A Feasibility Study of Assessing Bunkering Operations Through AIS Data
Master thesis
Permanent lenke
http://hdl.handle.net/11250/2614960Utgivelsesdato
2018Metadata
Vis full innførselSamlinger
- Institutt for marin teknikk [3437]
Sammendrag
This thesis is a result of utilising Automatic Identification System data to construct a framework for identifying bunkering operations and conducting a statistical analysis of identified operations. The two objectives of the project are to identify bunkering operations and derive a benchmark that can quantify operation performance. Three relevant fields are studied to gain fundamental knowledge prior to attempting to achieve the objectives. 1) A brief study of the bunker supplier industry, 2) a literature review, and 3) a study of the basics of AIS data.
Knowledge on the bunker industry is to a certain extent obtained from industry experts. Effects from lack of transparency are described - focusing on how this prevents monitoring and evaluation of the bunker process. Furthermore, reasons for increasing awareness are recognised. Literature was reviewed in previous work and is divided in two parts. First, literature is categorised according to topics and methods. Second, the most relevant literature is further assessed, where studies related to topics Risk Assessment and AIS data handling are considered the most relevant. Two additional studies regarding mapping of traffic patterns and benchmarking were appraised at later stages. The conclusion of the literature study is that no studies explore use of AIS data to evaluate bunkering operations, thus forming an unexplored academic field.
Three sub-problems attempt to achieve the objectives of the thesis. The first objective is covered by problem 1 in which a framework for identifying bunkering operations is constructed - resulting in a matching algorithm that matches ships with the barge that most likely performed a bunkering operation. Inputs are AIS data, measured fuel quality data from bunkerings and a list of officially approved bunker barges, restricted to Singapore. Output from the algorithm is bunkering operations that can be verified with high likelihood based on a certainty measure composed by proximity and alignment between ship and barge.
The second objective consists of problems 2 and 3. In problem 2, statistical analysis of identified bunkering operations is conducted. Aggregation and visualisation of data are conducted in the programming language Python, yielding numerous plots. This has mainly been restricted to scrutinising distributions of time spent at different stages during operations. In problem 3, results from the statistical analysis are utilised to establish an index that quantifies each bunkering operation with time spent before and after the bunkering operation as decisive parameters.
In conclusion, the algorithm is a "proof of concept" which proves that AIS data can be utilised to identify ship-to-ship operations. The index is a guide to evaluate operations, though insufficient to quantify performance by suppliers. Recommendations for further work are that a Geohashed AIS table should be implemented in the matching algorithm to filter nearby vessels more efficiently, and the algorithm should be expanded to identify perpendicular operations. Parallel data with known fuel quality from bunkering operations should be compared with results from the proposed index. Hence, index parameters can be optimised, and validity be verified. Lastly, methods used in this thesis can be expanded to other types of ship-to-ship operations.