Physics-informed performance intelligence
Know what your hull is costing you.
Your ships already send the numbers every day. Naval Brain reads them, applies the physics of how hulls actually behave, and tells you — in real dollars — where the money is leaking.
The problem
Fouling is continuous. Your view of it shouldn’t be twice a year.
Every hull degrades. Every day a vessel sails, a fraction of its fuel is paying for fouling rather than progress — and without performance monitoring, that number is invisible until drydock.
Naval Brain uses ISO 15016 trials data, ISO 19030 residuals, a Kalman filter over the residual series, and a physics-informed neural fit to return one defensible number per vessel per noon: how much of your fuel is paying for the hull.
It runs on data you already collect. It traces back to standards a class surveyor recognises. It’s built in Limassol, close to the fleets it serves.
Hull Intelligence — live dashboard · MV Naval Brain Demo
Kalman + Neural ODE · 95% CI · ISO 19030
Hardware
None
No sensor refits, no fleet hardware rollout. We work from the noon reports you already produce.
Output
$ / day
Fuel penalty in currency. Intervention date. The language the board already speaks.
Standards
ISO 19030
Every output traces to a published standard or peer-reviewed paper. Class surveyors recognise the method.
Diagnostic layer
Hull Intelligence.
The diagnostic layer of the platform. Each noon report becomes a new point on the hull’s degradation curve, with confidence bounds, data-quality flagging, and provenance back to every enrichment step. If the inputs are weak, the output says so.
Built around ISO 19030’s nb_drift, anchored to the vessel’s own sea-trial baseline, cleaned through a Kalman filter and smoothed with a Neural ODE.
Decision layer
Decision Support.
Hull Intelligence diagnoses. Decision Support acts. Every recommendation carries its physical basis, a confidence score, and an expected voyage saving in dollars — ready to issue as a standing order to the Master.
Trim optimisation, speed band, drydock timing. One click to issue, one click to dismiss. The Master, the superintendent, and the commercial team see the same object.
Built on
ISO 15016
ISO 19030
Holtrop & Mennen
Townsin 1981 · 2003
ITTC-78
Naval Brain
Limassol, Cyprus · Founded 2026
Technology
Physics-regularised regression. Not a black box.
Naval Brain combines a hybrid loss — data residuals, physics constraints from ISO 15016 and ISO 19030, and a Kalman smoothness prior — anchored to well-characterised industry standards. Mechanically, it’s a disciplined regression on top of a deterministic mapping layer, where every derived variable traces back to its source.
Hull Intelligence — one screen per vessel. The nb_drift trajectory sits beneath the KPIs.
Hull efficiency · Fuel penalty · Intervention day
How it’s built
No hand-waving. Every decision is defensible.
01
Physics foundation
ISO 15016 (speed-power trials) and ISO 19030 (hull and propeller performance) form the backbone. Baseline resistance from Holtrop & Mennen. Degradation priors from Townsin (1981, 2003). Every claim has a paper or a standard behind it.
02
Data architecture
A single canonical schema is the source of truth. Every enrichment output carries a source tag (reported, AIS-validated, CMEMS-derived, model-estimated) and a confidence score. Full audit trail from raw noon row to the final drift value.
03
Honest trade-offs
ISO 19030’s nb_drift conflates hull and propeller contributions — roughly 70–80% hull, 20–30% propeller, per ITTC and Townsin. We disclose this openly. Phase two separates the two once shaft power or torque telemetry is available.
04
Roadmap
Digitised stability booklet solver (hydrostatics, GZ curves) for trim. Age-aware intervention thresholds. Coating-type priors. A neural operator architecture for cross-vessel generalisation, once multi-vessel data justifies it.
Why it compounds
The derived variables are the product.
Any competitor can access the same raw feeds. The enrichment provenance chain, the hull-only drift, and the propeller separation are ours to build, vessel by vessel. We build data assets, not dashboards.
Team
Meet the people behind Naval Brain.
Contact
Let’s book a call.
Leave a short note and we’ll set up a 30-minute call. Technical, no slides.
Got it.
Thanks — we’ll be in touch within two business days to book a time. If you need us sooner, email [email protected] directly.