Predictive maintenance is maintenance triggered by a prediction of failure, based on measured equipment condition such as vibration, temperature, or noise. It aims to act before a failure, but not earlier than needed.
Maintenance strategies are often grouped in three. Reactive: fix after failure. Preventive: service on a fixed schedule, cycle count, or runtime. Predictive: service when data shows the asset is heading toward failure. NIST describes predictive maintenance as initiated by predictions made from observed data such as temperature, noise, and vibration.
The data comes from condition monitoring: vibration on rotating equipment, infrared thermography on electrical panels, oil analysis on gearboxes, ultrasound for leaks and bearings, and motor current. Analysis ranges from simple thresholds and trends to machine learning models that estimate remaining useful life. The research field is often called prognostics and health management (PHM).
It is not the same as condition monitoring. Monitoring detects a change; prediction estimates when it becomes a failure. Many programs stop at detection. Prediction needs failure history, which is scarce for reliable assets. It also only pays if a detected issue leads to a planned work order, parts, and a maintenance window.
Condition monitoring vs predictive vs prescriptive maintenance
| Condition monitoring | Predictive maintenance | Prescriptive maintenance | |
|---|---|---|---|
| What it tells you | The condition has changed | How long until failure | What action to take, when, with which resources |
| Typical output | Detection of a significant change | A prediction of failure, such as remaining useful life | Recommendations, such as replacing a bearing at the next stop |
| What it depends on | Clear alarm limits and a response process | Failure history, which is scarce for reliable assets | Failure history and cost data |
Questions about Predictive maintenance
What is the difference between preventive and predictive maintenance?
Preventive maintenance services equipment on a fixed schedule, cycle count or runtime. Predictive maintenance services it when measured data shows the asset is heading toward failure.
Is predictive maintenance the same as condition monitoring?
No. Condition monitoring detects that something has changed. Predictive maintenance estimates when that change becomes a failure. Many programs stop at detection.
When is predictive maintenance not worth it?
On cheap, redundant parts, where running to failure can be the rational choice. It also only pays if a detected issue leads to a planned work order, parts and a maintenance window.
Predictive maintenance in real plants
- Domtar and Waites: Domtar monitors equipment at its Kingsport, Tennessee mill with 748 Waites AI-assisted vibration sensors, with a Waites analyst and the mill reliability team reviewing alerts.
- Welsh Water and Samotics: Welsh Water signed a three-year agreement running until 2029 for Samotics SAM4 monitoring of critical wastewater assets; the system has monitored a selection of its assets since 2022.
- Anglian Water and Samotics: Anglian Water extended its condition monitoring framework with Samotics by two years; SAM4 monitors more than 250 clean water assets and will cover new Strategic Pipeline Alliance pumping stations.
- Coca-Cola Europacific Partners and Boston Dynamics: Spot runs thermal, ultrasonic and visual inspection routes on the four bottling lines at the Coca-Cola Europacific Partners plant in East Kilbride, feeding findings to the maintenance system.
- Yorkshire Water and Samotics: Yorkshire Water has around 2,500 pumping assets monitored by Samotics SAM4 electrical signature analysis and is adding 1,200 sewage pumping station assets and 300 clean water pumping assets.
Vendors linked to Predictive maintenance
Samotics (9) · Siemens (5) · Augury (4) · Gecko Robotics (4) · KCF Technologies (3) · AVEVA (2) · Boston Dynamics (2) · Nanoprecise (2) · Waites (2) · ABB (1) · AssetWatch (1) · Emerson (1)
Related terms
Condition monitoring
Condition monitoring is the ongoing measurement of machine parameters such as vibration, temperature, or oil quality to detect significant changes that indicate a developing fault.
Vibration analysis
Vibration analysis is the measurement and interpretation of machine vibration to detect faults in rotating equipment, such as imbalance, misalignment, looseness and bearing damage, before they cause failure.
Prescriptive maintenance
Prescriptive maintenance is a maintenance approach that goes beyond predicting when equipment will fail and recommends what action to take, when, and with which resources to prevent or limit the failure.
CMMS (computerized maintenance management system)
A CMMS (computerized maintenance management system) is software that keeps a database of a site's assets and maintenance work.
Anomaly detection
Anomaly detection is the identification of observations or patterns that deviate significantly from normal behavior in data.