As exploration and production of offshore oil moves into deeper waters, mooring lines become longer and heavier, with total weight increasing up to about 10% of the floater's total displacements. The effects of line dynamics on vessel motions and mooring line tensions become much more prominent. Coupled time domain dynamic analysis is used extensively to assess the response of the mooring system. If the responses are narrow-banded Gaussian processes with Rayleigh distributed peaks, the extreme values are readily available. However, the mooring system responses calculated based on coupled time domain dynamic analysis are more likely to be non-Gaussian due to the prominent nonlinear line dynamics effect. Generally, the extreme value computed from a single time domain simulation will vary about the Most Probable Maximum Extreme (MPME) value. Consequently, statistical fitting techniques are recommended by API RP 2SK to establish reasonable confidence in the predicted extreme responses.
In this paper, a fully coupled time-domain dynamic analysis is performed for a typical deepwater mooring system with 20 different seeds/wave profiles using Orcaflex. The most probable maximum extreme (MPME) responses under different wave profiles are predicted with Winterstein/Jensen method, Weibull fitting method and Gumbel fitting method, respectively. This paper investigates the characteristics of these three statistical fitting methods and examines the convergence of these methods under different random seed selections. Winterstein/Jensen method and Weibull fitting method may result in scattered values under certain conditions, while Gumbel fitting method always yields consistent values. Recommendations are provided to eliminate the divergent results from using Winstertein/Jensen method and Weibull fitting method.