Toward Energy-Efficient Navigation: A Data-Driven Approach for Flowmeterless Estimation of Ship's Engine Fuel Consumption in Real-Time

Reducing greenhouse gas emissions from ships is a significant challenge facing the maritime industry. In particular, energy-efficient navigation techniques allow shipowners to adapt their itineraries to minimize fuel consumption. However, the evaluation of the effectiveness of these techniques is slowed down by lack of data on the actual consumption of the engines. Direct measurement of fuel using a flowmeter seems an ideal solution to this problem. However, many shipowners do not adopt it since it requires significant modifications to the fuel inlet piping, reducing ships reliability and requiring significant initial investments. This paper proposes a new, less invasive, data-driven approach to estimating the real-time fuel consumption of engines without using a flowmeter on each ship or each engine of a fleet. The strategy is based on a lookup table fitted in a three-dimensional space composed of torque, rotational speed, and flow measurements gathered on real ships. The model is trained and validated with approximately 464 hours of actual measurements on tugboats operating in North American waters over the last five years. The developed model enables shipowners to evaluate at a low cost the effectiveness of operational measures to reduce their fuel consumption.