Using Model Test Data to Assess VIV Factor of Safety for SCR and TTR in GOM

Abstract

This paper presents results obtained as part of the DeepStar Phase 10 program on VIV Factors of Safety. The objective was to develop a general methodology to calibrate Factors of Safety for VIV-induced fatigue and to apply it to partially straked risers. This was achieved using reliability methods, accepted industry VIV prediction software and state-of-the-art model test experiments. Most oil companies use a Factor of Safety of 20 when predicting VIV damage using VIV software tools. There are numerous software tools currently in use in industry to predict VIV damage to straked risers and each of them will have different accuracy, and therefore an intrinsic level of conservatism. Understanding the level of conservatism in different VIV prediction software is therefore critical to determining what Factor of Safety to use. This study benchmarks the latest generation of industry accepted VIV design tools at the time of the study (2011): SHEAR7v4.6, VIVAv6.5 and VIVANAv3.7.24 against high quality VIV data from three separate straked riser experiments. A bias distribution (predicted to measured VIV damage results) is obtained for each software tool as a function of the strake coverage. A novel reliability framework approach is then developed to incorporate all uncertainties associated with VIV fatigue prediction into a limit state function, including variability in met-ocean conditions and variability in the fatigue resistance of the material characterized by a design S-N curve. The limit state function is analyzed using First Order Reliability Methods to develop Factors of Safety for target probabilities of failure. The general method is then applied on two case studies involving an SCR and TTR in Gulf of Mexico loop currents, but it can be easily extended to different locations and riser configurations. The resulting FoS range from about 1 to 15 for most software, and are lower than industry standards for VIV prediction. The FoS do not vary markedly for different riser configurations, indicating the possibility of reducing excess conservatism when predicting VIV damage on straked risers.