Preventive maintenance, from a schedule to a habit

Preventive maintenance is work done on a schedule to stop a failure before it happens, rather than after. Change the filter before the coil blocks. Grease the bearing before it seizes. That is the whole idea, and almost every team already has a schedule built on it.

Most of those schedules are behind. The reason is structural rather than moral: a broken thing is loud and a thing that is going to break is silent, so the loud one wins every week. This guide is about making the silent thing loud enough to compete.

The four kinds of maintenance, and where preventive fits

• •Reactive: something broke, go and fix it
• •Preventive: do the work on a date or a running-hours count, whether or not anything is wrong
• •Condition based: do the work when a measurement says the machine needs it
• •Predictive: watch a trend and act before the measurement crosses the line

These are not a ladder you climb. A good programme runs all four at once and chooses per asset. Run a light bulb to failure. Inspect a fire damper on a fixed interval because the law says so. Watch vibration on the pump that stops the production line.

How much preventive work is the right amount

There is a real answer and it is not "as much as possible". Every preventive job costs labour, costs a window when the equipment is unavailable, and carries a small risk of introducing a fault that was not there. Over-maintenance is a genuine failure mode, not a virtue.

The rough test: preventive work should be somewhere near half of all hours worked, and the proportion of jobs that find nothing wrong should not be close to everything. If a check has passed forty times in a row, the interval is too short.

Setting intervals without a reliability department

Most teams inherit intervals from a previous system, from a manufacturer’s manual written for the worst case, or from somebody who left in 2014. Three sources beat all three of those.

• •Your own failure history: what broke, and how long after its last service
• •The duty cycle: a fan running two shifts is not a fan running eight hours a week
• •What the law or the insurer requires, which is a floor rather than a plan

Where you have none of these, start with the manufacturer interval and review it after a year of your own data. That is a starting position, not an answer.

Make the slip visible

The single highest-leverage change in most programmes is putting late preventive work on a screen somebody opens in the morning, as a count, by building. Not a quarterly report. A number, today.

This works because it converts a vague unease into something specific that a person can be accountable for. Vague unease loses arguments about resourcing. A number does not.

Say how late, not just late

A list where this morning’s slip and last March’s look identical trains everybody to ignore the word. "Five months late" gets acted on. "Overdue" does not.

Build the plan on real durations

Most schedules are built from guesses about how long jobs take, which is why they are fiction by Wednesday. Once you are recording runs, use the middle of the range rather than the average. One bad afternoon should not move the plan, and the average is dragged around by exactly those outliers.

Prune once a year

Schedules accumulate. Somebody added a quarterly check for a machine removed in 2019 and nobody took it off. An annual review against the real equipment list usually removes enough work to make the rest achievable.

• •Remove work for equipment that no longer exists
• •Merge rounds that visit the same plant room twice
• •Lengthen the interval on anything that has passed forty times in a row
• •Shorten the interval on anything that has failed twice
• •Delete any step nobody can explain the purpose of

The measures worth watching

• •Preventive work past its date, by building, today
• •The proportion of all hours spent on planned rather than reactive work
• •Repeat faults on the same equipment within ninety days
• •The proportion of finished preventive jobs that arrived with evidence attached

What a working programme feels like

Nobody is surprised. The week is planned on Thursday and mostly survives to Friday. When something does break, somebody can say when it was last touched and what was found, without going to look for a folder. That is the whole return, and it is worth more than any percentage in a brochure.