Harvest Intelligence

September harvest to date (1–28 Sep) · forecasting, scheduling, labour and evacuation optimisationAll operations

Planned harvest (MTD)

54,308t

Month budget 58,186 t

Actual harvest (MTD)

52,306t

−2002 t vs plan

Harvest completion

96.3%

actual vs planned to date

Harvested today

1,848t

5 estates · as of 07:30

Loose fruit ratio

2.6%

of FFB weight · target ≤ 2.5%

Harvesting interval

9.8days

4 blocks on > 12-day rounds

Worker productivity

1,469kg/HK

1,962 harvesters · aggregate

Block productivity

2.13t/ha

FFB per planted ha, MTD

Unharvested area

1,159ha

past a 10-day round

Est. crop remaining

3,778t

ripe crop to month-end (AI)

Harvest Forecasting

Harvest Forecasting v2.0

Estate C −11.0% month-end

11,112 t forecast vs 12,485 t September budget

Confidence84%

Scheduling Optimization

Round Scheduler v1.3

Prioritise B29 → B17 → B22

4 blocks on > 11-day rounds; sequence restores a 10-day cycle in 6 working days

Confidence82%

Labor Optimization

Harvest Labor Optimization v1.1

+14 harvesters → Estate C for 6 d

From Estate D · Division 2; recovers ≈ 680 t FFB · ACT-004

Confidence79%

Crop Evacuation Optimization

Evacuation Router v1.4

Add 2 trucks to Estate C AM shift

CP dwell time 7.5 h → 4.0 h; restan at CP-C3/C5 cleared before 14:00

Confidence80%Open →

Ripeness Prediction

Ripeness CV + Phenology v2.2

418 t ripe in next 3 days

Across 50 blocks · peak Wed 30 Sep (post-rain ripening); fraction ripe 5–10 loose fruits/bunch

Confidence81%

Missed Harvest Detection

Drone RGB + CP ticket reconciliation

643 ripe bunches missed

B29, B22, B27, B17, B36 · ≈ 12.9 t at risk — re-pass within 48 h

Confidence77%

Estate harvest performance

September to date · month-end AI forecast

Estate	Planned	Actual	Completion	Month-end	Loose fruit	Round	kg/HK	Attendance	Crop left
Estate A	12,380	12,607	101.8%	+1.8%	2.1%	9.3 d	1,640	95%	896 t
Estate B	10,843	10,334	95.3%	−4.6%	2.6%	10.4 d(1 >12 d)	1,510	94%	749 t
Estate E	8,685	8,579	98.8%	−1.2%	2.3%	9.1 d	1,470	94%	614 t
Estate C	11,653	10,346	88.8%	−11.0%	3.6%	10.7 d(3 >12 d)	1,190	86%	766 t
Estate D	10,747	10,440	97.1%	−2.8%	2.4%	9.4 d	1,560	95%	753 t

Estate C is 11% behind

Confidence84%

Month-end forecast 11,112 t vs 12,485 t budget (−11%). Harvester attendance fell to 86% (group 94%) and 3 blocks (B22, B27, B29) are on 13–14-day rounds, pushing loose fruit to 3.6%. Reallocating 14 harvesters from Estate D Div 2 for 6 days recovers ≈ 680 t — see the planner below and ACT-004.

Daily harvest vs plan

t FFB per harvesting day · 1–28 Sep (Sundays excluded)

AI Harvest Planner Optimisation

Scheduling, labour and crop-evacuation optimisation for the next 2 working days

Estate & blocks

6 blocks · 203 t ripe crop (P50)

Predicted crop

Available harvesters80 HK

Include +14 reallocated from Estate D · Division 2

Weather

Road condition

Truck availability5 trucks

Set constraints and run the optimiser

The planner sequences blocks by ripeness and round length, right-sizes harvester gangs, and assigns trucks and collection points to minimise restan and harvest-to-mill time.

Reallocate 14 harvesters from Estate D · Division 2 for 6 days → Estate C · Blocks B22, B27, B29

Recovers ≈ 680 t FFB (half of the 1,370 t Estate C gap) · ≈ Rp 2.06 bn · Estate D Div 2 is 4% ahead of its round plan.

Confidence79%

Harvester productivity distribution

Share of harvesters by kg/HK band · September · aggregate by estate

Anonymised aggregate

<900 kg/HK900–1,200 kg/HK1,200–1,500 kg/HK1,500–1,800 kg/HK>1,800 kg/HK

Productivity analytics are aggregate and anonymised — no individual harvester is scored or ranked here. Lower bands usually reflect terrain, crop density, overdue rounds and tall palms rather than individual effort.

Estate C has 45% of harvesters below 1,200 kg/HK, consistent with long rounds (more loose fruit to collect) and 86% attendance — a capacity issue, addressed by reallocation rather than individual measures.

Human review required: any staffing, premium or disciplinary decision must be made by estate management under company HR policy. The model makes no automated employment decisions.