Master thesis : Vessel Performance Analysis using High Frequency Operational Data

Abstract

The increase in Greenhouse Gas (GHG) emissions due to shipping and its impact on climate change is a concern for the maritime industry. Urgent steps are required to achieve the International Maritime Organization (IMO) target of 50% GHG reduction by 2050. The recent boom in digitalization of the shipping industry can be vital to improving the energy efficiency of vessels. The high frequent operational data obtained from the onboard sensors can be used to analyse the decrease in ship’s performance over time due to aging and sufficient measures can be suggested timely to achieve the maximum possible efficiency. This master thesis investigates the performance of an intercontinental 14000 TEU container vessel using high-frequency operational data. The operational data has been ob tained from University of Rostock server. Statistical outliers are removed using ISO19030 standards. Then this data is further filtered for different environmental and operational conditions using simple filtering techniques and Machine Learning algorithms. Outputs of both methods are compared with the design data to analyze the vessel’s performance and the most suitable method is proposed. Moreover, the Energy Efficiency Operational Indicator (EEOI) of the vessel is investigated over time. Results show that the perfor mance of the vessel is constant over the investigated time period. A good agreement has also been found between the operational data with the design data. Furthermore, this study also compares the performance of different machine learning algorithms; Random Forrest, Decision Tree, Gradient Boosting, Multilayer Perceptron and Least square methods over the filtered high-frequency data set. Blind testing results show that the Random Forest algorithm has the best performance