This paper focuses on the short term probabilistic analysis of ice loads acting on a ship hull. The ice load data was obtained from full scale measurement onboard the Norwegian coast guard vessel KV Svalbard during the winter of 2007. The available data corresponds to discrete peak amplitude time histories of estimated ice impact loads as well as corresponding measurements of ice thickness in addition to ship speed and course. There were several number of sensors installed along the hull, either on the port side and starboard side of the bow part. The present paper focuses on the variation of the predicted extreme ice loads acting on the ship hull for a short time duration. The short term prediction of ice loads as an integral part of an Ice Loads Monitoring (ILM) system is very important in relation to the tactical navigation plan. An inexpensive ILM system would requires less number of sensors mounted on the hull. By addressing the variation of the extremes along the hull, it will be possible to make decisions regarding the minimum number of sensors and their location without loosing the accuracy of the predicted extremes. Three different approaches for predicting the short term extremes are considered, i.e. the classical extreme value distribution approach, the time window approach, and the up-crossing rate approach. In general, all the approaches involve the following two steps: (i) establishment of the estimated distribution model, (ii) calculation of the expected largest extreme ice impact load for an extrapolated duration. Comparison of the results obtained by the three different approaches is made, and some limitations of the various approaches are discussed.

This content is only available via PDF.
You do not currently have access to this content.