How to set a maintenance interval you can defend

A maintenance interval should come from three things: what the law requires, how hard the equipment works, and what your own history says about when it fails. Manufacturer manuals are a starting point, not an answer, because they are written for the hardest duty any customer might apply.

Start with the floor

Some intervals are not yours to choose. Fire systems, pressure vessels, lifting equipment, gas appliances and electrical testing all carry required frequencies, and the required frequency is a minimum rather than a target. Mark those in your schedule so nobody optimises them away in a pruning exercise.

Then adjust for duty

The same model of fan running two shifts a day in a dusty warehouse and eight hours a week in an office does not need the same interval. Where you can, schedule on running hours or cycles rather than dates, and let the machine tell you when it has done the work.

The pattern that fits most plant is "every 2,000 running hours or twelve months, whichever comes first". If your system cannot express that, you will over-maintain the quiet machines and under-maintain the busy ones.

Then use your own failures

For any equipment you have several of, sort the failures by how long after the last service they happened. Two patterns show up, and they mean opposite things.

• •Failures cluster well before the interval: shorten it, or the check is missing the real cause
• •Failures are spread evenly with no relationship to service: the schedule is not preventing anything, so look at condition monitoring or run to failure
• •Failures cluster just after a service: the maintenance itself is introducing the fault, which is more common than anybody likes

The forty-in-a-row rule

If a check has passed forty consecutive times and found nothing, one of two things is true: the interval is too short, or the step is not looking at the thing that fails. Both are worth an afternoon. Neither is fixed by doing it again.

Change one thing at a time

Lengthen an interval on one group of equipment, not across the estate, and write down the date you did it. A year later you will want to know whether the extra failures came from the change or from the weather, and the only way to know is to have changed one thing.

Write down why

Every interval should have a one-line reason attached: the regulation, the manual, or the failure that caused you to shorten it. Without that line, the next person to inherit the schedule cannot tell the required from the invented, so they change nothing and the schedule ossifies.