A quantitative ship risk model to select unannounced ship inspection - A tool to improve safety and environment?
MetadataShow full item record
- Institutt for marin teknikk 
Norwegian Maritime Directorate has been worked with development of the risk-based inspection. The Directorate has established a method for risk assessment at a general level. However, they do not have any concrete method for each vessel. Therefore, a method for quantitative evaluation of the vessel risk was established using risk influencing factors. In this master thesis, there has been done a literature study which describes the use of the Bayesian Belief Network (BBN) and risk influencing factors. The BBN is a method for risk analysis, and it is particularly suitable for handling uncertainty, non-deterministic and non-sequential relationships. The study clarifies the assumptions and limitations regarding the use of BBN, and develops a method for quantifying the interaction between risk influencing factors. The qualitative model developed in this work illustrates the interaction between the variables of human, organization, and technical risk factors. The model describes the cause and accident relationships between the factors with the use of arrows. The risk factors have been identified based on several literature surveys, expert knowledge together with the author s judgment. For the quantitative model, the input data has been based on the author s judgments and the results from several marine accident researches. The author s decisions took a considerable part when establishing the importance of each risk factor, and the extent of the factors influence to its child node. Some of the risk factors decided in the qualitative part of the model had to be excluded when building the quantitative model due to the limitation of real data. The model was tested with the inspection station at the NMD, the inspected vessel database from the year 2015, and the demonstration of the model in a practical use. The model gives expected overall results of the testing of the vessels from the year 2015 where all of the vessels have to be chosen for the unannounced inspection. The model therefore has a quantitative utility value, but some of the factors differ slightly from the first two cases discussed. The following discussions lead to the conclusion that the model has an objective value and it is suitable for the vessels with the inspection historic; meaning that the model is not for the new vessels since they do not have any deficiencies, yet. By using the model it is possible to monitor the risk level of vessels, thus prioritize them subsequently with the percentage of risk level. However, by implementing more risk factors related to the national vessels, and especially clarify between new and old vessels, the model could be refined giving a more accurate and desired result for the selection of unannounced inspection.