Abstract
Loop Current and Eddies (LCE) occur in the Gulf of Mexico and can cause fatigue of tubular systems such as drilling and production risers and wellhead/casing systems by vortex-induced vibrations (VIV). It is a standard practice to perform a cumulative fatigue assessment of risers and wellhead/casing systems using a site-specific scatter diagram of LCE. A scatter diagram consists of a number of full-depth current speed profiles and associated probabilities of occurrence for each profile. Scatter diagrams often consist of too many current profiles, which can make the VIV analysis computationally very expensive and intractable. On the other hand, if the higher-order modes are excluded in order to reduce the number of profiles, fatigue estimates may be inaccurate and sometimes unconservative.
In this paper, a more efficient binning procedure is presented for deriving the LCE scatter diagram using 1000 years of simulated data from the GEM42 hindcast model. The hindcast data consist of mode shapes and amplitudes of the first three empirical orthogonal functions (EOFs) of the LCE. The statistics of the amplitudes of the three EOF modes are carefully investigated to find relevant trends that guide the binning procedure. It is first shown that using all three modes is required since reconstructing current profiles with only two modes results in significant speed errors. Further, VIV fatigue analysis of a wellhead/casing system shows that using only two modes can result in significantly unconservative fatigue life estimates. The joint distribution of the last two mode amplitudes is analyzed and found to be polar. Therefore, the three mode amplitudes are mapped to a cylindrical coordinate system, and it is found that cylindrical binning is more efficient, by a factor of more than 2, as compared to traditional binning in the Cartesian coordinate system. Finally, cumulative VIV fatigue analysis of a drilling riser and wellhead/casing system is performed to investigate the effect of different binning choices (i.e., bin widths, coordinate system, and number of modes) on the end result, i.e., fatigue estimates. Fatigue life results show that cylindrical binning can be significantly more efficient without losing accuracy.
