Which assets suit predictive maintenance?
An asset is a good PdM candidate when three things are true together:
- Failure matters. It affects safety, environment, production, cost or quality enough to justify the effort.
- Failure gives a warning. Its dominant failure modes have a detectable P-F interval long enough to act on.
- Monitoring is cheaper than the alternative. The cost of sensors and analysis is less than the failures it avoids, or less than over-maintaining on a fixed schedule.
| Usually well suited | Usually less suited |
|---|---|
|
|
Build the asset register first
Every later decision depends on a clean asset register. For rotating equipment, capture at least:
| Field | Why it matters |
|---|---|
| Tag, description, location, area classification | Identifies the asset; hazardous areas need certified sensors |
| Asset type and driven equipment | Decides the failure modes and techniques |
| Driver: induction motor (direct on line or variable speed drive), DC motor, engine, turbine | Variable speed and VFDs change data collection and motor current analysis |
| Running speed and speed range | Decides frequency range, sensor type and record length |
| Number and type of bearings (rolling element or sleeve), bearing numbers if known | Decides sensor count and bearing defect frequencies |
| Gearbox stages and tooth counts, belt or coupling, impeller vanes or fan blades | Decides the highest frequency you must measure |
| Power, voltage, current, poles, rotor bars (motors) | Needed for motor current analysis and severity limits |
| Mounting (horizontal, vertical, overhung), foundation | Decides sensor orientation and placement |
| Duty: continuous, intermittent, standby; redundancy | Affects criticality and measurement triggering |
| Failure history and repair cost | Feeds the likelihood and consequence scores |
Scoring criticality
Criticality combines the consequence of a failure with its likelihood. A practical method used across industries:
- Agree a small number of consequence categories, usually safety, environment, production, repair cost and quality.
- Define what each score from 1 to 5 means in your plant, in concrete terms ("plant stop longer than 3 days" rather than "high").
- Weight the categories to reflect your priorities and calculate a weighted consequence.
- Multiply by likelihood (1 to 5) from failure history or engineering judgement.
- Apply an override: any asset whose failure could cause a fatality or major environmental damage is critical regardless of likelihood.
- Score as a team: operations, maintenance, safety and engineering. One person's view is always biased.
The weights and thresholds in the calculator are a reasonable starting point (tier A from a score of 12, tier B from 5). Calibrate them so that roughly 10 to 20% of assets end up in tier A, 30 to 40% in tier B and the rest in tier C. If half your plant is tier A, the definitions are too loose to guide decisions.
What each tier gets
| Tier | Typical monitoring | Techniques | Measurement interval | Response |
|---|---|---|---|---|
| A · critical | Online continuous; wired protection where a machine trip is needed, plus condition monitoring | Vibration plus one or two more matched to failure modes (oil, motor current, thermography, process data) | Continuous to hourly | Alert reviewed the same day; spares held; RUL estimate |
| B · important | Wireless online, or a portable route for accessible assets | Vibration and temperature; ultrasound for slow or grease-sensitive bearings; motor current for remote motors | Hourly to weekly (wireless) or monthly (route) | Alert reviewed within a few days; planned repair |
| C · low impact | Operator rounds, periodic route, or run to failure | Senses, handheld checks, temperature | Monthly to quarterly, or none | Fix when convenient or on failure |
Wired or wireless?
A widely used split: wired online systems for machinery protection on large turbomachinery and other assets where a fast automatic trip is needed, and wireless sensors for the balance of plant, the hundreds of pumps, fans, motors and gearboxes where hourly or daily data is enough and running cable would cost more than the sensor.
Spares, redundancy and lead time
Two assets with the same consequence of failure can need very different monitoring. A pump with an installed standby that starts automatically is less critical than the same pump alone. A gearbox with a 40-week lead time and no spare deserves more attention than one with a spare on the shelf. Include redundancy and lead time explicitly in the production and cost scores, and revisit them when the spares strategy changes.
Key takeaways
- Good PdM candidates matter, give warning, and cost less to monitor than to let fail.
- A clean asset register with speeds, bearings, gears and drivers drives every later step.
- Criticality = weighted consequence × likelihood, with an override for severe safety and environmental consequences.
- Aim for roughly 10–20% tier A. Match monitoring intensity to tier.
Try it in the apps
The Program Setup app scores criticality for each asset you add and uses the tier to choose techniques, wired or wireless sensors, and measurement intervals.