Fatigue damage prediction in deepwater marine risers due to vortex-induced vibration

Abstract

Slender marine risers used in deepwater applications often experience vortex-induced vibration (VIV). Fatigue damage associated with VIV is of great concern to offshore engineers; however, it has proven difficult to predict this fatigue damage using existing semi-empirical tools. Similarly, approaches based on theoretical and computational fluid dynamics (CFD) generally rely on simplified assumptions on the fluid flow fields and response characteristics. To gain an understanding of VIV and associated fatigue damage, full-scale field monitoring campaigns as well as reduced-scale laboratory experiments are often carried out, wherein the riser response in the form of strains and/or accelerations is recorded using an array of a limited number of sensors distributed over the length of the riser. Simultaneously, current velocities at a proximate location are also recorded. Such measurements generally reveal complex characteristics of the dynamic response of a riser undergoing VIV, including the presence of multiple vibration harmonics, non-stationary behavior, and the existence of sustained or intermittent traveling wave patterns. Such complex features, often not accounted for in some semi-empirical and theoretical approaches, are critical to take into consideration for accurate fatigue damage estimation.

In this study, several empirical methods are employed to first reconstruct the response of an instrumented riser and, then, estimate fatigue damage rates over the entire span of the riser based on a limited number of discrete measurements. The methods presented employ the measured data in different ways. One method, referred to as ``weighted waveform analysis'' relies on expressing the riser response as a summation of several weighted waveforms or riser modes; the mode shapes are ``assumed'' and time-varying weights for each mode are estimated directly from the measurements. The riser response over the entire span is reconstructed based on these assumed mode shapes and estimated modal weights. Other methods presented extract discrete mode shapes from the data directly. With the help of interpolation techniques, continuous mode shapes are formed, and the riser response is again reconstructed. Fatigue damage rates estimated based on the reconstructed strains obtained using the various empirical methods are cross-validated by comparing predictions against direct measurements available at the same locations (but not used in the analyses). Results show that the empirical methods developed here may be employed to accurately estimate fatigue damage rates associated with individual recorded segments of measurements.

Finally, a procedure for prediction of long-term fatigue damage rates of an instrumented marine riser is presented that relies on combining (multiplying) the fatigue damage rates associated with short recorded segments for specific current profile types, with the relative likelihood of different incident current profiles, and integration over all current profiles. It should be noted that the empirical approaches to fatigue damage estimation presented in this study are based only on measured data; also, they explicitly account for different riser response characteristics and for site-specific current profiles developed from metocean studies. Importantly, too, such estimation procedures can easily accommodate additional data that become available in any ongoing field monitoring campaign to improve and update long-term fatigue damage prediction.