The responses of a flexible riser caused by motions of a weather-vaning FPSO subjected to two cyclonic storm events are investigated numerically. The objective was to comparatively study probability distributions of the riser responses to determine their statistical properties and the distribution types appropriate for particular responses. This will contribute to a method for the prediction of the extreme responses of the flexible riser during cyclonic events. Simulations of the riser responses in two cyclonic storm events were conducted including effects of wind, waves and current. Several types of probability distributions were fitted to the generated datasets using maximum likelihood estimates. Distributions tested included Rayleigh, Normal, Lognormal, Weibull, Generalized Extreme Value, Gamma, Burr, Birnbaum-Saunders and Beta. In addition, Hermite moment models were fitted in an attempt to use a single probabilistic model for all responses. A comparison was conducted using the goodness of fit tests to determine the distribution that best represented the generated data. Hermite moment models based on four moments produced good fits across both storm cases and for the majority of responses. The findings of this study provide foundations for development of more bespoke distributions for riser responses. The implication of using appropriate parent distributions for short and long term response analysis and the association between parent and extreme value distributions in both time and frequency domain contexts are discussed.
Dynamic responses of flexible risers in storm events due to vessel motions, external pressure and internal flows can have a detrimental effect on riser performance, possibly leading to its failure. Analysis of flexible riser responses is critical to determine the potential impacts of such storm events on riser performance.
The design of offshore structures and systems relies on estimations of the effect on structures in design and survival conditions as presented in the relevant guidelines and recommendations, such as American Petroleum Institutes (API), DNV-GL and Joint Industry Project (Fergestad and Løtveit 2014) design guidelines. The underlying assumption of many guidelines is that the maximum system responses can be described by the Rayleigh law. For the non-linear systems, such as risers and moorings, this assumption, which implies that the system is linear and its response is a Gaussian process, was found to significantly underestimate the extreme responses.