Evaluation of uncertainties in the simplified fatigue method

Abstract

The use of external piping is becoming more common for semi-submersible platforms. Previous analyses have shown that uncertainties exist in how the methods used for assessing the fatigue limit state should be applied. One method of interest, which is offered by classification society rules for offshore structures, is the simplified fatigue method. The expression used for estimating fatigue damage only requires S-N parameters, distribution parameters of the stress range history and a design life. The objective of this thesis has been to identify and evaluate factors that have a great influence on the calculated fatigue damage in the simplified fatigue method when applied on appendages in the splash zone. The evaluation was conducted on a case study regarding external piping on a semi-submersible production platform. By using a Weibull model to represent the stress-range distribution the following critical factors could be identified: the shape parameter, the reference stress range, and the chosen SN parameters. Methods for assessing them individually were developed. An assumption of a linear relation between stress range and wave height was used. The distribution parameters could then be obtained using the results from a linear finite element analysis to scale the distribution of wave heights. The conclusions obtained through the devised methods showed that the estimated damage is heavily dependent on how stress ranges are represented in the Weibull fit. This is dependent on both the shape parameter and the reference stress range. Contributions from stress ranges with a high magnitude dominate even though the occurrences of such stress ranges are very low. A subdivision of stress-range distributions corresponding to, for example, different wave headings gives improved estimates. More Weibull-like distributions can be obtained and scaled with the corresponding stress ranges. It was also noted that small changes in load input data increases the estimated fatigue damage significantly. Thus, it is very important to have an understanding of the uncertainties in the hydrodynamic analysis used for deriving the loads. Moreover, the choice of S-N data also shows a large impact, although it is hard to determine which weld classification should be used when only one reference stress range is considered. It is important that future work is conducted in order to verify the simplified fatigue method. Different alternatives in deriving the wave-height distribution, or one of the other fatigue assessment methods offered by class could be used.