Preventive maintenance: what it is and how to build the plan

Preventive maintenance is carried out at predetermined intervals or according to prescribed criteria, while the machine is still working, to reduce the probability of failure. In Brazilian industry it is the most widely used regime, and it is also the one that suffers most from a badly built plan: it either turns into a task list nobody follows, or into replacing good parts every week.

This article shows how to build a plan that holds up: where to get the interval from, what to write for each piece of equipment, what goes into a route, how to tell whether the plan is alive, and where it wastes.

Where the interval comes from

This is the question that stops everyone. There are four legitimate sources, in this order of priority:

1. Legal requirements. Boilers, pressure vessels and piping have inspection deadlines set by the pressure-equipment regulations that apply to your plant, often by category and type of inspection, with mandatory records. There is no negotiating here, and equipment with an overdue inspection does not run.
2. The manufacturer’s manual. The starting point for lubrication, filters, belts and seals. The manual assumes typical conditions; dust, humidity, high temperature or frequent starts shorten everything.
3. Your own plant’s history. If the work orders show that the belt on that exhaust fan always breaks between 5,000 and 7,000 hours, the interval is 5,000, not the catalog figure.
4. Measured operating condition. Hour meter, start counter, filter differential pressure, oil analysis. Changing a filter based on differential pressure is already a step toward condition-based maintenance.

One rule solves half the problems: count operating hours, not calendar days, whenever the wear comes from use. A pump that runs 20 hours a day and another that runs 3 cannot have the same interval.

The plan, equipment by equipment

The plan is not a single plant-wide document. It is a set of sheets, one per piece of equipment, and each sheet needs six fields:

• Asset identification. Tag, location, criticality and nameplate data (power, speed, voltage, rated flow).
• Task. One line per task, written as a verifiable action: "replace the compressor air filter", not "check the compressor". A vague task produces no evidence.
• Interval and trigger. Hours, cycles, calendar or measured condition. Say which.
• Duration and labor. How long it takes and who does it (mechanic, electrician, lubrication technician, contractor).
• Parts and consumables. Part number, quantity, and whether it is in stock. A preventive task without a reserved part becomes a postponed task.
• Machine stopped or running. This determines whether the task depends on a production window, and it is the information production most wants to see.

Start with the critical assets and work down. A plan with 15 good sheets is worth more than 200 sheets copied from the manual.

What goes into a route

A route is a set of short tasks done in sequence along a path through the plant, almost always with the machine running. It is the cheapest and most profitable part of preventive maintenance:

• Lubrication: point, lubricant, quantity and interval. Too much grease ruins a bearing just as much as too little; record the quantity, not just the point.
• Sensory inspection: look, listen, touch and smell. Leaks, a new noise, visible looseness, a slack belt, dirty cooling fins, a loose anchor bolt, a bearing hot to the touch.
• Readings: hour meter, counter, filter differential pressure, bearing temperature, current with a clamp meter. Writing it down is what turns an inspection into history.
• Cleaning and tightening: fins, panel filters, terminals torqued to specification.

A good route has a set time, a fixed path and an owner. A route without records does not exist.

How to tell whether the plan is alive

Four indicators say almost everything, and all of them come out of your own work-order system:

• Schedule compliance: tasks completed on time divided by tasks scheduled. Below 80%, the plan is too big for the crew.
• Ratio of planned to emergency work: the more hours spent on emergencies, the less the plan is working (or it is working on the wrong machine).
• Rework: a failure that shows up right after an intervention. High rework means a bad procedure, a bad part or an unnecessary task.
• Failures on items covered by the plan. If the machine failed in the mode the plan was supposed to prevent, the interval is wrong or the task does not address the real failure mode.

Where preventive maintenance wastes

Two kinds of waste are structural, and recognizing them is what separates a mature plan from a plan run on inertia.

Discarded useful life. Replacing a component that still had half its life left throws away both the part and the downtime. The wider the spread of real service life, the greater the waste of any fixed interval.

Risk introduced by the intervention. A badly seated bearing, alignment lost after replacing the coupling, dirt that got in while the cover was open. There is a failure rate right after an intervention, and it grows with the number of interventions.

Add to that the classic finding of reliability-centered maintenance: in Nowlan and Heap’s 1978 study, most of the items analyzed had no age-related wear-out zone. For those items, replacing by the calendar does not reduce the chance of failure.

Where the rest goes

Not every task should move. The rule is simple:

• What stays preventive is anything with wear predictable by use and a cheap part: lubrication, filters, belts, seals, cleaning, and everything the law requires on a deadline.
• What moves to condition-based is anything that fails progressively and has a measurable symptom: bearings, alignment, looseness, electrical overheating, impeller fouling. On these items, the inspection becomes the measurement, and the task is triggered by the trend.

The practical path is to add measurement to the critical machines without touching the rest of the plan, and to let the first year of data decide which intervals can safely be extended.

What to do now

1. Draw up the asset list with tag, criticality and nameplate data.
2. Install or read hour meters on the machines that run on different duty cycles. Without operating hours, the interval is a guess.
3. Write the sheets for the ten most critical assets, with verifiable tasks and reserved parts.
4. Set up one lubrication route and one inspection route, each with a path and an owner.
5. Measure compliance and emergency work for three months before changing anything.

If you want to know what your plant needs, talk to one of our engineers.

---

Sources: EN 13306, Maintenance — Maintenance terminology; F. S. Nowlan and H. F. Heap, Reliability-Centered Maintenance, 1978; ISO 17359:2018.