Short term extreme statistics of local ice loads on ship hulls

This paper focuses on prediction of short term extreme values of local ice loads on ship hulls. The analysis is based on the exact classical method, which utilizes the selected statistical models as its initial distribution. A conventional approach, where the number of events is considered as a fixed value, i.e. equal to the expected number of events for a given duration, is employed. An alternative approach, which accommodates the number of ice load peak events as a random variable, is also investigated. An asymptotic approach, which is based on the one-minute maxima, is also investigated. When compared with the other two methods, the asymptotic approach tends to give similar level of extreme values.

► The effect of treating the number of local ice load peaks as a random variable. ► Different methods for extrapolation to extreme values. ► Comparison between different methods for prediction of characteristic values versus fractiles of extreme-value distribution.