COMPARATIVE STUDY OF EMPIRICAL, PHYSICAL AND AI MODELS FOR PREDICTING SHIP FUEL CONSUMPTION UNDER REAL HYDROMETEOROLOGICAL CONDITIONS

Mingyang Zhang^a*, Nikolaos Tsoulakos^b, Manderbacka Teemu^a, Remes Heikki^a

^a School of Engineering, Aalto University, Espoo, Finland
^b Laskaridis Shipping Co. Ltd., Athens, Greece

Presented at the 44th International Conference on Ocean, Offshore and Arctic Engineering (OMAE2025) – Vancouver, Canada – June 2025

ABSTRACT

With the global maritime industry striving to reduce fuel consumption and emissions, accurate prediction models have become essential for optimizing ship operations under varying environmental conditions. This paper presents a comparative analysis of artificial intelligence (AI), semi-empirical, and physical models for predicting ship fuel consumption, with a particular focus on hydrometeorological influences. While empirical and physical models have traditionally been relied upon for estimating ship resistance and fuel consumption, recent AI advancements offer potentially superior alternatives in accuracy and adaptability. This research examines AI potential to either complement or exceed the performance of established methods, potentially setting new solutions for maritime fuel consumption prediction. Case studies are conducted on a Kamsarmax bulk carrier operated by Laskaridis Shipping Co. Ltd. The results reveal that the semi-empirical model, grounded in theoretical principles and empirical relationships, provides moderate accuracy in predicting ship fuel consumption. The physical model significantly improves predictive performance by incorporating dynamic simulations of ship behavior under varying sea conditions, using detailed hydrodynamic principles. In comparison, the AI-based model achieves the highest precision, demonstrating its ability to capture complex, nonlinear interactions and adapt to diverse operational scenarios. These findings highlight the transformative potential of AI in ship fuel consumption modeling, showing its capacity to set new benchmarks for efficiency and decision-making in maritime operations. This study may provide valuable insights for improving weather routing and operational decision-making, affirming AI transformative impact in maritime engineering.