Quality deviations, investigated end to end. Chart to corrective action.
We work on making a plant’s history usable the moment a chart goes out of control.
SPC chart with an RCA link. Control chart of shoulder temp, end of steam on T2-B08, one point per cure cycle: centre 171.0 °C, limits 169.0 to 173.0 °C. Rule 1 · one point beyond a 3σ limit trips at cycle 2165 · TB-3302 (168.8 °C); the series stops there because TB-3302 goes on hold. Plant Sentinel flags it, the RCA Investigator opens INV-20533, and the hold under SOP-QA-001 waits for a person's disposition.
Scoping the hold.
The evidence for a deviation sits in several systems, and the experience sits with whoever is on shift. Deciding what to do with a suspect lot takes minutes, but finding every lot that shared its press takes days.
Lot genealogy belongs in the knowledge graph, so that question becomes a query. The engineer who owns the line still approves the hold.
- Readschema · production.batches0.4s
- Queried4 tables · 12,480 rows1.9s
- Comparedweek 41 against week 400.6s
- Checkedmaintenance_log, shift_roster0.8s
- Wroteanswer with 3 sources0.3s
Most of the drop is on the night shift after Tuesday's fixture change. Lines 1, 2 and 4 held steady.
From one point on a chart to the fix.
RCA investigation trace. Investigation INV-20533, anchored on: Blisters and uniformity rejects (7 of 48 tyres · TB-3302; press T2-B08 · on hold). The RCA Investigator agent worked 7 steps: 1. Prior knowledge: Lesson: curing press · steam dip → under-cure; 2. Query lot events: LotEvent: cure, uniformity, hold · TB-3302; 3. Read state history: Machine: T2-B08 · HP steam below warn, min 6; 4. Search SOP: Document: SOP-CUR-010 §4 · steam out of range → hold; 5. View X-ray image: Image: TB-3302 tyre 17 · blister at shoulder; 6. Twin check: Twin: equivalent cure · 0.86 at the shoulder (min 0.90); 7. Check inputs: LotEvent: CB-1187 age 31 h · bladder 212 of 400 cycles. Hypotheses: H1 Under-cure after a steam pressure dip, confirmed; H2 Compound CB-1187 aged past limit, refuted by LotEvent: CB-1187 · 31 h, limit 72 h; H3 Bladder near end of life, refuted by Machine: bladder 212 of 400 cycles; H4 Mould out of register, refuted by Measurement: harmonic ≠ mould segments. Root cause proposed by the agent: Under-cure after a steam dip, confirmed by a process engineer. Solution waiting as a pending proposal: No cure start below steam warn (CURE-PCR-SN2 v4 → v5 · approval requested). TB-3302 held · retest on UF-03, disposition off-flow under SOP-QA-001. Verified by a person: Next 6 batches in limits (TB-3303, TB-3304, TB-3305, +3).
Lot genealogy and traceability. Genealogy of 27 records in 5 steps: Materials, Components, Green tyres · TBM-07, Cured tyres · curing, Dispatch. T-3302/17 (uniformity reject · blister) is traced back along its made-from links: T-3302/17 ← G-3302/17 ← TR-0418 ← CB-1187. Source: CB-1187. Made from the same source and held too under SOP-QA-001: G-3302/15, T-3302/15, G-3302/16, T-3302/16, PAL-62, G-3302/18, T-3302/18.
Manufacturing knowledge graph. A plant knowledge graph: the classes Plant → Zone → Machine → Cure spec → Batch → Measurement → Defect → SOP, with 21 records linked under them. The highlighted path traces Defect Blister ← Measurement State of cure ← Batch TB-3302 ← Cure spec CURE-PCR-SN2 ← Machine T2-B08 ← Zone Curing hall ← Plant Tyre plant, and on to SOP-QA-001 for what to do. Preview: Machine T2-B08 (curing press): running, OEE 62%.