When to Conduct an RCA
Not every failure warrants a full RCA. Use criticality to decide the appropriate investigation depth:
| Event type | Investigation level | Method |
|---|---|---|
| Critical asset failure causing production loss >4 hours, safety event, or environmental incident | Full formal RCA | Fishbone + 5-Why or Fault Tree + documented report with CAPA |
| Repeated failure of same component (3+ times in 12 months) | Structured investigation | 5-Why analysis + corrective action tracking |
| High-cost repair (>£5,000 or site-defined threshold) | Investigation | 5-Why with documented corrective actions |
| Near miss with significant potential consequence | Full RCA | Same as critical failure |
| Routine failure of non-critical component, first occurrence | Observation only | Record in CMMS; monitor for recurrence |
5-Why Method
The 5-Why technique is structured iteration — ask "why" until you reach a root cause that, when addressed, prevents recurrence. The number of iterations needed varies; 5 is a guideline.
Example — centrifugal pump bearing failure:
- Why did the pump fail? → Bearing seizure (mechanical failure)
- Why did the bearing seize? → Insufficient lubrication (physical root cause direction)
- Why was lubrication insufficient? → Grease interval exceeded — last grease was 14 months ago (human/process issue)
- Why was the interval exceeded? → The lubrication route card didn't include this pump — it was installed 2 years ago during a plant extension (system/latent issue)
- Why wasn't it added to the route? → No management of change (MOC) procedure requires updating the lubrication routes when new equipment is commissioned (latent/systemic root cause)
Root cause: Absence of MOC procedure that links new equipment commissioning to lubrication route update. Corrective action: Implement MOC checklist that includes lubrication route review. Add pump to route card immediately.
Fishbone (Ishikawa) Diagram
The fishbone diagram organises potential causes into categories to prevent tunnel-vision analysis. Draw the effect (failure) at the right-hand "head", then draw "bones" for each cause category:
- Machine/Equipment: Design deficiencies, wear, installation errors, material defects
- Method/Process: Wrong procedure, no procedure, procedure not followed, incorrect sequence
- Material: Wrong specification, substandard quality, incorrect storage, contamination
- Man: Training gaps, fatigue, distraction, inadequate supervision
- Environment: Temperature extremes, vibration, contamination, corrosion
- Measurement: Incorrect instruments, calibration errors, wrong monitoring points
For each category, brainstorm all possible contributing causes. The team then investigates the most likely candidates using evidence from the failure site, CMMS history, maintenance records, and physical examination of failed components.
Three-Level Root Cause Framework
All complete RCAs should identify causes at three levels:
| Level | Definition | Example from pump failure above | Corrective action type |
|---|---|---|---|
| Physical root cause | The physical mechanism of failure — what actually broke and how | Bearing raceway fatigue from inadequate film thickness caused by grease starvation | Replace bearing; confirm correct grease type and quantity |
| Human root cause | The human act or omission that enabled or triggered the physical failure | Technician did not add pump to lubrication route when installed | Retrain on new equipment commissioning procedure; immediate route update |
| Latent root cause | The organisational or system failure that allowed the human error to persist uncorrected | No MOC procedure requires lubrication route review at commissioning | Implement MOC checklist; audit all recently added equipment for lubrication coverage |
RCA Report Structure
A formal RCA report should include:
- Event description: What failed, when, where; production impact; safety and environmental consequences
- Timeline: Chronological sequence of events leading up to the failure
- Evidence gathered: Physical examination findings; photo documentation; oil/vibration/process data; CMMS history
- Causal analysis: 5-Why chain or fishbone diagram results
- Root causes: Physical, human, and latent root causes clearly stated
- Corrective actions (CAPA): Specific, measurable, assigned, time-bounded actions to address each root cause
- Recurrence prevention: How the corrective actions will be verified as effective
- Lessons learned: Are other assets affected by the same latent root cause? Horizontal deployment plan