Measurement-based relationships between container ship operating parameters and fuel consumption

Abstract

Increases in fuel prices and the need to decrease emissions have made the optimization of fuel consumption on ships more critical. Developing a method to accurately estimate fuel consumption in ship operations is crucial for this reason. This article determined the mathematical relationships between fuel consumption and operation parameters, such as rotational speed, draught, trim, hull fouling time, wind speed, wave height, and seawater temperature, as well as the hierarchical impact of these parameters on fuel consumption. The study used a dataset of 105,790 measurements taken on a 4,800 TEU container carrier that was in operation for 1,187 days after hull cleaning. Data-driven approaches such as Artificial Neural Networks (ANNs) and a Multiple Nonlinear Regression (MNR) with Random Search were used to develop a relationship between the ship operating parameters and fuel consumption. A comparison of statistics between the two methods showed that both ANN and MNR provided highly accurate fuel consumption estimations, with MAPE of 6.1/6.41%, RMSE of 5.01/4.98 t/day, RRMSE of 7.22/7.12%, PCC of 0.964, and R2 of 0.93. The relationships presented here could be invaluable for ship route optimization, as well as transport and air pollution studies for similarly sized container carriers.