Built for people who build things

8D: DRAFT customer version

Controlled 8D template: not on file; standard eight-discipline structure used.

D1: Team (proposed roles; owner edits)

• Lead: quality engineering, unassigned
• Finder: inspection
• Production supervisor, unassigned
• Boring-operation operator, unassigned
• Metrology / calibration owner, unassigned
• Reviewer: quality review

D2: Problem description

• Requirement: DWG-PN-305 Rev B, Sheet 1, detail A, zone C3: bore Ø12.000–12.025 mm.
• Found 2026-11-05 09:10 EST, INSP-1105, by inspection: 4 of 8 sequentially sampled pieces from LOT-1105 (WO-305) above 12.025 mm (pieces 4, 5, 6, 8; max 12.034 mm, 0.009 mm over).
• Gauge: the inspection gauge; calibration unverified, so measurement validity is unconfirmed.
• Scope: whole lot suspect. Reported lot quantity: 200, unreconciled; wider suspect scope unknown. This does not establish that all pieces are defective.
• Prior nonconformance on the same part and characteristic: 2 Oct 2026. Whether its CAPA held is unknown.

D3: Interim containment (plan confirmed; execution unverified)

• Reported done (supervisor, 09:25, unverified): operation stopped; red hold tag on lot rack; all 200 reported onsite, none shipped.
• Open, owners unassigned: shipment-record check covering installed-at-customer, at-customer and in-transit (immediate); verify segregation, count lot, ERP hold (immediate); confirm none in finished goods (same day).
• Held: 100% inspection, until the inspection gauge is verified. Not authorized: disposition, release, restart, inspection execution.

D4: Root cause

Status: ROOT CAUSE NOT VALIDATED. Ranked candidates below; the leading candidate is weakly ahead and not confirmed.

Occurrence, candidate 1 (leading, weak): setup offset error after insert change

• Chain: insert changed 08:40 → offset adjusted → offset left bore cutting oversize → parts ran oversize until 09:10 sample.
• Link 1 insert changed. For: process note C, 08:40. Against: none.
• Link 2 offset adjusted. For: note C says adjusted. Against: no offset log; value unknown. WEAK POINT: no record of the value set.
• Link 3 bore cut oversize. For: 4 of 8 over (A); an operator-reported suspicion of a setting error (unvalidated). Against: pieces 1–3 and 7 in limits; readings rise across 1–5 rather than step; which pieces were cut after 08:40 is unknown; gauge unverified.
• Confirm: CNC offset history shows a changed bore offset at 08:40 AND retained post-08:40 pieces are oversize on a verified gauge.
• Rule out: retained pieces re-measure within limits on a verified gauge, or pieces cut before 08:40 are equally oversize.

Occurrence, candidate 2 (unresolved, cannot be excluded): gauge reads high

• Chain: the inspection gauge out of calibration or mis-set → readings biased high → in-limit bores reported oversize.
• Link 1 gauge out of cal. For: certificate unavailable (A). Against: unavailable is not lapsed.
• Link 2 readings biased high. For: none. WEAK POINT. Note: a uniform +0.009 mm bias would put all 8 readings within limits (lowest becomes 12.002 mm), so the data cannot exclude it. Against: no evidence of bias.
• Confirm: the inspection gauge reads a certified master outside its accuracy.
• Rule out: the inspection gauge reads a certified master within its accuracy.

Occurrence, candidate 3 (weakest): progressive drift

• Chain: thermal growth or edge build-up after restart → bore grows across the run → later pieces oversize.
• Link 1 drift mechanism. For: none. WEAK POINT. Against: new insert at 08:40 argues against wear; no sensor or thermal data.
• Link 2 bore grows across run. For: readings rise across pieces 1–5. Against: rise does not continue (pieces 6–8: 12.029, 12.021, 12.030); six-points-increasing run rule not met (longest run 5).
• Confirm: further sequential pieces on a verified gauge keep growing.
• Rule out: a verified-gauge sequence shows a stable offset.

Cheapest discriminating test

Check the inspection gauge against a certified master, then re-measure retained pieces 1–8 on a verified instrument. Separates Candidate 2 from Candidates 1 and 3, and every other conclusion depends on it.

Escape / detection, candidate, not validated

Why it was not caught before 09:10: first-piece check after insert change is required by the work instruction per note C, but no first-piece approval exists for 08:40. Whether it was skipped, done and not recorded, or not applicable is unknown; the instruction's document-control record was not supplied.

Systemic, candidate, not validated

The prior nonconformance on this part and characteristic (2 Oct 2026) was closed after offset correction and retraining; its corrective action's effectiveness check has no results. Effectiveness unknown, not proven failed. Whether today's cause matches October's is not established.

D5: Permanent corrective actions

Status: PROPOSED, NOT CHOSEN. No owners, no dates. Contingent on the D4 test results.

P1: breaks Candidate 1, link 2 (offset value unrecorded/unverified)

After any insert change, record the offset value set and require a measured first piece on a verified gauge before the run continues.

P2: breaks Candidate 2, link 1 (gauge out of calibration)

If confirmed: quarantine the inspection gauge, recalibrate, and assess every inspection it performed since its last valid calibration.

Moved to D3 (containment, not corrective)

100% inspection of the lot; ERP hold; shipment check.

Improvement, not corrective action

Review whether 8-piece sequential sampling is adequate for this characteristic.

D6: Implement and verify

Placeholder. Action implemented, implemented by and date, verification method and result, and verified by and date, all to be completed.

D7: Prevent recurrence

Status: PROPOSED, NOT APPROVED. The items below are QUESTIONS TO INVESTIGATE, not established facts. Missing evidence in the supplied extracts does not show how the prior CAPA was run.

Known (from supplied quality-history extract)

The prior nonconformance on this part (2 Oct 2026) was closed after offset correction and operator retraining. Its corrective action record lists retraining completed 4 Oct and an effectiveness field reading 'check next 3 lots'. No lot results, independent verification or closure evidence were supplied. Effectiveness is unknown, not proven failed.

Questions to investigate

• Q1. Effectiveness check. Did its corrective action's effectiveness check have an owner, a due date and a data source, and was it run? Records not supplied.
• Q2. Closure sequence. When was the prior nonconformance closed relative to the effectiveness check, and what closure criteria applied? Records not supplied.
• Q3. Scope of the prior action. Was retraining the only action under the prior corrective action, or were setup or first-piece controls also changed? The extract lists retraining; it may not be complete.
• Q4. Calibration visibility. Was the inspection gauge's calibration status visible and checked at the point of use? Only the certificate's unavailability is known.

Candidate system changes (conditional on the answers; none selected)

• If Q1 or Q2 shows a gap: require effectiveness evidence before CAPA closure.
• If Q3 shows training-only control: add a recorded first-piece release after insert or offset changes.
• If Q4 shows a gap: record gauge calibration status with each inspection result.

D8: Close and recognize

Placeholder. Closure approval, date, effectiveness evidence reference and team recognition, all to be completed.

Redaction diff

Gate 2: redaction diff approved by Quality Reviewer, 2026-11-05 10:07 EST. Seven rows of personnel names and internal identifiers removed. Does not approve the 8D conclusions, corrective actions or release. Draft; unsent; no customer identified.