Fuel Consumption Prediction for a Ferry using ML and In-service Data

Published in Ocean Engineering (Elsevier, 2023), this paper presents a machine learning approach for predicting ferry fuel consumption using only operational data collected during normal service — no controlled experiments or dry-dock measurements required.

Context

Ferry operators face a practical dilemma: fuel efficiency depends heavily on operating conditions (speed, loading, sea state, weather), but the relationship is non-linear and difficult to model analytically. Traditional approaches use computational fluid dynamics (CFD) models that require detailed vessel geometry and expensive experimental calibration.

We asked: can we achieve useful fuel consumption predictions using only the data already collected by the vessel's onboard systems?

Data and Method

We worked with in-service operational data from a BC Ferries vessel, including engine telemetry, GPS/speed logs, draft measurements (load proxy), weather data, and fuel flow meter readings. This data is routinely collected but rarely used for optimization.

Feature engineering: Speed through water, engine RPM, pitch/roll from IMU, wind speed/direction relative to vessel heading, draft (as loading proxy), and derived features (power-speed relationship, wind load estimate).

Model selection: We evaluated multiple regression approaches — gradient boosted trees, ridge regression, and a simple LSTM for temporal dependencies. Gradient boosting performed best on the validation set, with the LSTM competitive on steady-state passages.

Results

The ML model achieves meaningful accuracy on voyage-level fuel consumption prediction using holdout passages. The model's predictions are useful for route optimization: simulating the fuel cost of speed profiles before committing, and identifying high-fuel passages for crew review.

This work is part of a research collaboration between SFU and the National Research Council of Canada focused on marine vessel optimization.