Failure Modes and Effects Analysis¶
Tabular failure-analysis method — instantiates Failure Mode Anticipation
A tabular method that scores each failure mode on shared severity and detectability scales — combined with an occurrence input — into a single ranked priority, so many heterogeneous failures can be triaged by a common number.
Failure Modes and Effects Analysis (FMEA) is the method that makes a long, heterogeneous list of possible failures commensurable. Its distinguishing contribution among its siblings is the shared yardstick and the ranked score: it rates each failure mode on a standard severity scale (how bad the effect is) and a standard detectability scale (how likely existing controls are to catch it before harm), folds in an occurrence input, and combines the three into a single priority — the Risk Priority Number — so a hundred different failures can be sorted and the worst worked first. Where a scenario walk tells one failure's story and an incident review counts what has actually happened, FMEA's whole trick is the common scale: it turns "these all feel worrying in different ways" into "here is the ranked order, and here is why each landed where it did." The scales are agreed before scoring, which is what lets a corrosion failure and a software timeout be compared on one line at all.
Example¶
An automotive supplier runs an FMEA on a new electric-vehicle regenerative braking module before release. One row: regenerative torque fails to blend smoothly with friction braking during a fast pedal release. The team scores its severity high — the effect is a lurch or an unexpectedly long stop, a safety-relevant event — and its detectability low, because nothing in the current design flags the blend fault to the driver until a hard stop reveals it. The occurrence input, drawn from field data on a prior module, is moderate. Multiplied and ranked, this row jumps above a dozen cosmetic and comfort failures that the team had been treating as equally pressing. The FMEA doesn't fix anything; what it does is reorder the work — it moves "blend-fault detection" to the top of the action list precisely because the failure is severe and silent, a combination the low detectability score makes impossible to ignore.
How it works¶
- Build the scales first. A shared severity scale anchored to real consequence tiers and a shared detectability scale anchored to control strength are agreed before any row is scored, so entries stay comparable.
- Score each mode on the scales. Every failure mode gets a severity and a detectability rating; the occurrence input is brought in from evidence rather than invented at the table.
- Combine into a ranked priority. The ratings are multiplied (the Risk Priority Number) or mapped through an Action Priority table, and the rows are sorted so the highest priority surfaces first.
- Read detectability as a first-class factor. FMEA's inclusion of a detection rating — the factor most risk methods omit — is what lets a silent severe failure outrank a loud minor one.
Tuning parameters¶
- Scale granularity — a 1–10 scale versus a coarse 1–5. Finer scales discriminate more but invite false precision and endless argument over a point.
- RPN versus Action Priority — a raw multiplied number versus a lookup table that treats severity as dominant. RPN is simple but arithmetically dubious; Action Priority protects high-severity items from being averaged down.
- Action threshold — the priority level above which a mitigation is mandatory. Set it high and real risks slip under; set it low and the team drowns in low-value actions.
- Severity veto — whether a catastrophic-severity mode gets an action regardless of its score. Turning this on prevents a low occurrence from burying a rare-but-lethal failure.
- Occurrence sourcing — field data versus expert judgment for the occurrence input. Data is defensible but often unavailable for a genuinely new design.
When it helps, and when it misleads¶
Its strength is triage at scale: it makes heterogeneous failures comparable, forces the detectability question that other methods skip, and produces a defensible ranked order for scarce prevention effort. On a complex component with hundreds of possible modes, that ordering is the difference between systematic coverage and working whatever felt scariest in the meeting.
Its signature failure mode is false precision. The Risk Priority Number is an ordinal product with no real arithmetic meaning — a 720 is not "twice as bad" as a 360, and two identical RPNs can hide wildly different severities, so a coarse multiply can rank a catastrophic-but-rare failure below a trivial-but-frequent one.[n1] The classic misuse is scoring backwards — tuning the ratings until a mode lands just under the action threshold, so the number certifies a decision already made. The guarding discipline is to treat severity as a veto rather than one factor among three, to record the evidence quality behind each rating instead of trusting the digit, and to prefer an Action Priority table where the stakes justify it.
How it implements the components¶
severity_scale— builds the shared, consequence-anchored scale that lets the seriousness of otherwise incomparable failure effects be placed on one axis.detectability_estimate— rates how likely each mode is to be caught by existing controls before it causes harm — the factor FMEA is distinctively known for including.risk_priority— combines the ratings into the ranked priority (RPN or Action Priority) that orders the prevention work.
It does not estimate how often a mode actually occurs from real history — that empirical likelihood_estimate is Incident Pattern Review, its nearest data-twin, whose base rates FMEA merely consumes as its occurrence input — and it does not narrate the causal chain of any single failure (failure_cause, failure_effect), which is Failure Scenario Review.
Related¶
- Instantiates: Failure Mode Anticipation — FMEA supplies the ranked priority that tells the rest of the appraisal which failures to work first.
- Consumes: Incident Pattern Review supplies the occurrence base rates FMEA scores against; Failure Scenario Review can supply the failure modes it tabulates.
- Sibling mechanisms: Design Review · Failure Scenario Review · Incident Pattern Review · Safety Case · Fault Tree Analysis · Hazard Analysis · Premortem Workshop · Risk Register
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Failure Modes and Effects Analysis operates as a computation, comparison, model, or analytic representation used to infer, estimate, or choose because it a tabular method that scores each failure mode on shared severity and detectability scales — combined with an occurrence input — into a single ranked priority, so many heterogeneous failures can be triaged by a common number.
Independent corroboration: The frozen evidence defines Failure Modes and Effects Analysis as 'A tabular method that scores each failure mode on shared severity and detectability scales — combined with an occurrence input — into a single ranked priority, so many heterogeneous failures can be triaged by a common number', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Engineering & Design
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: FMEA is a canonical reliability-engineering method for tabulating failure modes, effects, occurrence, severity, and detectability.
Related originating lineages:
- Military & Strategic Studies — Military systems assurance materially contributed its early formal standardization.
Review resolution: Both reviewers agree that engineering_design is primary. I retain military_strategic_studies only as formative origin lineages; single_lineage is appropriate because the alternate domains informed practice without constituting independent ownership. Reach is multi_domain because the structure transfers across several fields but is not a near-universal human pattern, an applicability judgment kept separate from provenance. Encyclopedia synthesis is false because the artifact is already established enough that encyclopedia-specific synthesis is not required. No unresolved historical ambiguity remains after reconciling the secondary fields.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] The Risk Priority Number = Severity × Occurrence × Detection is FMEA's traditional scoring device, but its multiplication of ordinal ratings is mathematically ill-founded — which is why the AIAG-VDA FMEA standard replaced raw RPN thresholds with an Action Priority table that treats severity as dominant rather than just another multiplicand. ↩