Mold Cavity Tracking

Mold cavity tracking is the structured capture of which specific cavity within a mold produced each molded unit, together with the tool identity, cycle, and relevant process context (e.g., machine, shot number, lot, shift). In multi-cavity tools, each cavity is a micro-source of variability; associating the cavity attribute to each part creates a fine-grained genealogy that supports targeted quality controls, maintenance insight, and traceability. In regulated operations, cavity tracking feeds the Device History Record (DHR) or batch record to demonstrate that each finished device/subassembly was manufactured under defined controls and that nonconformances can be contained and investigated efficiently.

Cavity tracking can be realized by: (1) sensing/attributing the active cavity during the cycle (machine signals, hot-runner selection), (2) reading a cavity code embossed on the part via vision/OCR, or (3) deterministic mapping where part diverting/conveyance uniquely ties to a cavity chute or lane. The MES reconciles these signals with material lots and work orders, creating a persistent unit attribute used in routing decisions, in-process holds, SPC per cavity, and recall narrowing.

For medical devices, 21 CFR 820.184 requires complete Device History Records that demonstrate the device was manufactured in accordance with the Device Master Record; for certain devices, 21 CFR 820.65 requires traceability. ISO 13485 requires manufacturers to define and maintain traceability appropriate to product risk, with enhanced expectations for implantable and high-risk devices. While no regulation mandates cavity tracking per se, capturing cavity identifiers materially strengthens DHR completeness and risk-based traceability, supports complaint trending, enables targeted holds (only parts from suspect cavities), and reduces recall scope and cost.

Electronically maintaining cavity genealogy invokes electronic-records expectations under 21 CFR Part 11—controlled access, audit trails for configuration changes (e.g., cavity-to-chute mapping), system validations proportionate to risk, and electronic signatures where records are signed. A defensible program shows how cavity-level data are accurate, attributable, contemporaneous, original, and complete (MHRA data-integrity principles), and how they feed into CAPA and management review with trend analyses at the cavity level.

ISA-95 guides how to represent equipment, materials, and production records at Level 3 (MES), while ISA-88 informs procedural control and equipment hierarchies for batch-like operations. A practical approach is to model the mold as a primary equipment asset with equipment modules for cavity groups or individual cavities when signals support it. The cavity identifier is then a unit attribute of the produced material/serial captured in the production record and linked to the equipment genealogy. This allows material genealogy and equipment genealogy to intersect at the unit, enabling both backward and forward trace at the finest needed granularity.

Successful cavity tracking starts in master data and tool design. Each mold, insert set, and cavity needs a stable identifier that survives maintenance cycles and insert swaps. Cavity identifiers should be unique within the tool and human-readable on the part where vision is employed. Family molds require mapping of cavity-to-part-number. The MES should version-control the cavity map (cavity-to-chute/part), interlock production to approved maps, and tie each map version to tool maintenance history.

• Define a canonical cavity ID scheme: TOOL-771:CAV-01 … CAV-16.
• Engrave cavity codes on the parting-line region readable by vision (where applicable).
• Maintain a versioned cavity-to-diverter/chute map; require approval workflow before use.
• Model replaceable inserts with their own IDs; track insert swaps by timestamp and approver.
• For family molds, define cavity-to-part mapping and enforce order mix rules in MES.
• Tie cavity map versions to preventive maintenance (PM) and dimensional qualification events.

Cavity attribution may be deterministic, sensed, or inferred. Deterministic attribution uses physical segregation (e.g., each cavity ejects to a dedicated lane with a scanner at pick) so the lane implies cavity. Sensed attribution uses PLC tags: valve-gate states, cavity enable bits, reject diverter states, or hot-runner selection. Inferred attribution uses vision/OCR of cavity imprints or unique surface features. The MES must time-correlate signals to shots, debounce events, and handle short shots or blocked cavities. Store-and-forward is needed for network resilience; missing data must raise exceptions and drive controlled rework or quarantine.

1. Map machine/vision tags to MES attributes (cavity, shot, reject flag) with version control.
2. Synchronize time sources across PLC, vision, and MES; monitor clock drift.
3. Implement lane/chute validation scans to verify cavity-to-lane integrity after changeovers.
4. Capture first-article and in-process checks per cavity to verify code legibility and mapping.
5. Define fallback: if cavity cannot be determined, route to hold with reason code and review.

Cavity tracking data are GxP records when used to release devices or lots. Under 21 CFR Part 11, closed systems must enforce unique user credentials, role-based access, and audit trails of create/modify events (e.g., edits to cavity maps, overrides of unreadable marks). Electronic signatures should be bound to DHR/eBMR records at key steps (setup approval, lot release, NC disposition). The system validation approach should be risk-based (GAMP 5), focusing testing and controls on functions that could mis-attribute cavities or allow undetected data loss.

Data integrity controls should demonstrate that cavity attribution is attributable (to equipment and person), legible, contemporaneous (time-aligned to the shot), original (captured from authoritative sources), and accurate/complete. Typical measures include write-once event logs tied to shot numbers, detection of sequence gaps, hash-protected file storage for images, and periodic review of audit trails by QA with documented findings.

Per-cavity genealogy enriches both backward trace (from finished unit to resin lot, tool, cavity, shot, equipment state) and forward trace (from a suspect cavity/tool/time window to all affected units and downstream assemblies). Combined with serial traceability and lot controls, it enables highly targeted holds and CAPA scoping—often reducing recall scope by an order of magnitude compared to lot-level only. In regulated practice, the DHR shows which cavity produced each component and how exceptions were handled (rejected shots, rework flows, waivers).

1. Backward trace scenario: start from a complaint unit; read its serial; query eDHR → find cavity, shot, tool, machine; identify upstream material lots and parameter bands for that shot.
2. Forward trace scenario: maintenance finds wear on CAV-07 insert; query MES for all units produced by TOOL-771:CAV-07 during last validated interval; issue targeted WMS hold and notify QMS for assessment.
3. Containment: use per-cavity SPC triggers to auto-route parts from drifting cavities to 100% inspection; document in eDHR and open NC if trends persist.

Because cavities are parallel processes sharing a tool, per-cavity SPC and capability indices are often more sensitive than pooled charts. Track dimensionals, weight, ejector force, cosmetic defect rates, and cycle time per cavity. Use stratified control charts to detect localized drift (e.g., cooling channel blockage affecting only Group B). Build capability baselines by cavity after tool PM or insert replacement; compare to detect regression. Feed CAPA with cavity-specific trends and link maintenance records to post-PM performance to demonstrate effectiveness.

• X̄–R or EWMA charts per cavity for key characteristics; pooled comparison for between-cavity variance.
• Pareto of defects by cavity; correlate with tool temperatures/pressures from historian tags.
• Heat map dashboards of Cp/Cpk by cavity and time-window; flag crossings of action/alert limits.
• Rule-based diversion: if cavity defect rate exceeds threshold, divert to manual inspection, notify Maintenance, and require QA review.

Apply GAMP 5, 2nd ed., to classify components (e.g., COTS MES functions vs. configured workflows vs. custom interfaces) and tailor verification. Author a URS that specifies cavity attribution accuracy, timing tolerances, exception handling, audit trail needs, and performance (throughput under multi-cavity shots). Derive functional/design specifications detailing tag mappings, data models (cavity attribute on serial/lot), and security. Execute risk-based IQ/OQ/PQ: challenge time sync loss, tag remapping attempts, vision OCR misreads, and network outages (store-and-forward). Verify Part 11 controls—user management, e-signatures on setup approvals, and immutable audit trails for map versions.

Change control should govern all alterations to molds (insert swaps), map versions, and machine/vision tag points. Maintain traceability matrices from URS to test evidence, including negative and boundary tests (e.g., simultaneous multi-cavity disable/enable, rejected-shot handling). Periodic review and requalification after significant mold maintenance ensures attribution remains correct and documented, with QA approval captured in the eDHR batch or lot file.

Common failure modes include: remapped diverter chutes after tool changeover without updating the MES; vision reads of worn/low-contrast cavity marks; blocked cavities creating unplanned part mixing; and shared conveyors that defeat deterministic segregation. Shot rejections not correctly tied to specific cavities can leak defects downstream. Another pitfall is ignoring insert-level identity—after an insert swap, historical trends become incomparable without recording insert IDs.

• Use electronic setup checklists with enforced scans/tests to verify cavity-to-lane and vision read accuracy before start-of-run.
• Alarm and interlock on inconsistent signals (e.g., valve-gate open but no part detected on lane).
• Quarantine logic: any un-attributable part routes to a hold container with an NC opened automatically.
• Govern cavity-mark legibility with a preventive maintenance criterion; verify at first-article inspection.
• If using shared conveyors, introduce unit-level identifiers or in-lane scans immediately after ejection.

V5 models molds and cavities as equipment/equipment-modules aligned to ISA-95, stamping the cavity attribute on each unit/lot record at execution. It ingests PLC/vision tags, lane scans, and historian events, time-aligns them to shots, and persists immutable, Part 11–compliant records and images. Per-cavity SPC, automated holds, and maintenance triggers are native—so when a cavity drifts, V5 can auto-divert output, open a Nonconformance, schedule a Maintenance work order, and restrict WMS release while capturing approvals in the eDHR.