FMEA — Failure Mode and Effects Analysis
A complete interactive FMEA guide: RPN calculator, Severity/Occurrence/Detection ratings, a sample FMEA table, DFMEA vs PFMEA, the modern AIAG-VDA Action Priority (AP) method, Fault Tree Analysis (FTA), complementary tools (Fishbone, 5-Why, Pareto, 8D), and the key FMEA standards.
🧮 RPN Calculator
—
RPN = S × O × D · Maximum = 1000 · Action threshold: RPN>100 (general) / >50 (safety)
⚠️ RPN Limitations — why AIAG-VDA 2019 dropped it
Problem 1: S=9×O=1×D=1 = RPN of just 9 — but S=9 (life-threatening!) must always be addressed; a low RPN "hides" this.
Problem 2: S=2×O=5×D=5 = RPN 50 — the same number can arise from many completely different combinations.
Problem 3: No clear breakpoints — RPN 99 vs RPN 101 aren't really different, yet a "threshold of 100" treats them differently.
Problem 2: S=2×O=5×D=5 = RPN 50 — the same number can arise from many completely different combinations.
Problem 3: No clear breakpoints — RPN 99 vs RPN 101 aren't really different, yet a "threshold of 100" treats them differently.
The modern solution: AIAG/VDA Action Priority (AP) — see dedicated tab. High S = mandatory action, regardless of O/D!
📊 Severity (S) — impact severity
| S | Description | Example |
|---|---|---|
| 1 | No effect | Cosmetic change only |
| 2-3 | Slight — customer notices occasionally | Slight noise, peeling label |
| 4-6 | Moderate — performance degradation | Partial function loss, delay |
| 7-8 | High — loss of function | Product failure, return |
| 9 | Critical — safety without warning | Possible hazardous failure |
| 10 | Catastrophic | Certain hazardous failure |
⚠️ S cannot be reduced by controls — only a design change lowers S! S=9-10 = a design red flag that always demands attention.
📊 Occurrence (O) — failure occurrence frequency
| O | Description | Approx. Cpk |
|---|---|---|
| 1 | Almost impossible | Cpk>1.67 |
| 2-3 | Rare — well-controlled process | Cpk 1.33-1.67 |
| 4-6 | Occasional — reasonable process | Cpk 1.0-1.33 |
| 7-8 | High — unstable process | Cpk 0.67-1.0 |
| 9-10 | Almost certain | Cpk<0.67 |
Lowering O: process improvement, Poka-Yoke, design change. Based on real historical data/Cpk — not subjective estimation!
📊 Detection (D) — ability to detect before reaching the customer
| D | Description | Detection Method |
|---|---|---|
| 1 | Certain detection | Physical Poka-Yoke — cannot assemble incorrectly |
| 2-3 | High detection | 100% automated inspection |
| 4-6 | Moderate detection | Statistical inspection (SPC), sampling |
| 7-8 | Low detection | Visual inspection, operator-dependent |
| 9-10 | Almost undetectable | No control, only found in the field |
⚠️ Note the reversed direction! D=1 = excellent detection (good). D=10 = not detected (bad). Improving D is cheaper than improving O — but preventing is better than detecting.
📋 Sample FMEA Table — Shaft Bearing
| ID | Function | Failure Mode | Effect | S | Cause | O | Controls | D | RPN | Action |
|---|---|---|---|---|---|---|---|---|---|---|
| 1.1 | Shaft support | Bearing wear | Noise, vibration, shaft failure | 7 | High load/speed | 3 | Vibration check | 5 | 105 | ↑lubrication frequency |
| 1.2 | Shaft support | Cage failure | Shaft seizure | 9 | Sudden overload | 2 | Visual inspection | 7 | 126 | ↑C/P, ↓load |
| 1.3 | Shaft support | Corrosion | Dimensional growth | 5 | Moisture, lack of lubrication | 4 | Annual inspection | 4 | 80 | Grease upgrade |
| 1.4 | Sealing | Oil leak | Environmental hazard | 4 | Seal wear | 5 | Visual inspection | 3 | 60 | Routine monitoring |
📝 Full FMEA Form Structure — all required fields
Identification: Item number · Function · Requirement
Failure: Potential failure mode · Effect on customer · Severity (S)
Cause: Root cause · Failure mechanism · Occurrence (O)
Current control: Prevention control · Detection control · Detection rating (D)
Calculation: RPN (or AP) · Priority ranking
Recommended action: Responsible person · Target date · Action taken · New S/O/D after action
Failure: Potential failure mode · Effect on customer · Severity (S)
Cause: Root cause · Failure mechanism · Occurrence (O)
Current control: Prevention control · Detection control · Detection rating (D)
Calculation: RPN (or AP) · Priority ranking
Recommended action: Responsible person · Target date · Action taken · New S/O/D after action
DFMEA — Design FMEA
Examines the design — before the part physically exists.
Asks: How can the design itself fail? Wrong material, incorrect tolerance, unexpected load, stress concentration.
Performed by: Design engineer, during development — before production!
Output: Affects material selection, GD&T, geometry.
Asks: How can the design itself fail? Wrong material, incorrect tolerance, unexpected load, stress concentration.
Performed by: Design engineer, during development — before production!
Output: Affects material selection, GD&T, geometry.
PFMEA — Process FMEA
Examines the manufacturing process — how the part is actually produced.
Asks: Where in the process (machining/assembly/inspection) could a failure occur? Worn tool, wrong machine setting, operator error.
Performed by: Process/manufacturing engineer, before series production starts.
Output: Affects the Control Plan, inspections, in-line Poka-Yoke.
Asks: Where in the process (machining/assembly/inspection) could a failure occur? Worn tool, wrong machine setting, operator error.
Performed by: Process/manufacturing engineer, before series production starts.
Output: Affects the Control Plan, inspections, in-line Poka-Yoke.
🔗 The link between DFMEA and PFMEA
DFMEA precedes PFMEA chronologically. Critical characteristics identified in DFMEA (Key Characteristics) become mandatory control points in PFMEA. Both documents are living — updated with every design/process change and field failure.
🆕 AIAG/VDA 2019 — Action Priority (AP)
The new standard (2019) replaces the raw RPN calculation with a decision table: High / Medium / Low — based primarily on S, then O, then D.
| S | O | D | Action Priority |
|---|---|---|---|
| 9-10 | 4+ | any | High |
| 9-10 | 2-3 | 5+ | High |
| 9-10 | 2-3 | 1-4 | Medium |
| 7-8 | 4+ | any | High |
| 7-8 | 1-3 | any | Medium |
| 1-6 | 1-3 | 1-3 | Low |
⚠️ The key difference: S=9 with low O still comes out High — it doesn't "hide" behind a low RPN number like in the old method! This is the main reason for the switch.
High — action mandatory
A documented corrective action is required. If no action is taken, a written, reasoned justification is required.
Medium — action recommended
Action recommended at the team's discretion. Documenting the decision (act/don't act) is still required.
Low — no obligation
Can be left as-is, but continue monitoring. Doesn't mean "safe" — only that relative priority is low.
🌳 FTA — Fault Tree Analysis
▸ FTA method: from the top failure ↓ down to basic causes (Top-Down)
▸ FMEA method: from a component ↑ up to system effect (Bottom-Up)
▸ Gates: AND = both · OR = either
▸ P(AND) = P₁ × P₂ (independent)
▸ P(OR) = 1-(1-P₁)(1-P₂)
▸ MIL-STD-1629A: military/defense FMEA standard
▸ IEC 60812: civilian FMEA standard
▸ AIAG/VDA FMEA: automotive, 2019+
▸ FMEA method: from a component ↑ up to system effect (Bottom-Up)
▸ Gates: AND = both · OR = either
▸ P(AND) = P₁ × P₂ (independent)
▸ P(OR) = 1-(1-P₁)(1-P₂)
▸ MIL-STD-1629A: military/defense FMEA standard
▸ IEC 60812: civilian FMEA standard
▸ AIAG/VDA FMEA: automotive, 2019+
✅ Criticality Levels (MIL-STD-1629)
| Category | Description | Min SF |
|---|---|---|
| I — Catastrophic | Death/severe injury/system loss | 4-10 |
| II — Critical | Injury, major mission damage | 3-6 |
| III — Marginal | Minor injury, delay | 2-3 |
| IV — Minor | Inconvenience only | 1.5-2 |
🌳 Sample Fault Tree — oil supply failure
Top Event: Oil supply failed
├─ OR ─┬─ Pump failed
├─ AND ─┬─ Motor failed
└─ Backup not activated
└─ Filter clogged
└─ Pipeline cracked
├─ OR ─┬─ Pump failed
├─ AND ─┬─ Motor failed
└─ Backup not activated
└─ Filter clogged
└─ Pipeline cracked
OR gate = any single branch is enough to cause the overall failure. AND gate = both branches must occur together — true redundancy!
🐟 Fishbone / Ishikawa
Maps root causes across the 6M's: Man, Machine, Method, Material, Measurement, Environment. Systematic — doesn't get stuck on the first suspect.
❓ 5-Why
Ask "why" until you reach the true root cause — usually 5 times. Trap: "operator error" is almost never the root — why did the system allow it?
📊 Pareto Analysis
80% of problems come from 20% of causes. Focuses where to invest — by cost, not just count!
8️⃣ 8D Problem Solving
D1 Team→D2 Description→D3 Containment (24h!)→D4 Verified root cause→D5 Corrective action→D6 Implementation→D7 Systemic prevention→D8 Recognition.
🔗 When to use which tool?
| Situation | Recommended Tool |
|---|---|
| New product/process design (preventive) | DFMEA / PFMEA |
| Recurring fault, cause unclear | 5-Why → Fishbone |
| Many defect types, unsure what to tackle first | Pareto Analysis |
| Formal customer complaint (automotive/aerospace) | Full 8D |
| Severe safety failure, multiple failure paths | FTA |
| Standard | Domain | Notes |
|---|---|---|
| MIL-STD-1629A | Military/aerospace | The historical standard — Criticality Analysis (I-IV) |
| IEC 60812 | General civilian | The official international FMEA standard |
| AIAG/VDA FMEA (2019) | Automotive | Replaces RPN with Action Priority. Mandatory under IATF 16949 |
| ARP5580 | Aerospace (SAE) | Based on MIL-STD-1629A, updated |
| AS9100 | Aerospace — QMS | Requires FMEA as part of risk management |
Choose the standard per customer/industry requirement — most automotive today has moved to AIAG/VDA 2019, while aerospace still moves between MIL-STD-1629A and ARP5580.