Why it matters
Without this system, miscounted leaf input, improper withering, under- or over-fermentation, or incorrect drying can lead to misreported made tea outturn and poor conversion ratios, risking grade, auction value, and certification.
From design to safety
Design
AI informs design for conversion tracking
The system reads planned plucking rounds, clone behavior, and processing capacity to set measurement points, data schemas, and record templates for green leaf input and made tea output. It defines data dictionaries, integration points, and batch identifiers to support end-to-end traceability from garden to cup.
- CTC roller segment
- Rotorvane
- Withering trough
Build
Install sensors and data channels for throughput
Sensors capture leaf batch IDs, fresh leaf weight, wither percentage, fermentation duration, and drying temperature, feeding a central data model for ongoing validation. It specifies wiring maps, enclosure ratings, and calibration schedules for quality control.
- Fermentation bin
- Drying oven
- Made tea weighing scale
Operation
Real-time tracking of mass balance and conversion
AI calculates conversion ratio, identifies deviations between input green leaf and output made tea, and logs batch-level losses for traceability. It generates daily dashboards, exception lists, and batch certificates.
- CTC roller segment
- Withering trough
- Made tea weighing scale
Maintenance
Status checks of processing lines and logs
The model flags wear or drift in rollers, belts, misalignment in conveyors, and calibration needs for scales; prompts service windows, and keeps repair histories with parts replaced and labor hours.
- Rotorvane
- Drying oven
- Fermentation bin
Safety
Audit-ready safety and compliance records
Records for GHG or energy use, sanitation, and PPE usage are tagged to batches; triggers required certifications reviews and safety audits. It maintains incident logs and corrective actions linked to batch IDs.
- Made tea weighing scale
- CTC roller segment
- Withering trough
Part by part
- 1
CTC roller segment
- Watch for
- Roller wear pattern causing inconsistent leaf breaking and mass balance discrepancies.
- AI action
- Detect abnormal torque or temperature drift; halt line if threshold breached and notify supervisor.
- 2
Rotorvane
- Watch for
- Rotorvane misalignment or surface wear reducing leaf extraction efficiency.
- AI action
- Flag output inconsistency; adjust gap and alert maintenance team.
- 3
Withering trough
- Watch for
- Variations in wither percentage across batches indicating airflow or humidity imbalance.
- AI action
- Trigger climate control adjustment and log batch-level wither data.
- 4
Fermentation bin
- Watch for
- Uneven fermentation indicating temperature or microbial activity variance.
- AI action
- Rebalance temperature profile and flag out-of-spec batches.
- 5
Drying oven
- Watch for
- Uneven drying or over-drying seen in moisture readings across trays.
- AI action
- Modulate drying temperature and conveyor speed; record moisture trajectory.
- 6
Made tea weighing scale
- Watch for
- Scale drift causing inaccurate made tea outturn recordings.
- AI action
- Calibrate or replace scale; annotate batch records with measurement confidence.
Worked example
Catching a seal leak before a charter
Illustrative scenario with sample data.
Anomalous outturn data triggers QA review before dispatch of a chartered lot.
Signal in
batch_id=BT104, green_leaf_kg=1320, wither_pct=54, fermentation_min=36, drying_temp=95C, output_made_tea_kg=740, scale_id=MTW-02
System out
alert: outturn variance detected; batch BT104 flagged; QA to inspect line M-1; corrective action memo generated.
Human sign-off: Factory manager reviews anomaly, approves inspection, and authorizes adjusted calibration and retest.
How to roll it out
Phase 1
Phase 1
Baseline data model and sensor network deployed in one factory line and nursery connections.
Phase 2
Phase 2
Expanded data feed to multiple lines and standard operating procedures for data quality checks.
Phase 3
Phase 3
Full governance, audit trails, and integration with auction and certification records.
Where people stay in charge
- Decisions on calibration, maintenance actions, and disposition of suspect batches must stay with the factory manager or senior technical supervisor.
- All data used for safety and regulatory reporting must be reviewed by a certified quality auditor before submission.
- AI outputs used for grade or auction decisions must be corroborated by human tasters and documented in batch sheets.
Questions owners and crews ask
What data signals are needed to track made tea outturn effectively?
The system requires input leaf weight, batch IDs, wither percentage, fermentation duration, drying moisture, and output made tea weight to compute the conversion ratio and reveal losses.
Who approves changes to processing parameters based on AI alerts?
The factory manager reviews critical alerts, approves parameter changes, and records the decision alongside the batch record.
How is compliance with MRLs and certifications maintained?
All process records and testing results are time-stamped, stored with batch IDs, and reported to QA for certification audits.
What happens if a sensor shows drift during production?
Anomaly is flagged, line is halted if needed, calibration is scheduled, and the batch is quarantined until verification.