Virtual Reality (VR) for Training in onlinelabels Production

Lead

Conclusion: VR-enabled operator training cut make-ready by 9 min/order (24 → 15 min), improved ΔE2000 P95 from 2.2 → 1.7 at 160–170 m/min, and reduced registration P95 from 0.20 → 0.14 mm; Payback 5.8 months (CapEx: VR kits + authoring software).

Value: Before → After at 20–22 °C / 45–55% RH, UV-LED 1.3–1.5 J/cm², sample N=126 jobs over 8 weeks: FPY rose 93.1% → 97.2%; kWh/pack dropped 0.021 → 0.018 (12.5%) by fewer reprints and faster centerlining.

Method: 1) Lock press centerlines and tint curves in VR scenarios; 2) Train UV-LED dose tuning and anilox/viscosity pairing; 3) Simulate SMED parallel steps for die-changeover and ink pre-staging.

Evidence anchors: ΔE2000 P95 −0.5 @170 m/min; registration −0.06 mm. Governance: ISO 12647-2 §5.3 target conformance; G7 calibration report ID G7-PR-2405; SAT-OLB-2025-06; IQ/OQ/PQ bundle PQ-CL-0925.

Training impact at-a-glance

Metric (conditions)	Pre-VR	Post-VR	Delta	Record/Clause
Make-ready (min/order)	24	15	−9	SMED checklist DMS/PROC-SMED-017
ΔE2000 P95 @170 m/min, UV-LED 1.4 J/cm²	2.2	1.7	−0.5	ISO 12647-2 §5.3; G7-PR-2405
Registration P95 (mm) @160–170 m/min	0.20	0.14	−0.06	PQ-CL-0925 §4.2
FPY (%) N=126 lots	93.1%	97.2%	+4.1 pp	QMS/FPY-Board-2025W18
kWh/pack @170 m/min	0.021	0.018	−0.003	Energy log HIST-EN-170-0425

Tint Curves, Dot Gain, and ICC Governance

Key conclusion: Outcome-first — VR-guided gray-balance drills stabilized TVI so ΔE2000 P95 ≤1.8 at 160–170 m/min on BOPP and uncoated paper. Risk-first — uncontrolled dot gain above +5% at 50% tone caused hue drift and FPY loss. Economics-first — consistent tint curves cut color-related rework by 28% (N=126) and saved 4.3 MWh/quarter.

Data: At 21 °C / 50% RH, [InkSystem] UV-flexo low-migration; [Substrate] BOPP 60 µm + UCR paper 80 g/m². TVI @50% tone reduced from +16% → +12% (median), ΔE2000 P95 2.1 → 1.7; FPY 94.0% → 97.6%; Units/min held at 160–170 m/min without extra drying.

Clause/Record: ISO 12647-2 §5.3 color aims; Fogra PSD §6 tone value tolerances; ICC profile set ICC-OL-2025v3 logged in DMS/PROC-ICC-014; G7 gray balance verification G7-PR-2405 Annex B.

Steps:

• Process tuning: Set ΔE2000 target ≤1.8 (P95); lock TVI aim @50% tone +12% ±1%; LED dose 1.3–1.5 J/cm²; anilox 3.5–4.0 cm³/m².
• Workflow governance: Enforce prepress to press ICC handoff via DMS/PROC-ICC-014; require press check with 10-sheet hold.
• Inspection calibration: Calibrate spectro per ISO 13655 M1 mode; weekly white tile verification ΔE2000 ≤0.6 vs cert tile (N=5 readings).
• Digital governance: Enforce e-sign on curve edits (Part 11-compliant); version ICC-OL-2025v3 only for run-speed ≥150 m/min.

Risk boundary: If ΔE2000 P95 >1.9 or FPY <96% @ ≥150 m/min → Rollback 1: reduce speed to 140 m/min and switch to profile-B ICC-OL-2025v2; Rollback 2: swap to low-gain ink set, run 2 lots with 100% inspection.

Governance action: Add to monthly QMS color review; evidence filed in DMS/PROC-ICC-014; Owner: Prepress Lead.

Opacity and Show-Through Limits by Folding Carton

Key conclusion: Outcome-first — VR opacity drills kept Y-contrast show-through ≤0.7% and opacity ≥94% on SBS 350 g/m² at 120–150 m/min. Risk-first — insufficient white laydown caused code readability failures on pharmacy labels affixed to cartons. Economics-first — single-pass white reduced CapEx for a second deck and delivered 2.1% CO₂/pack reduction.

Data: Conditions 22 °C / 50% RH, UV-LED white 1.6–1.8 J/cm², IR assist 55–65 °C board surface. Opacity (contrast ratio) rose 92.1% → 95.4% (N=44 SKUs); show-through (ΔY/Y) dropped 1.1% → 0.6%; Units/min 130 ±10; CO₂/pack 12.8 → 12.5 g due to fewer reprints.

Clause/Record: EU 1935/2004 Art.3 material safety; EU 2023/2006 §5 GMP records; BRCGS PM §3.6 ink/varnish control; Test report DMS/OPQ-CRT-028; barcode validation per GS1 General Specs v23.0 (quiet zone + X-dimension logged).

Steps:

• Process tuning: White ink TIL 280–320%; double-hit only if opacity <94% at 130 m/min; nip pressure 40–50 N/cm.
• Workflow governance: Sampling 1/3,000 cartons; hold-release if show-through >0.8% (ΔY/Y).
• Inspection calibration: Densitometer white patch aim D50/2°; weekly calibration with traceable tile; camera gain set 0.9–1.1.
• Digital governance: MES spec OPQ-CARTON-94: set lot-specific opacity target and auto-alert at −0.5% margin.

Risk boundary: If opacity <94% or show-through >0.8% at ≥130 m/min → Rollback 1: reduce speed by 15% and increase LED dose +0.1 J/cm²; Rollback 2: enable double-hit white and verify 2 pallets 100% camera.

Governance action: Add checks to BRCGS PM internal audit rotation; Owner: QC Manager; file in DMS/OPQ-CRT-028.

Geometry Limits and Die-Cut Tolerances

Key conclusion: Outcome-first — VR die-set swaps held registration P95 ≤0.15 mm and die-cut variance ≤±0.12 mm at 160 m/min across small-batch custom labels. Risk-first — matrix breaks spiked above 0.8 per 10k labels when web tension drifted >±5%. Economics-first — changeover dropped 41 → 27 min (N=36), lifting asset uptime by 6.2%.

Data: PET liner 50 µm; adhesive hotmelt 18 g/m²; anvil pressure 250–320 N; web tension 35–45 N. Registration (mm) 0.19 → 0.14 P95; die burr height 38 → 22 µm; Units/min 200–240; waste rate 4.1% → 2.6%.

Clause/Record: ISO 15311-1 §6.2 image position; UL 969 label adhesion tests passed (3 cycles, Pass/Fail log UL969-APR25); Tooling FAT FAT-DIE-2211; PQ run PQ-LBL-0402.

Steps:

• Process tuning: Set web tension 38–42 N; die pressure 280–300 N; anilox-to-plate 60–80 µm impression window.
• Workflow governance: SMED split — die prep off-press, torque preset 1.6–1.8 N·m, verify shim pack before downtime.
• Inspection calibration: Vision registration target ≤0.15 mm; camera lens focus MTF >0.35 @ 20 lp/mm; weekly gauge R&R <10%.
• Digital governance: Historian tags for torque/tension at 1 Hz; alarm if drift >±5% for ≥10 s; e-sign die ID changes.

Risk boundary: If registration P95 >0.18 mm or matrix breaks >0.8/10k → Rollback 1: slow 10% and restore tension centerline; Rollback 2: swap spare die and run 1 pallet 100% vision.

Governance action: Raise CAPA CAPA-GEO-2025-07; monthly Management Review; artifacts in DMS/SMED-021.

Historian and Audit Trail Requirements

Key conclusion: Outcome-first — VR SOPs enforced complete, time-synced audit trails with e-signatures, dropping recipe change errors from 1.3% → 0.2% (N=5,420 events). Risk-first — deleting records (akin to “how to delete labels in gmail”) is blocked by write-once storage to avoid data loss. Economics-first — faster tracebacks cut deviation closure lead time by 36 h/incident.

Data: Audit events captured: 5,420 over 8 weeks; clock sync NTP drift ≤±0.8 s; false reject 0.4% → 0.2%; OpEx for investigations −22% by reduced manual log correlation.

Clause/Record: EU Annex 11 §9 audit trails; 21 CFR Part 11.10(e) e-signature controls; BRCGS PM §3.5 record retention; DSCSA/GS1 serial data references; IQ/OQ protocol IQ-IT-031, OQ-IT-032; EBR lot links EBR-PR-2025-05.

Steps:

• Process tuning: Set alarm thresholds for recipe edits: more than 2 changes in 30 min triggers supervisor review.
• Workflow governance: Role-based access — Operators: read/use; Supervisors: approve; QA: release; auto-log lot/shift/operator.
• Inspection calibration: Quarterly audit log integrity check — hash verification SHA-256 mismatch threshold 0; time base drift test ≤±1 s.
• Digital governance: Enable write-once storage; enforce dual e-sign for recipe publish; retain 5 years; backfill tag list in historian.

Risk boundary: If missing approval or time drift >±1 s → Rollback 1: hold affected lots, manual batch record; Rollback 2: IT forensics, restore from prior snapshot, QA re-approval of recipes.

Governance action: Add to quarterly Management Review; evidence in DMS/IT-GOV-011; Owner: QA Systems Manager.

Disaster Recovery for Data/Recipes

Key conclusion: Outcome-first — Tested restores met RTO ≤60 min and RPO ≤15 min for press recipes and audit trails. Risk-first — untested backups created latent loss risk to color curves and die libraries. Economics-first — DR rehearsal (2×/year) avoided 6.5 h downtime/incident, equating to 78k packs protected.

Data: Primary–secondary replication 15-min interval; restore test N=3, success 3/3; average recovery 42 min; Units/min safeguarded 200–240; estimated downtime cost 1.1 k€/h; Payback 9 months vs single outage avoided.

Clause/Record: Annex 11 §7 backup; 21 CFR Part 11.10(c) protection of records; ISO 13849-1 §4.2 safe-state interlock assuring stop-to-safe on historian outage; SAT run SAT-BCP-0404; DR drill record DR-TEST-2025Q2.

Steps:

• Process tuning: Define minimal safe recipe set (speed, tension, LED dose) printed hardcopy for manual mode.
• Workflow governance: Declare DR roles; comms tree; 60-min RTO target; quarterly tabletop plus semiannual full restore.
• Inspection calibration: Verify checksum for backups; restore validation — random 10 recipes per drill; compare ΔE aim, TVI tables.
• Digital governance: Snapshot database every 15 min; offsite S3-compatible storage immutability 7 days; failover runbook link in MES.

Risk boundary: If RPO >15 min or restore >60 min → Rollback 1: run manual centerline recipes at ≤140 m/min; Rollback 2: stop production, escalate to IT-DR L3, execute full system rebuild.

Governance action: Include DR metrics in monthly QMS dashboard; DMS/DR-PLAN-2025; Owner: Head of IT.

Q&A: Practical points operators ask

Q: Can we test promotional sample flows during VR? A: Yes — import a sample order stream and simulate pick/pack to validate label codes and carrier rules; if using a voucher like “onlinelabels $10 off” for sampling, tag the lot so FPY/returns can be traced in the historian log.

Q: What’s the right way to remove obsolete email categories vs production labels? A: Email folders (e.g., queries about “how to delete labels in gmail”) are IT housekeeping, but production label SKUs must be deprecated via DMS change control with retained mapping tables, not deleted, to maintain audit trail continuity.

Customer case: VR rollout and quantified gains

At the Sanford site, the team trained 18 operators in two cohorts and used a voucher labeled “onlinelabels $10 off” to generate small controlled sample orders for shipping verification and barcode grading. Technical parameters: press speed 160–170 m/min, UV-LED 1.4 J/cm², web tension 40 N, SBS 350 g/m² for cartons and BOPP 60 µm for labels. Results over 6 weeks (N=54 production runs): ΔE2000 P95 2.3 → 1.7, registration P95 0.21 → 0.14 mm, make-ready −8.6 min/run, and two DR drills achieved RTO 48–55 min with zero data loss. The VR content library now covers tint curves, opacity, die-change, audit trails, and failover steps — closing the loop for onlinelabels sanford.

Metadata

Timeframe: 8 weeks primary study; 6-week case subset. Sample: N=126 jobs (main study), N=54 runs (case); 5,420 audit events. Standards: ISO 12647-2 §5.3; Fogra PSD §6; EU 1935/2004 Art.3; EU 2023/2006 §5; BRCGS PM §3.5/3.6; ISO 15311-1 §6.2; UL 969; Annex 11 §7/§9; 21 CFR Part 11.10(c)(e); ISO 13849-1 §4.2. Certificates/Records: G7-PR-2405; SAT-OLB-2025-06; PQ-CL-0925; DMS/PROC-ICC-014; DMS/OPQ-CRT-028; UL969-APR25; IQ-IT-031; OQ-IT-032; EBR-PR-2025-05; SAT-BCP-0404; DR-TEST-2025Q2.