🧭 Why Step 6? #
You’ve prioritized risks with AP (High/Medium/Low) in Step 5. Step 6 is where you change the process to reduce risk—preferably by lowering Occurrence (O) via robust prevention, and lowering Detection (D) via earlier, stronger, and more reliable detection. Then you prove the improvement with data and re-evaluate the ratings.
🎯 Objectives #
- Define concrete, measurable actions for every AP = High (and selected Medium) row.
- Assign owner, due date, expected O/D impact, and evidence to collect.
- Pilot, validate (MSA/SPC/test), re-rate O and D, and update AP.
- Update Control Plan, PFD/WIs/MES, and record residual risk if any.
🚦 Action Prioritization (triage) #
Work in this order:
- Safety/Regulatory (S ≥ 9 or 🔺/🔧) with AP = High
- Other AP = High risks (especially weak D ≥ 6)
- AP = Medium that are frequent, costly, or systemic (affect many parts/stations)
Guiding principle: Prevention first (O ↓), then Detection (D ↓). Don’t swap them.
🧩 Action Types (tag each action) #
- P = Prevention (fixture, recipe lock, poka-yoke, capability, supplier control) → impacts O
- D = Detection (100% check, earlier check, automated interlock, MSA) → impacts D
- S = Systemic (traceability, MES gates, training, change control, PM/Calibration system) → enables both
🛠️ Worked Actions for EWP Line (from your Step-5 hot list) #
1) OP05 — Press-fit depth out of spec / signature out (S=10, AP=H)
- Issues: Depth window violations; force signature mismatches.
- Actions:
- P: Add precision hard-stop shim kit; lock press profile; add lube metering (volume-controlled).
- D: Tighten signature matcher (correlation threshold), add dual-OK interlock (depth and signature).
- S: Add auto-hold in MES when two SPC rules break.
- Evidence: LVDT GR&R ≤10%, depth Cpk ≥ 1.67, false-accept study on signature.
- Expected Impact: O: 4 → 2, D: 3 → 2.
2) OP06 — Seal mis-orientation / surface roughness high (S=8, AP=H)
- Issues: Orientation errors; Ra out-of-spec detected late.
- Actions:
- P: Upgrade guided nest with chamfer + stop pin; add incoming Ra COA gate + in-line Ra quick check (stylus sample/shift).
- D: Vision orientation add presence-of-lip feature; pre-leak moved earlier (post-seal) for earlier detection.
- Evidence: Vision AQL FA rate <0.5%; Ra audit pass rate; pre-leak defect discovery moved upstream.
- Expected Impact: O: 4 → 3, D: 6 → 3.
3) OP07 — ESD event uncontrolled (S=9, AP=H)
- Issues: Manual strap checks; latent failures.
- Actions:
- P: Continuous ESD monitor with audible alarm + station interlock; humidity control (≥40% RH); ionizer at bench.
- D: Start-of-shift & mid-shift strap/mat auto-log with no-pass/no-go; weekly ESD audits in MES.
- Evidence: ESD logs 100% captured, audit pass ≥98%; correlation to field returns reduced.
- Expected Impact: O: 5 → 3, D: 8 → 3.
4) OP07 — Potting mass low/voids/under-cure (S=7, AP=H)
- Actions:
- P: Closed-loop dispenser with auto-weigh check per unit; resin FEFO with expiry interlock; oven dual-probe cure verification.
- D: Vision volumetrics (simple area proxy) or in-process mass record; cure profile auto-review.
- Evidence: Scale GR&R ≤10%; potting Cpk ≥1.33; cure profile trend report.
- Expected Impact: O: 4 → 2, D: 7 → 4.
5) OP09 — Under/over-torque, cross-thread (S=8–9, AP=M)
- Actions:
- P: Thread-start guide & pre-thread sensor; add RFID socket ID with wrong-socket lockout; strengthen PM cadence.
- D: Curve shape analytics (angle vs torque) with auto-NOK; escalate to 100% visual thread check for 2 lots post-changeover.
- Evidence: Tool audit logs; torque Cpk ≥1.67; NOK rate drop.
- Expected Impact: O: 3 → 2, D: 3–4 → 2–3.
6) OP10 — Pre-leak bench recipe/fixture issues (S=8, AP=H/M)
- Actions:
- P: Recipe lock and checksum; fixture PM with seal life counter; dry-air quality alarm.
- D: 2-point master (low/high leak) daily, auto-block if drift; plausibility vs. OP12.
- Evidence: Master trends stable; false-accept <0.3%.
- Expected Impact: O: 4 → 2, D: 4 → 3.
7) OP12 — Final test bypass/recipe error; flow meter mis-cal (S=10, AP=H)
- Actions:
- S: “No scan → no progress” can’t be overridden except by two-person role with audit trail; bench heartbeat watchdog to MES.
- P: Controlled parameter download (hash check).
- D: Dual-check: flow vs. ΔP & RPM sanity; daily two-point master with temp compensation; auto-quarantine on master fail.
- Evidence: Audit logs; master drift charts; bypass attempts recorded = 0.
- Expected Impact: O: 2 → 1, D: 8/6 → 3.
8) OP01 — Wrong variant in kit (S=10, AP=M)
- Actions:
- P: Dual-scan with shape/vision variant verify; kitting bins poka-yoke geometry.
- D: Hard gate at OP05 (model ID = expected).
- Evidence: Miss-kit PPM → near zero; vision accuracy report.
- Expected Impact: O: 3 → 1–2, D: 3 → 2.
9) Interface — MES gate disabled (S=10, AP=H)
- Actions:
- S: Role policy: two-factor + dual-approval for gate changes; real-time alerts to QE/Process Eng on overrides; weekly audit.
- Evidence: Override count=0; alert tests pass; audit logs complete.
- Expected Impact: O: 2 → 1, D: 9 → 3.
📋 Optimization Tracker (working template) #
| PFMEA Row ID | SC | Action (P/D/S) | Owner | Due | Evidence to Collect | Expected ΔO/ΔD | Status |
|---|---|---|---|---|---|---|---|
| PF-05-01 | 🎯 | Hard-stop shim; profile lock (P) | Mfg Eng | 20-Sep | LVDT GR&R, Depth Cpk, Signature FA | O-2, D-1 | In progress |
| PF-06-02 | 🔧 | Ra quick-check; pre-leak earlier (P/D) | QE | 25-Sep | Ra audits; pre-leak catch rate | O-1, D-3 | Planned |
| PF-07-01 | 🎯 | Continuous ESD monitor + interlock (S/P/D) | EHS/IT | 18-Sep | ESD logs, alarm tests | O-2, D-5 | In progress |
| PF-07-03 | – | Closed-loop potting, dual-probe cure (P/D) | Mfg Eng | 22-Sep | Scale GR&R, cure trends | O-2, D-3 | Planned |
| PF-09-01 | 🎯 | Thread-start guide; curve analytics (P/D) | PE | 28-Sep | Curve OK rate, tool audit | O-1, D-1 | Planned |
| PF-10-01 | 🔧 | 2-pt master; seal life counter (P/D) | Test Eng | 17-Sep | Master results, PM logs | O-2, D-1 | Active |
| PF-12-01 | 🔺 | Two-person override; heartbeat watchdog (S/D) | IT/MES | 15-Sep | Override logs=0, alerts | O-1, D-3 | Active |
| PF-01-02 | 🔺 | Dual-scan + shape vision (P/D) | IE | 30-Sep | Miss-kit PPM, vision AQL | O-1, D-1 | Planned |
Tip: Keep this tracker inside the PFMEA workbook and reference row IDs in the Control Plan.
🔁 Re-rating After Implementation (show your math) #
| Row | S (fixed) | O Before → After | D Before → After | AP Before → After | Evidence Snapshot |
|---|---|---|---|---|---|
| OP05 depth/signature | 10 | 4 → 2 | 3 → 2 | H → M/L | Depth Cpk 1.82, GR&R 8.5%, signature FA ↓ |
| OP06 Ra/orientation | 8 | 4 → 3 | 6 → 3 | H → L | Pre-leak moved; vision FA <0.3% |
| OP07 ESD | 9 | 5 → 3 | 8 → 3 | H → M | 100% logs; alarms tested weekly |
| OP07 potting | 7 | 4 → 2 | 7 → 4 | H → M/L | Scale GR&R 7%; cure profile in spec |
| OP09 torque | 9 | 3 → 2 | 3 → 2–3 | M → L | Torque Cpk 1.72; thread rejects ↓ |
| OP10 pre-leak | 8 | 4 → 2 | 4 → 3 | H/M → L/M | Master trends stable |
| OP12 bypass/meter | 10 | 2 → 1 | 8/6 → 3 | H → M | Overrides=0; dual-check in MES |
| OP01 kitting | 10 | 3 → 1–2 | 3 → 2 | M → L | Miss-kit PPM ~0 |
Keep Severity (S) constant unless the end-user effect truly changes (rare in PFMEA).
🧪 Evidence Pack (what to attach before re-rating) #
- MSA results (GR&R %, ndc, bias/linearity if relevant)
- SPC/Cpk before vs. after; rule violations trend
- Test bench masters results, calibration certificates, false-accept study
- Screenshots: recipe locks, MES gates, interlocks, alerts
- PM/Calibration logs; audit checklists; training sign-offs
- Pilot/Run@Rate defect Pareto showing delta
🧷 Control Plan, PFD, WI & MES Updates (don’t skip) #
- Add/modify characteristics, methods, frequency, records, reaction plans to the Control Plan.
- Update PFD (inspection moved earlier, new checks).
- Update Work Instructions (photos, torque/press strategy, ESD steps).
- Lock MES changes (recipes, gates, alerts) and version records.
🧾 Residual Risk Acceptance (if AP not Low) #
If after reasonable actions an item remains AP = Medium/High:
- Document why further reduction isn’t feasible (technology/cost/timing).
- Record acceptance by authorized person (e.g., Chief Engineer/Quality Director).
- Define monitoring: tighter audits, layered process audits (LPA), containment windows.
🧰 Rapid “Optimization Storyboard” (8 steps to closure) #
- Define gap (S/O/D/AP + cost/PPM).
- Hypothesize root causes (5-Why/Fishbone).
- Select P/D/S countermeasures.
- Pilot on a small batch; instrument for data.
- Validate (MSA, SPC, masters).
- Re-rate O/D, update PFMEA & Control Plan.
- Train, standardize, and release.
- Monitor KPI (FPY, PPM, Cpk), capture Lessons Learned.
✅ Outputs of Step 6 #
- ✅ Action tracker with owners/dates/evidence.
- ✅ Implemented prevention/detection/systemic changes.
- ✅ Re-rated O/D and updated AP.
- ✅ Updated Control Plan/PFD/WIs/MES.
- ✅ Residual risk documented (if any) and LL captured.
⚠️ Common Pitfalls (and fixes) #
| Pitfall | Fix |
|---|---|
| “Action” = vague (“improve process”) | Make it measurable (e.g., add hard-stop, Cpk target, GR&R limit). |
| Claiming D improvement via prevention | Only checks/tests reduce D. Keep prevention in O. |
| Re-rating without data | Attach MSA/SPC/master evidence before changing O/D. |
| Not updating Control Plan | Every risk change must reflect in Control Plan & WIs. |
| Ignoring system risks (MES gates) | Treat interfaces as first-class risks; harden access & alerts. |
🔗 Next #
Move to Lesson 5.7 — Step 7: Results Documentation, where you’ll compile the PFMEA Summary Table, finalize the Control Plan, build the PPAP pack, and publish Lessons Learned.
🧠 Pro Tip #
Optimization isn’t finished when the action is done—it’s finished when the data proves O/D have dropped, the Control Plan is updated, and the system keeps it that way (gates, alerts, audits).
