Analysis of Random Wave Interaction with Cylinders using Extremal Statistical Methods

Abstract

Long slender cylindrical structures are widely used in offshore structural designs. Their interactions with random waves are very complicated and it is extremely difficult to model their motions in a direct analytical approach. Accurate characterization of the observed extreme values in their response behavior is of vital importance for the safety of the designed structure. In this study a general statistical methodology based on generalized extreme value (GEV) family of distributions is formulated for the characterization of both laboratory and field data associated with the wave-cylinder interactions in random waves. An iterative process is developed to determine the most appropriate block size and the corresponding statistical model for the block maxima constructed from the time series of the target variable. The model's capability to fit the data is assessed using the Anderson-Darling (AD) test criterion, visually with quantile plots and histograms. Special attention is paid to the quality of fit in the upper tail of the distribution, which corresponds to the data points of the extreme statistical events. Case studies are performed that investigate the performance of the statistical methodology in characterizing the measured data from two industrial scale model basin test programs. The first case study addresses data on the in-line interactions of closely spaced deep-water cylinder arrays for two- and three-cylinder configurations and the second considers the VIV response amplitudes of a slender horizontal cylinder. In both test programs the random seaways were generated using JONSWAP wave amplitude spectrum. The effectiveness of this statistical methodology in characterizing the response behavior observed in the model basin experimental data is discussed in detail using both tabular and graphical interpretations.