After having relied for decades totally on small scale model tests, the design of anchoring systems for moored floaters like FPSOs is now widely performed numerically. Estimation of design maxima during mooring analyses requires calculating system response for a large number of sea states in order to screen all possible scenarios between wind, waves and current parameters. In addition, the slow-drift response motions of the system constituted by the floater and the anchoring system are highly dependent of the wave elevation realization, which is not an input parameter of the simulations and can lead to extremely variable responses.
This is generally addressed by designers by performing the analysis for the same sea state and varying the wave group spectrum (or wave components phases) a large number of times N (20–50 realizations is a typical range, see Refs , ). For these N realizations, N response maxima are extracted, and a distribution of response extremes is derived, from which the response level is extracted. In terms of computational cost, performing N 3-hour simulations to derive N values of extreme response is extremely expensive.
The paper will focus on methods that can be employed to reduce the computational cost of analyses. In a first step, the rapidity of statistical convergence of response estimates depending on the system will be investigated. This will allow pre-determining the number of sea states realizations required to reach a satisfactory convergence of response. In a second step, a mean of improving the computational efficiency of calculations carried-out to reach the statistical convergence will be proposed.