For the purpose of design of an ice-going vessel, the L-year maximum value of the iceinduced load acting on the ship hull is important. Such a maximum value can be obtained by means of classical statistical analysis for which it required to establish a long term distribution. In the present paper, a methodology is proposed which is referred to as the m-nautical mile maximum approach. The approach relies on the conditional distribution of ice load for given stationary ice conditions and/or vessel's speed. Two statistical models were applied and compared, i.e. the Weibull and the three-parameter exponential models. Full scale measurement data, which was obtained on board KV Svalbard during the winter of 2007 in the Northwestern part of the Barents Sea, was employed in order to demonstrate the application of the procedure. It was found that for this particular data set application of the Weibull model as the initial distribution leads to less conservative extreme values as compared to the measured maximum values. Predictions which were on the safe side of the observed maximum value were produced by instead applying the three-parameter exponential model as the initial distribution. The key benefit of the present proposed method is that it requires a quite limited amount of ice load data as long as the data is accompanied by a simultaneous ice thickness record.
展开▼