Sea State Duration and Probability of Occurrence of a Freak Crest

For design purpose extreme wave crests can be determined by direct extrapolation or by estimating an extreme sea state condition. In the latter procedure it is common to estimate a n-year Hs level as a fractile of a marginal long-term distribution of significant wave height, combined with adequately chosen characteristics values for the other sea-state parameters. In this approach a sea state duration needs to be assumed when extremes are calculated. In the present study sensitivity of extreme crest characteristics to an assumed sea state duration is investigated for wave records which include a freak event. The wave time series recorded at the Draupner platform, January 1, 1995, and the data registered at Ekofisk, October 25, 1998, are used in the analysis. Predictions of several models are compared with the prediction given by the 2nd order Forristall crest distribution. Implications of the results for the current design practice are discussed.