything when they fail.

Diagnostics. Equipment rarely fails without warning. It fails after a trend that nobody was watching: a rising differential pressure, a creeping vibration reading, a temperature that drifts over weeks. Condition monitoring turns an unplanned outage into a planned one, which is the single cheapest availability improvement available to a plant that already exists.

Competence. Someone has to diagnose the problem at three in the morning, with the documentation that was handed over five years ago, and decide what to do. Whether that is possible depends on how legible the design and its records are — not on how clever the process is.

DESIGN DECISIONS THAT PAY FOR THEMSELVES

The interventions are unglamorous and mostly well understood.

Duty and standby on the components that stop the plant. Isolation valves and bypass routes so a section can be worked on while the rest continues. Lifting points, laydown space and removable panels, so maintenance is a planned activity rather than a rigging project. Condition-based maintenance schedules, so equipment is opened when the data says so rather than on a calendar. A spares strategy that is derived from failure modes and lead times rather than from a budget line. And a documentation set written for the person who will use it, not for the person who approved it.

None of this is novel engineering. It is the difference between a design that anticipates being maintained and one that has never considered it.

THE HONEST TRADE

Designing for availability costs money, and pretending otherwise would be dishonest.

More valves, more space, more instrumentation, redundant equipment and a stocked spares room all consume capital, and all of it is visible in the budget at the moment when budgets are most scrutinised. Availability, by contrast, is invisible in that moment. It is an assumption in a spreadsheet, usually inherited rather than derived.

That asymmetry explains most of the industry's availability problems. The cost of maintainability is paid up front, in public, and the benefit arrives years later, quietly, in the form of outages that did not happen. Projects that get this right made the trade deliberately, and can explain why.

WHAT IT MEANS FOR A BUSINESS CASE

If a financial model assumes a high availability figure with no stated basis, the assumption is doing more work than the technology is.

Assumed availability should be derived from something: a comparable operating asset, a maintenance plan with defined intervals, a spares strategy with actual lead times, and an honest assessment of the unplanned failures that the condition monitoring is expected to catch. Where those do not exist, the number is a hope with a decimal point on it.

This matters beyond the operator. Availability assumptions drive revenue projections, debt service capacity and payback periods, and they are usually the least examined inputs in the entire model.

THREE QUESTIONS WORTH ASKING

What availability is assumed, and what is that number based on?

Which components contribute most to unplanned downtime, and how long does each take to replace?

What is the spares strategy, and what are the actual lead times on the critical items?

A design team that can answer those three from memory is maintaining its plant on paper before it exists. One that cannot has designed a plant that will be maintained by improvisation.

PERFORMANCE IS WHAT YOU GET WHEN IT RUNS

There is a version of this industry that competes on performance figures, and there is a version that competes on keeping plants running. The second one is less discussed, harder to market, and ultimately more valuable — because a plant that is stopped produces nothing at all, elegantly or otherwise.

If you are reviewing a design or a business case and the availability number has never been questioned, that is worth a conversation.

Talk to us at zero-x.co