Route and speed decision under weather-conditioned vessel behaviour.
P1-Maritimo is a physics-based voyage optimisation engine for container shipping. It computes the route-and-speed configuration that minimises fuel consumption within safety and schedule limits, then recomputes it continuously against the weather field as it evolves. The result is a routing decision that is always the current one, corrected and recomputed against the conditions present at the moment it is issued.
Continuous decision.
Most routing systems use one weather forecast across the entire route, treating it as equally accurate at every point along the way. In practice, forecast accuracy varies by location, terrain, and time. A model that is reliable in aggregate can be systematically off at a specific point of the corridor. Maritimo corrects for this variation. At each point along the route, a dedicated calibrator measures how the upstream forecast has been performing against actual observations, and applies the correction before the routing layer. The system runs this correction every hour, so the forecast underneath a Maritimo decision is never the raw model output; it is the model output corrected against the most recent inputs at every point the vessel will cross.
A route is not one forecasting problem. It is a thousand forecasting problems, each with its own error pattern. Maritimo is built around that structure. Location-specific calibrators run independently on their own compute, refreshed every hour, and the routing optimisation runs over the corrected field they produce. Where traditional routing ends, Maritimo begins.
Observed across evaluated voyages in replay analysis
Blended across vessel classes at current fuel prices
Asia-Europe to Transatlantic
Completed voyages, replayed against the live decision engine
Acceptance corridor
Every voyage operates inside a corridor of admissible routes, bounded by safety, schedule, and operational constraints. Existing routing systems consume the upstream forecast as-is; the weather along the corridor varies as the upstream model reports it, but the accuracy of that variation is not measured at any specific point. Maritimo treats each point of the corridor as a separate forecasting problem. A dedicated calibrator at each point measures how the upstream forecast has been performing against observations at that specific location, and applies the correction before the routing layer computes over it.
The system recomposes itself every cycle. Forecast precision at any given point of the corridor reflects that point's own observation record and correction mechanism. Routing decisions are made over this point-wise calibrated field, recomputed at every cycle as the field itself is recomputed.
Weather-conditioned environmental modelling
Wind, wave, and current forecasts are corrected against actual observations at every point on the route, every hour. Open weather models are accurate at scale but consistently wrong in known, local ways. Maritimo learns those local errors and removes them before the routing layer makes a decision. The quality of the routing is bounded by the quality of the inputs. This is where the system spends most of its effort.
Vessel performance modelling
Hydrodynamic resistance and propulsion models calibrated to individual vessels, not class averages. Two ships built to the same specification will burn fuel differently in the same conditions, and the same hull changes over time as it ages and accumulates fouling. Maritimo's fuel model is built per vessel and refreshed against observed performance. The model the routing decision is computed against is the model of that ship as it is now.
Constrained route optimisation
Speed choices on one leg of a voyage compound into fuel and schedule consequences on every leg that follows. Maritimo evaluates each leg in the context of every other leg, then returns the speed profile that produces the lowest total fuel burn within the active safety and schedule constraints. The optimisation runs hourly, against the field as it then exists, so the recommended profile is the one the system produces in the current cycle, not the one it produced at voyage start.
Weather Corridor
The map below shows the corridor for the selected voyage, the space of admissible routes, populated by the calibrated weather field. Each point is a forecasting node with its own calibrator, refreshed hourly against the latest observations at that location. The optimisation layer reads across these nodes and computes the route that produces the lowest fuel cost while remaining within operational limits. When adverse conditions appear at any point in the corridor, they are the conditions the recommendation is being computed against for that segment of the voyage.
Voyage replay
Every voyage this fleet has sailed is a data point Maritimo can be tested against. For each completed voyage, the system computes the recommendation it would have produced at every cycle of that voyage, using only the forecasts and observations available at each moment. The recommendation is compared to the route actually sailed and the fuel actually burned. The system is not shown the outcome. It is shown the voyage as it unfolded, and asked what it would have decided.
Routing economics
Select a fleet composition to see the projected annual saving across vessel classes. Figures are per-vessel and per-year, computed from the saving rate observed in the replay analysis for this fleet. Total fleet impact scales linearly with the number of vessels in each class. Every figure is traceable to the forecast precision, vessel model, constraint set, and decision cycle that produced it.
Current operating baseline
The system operates on high-resolution global forecast data. Replay analysis is conducted in the same way the live system operates: at every cycle of a past voyage, Maritimo computes the recommendation it would have produced using only the forecasts and observations available at that moment. Nothing about the analysis is retroactive except the fact that the voyage has already been sailed.
Current results are a baseline, not an upper bound. The system outperforms existing routing solutions on standard forecast inputs before any operational data integration.
Expansion with operational data
Integration with vessel-level data sources including onboard sensors, vessel-specific performance telemetry, and proprietary weather observations increases modelling precision and further improves routing outcomes.
Maritimo is the first full-stack Principia application, a domain-specific instantiation of the Intelligence OS architecture applied to maritime operations. Calibrated environmental intelligence from P1-Tempo feeds directly into the routing optimisation layer. Forecast precision at the Tempo layer determines the quality of the field Maritimo optimises over, which determines the speed and route choices on each leg, which determine the realised fuel outcome. Routing decisions are traceable to the forecast precision, vessel model, and constraint set that produced them.