Ocean AI
Solutions

Biofouling Monitoring

Continuous hull and propeller condition monitoring from your existing vessel data. Know exactly when a cleaning is economically justified — not when the calendar says so.

How it works

From sensor data to maintenance decision.

Predict → Detect → Act
Vessel sensors

Shaft power, GPS speed, draft, wind, IMU — existing data streams

Predict
Physics model

Physics-based speed model — the speed a clean hull should make in these conditions

Speed gap

How far actual speed sits below a clean hull — zero when the hull is clean

Detect
Condition tracker

Hull condition with a calibrated confidence range — not just a single number

Alert fusion

Multiple checks must agree — over 99% of single-sensor spikes are ignored

Act
Recommendation

Clean now / Watch / Alert / Schedule — with expected savings and confidence

Capabilities

Four things no calendar can do.

Zero hardware install

Your vessel already has everything we need.

We connect to the 40 sensor channels your vessel already logs — shaft power, GPS speed, draft, wind, IMU motion, fuel telemetry. No new equipment. No dry-dock installation. No operational disruption. The moment you share a data feed, monitoring begins.

40
sensor channels
0
hardware required
15 min
data cadence
6 m
to first insight
LIVE FEED · 40 sensor channels · 15-min cadence · no new hardware

GPS Speed

12.8kn

Shaft Power

4,336kW

Wind Speed

7.2m/s

Mean Draft

7.9m
+36 additional channels · shaft rpm · torque · fuel density · rudder · roll · pitch · yaw86,925 records
Cause attribution

It knows weather from fouling — and hull from propeller.

Rough seas and heavy loading slow any vessel and burn extra fuel. Our physics model accounts for those effects first, then reads what is left as genuine fouling — and tells hull fouling apart from propeller fouling, because each leaves a different mark across speed and power.

Two signals, not one
Lost speed and extra fuel consumption are tracked together. A real fouling trend shows up in both — and both must agree before we act.
Weather, removed
Transient weather and loading effects are separated out, not mistaken for fouling — so a rough crossing never looks like a dirty hull.
Hull vs propeller
Distinct signatures let us localise the problem and recommend the right action — a hull clean, a propeller polish, or both.
Telling the causes aparttwo signals · weather vs hull vs propeller
What the vessel shows usmeasuredSpeed lossWeatherHullProp.Fuel burn ↑WeatherHullProp.weather & loading removed → genuine fouling, localisedAct onHull foulingPropellerWeather & loading (filtered out)Hull foulingPropeller fouling

Two independent fingerprints — lost speed and extra fuel burn — are read together. The physics model strips out what weather and loading explain, then attributes the remainder to the hull or the propeller. Both signals must point the same way before we act.

Why false alarms stay rareindependent checks must agree before an alert fires
1 of 3 · ignored3 of 3 agreeCheck 1Check 2Check 3AlertquietALERTtime →
Continuous monitoring

Updated every voyage. Not every dry-dock.

Multiple independent signals watch your hull condition simultaneously. They must all agree before an alert is raised — so you act on evidence, not on noise.

Always onContinuous baseline comparisonActual speed compared against what physics says it should be — at every data record, every voyage.
Early warningDetects onset, not aftermathCatches the start of fouling as soon as the data show it — before fuel loss compounds.
Weather-filteredFouling vs rough weatherSeparates a slow fouling trend from a temporary rough-weather slowdown. No false calls after a storm.
Multi-sourceNo single point of failureMultiple independent signals must agree before an alert is raised. One anomaly is never enough.
>99%
of sensor noise rejected

A single noisy reading never triggers a cleaning call — only sustained agreement across independent checks does. You act on evidence, not noise.

Calibrated confidence

Every recommendation comes with a confidence level.

Every recommendation includes a confidence range, not just a single number. In rough weather or unusual loading, the range widens — and we wait for clearer evidence before telling you to act.

Knows when it is uncertain
Unusual conditions — rough seas, rare loading, extreme weather — are detected automatically. The system widens its confidence range instead of guessing.
Acts only on strong evidence
A cleaning recommendation fires only when the pessimistic end of the confidence range crosses the threshold — not just the best-case estimate.
Always gives an answer
Even in difficult conditions the system gives a recommendation — just with a wider confidence range so you know how much to trust it.
Predicted speed with confidence range · unusual-conditions flag
Unusual ↑Unusual ↑Confidence range10 kn12 kn14 kn16 knUnusual-conditions scorethresholdObserved GPS speedModel prediction
90%
Confidence range
calibrated, every reading
Flag
Unusual conditions
automatically trusted less
±0.05 kn
Typical range width
calm water, full sensor coverage
The problem

Calendar maintenance wastes money every cycle.

Vessels run at heavy fouling penalty for months before the schedule permits cleaning. The schematic shows the real cost in fuel.

Extra fuel burn over time · calendar cleaning vs Ocean AI
Recommend cleaning (13%)4.6 months excess foulingWasted fuelOcean AIt = 8.4 mCalendart = 13 mhull resets → regrows0%5%10%15%20%0 m3 m6 m9 m12 m15 mFuel penalty (%)Months since last cleaningCalendar (no trigger)Ocean AI (post-clean regrowth)

Calendar cleaning runs on a fixed interval — the hull sits at a 13–16% fuel penalty for over four months before anyone acts. Ocean AI cleans the moment the cost of waiting outweighs the cost of cleaning, removing the wasted-fuel zone entirely. After each clean the hull resets and fouling regrows (dashed blue line).

24-month comparison · colour shows fouling severity and fuel penalty
Calendar-based (fixed 12-month interval)clean ↑ 12mclean ↑ 24mheavy foulingOcean AI (threshold-triggered optimal cleaning)clean ↑ 8.4mclean ↑ 16.8mlow foulinglow fouling0m3m6m9m12m15m18m21m24mClean hullLight foulingModerateHeavy fouling

Both scenarios use the same vessel and fouling model. Calendar approach runs at heavy fouling (orange-red) for ~4 months before each cleaning. Ocean AI triggers at the economic threshold, keeping the hull in the green-yellow zone throughout. Same number of cleanings — less fuel waste.

Product output

This is what you receive.

A live condition estimate, a calibrated alert tier, and actionable metrics — updated continuously from your vessel's operational data.

LIVE · Confidential vessel · North Atlantic transit
Updated 2 h 14 m ago

Hull condition — extra fuel burn

+8.0%-0.31 kn speed loss

confidence range [-0.38, -0.24] kn

NominalClean ↑Drydock ↑
30-day condition trend−0.44 kn / month drift
30 days agoToday

Status

Recommend Cleaning

Fuel penalty

+8.0%

vs clean baseline

Annual loss

$145K

at $5K/day fuel

Payback

38 days

after cleaning

Confidence

High 92%

calibrated confidence

Four-tier maintenance framework.

We act only when the confidence range — not just the best guess — clears the threshold, so rough-weather noise never triggers a false cleaning call.

ActionTriggerSpeed lossFuel penalty
WatchAlert score ≥ 1< 0.5%< 1%
AlertSpeed loss ≥ 0.07 kn~0.5%~1.5%
Recommend CleaningSpeed loss ≥ 0.28 kn or alert score = 3~2%~4.3%
Schedule DrydockSpeed loss ≥ 0.70 kn~5%~10.7%

Fouling severity reference

LevelFuel penaltySpeed loss
Light5%−0.5 kn
Moderate~10%−1.0 to −1.5 kn
Heavy20%+−2.0+ kn

IMO 2023 Biofouling Guidelines; Schultz et al., Biofouling, 2011

Ready to monitor your fleet?

Get in touch to discuss data requirements, integration, and pricing.

Request a Demo