Insights

Every weather-routing engine is, at its core, a graph. The geometry of that graph determines whether the routing solution can capture meteorological reality, respect coastal safety margins, and remain computationally tractable. Most providers compromise on at least one. This paper explains how VesselFront's adaptive 22M-node grid resolves all three — and why the result is structurally what you should be after.

Weather routing (WR) is a critical component of maritime operations, directly impacting fuel efficiency, voyage duration, and navigational safety. We present VFWR (VesselFront Weather Routing), an optimization framework for ship weather routing based on the A* path-finding algorithm with a weather-aware heuristic incorporating real-time forecast, wave height threshold, maximum latitude constraint, no-go areas, and minimization of the duration of time within emission control area (ECA). To ensure accuracy based on the updated forecasts, re-routing is performed at six-hour intervals throughout the trip. VFWR is validated against a dataset of 100 historical voyages using hindcast weather data, while calculating optimized routes in a median of 23 seconds and consistently producing shorter distances and durations than real-world trips across all tested voyages.

Our work with TEN produced an ANN-based FOC prediction pipeline validated on real high-frequency vessel telemetry produced for a period of over four years. The work on voyage segmentation, drifting/canal/port removal, fuel type filtering produced a clean RPM signal and a strong data engineering moat.