Predictive ETA that beats carrier estimates for a terminal
A container terminal operator (anonymized)
Illustrative case study. Client identity is anonymized and figures represent the class of outcome MarineAware targets, not a specific contracted result.
Challenge
Carrier-reported ETAs were too noisy to plan berths and yard against, driving demurrage, anchorage queues and wasted at-berth emissions. The terminal needed arrival predictions it could trust across a 72-hour planning horizon.
Approach
We trained ML ETA regressors on historical AIS voyages and layered physics-aware weather routing on top — metocean forcing from CMEMS currents, WaveWatch III waves and ECMWF wind, constrained by GEBCO and EMODnet bathymetry. A congestion model on port-call time series projected queue build-up 72 hours out, enabling just-in-time arrival coordination with inbound carriers.
Outcome
Berth planners now work from a reliable arrival horizon rather than carrier guesswork, cutting anchorage waiting and at-berth emissions and smoothing yard utilisation.
This engagement is representative of MarineAware’s voyage and port optimization work. Figures are illustrative of the class of outcome we target and are anonymized to protect client confidentiality.
Carrier ETAs are not a plan
A carrier’s reported ETA is an estimate made against the carrier’s incentives, not the terminal’s. It updates late, ignores the weather the vessel is actually sailing into, and cannot be planned against with confidence. The result is berths held empty, yards congested, and vessels burning fuel at anchor.
Physics plus history
A pure machine-learning regressor learns the patterns in past voyages. A pure routing model knows the physics of wind, wave and current. The accurate prediction comes from both: an ML estimate corrected by the metocean conditions on the actual route, bounded by the depth the vessel can transit. That is what turns an ETA into something a berth plan can rely on.