FMEA Ratings Guide
Detailed guidance on assigning Severity, Occurrence, and Detection ratings.
Severity (S)
How severe is the effect on the customer? (1-10 scale)
Rating Table
| Rating | Effect | Criteria |
|---|---|---|
| 10 | Hazardous without warning | Death or serious injury, no warning |
| 9 | Hazardous with warning | Death or serious injury, with warning |
| 8 | Very high | Vehicle/product inoperable, loss of primary function |
| 7 | High | Vehicle/product operable but reduced, customer dissatisfied |
| 6 | Moderate | Vehicle/product operable but comfort/convenience degraded |
| 5 | Low | Fit and finish/squeak and rattle item, noticed by most customers |
| 4 | Very low | Fit and finish/squeak and rattle item, noticed by discriminating customers |
| 3 | Minor | Fit and finish/squeak and rattle item, noticed by very few customers |
| 2 | Very minor | No effect noticed by customer |
| 1 | None | No effect |
Important Notes
- Severity cannot be reduced through controls or testing
- Only design changes can reduce severity
- Always rate worst-case scenario
- Customer perspective only
Occurrence (O)
How frequently does the cause occur? (1-10 scale)
Rating Table
| Rating | Probability | Cpk | Failure Rate |
|---|---|---|---|
| 10 | Very high: Failure almost inevitable | < 0.55 | > 1 in 2 |
| 9 | High: Repeated failures | 0.55 | 1 in 3 |
| 8 | High: Frequent failures | 0.78 | 1 in 8 |
| 7 | Moderate: Occasional failures | 0.86 | 1 in 20 |
| 6 | Moderate: Occasional failures | 0.94 | 1 in 80 |
| 5 | Low: Relatively few failures | 1.00 | 1 in 400 |
| 4 | Low: Relatively few failures | 1.10 | 1 in 2,000 |
| 3 | Very low: Isolated failures | 1.20 | 1 in 15,000 |
| 2 | Remote: Failure unlikely | 1.30 | 1 in 150,000 |
| 1 | Nearly impossible | 1.67 | < 1 in 1,500,000 |
How to Estimate
- Use data when available (test results, field data, warranty claims)
- Consider process capability (Cpk)
- Account for current design and controls
- Be realistic, not optimistic
Detection (D)
How likely are current controls to detect the failure before delivery? (1-10 scale)
Rating Table
| Rating | Detection | Likelihood | Criteria |
|---|---|---|---|
| 10 | Absolute uncertainty | No chance | No inspection/control |
| 9 | Very remote | Very low | Random inspection only |
| 8 | Remote | Low | Visual inspection, manual |
| 7 | Very low | Moderate | Double visual inspection |
| 6 | Low | Moderately high | Manual gauging/measurement |
| 5 | Moderate | High | SPC with immediate reaction |
| 4 | Moderately high | Very high | SPC with reaction plan |
| 3 | High | Very high | Automated gauging, 100% inspection |
| 2 | Very high | Almost certain | Error-proofing, automated 100% inspection |
| 1 | Almost certain | Certain | Defect cannot be made (design prevents it) |
Important Considerations
- Rate detection before customer receives product
- Consider all test and inspection methods
- Lower is better (easier to detect)
- Don't assume future improvements
RPN Calculation
RPN = Severity × Occurrence × Detection
RPN Priority Matrix
| RPN Range | Priority Level | Action Required |
|---|---|---|
| 125-1000 | Critical | Immediate corrective action required |
| 50-124 | High | Action plan required with specific timeline |
| 10-49 | Medium | Action recommended, monitor closely |
| 1-9 | Low | No action required, continue monitoring |
Mitigation Strategy
To reduce RPN, focus on:
-
Occurrence (easiest to reduce)
- Improve design robustness
- Add redundancy
- Prevent root cause
-
Detection (moderate difficulty)
- Add automated inspection
- Improve test coverage
- Error-proofing
-
Severity (hardest to reduce)
- Redesign to eliminate/reduce effect
- Add safety features
- Change architecture
Focus Areas
A reduction in any factor reduces RPN. For example:
- S=10, O=6, D=7 → RPN=420
- Reduce O to 2: RPN=140 (67% reduction!)
- Reduce D to 3: RPN=60 (86% reduction!)
Common Mistakes
Over-Rating
❌ Don't:
- Rate every effect as severity 10
- Assume worst-case for occurrence
- Ignore existing controls
✅ Do:
- Use rating tables consistently
- Base on data when possible
- Consider current design state
Under-Rating
❌ Don't:
- Underestimate severity to avoid action
- Assume controls work perfectly
- Rate based on desired state
✅ Do:
- Be honest about risks
- Challenge assumptions
- Focus on facts
Examples
Example 1: Battery Overheat
- Severity: 10 (fire hazard, injury risk)
- Occurrence: 4 (rare, 1 in 500 charges)
- Detection: 6 (temperature monitoring, but not fail-safe)
- RPN: 240 (CRITICAL)
Action: Add automatic charging cutoff → O=2, D=3 → RPN=60
Example 2: Weld Defect
- Severity: 7 (product recall, safety risk)
- Occurrence: 5 (moderate, 1 in 100)
- Detection: 8 (visual spot check only)
- RPN: 280 (CRITICAL)
Action: 100% ultrasonic testing → D=3 → RPN=105 (still high, improve process)
What's Next?
- Creating FMEA - Apply ratings to your FMEA
- Actions and Mitigation - Reduce high RPNs