PFMEA Assembly Process Example
Process overview
The scope is a manually loaded, fixture-assisted assembly of an electromechanical unit (e.g. a motor controller with display, harness and external connectors) at a six-station flow line. Each station has a dedicated fixture, a torque tool, a barcode scanner and an MES interlock to the next station. Functional test runs at station 5; conformal labeling and packaging at station 6. The PFMEA reflects the realistic ratio of human-error failures (torque, routing, missed steps) to equipment-driven failures.
Process flow
- Kitting — pick-by-light verified against work order
- Station 1 — sub-assembly screwing
- Station 2 — wire harness routing
- Station 3 — connector mating
- Station 4 — torque application on structural fasteners
- Station 5 — functional test (power-up, communication, output verification)
- Station 6 — label application, accessory pack, sealing
- Outgoing inspection — sampling + weight check
Common failure modes
- Wrong part in kit — the most common high-severity failure on assembly lines and the most preventable with pick-by-light.
- Cross-threading and under/over-torque — operator-induced, mitigated by tool design and torque/angle monitoring.
- Wire pinch and routing damage — invisible until field failure; only an isolation test reliably catches it.
- Half-mated connectors — designed-in by connector choice; tactile feedback and Poka-yoke matter.
- Test bypass — process-discipline failure, not a technical failure. MES interlock is the only real control.
- Label and packaging errors — low severity individually but high frequency, drives customer dissatisfaction.
Representative PFMEA table
| Process Step | Failure Mode | Effect | Cause | Prevention | Detection | S | O | D | AP |
|---|---|---|---|---|---|---|---|---|---|
| Kitting | Wrong sub-component pulled into kit | Wrong part assembled; functional failure | Bin mislabeled or location swap during 5S | Pick-by-light + barcode verification at kitting | End-of-line cross-check against BOM | 8 | 3 | 4 | High |
| Sub-assembly screwing | Cross-threaded fastener | Joint failure in vibration / service | Operator misalignment of fastener to thread | Self-piloting screws + fixture guides | Torque-angle signature monitoring | 7 | 4 | 3 | Medium |
| Wire harness routing | Wire pinched between housing halves | Insulation damage → short circuit in service | Free routing with no guide; operator variation | Routing channels in assembly fixture | End-of-line hipot / isolation test | 9 | 3 | 4 | High |
| Connector mating | Connector half-mated (latch not engaged) | Intermittent loss of signal in service | No tactile click; operator over-confidence | Connector design with audible click + Poka-yoke fixture | End-of-line continuity test | 8 | 4 | 3 | High |
| Torque application | Under-torqued bolt | Joint loosening under vibration | Torque tool calibration drift | Daily torque tool calibration check | DC torque tool data logging per joint | 8 | 3 | 2 | Medium |
| Torque application | Over-torqued bolt | Thread strip or housing crack | Wrong tool program selected for joint | Program selection tied to fixture barcode | Inline torque trace SPC + reject on out-of-band | 7 | 3 | 3 | Medium |
| Functional test | Test station bypassed (skip) | Defective unit shipped to customer | Operator override; flow disruption | Mandatory serial-number gating at next station | Daily golden-unit verification of test station | 9 | 2 | 3 | Medium |
| Label application | Wrong label printed on unit | Traceability lost; wrong product data shipped | Label printer queue misaligned with MES serial | Label print triggered by serial scan at the station | End-of-line label barcode scan vs MES record | 6 | 3 | 2 | Low |
| Final assembly inspection | Cosmetic damage on visible surface missed | Customer return / cosmetic rejection | Inspector lighting / fatigue | Standardized lighting + inspector rotation | Sampling re-inspection by quality at packout | 5 | 4 | 5 | Medium |
| Packaging | Missing accessory in box | Customer complaint, replacement shipment | Picker error during packout | Pick-list verification + weight check on sealed box | Outgoing weight check vs reference | 5 | 4 | 3 | Medium |
Example failure chains
| Trigger | Failure chain | Customer impact |
|---|---|---|
| Operator routes harness over the housing rib instead of through the channel | Housing closes on the wire → insulation cut by sharp edge → intermittent short to chassis weeks later | Field failure, potentially safety-related on automotive |
| Connector latch not fully engaged at station 3 | Unit passes functional test (signal present) → vibration in shipping unseats latch → no signal at customer install | Customer install delay + field service callout |
| Operator skips functional test station during a stoppage | Defective unit moves to packaging without test → shipped → fails at customer power-up | RMA and audit non-conformance against control plan |
Current controls
- Prevention: pick-by-light, fixture geometry, self-piloting fasteners, torque program tied to barcode, MES station interlock, operator certification matrix.
- Detection: torque/angle data logging, end-of-line isolation test, continuity test on connectors, functional test, label barcode scan, weight check at packout.
Recommended actions
- Add vision system at station 2 to verify harness routing before housing close.
- Mandate "pull test" on connectors before functional test — small operator action, drops Detection significantly.
- Convert torque data logging from passive log to active reject on out-of-band torque/angle.
- Replace fixed label printer with print-on-demand triggered by serial scan at the station.
Action Priority discussion
Three rows promote to AP High: wrong kit, wire pinch, half-mated connector. These share the same structural pattern — Severity 8–9, Occurrence 3–4, Detection 3–4. The AIAG-VDA AP table is explicit that any High Severity row with non-trivial Occurrence and Detection above 2 belongs in the High bucket. The under-torque row sits at AP Medium because Detection is 2 (every joint is logged) and Occurrence is well-controlled; a team that drops the daily calibration check would see Occurrence rise to 5 and promote this row to High.
Common mistakes in assembly PFMEAs
- Listing "operator error" as a cause. Operator error is not a cause; the cause is the process design that allowed the error.
- Scoring Detection 2 for "operator visual" on every row. Realistic floor is Detection 5.
- Missing the bypass / skip failure mode entirely. It is the single most common cause of escapes on flow lines.
- One row per station instead of one row per failure mode. A station can have 6–10 distinct failure modes.
- Treating packaging as out-of-scope. Customer complaints about packaging defects show up in the same warranty system as functional failures.
Audit considerations
IATF 16949 and customer assembly-line audits specifically look for: torque program traceability per joint, MES interlock evidence (no station skip without exception record), end-of-line test correlation to golden unit, and operator certification matrix tied to PFMEA-identified critical steps. Build the PFMEA so each High-AP row can be pointed at a specific control plan line item and a specific work instruction.
PFMEA — Assembly Process in the Qhubio knowledge graph
How PFMEA — Assembly Process connects to other FMEA concepts, standards, examples and software.
Frequently asked questions
Learning path
Three tiers, automatically derived from the Qhubio knowledge graph. The tier containing this guide is highlighted.
Further reading
Curated next steps — methodology guides, worked examples, and the relevant tool.
Qhubio FMEA solutions
Move from theory to a working FMEA. Same methodology, three entry points depending on how you want to start.
Generate a structured FMEA in minutes
Qhubio applies the AIAG-VDA methodology automatically — no Excel formulas, no inconsistent rating scales, no scattered spreadsheets.
Generate your first FMEAHelp us improve Qhubio. Analytics help us understand which FMEA features are useful, which pages need improvement, and where users encounter problems. No advertising. No selling personal data. See our Cookie Policy.