Reliability centred maintenance, the useful parts

Reliability centred maintenance starts from a question most schedules never ask: what does this equipment have to do, and what should we do about each way it stops doing it? The answer is often "nothing", and that is the part of RCM that saves money.

The decision, in four branches

For each failure mode, RCM asks what task would prevent or detect it, and whether that task is worth doing. There are only four outcomes.

• •A scheduled task that restores or replaces before the failure, where wear is predictable
• •A scheduled check that detects the failure developing, where there is warning to find
• •A scheduled test of a protective device, where the failure would otherwise be hidden until needed
• •No scheduled task, and either accept the failure or redesign the equipment

The hidden failure branch is the one people miss

Protective devices fail silently. A relief valve, an interlock, a smoke detector, a standby pump, an emergency light. Nothing tells you it has failed until the day it is needed, so the only sensible maintenance is a test that proves it still works.

Walk any building and you will find protective devices that are visually inspected and never functionally tested. That is the highest-value finding an RCM exercise typically produces, and you do not need the full method to go and look.

Why full RCM is rarely finished

A complete analysis takes a facilitator, a team off the tools for days, and produces documentation proportional to the effort. Programmes that start on the whole plant almost always stall, and the analysis that was completed goes stale before it is implemented.

The version that works

• •Pick the ten assets that hurt most, by callouts or by consequence
• •For each one, list the failure modes you have actually seen, not every conceivable one
• •Ask the four questions and write down the answer
• •Change the schedule that week
• •Come back in a year with the failure data and revise

What you should expect to find

Roughly the same things every time: intrusive scheduled work on equipment that fails randomly, which should stop. Protective devices never functionally tested, which should start. And two or three assets where the honest answer is that maintenance cannot fix the problem and the equipment needs replacing.