What is Predictive Maintenance?

• Reactive. Fix it when it breaks. Cheapest per intervention and by far the most expensive per failure, because the failure happens wherever the vehicle happens to be, usually loaded and far from a workshop.
• Preventive. Service on a fixed schedule, by kilometres or by calendar. Predictable and much better than reactive, but it treats every vehicle as average: a truck on smooth highway running gets the same interval as one on mine haul roads, so one is over-serviced and the other fails early.
• Predictive. Service based on the actual condition of this vehicle and the failures its data suggests are developing. The vehicle earns its own interval.

Most Indian fleets sit between reactive and preventive. Moving to predictive is usually less about algorithms than about having condition data at all.

Predictive maintenance is not a single technique. It is a set of signals, each of which fails in a characteristic way before it fails completely.

• Fault codes are the most direct signal. A recurring pending code is the vehicle reporting a developing problem before the driver notices anything.
• Trend rather than threshold. Coolant temperature creeping up over three weeks matters even while every individual reading is within limits. This is the signal a workshop inspection cannot see and continuous data can.
• Duty cycle. Engine hours, load, gradient and idling describe how hard this vehicle has actually worked, which is what wear tracks rather than odometer reading.
• Driver behaviour. Consistent harsh braking and over-revving show up as brake and driveline wear long before they show up as a breakdown.
• Fuel efficiency drift. A steady decline in kilometres per litre on the same route with the same load is a mechanical symptom, commonly injectors, filters or tyre pressure.

The direct saving from catching a fault early is the difference between a workshop repair and a roadside failure. That is real, but it is the smaller half.

The larger half is the trip. A breakdown mid-route means a loaded vehicle immobilised, a delivery missed, a customer commitment broken, possible detention charges, a recovery vehicle dispatched, and a driver stranded, sometimes overnight, sometimes on a highway shoulder where the cargo itself is at risk.

On long-haul routes the recovery logistics alone routinely exceed the repair. This is why predictive maintenance economics look different for a fleet running 400 kilometre highway rotations than for one running city distribution: the further from base a failure occurs, the more the prediction is worth.