What predictive maintenance detects that preventive maintenance does not
The difference between the two regimes is not one of sophistication, it is one of information. One acts on the calendar, the other acts on the measured condition of the asset.
Three regimes, not two
The conversation is usually reduced to preventive against predictive, when in practice three regimes coexist. Corrective maintenance acts after the failure and still makes sense for low consequence items. Preventive maintenance acts on calendar or operating hours, regardless of the real condition of the equipment. Predictive maintenance acts on condition, that is, it intervenes when measured evidence indicates degradation rather than when the calendar says so.
The difference is not one of sophistication, it is one of information. Preventive work assumes that degradation is a function of time, which holds for abrasive wear and elastomer ageing but is false for most failure modes in rotating machinery and in assemblies under variable load. Because of that, replacing a healthy component on a calendar not only wastes money, it introduces risk, since a good share of premature failures is born of the intervention itself.
What predictive work detects and preventive work does not
Degradation that does not follow the clock
Misalignment, unbalance, mechanical looseness, soft foot and bent shafts evolve with operating condition rather than with elapsed time. Vibration analysis identifies those signatures long before there is audible noise or perceptible heating, so the intervention can be scheduled for the next window instead of happening as an emergency shutdown.
Incipient bearing failure
The bearing is the classic case, since the failure travels a long path between the first microscopic defect and the break. Monitoring through vibration envelope and ultrasound detects the early stage, when there are still weeks or months of safe operation, whereas calendar based preventive work either replaces a good bearing or arrives after the failure.
Efficiency loss that never becomes a shutdown
Fouling, impeller erosion, internal recirculation and passing block valves rarely stop the unit, yet they consume energy and capacity continuously. Predictive work sees that silent cost through performance curves, specific consumption and thermography, while preventive work has no instrument to notice it.
Electrical behaviour of the drive
Broken rotor bars, eccentricity, insulation degradation and phase imbalance appear in the motor electrical signature well before there is any thermal symptom, which allows the drive to be treated and not only the driven machine.
Leakage and passing in valves
Ultrasound applied to block valves and steam traps identifies internal passing without strip-down, which allows intervention to be prioritised by what is actually passing rather than by the order of the list.
Where preventive work is still right
It would be a mistake to replace one dogma with another. Preventive work remains appropriate for wear components with predictable life and low cost, for items with a legally required periodicity, for assemblies whose failure has severe consequence and whose condition is not observable with the available instrumentation, and for equipment where the cost of monitoring exceeds the cost of simply replacing. The correct design is therefore hybrid, with predictive work concentrated where consequence is high and condition is observable.
What implementation really demands
The technology is the simplest part. What is usually missing is everything else, namely the definition of which assets enter the programme and why, the baseline measured with the machine in a known condition, the alarm criterion tuned to the equipment rather than to the system factory default, the flow that turns an alert into a work order with an owner and a date, and the record of what was found at strip-down, which is what calibrates the next diagnosis.
Without that closed loop the programme produces reports and not decisions. With it, predictive maintenance changes the nature of the conversation with operations, since the question stops being when to stop as a precaution and becomes how much time is left before stopping is necessary.
How this connects to the rest of the operation
Condition monitoring directly changes the spare parts policy, covered in how to size MRO inventory for critical items, and the repair or replace decision, covered in when to repair and when to replace a safety valve, since anticipating degradation creates time to buy well instead of buying in a hurry. The asset integrity and predictive maintenance front is described in Solutions.
Frequently asked questions on predictive maintenance
Does predictive maintenance replace preventive maintenance?
No. The design that works is hybrid, with predictive work concentrated on assets of high consequence and observable condition, and preventive work kept for wear components, for items with a periodicity required by standard and wherever the cost of monitoring exceeds the cost of replacement.
What is the main obstacle to implementation?
Rarely the technology. What is usually missing is the loop that turns an alert into a work order with an owner and a date, plus the record of what was found at strip-down, which is the data that calibrates the next diagnosis.
Does predictive work reduce spare parts inventory?
It changes the nature of the inventory. By anticipating degradation it creates time to buy under normal conditions instead of in an emergency, which allows coverage of observable items to be reduced and capital to be concentrated on critical items whose condition cannot be monitored.
Want to assess which assets enter condition monitoring?
We select assets by consequence and observability, define baseline and alarm criteria and design the flow that links alert to work order.