KPT logo KPT
← Back to KPT

Industry

Ports

Where every hour a vessel waits costs $50k+, and every crane move multiplies up the chain.

Container terminals run a tightly coupled system of berth allocation, ship-to-shore crane assignment, yard sequencing, and gate flow — against a clock where a single waiting vessel burns $50k+ per hour. The combinatorial space is too large for spreadsheets and too dynamic for off-the-shelf TOS optimizers. KPT mathematically models berth + crane + yard decisions together, predicts equipment failures on the capex-heaviest assets, and integrates with your existing TOS / SAP setup without customization.

Pipeline prospect Target sector — South American terminal operators (Santos Brasil, Wilson Sons, DP World, Empresas Portuarias Chile).

How KPT helps

A decision-support layer for terminal operations.

Port operations are the canonical multi-resource scheduling problem: berths, ship-to-shore cranes, RTGs, yard slots, and gate lanes all interact, and the cost of getting it wrong is denominated in waiting-vessel hours at $50k+ each. KPT's optimizer attacks the terminal from three directions at once:

  • Allocate berths and ship-to-shore cranes jointly across the rolling vessel schedule.
  • Sequence yard moves and RTG assignments to minimize re-handles and gate wait.
  • Predict bearing, hoist, and trolley degradation on cranes + RTGs 7–14 days before failure.

All recommendations flow through an operator-approval UI. No write-back to TOS / SAP happens without explicit human sign-off. Shadow-run trust gates A/B-test every variable for 30 days before promotion — the same glass-box discipline that runs at the PVM reference deployment.

Specific outcomes KPT delivers

  • Berth + Crane Joint Allocation

    Cut vessel waiting time and ship-to-shore crane idle simultaneously across the rolling schedule.

  • Yard Sequencing + RTG Assignment

    Reorder container moves to minimize re-handles and balance load across RTGs.

  • Vibration-Anomaly Predictor

    Flag bearing, hoist, and trolley degradation on cranes 7–14 days before unplanned failure.

  • Probabilistic Demand Forecaster

    P10 / P50 / P90 vessel-arrival and cargo-volume quantiles per terminal per week.

Where to start

A 2-week PoC on a single terminal, with full read-only deployment and an operator-approved-only write path. Atlas leads — berth + crane joint allocation is the highest-leverage entry point. One avoided hour of vessel waiting pays for a quarter of pilot.

Port Terminal PoC

2 weeks to a live PoC.

One terminal. One measurable KPI — vessel waiting hours, gross moves per crane hour, or re-handle rate. A 2-week assessment, a shadow-run validation, and measured impact in dollars. The same playbook that proved out at PVM, retargeted to the quay.