Preventive, corrective or predictive maintenance: which policy fits your fleet?.
Corrective maintenance is what happens when it is already too late. Preventive is what sits in the plan. Predictive is what every trade show promises. Here is what is realistic with the data you have today.
Every maintenance department knows the three words, but in practice they blur. This article sets out the definitions and, more importantly, explains which policy is realistic for a fleet with machines of different brands and ages.
What is corrective maintenance?.
Corrective maintenance is work carried out after a failure has occurred: the machine stops, the technician is dispatched, the part is replaced. It is the most expensive form of maintenance, not because of the repair itself but because of everything around it: a stalled site, an emergency parts delivery, a rented replacement machine and a customer waiting.
A rule of thumb from the industry: an unplanned failure costs three to five times more than the same job done as planned work.
What is preventive maintenance?.
Preventive maintenance runs on fixed intervals - for heavy equipment expressed in engine hours: a service every 250, 500 or 1,000 hours as the manufacturer prescribes. Filters, oil, inspection points: all to schedule, before anything breaks.
Preventive maintenance is the norm in the equipment world, but it lives or dies on reliable engine hours. If hours are reported by operators or read once a month, you plan maintenance on stale numbers. A machine working harder than expected slips past its interval unnoticed, and you are back to corrective work.
What is predictive maintenance?.
Predictive maintenance uses live machine data - fault codes, temperatures, pressures, consumption trends - to see a failure coming before it happens. Instead of every 500 hours it becomes: this machine is behaving abnormally, schedule an intervention.
Predictive maintenance is not all or nothing. The realistic first step is not an AI model that hears bearings wear out, but fault codes and machine data that arrive automatically and are immediately understandable. A coolant temperature creeping up week after week, or a recurring fault code, is often all the prediction you need.
Which policy fits your fleet?.
The right question is not which form is best, but which form you can deliver today with the data you have. The order is almost always the same.
- Get engine hours in automatically and correctly for every machine, including the old ones and the smaller brands
- Add faults and machine parameters, so planning shifts from by interval to by signal
- Measure your share of unplanned interventions; that number tells you whether the policy works
Corrective work never disappears entirely. But on a data-driven fleet, unplanned work is the exception rather than the rule.
The obstacle nobody mentions: mixed fleets.
Most articles on maintenance strategy assume one factory, one machine type, one data source. A real equipment fleet looks different: dozens of brands, build years from 2005 to today, and OEM portals that do not talk to each other. Predictive maintenance therefore starts with connectivity, not algorithms: every machine connected, all data on one screen.
