Failure Mode and Effects Analysis
Also known as: FMEA
Bottom-up identification of potential failure modes and their consequences.
Failure Mode and Effects Analysis (FMEA) is the systematic bottom-up reliability-engineering tool for identifying potential failure modes of components or process steps, analyzing their effects on the broader system, and prioritizing them for risk-mitigation action. The technique was substantially developed by US military reliability-engineering practice in the late 1940s (US Military Procedure MIL-P-1629 'Procedures for Performing a Failure Mode, Effects and Criticality Analysis' 1949) and substantially extended through NASA's substantial 1960s Apollo program reliability work, Ford's substantial post-Pinto safety adoption, and substantial automotive industry codification (Ford's 1977 internal FMEA standard, AIAG/VDA harmonized FMEA handbook 2019). FMEA's central methodology: (1) decompose system into components or process steps; (2) for each, identify potential failure modes (ways it could fail); (3) for each failure mode, identify effects on broader system, causes, and detection mechanisms; (4) score severity (S — magnitude of effect if failure occurs), occurrence (O — likelihood of failure), and detection (D — likelihood failure would be detected before causing harm); (5) compute Risk Priority Number (RPN = S × O × D) prioritizing risks for mitigation; (6) develop and track mitigation actions reducing risk. FMEA distinguishes between Design FMEA (DFMEA — analyzing product designs) and Process FMEA (PFMEA — analyzing manufacturing processes). The technique is foundational across automotive (mandatory in many supplier-quality requirements), aerospace, medical devices, pharmaceutical manufacturing, and other safety-critical industries. The 2019 AIAG/VDA harmonized FMEA substantially modified traditional RPN approach by replacing it with Action Priority (AP) ranking that addresses substantial criticism that RPN's multiplicative scoring obscures distinctions between high-severity-low-occurrence and low-severity-high-occurrence failures. Critics including substantial reliability engineers note that FMEA as practiced often produces compliance-style documentation rather than substantive risk reduction, that scoring is genuinely subjective and inconsistent across analysts, and that the technique's bottom-up perspective can miss systemic and emergent failures.
Core components
- Bottom-up analysis: components/steps → failure modes → effects → causes → detection
- Severity (S), Occurrence (O), Detection (D) scoring
- Risk Priority Number RPN = S×O×D (traditional)
- Action Priority (AP) ranking (2019 AIAG/VDA harmonized replacement)
- Distinction between Design FMEA (DFMEA) and Process FMEA (PFMEA)
- Mitigation action tracking
- Substantial automotive, aerospace, medical-device, pharmaceutical adoption
- Foundation in US military reliability engineering
Primary use case
Foundational reliability-engineering tool across automotive, aerospace, medical devices, pharmaceutical manufacturing, electronics, defense; basis for substantial supplier-quality requirements particularly in automotive (mandatory at many OEMs); reference framework in reliability-engineering education; foundation for substantial commercial FMEA-consulting and software industry; integration with broader quality-management and reliability frameworks; pedagogical foundation in engineering and quality curricula; influence on healthcare risk management, food safety, and other safety-critical contexts; foundation for ISO/IEC 31010 risk-management standard.
Common criticisms
- FMEA as practiced often produces compliance-style documentation rather than substantive risk reduction — automotive supplier-quality requirements mandate FMEA artifacts that may not substantively drive design improvement
- scoring is genuinely subjective and inconsistent across analysts — the same failure mode can receive substantially different RPN scores from different teams
- traditional RPN's multiplicative scoring substantially obscures distinctions between high-severity-low-occurrence and low-severity-high-occurrence failures (severity 10 × occurrence 1 × detection 5 = 50 same as severity 5 × occurrence 5 × detection 2 = 50, but high-severity-low-occurrence may warrant more attention) — 2019 AIAG/VDA harmonized FMEA addresses this through Action Priority (AP) ranking
- bottom-up perspective can miss systemic and emergent failures that involve component interactions or environmental conditions not captured in component-level analysis
- substantial commercial FMEA software has produced compliance-style adoption with varying analytical fidelity
- cross-cultural application varies — automotive FMEA requirements substantially shaped by Japanese and German automotive supply chains may not transfer cleanly
- integration with substantively different risk-analysis tools (Fault Tree Analysis, Bowtie Analysis) creates which-when ambiguity
- cybersecurity threats fit awkwardly into traditional FMEA framework — substantial subsequent development of cybersecurity-specific FMEA variants
- the technique works better for hardware components with well-understood failure modes than for software, where failure modes are often emergent and configuration-dependent
- AI/ML-enabled systems raise substantial questions about whether traditional FMEA framework adequately addresses emergent-behavior risks.
Lineage
- Child of
- Reliability Engineering
- Siblings
- Fault Tree Analysis, Bowtie Analysis
- Derived from
- Reliability Engineering