FMEA RPN calculation: severity × occurrence × detection, done right.
The Risk Priority Number is how a FMEA turns a list of possible failures into a ranked action list. It is three 1-to-10 ratings multiplied together — but the number is only as honest as the ratings behind it, and a high RPN is not always the failure you should fix first. Here is the formula, a worked example, and why AIAG-VDA replaced RPN with Action Priority.
What RPN is
RPN stands for Risk Priority Number. In a Failure Mode and Effects Analysis you list every way a process or product can fail, and for each failure mode you score three things: how bad the effect is (severity), how often the cause happens (occurrence), and how likely your controls are to catch it before it reaches the customer (detection). RPN multiplies the three into a single risk score that lets the team rank where to act.
The whole point is prioritisation. A process FMEA on a machined part can throw up forty failure modes; you cannot fix all forty at once. RPN sorts them so engineering effort goes where the risk is highest rather than where the loudest person in the room wants it.
The RPN formula
Because all three factors share the same 1-to-10 scale but mean very different things, two failure modes can reach the same RPN by completely different routes. An RPN of 200 from severity 10 is a safety problem; an RPN of 200 from detection 10 is an inspection problem. The number alone hides which — which is exactly the weakness that Action Priority was created to fix.
Rating the three factors
Consistent ratings are everything. A team that scores severity loosely produces RPNs that cannot be compared across failure modes. Use defined rating tables and agree them before scoring:
| Rating | Severity | Occurrence | Detection |
|---|---|---|---|
| 1 | No noticeable effect | Failure unlikely (<1 in 1,500,000) | Control almost certainly detects |
| 2–3 | Minor, slight annoyance | Low (1 in 150,000) | High chance of detection |
| 4–6 | Moderate, customer dissatisfied | Moderate (1 in 2,000) | Moderate chance |
| 7–8 | High, loss of primary function | High (1 in 50) | Low chance of detection |
| 9–10 | Safety / regulatory, no warning | Very high (1 in 8) | Control cannot detect |
Severity is fixed by the effect and can only be reduced by a design change, not by better inspection. Occurrence is reduced by attacking the root cause. Detection is reduced by better controls — but adding inspection is the weakest of the three improvements because it catches failures instead of preventing them.
A worked PFMEA example
Take a drilling operation on a Pune auto-component line. The failure mode is "hole drilled undersize", the effect is "bolt will not assemble", and the cause is "drill wear not detected".
| Factor | Rating | Reasoning |
|---|---|---|
| Severity | 7 | Loss of assembly function at the customer |
| Occurrence | 6 | Drill wear is moderately common without a change schedule |
| Detection | 8 | Only a final visual check; no in-process bore gauge |
RPN = 7 × 6 × 8 = 336. That is high. Now look at where the leverage is: severity (7) is locked unless the design changes, but detection (8) and occurrence (6) are both addressable. Adding an in-process go/no-go bore gauge pulls detection from 8 to 3, and a tool-change schedule pulls occurrence from 6 to 3. New RPN = 7 × 3 × 3 = 63 — an 81 percent risk reduction without touching the design.
RPN vs Action Priority (AIAG-VDA)
The 2019 AIAG-VDA FMEA handbook replaced RPN with Action Priority (AP) for a reason: a raw RPN treats all three factors as equal, but they are not. A failure with severity 10 deserves attention even at a low RPN, because the consequence is catastrophic. AP is a lookup table that weights severity first, then occurrence, then detection, and outputs a priority of High, Medium or Low.
| Method | Output | Weakness it fixes |
|---|---|---|
| RPN | Number 1–1000 | — (the legacy method) |
| Action Priority | High / Medium / Low | Stops a high-severity item hiding behind a low RPN |
Common FMEA RPN mistakes
- Treating RPN as a pass/fail gate. There is no magic threshold; RPN ranks, it does not certify. Chasing "RPN below 100" invites rating manipulation.
- Improving detection instead of occurrence. Adding inspection lowers the number but the failure still happens. Attack the cause first.
- Inconsistent rating scales. Different teams scoring severity differently make RPNs incomparable. Use one agreed rating table.
- Ignoring high-severity, low-RPN items. The exact failure Action Priority was built to catch.
- Never updating the FMEA. RPN is a living estimate; recalculate after every process change or field failure.
Connecting FMEA to inspection and control
The detection rating in a FMEA is a direct claim about your inspection system — and that claim is only credible if the inspection is real and traceable. When a high-detection-rating failure mode demands a new check, that check has to land on the inspection sheet against a specific characteristic. CadNexa's auto-ballooning tool turns a drawing into a numbered characteristic list, so the controls your FMEA promises actually exist on the floor rather than on paper.
FMEA feeds directly into the control plan, and the detection ratings depend on capable measurement — verify yours with a Gauge R&R study. Standard FMEA and control-plan forms are on the templates page.