WindMar: Routing around Typhoon Dolphin
Download the full illustrated study (PDF)
Southwest of Japan, Typhoon Dolphin lay directly across a laden passage. This is a worked case study of how WindMar’s isochrone optimiser handles heavy weather, what it saves, and, more importantly, where it puts the ship.
The situation
At the analysis time (forecast valid 07 August 2026), a compact, deepening typhoon lay southwest of the Ryukyu Islands near Okinawa, central pressure about 968 hPa, gusts to Beaufort 12, and significant wave heights above 10 m along the direct line, tracking from south of Japan towards Taiwan. The centre sat astride the direct track between 34°N 137°E and 28°N 122°E, so the rhumb-line route crossed the core.
The passage is laden, at a constant 14.0 kn through-water speed (Best-ETA), departing 05 August 2026.
Baseline versus optimised
Both routes are evaluated over the same weather, at the same constant engine order.
| Route | Fuel | Δ vs baseline | Passage | Distance | Avg SOG | Max roll |
|---|---|---|---|---|---|---|
| Baseline, held through the storm | 71.9 MT | , | 69.2 h | 943 nm | 13.6 kn | 33.9° |
| Optimised, isochrone | 68.4 MT | −4.9% (−3.5 MT) | 69.0 h | 981 nm (+38) | 14.2 kn | 17.0° |
The optimised track is 38 nm longer yet arrives 0.2 h earlier on 3.5 MT less fuel. There is no contradiction: the engine order is unchanged, but in the seaway the direct route is dragged to 13.6 kn average speed-over-ground by added resistance, while the optimised track holds ~14.2 kn in calmer water. The saving is routing, not slow-steaming.
What the case shows
The fuel is the side story. The headline is where the optimiser put the ship. It guided her into materially safer water, significant wave 7.4 m → 5.8 m at the crux, and out of the Force-10 wind core, for the same ETA. Held on the direct track, the passage sees sustained Force 10 leg after leg, seas past 10 m, and roll to 33.9°.
The dodge is a hard exclusion. WindMar’s isochrone wavefront prunes survivors by distance-to-destination; a graded cost penalty only breaks ties. The storm avoidance is therefore produced by the hard severity wall, wavefront cells inside the storm (significant wave > 6 m or wind > 45 kn) are dropped from the search, equivalent to extending the land mask. That hard exclusion is the reliable route-bender; the soft weather-cost term refines the chosen path but does not, on its own, divert.
Seakeeping
WindMar’s seakeeping module screens the chosen route for resonant rolling and flags the synchronous-roll legs, where the wave encounter period meets the ship’s own natural roll period. On this passage peak roll falls from 33.9° to 17.0°, though the early legs still pass through synchronous water: the passage remains challenging, not benign. The figure is an IMO MSC.1/Circ.1228-class screen on a GM roll model, a resonance indicator, not a validated 6-DOF motion prediction, so read it as a relative safety margin, not a guarantee.
Method and honesty
- The baseline is the direct track held through the storm, weather-evaluated at the same constant 14.0 kn, not an actual sailed voyage. Every saving is quoted against that naive straight run, never against what a prudent master would sail, who would also dodge.
- Same engine order throughout, so the gain is routing alone, not slow-steaming.
- The route bends via hard exclusion of the storm core, the only reliable lever in the engine; graded costs bias tie-breaks only.
- Seakeeping is a screen, not a prediction, GM-based, coefficients unvalidated against the real ship’s motions.
- Configuration: WindMar v0.2.3 (build b193), Mode A constant-power isochrone. Winds NOAA GFS 0.25°; seas CMEMS WAV 1/12°; currents SMOC. Forecast vintage 05 August 2026.
Longer route. Less fuel. Same ETA. And, above all, a steadier, safer ship.
Decision-support only; not for navigation.