Statistics of local ice load peaks on ship hulls

This paper focuses on statistical inference in relation to the peak process of local ice loading on ship hulls. Such inference comprises selection of a statistical model (by means of P–P plot, Q–Q plot, and probability paper), parameter estimation (by means of LSM, MoM, and MLE), evaluation of the confidence interval of the estimators, and goodness-of-fit testing of the initial distribution. Several different approaches are considered for each of the steps and numerical experiments were performed in order to reaffirm suggestions from previous works and extend the analysis. A particular data set provided by full scale measurements on the coast guard vessel KV Svalbard during the Winter 2007 was analyzed by means of the present approach. Four probability models were tested and the best fit to the data was provided by the Weibull distribution with the shape parameter being located in the range 0.7 < k < 1.0.

▸ A systematic manner of statistical inference of ice load peaks on ship hull is outlined. ▸ Therefore, a well established model can be taken confidently as probabilistic model. ▸ Ice-induced load data from the Barents Sea is presented. ▸ The Weibull's is a good model when the shape parameter is in the range 0.7 < k < 1.