Who this is for: Maintenance and reliability engineers, planners, supervisors, and plant managers selecting maintenance tasks for pumps, fans, gearboxes, conveyors, and other equipment.
1. Preventive, predictive, and other maintenance approaches
Preventive maintenance (PM) uses predefined intervals or usage triggers to inspect, service, replace, or restore an item. Predictive or condition-based maintenance (PdM/CBM) uses observed condition and degradation evidence to decide when investigation or intervention is needed. The names are sometimes used differently by organizations: the critical distinction is whether the action is interval-directed or condition-directed.
| Approach | Trigger | Most suitable when | Main limitation |
|---|---|---|---|
| Time / usage-based PM | Calendar, cycles, operating hours | Effective interval is supported by wear, warranty, regulation, or test history | May replace parts too early or miss random failures |
| Condition-based / predictive | Trend, inspection finding, validated alarm | Failure develops measurably with enough lead time to act | Needs reliable measurements, interpretation, and response |
| Failure-finding | Scheduled proof test | Hidden protective functions could fail without obvious symptoms | Test method and interval must be justified |
| Planned run-to-failure | Functional failure | Low consequence, safe and economical to replace after failure | Not suitable when failure has unacceptable consequences |
| Redesign / redundancy | Formal risk and technical review | No practical task sufficiently controls an unacceptable failure consequence | Requires engineering validation and investment |
2. Select tasks by failure mode, not by equipment name alone
A motor, fan, or conveyor can have several distinct failure modes. Bearing lubrication deterioration, coupling misalignment, motor winding failure, and a failed emergency-stop circuit do not automatically deserve the same maintenance approach.
- State the function: Define the required duty, operating conditions, and acceptable performance.
- Identify the failure mode: What physically goes wrong, and how would the function be lost?
- Assess consequences: Personnel safety, environment, lost output, product quality, repairs, and collateral damage.
- Check for an effective task: Is there a credible interval-based action? Is degradation detectable before functional failure?
- Verify the response window: A monitoring alarm must allow enough time to inspect, obtain parts, plan a stop, and safely intervene.
- Define responsibility: Assign method, interval or trigger, alert route, records, and periodic review.
Safety and regulatory tasks: Continue mandatory tests, inspections, protective trips, and approved procedures irrespective of any simplified score or recommendation below.
3. Example selection matrix for industrial equipment
| Asset / failure mode | Possible strategy | Useful evidence | Decision check |
|---|---|---|---|
| Process fan bearing degradation | Vibration and temperature trending + lubrication PM | Comparable baseline, speed, load, spectral data, oil condition | Reliable warning and adequate action time? |
| Gearbox lubricant contamination | Oil inspection / analysis + controlled service | Water, particles, wear debris, operating temperature | Can condition be measured before harm? |
| Scheduled filter servicing | Differential-pressure condition check or interval PM | Pressure drop, manufacturer limits, contamination history | Can the measurement distinguish loading from process changes? |
| Conveyor belt damage | Inspection-based corrective planning | Damage mapping, splice condition, alignment, tracking | Can defects be found and repaired before rupture? |
| Low-cost, noncritical indicator lamp | Possible planned run-to-failure | Consequences, redundancy, replacement availability | Does its failure conceal any safety-critical state? |
| Emergency shutdown / protective device | Approved proof testing and statutory controls | Design requirements, safety analysis, inspection records | Never select run-to-failure from this guide |
4. Two worked decision examples
Example A — Critical fan bearing
Hypothetical case: A production fan runs continuously. An unexpected bearing failure could stop production for 20 hours. The team has stable vibration measurement locations and can review trends weekly. Recent readings are trending upward under comparable speed and load conditions.
Reasoning: A condition-based inspection is credible because the symptom is measurable, operational impact is high, and a planned corrective window may be available. Retain lubrication and manufacturer-mandated tasks; confirm the trend with appropriate diagnostics before deciding on bearing replacement.
Simple consequence illustration: At an assumed interruption contribution of $1,200 per hour, 20 hours represents $24,000 of potential lost contribution, excluding repairs and restart losses. This is an illustrative economic exposure, not a measured site saving or proof that a monitoring program will prevent failure.
Example B — Inexpensive noncritical utility component
Hypothetical case: A small accessible component costs $40 to replace, has an installed spare, and failure does not affect safety, quality, the environment, or production continuity. Replacement can occur without significant outage.
Reasoning: A planned run-to-failure strategy may be reasonable after validating redundancy, actual failure consequence, and the cost of periodic inspection. Introducing online sensors would be difficult to justify without additional evidence.
Example C — Hidden safety function
A protective interlock could fail silently. Its apparent lack of faults does not demonstrate reliability. It requires the proof-testing and verification regime defined by the approved safety engineering process; a simple condition trend or economic comparison cannot replace that obligation.
5. Interactive preliminary strategy decision aid
Use the questions below to explore a starting strategy for discussion. This is not a validated risk-scoring model. A qualified reviewer must assess the failure mode and the result before adopting any maintenance task.
6. Validate and periodically improve maintenance strategy
Task selection is not finished when a PM schedule is published. Each planned task should have a clear failure mode, execution standard, and feedback route. Track evidence such as:
| Indicator | What to examine | Warning sign |
|---|---|---|
| PM findings | Defects found per task and trend over time | Repeated invasive tasks producing no actionable findings |
| Monitoring alerts | Validated detections, false positives, response time | Alarms repeatedly ignored or never linked to work orders |
| Unplanned breakdowns | Repeated failure mechanisms and lost hours | Same failure returns despite task completion |
| Work-order quality | Defect descriptions, parts, as-found/as-left condition | “Completed” records without evidence of equipment condition |
| Risk control | Required statutory and safety tasks completed | Overdue critical tests or unsupported interval changes |
When condition monitoring indicates deterioration, verify measurement quality and operating conditions first; decide whether to intensify monitoring, perform targeted inspection, or schedule a controlled intervention under approved plant procedures.
7. Further reading and related resources
The approach is consistent with reliability-centered maintenance principles, which integrate interval-directed, condition-directed, failure-finding, and other actions according to failure mechanisms and consequences.
- NASA Reliability-Centered Maintenance Guide for Facilities and Collateral Equipment (PDF)
- NASA Facilities Maintenance and Operations Management — RCM chapter
- Industrial Maintenance Shutdown Planning — KHTech
- Condition Monitoring for Rotating Equipment — KHTech
This guide uses hypothetical examples for education. It does not present KHTech customer performance results or claim certification of the decision tool.