Extreme value behaviour of slow-drift motion of moored floaters is investigated. A need for prediction methods other than the "standard" Rayleigh model is identified, as the Rayleigh model will normally underpredict the largest maxima. This is related to the fact that the exciting drift forces are approximately exponentially distributed. An existing, general method for expressing the response statistics of 2nd order systems as a combination of purely exponentially distributed contributions is, based on certain assumptions on the quadratic transfer function (QTF), reduced to a simple extreme value estimation method. Thus the deviation from the Rayleigh model for extreme value behaviour is determined simply by the bandwidth of the slowdrift response spectrum. The method predicts that for practical use, extreme values in cases with high motion damping (> 20 %) should be predicted as if the response were purely exponentially distributed, with expected extreme amplitudes (around the mean offset) from 50 to 100 % higher than predictions based on the Rayleigh model. For lower damping, the deviation from the Rayleigh model is somewhat reduced, but the use of the Rayleigh predictions is not recommended unless the damping is very low « 2 %). Preliminary comparisons to numerical simulations and a model test experiment show a reasonably good agreement. Further comparison studies are recommended.
The horizontal motion of moored floating vessels and structures in irregular waves is often dominated by a significant lowfrequency contribution excited by non-linear wave drift forces, see e.g. [1–3]. The importance of these responses arises as a result of low natural oscillation frequencies of the moored systems, and also because the motion damping is often rather low. The extreme value estimation of linear (wave frequency) responses is normally made directly from the standard deviation by assuming Rayleigh distributed maxima (Gaussian statistics) of the responses.
