FMEA Template Guide: Process FMEA for UK Manufacturing SMEs
What Is FMEA?
Failure Mode and Effects Analysis (FMEA) is a structured technique for identifying ways a process or product can fail, working out how bad each failure would be, and prioritising the riskiest failures for action before they happen. Unlike a fishbone diagram or a 5 Whys (which look at problems that already occurred — see our root cause analysis comparison), FMEA is preventive — it asks "what could go wrong here, and what would the impact be?"
There are two main FMEA variants used in UK manufacturing:
- Design FMEA (DFMEA) — applied to a product design before it is released. Identifies design weaknesses that could lead to field failures.
- Process FMEA (PFMEA) — applied to a manufacturing process. Identifies process steps where defects, errors, or variation could occur.
This guide focuses on Process FMEA because most UK SME manufacturers buy or license designs from customers and own the manufacturing process. The same scoring logic applies to DFMEA — only the failure modes change.
A downloadable PFMEA template (Excel + PDF) is in development for the next content cycle. This guide explains the scoring methodology so the template makes sense before you fill it in.
When to Run a PFMEA
UK manufacturers typically run a PFMEA in five situations:
- New product introduction — before first production run, the engineering team works through every process step to identify where things could go wrong
- Process change — if a process is materially changed (new equipment, new sequence, new operator team), the existing PFMEA is reviewed and updated
- After a serious quality event — a customer complaint, recall, or repeated non-conformance often triggers a PFMEA review focused on the failure area
- Customer requirement — automotive customers (IATF 16949), aerospace customers (AS9100), and medical device customers (ISO 13485) often require a PFMEA as part of supplier approval
- Periodic review — many ISO 9001 management systems schedule annual PFMEA reviews on critical processes
The PFMEA is a living document. The first version is rarely the last.
The PFMEA Structure
A PFMEA is built around a single table where each row captures one potential failure mode. The minimum columns are:
| Column | Purpose |
|---|---|
| Process step | The specific step being analysed (e.g. "drill 5mm hole", "apply primer coat", "torque M8 fastener") |
| Potential failure mode | How this step could fail (e.g. "hole drilled in wrong position", "primer applied to dirty surface", "fastener under-torqued") |
| Potential effect of failure | What happens if this failure occurs (e.g. "part rejected at inspection", "coating peels in customer use", "joint loosens during operation") |
| Severity (S) | How serious the effect is — scored 1 to 10 |
| Potential causes | Why this failure might happen (e.g. "drill bit wear", "operator skipped clean step", "torque wrench out of calibration") |
| Occurrence (O) | How often this cause is expected to produce the failure — scored 1 to 10 |
| Current controls | What is in place today to prevent or detect this failure (e.g. "in-process gauge check", "supervisor pre-flight inspection", "torque verification on every 10th part") |
| Detection (D) | How likely current controls are to catch the failure before it reaches the customer — scored 1 to 10 |
| RPN | Risk Priority Number = S × O × D |
| Recommended action | What to change to reduce S, O, or D |
| Action responsibility + due date | Who owns the change, by when |
| Revised S, O, D, RPN | Re-scored after action is implemented |
The first eight columns describe today's state. The remaining columns drive the improvement loop.
The Three Scoring Scales
The 1-10 scoring is where most PFMEAs go wrong. Teams either treat the scales casually (everything gets a 5) or argue about half-point differences. UK manufacturing SMEs typically adopt a published scale once and stick with it — the relative scoring within one PFMEA matters far more than absolute precision across PFMEAs.
Severity (S) — Impact on the Customer
| Score | Description (UK manufacturing SME interpretation) |
|---|---|
| 1 | No discernible effect — customer unlikely to notice |
| 2-3 | Minor effect — slight customer inconvenience, no functional loss |
| 4-6 | Moderate effect — customer dissatisfied, partial loss of function, may require rework or return |
| 7-8 | High effect — customer experiences major function loss; product non-conforming to specification |
| 9 | Very high effect — non-compliance with regulation, safety risk possible, warranty claim likely |
| 10 | Hazardous effect — endangers operator, end user, or compliance status; potential recall |
Severity is set by the failure effect, not the cause. A drilled hole 0.1mm off-centre that produces a part the customer can still fit may score 3. The same drilling step producing a hole that cracks the part may score 8 — same process step, different failure mode, very different severity.
Occurrence (O) — How Often the Cause Produces the Failure
| Score | Description | Typical PPM (parts per million) |
|---|---|---|
| 1 | Failure unlikely — comprehensive controls; cause virtually eliminated | < 1 PPM |
| 2-3 | Low probability of occurrence | 1-100 PPM |
| 4-6 | Moderate probability — occasional failures expected | 100-5,000 PPM |
| 7-8 | High probability — repeated failures likely | 5,000-50,000 PPM |
| 9 | Very high probability — failure almost inevitable | 50,000-500,000 PPM |
| 10 | Failure certain — no controls preventing the cause | > 500,000 PPM |
The PPM values are not legal definitions — they are anchor reference points published in AIAG and SAE FMEA handbooks. Some UK SME PFMEAs use process data (Cpk values, in-process reject rates) to set occurrence; others use engineering judgement when no production history exists. Either approach works, provided the same logic is applied to every row.
Detection (D) — How Likely Current Controls Catch the Failure
| Score | Description |
|---|---|
| 1 | Almost certain detection — automated 100% inspection, error-proofing (poka-yoke), or inspection physically impossible to skip |
| 2-3 | Very high detection — automated 100% inspection with high reliability |
| 4-6 | Moderate detection — sample inspection or operator visual check |
| 7-8 | Low detection — controls rely on operator vigilance or downstream catches |
| 9 | Very low detection — controls unlikely to catch the failure |
| 10 | No detection — no controls exist, or controls cannot detect this failure mode |
Detection is counter-intuitive — a higher score means worse detection. This is so that the RPN multiplication (S × O × D) yields a higher score for higher-risk failures.
Calculating the Risk Priority Number (RPN)
RPN = Severity × Occurrence × Detection.
The minimum RPN is 1 × 1 × 1 = 1. The maximum is 10 × 10 × 10 = 1000. Most PFMEAs find RPNs distributed between 20 and 300, with a handful of high-risk rows above 300.
There is no universal "action threshold" — different industries use different cutoffs. UK manufacturing SME PFMEAs typically use:
- RPN ≥ 100 — action recommended
- Severity ≥ 9 — action required regardless of RPN (because a low-RPN safety/recall risk is still a risk)
- Occurrence ≥ 7 — process capability action required regardless of RPN
- Detection ≥ 7 — control improvement required regardless of RPN
This approach prevents the most common PFMEA failure mode: dismissing a Severity 10 row because its RPN is "only" 80 (S=10, O=2, D=4) when the issue is a safety risk that should be designed out regardless of probability.
The AIAG-VDA harmonised FMEA approach (introduced in the AIAG-VDA FMEA Handbook, 2019) replaces RPN with Action Priority (AP) — High, Medium, or Low — based on a lookup table combining S, O, and D. UK automotive Tier 1 suppliers usually follow the AIAG-VDA approach; non-automotive SMEs typically still use RPN. Either is acceptable provided the chosen method is documented.
A Worked Example
Process step: applying adhesive to a bonding surface before final assembly.
| Item | Detail |
|---|---|
| Process step | Apply adhesive (2-part epoxy) to bonding face |
| Potential failure mode | Adhesive applied to contaminated surface |
| Potential effect | Bond fails in service — joint separation under load |
| Severity | 8 (major customer function loss, potential warranty claim) |
| Potential cause | Operator skips solvent wipe before adhesive application |
| Occurrence | 5 (moderate — relies on operator following work instruction; observed 3 times in last 12 months) |
| Current controls | Visual check by operator; final assembly torque test catches gross bond failures |
| Detection | 7 (visual check unreliable; torque test only catches catastrophic bond failures, not partial contamination) |
| RPN | 8 × 5 × 7 = 280 |
| Recommended action | Add poka-yoke: solvent wipe station with operator-presence sensor and timed wipe sequence before adhesive station opens |
| Owner / Due | Production engineer / 8 weeks |
| Revised S, O, D, RPN | 8 × 2 × 3 = 48 |
The action does not change severity — the failure effect (joint separation) is unchanged. The action reduces occurrence (poka-yoke prevents the operator skipping the wipe) and improves detection (presence sensor catches process skip in real time). RPN drops from 280 to 48 — risk significantly reduced.
This worked example shows the PFMEA logic in practice. Real PFMEAs typically contain 30-150 rows for a moderately complex process.
Common PFMEA Mistakes
UK certification body auditors and customer supplier-development teams routinely find the same PFMEA weaknesses:
- Generic failure modes — "incorrect dimension" with no specification of which dimension, what specifically goes wrong, or what the effect is. Each failure mode should be process-step-specific.
- Cause and failure mode confused — "drill bit broken" is a cause, not a failure mode. The failure mode is "hole undersized" or "hole missing entirely". The cause is why.
- Severity changed after rework — severity is a property of the failure effect, not the process. Rework reduces occurrence and improves detection but does not change severity.
- Detection scored against the inspection rather than against the failure mode — a Cpk study is not a control if the failure mode is operator error; an operator visual check is not strong detection regardless of how thorough.
- Action plans without owners or dates — a recommended action without responsibility and a target completion date is a wish list, not a PFMEA.
- PFMEA never updated — the document was created for an ISO audit and never touched again. PFMEAs are living documents that should be updated when processes change.
How PFMEA Fits into the Wider Quality System
PFMEA is one of several risk-management techniques in ISO 9001:2015's clause 6.1 (Actions to address risks and opportunities). Other techniques include hazard analysis, control plans, and process capability studies. PFMEA is the most widely used in manufacturing because it directly links risk to action.
The PFMEA also drives:
- The Control Plan — every row with a recommended detection method becomes a control plan entry
- Work instructions — process steps with poka-yoke or high-control requirements drive specific work instruction content
- Operator training — high-severity steps require documented operator training records
- Calibration plans — measurement systems used for high-severity detection must be calibrated to the SI traceability standard
- Internal audit plans — high-RPN areas attract more frequent internal audit attention
- Supplier quality — incoming inspection plans are built from supplier-process PFMEAs where available
A PFMEA done well drives all of these downstream documents. A PFMEA done badly is a one-off Excel exercise that gathers dust until the next surveillance audit.
Linking PFMEA to Engineering Change Management
A PFMEA review is one of the most common triggers for an Engineering Change Request (ECR). When the PFMEA identifies a high-RPN failure mode that requires a design change (rather than just a process or control change), the engineering team raises an ECR with the PFMEA row as supporting evidence.
Similarly, every approved engineering change should trigger a PFMEA review on the affected process. If a Change Control Board approves a design change, the production engineer responsible for the affected step should ask: "what new failure modes does this introduce? what failure modes does it remove? do any existing severity, occurrence, or detection scores change?"
The PFMEA-to-ECR-to-PFMEA loop is one of the audit traces ISO 9001 surveillance audits look for. UK certification bodies expect to see PFMEA versions linked to specific engineering changes, with the new PFMEA dated after the change implementation date.
PFMEA Standards and References
PFMEA methodology comes from several published sources. UK manufacturers typically reference:
- AIAG-VDA FMEA Handbook (2019) — the harmonised approach used by automotive Tier 1 suppliers worldwide. Available from AIAG (paid).
- SAE J1739 — the original automotive FMEA standard, still cited by some UK aerospace and defence suppliers. Available from SAE (paid).
- ISO 9001:2015 clause 6.1 — risk-based thinking. PFMEA is one accepted method for meeting this clause. ISO standard available from ISO (paywalled).
- IEC 60812:2018 — Failure modes and effects analysis (FMEA and FMECA). The general (non-automotive) international standard. Available from IEC (paid).
For UK manufacturers without automotive customer requirements, IEC 60812 or the methodology in this guide is sufficient. Automotive Tier 1 suppliers should follow AIAG-VDA. Aerospace suppliers should follow whichever method their AS9100 customer specifies (usually AIAG, but verify).
How ChangeRoute Fits
PFMEA reviews surface design and process risks that often require engineering changes. ChangeRoute will link PFMEA findings into the engineering change workflow — a high-RPN row can raise an ECR directly, with the PFMEA reference attached as evidence, and the resulting ECO will trigger a PFMEA review on completion.
If you run a UK manufacturing SME and want to spend less time on PFMEA-to-ECR-to-PFMEA paperwork, join the waitlist for early access.
Sources
- AIAG-VDA FMEA Handbook (2019) — the harmonised automotive FMEA standard (paid).
- SAE J1739 — automotive FMEA standard (paid).
- IEC 60812:2018 — Failure modes and effects analysis (paid).
- ISO 9001:2015 — Quality management systems (paywalled).