Failure-mode-based preventive maintenance asks what can fail, what consequence matters, whether deterioration can be detected, and which maintenance strategy can actually control the risk.
Reviewed and authored by Joshua Rivera, XRVE Reliability founder and maintenance & reliability leader.
Legacy PM programs often accumulate tasks because equipment was new, a failure happened once, a vendor recommended a generic interval, or someone wanted to “be safe.” Over time the program becomes a mixture of valuable work, vague inspections, duplicates, unsupported frequencies, and tasks disconnected from current asset risk.
A failure-mode-based review reverses the logic. First define the function and the way it can fail. Then choose the task, inspection method, condition threshold, interval, or alternative strategy capable of managing that failure.
Use when deterioration is meaningfully age- or usage-related and a replacement, restoration, or service interval is technically defensible.
Use when deterioration can be detected with enough warning to plan corrective action. Examples can include vibration, oil analysis, thermography, ultrasonic inspection, or measured wear.
Use for hidden protective functions where the failure may not be evident during normal operation and periodic functional testing is needed.
Use when failure consequence is unacceptable and no feasible maintenance task controls the mechanism well enough.
Use deliberately when consequence is low, failure is evident, restoration is straightforward, and proactive maintenance does not create enough value.
Some failure risk is better controlled through cleaning, lubrication, operating discipline, basic care, setup, or process control rather than a maintenance PM.
“Inspect motor,” “check pump,” and “service conveyor” do not tell technicians what condition to evaluate or what action to take when the condition is unacceptable.
| Task element | Question it should answer |
|---|---|
| Component / point | Exactly what is being inspected, measured, tested, serviced, or restored? |
| Method | How should the technician perform the task? |
| Acceptance criteria | What does acceptable condition look like, and what authority supports the limit? |
| Response | What happens when the condition is outside the acceptable range? |
| Frequency basis | Why is the task performed at this interval? |
| Safety / access | What permits, isolation, guarding, sanitation, or access requirements apply? |
If a critical bearing develops a detectable condition with a usable P-F interval, condition monitoring may provide better warning than frequent intrusive replacement.
Replacing the coupling more often does not control the failure mechanism. The strategy should address alignment, base condition, thermal growth, installation practice, or other verified contributors.
If a protective device can fail without becoming obvious during normal operation, periodic functional testing may be more appropriate than visual inspection.
If failure is evident, spares are available, and consequence is negligible, planned replacement may create more work than deliberate run-to-failure.
During PM optimization, label tasks as retain, revise, combine, remove, convert to condition-based, convert to failure-finding, escalate for engineering review, or replace with another maintenance strategy. Record the reason so the program does not drift back into inherited work with no technical basis.
Never invent OEM values or engineering limits to make a task appear complete. If a torque value, lubricant specification, temperature limit, vibration alarm, clearance, or test criterion requires authority, source it or flag it for review.
Review the full PM library using task quality, history, evidence, and technical authority.
Determine whether task intervals are supported by failure behavior and detection evidence.
Understand why completing PMs on time does not prove the maintenance strategy is working.
Use the method yourself, or apply the same reasoning to your facility's CMMS history through XRVE Reliability.
Free maintenance maturity assessment See sample analysis