Why it matters
Without AI, misaligned leaf flow and uneven spreading cause over-withered or under-withered leaf, inefficient trough use, and inconsistent quality.
From design to safety
Design
AI-guided layout for shedding
Model predicts optimal leaf flow paths, trough density, and turning zones from design data to balance moisture removal and throughput, aligning with shed geometry, airflow, shade distribution, and material handling constraints.
- Leaf loading conveyor
- Withering trough
- Spreader grid
- Trough selector solenoid valve
Build
Configuration of withering assets
Data from nursery and factory layout informs sensor placements, trough controls, conveyors, shade mesh alignment, and electrical cabinets for consistent commissioning and test runs under heat-sink simulations.
- Leaf loading conveyor
- Withering trough
- Spreader grid
- Trough selector solenoid valve
Operation
Real-time allocation and turning
AI monitors leaf load, spread density, trough occupancy, humidity, and airflow to allocate troughs and trigger turning cycles, while logging variances, corrective actions, and QA annotations for traceability and QA review.
- Leaf loading conveyor
- Withering trough
- Spreader grid
- Turn and flip mechanism
Maintenance
Predictive upkeep of sheds
Continuous data on trough performance, sensor drift, humidity, cell integrity, and shade wear informs proactive servicing, calibration, replacement planning, and spare-part pooling with maintenance windows.
- Withering trough
- Spreader grid
- Turn and flip mechanism
- Humidity sensor array
Safety
Safety and traceability in processing
System logs ensure auditable records of allocations, turning events, deviations, ergonomic risk checks, and PPE compliance checks for audits, worker protection, and certification readiness across all plants.
- Humidity sensor array
- Trough selector solenoid valve
- Leaf loading conveyor
- Withering trough
Part by part
- 1
Leaf loading conveyor
- Watch for
- Conveyor jams or misalignment causing uneven leaf arrival.
- AI action
- Signal backpressure, reroute leaf flow to balance load and prevent bottlenecks.
- 2
Withering trough
- Watch for
- Trough occupancy exceeding safety or capacity thresholds.
- AI action
- Reallocate leaf to adjacent trough and adjust duration based on humidity and density.
- 3
Spreader grid
- Watch for
- Inconsistent spread density across trough width.
- AI action
- Adjust grid gap or tilt to normalize density and enhance contact area.
- 4
Trough selector solenoid valve
- Watch for
- Valve drift causing wrong trough allocation.
- AI action
- Lock to correct trough channel and flag for maintenance, with audit tag.
- 5
Turn and flip mechanism
- Watch for
- Turning delays or jams impacting uniform withering.
- AI action
- Trigger turning cycle, record timestamp, and QA-check leaf surface after turn.
- 6
Humidity sensor array
- Watch for
- Sensor drift or moisture out of range.
- AI action
- Calibrate or raise alert for field check, initiate partial drying bypass if needed.
Worked example
Catching a trough jam before a charter
Illustrative scenario with sample data.
During a peak plucking day, an overloaded trough risks over-wetting and spillage unless turn rate is adjusted in real time.
Signal in
leafLoadKg=4200; troughOccupancy=85; spreadDensity=0.72; humidity=78; turnInterval=6min; shedTemp=28.5C; airflowRate=1.2m/s
System out
alert: trough 3 occupancy high; action: reallocate to trough 4; adjust turnInterval to 8min; QA to check trough 3 after shift; logs show variance corrections with timestamp.
Human sign-off: Factory supervisor approves reallocation and turn adjustment; QA verifies spreading uniformity and logs the incident; safety officer signs off on corrective actions.
How to roll it out
Phase 1
Weeks 1 to 4: Baseline
Install sensors, calibrate conveyors, establish initial trough mapping, layout power circuits, and test AI rule foundations with dry runs.
Phase 2
Weeks 5 to 8: Pilot
Run live data capture, validate dynamic allocation, adjust thresholds by operator feedback, and record QA annotations for audit readiness.
Phase 3
Weeks 9 to 12: Scale
Expand AI across sheds, integrate with QA, traceability, and certification records; implement governance for data integrity and change management.
Where people stay in charge
- Decisions about when to override AI allocation must stay with the factory manager and QA supervisor to ensure safety and traceability.
- Changes to sensor calibrations and thresholds require written sign-off by the estate manager and safety officer.
- All turning and allocation actions must be auditable and reviewed in the weekly safety and quality meeting.
Questions owners and crews ask
How does AI decide where to allocate troughs during peak plucking days?
The AI cross-references leaf load, trough occupancy, spread density, humidity, and airflow to assign troughs and trigger turning, while logging decisions for audit and QA traceability.
What happens if a sensor reports out-of-range moisture?
The system flags the anomaly, initiates local recalibration where possible, alerts the factory manager, and routes to manual inspection to verify sensor integrity and adjust as needed.
Can this affect made tea outturn or quality grades?
Yes, by balancing withering consistency, which influences moisture removal, aroma development, and eventual grade consistency under QA supervision and certification checks.
What records are kept for certification audits?
Trough allocations, turning events, sensor readings, maintenance logs, QA checks, and safety checks are stored for traceability and compliance across audits.