Application and improvement of iterative methodology for response based analysis of an FPSO mooring system in tropical cyclones

Abstract

Response based analysis (RBA) is an advanced alternative to traditional design approach for mooring systems of floating structures, which reduces uncertainty associated with extreme responses. In the traditional approach, mooring analysis is performed for specified sets of extreme metocean conditions, representing typically a return period of 100 years. In RBA, the dynamic analysis is carried out over a long-term database of metocean conditions and statistics of the responses are used to identify responses associated with a return period of 100 years. Whilst the RBA has a more rigorous probabilistic basis, it is also computationally demanding. An “Iterative methodology” has been developed recently, comprising a combination of time domain (TD), frequency domain (FD) and probabilistic analyses of the structural system into a single iterative process, to minimise the number of time-consuming simulations through distinguishing between “mild” and “severe” response events. In the study, the “iterative methodology” is applied for RBA of the mooring system of a turret moored FPSO in the tropical cyclone environment of the North West Shelf of Australia. In the first stage, the time-efficient FD analysis is utilised to identify the contribution of all the storms to the 100-year response. In the second stage, an iteration process is set up, in which only the most contributing response events (tropical cyclones) and sea states are analysed by TD simulations. The iterations proceed by selecting the next most contributing storm until the convergence of the 100-year response is achieved. Application of the iterative methodology leads to significant reduction in computational time and makes RBA feasible as a part of the usual design process by using fully coupled TD analysis. Further optimisation can be achieved through a new method of applying thresholds to metocean parameters, which can reduce the analysis effort further. In the case study considered, it is found that 61% of metocean time history could be excluded from any dynamic analysis by applying a variable storm-dependent threshold. Among the remaining intervals, only 5.2% need to be re-analysed by TD simulations. Several methods for assessing errors in the convergence process are compared, and a sensitivity analysis on the number of storms that should be selected at each iteration step is performed.