Preventive maintenance: what it is, the types, a PM schedule example and PM compliance

What is preventive maintenance?

Preventive maintenance means servicing equipment on a plan so it keeps running, instead of waiting for it to stop. Each PM is a defined job on one asset: change the inlet filter on compressor 2, check the oil and listen to the gearbox on conveyor 4, replace the vacuum cups on the case packer. The job has a trigger, an interval, a time estimate and a standard to check against.

Most PM work falls into five kinds of task. Inspect: look, listen, feel, and take a reading against a limit. Clean: remove the dirt, dust and product build-up that hide leaks and cause heat. Lubricate: the right grease or oil, in the right amount, at the right point. Adjust: belt tension, chain slack, guide clearances, sensor positions. Replace: filters, seals, belts and other parts that wear at a known rate.

Preventive and preventative mean the same thing. Preventive is the more common spelling in maintenance standards and job titles, and PM is the shorthand on the shop floor, as in "Line 3 is down for PM on Saturday".

Why preventive maintenance matters, and what it costs

A planned stop is cheaper than a breakdown. When a gearbox seizes in the middle of a run, the line stops without warning, the order is late, the technician starts by looking for the cause, and the part may not be on the shelf. When the same gearbox has its oil checked every week, the low level shows up on a Tuesday, the leak is fixed in a planned half hour, and the line keeps its plan.

PM has its own costs. Parts are sometimes changed with life left in them. The machine has to stop for most tasks, so PM is planned downtime the production schedule must allow for. Every PM takes technician hours, and a badly written PM can do harm: a bearing over-greased or a guard put back loose. The aim is not the most PM, it is the PM that prevents a failure you can see coming.

A preventive maintenance schedule example, worked out in numbers

This example is illustrative. A packaging line runs two 8-hour shifts, 5 days a week, for 50 weeks a year: 80 running hours a week and 4,000 a year. The case packer makes 10 cycles a minute while the line runs, which is 600 cycles an hour and 48,000 a week. The compressor serves only this line and runs when it runs. The team has five PM tasks on the line's critical assets.

Three things come out of the numbers. First, the technician load is small on average, 40.5 hours over 50 weeks is about 49 minutes a week, but it is lumpy. In a week where the gearbox check, the case packer, the filler and the wrapper all fall due, the line needs 12 + 45 + 90 + 60 = 207 minutes, almost 3.5 hours. Keeping the wrapper and the compressor filter out of the filler's week brings the worst week down to the filler, the case packer and the gearbox check: 90 + 45 + 12 = 147 minutes, under 2.5 hours.

Second, most of these PMs stop the line. The gearbox check is done running, at the start of the shift, but the other four need the machine stopped: 1.0 + 7.5 + 18.0 + 4.0 = 30.5 hours of planned downtime a year, which is 30.5 / 4,000 = 0.76% of the running time. Production needs those windows on its plan, and the OEE calculation needs a rule for whether they count against availability.

Third, the usage intervals depend on the plan. If the line goes to three shifts, 120 hours a week, the compressor filter falls due every 2,000 / 120 = 16.7 weeks and the case packer every 250,000 / 72,000 = 3.5 weeks, so every 3. A calendar schedule copied from a two-shift year is now late on both. Usage-based PMs should be re-worked whenever the shift pattern changes, or driven from the meter itself.

Fixed or floating due dates

When a PM is done late, the next due date can be set two ways. A fixed schedule keeps the original dates: the PM is always due in weeks 5, 10, 15 and so on, whenever the last one was done. A floating schedule counts from the date the work was actually done.

When preventive maintenance does not help

A scheduled replacement only prevents failures that get more likely as a part ages: wear, fatigue, corrosion, a filter loading up. Many failures do not behave that way. Some are random, as likely in the first week of a part's life as in the fiftieth; electronics often fail like this. Others come from a bad repair and are most likely just after the work is done. The airline studies that led to reliability-centred maintenance (F. Stanley Nowlan and Howard Heap, Reliability-Centered Maintenance, 1978) made this point, and it is why changing a part on a timer does nothing for some failures except add the risk of a bad reassembly.

For those failures, an inspection that spots the early warning, a predictive reading, or a spare on the shelf with a fast changeover does more than a time-based replacement. The maintenance strategy selector asks how an asset fails and what failure costs, and suggests which strategy fits. Reliability-centered maintenance goes through the full method with a worked example.

Two signs that a PM is in the wrong place: the technician finds nothing to fix at PM after PM, so the interval can be stretched, or the machine still breaks down between PMs with the same failure, so the task is the wrong one or the interval is too long.

Common preventive maintenance mistakes

• PM on everything. Putting every asset on PM spreads the crew thin. Rank by criticality and let cheap, non-critical items run to failure with a spare ready.
• Tasks without a limit. "Inspect motor" gets a tick whatever the motor looks like. Write what to check and the value that is out of limit.
• Ticks without readings. A tick says the task was done; a reading shows the trend that warns of the next failure.
• Rounding usage intervals up. Converting 5.2 weeks to 6 runs the part past the maker's interval every time.
• Skipping PMs when production is busy. A skipped PM is a late PM until it is done. Move it to the next window, agree that window with production, and keep it on the overdue list.
• Findings that go nowhere. A leak or a worn coupling found during a PM and not fixed needs its own work order and priority, or it waits for the breakdown.
• Never reviewing intervals. Intervals copied from the manual and never changed are too short for some machines and too long for others. Your own breakdown and reading history should move them.