Bayesian analysis of short term extreme mooring tension for a point absorber with mixture of gamma and generalised pareto distributions

Abstract

Model tests have been conducted to study the performance of slack, catenary and hybrid mooring systems for a point absorber in survival sea states. The snap loads can be found in mooring tension time series of all the tested mooring configurations. The classical Weibull distribution is applied to model the extreme dynamic mooring tensions. In addition to the Weibull distribution, the Bayesian inference method with a new Markov chain Monte Carlo sampling procedure is proposed to study the short term extreme mooring tensions based on model test results. Before applying this method to analyse extreme mooring tension, its accuracy is validated. The probability densities as well as short term extreme dynamic tension estimated by a mixture of Gamma and Generalised Pareto Distributions models are compared with the two-parameter Weibull model results. In the end, an extensive discussion of about the mixture of Gamma components, peak definitions, threshold and Markov chain Monte Carlo iteration on Bayesian inference is carried out. It is validated that the mixture model proposed in this paper shows good performance in predicting the extreme mooring tensions even when the snap loads occur, while the Weibull distribution fails to give accuracy predictions.