Overview
The PFMEA Detection scale measures how well manufacturing controls can identify when a process failure has occurred. Unlike DFMEA Detection, which focuses on design-level detection methods, PFMEA Detection specifically addresses production process controls such as automated gauging, Statistical Process Control (SPC), and visual inspection at manufacturing stations.
Detection ratings use an inverted scale: lower values (1-3) represent high detection capability through error-proofing and automation, while higher values (8-10) represent poor or absent detection capabilities relying on manual inspection or random sampling.
Detection Rating Scale
| Rating | Name | Detection Method | Detection Capability | Typical Escape Risk |
|---|
| 0 | Unanalyzed | Not yet assigned | Unknown | — |
| 1 | Error-Proofing (Poka-Yoke) | Product/process design prevents defect creation | Almost Certain (99%+) | <0.1% |
| 2 | Automatic Gauging + Auto-Stop | Automated measurement system halts production on defect | Very High (95-99%) | 0.1-1% |
| 3 | Multi-Layer In-Process Detection | Multiple verification points (in-station + subsequent ops) | High (85-95%) | 1-5% |
| 4 | Subsequent Operation Detection | Setup inspection or first-piece detection in downstream operation | Moderately High (70-85%) | 5-10% |
| 5 | Variable/Go-No-Go Gauging | Measurement or attribute inspection after station, sampling-based | Moderate (50-70%) | 10-20% |
| 6 | SPC Charting Control | Statistical Process Control with trend monitoring | Low (30-50%) | 20-40% |
| 7 | Double Visual Inspection | Two independent visual inspections without measurement | Very Low (10-30%) | 40-70% |
| 8 | Single Visual Inspection | One visual inspection per piece, no measurement | Remote (5-10%) | 70-80% |
| 9 | Indirect or Random Checks | Sampling-based or indirect indicator monitoring | Very Remote (1-5%) | 80-99% |
| 10 | No Detection | No inspection or control exists | Almost Impossible (<1%) | >99% |
Detection Method Categories
Prevention-Based Controls (Ratings 1-2)
Error-Proofing (Rating 1) — The highest detection capability where process or product design makes it physically impossible to create a defective part:
- Poka-yoke principle: design eliminates the failure mode at the source
- Examples: keyed connectors, fixture nesting, part geometry constraints
- RPN impact: Dramatically reduces escape risk; often combined with low severity ratings
Automatic Gauging with Auto-Stop (Rating 2) — Automated measurement systems that halt production immediately when specifications are violated:
- In-process automated inspection: laser gauging, pneumatic comparators, vision systems
- Prevents any out-of-spec parts from advancing downstream
- High cost/complexity but near-perfect detection capability
- Requires calibration and maintenance planning
In-Process Detection (Ratings 3-5)
Multi-Layer In-Process Detection (Rating 3) — Multiple verification checkpoints create redundant detection:
- Examples: 100% in-station measurement + second-operation verification
- Catches errors at first opportunity, with backup detection
- Typical in safety-critical or high-complexity processes
Subsequent Operation Detection (Rating 4) — Errors detected in downstream operations:
- Setup approval / first-piece inspection
- Functional test in next operation
- Trade-off: defects caught before customer, but downstream rework costs
Variable/Go-No-Go Gauging (Rating 5) — Measurement or attribute inspection with sampling:
- Go/No-Go gauging (binary attribute checks) at the station
- Variable measurement (continuous, e.g., 0-10 mm)
- Sampling plans reduce detection in proportion to sample size
- Cpk relationship: high-Cpk processes reduce occurrence rating, justifying moderate detection
Statistical & Visual Controls (Ratings 6-9)
| Rating | Method | Advantages | Limitations |
|---|
| 6 | SPC Charting | Predictive trend detection, process capability insight | Reactive (detects drift after samples plotted) |
| 7 | Double Visual | Low cost, no equipment needed | Error-prone; depends on training/attention |
| 8 | Single Visual | Simple implementation | High escape risk; inconsistent standards |
| 9 | Random Sampling | Low labor cost | Very limited detection; only acceptable for stable, high-Cpk processes |
Rating 6 (SPC) and Ratings 7-9 (visual inspection) are not recommended for new PFMEA analyses. Modern manufacturing standards (IATF 16949, AIAG-VDA) favor error-proofing and automated detection. Use only when justifiable by:
- Process stability data (demonstrated Cpk ≥ 1.67)
- Historical defect rates <10 ppm
- Risk assessment showing acceptable residual risk
Detection and RPN Calculation
PFMEA Detection combines with Severity and Occurrence to calculate Risk Priority Number (RPN):
RPN = Severity × Occurrence × Detection
RPN Decision Matrix
This matrix shows how detection rating affects risk classification for a sample process failure:
| Severity | Occurrence | Detection Rating | RPN | Risk Level | Recommendation |
|---|
| 8 | 6 | 1 (Error-proof) | 48 | Low | Acceptable; routine monitoring |
| 8 | 6 | 3 (Multi-layer) | 144 | Medium | Monitor; consider controls if Occ high |
| 8 | 6 | 6 (SPC) | 288 | High | Unacceptable; require auto-gauging |
| 8 | 6 | 9 (Random) | 432 | Critical | Unacceptable; implement error-proofing |
RPN Thresholds
TestAuto2 — Automotive Safety Solution follows AIAG-VDA FMEA thresholds:
| RPN Range | Classification | Action Required |
|---|
| 1–40 | Low | Routine process control; document in control plan |
| 41–100 | Medium | Evaluate control improvements; may require risk control |
| 101–200 | High | Implement additional controls; risk control mandatory |
| >200 | Critical | Unacceptable; implement immediate mitigation |
Detection Improvement Workflow
Detection ratings evolve during PFMEA lifecycle:
Pre-Mitigation Analysis Control Plan Implementation
───────────────────────── ────────────────────────────
Initial Detection Rating Updated Detection Rating
(Current Controls) (Proposed Controls)
↓ ↓
RPN Calculated New RPN Shows Risk Reduction
(Often High) (RPN should drop 50-80%)
↓ ↓
High-Risk Items Post-Mitigation Tracking
Identified in pfmRPNPost field
- Initial PFMEA Phase: Assess detection using current manufacturing controls
- Risk Assessment Phase: Calculate pre-mitigation RPN
- Control Plan Phase: Define improved detection methods (e.g., upgrade from Rating 6 to Rating 2)
- Post-Mitigation Phase: Update detection rating, recalculate RPN to verify acceptability
Configuration in TestAuto2
Risksheet Integration
PFMEA Detection appears in Risksheet PFMEA documents as the pfmDetection field:
# Column name in PFMEA Risksheet
- fieldName: pfmDetection
type: enum
label: "Detection (Pre)"
enumRef: processFailureMode-detection-enum
required: true
description: "Current manufacturing control detection effectiveness"
pfmDetectionPost — Post-mitigation detection rating (after controls implemented)pfmSeverity + pfmOccurrence — Combined with detection to calculate pfmRPN
Configuration Properties
| Property | Value | Note |
|---|
| Enumeration | processFailureMode-detection-enum | 11 values (0-10) |
| Work Item Type | processFailureMode | PFMEA failure mode records |
| Field Name | pfmDetection | Pre-mitigation detection rating |
| Post-Field Name | pfmDetectionPost | Post-mitigation rating tracking |
| RPN Formula | pfmSeverity × pfmOccurrence × pfmDetection | Calculated field |
| Default Value | 0 (Unanalyzed) | Must be set before RPN calculation |
Links to Related Pages
Best Practices
- Start with current-state detection: Assess what controls actually exist today, not what you wish existed
- Document the control method: In control plan module, describe the gauge/process used (e.g., “100% laser gauging at Station 3B”)
- Upgrade detection before occurrence: It’s typically easier to improve detection (add gauging) than reduce occurrence (process redesign)
- Error-proofing first: If RPN is unacceptable, prioritize moving from Rating 6-9 → Rating 1-2 rather than accepting high escape risk
- Post-mitigation tracking: Always update
pfmDetectionPost and pfmRPNPost when implementing controls to show risk reduction effectiveness
- SPC as complement, not primary: Use Rating 6 (SPC) only in combination with sampling inspection, never as sole detection method for new design
Enumeration Values Reference
<!-- .polarion/tracker/fields/processFailureMode-detection-enum.xml -->
<option id="pfmDetection-0">
<name>0 - Unanalyzed</name>
<description>Detection method not yet assigned</description>
</option>
<option id="pfmDetection-1">
<name>1 - Error-Proofing (Poka-Yoke)</name>
<description>Design prevents defect creation; almost certain detection</description>
</option>
<option id="pfmDetection-2">
<name>2 - Automatic Gauging + Auto-Stop</name>
<description>Automated measurement halts production on defect</description>
</option>
<!-- ... Ratings 3-10 follow similar pattern ... -->
See Also
