What is FMEA?
FMEA (Failure Mode and Effects Analysis) is a structured way to ask three questions about a design or a process before anything goes wrong:
- How could it fail? (the failure mode)
- What happens if it does? (the effect)
- Why would it happen, and what stops it? (the cause and your current controls)
Then you rate the risk, rank the list, and fix the worst items first. That's the whole idea. Everything else is detail.
FMEA is a prevention tool. Root cause analysis and 8D are what you use after a problem has reached a customer. FMEA is what you use so it doesn't get there in the first place.
DFMEA vs. PFMEA
You'll mostly see two kinds:
| Type | Looks at | Typical question |
|---|---|---|
| DFMEA (Design FMEA) | The product design: parts, materials, dimensions, interfaces | "What if this bracket is too thin to carry the load?" |
| PFMEA (Process FMEA) | The manufacturing or assembly process, step by step | "What if the operator installs the seal upside down?" |
The worksheet columns are almost the same. The difference is what you're analyzing. If you work on a production floor, you'll spend most of your time in PFMEAs, which is what the free template on this page is set up for.
The key terms (plain language)
| Term | What it means |
|---|---|
| Function / requirement | What the step or part must do, ideally with a spec. "Press bearing to a depth of 12.0 ± 0.2 mm." |
| Failure mode | The way it fails to do that. "Bearing pressed too shallow." |
| Effect | What the next operation, the customer or the end user experiences. "Noise, early bearing wear." |
| Cause | Why the failure mode would happen. "Press stop worn." |
| Prevention control | What stops the cause from happening. "Press stop checked weekly by maintenance." |
| Detection control | What catches the failure (or the cause) before it ships. "Depth gauge check on 100% of parts." |
| S | Severity, 1 to 10: how bad the effect is. 9–10 usually means a safety or legal issue. |
| O | Occurrence, 1 to 10: how likely the cause is, given your prevention controls. |
| D | Detection, 1 to 10: how likely you are to miss it. Low is good: 1 means you'd almost certainly catch it. |
Action Priority: how risk gets ranked
Once every row has its three ratings, you need to decide which rows to work on first. The AIAG-VDA FMEA Handbook (released jointly by the North American and German automotive groups in 2019) does this with Action Priority (AP).
S, O and D → AIAG-VDA AP table → High / Medium / Low
AP doesn't multiply anything. You look up the combination of Severity, Occurrence and Detection in the handbook's table, and it gives the row a High, Medium or Low priority. The table weights Severity first, then Occurrence, then Detection, so a serious effect with a likely cause comes out High even when your inspection looks good.
- High: the team must act to improve prevention or detection, or write down why the current controls are enough.
- Medium: the team should act, or document why not.
- Low: the team could act.
How to do an FMEA in 7 steps
The AIAG-VDA handbook formalizes FMEA into seven steps (Planning & Preparation, Structure Analysis, Function Analysis, Failure Analysis, Risk Analysis, Optimization, Results Documentation). Here's the same flow in plain words:
- Plan it. Pick the scope (one process or one design), the team, and the date. A good team mixes the people who run the job with the people who engineer it: operator, engineer, quality, maintenance.
- Map the structure. List the process steps in order (or the parts of the design). One row group per step.
- Write the function of each step. What must it achieve? Add the spec where you can.
- Find the failures. For each function, ask "how could this fail?" Then write the effect and the cause for each failure mode.
- Rate the risk. Write your current prevention and detection controls, then rate S, O and D. Rate what's actually in place today, not what you hope to have.
- Act on the worst items. High Action Priority and Severity 9–10 first, then Medium. Every action gets an owner and a date.
- Re-rate and keep it alive. When the action is done, re-rate. Update the FMEA whenever the process changes, a customer complains, or an 8D finds a new cause.
A worked example
Here's one row from a process FMEA for tightening wheel nuts on an assembly line. The ratings are illustrative, made up to show the method.
| Field | Before action | After action |
|---|---|---|
| Function | All 5 nuts torqued to spec | |
| Failure mode | Nut under-torqued | |
| Effect | Wheel can loosen in use; driver safety risk | |
| Cause | Wrong torque program selected on the tool | |
| Prevention | Program checked at start of shift | Tool reads the vehicle ID and loads the right torque program automatically |
| Detection | Operator listens for the tool "click"; 1 car per shift audited | Torque-controlled tool with an interlock: the car can't move on until all 5 nuts read OK |
| S / O / D | 10 / 4 / 7 | 10 / 2 / 2 |
| Action Priority | H (High) | L (Low) |
Three things to notice:
- Severity stays at 10. A loose wheel is just as bad after the fix. Severity normally only drops if the design changes.
- The actions lowered Occurrence and Detection. Automatic program selection fixes the cause, so the wrong program can't be picked (O 4 → 2). The interlock is the safety net, so a bad nut can't leave the station (D 7 → 2).
- AP drove the action. Severity 10 with Occurrence 4 is High priority however good the inspection looks. After the actions, the table gives Low. Look up real values in the AIAG-VDA table for your own work.
This exact row is on the "Example" tab of the free worksheet.
Common FMEA mistakes
- One person fills it in alone. The operator on the line knows failure modes an engineer at a desk never sees.
- Rating to make the numbers look good. A low Occurrence with no real prevention control behind it only hides risk.
- Chasing a better Detection score. Adding one more visual inspection to lower D looks good on paper. Fixing the cause (lowering O) is usually the better action, and the AP table rewards it.
- Mixing up effects, modes and causes. Keep the chain straight: cause → failure mode → effect.
- Write it once, never look at it again. An FMEA is a living document. When an 8D finds a cause that wasn't on the list, add it.
FMEA questions people ask
What does FMEA stand for?
Failure Mode and Effects Analysis.
What is Action Priority and why did it replace RPN?
Action Priority (AP) is how the 2019 AIAG-VDA FMEA Handbook ranks risk: you look up the Severity, Occurrence and Detection ratings in a table that returns High, Medium or Low. It replaced the Risk Priority Number because multiplying S × O × D treats the three ratings as equally important, so very different risks could get the same score and a Severity 10 item could hide behind a low number. AP weights Severity first, then Occurrence, then Detection.
What's the difference between FMEA and FMECA?
FMECA (Failure Mode, Effects and Criticality Analysis) adds a formal criticality analysis on top of FMEA. You'll see it more in military, aerospace and reliability work. In manufacturing and automotive, FMEA is the common form.
Who should be on an FMEA team?
The people who know the product and the process: typically an engineer, someone from quality, an operator or line lead, and maintenance. Bring in others (suppliers, design) for the steps that involve them.
How does FMEA connect to the control plan?
The controls you list in the PFMEA should show up in the control plan: what to check, how often, and what to do when it's out. If a control is in one and not the other, something's missing.
