What FMEA actually is
FMEA (Failure Mode and Effects Analysis) is a structured, team-based method for identifying the ways a product or process could fail, judging the risk of each, and acting on the highest risks before they reach the customer. It is one of the five IATF 16949 core tools and the analytical engine behind the control plan: the failure modes it prioritises become the characteristics the control plan then controls.
Every FMEA answers three questions for each potential failure: what is the effect and how bad is it (severity), what is the cause and how often will it happen (occurrence), and how likely are we to catch it before it escapes (detection). Those three judgements, each on a 1–10 scale, drive the prioritisation.
DFMEA vs PFMEA
The two FMEAs most automotive suppliers run differ in what they analyse. A DFMEA examines the design; a PFMEA examines the manufacturing process. They are complementary — the DFMEA is usually done first and hands special characteristics to the PFMEA.
| Aspect | DFMEA (Design) | PFMEA (Process) |
|---|---|---|
| Focus | How the design could fail to meet its function | How the manufacturing process could produce a non-conforming part |
| Failure mode | e.g. shaft fractures under load | e.g. bore machined oversize |
| Cause | Material choice, geometry, tolerance stack-up | Tool wear, fixturing, setup, machine drift |
| Controls | Design verification, analysis, testing | Prevention (poka-yoke) and detection (inspection, SPC) |
| Owned by | Design / product engineering | Process / manufacturing engineering |
| Output | Special characteristics, design actions | Special characteristics onto the control plan |
Severity, occurrence and detection
Each failure mode is rated on three 1–10 scales, where 10 is always the worst case:
- Severity (S) — the seriousness of the effect, from 1 (no discernible effect) to 10 (a safety or regulatory hazard with no warning). Severity is a property of the effect and cannot be reduced by more inspection — only a design change lowers it.
- Occurrence (O) — how frequently the cause is expected to happen, from 1 (extremely unlikely) to 10 (almost inevitable). Occurrence is lowered by prevention controls that attack the cause.
- Detection (D) — the ability of current controls to catch the failure before it escapes, from 1 (certain detection) to 10 (no control, cannot detect). Better inspection or SPC lowers detection.
RPN and Action Priority (AP)
The classic AIAG method multiplies the three ratings into a Risk Priority Number: RPN = S × O × D, ranging from 1 to 1000. Teams historically set an RPN threshold above which action is required. The weakness of RPN is that very different risks can share a number — an RPN of 100 from S=10, O=2, D=5 (a safety issue) is not the same as 100 from S=4, O=5, D=5, yet the raw number treats them alike.
The 2019 AIAG-VDA harmonised FMEA replaced the RPN threshold with Action Priority (AP). AP uses a lookup table across the S, O and D ratings to assign each failure mode a priority of High, Medium or Low — and it always weights severity first, so a high-severity failure gets attention regardless of how rare or detectable it is. Many suppliers still record the RPN for continuity while using AP to decide what to work on.
| Method | How it prioritises | Key limitation / strength |
|---|---|---|
| RPN (classic AIAG) | S × O × D, compared to a threshold | Simple, but different risks share the same number |
| Action Priority (AIAG-VDA) | Lookup table → High / Medium / Low | Weights severity first; more defensible prioritisation |
The AIAG-VDA 7-step method
The harmonised standard frames the FMEA as seven steps rather than a single spreadsheet: 1) Planning and preparation (scope and boundaries), 2) Structure analysis (break the system into elements), 3) Function analysis (what each element must do), 4) Failure analysis (failure modes, effects and causes), 5) Risk analysis (rate S, O, D and derive AP), 6) Optimisation (define and implement actions to reduce risk), and 7) Results documentation (record the residual risk and communicate it). Steps six and seven make clear that an FMEA is a living document, re-scored whenever a corrective action changes the process.
From FMEA to the control plan
The FMEA is not an island. A high-risk failure mode — especially one tagged as a special (significant) characteristic because it affects safety, fit or function — must flow onto the control plan, where it becomes a defined check with a gauge, a frequency and a reaction plan. When that characteristic later fails on the line, the defect code links the rejection back to the FMEA, and the 8D that follows re-evaluates the occurrence and detection ratings. Break any link in that chain and the FMEA stops reflecting reality.
How Fast Quality Software runs FMEA
Fast Quality Software carries DFMEA and PFMEA as core-tool records linked to the item and its specifications, so the analysis, the control and the reaction stay connected.
Because the FMEA, the control plan and the defect masters share one platform, a special characteristic flagged in the PFMEA is the same characteristic the control plan verifies and the same defect code a rejection carries — so the loop back to risk is always traceable. See it applied in casting and plastics and precision machining, or review pricing.
Frequently asked questions
What is the difference between DFMEA and PFMEA?
A DFMEA (Design FMEA) analyses how a product design could fail to perform its intended function, looking at causes such as material, geometry and tolerance stack-up. A PFMEA (Process FMEA) analyses how the manufacturing process could produce a non-conforming part, looking at causes such as tool wear, fixturing and setup. The DFMEA is usually completed first and hands its special characteristics to the PFMEA, which in turn feeds the control plan.
How is RPN calculated in an FMEA?
RPN, the Risk Priority Number, is calculated by multiplying the three 1-to-10 ratings: RPN = Severity x Occurrence x Detection, giving a value from 1 to 1000. A higher RPN indicates higher risk. The main limitation is that very different combinations of S, O and D can produce the same RPN, which is why the AIAG-VDA standard introduced Action Priority as a more defensible alternative.
What is Action Priority (AP) in FMEA?
Action Priority is the prioritisation method introduced in the 2019 AIAG-VDA harmonised FMEA. Instead of a single RPN threshold, AP uses a lookup table across the severity, occurrence and detection ratings to assign each failure mode a priority of High, Medium or Low. AP always weights severity first, so a high-severity failure receives attention regardless of how rare or detectable it is.
What are severity, occurrence and detection?
They are the three 1-to-10 ratings in an FMEA. Severity rates how serious the effect of a failure is; occurrence rates how frequently the cause is expected to happen; and detection rates how likely current controls are to catch the failure before it escapes. In every scale, 10 is the worst case. Severity can only be reduced by a design change, occurrence by prevention controls, and detection by better inspection or SPC.
What are the 7 steps of the AIAG-VDA FMEA?
The AIAG-VDA FMEA has seven steps: 1) planning and preparation, 2) structure analysis, 3) function analysis, 4) failure analysis, 5) risk analysis (rating S, O and D and deriving Action Priority), 6) optimisation (defining and implementing risk-reduction actions), and 7) results documentation. The last two steps make the FMEA a living document that is updated whenever corrective action changes the process.