Aurora Foods Lowered ppm Defects to 210 on ASFL, FPY 98.2%

Aurora Foods Lowered ppm Defects to 210 on ASFL, FPY 98.2%

Aurora Foods cut pack-out defects to 210 ppm on three ASFL lines while lifting First Pass Yield to 98.2% and OEE to 86.9% in 12 weeks (N=3 lines, 28 lots). Changeover moved from 38 to 24 minutes; energy intensity from 0.064 to 0.056 kWh per pack; MTBF rose from 9.6 to 14.1 hours and MTTR fell from 38 to 24 minutes. We used Single‑Minute Exchange of Die parallelization, recipe locks with electronic signatures, and airflow re‑zoning at the shrink tunnel. Evidence anchors include ISO 13849‑1 Performance Level d guarding, HACCP plan CCP‑Seal‑02, and SAT record SAT‑24‑118 with GS1 case aggregation enabled. Payback modeled at 7.5 months based on scrap, labor, and rework deltas. Validated during OQ‑23‑441 and PQ‑23‑447 runs onsite.

Results and Evidence

Metric	Baseline	Result	Window / N	Confidence/Range	Evidence
OEE	78.4%	86.9%	12 weeks / 3 lines	90% CI ±1.2%	SAT‑24‑118; OQ‑23‑441
FPY	93.5%	98.2%	28 lots	90% CI ±0.6%	PQ‑23‑447 lots 1–28
ppm Defects	710 ppm	210 ppm	28 lots	Range 180–240 ppm	CCP‑Seal‑02 records
Changeover	38 min	24 min	18 events	Interquartile 22–26 min	SMED worksheets
kWh/pack	0.064	0.056	12 weeks	±0.002	Energy log EL‑24‑Q2
MTBF	9.6 h	14.1 h	12 weeks	±0.8 h	CMMS trend TR‑MB‑24
MTTR	38 min	24 min	32 tickets	±4 min	CMMS tickets
Payback	—	7.5 months	1 site	Sensitivity ±1.2 mo	Econ model FIN‑ASFL‑24

Economics

Item	Value	Notes
CapEx	€18,000	Sensors, guards, HMI licenses
OpEx delta	+€2,400/yr	Calibration and spares
Scrap avoidance	€140,000/yr	Defects 710→210 ppm
Rework labor	€52,000/yr	MTTR 38→24 min
Energy	€24,000/yr	0.064→0.056 kWh/pack
Changeover labor	€62,000/yr	38→24 min; N=18/wk
Net annual	€275,600	Payback 7.5 months

Compliance Mapping

Clause/Standard	Control / Evidence	Audit cadence
ISO 13849‑1 PL d	Interlocks, estops; SAT‑24‑118	Annual
HACCP/HARPC CCP‑Seal‑02	Decay/peel logs; WI‑ASFL‑12	Per lot
GS1 aggregation	Case serials; verifier logs	Daily
Annex 11 / Part 11	e‑signatures; HMI audit trail	Quarterly

Target Markets and Demand Variability

Demand across soups, ready meals, and nutraceutical sachets fluctuated weekly; we aligned maintenance triggers to forecasted changeovers to stabilize uptime. On the Aurora Foods case, each line averaged 17 changeovers per week; baseline OEE 78.4% and FPY 93.5% reflected setup drift and seal variability. We created a market-linked maintenance cadence with a 72-hour lubrication cycle on conveyors, a 168-hour vacuum system integrity check, and a seal jaw swap every 500,000 packs. Record HARPC-DA-19 documented hazard controls tied to allergen changeovers. Key conclusion: linking preventive tasks to demand variability cut unplanned stops from 11 to 6 per week in four weeks. Risk boundary: lockout and tagout applied before jaw removal; residual pressure verified ≤0.1 megapascal at manifold gauge and recorded readings.

Market-driven pack sizes required tool-less rail changes and consistent film tracking. We standardized a pre-change kit that staged seal bars, O-rings, knives, and certified torque tools beside line-side bins. For liquids, the soup cell operates a liquid ASFL vacuum sealerealer with a vented nozzle; material lists note liner size by recipe to prevent cavitation. Operators log film width, jaw temperature, and bag lot to enable traceability. When running vacuum sealer bags for family packs, we enabled a 9-point web alignment check to avoid wrinkled seals. Clause reference: QMS work instruction WI-ASFL-12 governs film threading. Risk boundary: do not heat-soak new jaws without a verified thermocouple; enforce 20 minutes conditioning before rate-up to avoid unstable seals during the first pallet run.

1. Publish a weekly changeover forecast tied to maintenance windows.
2. Pre-stage seal and knife kits with torque certificates.
3. Inspect vacuum hoses and clamps; replace per hour meter.
4. Verify seal jaw wear and plan swaps at 500k packs.
5. First-article inspection before rate-up; QA release by lot.

Leak Rate and Seal Integrity Baseline

Key conclusion: a clear leak-rate baseline allows swift isolation of seal faults versus vacuum faults. At kickoff we measured vacuum decay at 0.8 millibar per second across 20-pack samples, while dye test failure averaged 710 ppm. After jaw rebuilds and vacuum manifold reseals, decay tightened to 0.3 millibar per second and dye failures tracked 210 ppm over 28 lots. Customer case: the soup line (ASFL vacuum sealerealer for soup) needed vent profile changes to release entrained steam. Q: Can the unit run thin broths without corner leaks? A: Yes, with a 12° jaw bevel and 0.4 second extra dwell at 74 °C, verified in OQ-23-441. Risk boundary: never exceed 80 °C at the jaw thermocouple during validation under any condition.

We standardized three tests per shift: vacuum decay, bubble immersion, and tensile peel. Data are logged to MES with electronic signatures per Annex 11/Part 11 Clause 12, retaining raw curves for audit. HACCP Critical Control Point CCP‑Seal‑02 defines acceptance at ≤0.4 millibar per second and peel force 18–24 newton. To train operators, the HMI hosts a concise card titled vacuum sealer how to use that maps each alarm to a maintenance action. Evidence from SAT‑24‑118 shows 95% of faults isolated within two test cycles. Risk boundary: do not submerge packs above 60 °C in bubble tests; handle scalding risk. For allergen runs, verify twin seals and serialized labels per GS1 before releasing the first case on each shift report template.

1. Calibrate vacuum transducers and thermocouples prior to tests.
2. Run vacuum decay on 20 packs; record mbar/s values.
3. Execute bubble and dye tests on any suspect lot.
4. Trend peel force; set alerts at 18 and 24 newton.
5. Escalate to root-cause checklist; document corrective actions.

HMI Design, Roles, and Permissions

Key conclusion: locking recipes with roles and signatures eliminates drift that masquerades as machine faults. We created roles for Operator, Maintainer, and QA with electronic signature per Annex 11/Part 11 Clause 7 and dual verification on critical parameters. Recipe setpoints moved from free‑edit to request‑change workflow with audit trails. In four weeks, unauthorized edits dropped from 22 to 1 across three lines; FPY stabilized near 98.2% (N=28 lots). HMI alarms reference maintenance guides and spares. SAT‑24‑118 screenshots demonstrate traceable approvals. Risk boundary: only QA can release sealed parameter libraries; temporary deviations expire automatically at lot close, preventing latent carryover to allergens. Backups replicated to two servers and a local historian to sustain recovery targets during validated disaster recovery drills quarterly.

To reduce configuration time, we templated HMI layouts and enforced consistent color codes and iconography. Operators previously learned on consumer models like an inkbird vacuum sealer, so we mapped plain-language prompts to industrial terms. Mean time to configure a new recipe dropped from 19 to 9 minutes with SMED-style kitting (N=18 changeovers). Clause reference: SOP‑HMI‑21 details button hierarchy; Annex 11 Clause 9 covers user access reviews. Record CAPA‑24‑031 closed training gaps and archived quizzes. Risk boundary: do not grant Maintainer role to temporary contractors; grant time‑bound tokens only. When deviating a recipe, require QA co‑sign within 15 minutes or the system reverts to last qualified parameters, avoiding unsupervised drift across shifts and maintaining audit readiness across seasonal staffing peaks companywide.

1. Define roles; enforce least privilege by task.
2. Lock recipes; enable e‑signatures and reason codes.
3. Map alarms to maintenance actions with part numbers.
4. Review audit trails weekly; close deviations.
5. Back up and checksum recipe libraries nightly.

Gauge R&R and Calibration Routines

Key conclusion: measurement systems with known repeatability isolate causes faster. We executed Gage Repeatability and Reproducibility on vacuum transducers and jaw thermocouples. Vacuum sensor R&R was 8.4% and jaw temperature R&R 7.1% across three operators using 10 parts and 3 trials per part. Resolution met 10:1 against tolerance. Calibration intervals set at 90 days for vacuum, 30 days for temperature. IQ‑23‑117 and OQ‑23‑441 capture calibration certificates and as‑found/as‑left data. Data showed drift of +0.5 °C per week on Zone 2; a bracket was redesigned. Risk boundary: do not reuse damaged probes; apply food‑grade thermocouple grease only. Verify instrument IDs before logging; using wrong device IDs corrupts the audit trail and triggers retraining for affected maintenance technicians when detected and documented immediately.

Seal force and vacuum accuracy mean little without packaging identity control. We installed GS1 barcode verification at case pack with ISO/IEC 15416 grading, linked to the machine’s serial number. Correlation between peel failures and grading D or worse highlighted film tension errors. Calibration records reside in CMMS with alerts; PQ‑23‑447 validated the end‑to‑end chain. Data: grading C or better correlated with FPY above 98% across 28 lots. Clause: Part 11 Clause 17 enforces time‑stamped audit trails. Risk boundary: never bypass verification on rework cases; serialize reintroduced units. For trace recalls, export aggregation files daily at 18:00 to the QMS repository, sustaining retrieval within Service Level Agreement targets during audits and preventing mislabel release during weekend unattended shifts across all lines.

1. Select instruments with 10:1 resolution to tolerance.
2. Conduct R&R with 10 parts, 3 operators, 3 trials.
3. Set calibration windows; attach certificates to assets.
4. Trend drift; trigger engineering change if repeatable.
5. Verify GS1 grading before first-case release.

Parameter Table (Target/Current/Improved)

Parameter governance is the backbone of repeatable seals. We published a parameter table showing current, target, and improved values with ranges and lot notes. The matrix applies to dry runs and to a liquid ASFL vacuum sealerealer, including soup recipes needing vent time. Target vacuum level −78 to −82 kilopascal, seal bar 72–76 °C, dwell 1.6–2.0 seconds, and belt 18–22 meters per minute. Data from 28 lots show packs within these windows hold ≤0.4 millibar per second decay. Clause: SOP‑PROC‑07 defines change thresholds. Risk boundary: never chase defects by pushing temperature outside target; diagnose root cause first. Payback sensitivity shows material and rework deltas dominate, so parameter discipline remains the most controllable lever for high-mix seasonal demand at Aurora Foods.

For change control, we mapped each parameter to a gage, a calibration source, and a verification test. The ASFL vacuum sealerealer for soup uses a dual-vent recipe; target jaw offset is 0.3 millimeter to bleed vapor before full contact. Data: with offset enabled, we saw peel strength center-to-corner Cpk move from 1.09 to 1.46 over 12 lots. Clause: Engineering Change Order ECO‑24‑112 covers parameter promotions. Risk boundary: promote only after three consecutive lots meet limits and audit trail is closed. When parameters change, serialize the recipe version in the QR case label for line clearance. This creates a clear audit trail tying the run to the validated window and maintains parameter lineage across shifts, sites, and maintenance interventions without ambiguity.

1. Publish parameter windows with min/max and rationale.
2. Train operators; pin quick-reference cards on HMI.
3. Control-chart vacuum decay and peel by lot.
4. Approve changes via ECO; link OQ/PQ evidence.
5. Serialize recipe version on the GS1 case label.

Parameter Matrix

Parameter	Current	Target	Improved (median)	Sample/Window	Notes
Vacuum level (kPa)	−74 to −76	−78 to −82	−80	28 lots / 12 wk	Decay ≤0.4 mbar/s
Seal bar temp (°C)	68–70	72–76	74	28 lots	Zone 2 drift bracket updated
Dwell time (s)	1.2–1.4	1.6–2.0	1.8	28 lots	Liquid recipes +0.4 s
Belt speed (m/min)	24–26	18–22	20	12 wk	Stabilizes jaw cooling
Jaw offset (mm)	0.0	0.3	0.3	12 lots	Vapor bleed for soups

These controls replicate across sites because they standardize how technicians verify sealing physics, HMI authority, and measurement integrity. If your ASFL runs multi‑recipe portfolios, adopt the same cadence: link maintenance to demand, baseline leak rate, lock recipes, certify gauges, and govern parameters with evidence. The ASFL gains are sustained when every change carries an audit trail and a measurable window.