What is root cause analysis in maintenance?
Root cause analysis in maintenance is a structured investigation used to explain why a significant or recurring equipment failure occurred, identify the physical and systemic causes that were necessary or contributory to the event, select corrective actions that address those causes, and verify through later operating evidence that the actions reduced recurrence or consequence. RCA should be reserved for problems whose risk, recurrence, uncertainty, or business impact justifies deeper investigation.
Use RCA when the consequence or recurrence justifies investigation depth.
Not every corrective work order needs a formal RCA. A failed lamp, a known consumable replacement, or a one-off low-consequence defect may only need good closeout and normal corrective work. Formal analysis earns its cost when failures are consequential, repeatedly returning, poorly understood, safety- or quality-relevant, or resistant to prior fixes.
A practical trigger combines recurrence with consequence. Bad-actor analysis can identify the assets consuming disproportionate downtime or maintenance capacity, while criticality protects low-frequency but high-consequence failures from disappearing behind event counts.
Do not launch RCA because an asset is loud. Launch it because the organization needs a better causal explanation before choosing the next intervention.
Define exactly what failed before asking why.
Start with the required function, the observed loss or degradation of that function, the asset boundary, operating state, time sequence, and consequence. Statements such as motor failed or pump broke are usually too broad to support causal reasoning.
Preserve the difference between symptom, failure mode, mechanism, and cause. A tripped motor may be the symptom. Bearing seizure may be the physical failure mode. Lubricant contamination may be a mechanism. A damaged seal, wrong storage practice, ingress path, or maintenance error may be causal contributors. Do not collapse those layers into one label.
Build the causal model from evidence, not familiarity.
Useful evidence can include failed parts, photographs, work orders, historian or condition-monitoring trends, alarm sequences, process conditions, operator statements, technician findings, PM history, installation records, drawings, specifications, prior repairs, parts history, environmental conditions, and controlled technical references.
Separate confirmed evidence from inference. A plausible cause is not a confirmed cause until the evidence supports the causal link and competing explanations have been tested reasonably. When evidence was destroyed during restoration, say so and preserve the uncertainty instead of manufacturing certainty.
- Preserve the timeline before memories drift.
- Record what was observed before disassembly changed the condition.
- Use failure history to test whether the same pattern existed previously.
- Document evidence gaps explicitly.
Corrective actions should break a causal path, not merely document the investigation.
Actions should map to the causes they are intended to control. Replacing the failed bearing may restore function but does not eliminate recurrence when contamination, misalignment, installation error, overload, or lubrication control remains unchanged.
A strong action register states the causal condition, action, owner, required authority, completion evidence, and verification measure. Engineering, safety, quality, OEM, and regulatory controls remain authoritative where applicable.
RCA is not complete when the meeting ends.
Verification asks whether the intervention changed the failure behavior over a meaningful exposure period. Depending on the failure, evidence might include recurrence count, time or cycles without repeat failure, condition trend, defect rate, downtime, inspection findings, quality losses, or the absence of the prior causal condition after controlled review.
If recurrence persists, reopen the causal model. An RCA that cannot be challenged by new operating evidence becomes a story rather than a reliability process.
What should a maintenance RCA produce?
It should leave behind a verified event definition, evidence trail, causal model, corrective actions tied to those causes, named owners and authority boundaries, and a future verification method that can prove whether recurrence or consequence changed.
- Distinguish symptom, failure mode, mechanism, and cause.
- Preserve uncertainty when the evidence cannot support a stronger conclusion.
- Assign actions to causal conditions, not to presentation slides.
- Verify effectiveness with later operating history.
Technical references
- NASA Reliability-Centered Maintenance Guide — Defines root cause failure analysis within proactive maintenance and recurrence control.
- NASA Facilities Maintenance and Operations Management, Chapter 7 — Describes RCM as an ongoing process using operating data to improve maintenance and design.
Turn repeated failure into a controlled defect-elimination loop.
Use recurrence, consequence, evidence quality, and causal verification to decide where RCA deserves time and how corrective actions will be proven effective.
Bad Actor Analysis See facility-specific XRVE services