FMEA vs FMECA: When Criticality Analysis Is Worth the Extra Effort

A defense prime hands back your analysis with one line in the margin: “Where’s the criticality analysis?” You thought you’d delivered it — your worksheet has severity, occurrence, detection, and an RPN column for every failure mode. But the “C” in FMECA is a separate calculation, and it needs data your RPN never touched. FMEA and FMECA are not two names for the same deliverable, and finding that out at a customer review is an expensive way to learn it.

This post compares FMEA and FMECA on the things that actually decide which one you owe: what each produces, what data each demands, and where criticality analysis is required versus where it’s effort you won’t get back. If you already rank risk with RPN or AIAG-VDA Action Priority, you’re closer to FMECA than you think — but not all the way there.

What actually separates FMEA from FMECA

FMECA is FMEA plus a criticality analysis — the failure analysis is identical up to the point where you rank risk. The split is in how the ranking is built. FMEA ranks risk qualitatively or semi-quantitatively: you rate severity, occurrence, and detection on 1–10 scales and either multiply them into a Risk Priority Number or look up an Action Priority. FMECA adds a criticality step that ranks failure modes using failure-rate data and a defined consequence probability.

The current international standard, IEC 60812:2018, “Failure modes and effects analysis (FMEA and FMECA)”, covers both methods in one document — a useful signal that the difference is a step, not a different discipline. The quantitative criticality math most reliability engineers cite still traces back to MIL-STD-1629A, the US military procedure that defined the criticality number. (MIL-STD-1629A was formally cancelled by the DoD in 1998, but it remains the de facto reference for how criticality is computed, and many aerospace and defense contracts still invoke it by name.)

FMEA: qualitative risk ranking you can run without field data

FMEA’s strength is that it works when you have no failure-rate data — which is most of the time on a new design. The team assigns occurrence from engineering judgment, prior-generation experience, or process capability, and the ranking falls out of the ratings. That’s also its weakness: multiplying three ordinal scales into an RPN treats a severity-10/occurrence-1/detection-1 failure (catastrophic but rare and caught) the same as a severity-1/occurrence-10/detection-1 nuisance. Both land at RPN 10.

The AIAG-VDA handbook fixed this for automotive by replacing RPN with Action Priority, a lookup table that weights severity first so a safety-critical failure is never buried by a low occurrence score. If you want the mechanics, see why AIAG-VDA Action Priority replaced RPN and how RPN is calculated from S, O, and D. The point for this comparison: Action Priority is itself a qualitative criticality logic. Automotive teams already do a form of criticality — they just don’t call it that, and they do it without failure-rate numbers.

FMECA: criticality adds a quantitative axis

FMECA can be run two ways. Qualitative criticality plots each failure mode on a severity-versus-probability-of-occurrence matrix — conceptually close to an Action Priority grid, and the right choice when you lack failure-rate data. Quantitative criticality computes a criticality number per mode, and that’s what a defense or space customer usually means when they ask for FMECA.

The mode criticality number from MIL-STD-1629A is:

Key Formula — Mode Criticality (C_m) $$C_m = \beta \cdot \alpha \cdot \lambda_p \cdot t$$

The terms separate cleanly. β is the conditional probability that the failure mode produces the effect you’re ranking — 1.0 means the effect is certain if the mode occurs. α is the failure mode ratio: the fraction of the part’s total failure rate attributable to this mode. λ_p is the part failure rate, and t is the operating time or mission duration. Item criticality is the sum of its modes: $C_r = \sum C_m$.

Worked Example Take a solenoid valve whose “fails-to-open” mode causes loss of function with certainty, so β = 1.0. That mode is 25% of the part’s failures, so α = 0.25. The part failure rate is λ_p = 0.6 failures per million hours (0.6 × 10^-6/hr), and the mission runs t = 1,000 hours. Then C_m = 1.0 × 0.25 × 0.6 × 10^-6 × 1,000 = 1.5 × 10^-4 expected mode-induced failures over the mission. Rank every mode this way and the highest C_m values are where reliability effort buys the most.

Common Mistake Quantitative criticality is only as good as λ_p. If your failure rate is a guess dressed up as a number, the C_m ranking inherits that guess and hides it behind two decimal places. With no defensible failure-rate source — field data, a reliability prediction, or a supplier’s reliability handbook — run qualitative criticality instead and say so. A hedged qualitative matrix beats a fabricated quantitative one.

Side-by-side: FMEA vs FMECA

Criterion	FMEA	FMECA
Deliverable	Failure modes, effects, causes, controls, risk ranking (RPN or AP)	Everything in FMEA plus a criticality ranking (matrix or C_m/C_r)
Risk metric	RPN (S×O×D) or Action Priority (H/M/L)	Criticality number, or severity-vs-probability matrix
Data required	Ratings from judgment, prior parts, or process capability	Quantitative form needs part failure rates and mode ratios
Quantification	Qualitative / semi-quantitative	Qualitative matrix or fully quantitative
Typically required by	IATF 16949 automotive, ISO 9001 risk thinking	Defense, space, and aerospace contracts; reliability programs
Effort beyond FMEA	—	Added: sourcing failure rates, computing and reviewing criticality

When the criticality step is worth the effort

Hedge this decision by who you are, not by “it depends.”

If you’re an automotive or IATF 16949 supplier with no field failure-rate data, stay with FMEA and Action Priority. AP already weights severity the way criticality intends, and bolting on a quantitative criticality number you can’t feed with real λ_p data adds review time without adding signal.

If you’re in aerospace, defense, or space and the contract names MIL-STD-1629A or asks for criticality, run FMECA. Use quantitative criticality where you have failure-rate data — a reliability prediction or field history — and a qualitative criticality matrix where you don’t. This is the same world that drives criticality analysis in aerospace FMEA programs.

If you’re a medical device team working under ISO 14971, you’re already doing criticality in spirit: risk is severity combined with probability of occurrence, scored against an acceptability matrix. FMECA’s qualitative form maps cleanly onto that; see linking failure modes to the ISO 14971 risk management file.

If you’re a reliability engineer with failure-rate data on a high-consequence system, FMECA is the point of the exercise — the criticality number is how you defend where the reliability budget goes.

Whichever you run, the rating work underneath is the same. If you want to pressure-test severity, occurrence, and detection before you layer criticality on top, our RPN and Action Priority calculator takes your S/O/D inputs and returns both the RPN and the AIAG-VDA AP level. That lets you see where a mode lands before deciding whether it earns a criticality calculation. Criticality analysis is worth its cost exactly when a failure-rate-weighted ranking would change what you act on first — and not before.