K↗KASYNETSOPERATIONS / DECISION INTELLIGENCE

06 / DAILY OPERATIONS · OMS · SAFETY

One day.
One accountable plan.

Compare three staffed shifts. See what gets picked, what reaches handover and what the next crew inherits.

View live operations ↓
ACTIVE DAILY PLAN / BASE UNITS300,000
Sorter 280,000 Classic 20,000
OUTER cartons × SKU pack + INNER pieces. Classic uses individual pieces.
PREPARING06:00:000.00 / 24 h elapsed

Preparing the first paired day…

Local shift time06:00–14:0014:00–22:0022:00–06:00 +1

Waiting for a validated plan.

PICKED / PLAN—
HANDED OVER—
AWAITING HANDOVER—
ORDERS ON TIME AND IN FULL—
LIVE / SAME EVENT MODELDIGITAL TWIN
THE CREW RELAY

Three shifts. Continuous work.

Backlog, stock and carriers persist.
Shift / timeSorter plan → pickedClassic plan → pickedOverdue picking remainder
SAME DEMAND / SAME CLOCK

Compare the whole day.

Interim result
MeasureOld systemDigital TwinChange

Run the paired day to assess coverage and the remaining bottlenecks.

Daily evidence & measurements

Put timings against observations.

Daily orders, locations, traffic and capacity are simulated. Imported samples can inform stop times, Classic quantities and carrier placement. Validation rows report fit error for those components. A full-day prediction needs independent warehouse outcomes.

Reference timings are active until a plan with observations is applied.

CSV format and fitting rules

sample_id,group_id,split,phase,quantity,duration_s

  • Use anonymous sample and group IDs. Give each group one split, calibration or validation.
  • Clean stops use outer (cartons), bin (pieces) or classic (pieces). Include ordinary travel from the previous stop through confirmation. Record carton cutting, queues and other interruptions outside this interval.
  • For sorter_handover, classic_handover, sorter_wait and classic_wait, use quantity 1. Handover measures one carrier placement after the slot is available. Wait rows provide a reference, while queues are generated by the simulation.
  • A phase needs eight calibration rows in two groups to apply a median. A stop curve also needs two distinct quantities. Within the observed range the model interpolates between medians. Other quantities use the reference range. Holdout error uses validation rows only.